Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Coronary revascularization in British Columbia, 1979-1988 Gait, Jennifer Mary 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1992_spring_gait_jennifer.pdf [ 13.61MB ]
JSON: 831-1.0086238.json
JSON-LD: 831-1.0086238-ld.json
RDF/XML (Pretty): 831-1.0086238-rdf.xml
RDF/JSON: 831-1.0086238-rdf.json
Turtle: 831-1.0086238-turtle.txt
N-Triples: 831-1.0086238-rdf-ntriples.txt
Original Record: 831-1.0086238-source.json
Full Text

Full Text

CORONARY REVASCULARIZATION IN BRITISH COLUMBIA: 1979 - 1988 By  JENNIFER MARY GAIT B.S.N. The University of British Columbia, 1979  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE (Health Services Planning and Administration) in THE FACULTY OF GRADUATE STUDIES Department of Health Care and Epidemiology  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA April 1992 © Jennifer Mary Gait, 1992  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature)  Department of  1 1t..oLtql Cal-k.... ex 2?‘ 6Ur I'M0 10t\-  - -  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  M a t 1 1 l 0l g 2,  -  -  1  ii  ABSTRACT Since the introduction of coronary artery bypass surgery (CABS) in the late sixties, the increase in the incidence rates has aroused controversy in the literature. Recent studies in the United States and Canada have documented both large rate increases in the elderly and geographic variations in incidence rates. This study was undertaken to discover whether similar patterns exist in British Columbia. Data from the British Columbia Hospital Morbidity Database, for fiscal years 1979 through 1988, were used to calculate age-sex adjusted small-area incidence rates based on the school district of residence. Results showed a 1.2 fold increase in overall annual rate with a two-fold increase in the elderly. The greatest increase, almost nine-fold, was seen in the population aged 75 and over. In addition, the percentage of patients with either diabetes or chronic obstructive pulmonary disease increased from three to twelve percent of annual cases. Over the study period, extreme variability in annual rates was seen both within and among school districts. Within school districts, most variability was seen in districts with populations below 10,000. Poisson regression (which weighted school districts according to population size) showed that variation among school districts was highly significant (p<0.0001). In an attempt to explain the variation in small area rates, the CABS rates for each school district from 1983 to 1988 were regressed on six ecological variables (distance from cardiologist, distance from internist, distance from centre, income, employment rate and graduation rate) with year and year-squared forced in. Income, distance from cardiologist, distance from centre and their first-order interactions were found to be important explanatory variables (R 2 = 0.21). While income and distance from cardiologist had a negative effect on the CABS rate,  111  distance from centre had a surprising positive effect, which did not appear to be accounted for by colinearity with distance from cardiologist. The model was then refitted, using the CABS rate adjusted for morbidity in the school districts as the dependent variable. In this model distance from cardiologist and income changed in relative importance, and distance from centre and the interaction between variables were no longer important (R 2 = 0.30). Refitting this model to account for mobility to Alberta showed that distance from cardiologist and income explained more of the variation in rates (R 2 = 0.34). The presence of small-area variations in CABS rates, together with differences between centres in the number and type of procedures performed, suggest that there are inequities in cardiac care within B.C. These inequities appear to arise from complex relationships between distance from services and morbidity rates and average income in the school district of residence. In addition, it appears that the surgical centre referred to may also contribute to the variation in smallarea rates, although this was not tested. It is clear that inequities in cardiac care cannot be redressed by simple solutions. Policy implications and suggestions for further research are discussed.  iv TABLE OF CONTENTS  ABSTRACT^  II  TABLE OF CONTENTS^  IV  LIST OF MAPS^  IX  LIST OF TABLES^  X  LIST OF FIGURES^  XII  ACKNOWLEDGMENTS^  XV  INTRODUCTION^  1  CHAPTER 1^CORONARY ARTERY DISEASE^  4  INTRODUCTION^  4  THE PROBLEM OF CAD IN CANADA^4 CAUSE AND EFFECT OF CORONARY ARTERY DISEASE^ MANIFESTATIONS OF CORONARY ARTERY DISEASE^ Myocardial Infarction^ Chronic Stable Angina^ Unstable Angina^  5 8 8 9 10  RISK FACTORS^ 21 Stages of Disease^ 12 Primary and Secondary Risk Factors^13 14 Hypercholesterolemia^ Hypertension^ 14 Tobacco Smoking^ 15 Diabetes Mellitus^ 15 High Density Lipoproteins^ 16 Obesity^ 16 Family History^ 16 Physical Activity^ 17 Age and Sex^ 17 TRENDS IN CORONARY ARTERY DISEASE^18 MORTALITY  V  DIAGNOSIS IN CAD^ Coronary Angiogram^ Exercise Stress Test^  21 21 23  PREVENTION AND TREATMENT OF HEART DISEASE^ Prevention^ Treatment^  25 25 29  CONCLUSIONS^  30  REFERENCES^  32  CHAPTER 2^CORONARY REVASCULARIZATION TECHNIQUES 37 PART 1 CORONARY ARTERY BYPASS SURGERY 37 HISTORY^ Growth and Utilization^ Assessment^ Changes in CABS Procedure ^ Changes in Medical Treatment^ Changes in CABS Patients^  38 38 40 42 43 43  EFFICACY OF CABS^ 44 The Intent-to-Treat Principle^ 45 Chronic Stable Angina^ 46 Unstable Angina Asymptomatic Patients with "Silent" Myocardial Ischemia^ 63 Post-Myocardial Infarction^ 64 Evolving MI^ 66 Emergency CABS after Failed PTCA ^68 Re-operations^ 69 The Elderly^ 70 Gender^ 76 RISKS^ Operative Mortality^ Morbidity^ Other Risks^  78 78 81 82  COSTS^  85  CONCLUSION^  86  CORONARY REVASCULARIZATION TECHNIQUES 98 REFERENCES^  90  vi CHAPTER 2^PART II PERCUTANEOUS TRANSLUMINAL ^98 CORONARY ANGIOPLASTY HISTORY^ Trends in Assessment^ Changes in PTCA^  98 98 100  MECHANISM OF ACTION^  101  101 EFFICACY OF PTCA^ PTCA in Single Vessel Disease^102 108 Multi-Vessel Disease^ PTCA in Unstable Angina^ 112 PTCA in Left Ventricular Function ^113 Post-Myocardial Infarction ^ 113 114 Gender^ 115 The Elderly^ 117 Unsuccessful PTCA^ 119 Summary^ RISKS^ Hospital Mortality^ Emergency Bypass Surgery^ Other Factors Affecting Risk^  120 122 123 124  COSTS^  126  CONCLUSION^  128  REFERENCES^  131  CHAPTER 3^REGIONAL VARIATIONS^ INTRODUCTION^  135 135  PROCEDURAL VARIATIONS IN GENERAL ^135 Factors Associated with Procedural Variations^138 VARIATIONS IN CABS UTILIZATION RATES^140 ISSUES IN VARIATIONS RESEARCH ^146 CONCLUSION^  152  REFERENCES^  153  vii CHAPTER 4^RATIONALE AND METHODOLOGY^156 RATIONALE^  156  Questions^  157  METHODS^ Study Design^ Independent Variables^ Data Sources^ Study Population and Analysis ^ 1. Descriptive Analysis^ 2. Regression Analysis^  157 157 158 159 160 160 162  CHAPTER 5^REVASCULARIZATION IN BRITISH COLUMBIA 166 1979-1988: RESULTS CORONARY ARTERY BYPASS SURGERY ^166 Characteristics of the CABS population^166 171 Re-operations^ Region of Residence of CABS Population ^173 180 Referral Patterns^ ANGIOPLASTY^  184  REGRESSION ANALYSIS^  185  SUMMARY^  194  REFERENCES^  195  CHAPTER 6^REVASCULARIZATION IN BRITISH COLUMBIA 197 1979-1988 DISCUSSION OF RESULTS AND POLICY IMPLICATIONS OF STUDY DISCUSSION^ DESCRIPTIVE STUDY^ Incidence Rates^ Regression Analysis^  196 196 196 203  LIMITATIONS^  207  POLICY IMPLICATIONS^ 210 Organization Of Health Care in B.C.^211 Implications of Literature and Present Study ^211 Cost Effectiveness^ 213 214 Inappropriate Treatment^ 217 Other^  viii RESEARCH^ REFERENCES^ ^  APPENDIX A ^ APPENDIX B  218 221 222 265  ix  LIST OF MAPS  Map A^British Columbia School Districts ^  279  Map B^School Districts Sending a Plurality of Cases to Centre ^280 Map C^School Districts Sending 90% or More of Cases to One Centre ^281  x  I 2 3 4 5 6 7 8  LIST OF TABLES Inclusion criteria for the three major RCTs Characteristics of patients in RCTs for stable angina Outcomes in RCTs for stable angina Inclusion criteria for RCTs of unstable angina Randomized trials for unstable angina Outcomes in RCTs for unstable angina Percentage of patients surviving five or more years after reoperation Unadjusted cumulative 6-year survival in age subgroups of the Over-65's Outcomes of CABS in CASS registry patients  9 10 Predictors of operative mortality 11 Cost effectiveness of CABS as a function of the severity of angina 12 Comparison of outcomes in controlled trial of ptca versus CABS 13 Comparison of old- and new NHLBI registry characteristics and outcomes by extent of disease 14 Comparison of one-year outcomes in elderly and non-elderly patients after successful and unsuccessful PTCA 15 Regression models 16 Annual CABS procedures for B.C. and out-of province residents 17 Annual CABS procedures by sex 18 Annual mean age of CABS procedures per year 19 Annual standardized incidence ratios and CABS rates 20 Age-sex specific rates for coronary artery bypass surgery 21 Distribution of cases by age-group 22 Annual CABS procedures performed on sub-groups of the over-65 population 23 Annual standardized incidence ratios and sex adjusted rates for subgroups of the over-65 populations 24 Annual numbers of patients with comorbidity receiving isolated CABS 25 Distribution of comorbidity by age group isolated coronary artery bypass 26 Out-of-province revascularization services for B.C. Residents  48 50 53 58 59 61 71 72 74 79 86 107 109 117 165 224 224 225 167 168 168 226 226 227 227 228  xi 27 Numbers and mean age of B.C. Residents receiving CABS in Alberta 228 28 Annual distribution of isolated CABS by school district ^229 29 Annual populations, rates and standardized incidence^231 ratios by school district 30 Variability of CABS and standardized incidence ratios ^256 within school districts 31 Standardized incidence ratios for five and ten-year periods ^258 32 Effect of migration to Alberta for CABS procedure on^260 standardized incidence ratios in B.C. school districts 33 Percentage of CABS cases from each school district using B.C.^261 centres 34 Coronary artery bypass surgery per centre per year ^ 263 35 Percent of open heart surgery devoted to CABS annually by centre ^263 36 Distribution of CABS patients with comorbidity between centres ^182 38 Increase in non-CABS revascularization procedures^182 39 Mean age by sex "no-CABS" and angioplasty populations ^264 183 40 Numbers and percentages of revascularization^ procedures by centre 41 Growth and decline of the 4800 CCP code by centre^ 265 41(a) Distribution of angioplasty (CCP codes 4801-4805) by centre^265 267 42 Independent variables simple statistics^ 43 Independent variables Pearson correlation coefficients^268 44 Poisson regression variables explaining variation^ 269 in CABS rate across all school districts 45 Poisson regression variables explaining variation in ^270 CABS rate across school districts with adjustment for mobility 271 46 Poisson regression saturated models ^ 47 Poisson regression variables explaining variation in morbidity-adjusted CABS rate across all B.C. school districts ^272 48 Poisson regression variables explaining variation in ^273 morbidity-adjusted CABS rate across school districts with adjustment for Alberta 49 Regional divisions used in age-sex-year-region 274 regression analysis^ 50 Poisson regression results interactions between age, sex, year ^275 and region for metropolitan/urban/rural/remote regions 51' Poisson regression results interactions between age, sex, ^277 year and region for geographic regions  xii LIST OF FIGURES  84  1  Outcome for patients referrd for second opinion  2  Annual age-sex adjusted cabs rate per 10,000 population  167  3  Age-use curves B.C.  170  4  Scattergraph of coefficient of variation of observed CABS in school district by school district population  175  5  Range of standardized incidence ratios for isolated CABS by school district  176  6  Growth of revascularization procedures  183  ACKNOWLEDGMENTS  I would like to acknowledge the help of my committee members Dr. Geoff Anderson and Dr Charles Wright for the many hours they spent reviewing and for their helpful comments and suggestions during the planning and implementation of this thesis. Especial thanks go to my thesis supervisor, Dr Sam Sheps, whose lively interest and direction kept me going when I felt overwhelmed, and to committee member Dr Stephen Marion for his patient help and guidance in the statistical analysis. Thanks also go to Sandi Wiggins who helped me through my initial encounters with the mainframe computer and SPSS-X, and to Dr Ken Benson for his detailed and helpful critique of an early draft of Chapter One. I also want to acknowledge the support, both emotional and tangible, of all my friends, especially Jude and Ben Platzer and Linda Brown who shared their homes during my stays in Vancouver, and Judy Flagg who provided child-care during my absences from home. My most heartfelt thanks go to my family, Carl and Kate, without whose patient support, and ability to keep the home fires burning during my frequent absences from Victoria, this thesis would not have been completed.  1  INTRODUCTION Over the past two decades there has been a vast number of studies in the medical literature on coronary revascularization techniques. The majority of these studies have been clinically based, comparing the outcomes of revascularization with those from medical treatment. A few studies have documented, and tried to account for, regional variations in the population-based rates of revascularization procedures. A drawback to many of these latter studies is that they have based their conclusions on a single years' data. Therefore, any variation found could be an artifact, peculiar to that year only. Also many studies did not account for the possibility of variation in the incidence of coronary artery disease in the regions studied. A further difficulty for policy makers in the Province of British Columbia (B.C.) is that health services and topography in the regions studied are very different from those in B.C. where revascularization procedures are centralized, the numbers performed controlled by the Ministry of Health and where patients may be isolated without easy access to heart specialists or hospitals. The factors associated with procedural variations in the United States (US) might not be relevant in B.C. Over the past two years in B.C. a number of changes have either occurred or been proposed which will impact on future revascularization rates. In 1990, media attention to the length of waiting lists, prompted the government to start a registry of patients waiting for cardiac surgery, and to make arrangements with hospitals in Seattle, Washington, for some B.C. patients to receive open-heart surgery there. In the spring of 1991, a fourth hospital in B.C. started an open-heart program and the cardiac surgeons and cardiologists in the province set-up a clinical data-base to include all patients receiving coronary artery bypass surgery in  2  the province. At the University of British Columbia, preparations were started for a consensus process to develop indicators for appropriate use of coronary artery bypass surgery (CABS). The time seemed right to document the trends in revascularization procedures in B.C. The purposes of this study are to: i.  describe the Provincial and regional trends in CABS and percutaneous transluminal coronary angioplasty (PTCA) in B.C. between 1979 and 1988.  ii.  determine if there are variations in procedural rates among geographic regions, and  iii. attempt to account for any regional variations that are found. This thesis, which describes the above study, is organized as follows: Chapter 1: reviews coronary artery disease as a problem in Canada and elsewhere, very briefly describes the causative theories and describes the manifestations of the disease, examines the role of risk factors, discusses the diagnosis of the disease and the use of the coronary angiogram and exercise "stress test", discusses the primary and secondary prevention of the disease and touches on the medical treatment. The purpose of Chapter 1 is to provide information that will help the lay reader to understand the later chapters. Epidemiological and medical concepts considered to be useful as background information are included, as footnotes or within the text itself. Chapter 2: Part I describes the history of CABS with respect to trends in utilization and the changes that have occurred in the procedure and in the patient population receiving it. This section also reviews the findings of the clinical studies of CABS as they relate to the effectiveness of the procedure in chronic  3  stable angina, and in other conditions. Costs, complications and mortality are reviewed, both generally and as they relate to the above populations. Part II follows the same format for PTCA and reviews the comparisons of PTCA and CABS. Chapter 3: describes regional variations in CABS and other procedures and examines the factors which may account for such variation. Some of the methodological problems associated with studying regional variation are discussed. Chapter 4: describes the rationale and methodology used in this study, Chapter 5: outlines the results of this study, Chapter 6: discusses the results, policy implications and further studies that need to be done to clarify the issues raised.  4  CHAPTER ONE CORONARY ARTERY DISEASE INTRODUCTION Coronary artery disease (CAD) is the only condition for which coronary revascularization procedures are performed and, therefore, some knowledge about CAD is required in order to place the recent trends in coronary artery bypass surgery (CABS) and percutaneous transluminal coronary angioplasty (PTCA) in context. This chapter will define and describe CAD and will discuss the major findings in the literature in relation to the causes, risk factors, incidence, diagnosis, prevention and medical treatment of the disease. This discussion is not intended to represent an exhaustive review of the extensive literature on this disease but is intended to simply set the stage for the later discussions on CABS, PTCA and on the implications of the findings from this study. THE PROBLEM OF CAD IN CANADA Coronary Artery Disease, also referred to as coronary heart disease (CHD) or ischemic heart disease (IHD), ties with cancer as the leading cause of death in Canada. In 1987 CAD was responsible for 46,000 deaths, one quarter of all deaths, in this country and over half these deaths were from acute myocardial infarction (AMI). In 1987 British Columbia (B.C.) was among the three provinces having the lowest age-standardized mortality rates for CAD, with rates of 162 per 100,000 males and 77 per 100,000 females. Newfoundland, the province with the highest rates had a rate of 225 per 100,000 males and 115 per 100,000 females (Nair et al, 1990). CAD creates a significant amount of morbidity as well as mortality. Together with stroke and other cardiovascular diseases (responsible for 17 percent  5  of all deaths) CAD accounted for 21 percent of all hospital days in Canada in 1985 at a cost, excluding doctors and surgery fees, of over $3 billion (Nicholls et al, 1986).Despite the magnitude of the present problem, CAD mortality has been declining in Canada, and in much of the Western World, since the 1960's. These trends will be described later in this chapter. CAUSE AND EFFECT OF CORONARY ARTERY DISEASE The cause of coronary heart disease is artherosclerosis of the coronary arteries. The right and left coronary arteries, the first two branches of the aorta, supply freshly oxygenated blood to the heart muscle (myocardium). The left coronary artery has two branches from the main stem - the left anterior descending artery and the left circumflex artery. Thus, in effect, there are three coronary arteries supplying the heart muscle, and the terms one- two- or three-vessel disease indicates how many of these vessels are significantly affected by CAD 1 . Left main stem disease, stenosis of the left main stem before the bifurcation, affects the blood supply to both the left anterior descending artery and to the left circumflex. This has a more far-reaching effect than a comparable stenosis elsewhere. Although atherosclerosis most commonly involves the aorta and its major branches, the disease also frequently affects vein grafts such as those used in coronary artery bypass surgery. The build-up of collections of abnormal fats, cells and debris (atherosclerotic plaque) under the endothelial layer of the artery gradually reduces the cross-sectional area of the artery in the affected segments. When the lumen of the artery is reduced by approximately 75 percent, increases in the myocardial demand for oxygen (e.g., in exercise) cannot always be met and myocardial ischemia results. An 80 percent reduction in the lumen may reduce 1 In most studies an artery is considered to be significantly affected by CAD if the an atherosclerotic plaque occludes 70 percent, or more of the lumen. Some studies set this figure at 50 percent occlusion.  6  blood flow to the myocardium even when the body is at rest. The clinical symptoms of CAD, therefore, are those of vascular insufficiency which may result from abrupt closure of the artery due to plaque rupture and thrombosis formation (usually resulting in myocardial infarction) as well as from the more gradual luminal narrowing by the atherosclerotic plaque. (Rolak and Rokey 1990). The mechanisms involved in the development of the atherosclerotic plaque are not clearly understood. Because CAD is associated with hypercholesterolemia, cholesterol is generally believed to be implicated in the development of atheroclerosis. The hypothesis of cholesterol as a causative agent accounts for the increased incidence of CAD seen in societies having high saturated fat consumption and for the higher relative risk of cardiovascular events in individuals with high cholesterol levels, but it does not account for the importance of cardiovascular risk factors such as hypertension and smoking. There are several other theories of athero8._nesis but one of the most popular, though unproven, is the 'response to injury' theory which proposes that atherosclerotic lesions form on injured areas of the epitheliun. As the lesion grows it alters blood flow thus placing the endothelium at a greater risk for injury. This theory would account for the high incidence of atherosclerotic lesions at vessel branch points where shearing forces, which could induce endothelial injury, would be higher. Hypertension as a risk factor could also be mediated by the higher stresses placed on the endothelium (Rolak and Rokey 1990). As indicated above, blood flow limitation due to plaque progression is the most direct mechanism by which atherosclerosis results in symptomatic disease. An increase in the myocardial demand for oxygen above the level that the compromised coronary artery can provide, results in myocardial ischemia and the symptoms of classic angina, i.e., visceral discomfort (located appropriately for cardiac origin and usually sub-sternal) which is precipitated by increased cardiac  7  work and which is relieved promptly by rest. In recent years it has been recognized that mental, as well as physical, stress can provoke myocardial ischemia and angina (Rozanski et al. 1988). It is also recognized that factors other than the mechanical obstruction of the atheromatous plaque contribute to myocardial ischemia in patients with CAD. Angiographic studies of coronary vasomotion have demonstrated that while normal coronary arteries dilate in response to stimuli such as cold or exercise, arteries with atheromatous irregularities paradoxically constrict. The cellular mechanisms which lead to this response are not, however, completely understood (Nabel et al, 1988). Because the symptoms of clinical coronary artery disease result from myocardial ischemia, the terms coronary artery disease and ischemic heart disease are usually used synonymously, although they are not in fact synonymous. The archetypal patient with CAD will have angina resulting from myocardial ischemia. However, myocardial ischemia may be caused by disease other than CAD and it is also possible to have angina, or chest pain resembling it, without having CAD or even myocardial ischemia. People who have 'silent myocardial ischemia' have CAD and myocardial ischemia but no anginal symptoms. In recent years there has been considerable research on this condition as the potential hazards have become apparent (Epstein, Quyyumi and Bonow 1988). Patients with anginal symptoms may also have episodes of silent ischemia. Deanfield et al (1983) showed that in patients with stable angina pectoris up to 70 percent of ischemic episodes are silent. Finally, there are people who are found, on angiography or at autopsy, to have CAD with no previous history of functional disturbances or symptoms. In summary, while angina is an important warning of the probable presence of myocardial ischemia resulting from CAD, not all people with angina have  8  myocardial ischemia and/or CAD, nor does the absence of angina indicate the absence of ischemia and/or CAD. MANIFESTATIONS OF CORONARY ARTERY DISEASE Although angina is the classical symptom of CAD it is, unfortunately, not the most common 'cardiac event' heralding the onset of the disease in men. Rolak and Rokey (1990) state that one percent of previously asymptomatic males aged 30 to 62 will develop clinical manifestations of CAD each year. Of these, 42 percent will present with an acute myocardial infarction, 38 percent with stable angina pectoris, 13 percent with sudden death and 7 percent with unstable angina. Myocardial Infarction: Myocardial infarction (MI), death of a portion of the heart muscle, occurs when a relatively sudden occlusion of a coronary artery, or one of its branches, interrupts the blood supply to the muscle. Such an occlusion often results from thrombus formation on a plaque. Data from the Framingham study (Weiner and Kannel 1987) indicates that, for survivors from an initial MI, one third will have a reinfarction within ten years, for men and women respectively 30 percent and 40 percent will develop angina, 16 percent and 24 percent will suffer stroke, 27 percent and 31 percent will develop cardiac failure and 20 percent and 10 percent will die suddenly. Overall, 60 percent will die within ten years. Kannel (1990) points out that these figure apply to untreated or to "primitively" treated CAD. The survival rate with modern medical treatment is likely to be better than this. Evidence for better survival in more recent times, comes from a community based investigation which was part of a World Health Organization collaborative study. Data from the Perth coronary register on patients who  9  suffered an MI between 1971 and 1979 and who survived the first 28 days, showed one, five and nine year survival rates to be 88, 67 and 52 percent respectively (Martin et al. 1983). The increased use of beta-blockers since that time is likely to have improved survival over these figure as wel1 2 . Chronic Stable Angina: Angina which is predictable in frequency and duration and which can be relieved by nitrates and rest is termed "chronic stable angina". Framingham study data (Weiner and Kannel 1987) showed that 30 percent of men and 40 percent of women died within ten years of developing angina; a mortality rate 1.6-1.9 fold that of the general population of the same age. Again, modern treatment may well improve the survival chances of angina patients. Stable angina has been graded, according to severity of symptoms, by the Canadian Cardiovascular Society as follows: 1.  Class I Angina occurs only with strenuous or prolonged exertion at work or recreation and does not occur with ordinary physical activity.  2.  Class II Angina occurs with walking rapidly on level ground or a grade and with rapidly walking upstairs. Ordinary walking for less than two blocks on the level or climbing one flight of stairs does not cause angina except during the first few hours after awakening, after meals, under emotional stress, in the wind or in cold weather. This implies slight limitation of ordinary activity.  3. Class III Angina occurs when walking less than two blocks on level ground at normal pace, under normal conditions or when climbing  2 By reducing the incidence of life-threatening cardiac arrythmias, beta-blockers probably reduce the incidence of sudden death which generally results from a cardiac arrythmia.  10  one flight of stairs. This implies marked limitation of ordinary physical activity. 4. Class IV Angina occurs with even mild activity, and may occur at rest but must be of brief (less than 15 minutes) duration. This implies inability to carry out even mild physical activity. Unstable Angina:  The term "unstable angina" has been applied to several syndromes, which are taken as clear evidence of important, but reversible, myocardial ischemia. The American College of Cardiology/American Heart Association Task Force on Assessment of Diagnostic and Therapeutic Cardiovascular Procedures in reporting their guidelines and indications for CABS (Kirklin et al, 1991) used the following criteria for defining unstable angina. Patients with severe and persistent angina on presentation to the physician (or hospital) with electrocardiographic evidence of myocardial ischemia and only minor enzyme evidence (available later) of myocardial infarction. Also patients with new onset angina (Canadian Class IV) within two months of presentation or recurring or prolonged (greater than 15 minutes duration) angina within 10 days of presentation even if such angina was not new. They also applied the term unstable angina to those patients who had severe angina within two weeks of a myocardial infarction. There was also a requirement that in all groups there would be ECG evidence of myocardial ischemia during the severe pain but no evidence of more than minimal myocardial necrosis. Braunwald (1989) has devised a classification for unstable angina but it is uncertain whether this is in common use.  11  Atherosclerosis in Non-coronary Vessels: So far this discussion has centered on the effects of atherosclerosis on the coronary vessels. However, it should be recognized that atherosclerosis is not just restricted to the coronary vessels but may occur in other vessels of which the cerebral vessels and those of the lower extremities are the most common. It may be expected that the presence of atherosclerotic disease in the arterial system of one part of the body would herald or be associated with the same disease in another part of the body. The Framingham study (Dawber 1980) found that the rate of claudication, a symptom of peripheral vascular disease, was seven times greater in subjects with angina than in those with no anginal symptoms. Also the risk of myocardial infarction was found to be four times greater in those who had previously had a cerebrovascular accident (stroke) than in those who had not. Dawber concludes that the presence of one manifestation of atherosclerotic disease greatly increases the probability of developing another and that frequently the same risk factors are involved in contributing to the different categories of atherosclerotic disease. RISK FACTORS Factors whose presence is associated with an increased probability that a particular disease will develop later are called "risk factors" for that disease 3 . Such an association does not necessarily imply that the risk factor always causes the disease, that all patients with risk factors will develop the disease or that eliminating, or decreasing, the risk factor will prevent the disease from occurring. The "risk" measured in risk factors is the "relative risk" (the incidence rate among the exposed/the incidence rate among the unexposed) which is derived from 3 The term "risk marker" is often used to refer to risk factors, such as age and sex, which cannot be changed.  12  measurement of the presence of risk factors and disease in groups. Factors which are powerful predictors for a group may be weak predictors for individuals. Stages of Disease:  Understanding the concept of risk factors requires an understanding of the natural history of disease. Most diseases do not arrive out of the blue but are rather a culmination of a long process, interactions between environmental and individual factors, which eventually result in disease. The natural history, the course of the untreated disease over time, is different for each disease but may be modeled as a series of stages (Mausner and Kramer 1985). 1. Stage of Susceptibility. In this stage the groundwork for CAD is laid by the presence of risk factors for the disease. Some of these risk factors, such as sex, race, and family history, are not alterable but others, such as smoking and dietary habits, can be changed. 2. Stage of Presymptomatic disease. In this stage pathologic changes, such as atherosclerotic changes in the coronary arteries, have started to occur although there are no clinical symptoms. 3. Stage of Clinical Disease. This stage starts when the signs and symptoms of the disease become discernible and continues for as long as the patient is clinically ill from the disease. The diverse manifestations and outcomes of CAD mean that patients with the disease may have very different symptoms, prognoses and requirements for treatment. Consequently, patients in this stage are usually placed in homogenous subgroups for the purposes of therapeutic management or study. For clinical study this may be done on the grade of angina experienced, the location of the atherosclerotic block, the number of vessels that are blocked, the functional status of the left ventricle, or on some  13  combination of these factors, all of which have been shown to influence prognosis and the response to treatment. Subgroups in epidemiologic trials are generally categorized according to their risk factors or to symptomatology. 4. Stage of Disability. CAD may give rise to a residual defect which can reduce patients' activity to the state where they are functionally disabled. However, because it is so often fatal, CAD may result in less individual disability and community disruption than other cardiovascular diseases . Risk factors for CAD may be of greater or lesser importance depending on the stage of disease. Some may make the person more susceptible to the development of atheroma while others may cause the atheroma to form (Pearson 1991). Yet others may increase the likelihood of outcomes such as myocardial infarction or sudden death (Dawber 1980). Confounding factors, those associated with a disease only because of their association with a true risk factor for the disease, may confuse the picture.  Primary and Secondary Risk Factors: The results of numerous studies have shown that the major risk factors for developing clinically detectable coronary artery disease are hypercholesterolemia, hypertension, cigarette smoking and diabetes mellitus. These "primary" risk factors are independent of one another in that they make a separate contribution to the risk of developing CAD but they can also be additive, because the presence of more than one further increases the risk (Dawber 1980). Other variables, secondary risk factors, appear to accentuate the action of the primary risk factors, and include obesity, family history, physical inactivity, age, male gender and decreased levels of HDL (Rolak and Rokay 1990).  14  The major findings surrounding primary and secondary risk factors are briefly outlined below.  Hypercholesterolemia: The Framingham Study (Dawber 1980) showed that risk of developing CAD in males appears to be continuous with a gradient of increased risk of the disease with increases in the total cholesterol level For men in their thirties, the relative risk for those with cholesterol levels of 260 milligrams per decilitre (mg/dl) or more, was over four times that for those with levels below 200. The trend was consistent, though not as great, for men in the forties and fifties. The Multiple Risk Factor Intervention Trial (MRFIT) demonstrated that with each 50mg/ dl increase in cholesterol the coronary event risk doubled, implying a linear relationship between the level of cholesterol and risk (Multiple Risk Factor Intervention Trial Research Group 1982). Other studies have suggested a nonlinear relationship with a "threshold", between 200 and 220 mg/dl, at which disease will develop (Pooling Project Research Group 1978). In any event, in males the effect of elevated total cholesterol diminishes with age. In women, the picture is somewhat different. The Framingham study showed a gradient of risk only for certain outcomes in certain age-groups. For example, higher levels of total cholesterol were associated with increased risk of MI in women age 40-49, and with an increased risk of CAD in general for women age 50-59. In women, unlike men, the total cholesterol level tended to increase with age.  Hypertension: The risk of CAD increases progressively as both systolic and diastolic blood pressure increase. (The Pooling Project Research Group 1978, Dawber, 1980). In  15  The Framingham Study men who were in their thirties at the time of entry to the study, had an incidence rate of total CAD six times greater in those with systolic blood pressure over 180 than for those below 120 millimeters of mercury (mm Hg). Also in men, the age at which CAD appeared decreased as blood pressures increased, with an overall difference of 2 years in the age of onset. Hypertension also showed a significant correlation with the clinical manifestations of angina pectoris, myocardial infarction and sudden death (defined as death within one hour of being well). Tobacco Smoking: Tobacco smoking is an independent risk factor for CAD in both males and females but its contribution to the development of CAD is accentuated by the presence of other risk factors (Dawber 1980, Pooling Project Research Group 1978, Doll et al 1980). The highest risk comes from smoking cigarettes but increased risk has been shown for pipe and cigar smokers as well (Pooling Project Research Group 1978). While the risk for smokers increases for all manifestations of CAD the greatest risk is for sudden death and MI. The risk declines with advancing age so that by the age of 65 there is no added risk for smokers, except from sudden death where the risk is still present but diminished (Dawber 1980). Diabetes Mellitus: The Framingham results (Dawber 1980) showed that diabetes contributes significantly to all manifestations of CAD. The risk for male diabetics was more than twice that for non-diabetic men, while diabetic women had a risk five times greater than that of non-diabetics. Although other risk factors (hypertension, obesity and elevated cholesterol) were significantly related to diabetes, these factors  16  did not wholly explain the added risk observed in diabetics. CAD is the primary cause of death in overt diabetics and this correlates best with the age of onset and duration of the diabetes rather than with the severity of the disease (Rolak and Rokay 1990). High Density Lipoproteins:  Rolak and Rokey (1990) cite numerous studies which have shown a negative correlation between high density lipoproteins (HDL) and CAD. These authors report that, because low density lipoproteins (LDL) and HDL have opposite effects on the development of CAD, the ratio of LDL to HDL in an individual can influence the development of disease. Patients with ratios of less than two are at lowest risk while those with ratios above five are at greatest risk of developing CAD. Obesity:  The Framingham Study (Dawber 1980) showed a clear relationship between relative weight (actual weight/median weight for that age-sex group) and the incidence of CAD. This finding has been confirmed in other studies (Lew and Garfinkel 1979) although a direct causative independent role for obesity in the development of CAD has not been established. Obesity is associated with several other risk factors for CAD, notably hypertension and abnormal lipid profiles, and these may contribute the majority of the increased risks seen in obese people (Rolak and Rokey 1990). More recent research suggests that the distribution of body fat may play an independent role in the development of CAD (Wingard 1990).  17  Family History: Family history of heart disease below the age of 65 appears to predispose to CAD with a two to four-fold increased risk of MI for the immediate relatives of MI victims. Because many other risk factors (including hypertension, hyperchol ' erolemia, obesity, and physical inactivity) are found in families with CAD, -  an independent role for genetic factors cannot be established (Rolak and Rokey 1990). Physical Activity: The evidence that physical activity plays a direct independent role in the prevention of CAD is difficult to establish; in part because of difficulties in standardizing physical activity and in part because of its effect on other cardiac risk factors.. Both Gau (1985) and Rolak and Rokey (1990) cite many studies which show an inverse relationship between physical activity and the incidence of CAD, MI and mortality from heart disease. The Harvard Alumni Study (Paffenberger and Hyde 1984) showed that the benefit in reduced CAD incidence and in reduced mortality obtained from exercise in excess of 2000Kcal per week, was independent of smoking, obesity, hypertension and family history of CAD. Of the alumni who developed CAD during the 12 to 15 year follow-up, those who exercised in excess of 2000Kcal per week had 71 percent of the coronary artery disease mortality of the inactive alumni. Age and Sex: The incidence of CAD is greatest in males, particularly in the 35-44 year age group where the risk is three to ten times that for females. With increasing age, particularly past 60, the risk gradient between the genders diminishes. Postmenopausal women have three times the incidence of CAD as their age  18  counterparts who are still menstruating, but the role of hormones in protecting women is still unknown. The lower total cholesterol, lower LDL's and higher HDL's found in women may contribute to their lower risk (Rolak and Rokey 1990). In conclusion, the fact that risk factors are poor predictors of future disease in individuals should not detract from their importance in the development and progression of CAD. The role of risk factors as "triggers" in the manifestation of cardiac events such as MI or sudden death does not appear to be understood and should be an avenue for further study. TRENDS IN CORONARY ARTERY DISEASE MORTALITY Although deaths with the characteristics of acute coronary artery disease have been described as early as 3000 B.C. (MacKinnon 1987) it was not until the beginning of this century that they became more than a rare event. Numerous studies in North America and Europe have shown that CAD mortality rose dramatically from the early 1900's and then started a decline which is still continuing (Nair et al 1990, Mackenbach et al, 1989, Thom and Maurer 1988, Slater et al 1985, Gillum, Folsom and Blackburn 1984, Feinleib 1984,). These studies show that the onset of the decline, between the mid-sixties and 1972, varied between countries and between geographic regions within countries. Several studies showed a gender variation with the slowing or decline in acute coronary deaths beginning in women up to a decade before it was seen in males (Gillum, Folsum and Blackburn 1984, Nicholls, Jung and Davies,1981, Slater et al 1985). There are several problems associated with using cause-specific mortality rates 4 to estimate the rise or fall in incidence of a disease. One is that a decline or 4 The cause-specific mortality rate is calculated by dividing the number of people dying of a specific  cause in a certain time period by the number of people at risk. This rate is influenced by the age distribution of the population, the proportion of people dying of various causes and their average age at death by cause.  19  rise in other causes of death, such as tuberculosis or influenza, may cause an increase or decrease in the mortality rate of the disease of interest without any change occurring in the incidences or prevalence 6 of that disease. Cause-specific death rates are also vulnerable to conventions and fads in the classification of causes of death and to classification problems such as the distinction between primary versus underlying cause of death and periodic nomenclature revisions in the International Classification of Disease (ICD). These problems have led to concerns that the rise in coronary artery disease mortality in the first half of this century was just "a paper epidemic" (Slater et al 1985). However, the use of analytic techniques which account for the problems outlined above (Slater et al 1985, Thom and Maurer 1988) have shown that the epidemic was real and that the decline in mortality rates is also real. The reason for the decline, however, is not clear. It may be that the incidence of CAD has not changed but that the incidence of the more lethal manifestations, such as MI or sudden death, have declined. Some authors have reported a decline in the case-fatality rate from MI (Kovar et al 1988, Gillum, Fossum and Blackburn 1984) but whether this is due to better medical care or to a "detection bias" 7 is not known. If, indeed, the CAD fatality rate is declining while the incidence of the disease remains unchanged, the result will be an increasing The incidence rate of a disease measures the probability that healthy people will develop that disease during a specified period of time. That is, it is the number of new cases of the disease within the given time period divided by the number of people at risk for that disease. 5  The prevalence rate of a disease measures the number of people in a population who have the disease at a given time. It is calculated by dividing the number of existing cases of the disease at a given time by the population at that time. 6  7 A detection bias may occur when improved diagnostic procedures, or the more frequent use of existing techniques, leads to the identification of cases which would previously not have been identified. Up to 40 percent of MI's are "silent" and are detected only by repeat ECG examinations of asymptomatic individuals. More frequent examinations could result in a gradual shift of these undetected MI's to clinical diagnosis and treatment (Kuller, 1988).  20  prevalence of the disease. Evidence for increasing prevalence of CAD in three regions of the US is shown by Feinleib et al (1988) Studies and reviews which have attempted to account for the declining CAD mortality rate have reported a declining incidence of out-of-hospital CHD death, a decline in the age-adjusted incidence of hospital admissions for MI and a decline in the MI case fatality rate. (Thom and Maurer 1988, Kovar et al 1988, Feinleib et al 1988, Goldberg et al 1986). These authors suggest that the declining mortality rate may be due to changes in medical care. Slater et al (1985) argue that dietary-changes and medical care improvements, which would affect men and women equally, cannot account for their results, which show different patterns in the rise and fall of acute CAD incidence for males and females. Instead they claim that the changes in the male-female differential for acute CAD are compatible with changes in smoking behaviour which has also been gender-related. The majority of the authors cited above called for better national data on CHD incidence, severity, case fatality, suddenness and place of death. Goldman and Cook (1984), basing their estimate on reasonable assumptions drawn from the literature, conclude that over 50 percent of the decline in CAD mortality is due to reductions in serum cholesterol levels and cigarette smoking. A further 40 percent of the decline is due to specific medical interventions including coronary care units, medical treatment (especially beta-blockade therapy) of clinical CAD, and the treatment of hypertension. Killip (1988) estimates that coronary artery bypass surgery (CABS) accounts for only one to two percent of the decline in CAD mortality in the US. In summary, a true epidemic of acute CAD began in the first half of this century and reached a peak in mid-century. Mortality from the disease is now on the decline for reasons unknown but thought to be related to changes in medical care, changes in lifestyle and treatment of hypertension. It is still not known if the  21  incidence of CAD has declined; if not the decline in mortality will lead to an increased prevalence of the disease. Investigation into CAD incidence and mortality in the US. is hampered by the absence of valid and reliable national databases. It is likely that this problem also exists in Canada. DIAGNOSIS IN CAD Coronary Angiogram: The gold standard 8 for the diagnosis of coronary atherosclerosis is the pathologist's examination. Since this is not possible in most instances, a more readily available technique is required as a proxy. The coronary angiogram is usually accepted as such a proxy, since it has good, but not perfect, correlations with anatomic findings (Hlatky et al 1989). Comparisons of the findings by coronary angiography with those in subsequent post-mortem examinations have shown that significant under-estimations of atherosclerosis can occur. Also, significant intra-observer variability may be seen with conventional coronary angiography. Quantitative techniques (e.g. digital subtraction angiography and the use of computer automated techniques) to reduce this variability have been developed, although their application is not widespread (Fisch et al 1987).  8 A diagnostic test which is generally accepted as the most accurate test for diagnosis of the disease  and to which all other diagnostic tests are compared, is termed the "gold standard" for that disease. Hlatky (1989) distinguishes between two different comparisons that can be performed using a gold standard. The first occurs when two tests measuring the same phenonomen are compared with one another. e.g the pathologists examination at autopsy and coronary angiogram are both measuring anatomic coronary disease. In this case a specific hierarchy can be established in which one test, the gold standard, is seen as "more correct" than the other and will overrule the other if there is disagreement between them. In the second type of comparison an explicit hierarchy cannot be established because the tests are measuring inherently different phenomena. Thus, in a patient with atypical syptoms has a normal exercise test (functional test for ischemia) but an abnormal coronary angiogram (anatomic test) the clinician cannot determine whether the exercise test is a false negative or a true negative for myocardial ischemia.  22  The problem with the use of coronary arteriogram as a diagnostic test is that it is invasive, expensive and has several potentially life-threatening complications, including death, MI, cerebro-vascular accident (CVA), ventricular arrythmias, local vascular complications and contrast agent reactions. Stewart et al (1990) found a 0.0024 rate for major complications (requiring immediate CABS) in a population of 5781 low risk patients, whose risk was determined retrospectively 9 . The overall mortality rate is about 0.2 percent but patients with significant left main coronary stenosis, severe three-vessel disease, multi-vessel disease with left ventricular dysfunction, advanced age or unstable angina are at significantly higher risk than patients without these complications (Chassin et al 1986, Fisch et al 1987). However, it is patients with these conditions (with the exception of advanced age) who are most likely to require angiography. The primary purpose of coronary angiography is to define the anatomy of the coronary arteries when such definition is needed for patient management. The procedure may also be used to assess the results of therapy and to help formulate prognosis in patients with CAD, although prognosis has several other determinants which are not discernible by angiography. Guidelines for the use of angiography have been developed by the American College of Cardiology in conjunction with the American Heart Association (Fisch et al 1987). As ii.aicated above, coronary angiogram is not necessarily required in all cases of CAD. The diagnosis of angina may be reliably determined by a physical examination and history taken by an experienced clinician (Hlatky 1989). Other initial examinations for patients with chest pain suggestive of angina would include resting electrocardiogram, blood lipids and routine blood chemistry 9 Given an approximate annual figure of 400,000 coronary angiograms in the U.S. (Chassin et al 1986), there would be at least 960 patients receiving emergency CABS as a result of complications from angiography. The actual number of patients experiencing major complications would likely be higher since this figure was calculated from the rate for low risk patients.  23  determination. For patients believed to have chronic stable angina who continue to be minimally or moderately symptomatic after initial treatment to control risk factors and the use of nitrates, beta-blockers or calcium antagonists, an exercise stress test should be performed (Silverman and Grossman 1984).  Exercise Stress Test: The exercise electrocardiogram ("stress test") is a functional test of  myocardial ischemia 10 . The main purpose of the test is to identify those patients who may at high risk for future cardiac events (and exclude those who are not) so that patients may be appropriately referred for angiography. When compared to coronary angiography findings, the sensitivity" of the stress test is between 60 and 70 percent and the specificity 12 is 90 percent (Rolak and Rokey 1990). 13 However, patient traits such as age, sex, presence of typical angina and maximal exercise heart rate have been shown to have independent effects on the sensitivity of the stress test (Chassin et al 1986). Restriction of the test to those patients whose history and initial examination indicate a high pretest likelihood of untoward cardiac events, would likely reduce false positives and may minimize the number 10 There is no universally accepted "gold standard" for measuring myocardial ischemia. Instead functional tests, such as the exercise electrocardiogram, thallium-201 scintigam, radionucllide angiogram or positron emission tomogram, are referenced to the coronary angiogram, an anatomic gold standard. Without a single gold standard for myocardial ischemia, physicians have to rely on the concordance of two or more functional tests to determine the presence or absence of ischemia (Hlatky, 1989). 11 Sensitivity refers to the ability of a test to correctly identify those who have the disease in question. When sensitivity is high, the number of 'false-negatives' (i.e. those who have the disease but who have a negative test result) is low.  12 Specificity refers to the ability of a test to correctly identify those who do not have the disease in question. When specificity is high the number of "false-positives" (i.e. those who do not have the disease but who have a positive test result) is low. 13 The predictive value of a test is generally considered to be more important than the sensitivity or specificity but was not discussed in the literature reviewed here.  24  of patients who receive unnecessary coronary angiograms. Use of the stress test for screening apparently healthy individuals is not recommended (Froelicher et al 1988). Patients with a negative stress test should continue to receive therapy but require no further evaluation at that time. Patients with a positive response but with no indications of left main or three-vessel disease, likely also require no further evaluation though a more aggressive approach may be favoured in the younger more active patients. Patients whose stress test shows evidence of left main or three-vessel disease should generally undergo coronary angiography (Silverman and Grossman 1984). Patients who may be referred directly for angiography without a prior stress test may include those with chronic stable angina resistant to maximal medical therapy (Silverman and Grossman 1984), unstable angina when pain does not respond to medical management (Chassin et al 1986), candidates for intracoronary thrombolysis when less than six hours have elapsed since the onset of chest pain, convalescent MI patients in whom angina develops at rest or on minimal exertion, and candidates for valvular surgery who are males over 35 years, postmenopausal females or those in whom CAD is suspected (Fisch et al 1987). In summary, although the coronary angiogram is accepted as the gold standard for the diagnosis of CAD and its prognostic indicators, the test is not appropriate for all patients with suspected CAD because of its invasiveness, potential for mortality or serious complications and its expense. Many patients can be diagnosed on the basis of the history, resting ECG and simple laboratory tests. Those in whom the diagnosis is equivocal or whose symptoms are not controlled with medical therapy should be further evaluated by means of a stress test to identify those at high risk. These high risk patients are those who will require definition of coronary anatomy by coronary angiogram.  25  PREVENTION AND TREATMENT OF HEART DISEASE The increasing prevalence of CAD over the years with the consequent high costs to the individual and the health care system have led to a plethora of research into prevention and treatment of the disease. While the concepts of prevention and treatment would appear to be separate, in the case of CAD, and other diseases, there is not always a clear distinction between them. Much of the 'treatment' of CAD is aimed at prevention of progression of the disease and of outcomes such as MI or sudden death. Moreover, many of the same methods are used in both prevention and treatment  Prevention: As indicated above, the term prevention may mean more than the inhibition of the development of a disease before it occurs but may include measures that interrupt or slow the progression of the disease. For this reason, prevention may be divided into levels which have different expected outcomes and which may be used in different stages of the disease (Mausner and Bahn 1985). Primary prevention is used in the stage of susceptibility and is aimed at preventing the development of atherosclerosis and CAD or, at least, delaying its onset. There are two major approaches which may be used here. The first is the population approach; the universal adoption of health education measures to encourage and motivate individuals to modify their lifestyle in the hope that such changes will decrease the incidence of CAD. Oliver (1987) states that the problem with this approach is that there is no conclusive evidence that universal lifestyle changes would affect the incidence of CAD. However, because lifestyle changes have been shown to affect the progression of atheromatous lesions, both by causing regression of the lesions and by slowing progression (Ornish et al, 1990), it  26  appears likely that the population approach will delay the onset of CAD in many individuals and may prevent some from developing it. The second approach, also used in the susceptibility and pre-clinical stages, is that of screening for risk factors combined with measures to reduce risk factors where present. Such screening measures include estimations of total cholesterol from blood samples and blood pressure measurement but could also include weight, dietary and smoking history and family history of heart disease. Several problems are associated with this approach, the first being that the specificity 14 of risk factors is low. This raises certain ethical questions abo ,t how to deal with those whose results show borderline risk of CAD, especially because treatment of hypertension and of hypercholesterolemia may requirethe use of drugs which themselves carry risks from side effects. These drugs are also expensive and may, therefore, not be acceptable to asymptomatic people. Another problem with screening is that reduction of hypertension in high risk individuals has not been shown to significantly reduce their incidence of either heart disease or mortality from cardiac events (Mitchell, 1987), although this may be due to the negative effects of the commonly used hypertensive drugs (betablockers and diuretics) on other risk factors (Kaplan 1991). There is little use in screening if there is no effective intervention for those found to be at risk. However, the most recent report on mortality for hypertensive participants in MRFIT (Multiple Risk Factor Intervention Trial Research Group 1990) has shown that CAD mortality for the treatment group, after 10.5 years of follow-up, was 15 percent lower than for the control group. This reversal of unfavourable trends for  14 As a "test" of who will develop CAD, risk factors are poor predictors in individuals. For example, two-thirds of a group of healthy men age 40-55 years who are at the highest risk (above the 80th percentile) as a result of hypercholesterolemia and hypertension can be estimated to remain well over the next 25 years (Oliver, 1987). While the relative risk for these individuals is high their absolute risk is quite low.  27  the experimental group during the trial is attributed, in part, to a mid-trial change in the diuretic treatment for the experimental group. The use of drugs to reduce blood-lipids does appear to be effective according to a review by Blankenhorn (1989). The results of two randomized double-blind placebo-controlled studies (The Lipid Research Clinics Coronary Primary Prevention Trial and the Helsinki Heart Study) showed that reducing blood lipids resulted in reduced mortality and morbidity from heart disease. Turnstall-Pedoe (1987) reports that the cost per coronary death prevented in the Lipid Research Clinics Trial was £1,000,000 in drug costs alone. The four randomized controlled trials, reviewed by Blankenhorn, that used diet alone to reduce serum cholesterol in post-MI patients produced conflicting results. The trials were small, however, and the trial which achieved the largest reduction in cholesterol also found a significant reduction (33 percent) in CAD events over 11 years of follow-up. It should be noted that these latter trials were conducted on people who had already shown manifestations of CAD. It is possible that similar dietary measures would have an even greater effect on those in the pre-clinical or susceptible stages of the disease. The final problem with screening lies in the cost of the screening itself. While screening for hypertension can be done relatively inexpensively by General Practitioners (GP's) on an opportunistic basis, screening for hypercholesterolemia is costly and is likely to be prohibitive if the whole adult population is screened. Other alternatives are to screen for cholesterol only in those people with other risk factors or to screen family members of people with known CAD.  Secondary Prevention is applied in the early clinical stages of disease and refers to the early detection and prompt treatment of disease. The goal is to halt or slow the progression of the disease, to prevent complications and to limit disability. In the case of coronary artery disease the term "secondary prevention" is  28  also used to describe the strategy aimed at reduction of recurrence of MI (although this is actually tertiary prevention since the disease has already occurred and left residual damage). In this latter case, prevention includes reduction of risk factors, treatment by medication or surgery and may include attempts at psychological change (Mathes 1985). In secondary prevention, screening for early ischemia may result in a better pay-off than risk-factor screening. Rose (1987), reviewing data from the Whitehall Study, concluded that minor disease is a better predictor of major disease in individuals than are risk factors. This conclusion is not surprising since when minor disease is present the disease process has already started, but risk factors may be present without disease. Again, the problems associated with screening, as outlined above, apply. A recent randomized controlled US. study (Ornish 1990) has shown that angiographically determined regression in atherosclerotic lesions can be achieved by means of a rigorous life-style change program involving diet, exercise and the use of relaxation techniques. Eighty-two percent of the experimental group achieved regressive changes in only one year while 53 percent of the control group had progressive changes. The authors state that the purpose of their study was to determine what could be done, rather than what would be practicable in a larger population of patients. While it seems likely that the regression of atheroma would reduce the incidence of CAD, this has yet to be proven. Tertiary Prevention consists of limitation of disability and rehabilitation in those cases where disability has already occurred. Examples of tertiary prevention in CAD are rehabilitation following MI, treatment of angina, and treatment to reduce risk factors in patients who already show manifestations of CAD.  29  Treatment: Treatment of CAD may be divided into acute and chronic management of the disease. Acute management refers to treatment during and immediately after a myocardial infarct. Over the past few years the focus in acute management of MI has changed from treating the mechanical and electrical complications to salvaging jeopardized myocardium, thus decreasing infarct size. This may be achieved by increasing myocardial oxygen supply, utilizing pharmacologic or mechanical revascularization within four to six hours of the onset of symptoms (Rolak and Rokey 1990).  Chronic medical management of patients who have experienced symptoms of CAD, involves risk factor control (including dietary changes, cessation of smoking and exercise) and may involve the use of drugs to prevent platelet aggregation 15 , myocardial arrhythmias 16 and angina 17 . Patients in certain clinical sub-groups, or those whose angina is not adequately controlled by medical treatment, may benefit from revascularization procedures such as coronary artery  15 Platelet adhesion and aggregation are the precursors to the development of coronary thrombi which may result in MI. The use of asprin as an antiplatelet agent has been shown to significantly reduce the incidence rate of MI in patients with unstable angina (Theroux et al, 1988) and in elderly patients with a previous MI (Sixty-Plus Reinfarction Study Research Group, 1980). It is not reccommended that asprin be used in the primary prevention of death or MI because of the risk of heamoorhagic stroke. Studies of prophylactic asprin use in healthy people have been inconclusive (The Steering Committee of the Physicians Health Study Research Group, 1988, Peto et al, 1988). Rolak and Rokey suggest that use of asprin in patients with stable angina is probably justified to protect against both MI and unstable angina. 16 Hjalmarson (1985) reports that several randomized double-blind trials have demonstrated the influence of certain beta-blockers on the mortality and reinfarction rate of post-MI patients. He attributes these outcomes, in part, to the anti-arrythmic properties of beta blockers. Calcium antagonists also have anti-arrythmic effects and are increasingly being used in secondary prevention in CAD. 17 Coronary vasodilating drugs (usually nitrates or nitrites) are used to prevent or relieve angina pectoris. A retrospective study by Bussman (1985) has indicated that the regular use of nitrates in CAD patients may decrease mortality.  30  bypass surgery (CABS) or percutaneous transluminal coronary angioplasty (PTCA). The use of these procedures will be addressed in the next chapter. In summary, prevention of CAD through screening for, and controlling, risk factors in the general population, though intuitively appealing, presents ethical questions about the use of potentially harmful and possibly ineffective treatments in asymptomatic patients, and about screening costs. On the positive side, there is evidence that control of risk factors by means of diet or drugs is efficacious in reducing CAD mortality and that rigorous lifestyle change can produce a reversal of atheroma. In the absence of a "cure" for CAD, treatment is aimed at halting the progression of atherosclerosis through risk factor control and at prevention of "cardiac events" such as MI and sudden death. CONCLUSIONS Coronary artery disease, a chronic, progressive, "lifestyle" disease with a relatively high mortality rate, is a major problem in Canada in both human and economic terms. The declining mortality, unless accompanied by an equal or greater decline in incidence, is likely to result in an increased prevalence of CAD, having an even greater economic impact on the health services. Research into the incidence of CAD in various population sub-groups, has identified four primary risk factors which have both an independent and an additive effect on the development and progression of the disease. Attempts to reduce CAD incidence and/or mortality by the control of these risk factors, through drugs or lifestyle change, have had variable results. Overall, it appears that it is possible to reduce risk factors in individuals, and that these changes can lead to reductions in the incidence of MI and CAD mortality among those treated. However, programs to achieve these ends are expensive and may not always be effective. Furthermore, the risks from the drugs to reduce hypertension or  31  hyperlipidemia may be greater than the risks from CAD in asymptomatic individuals. Treatment of CAD is aimed at the general reduction of risk factors and at the prevention of disease progression and cardiac "events" such as MI and sudden death. It is interesting that research on risk factors has not focused to any extent on factors which may trigger cardiac events or those which may protect high risk people from such events. In a society such as ours where there is a high prevalence of the classic risk factors it is perhaps surprising that more people do not show manifestations of CAD; the reasons why they do not should be investigated.  32  REFERENCES Blankenhorn, D.H., 1989. Prevention or reversal of atherosclerosis: review of current evidence. Am T Cardiol 63 (May):38H-41H. Braunwald, E., 1989. Unstable angina: a classification. Circulation 80 (August):410-414. Bussman, W.D., 1985. The influence of nitrates on prognosis in patients with coronary heart disease. Chapter 14 in Secondary Prevention in Coronary Heart Disease and Myocardial Infarction. ed. P. Mathes, Boston: Martinus Nijhoff Publishers. Chassin, M.R. et al., 1986. Indications for Selected Medical and Surgical Procedures - A Literature Review and Ratings of Appropriateness: Coronary Angiography. Santa Monica: Rand Corporation. R-3204 /1CWF/HF/HCFA/PMT/RWJ. Dawber, T.R. 1980. The Framingham Study: The Epidemiology of Atherosclerotic Disease. Cambridge, Mass.: Harvard University Press. Deanfield, J.E. et al., 1983. Myocardial ischemia during daily life in patients with stable angina: its relation to symptoms and heart rate changes. Lancet 2:753758 quoted in Stephen Epstein, Arshed Quyyumi and Robert Bonow, 1988. Myocardial ischemia - silent or symptomatic. N Engl T Med 318 (April):10381043. Doll R., et al. 1980. Mortality in relation to smoking: 22 years of observation on female British doctors. Br Med T 280:967. Epstein, S.E., A.A. Quyyumi, and R.O. Bonow, 1988. Myocardial ischemia - silent or symptomatic. N Engl I Med 318 (April):1038-1043. Feinleib, M., 1984.  Changes in cardiovascular epidemiology since 1950. Bulletin of the New York Academy of Medicine 60 (June):449-464.  Feinleib, M. et al. Regional Variations in Coronary Heart Disease Mortality and Morbidity. In Trends in Coronary Heart Disease Mortality: The Influence of Medical Care ed. Millicent W. Higgins, and Russell V. Luepker, 31-43. New York: 1988. Froelicher, V.F. et al., 1988. The prognostic value of the exercise test. Disease-aMonth XXXIV(11):727. Gau, G.T., 1985. The role of physical exercise in secondary prevention. evidence to date. In Secondary Prevention in Coronary Heart Disease and Myocardial Infarction. ed. Peter Mathes, Boston: Martinus Nijhoff Publishers: 143-150 Gillum, R.F., A.R. Folsom, and H. Blackburn, 1984. Decline in coronary heart disease mortality. Old questions and new facts. The American Journal of Medicine 76:1055-1065.  33  Goldberg, R.J. et al., 1986. Recent changes in attack and survival rates of acute myocardial infarction. The Worcester heart attack study. TAMA 255 (May):2774-2779. Goldman, L., and E.F. Cook, 1984. The decline in ischemic heart disease mortality rates: An analysis of the comparative effects of medical interventions and changes in lifestyle. Annals of Internal Medicine 101 (December):825-836. Hjalmarson A., 1985. Beta-blocking agents in secondary prevention. Chapter 9 of Secondary Prevention in Coronary Heart Disease and Myocardial Infarction. ed. P. Mathes, Boston: Martinus Nijhoff Publishers. Hlatky, M. A., 1989. Angina, myocardial ischemia and coronary disease: gold standards, operational definitions and correlations. 1 Clin Epidemiol 42: 381-384. -  Killip, T., 1988. Has coronary artery bypass surgery influenced mortality from cardiovascular disease in the United States? Chapter 27 of Trends in Coronary Heart Disease Mortality: The Influence of Medical Care, ed. M.W. Higgins, and R.V. Luepker, New York: Oxford University Press: Kovar, M. G. et al. Trends in the Availability and use of Medical Care for Coronary Heart Disease and Related Diseases. In Trends in Coronary Heart Disease Mortality: The Influence of Medical Care ed. M.W. Higgins, and R.V. Luepker, 16-30. New York:1988. Kuller, L.H. Issues in measuring coronary heart disease mortality and morbidity. In Trends in Coronary Heart Disease Mortality: The Influence of Medical Care ed. M.W. Higgins, and R.V. Luepker, 44-53. New York:1988. Lew, E.A. and L. Garfinkel, 1979. Variations in mortality by weight among 750,000 men and women. T Chron Dis 32:563, quoted in L. A. Rolak and R. Rokey, 96. Coronary and Cerebral Vascular Disease: A Practical Guide. New York: Futura Publishing Company, Inc. Mackenbach, J.P. et al., 1989. Geographic variation in the onset and decline of male ischemic heart disease in the Netherlands. ATPH 79 (December):1621-1627. MacKinnon, A.U., 1987. The origin of the modern epidemic of coronary artery disease in England. Journal of the Royal College of General Practitioners 37 (April):174-176. Martin, C.A. et al., 1983. Long-term prognosis after recovery from myocardial infarction: a nine year follow-up of the Perth Coronary Register. Circulation 65 (November):961-969.  34  Mausner, J.S. and S. Kramer, 1985. Epidemiology - An Introductory Text. Philadelphia: W.B. Saunders Company. Mitchell, J.R.A. 1987. What effect will screening have on CHD incidence. Chapter 2 (Session III) of Screening for Risk of Coronary Heart Disease: Proceedings of a Workshop on Strategies for Screening for Risk of Coronary Heart Disease, ed. M. Oliver, M. Ashley-Miller and D. Wood, 67-75. Chichester: John Wiley and Sons Ltd. Multiple Risk Factor Intervention Trial Research Group, 1982. Multiple-risk factor intervention trial: Risk factor changes in mortality results. TAMA 248:1465 Multiple Risk Factor Intervention Trial Research Group, 1990. Mortality after 10 years for hypertensive participants in the multiple risk factor intervention trial. Circulation 82 (November):1616-1628. Nabel E.G. et al. 1987. Asymptomatic ischemia in patients with coronary artery disease. TAMA 257:1923-1928. Nabel E.G., P. Ganz et al. 1988. Dilatation of normal and constriction of atherosclerotic arteries caused by the cold pressor test. Circulation 77:43-52. Nair C. et al., 1990. Cardiovascular disease in Canada. Health Reports 1(1):1-22. Nicholls, E.S., C. Jung, and J.W. Davies. 1981. Cardiovascular disease mortality in Canada. CMA Journal 125 (November): 981-992. Oliver, M.F., 1987. Problems and limitations. Chapter 1 of Session I Screening for Risk of Coronary Heart Disease: Proceedings of a Workshop on Strategies for Screening for Risk of Coronary Heart Disease, ed. M. Oliver, M. AshleyMiller and D. Wood. 3-10. Chichester: John Wiley and Sons Ltd. Ornish, Dean et al. 1990. Can lifestyle reverse coronary heart disease? The life style heart trial. Lancet 336 (July):129-133. Paffenberger, R.S. and R.T. Hyde. 1984. Exercise in the prevention of coronary disease. Preventive Medicine 13:3-22. Paffenberger R.S., W.E. Hale , R.J. Brand, and R.T. Hyde. 1977. Work energy level, personal characteristics and fatal heart attack: A birth-cohort effect. American Journal of Epidemiology 105:200-213, Pearson, Thomas A., 1991. What is a risk factor and what is not - theoretical and practical implications. American Journal of Medicine 90(suppl 2A) (February):2A-50S-2A-52S. Peto R. et al., 1988. Randomized trial of prophylactic aspirin in British male doctors. Br Med T 296:313, quoted in Loren A. Rolak and Roxann Rokey,  35  Coronary and Cerebral Vascular Disease: A Practical Guide. New York: Futura Publishing Company, Inc.: 291. Pooling Project Research Group. 1978. Relationship of blood pressure, serum cholesterol, smoking habit, relative weight, and ECG abnormalities to incidence of major coronary events. Final Report of the Pooling project. L Chronic Dis 31:201. Report of the Sixty Plus Reinfarction Study Research Group, 1980. A double blind trial to asses the long-term oral anticoagulant therapy in elderly patients after myocardial infarction. Lancet 2:989, quoted in L.A. Rolak and R. Rokey, Coronary and Cerebral Vascular Disease: A Practical Guide. New York: Futura Publishing Company, Inc.: 293. Rolak, L.A. and R. Rokey. 1990. Coronary and Cerebral Vascular Disease: A Practical Guide. New York: Futura Publishing Company, Inc. Rose, G.A., 1987. CHD risk factors as a basis for screening. Chapter 2 (Session I) of Screening for Risk of Coronary Heart Disease: Proceedings of a Workshop on Strategies for Screening for Risk of Coronary Heart Disease, ed. M. Oliver, M. Ashley-Miller and D. Wood, 11-16. Chichester: John Wiley and Sons Ltd. Ross, John Jr. et al., 1987. Guidelines for coronary angiography. A report of the American College of Cardiology/American Heart Association task force on the assessment of diagnostic and therapeutic cardiovascular procedures (subcommittee on coronary angiography). JACC 10 (October):935-950. Rozanski, A, et al. 1988. Mental stress and the induction of silent myocardial ischemia in patients with coronary artery disease. N Engl T Med 318:10051012 Silverman, K.J. and Grossman, W., 1984. Angina pectoris: natural history and strategies for evaluation and management. N Engl T Med 310 (June):17121717. Slater, C. H. et al., 1985. Ischemic heart disease: footprints through the data. Am T Clin Nutr 42 (August):329-341. Stewart, J.T et al., 1990. Major complications of coronary arteriography: the place of cardiac surgery. Br Heart T 63:74-77. The Steering Committee of the Physician's Health Study Research Group, 1988. Preliminary report: Findings from the aspirin component of the ongoing physician's health study. N Engl T Med 318:262, quoted in Loren A. Rolak and Roxann Rokey, Coronary and Cerebral Vascular Disease: A Practical Guide. New York: Futura Publishing Company, Inc.: 291.  36  Theroux, P. et al., 1988. Aspirin, heparin or both to treat unstable angina. N Engi Med 319:1105, quoted in Loren A. Rolak and Roxann Rokey, Coronary and Cerebral Vascular Disease: A Practical Guide. New York: Futura Publishing Company, Inc.: 292. Thom, T.J. and J. Maurer, 1988. Time trends for coronary heart disease mortality and morbidity. In Trends in Coronary Heart Disease Mortality: The Influence of Medical Care, ed. M.W. Higgins, and R.V. Luepker, 7-15. New York: Oxford University Press. Turnstall-Pedoe, H.D., 1987. Yield versus effort. Chapter 1 (Session III) of Screening for Risk of Coronary Heart Disease: Proceedings of a Workshop on Strategies for Screening for Risk of Coronary Heart Disease, ed. M. Oliver, M. Ashley-Miller and D. Wood. 53-66. Chichester: John Wiley and Sons Ltd. Weiner, D.A. and Kannel W.B., 1987. A comparison of surgical and medical therapy for coronary artery disease. CUR & R 37-46 quoted in Kannel, W.B., 1990. Coronary artery surgery revisited: Limitation of the intent-to-treat principle. Circulation 82(5):1859-1862. Wingard, Deborah L., 1990. Sex differences and coronary heart disease. A case of comparing apples and pears? Circulation 80 (May):1710-1712.  37  CHAPTER TWO CORONARY REVASCULARIZATION TECHNIQUES PART I  CORONARY ARTERY BYPASS SURGERY As indicated in Chapter One, the treatment of coronary artery disease is generally aimed at the prevention of the serious sequelae of the disease, e.g., MI or sudden death, as well as at control or reduction of angina. Although these aims may often be attained through risk factor reduction and the use of medication, increasingly surgical revascularization techniques are being used. These procedures revascularize the heart muscle by either bypassing or removing the stenosis in the coronary artery. Part I of this chapter will discuss the first of these surgical techniques coronary artery bypass surgery (CABS). Since its introduction in 1968 this operation has become one of the most investigated surgical procedures in the medical literature. Twelve randomized controlled trials and numerous prospective and retrospective studies compare short and long-term outcomes of CABS versus medical treatment for a variety of patient conditions and ages. There are a multitude of other studies on almost all aspects of the procedure. Following a brief history of the growth and utilization of CABS and the changes in the procedure over the years, this chapter will present the major findings from the literature on the effectiveness in various clinical conditions and patient characteristics, the risks and monetary costs of CABS.  38  HISTORY Growth and Utilization: Although the earliest CABS procedure was reported (in 1974) as being performed in 1962, it was not until 1968 that the first clinical trials began at four centres in the U.S. The initial reports from these trials appeared in the literature in 1969 (Miller 1977) and by 1970, when 3000 procedures were performed in the U.S., bypass grafting was by far the most widely practiced type of direct coronary artery surgery (Preston 1989). In describing the early growth of CABS, Miller (1977) states: "Such a rapid and large-scale application of a new operation is without parallel in the history of surgery". The rapid growth in the U.S has continued, rising from 57,000 procedures (26 per 100,000 population) in 1975 to 284,000 procedures (122 per 100,000 population) in 1986 (Preston 1989) In Canada the trend is also to increased growth of CABS but at a slower pace than in the U.S. Peters et al (1990) report that between 1981 and 1986 the number of CABS performed in Canada rose from 7,825 (31 per 100,000) to 10,865 (43.2 per 100,000), while in B.C. they rose from 1,042 (37.7 per 100,000) to 1,140 (39.0 per 100,000). In the six years reported by Peters, B.C. fell from being the province with the highest unadjusted CABS per population rate to the third lowest. These aggregate rates obscure different trends in time and among the age groups. In the Canadian data Peters et al., found three distinct phases in the growth of numbers CABS. From 1981 to 1983 there was an increase of 29.4 percent followed by a 2.2 percent decline over the next two years. The third phase, 1985 to 1987, showed a moderate increase around 7 percent. Gillum (1987), who compared patterns of CABS utilization across the U.S. from 1979 to 1983, found similar phases though they were different for women and men. He also found that blacks had substantially lower utilization rates for CABS than did whites and that these  39  lower rates were not consistent with mortality and prevalence data, which, overall, are not substantially different for the two races. This discrepancy between blacks and whites is also borne out by Ford et al (1989). Both Peters and Gillum found that rates were highest for males in the 55 to 64 age group. In the Canadian data Peters found that the numbers of CABS performed in the youngest age group, 34 to 55-year old's, showed a sustained decline over the time period for both sexes. The 55 to 64 year olds show no growth after 1983 while growth for both male and female over-65-year olds increased steadily throughout the times studied. Gillum did not report on trends over time for different age groups but notes that the patterns of CABS by age and sex correspond more closely with the patterns of coronary heart disease prevalence than to coronary heart disease mortality or incidence. He reports that the peak rates for mortality and incidence occur in the over-65's while those for prevalence are in the 55 to 64-year old group for men and in the 65 to 74-year old group for women; these age-sex groups correspond to those with the peak CABS rates although the male:female ratio for CABS is substantially higher than similar sex ratios for mortality or for prevalence. Feinleib et al (1989), in their analysis of CABS and angioplasty discharges in National Hospital Discharge Survey Data, also indicate that CABS is performed for chronic, rather than acute, ischemic heart disease. In both Canada and the U.S., the rates for the elderly have increased dramatically. Gillum (1987) reports a 138 percent increase in rates for the over-65's in the U.S. between 1979 and 1983. In Canada, Roos and Cageorge (1987) found that CABS rates for the over-75 population in Manitoba had more than doubled between 1980 and 1983. In Ontario, the rates for the over-70 population increased five-fold between 1979 and 1985 although the overall CABS rate in the province increased by only 39 percent during this period (Anderson and Lomas 1988).  40  In summary, after its inception in 1968 CABS showed a phenomenal  growth, unparalleled by any other procedure in the history of medicine. The initial rapid growth during the seventies and early eighties started to plateau in the mid-eighties as the procedure became diffused. The moderate increases seen after the plateau period appear to be due to the increase in the numbers of procedures done on the elderly. According to Peters et al (1990) this may indicate an increasing acceptance of the procedure for elderly patients with the result that the procedure may not be completely diffused in this group. Assessment:  In the more than twenty years which have passed since CABS was first introduced, the methods used to assess the efficacy of new procedures have increased in sophistication. The early reports on the efficacy of CABS were based on observation of, and anecdotal reports from, patients and comparison of outcomes after surgery with the natural history of coronary artery disease. Patients, serving as their own controls, were assessed before and after surgery and were generally found to be improved. The assumption was made that this improvement was due to specific effects of CABS. Preston (1990) points out that this assumption ignored earlier reports that angina may naturally abate in some patients over time. Yeaton and Wortman (1985), in an analysis of 90 studies to determine changes over time in CABS assessment and outcome, found that authors of observational studies expressed significantly more enthusiasm for surgery as favourably influencing survival, than did the authors reporting RCTs. Preston states that, by the early seventies some institutions had accumulated enough cases to make retrospective analyses based on comparison of the surgical group with a non-surgical control group. Many of these control groups had been diagnosed and treated at a different time from the surgical group, and so were not a  41  valid comparison, though many physicians did not recognize that at the time. Later studies attempted to create more valid control groups by matching treatment and control patients on important parameters. However, differences in important but unmatched prognostic indicators remained in some studies (Preston 1990). The first randomized controlled trials (RCTs) comparing CABS with medical treatment, were started in 1971 and 1972 (see Table I). By randomly assigning patients to either medical or surgical treatment, they eliminated the bias that occurs when physicians choose medical or surgical treatment for the patient, and also maximized the likelihood that baseline characteristics would be the same in both groups. Early, small RCTs (from Buenos Aires, Oregon and Vermont) showed no difference in survival between surgical and non-surgical patients; results which were the opposite of those from previous non-randomized studies (Preston 1990). These first studies, whose sample sizes were likely too small to show differences in mortality (Chassin et al 1986), were quickly followed by several larger studies, all of which used different selection criteria (see Table 1). The Veteran's Administration cooperative (VA) study (Takaro et al 1976, Murphy et al 1977) was the first study to divide randomized patients into subgroups for analysis and the first study to find increased survival in any subgroup. Retrospective placement of patients into sub-groups showed improved survival in patients with left main disease who received surgery. This study showed the importance of sub-group analysis, a technique which was incorporated into all subsequent RCTs (Preston 1990). The last RCT comparing CABS with medical treatment (The Coronary Artery Surgery Study - CASS) randomized patients between 1975 and 1979 and issued its first reports in 1983. All studies on the outcomes of CABS since then have been either retrospective analyses or non-randomized prospective studies with or without controls. Preston (1990) states that, unfortunately, there appears to  42  be a consensus among clinicians that the definitive studies on outcomes have been done and there is no need for further RCTs, he goes on to say: "If anything we are regressing. We are like natives in a jungle hospital abandoning the newer methods after the Westerners have left". Changes in CABS Procedure:  When interpreting the literature on the efficacy of CABS, it is important to know the changes which have occurred in the procedure over time. The procedure now is very different from that in 1979 when the last randomized controlled trial finished. These differences have resulted in decreased intraoperative and post-operative mortality and morbidity and improved long-term outcomes from the procedure. Intra-operative changes which have contributed to better short-term outcomes are cold cardioplegia, anaesthetic techniques which help to preserve the myocardium, and shorter operative time. The use of medications such as postoperative aspirin, may also contribute to short-term outcome. Improved long-term outcomes have been obtained by the use of the internal mammary arteries (IMA). These conduits were always favoured by some surgeons (Preston 1990) but did not become routinely used until the early-to-mid eighties when reports of the increased long-term patency of these grafts compared to vein grafts, began to appear (Barner et al. 1982, Loop et al. 1986). The highest patency rates, approximately 95 percent after 10 years, are achieved by the use of the IMA to bypass stenosis of the left descending artery. IMA grafts to other coronary vessels appear to have lower patency rates and these may be no greater than those of vein grafts (Huddleston et al 1986). In contrast to the IMA patency rate, the long-term patency rates for saphenous vein grafts are highly variable; some studies report only 50 to 60 percent  43  of grafts still patent at 10 years. Kroncke et al (1988), analyzing data from the Veteran's Administration randomized trial, found that progression occurred in 74 percent of arteries with patent grafts at five years. Progression occurred in 38 percent of ungrafted arteries in surgical patients and in 36 percent of arteries in medical patients. After adjustment for baseline disease severity the risk of progression was three to six times higher in grafted arteries than in ungrafted arteries. It appears as though CABS, while relieving the patient of one arterial obstruction, may hasten the onset of others. Changes in Medical Treatment:  The majority of studies looking at the efficacy of CABS, compare the shortand long-term outcomes following the procedure with those in patients receiving medical treatment. Advances in the medical treatment of CAD over the last decade have likely reduced difference between these outcomes. Consequently, some of the findings of early studies (including the RCTs) may not be relevant today. For example, all the randomized trials pre-date the wide-spread use of beta blockers post MI, and the use of calcium channel antagonists (only introduced into North America in the late-seventies), long-acting nitrates and platelet-inhibiting agents which have allowed for better control of CAD so that it is now seldom impossible to relieve angina (Kannel 1990). Changes in CABS Patients:  Changes in the type of patients receiving the procedure over time are of interest, both because they indicate continued diffusion of the procedure and because they indicate how relevant the findings of earlier studies are to later patient populations. Advances in cardiology and in the CABS procedure appear to have resulted in a change in the profile of patients undergoing CABS. More  44  patients today are over the age of 65, have three-vessel disease, left ventricular dysfunction, associated medical diseases, emergency surgery or are receiving reoperations. A U.S. study (Naunheim et al 1988) found an increase in operative mortality from one percent to eight percent between 1975 and 1985. Christakis et al (1989), found overall operative mortality in Toronto to be 3.7 percent with no significant change in mortality between 1982 and 1986, although in-hospital morbidity increased from 10.1 percent 13.3 percent. Neither of the above studies explored the reasons for the increase in highrisk patients undergoing CABS but both suggest that it is, at least in part, due to the increase in the use of percutaneous transluminal coronary angioplasty (PTCA). Other factors will likely also play a part. Advances in medical therapy may mean that patients come later to surgery than in earlier years. Also the increased comfort and proficiency with CABS that cardiac surgeons gain with experience, together with the technical advances in the procedure, may make them more likely to accept sicker patients. EFFICACY OF CABS The efficacy of a procedure is a measure of whether it can do what it is supposed to do. Efficacy is related to, and is required for, effectiveness, a measure of how well the procedure can achieve the desired end under real world conditions 14 . Efficacy of CABS is assessed by randomised controlled trials (RCTs). That is, by comparing outcomes in patients randomly selected, from the same population, to receive either medical treatment (the conventional treatment) or 14 The distinction between efficacy and effectiveness becomes important in evaluating the usefulness of a medical treatment. If a treatment is not efficacious then it is of no use. It is efficacious but not effective under current conditions of use then it may become effective if the conditions are changed. For example, patients may need special support in order to comply with lifestyle changes generally required for the effective treatment of CAD. Without these lifestyle changes, efficacious medication may not be effective.  45  CABS. Ideally, the treatments for each group would be standardized as far as possible so that physicians would follow a protocol in, for example, prescribing medication or deciding on further assessment. The major outcomes of interest in assessing CABS are death, MI and angina. When, and to what extent these outcomes occur in a group of patients will depend to a great extent on their clinical condition. Consequently, proven efficacy for one condition cannot be assumed for another. The second condition should also be subjected to rigorous evaluation in a randomized controlled trial. This section will review the literature on the efficacy of CABS for a number of clinical conditions which result from CAD. In situations where there are no RCTs testing efficacy in a clinical population, or where the findings of RCTs conflict, evidence from non-randomized controlled studies (prospective or retrospective) will be presented. In addition, observational studies will be used to look at outcomes in surgical populations who have not been studied in a controlled trial, and to add to information on surgical risk. The Intent-to-Treat Principle: The three major RCTs 15 were analyzed on the intent-to-treat principle. That means that patient data were analyzed according to the group to which the patient was originally randomized. Thus, patients randomized to surgical treatment who never received surgery were analyzed in the surgical group, and medical patients who later received surgery, "crossovers", were analyzed as part of the medical group. The problem of crossovers becomes a major one in long-term follow-up. The 10-year follow-up in CASS showed that 40 percent of patients had crossed over  15 The Veteran's Administration Trial, the European Coronary Surgery Study and the Coronary Artery Surgery Study.  46  while, in the VA study, 47 percent of medical patients with left main disease had crossed over by the 11 year follow-up. The result of analysis according to the intent-to-treat principle is that the comparison is not surgical therapy versus medical therapy but rather the clinical strategy of immediate surgery (followed by medical therapy as required) versus medical therapy with the option of later surgery (Gersh 1989). This is more an evaluation of effectiveness than of efficacy. In this situation, analysis by the intentto-treat principle minimizes the difference in long term outcome. Kannel (1990) points out that with both groups (medically and surgically assigned) of patients in CASS receiving medical treatment and a 40 percent crossover to surgery from the medical group, the 10-year summary finding of no difference in survival or infarction rate may be due to the fact that at 10 years there was little difference in treatment between groups. Analysis according to treatment received would give a better indication of long-term efficacy, as would censoring 16 patients crossing over from medical to surgical therapy (providing that these patients had an acceptable surgical mortality). Chronic Stable Angina:  All three major RCTs showed that CABS was extremely effective in ameliorating angina pain in patients with chronic stable angina (CSA), and this finding was also supported by the results of the small RCTs. Therefore, the discussion on the effectiveness of CABS in patients with CSA will focus on the findings related to survival; less focus was put on the incidence of MI in these studies. Only the three major RCTs will be discussed in detail because, as indicated  16 Patients who are "censored" are analysed as though they were lost to follow-up.  47  earlier, the smaller RCTs had populations which were too small to detect differences in survival between treatment and control groups. The earliest randomized trial, the Veterans Administration cooperative study of coronary artery surgery (VA trial), was initially started in 1970 as an evaluation of surgical treatment for CAD and included the Vineberg implant. Because the majority of the patients were randomized in the CABS protoco1 17 , the study was restarted in 1972 and the Vineberg procedure was dropped. Patients were randomized from January 1972 to December 1974. Patient criteria for inclusion are presented in Table 1; excluded were patients with MI less than six months previously, those with other serious cardiac diseases, other major diseases which may affect five-year life expectancy, or those with unstable angina (Detre, Hultgren and Takero 1977). The first results from the VA trial were reported for the subgroup of patients with significant left main disease (Takero et al 1976). These results showed a significant advantage for the surgical group at 30 months; survival for the 60 surgically treated patients was 85 percent compared to 65 percent for the 53 medically treated patients. These patients were, however, assigned to subgroups retrospectively rather than being prospectively randomized to them. The results were still significant at five years of follow-up but were not reported at ten years because by then 91 percent of patients in the medical subgroup had either died or crossed over to surgery. Unfortunately, insufficient data were presented in the report to allow calculation of the ten-year outcomes according to the intent-to treat principle.  17 Obviously the procedure for randomizing patients was not effective!  48  TABLE 1 INCLUSION CRITERIA FOR THE THREE MAJOR RCTs Study  VA Trial  ECSS  CASS  Dates  1/72-12/74  9/73-3/76  8/75-5/79  No of Patients  686  768  620  Age and Sex Angina  Male Veterans* Male under 65 Present^at^least^6 Present^at^least^3 months with medical months treatment at least 3 months  Disease extent  Reduction^at^least 50% in at^least one artery^with graftable^distal segment 25% or more  LVEF **  Under 66 CCS Class I or II only or postMI (at lest 3 weeks prior to randomization) with no angina.  Obstruction 50% or Narrowing of 70% or more in more in at least two at least one major artery with operable distal segment. major arteries. 50% or more  35% or more  *All ages included. The oldest patient was 67. **Left ventricular ejection fractionl 8 Results for all patients at 30 months showed no differences in survival in patients with one, two or three-vessel disease as a whole (Murphy et al 1977). Later retrospective analysis, however, showed that when data from three hospitals with an aggregate mortality of 23 percent (and performing only 13 percent of the surgery) were excluded 19 , the 6-year survival rate of surgically treated patients with three-vessel disease was significantly better than that of medically treated patients (Takero et al 1982). Retrospective analysis at six years also showed that patients  18 The left ventricular ejection fraction, the percentage of blood that the left ventricle expels per heartbeat, is a measure of left ventricular function. 19 We are not given any information about these patients, hospitals or surgeons so cannot determine whether the high mortality was due to patient mix, the surgeons' experience or to some other factor. Nor is it shown that the mortality rates of these three hospitals were clearly outliers when compared to mortality rates in the other hospitals used in the RCT. Consequently, it cannot be established whether Takero et al were justified in excluding patients in these hospital from the analysis.  49  without left main disease, but considered to be at high clinical and angiographic risk 20 had significantly better survival with surgery than with medical treatment; surgical patients at high risk angiographically but not clinically had a slight, nonsignificant, survival advantage. Patients in low risk groups, however, had a marginally significant advantage with medical therapy. Eleven-year follow-up showed no significant treatment advantage overall (see Table 3). There was still a significant survival benefit for surgical treatment in high-risk patients (measured both clinically and angiographically, i.e., with three vessel disease and impaired LVF). Patients with normal LVF and low clinical risk showed a survival disadvantage with surgical treatment although this was only significant for patients with two-vessel disease. The investigators concluded that among patients with stable ischemic heart disease those with a high risk of dying benefit from surgical treatment but that this benefit gradually diminishes after seven years (The Veterans Administration Coronary Artery Bypass Surgery Study Group 1984). The VA trial has been criticized for its high operative mortality rate, 5.8 percent, which likely handicapped the surgical group. In response to these critics, Takero et al (1983) cite several studies which show similar mortality rates during the same time period, an era in which CABS was just becoming established. The average annual mortality rate for patients without left main disease during the first seven years of follow-up was 3.3 percent for the surgical group and 4.0 percent for the medical group. Between seven and eleven years annual mortality for these groups were 4.8 and 3.5 percent respectively. The increasing mortality for the Clinical risk, high medium or low, was determined on the basis of a multivariate risk function using four clinical variables which were measured at base-line. These variables were: amount of angina according to the New York Heart Association classification, history of hypertension, history of myocardial infarction and an ST-segment depression on the resting ECG. Angiographic risk was determined by left ventricular function and by the number of diseased vessels (The Veterans Administration Coronary Artery Bypass Surgery Study Group 1984) 20  50  surgical group after seven years is likely due to progression of the disease leading to late graft closures (Campeau et al 1983).  TABLE 2 CHARACTERISTICS OF PATIENTS IN RCTs FOR STABLE ANGINA Study Dates Characteristic  CASS (N=780) 1972-74 %  VA (N=686)  ECSS  (N=768)  1975-79  1973-76 %  %  Male Angina none Class I, II Class III, IV  90  100  100  26 74 0  0 42 58  0 57 43  Prior MI  60  61  46  LVEF < 0.5  21  26  0  Data from Alderman et al (1990)  The European Coronary Surgery Study (ECSS) randomized 768 males under the age of 65 years between September 1973 and March 1976. Inclusion criteria are shown in Table I; no exclusion criteria were reported. Characteristics of patients included in the study are shown in Table 2. Survival data reported at two years of follow-up showed that there was no significant treatment advantage overall though there was a significantly higher survival rate with surgery for the subgroup with three vessel disease (European Coronary Surgery Study Group 1979). This latter finding still held at five years of follow-up, and in addition a significant improvement of survival was found overall for the group randomized to surgery and for the surgical sub-group with left main disease (The European Coronary Surgery Study Group 1980). By 10 to 12 years of follow-up the benefit of surgery for the overall group was still significant at the .05 level but the difference in survival between medical and surgical groups was much less than it had been at five years.  51 In addition there was a significant survival advantage overall for surgical patients with stenosis of the left anterior descending (LAD) artery. However, when these patients were stratified by extent of disease, those with two-vessel disease did not benefit significantly from surgery (Varnuskas et al 1988). The differences between the overall results from the VA trial and ECSS may be due to the differences in operative mortality between the two studies. Operative mortality in the ECSS was 3.6 percent overall (1.5 percent in the last third of patients) compared to 5.8 percent in the VA trial. Six of the ECSS surgical patients died before surgery. These studies also differed on the effect of surgery in patients with normal LVF and three vessel disease. Chassin et al (1986) report that the ECSS results prompted the reanalysis of the VA data, excluding the hospitals with extremely high surgical mortality. These authors comment: "This reanalysis does not alter the original study's conclusion that CABS failed to improve survival in the study population with three-vessel disease. It is, however, a useful exercise in the effort to resolve differences among RCTs on this important issue." The National Heart, Lung and Blood Institute's coronary artery surgery study (CASS) randomized 780 patients between 1975 and 1979. The population from which these patients were drawn consisted of 94 percent of all patients at 15 participating centres patients who had coronary angiography for suspected CAD. Of the 16,626 eligible and consenting patients, 2,099 were eligible for randomization but only 37 percent of these were randomized 21 . The non-randomized patients were enrolled in the CASS registry and were followed-up by essentially the same methods used to follow the randomized patients (Principle Investigators of CASS and their Associates 1981).  21 It is not clear why more of the eligible patients were not randomized.  52  Criteria for inclusion in CASS are given in Table I. Excluded were those with previous CABS, the presence of another illness which would reduce the 5year life expectancy, unstable angina or obstruction of the left coronary artery of more than 70 percent. There were no randomized patients with left main disease. Patients in the medical and surgical groups had similar clinical and anatomic features. Follow-up at five, seven and ten years did not demonstrate any significant treatment advantage overall or for sub-groups stratified either on the basis of one, two or three-vessel disease or according to LAD stenosis (CASS Principle Investigators and their Associates 1983). For patients with a normal ejection fraction, long-term survival was best for patients randomized to medical therapy. Surgical patients in this cohort experienced a greater frequency of coronary events (death or MI) after 4-5 years. In patients with an impaired ejection fraction and three-vessel disease, however, survival was significantly improved in the surgical group. At ten years of followup surgical patients in this sub-group showed 80 percent survival compared to only 59 percent in medical patients. This difference in outcome started to become apparent by year 3 and gradually increased until year 10 (Alderman et al 1990). This is in contrast to the VA study where the surgical advantage for the high-risk subset reached a maximum at seven years and was no longer apparent by eleven years. All of the above RCTs were multicentre trials and, with the exception of CASS, none of them standardized treatment. Consequently it is impossible to say with certainty how CABS compares with maximum medical treatment. Multicentres also posed problems for standardization of surgery. There was a wide range in the operative mortality rate among hospitals in the VA trial and in the CASS. Finally, as the authors point out, the sample size of the CASS was likely too  53 TABLE 3 OUTCOMES IN RCTs FOR STABLE ANGINA  Outcome Operative death  CASS Med^Surg (N=390)^(N=390) % % 1.4  Perioperative MI  8  6.4  Survival 5 year 10 year  92 79  95 82  LMD 5/6 year 10 year LAD stenosis 5/6 year 10 year  (n=6) (n=275) 92 78  (n=8) (n=277) 94 82  Impaired LVF 5/6 year 10/11 year 3-vess dis. 5/6 year 10/11 year  (n=82) 85 59  (n=78) 96 80  3-vessel disease 5/6 year 10/11 year  VA TRIAL Med^Surg (N=354)^(N=332) % % 5.8  85 57 (n=53) 70  9.9  -  <8.0  90 58  83 70  92* 76*  (n=60) 95* -  (n=31) 65 61 (n=240) 79 65  (n=28) 79* 64 (n=262) 90* 76  (n=75) 80 53  (n=0) N/A N/A  (n=0) N/A N/A  N/A N/A  N/A N/A  (n=186) 82 68  (n=220) 94* 78*  (n=72) 73 49  78 57  91 75*  66 38  83 50  (n=135) 90 75  (n=123) 93 76  (n=156) 75 50  (n=135) 81 56  Reoperations  -  9.7  Crossovers Graft patency 18 month 60 month  38  7**  -  90 82  ECSS Surg Med (N=373) (N=394) % % 3.2  11  -  9.9  38  6**  36  5.8**  70 67  -  -  75 69  LMD left main disease, LAD left anterior descending artery, LVF left ventricular dysfunction Data from Alderman et al 1990, Varnuskas et al 1988, Takero et al 1982. * significant at 0.5 level or less ** randomized surgical patients not receiving CABS.  54  small to detect a difference in mortality, given the minimal difference in outcome between the medical and surgical groups. Taken as a whole the data from the three randomized trials indicates a definite surgical advantage for patients with left main disease and for those with a moderately impaired ejection fraction with three-vessel disease. In addition the ECSS showed a significant surgical advantage for patients with a normal ejection fraction and three vessel disease. The VA results can be reconciled with this but the CASS findings are in conflict. In an editorial, Killip (1988), states that the outcome differences between CASS and the two earlier trials may be reconciled by comparison of their patient populations. The CASS patients likely had less severe disease than patients in either of the other trials; they were specifically selected because they were at low risk, on the grounds that high-risk patients should not be denied the chance of immediate surgery. The medical mortality of only 2 percent per year is testimony to the low-risk status of these patients. Killip states that low-risk patients will do well whatever the method of treatment, provided that it is not actually harmful. Chassin et al (1986), in an in-depth review of CABS investigations prior to 1981, tentatively propose that the surgical survival benefit seen in the earlier trials was due entirely to the benefit of surgery for angina Class III and IV patients. There were no such patients in the CASS trial but they made up 58 and 42 percent of the VA trial and ECSS respectively. The results from other studies also appear to conflict on this point. The Seattle Heart Watch, a community-wide observational study of patients who underwent angiography between 1969 and 1974, showed improved surgical survival in patients who had three-vessel disease and impaired LVF, but not in those with normal LVF (Hammermeister, DeRouen and Dodge 1980). Analysis of data from the 5,809 patients in the Duke registry indicated that in 1970 surgery for  55  one- or two-vessel disease was associated with poorer survival, and no significant treatment difference was found for three-vessel disease. By 1984, however, patients with two- or three-vessel disease had significantly longer survival with surgery (Califf et al 1989). The surgical advantage for three-vessel disease in this study was already noticeable by 1977; it may be that the differences in time between the VA and ECSS trials accounts for the different outcomes for patients with normal LVF and three-vessel disease. It should be noted though, that Califf and colleagues did not look at the extent of disease according to degree of LVF (i.e., subgroup analysis) but instead adjusted for baseline differences in ventricular function. This may have affected the results. Despite the diverse findings from the literature on CABS for patients with CSA and three-vessel disease with normal LVF, many studies advocate it as an indication for CABS (Kirklin et al 1991, Gersh et al 1989). The likelihood is that, as time passes and the RCTs recede further in to the past, three-vessel disease will become an accepted indicator for elective CABS. For patients with stable angina a number of other variables besides the number of vessels involved, affect survival. Of these the most important are left ventricular function and severity of angina.  Left ventricular function alone has been shown to be the most powerful determinant of long-term survival in chronic stable angina. Gersh et al (1989) report that the substantial benefit of surgery demonstrated by medical registry studies for patients with impaired LVF, have changed this characteristic from a relative contraindication for CABS to a major indication for the procedure. Kirklin et al (1990) state that although early and late results of CABS are worse in patients with left ventricular dysfunction, the comparative benefits (i.e., between medical and surgical treatment) are better. These comparative benefits are  56  especially great when other risk factors, such as extensive coronary disease and severe ischemia, are also present. Left ventricular dysfunction is also a significant predictor of operative mortality in CABS. Parsonnet, Dean and Bernstein (1989) found that operative mortality increased four percent when the ejection fraction was less than 30 percent. In a prospective, multi-institution study in Toronto between 1982 and 1986, Christakis et al (1989), using multivariate analysis, found that a LVEF of less than 40 percent was the second most important independent predictor of operative mortality overall, but that this changed over time. In 1982 LVEF was the most important predictor of mortality but it had fallen to fourth place by 1986. Overall operative mortality for these patients was not reported, but for those with LVEF less than 20 percent, operative mortality was 11.7 percent compared to 2.5 percent for those with LVEF greater than 40 percent. Kirklin et al (1991) state that extreme left ventricular dysfunction probably results from myocardial scarring and consequent irreversible ischemia. In such a case the prognosis is limited whatever the treatment. The level of LVEF which indicates lack of surgical benefit is not known but is probably below two percent. Severity of angina appears to be a powerful prognostic factor in patients  treated medically but not in those receiving CABS. The benefit of surgery on survival is greatest in patients with severe symptoms (Kirklin et al 1991, Gersh et al 1989). This is likely because the severity of angina is generally a reflection of the severity of myocardial ischemia. Ischemia on exercise testing has been associated with lower survival and a higher rate of infarction in medical patients (Hlatky et al 1987, Weiner et al 1987) even in the absence of severe symptoms (Bonow et al 1984). In summary, for men under 65 years of age, the major issues to be taken into account when considering surgery are the degree of LVF, the location and severity  57  of the stenoses, the extent of disease and the severity of symptoms. Patients with left ventricular dysfunction and/or left main disease will receive a significant survival benefit from initial surgery while those with three-vessel disease, severe symptoms or with stenosis of the LAD artery may experience a survival benefit. Patients with one- or two-vessel disease, normal ventricular function, mild to moderate symptoms and no exercise-induced ischemia have an excellent prognosis with medical treatment and initial surgery is probably not indicated. Unstable Angina:  There have been four randomized trials which compared medical and surgical treatment for unstable angina, but only two of them, the National Cooperative Study and the Veterans Administration trial will be discussed as the others were too small to detect significant differences in survival. The National Heart, Lung, and Blood Institute (NIH) conducted a trial in nine cooperating centres between 1972 and 1976. Inclusion criteria are given in Table 4. Excluded were patients who had recent MI, left main disease, LVEF of less than 30 percent or a comorbid disease which would reduce life-expectancy to less than five years (Russell et al 1976). Two hundred and eighty-eight patients were randomized to medical or surgical treatment. In-hospital (operative) mortality was 3 and 5 percent for the medical and surgical groups respectively. The greatest operative mortality was seen for surgical patients with three-vessel disease, although this was not significant. Non-fatal in-hospital MI's were experienced by eight percent of the medical group and 17 percent of the surgical group; a significant difference (Russell et al 1978). At 30 months of follow-up, (Russell et al 1978), 36 percent of the patients assigned to medical treatment had crossed over. Overall survival and MI rates  58  were similar for medical and surgical groups (see table 5). Significantly more patients in the medical group had class III or IV angina, by the New York Heart TABLE 4 INCLUSION CRITERIA FOR RCTs OF UNSTABLE ANGINA  Criteria  NIH Study  VA Trial  Dates  1972-1976  1976-1982  # of patients  288  468  Sex and age  <70  male <70  LVEF  >30%  >30%  Signs and Symptoms  One or more attacks of severe coronary pain at rest, associated with S-T or T wave changes or both. Admission to a coronary care^unit^for^suspected impending infarction.  Type I angina: present > 2 months Increase^in^severity, frequency or angina at rest within 8 weeks prior to entry. Pain within 10 days of admission. ECG changes with pain at rest or on exercise stress test. Type II angina: recurring episodes^of^chest^pain resistant^to^nitrates.^At least one episode (>15 min) within 10 days of entry. ECG changes during at least one episode.  Association classification (NYHA), in the first year but during the second year this finding was only significant for those patients with one-vessel disease; possibly bec I use the patients with more severe angina had crossed over and received surgery which reduced their symptoms. Four-year follow-up on patients with 70 percent or greater obstruction of the LAD (Russell et al 1981), showed no significant difference in incidence of MI or mortality despite a large operative mortality in the surgical group (9 percent versus 2 percent, non-significant). In a later analysis of a sample of the patients, Russell et al (1980) found that a larger percentage of medical than surgical patients had returned to work. This overall difference was not significant but when this was expressed as a ratio of the  59  percent employed at time of follow-up to the percent employed 3 months prior to entry into the study, the difference was significant at the 0.01 level. TABLE 5 RANDOMIZED TRIALS FOR UNSTABLE ANGINA PATIENT CHARACTERISTICS Trial  Mean Age Previous MI CAD severity 1-vessel 2-vessel 3-vessel LVEF 50+ 30-49 LAD  NIH Med (N=147) 52.7  Trial Surg (N=141) 53  VA Med (N=237) 56.3  Trial Surg (N=231) 55.7  27  35  42.6  41.7  24 37 39  24 33 43  18.6 33.1 48.3  18.8 36.7 44.5  62 23 15  65 23 12  71.3 28.7 -  71.4 28.6  69  71  -  -  -  LVEF left ventricular ejection fraction, LAD left anterior descending artery. Data from Russell et al 1978, Lurchi et al 1987, Parisi et al 1989  The Veterans Administration Cooperative Study on Unstable Angina (VA Trial) randomized patients between June 1976 and June 1982, from eligible patients admitted to one of the participating hospital with chest pain. Inclusion criteria are shown in Table 4; excluded were patients with a diagnosis of acute MI, left main disease, a LVEF below 30 percent, a history of previous surgery for angina, or those who were participants in another study, Of the 3,159 patients screened, 2,433 were excluded by virtue of the criteria desceibed above and 468 patients were randomized. Prior to randomization, patients were stratified according to ventricular function and type of angina, and were booked for surgery. Therapy for medical patients was assumed to start on the date that surgery had been scheduled. Patient characteristics are shown in Table 5. The mean interval between randomization and surgery was 9.3 days in the surgical group and 3.6 days in the  60  medical group 22 ; no deaths occurred in this interval. Operative mortality was 4.3 in patients with Type I angina and 2.1 in those with Type 2. Eleven patients did not have surgery and were classified as crossovers to medical therapy. In the medical group five patients died within 30 days of randomization; a mortality rate of 1.6 percent for type I and 4.3 percent for Type II angina. The crossover rate to surgery in the first 30 days was 19.1 percent for Type II and 5.8 percent for Type I disease; thereafter the crossover rates were the same. By the 2 year follow-up, the cumulative crossover rate was 34 percent. Operative mortality in patients who crossed-over to surgery was much higher than for the assigned surgical group but the perioperative MI rate was the same (Luchi et al 1987). At two years of follow-up (Luchi et al 1987), no treatment advantage was seen overa11 23 but a significant surgical survival benefit was found for the subset of patients who had LVEF between 0.3 and 0.59 percent. This benefit was only significant when LVEF was treated as a continuous variable, and so was not found on analysis using the life-table method where LVEF was entered as a categorical variable 24 . At five year follow-up the surgical benefit was still present but was no longer significant. However, significant surgical advantage was seen for other subgroups at five years. Surgical patients with three-vessel disease had significantly lower mortality than medical patients (11 versus 24 percent); this survival advantage was increased for those surgical patients with three-vessel  22 No reason was given for this difference in randomization to surgery time between groups. Possibly the surgical patients had their surgery rescheduled. 23 Analysis was performed using both treatment assigned (intention-to-treat) and treatment received (with crossovers withdrawn at the time of crossover) and no statistically significant differences were noted. 24 Presumably the LVEF range for each categorical variable had clinical significance but we are not told what they were.  61  disease and abnormal LVF (9 percent mortality versus 29 percent medical mortality) (Parisi et al 1989). TABLE 6 OUTCOMES IN RCTs FOR UNSTABLE ANGINA  Outcomes  NIH STUDY Surg Med (N=141) (N=147) % %  Med (N=237) %  VA TRIAL Surg (N=231) %  Mortality 30-day 2-year 5-year  3 9 -  5 10 -  2.1 9.3 19  4.1 7.8 16  Non-fatal MI 30-day 2-year 5-year  8 11 -  17 13 -  4.6 12.2 17.7  11.7 11.7 15.6  Angina ** 1-year 2-year  36 19  11.3 10.5  -  -  No angina*** 5-year  -  -  32.9  54.8*  From Russell et al (1978), Parsi et al, (1989) *Significant at or below 0.05 **Class III or IV angina - data calculated from Russell et al (1978) *** Data from Booth et al (1991).  Multiple regression analysis showed LVEF to be a major predictor of mortality for medically treated patients. For those with ejection fractions between 0.3 and 0.49 survival was significantly better for the surgical group. There was no significant treatment difference for LVEF between 0.5 and 0.69, while for the subgroup with LVEF above 0.7, survival was significantly better in the medical group (Scott et al 1988). Relief of angina was significantly better in the surgical group. At five years of follow-up, approximately 33 percent of patients in the medical group were free  62  supported by the fewer medications used by the surgical group. At one and three years the number of patients not taking propanalol was significantly greater in the surgical group, but this difference was no longer significant at five years (Booth et al 1991). In summary, the findings of the two randomized trials are congruent. There is no significant treatment advantage overall with respect to survival or the incidence of MI, but the short-term relief of angina is significantly better with initial surgery. In addition the VA trial found long-term survival advantage for surgical patients with three-vessel disease and those with three-vessel disease and abnormal left ventricular function. The NIH study was terminated after 30 months and so did not look at long-term survival except in patients with LAD stenosis. There are a number of factors which may affect how the above findings are interpreted today. Firstly, in view of the relatively high survival of the medical group in both trials, sample sizes may not have been sufficient to detect a difference between the treatment groups, especially in the smaller sub-group of Type II patients (Parisi et al 1989). In other words, there may have been a treatment difference which the study did not have the power to detect. The second group of factors relate to the advances in non-surgical treatment of angina since these studies were carried out. Calcium-channel blockers were not introduced in the U.S. until after the NIH study and half-way through the VA trial; their use, together with beta-blockers and long-acting nitrates have reduced the number of failures in medical therapy (Gerstenblith et al 1982). Also the use of aspirin for platelet inhibition, the introduction of thrombolysis and the growth in the use of coronary angioplasty for unstable angina have likely increased the effectiveness of medical therapy.  63  Asymptomatic Patients with "Silent" Myocardial Ischemia: There have been no RCTs which looked at the effectiveness of CABS in asymptomatic patients, although the clinical significance of silent ischemia has been the subject of much debate in the literature. One problem with interpreting the literature on silent ischemia is that few, if any, authors have studied the phenomenon in completely asymptomatic patients. Most studies look at patients who have exercise-induced or holter-monitor detected silent myocardial ischemia in addition to angina. Intuitively, it appears likely that completely asymptomatic myocardial ischemia will have a different prognostic significance than that in patients in whom angina is also present. It appears that silent myocardial ischemia does have prognostic significance in patients with documented CAD. Weiner et al (1987) analyzed stress test and mortality data from 2,982 patients with proven CAD in the CASS registry. Those with silent ischemia (S-T depression but no chest pain) on exercise testing (Group 1), had a 7-year survival rate (76 percent) similar to those with angina but no ischemia (77 percent) and those with angina and ischemia (78 percent). Patients with neither angina nor ischemia had a seven-year survival of 88 percent, while a control group of patients who had no CAD but who had ischemia without angina had a survival of 95 percent. Among Group 1 patients, survival was related to the extent of CAD; survival was 86 percent, 73 percent and 57 percent respectively for patients with one-, two-, and three-vessel disease. Because quality of life, at least as far as symptoms are concerned, cannot be altered following CABS in completely asymptomatic patients, survival data becomes most important in assessing benefit. Unfortunately there appear to be no survival data for completely asymptomatic patients who have not had a myocardial infarction. Patients in this latter category will be discussed later.  64  Josphson (1990) believes that determining the effect of CABS on silent ischemia can provide useful information on the effectiveness of revascularization. He cites several studies which show that CABS reduces, and occasionally eliminates, silent myocardial ischemia. However, in the absence of controlled survival data, one cannot conclude that treatment of silent ischemia by revascularization provides a survival benefit over patients who are treated medically. Post-Myocardial Infarction: Two randomized controlled trials have compared medical and surgical treatment in patients who have had at least one M.I. One hundred and sixty asymptomatic post-MI patients were included in the CASS randomized trial. There was no significant difference seen in survival between treatment groups; at ten years 69 percent of the medical group and 81 percent of the surgical group were still alive. Those alive and free of MI at ten years were 62 and 68 percent of the medical and surgical groups respectively (Alderman et al 1990). Between 1972 and 1979 in New Zealand, Norris et al (1981) randomized 100 patients, aged 60 or younger, who had survived two or three myocardial infarcts which were at least one month apart. Excluded were men with congestive heart failure, left main disease and disabling angina. Patients were described as having "few symptoms with severe coronary disease and favourable vessels for grafting". Approximately two-thirds of the patients had a LVEF below 50 percent. At a mean follow-up time of 4.5 years six patients in the surgical group and five in the medical group had died, and all but one of these were cardiac deaths. The crossover rate, from medical to surgical treatment, was 14 percent. The results of these two studies indicate that CABS is not more effective than medical treatment for asymptomatic MI survivors who are at least 30 days  65  post-MI. These studies, however, did not distinguish between asymptomatic patients who have silent myocardial ischemia and those who did not. In a chapter reviewing the literature of silent ischemia in survivors of MI, Deedwania (1988) concludes: Uncontrolled observations indicate ... that the prognostic implications of silent and symptomatic episodes in the survivors of acute MI are essentially the same. Thus, treatment regimen designed to alter the adverse consequences of persistent myocardial ischemia should not distinguish between symptomatic and asymptomatic ischemia. However, controlled clinical trials will be needed [Italics mine] to determine whether the elimination of persistent asymptomatic myocardial ischemia in the survivors of acute myocardial ischemia [sic] will reduce the short- and longterm morbidity and mortality. (p171) . For patients who develop angina after MI the situation is also unclear. Both the VA trial and the ECSS include patients who had an MI but they did not report on them as a separate sub-group. However, Chassin and colleagues (1986), in discussing these trials, state that "it is reasonable to presume that their results apply to patients who develop CSA following an MI". Gardner et al (1989) studied 300 patients who received CABS after they developed postinfarction unstable angina. Hospital mortality was five percent, survival was 96 percent and 88 percent at one and five years respectively; there was no control group. However, the hospital mortality was higher than the two percent generally accepted for patients with unstable angina (Scott et al 1988). This may reflect the fact that CABS was carried out early after the infarction; the mean time from onset of MI to surgery was only 11.7 days. A similar retrospective study carried out by Naunheim et al (1988) on 336 patients who underwent surgery within 30 days of MI. Overall operative mortality was 7.7 percent but this differed markedly between subgroups. Operative mortality was 2.3 percent for those with stable or no angina, 6.1 percent for those with angina at rest, 9.5 percent for patients with severe angina requiring intra-aortic balloon  66  counterpulsation and 47.8 percent for those in cardiogenic shock. The authors, somewhat optimistically, conclude that early postinfarction CABS may be performed on the stable patient at any time with a mortality similar to that of elective CABS. However, an operative mortality of 2.3 percent is higher than the less than one percent found in other series in the early to mid- eighties (Gersh et al 1989). On multivariate analysis, Naunheim and his colleagues found that age, left ventricular function and the hemodynamic state of the patient were significant predictors of operative mortality. Time between the onset of MI to CABS was a predictor of mortality in univariate analysis but not in multivariate. Hochberg at al (1984) also looked at the question of how soon CABS should be performed after MI. Overall analysis showed that operative mortality was dependent on the time between surgery and the MI; those operated on in the first post-MI week had a mortality of 46 percent, compared to 7 percent in the seventh week. However, when patients were stratified according to ventricular function these time dependencies changed. Patients with LVEF at or above 50 percent survived at whatever time they were revascularized. Those with LVEF below 50 percent had better survival the more remote their surgery was from the infarction. The question of whether CABS is effective in either symptomatic postinfarction patients, asymptomatic patients postinfarction with myocardial ischemia, or early after MI is unanswered because of a lack of comparative data. Either a randomized controlled trial or a well-matched, non-randomized study is required for all situations. Evolving MI:  Revascularization of the myocardium within 4-6 hours of the onset of an MI can prevent irreversible myocardial ischemia and, consequently, infarction.  67  There is only one RCT comparing medical and surgical treatment of evolving MI reported in the literature. Koshal et al (1988) randomized 68 patients presenting within 4 hours of onset of chest pain, to receive CABS or medical treatment that did not include thrombolysis. Patients older than 65 years, those with serious comorbidity, cardiogenic shock, or prior MI were excluded. Thirty-day mortality was 2.9 percent for surgical patients compared to 8.8 percent in the medically treated group; 18-month mortality was unchanged for the surgical group but 20.6 percent for those medically treated. Although the size of the study was too small to provide statistical significance of the results, the study demonstrates that prompt reperfusion can be carried out with a low operative mortality and encouraging long-term survival (Ryan 1990). Results from a case-series in Spokane, where reperfusion early in acute MI has been carried out since 1971, support Koshal's findings. Between 1972 and 1976, surgical patients receiving CABS within 24 hours of peak symptoms experienced a 4.4 percent operative mortality. Other studies though have shown less optimistic findings. An observational study of 83 patients who received CABS during evolving MI showed an operative mortality of 15.6 percent, compared to mortality of 13.5 percent of patients in the same institution who were treated by thrombolysis, angioplasty or both. Significant predictors of operative mortality were cardiogenic shock, age over 65 years, LVEF less than 0.3 and absence of collateral vessels (Athanasuleas et al 1987). Differences in patient populations doubtless account for some of the differences between these and Koshal's results. There is no evidence to show the efficacy of CABS after thrombolysis although approximately 20 percent of patients suffer a reocclusion within the first week of thrombolytic therapy (Ryan 1990). CABS is evidently being used early after thrombolysis because Naylor and Jagdal (1990) found a significant short-term  68  increase in revascularization for patients treated with thrombolysis for acute MI compared to those treated conventionally. Ryan (1990) speculates that it is unlikely that CABS or PTCA will ever become the primary modes of therapy for MI, given the high cost and the logistical problems in getting the patient to surgery within the time before infarction takes place. He notes that probably 50 percent of the population who undergo acute MI will be ineligible for thrombolysis because of age or other contraindications, and, therefore, it will be imperative to clearly establish the role of primary revascularization procedures in the treatment of acute MI. In summary, there is evidence that CABS can reduce mortality if performed during an evolving initial MI on patients under 65 who are not in cardiogenic shock, although this evidence comes from an RCT which did not have sufficient power to measure a clearly significant difference and which excluded the highest risk patients and those over 65. Further research is required to determine which patients will benefit, and when, from revascularization during MI. Emergency CABS after Failed PTCA:  A number of observational studies have looked at outcomes of emergency CABS for failed coronary angioplasty. The incidence of failed PTCA requiring surgical revascularization ranges from 0.9 to 21 percent (Kahn et al 1990). Failure appears to be related to the year in which the procedure was performed and to the number of procedures performed in the institution (Parsonnet et al 1988). Outcomes for these patients appear to be worse than for those patients receiving elective CABS. Operative mortality ranges from 1.4 percent to 15 percent, perioperative MI from 20 percent to 51 percent. Parsonnet et al (1988) found a complication rate that was significantly higher than for patients undergoing elective CABS (61 percent versus 45 percent). Major complications arose in 45  69  percent of the emergency group. These authors conclude that in view of the higher mortality and morbidity rates, increased length of stay (15.3 versus 13.4 days) and consequent increased use of hospital resources, emergency CABS following failed PTCA cannot be equated with elective CABS. Re-operations: There are three periods during which reoperation may be considered necessary. The first is early, during the first or second year, and is generally due to a technical fault or to compromised blood flow which eventually leads to graft occlusion. The latter problem may be averted through the use of platelet inhibitors. The second period occurs four or five years post-operatively and is generally due to progression of disease in ungrafted arteries or to incomplete revascularization at surgery. Late revascularization, which occurs eight to twelve years postoperatively, is generally performed because of graft disease which is often associated with disease in ungrafted vessels. Nearly 70 percent of reoperations occur in this later time period which is the more hazardous because of the greater age and worse condition of the patient, anatomical changes resulting from the first operation and the presence of atherosclerotic grafts which have the potential to embolize during the operation (Grondin et al 1990). The incidence of reoperation is about 3 percent during the first postoperative year (Foster et al 1984) rising to about 17 percent by twelve years (Cosgrove et al 1986). Factors which have been identified as predictors of reoperation are total cholesterol level, smoking, hypertension, incomplete revascularization and lack of an internal mammary artery graft (Solymoss, B.C. et al 1988, Cosgrove et al 1986). Operative mortality ranges from three to nine percent, depending partly on the era of reoperation and the experience of the surgeon (Grondin et al 1990). Analysis of CASS Registry data (Foster et al 1984), showed that  70  the risk of death was highest on the operative day; in this series the incidence of death for subsequent days was not significantly higher than that for the primary procedures. The incidence of perioperative MI is also higher than at initial operation and Foster and colleagues found that the complications of perioperative MI, while being no more frequent than those in the primary procedure, were more frequently fatal. There have been no RCTs which have looked at the effectiveness of CABS versus medical treatment for second or subsequent reoperations. In fact there do not appear to be any studies of reoperation that used control groups. Evidence of long-term effectiveness of reoperation must, therefore, come from uncontrolled series. Table 7 shows that long term survival following reoperation is better than that reported for primary operations in the RCTs. However, these studies were carried out later in time than most of the RCTs, and more than 50 percent of patients in the Cleveland study received IMA. While it appears that reoperation has the potential for long-term benefit the lack of comparative data to show how patients receiving medical treatment rather than reoperation fared in these institutions over the same time period means that such a conclusion is only tentative at best. The Elderly:  This literature review has thus far looked at the effectiveness of CABS in various clinical conditions. It is, however, important to note that the populations studied in the RCTs were mainly males under the age of 65 years 25 ; consequently the efficacy and effectiveness of CABS in females and in older patients is not 25 The NIH trial criteria included males of any age but the oldest patient in the study was 67. Women made up 10 percent of the population in CASS.  71  known. In fact, older age was found to be a significant predictor of mortality in almost all studies which used multivariate regression to identify predictors of mortality or morbidity post-CABS, although the relative importance of age differed between studies. TABLE 7 PERCENTAGE OF PATIENTS SURVIVING FIVE OR MORE YEARS AFTER RE-OPERATION Results^Mayo Clinic *^Cleveland Clinic ** 5 year^10 year^7 year^10 year Survival^94^89^90^75 Event free^ 63 ***^63*"  76^48  Asymptomatic^28^-^52 Years  ^  1969-1980^  1967-1984  Adapted from Grondin et al (1990) *N=160 **N=1,500 ***Excludes Class I-II angina  There are several reasons why the over-65's may have different outcomes than younger patients. Older people with CAD generally have poorer left ventricular function, more extensive coronary artery disease, a higher incidence of unstable angina, and more associated medical diseases, all factors which have been associated with higher mortality and morbidity (Gersh et al 1983). In addition, older people are physiologically less able than younger patients to respond to the stress of major surgery, and their response to medication may also be different. These differences mean that the results of studies in the under-65 year olds cannot necessarily be applied to those over-65. The efficacy of CABS in people over-65 can only be established by a RCT which draws its sample from this population. To date there are no RCTs which do this.  72  In the absence of data from RCTs, information on the outcomes of CABS in the elderly must come from less rigorous studies. The only controlled study comparing surgical and medical therapy in CAD patients 65-years of age or older, analyzed data from the CASS Registry of patients receiving angiography between 1974 and 1979 (Gersh et al 1985). Excluded from this analysis were patients with left main disease, those who did not have angina as the primary symptom and those with less than 70 percent stenosis of a major vessel. Patients who received CABS were assigned to the surgical group if they received the procedure within the time that 95 percent of all patients, in the first year of the registry, undergoing CABS in the same institution received their surgery. All other patients were in the medical group which included more women and significantly more patients with associated medical diseases and impaired LVF. On long-term follow-up the cumulative unadjusted six-year survival was significantly better in the surgical group (80 percent compared to 63 percent in the medical group), despite an operative mortality of 4.3 percent. When survival curves were adjusted for the major prognostic variables, survival was still better in the surgical group (79 percent versus 64 percent). Unadjusted six-year survival rates for age subgroups are shown in table 8. TABLE 8 UNADJUSTED CUMULATIVE 6-YEAR SURVIVAL IN AGE SUBGROUPS OF THE OVER-65'S p  Medical N^% survival  Surgical N^% survival  65-69  459  67  625  81  0.0001  70-74  129  51  200  77  0.0001  75+  42  56  36  75  0.14  Age  73  Analysis of functional status showed that symptoms in both medical and surgical groups at five years were improved over baseline. In part this may reflect a relatively higher death rate in patients with more severe symptoms, but it also likely indicates improvement in angina over time. Gersh et al (1983) had earlier reported on a series of patients over the age of 65 from the CASS registry. This study compared outcomes in 1,086 patients over65 to 7,827 CASS registry patients under-65. As can be seen in table 9, both operative mortality and long-term survival were significantly worse in the over-65 patients and mortality increased with increasing age. The presence of associated comorbid conditions had a significant effect on 5-year survival in the older agegroup, as did the presence of left ventricular dysfunction, and a pre-operative history of hypertension. These variables were not reported for the under-65s. Surprisingly, event-free survival was found to be higher in the older group but was significantly poorer in women who had less symptomatic relief than men. The data indicated that recurrence of angina was lower in older patients but Gersh and colleagues suggest that this could be due to lower levels of activity in older patients. Taken together these studies indicate that although CABS patients over-65 have a greater surgical risk and poorer long-term survival than those under-65, long-term survival and freedom from angina may be better in those treated surgically than in those on medical treatment. Increased risk for poorer long-term outcome is seen in those with associated co-morbid diseases, impaired LVF, a history of hypertension and in older age-groups. However, it should be remembered that these patients were not randomized to treatment and so selection bias will preclude generalization of these results to all over-65s with CAD.  74  TABLE 9 OUTCOMES OF CABS IN CASS REGISTRY PATIENTS Outcomes Operative Mortality Age 65-69 (N=803) Age 70-74 (N=241) Age 75+ (N=42) 5-year survival Age 65-69 Age 70-74 Age 75+ LVEF <35* 36-50 51+  Under-65 (N=7827) % 1.9 91 -  Over-65 (N=1086) % 5.2 4.6 6.6 9.5 83 84 80 70 47 79 88  Comorbid conditions None (N=489) One Two or more  -  89 80 71  Event-free  39  47  69 5 26  53 14 33  Cause of death Cardiac related Non-cardiac, CAD rel Other non-cardiac  Data from Gersh et al 1983 *4 percent of those tested There are numerous observational studies in the literature which describe outcomes of CABS in various age subgroups of the elderly. Those looking at septuagenarians showed an operative mortality ranging from 3 to 22.1 percent (Meyer at al 1975, Richardson and Cyrus 1984,), while the range for those 80 years old and older was 6.3 to 24 percent (Edmunds et al 1988, Mullany et al 1990,). Mullany et al (1990) found that operative mortality in octogenarians was only 4 percent in those with isolated coronary disease compared to 13.8 percent in those with associated diseases, Goldman et al (1987) report on a prospective  75  observational study of 3,327 patients undergoing CABS at the Toronto General Hospital between 1982 and 1986. Over the four years of the study patients over 70 years of age increased from 4.5 to 14.2 percent of total bypass patients. Significantly more of the 340 patients over the age of 70 had high risk characteristics than did those under 70. One third of the older group had urgent surgery compared to 17 percent of those under 70. Clearly the older patients as a group were sicker than the younger ones and this is reflected in their poorer outcomes; an operative mortality of 6.1 compared to 2.9 percent, a higher incidence of stroke (4.4 versus 1.3 percent), and post-operative low-output state. Operative mortality was seen to increase with increasing age over the whole population, rising from one percent in those 30 to 39 years old to 7.3 percent in those aged 75 to 80 years. There were no operative deaths in the six patients aged over 80. In the above study, Goldman and colleagues also retrospectively reviewed a random sample of patients from the original cohort and found that the patients over the age of 70 required longer stays in the intensive care unit, longer ventilatory support and longer overall hospitalization (18.9 ± 11 versus 15.5 ± 10 days). Edmunds et al (1988), looking at open-heart surgery in octogenarians, found that those with complications were more resource intensive; mean length of stay (LOS) was 24.9 ± 19.6 days for those with complications and 11.5 ± 3.7 days for those without. These observational studies confirm the CASS registry findings that elderly patients are at higher risk of operative death and morbidity. Differences in operative mortality between studies likely reflect the era in which surgery was performed, patient selection or both. In addition these studies confirm that older patients undergoing CABS use more hospital resources than younger ones; an unsurprising finding given the higher morbidity in these patients.  76  Several authors comment that elderly patients should be investigated sooner in order to allow revascularization before their condition deteriorates to the point where they are at high risk for post-operative death or complications. While this would be a laudable action if CABS was proven to be efficacious in the elderly, the fact that efficacy is not proven means that caution in subjecting patients over-65 to the risks of angiography and surgery is justified.  Gender: A number of studies have reported greater operative mortality in women (Bolooki et al 1975. Fisher at al, 1982, Loop et al 1983) which has been attributed to smaller body size (Loop et al 1983) and to smaller cardiac size leading to more difficulties during anastomosis and earlier closure of vessels (Fisher et al 1982). Tobin et al (1987) found that, following a positive nuclear exercise test, men were ten times more likely than women to have cardiac catheterization and suggested that this disparity may be caused by a sex bias. Kahn et al (1990) were led by Tobin's work to investigate whether differences in the pre-operative status of women could account for their higher operative mortality. They investigated 2,297 consecutive patients (1,815 men and 482 women) undergoing isolated CABS at one institution after 1982. They found that women were significantly older than the men, more often had a history of hypertension and diabetes mellitus, and more frequently had unstable angina, post-MI angina and pre-operative symptoms of heart failure. There was a trend for women to be referred in cardiogenic shock more frequently but this was not significant. Women were referred significantly more often than men with severe symptoms of coronary artery disease (heart failure, unstable angina, post-MI angina, cardiac arrest or cardiogenic shock) while men were referred for surgery significantly more frequently with an abnormal exercise test.  77  Operative mortality was 2.6 percent for men and 4.6 percent for women. Operative mortality differed significantly by referral symptom; patients referred for either previous MI or post-MI angina had significantly higher rates while those referred for an abnormal exercise test had significantly lower mortality rates. Multivariate analysis showed no independent effect of gender, body surface area, height or weight on operative mortality. Instead, the higher operative mortality of women was explained entirely by their greater age and higher NYHA functional class. The authors conclude that women are referred for bypass surgery later in their disease and for different reasons than men, and this later referral may significantly increase their chance for operative death. Despite a higher operative mortality in women, many studies have shown no difference in long-term outcome between male and female surgical survivors (Killen et al 1982, Loop et al 1983, Tyras et al 1978, Eaker et al 1989). Unfortunately the number of women in CASS, the only randomized trial to include women, was too few to draw meaningful conclusions about the effectiveness of CABS in this sex. There is, however, evidence from less rigorous studies that they have less symptom relief after surgery than men (Bolooki et al 1975, Douglas et al 1981). It would be useful to know whether men and women treated medically for CABS also show this difference, but such data do not appear to be available. Overall the evidence indicates that while male and female surgical survivors show no difference in long-term survival, women face an increased operative risk and receive less symptomatic relief than men. There is, however, no evidence to show that CABS is more effective than medical treatment in women with CAD.  78  RISKS  Operative Mortality: The major risk for patients undergoing CABS is operative mortality. As has been shown above, operative mortality differs among different conditions and across time. Chassin et al (1986) used meta-analysis to calculate average mortality rates for two time periods. Before 1973 the average operative mortality rate was 5.1; between 1973 and 1981 it was 2.0. Results from the Cleveland Clinic show an operative mortality of 0.8 for 7,105 patients operated on between 1980 and 1982. Gersh (1989) reports similar results for the mid-to-late eighties from other large registries but these rates are considerably lower than those reported in many observational studies. It is clear, however, that operative mortality has decreased substantially over the years despite the worse pre-operative condition of presentday patients coming to surgery. Patient selection likely accounts for much of the difference seen in mortality rates for the same era. A number of clinical, anatomic, and physiologic patient conditions and operative factors have been shown to be risk factors for operative mortality, although only a few factors have consistently been demonstrated as important predictors. Moreover, the importance of some predictors appears to have changed over time. Christakis et al (1989) found that between 1982 and 1986, urgency of surgery, age and reoperation became more significant predictors of mortality with respect to time, while female sex, LVEF and left main disease became less significant. Cosgrove et al, in an analysis of CABS at the Cleveland Clinic between 1970 and 1982, found that over time, incomplete revascularization emerged as a new factor, and impaired LVF became less significant. Operative mortality rates for factors predictive of operative mortality at the Cleveland Clinic and Toronto General Hospital are shown in table 10. In reading this table it is important to compare rates, with and without factors, within  79 institutions only. The figures are not comparable between institutions because of differences in defining the factors and different methods of calculating the rates. Parsonnet, Dean and Bernstein (1989) devised an additive model of 14 risk factors to stratify patients into levels of predicted operative mortality following CABS. They included patients receiving other procedures along with CABS so the results may not be directly applicable to those having an isolated coronary bypass. The greatest weight (10-50) in the model was assigned to catastrophic states (e.g., cardiogenic shock); this was followed by age over 80 (weight 20), age 75 to 79 (weight 12), and second operation, dialysis dependency, and emergency surgery (all with weight 10). The correlation coefficient of anticipated and observed operative mortality was 0.99.  Predictive Factor  TABLE 10 PREDICTORS OF OPERATIVE MORTALITY Toronto 1982-86** Operative Mortality  Cleveland 1980-82* Operative Mortality Without  LVEF <20  1.6  0.7  11.7  2.5  Age >70  2.2  0.9  7.4  1.8  Urgent/Emerg. Surgery  4.9  0.7  8.0  1.9  Female sex  1.9  0.6  6.0  3.0  Reoperation  -  -  9.0  3.2  Left main CAD  0.8  0.8  5.8  3.2  Incomplete revasculariz. CHF  1.3  0.6  7.3  0.7  -  IMA graft  -  -  4.0  * Cosgrove, Loop and Sheldon (1982) ** Christakis et al (1989)  With factor  Without  With factor  -  -  2.8  80  Operative mortality rates have also been shown to be related to the institution where the surgery is performed. Data from the CASS registry showed a 22-fold difference in operative mortality between institutions (range 0.3 to 6.6 percent). After controlling for risk factors for operative mortality, this increased to a 31.5 fold difference between the institutions with the highest and lowest rates. Other investigators have found that hospital mortality is lower in hospitals which perform more than 200 CABS procedures per year (Luft, Bunker and Enthoven 1979). Showstack et al (1987) found that in California in 1983, higher volume hospitals (more than 100 procedures per year) had a lower in-hospital mortality (3.5 vs 5.3 percent in low-volume hospitals) adjusted for case mix. This association of volume with mortality was seen to the greatest extent in patients having 'non-scheduled' CABS. Higher-volume hospitals also had shorter LOS and fewer patients with stays over 15 days. The authors conclude that the greatest improvement in average outcomes of CABS would come from closure of lowvolume surgical units. A recent study from New York state, found that physician volume was also a factor (Hannan et al 1989). Physicians performing more than 116 procedures per year had risk adjusted hospital mortality rates 22 percent lower than surgeons with lower volumes. Low-volume surgeons in high-volume hospitals had lower adjusted mortality rates than high-volume surgeons in low volume hospitals 26 . These results do not indicate whether high-volume surgeons have a lower mortality due to greater experience with the procedure or whether low-mortality surgeons attract more referrals. The above studies obtained their data from discharge abstracts and, therefore, measured hospital mortality rather than 30-day "operative" mortality. This finding likely reflects the influence that other health-care staff in the hospital (e.g., anaesthetists, pump technicians, nurses, physiotherapists) have on hospital mortality rates. 26  81  Also the measures of risk would not be as accurate as risk factors measured prospectively or taken retrospectively from the chart. Even so, the fact that several studies have found an association between institutional volume for CABS and mortality rate, lends credence to the result. Morbidity:  Other potentially serious, and not uncommon, complications of CABS are perioperative MI, stroke, and neuro-psychological complications. Kirklin et al (1991) indicate that perioperative MI in patients with chronic stable angina has decreased over the years from about five to eight percent in the 1970's to about 2.5 percent now, and they attribute this change to improved methods of myocardial management. The rates appear to be higher in unstable angina, ranging from 3.8 to 17 percent (Kaiser et al 1989). Perioperative MI appears to be associated with operative mortality but most studies find no association with long-term mortality (Chassin et al 1986). Few studies report the incidence of post-operative stroke. Goldman et al (1987) found a perioperative stroke rate of 1.3 percent for those under age 70 compared to 4.4 percent for those over 70. Mullany (1990), reporting on a series of those aged 80 or over, found a 2.5 percent rate of perioperative stroke. Neither study defined their interpretation of stroke. Kirklin et al (1990) report that the prevalence is about 0.5 percent in relatively young patients, and about 5 and 8 percent in patients over 70 and 75 years of age respectively. Subtle neurological defects have been reported in up to 75 percent of patients at 8 days post-CABS but only 10 to 30 percent of patients still exhibit them at 3 months. These defects may not be apparent unless the patient is specifically tested for them, and most patients are not handicapped by them (Kirklin et al 1991).  82  The overall incidence of perioperative morbidity after CABS is, again, not often reported. Parsonnet et al (1988) found a 45 percent morbidity rate for patients having elective CABS between 1980 and 1986. Christakis et al (1989) found an overall morbidity of 10.9 percent in his series of 7,334 patients. Morbidity was higher for those undergoing urgent surgery for unstable angina (19.3 percent versus 6.9 percent), for those with LVEF under 20 percent (20.4 percent), those over 60 years of age (15.1 percent) and those undergoing repeat CABS (16.2 percent). While not all post-operative complications will have long-term effects on the patient, it is clear that CABS carries a significant risk of long-term disability over and above the risk of death. Also, given the relatively large numbers of patients who are receiving the procedure, the effect of even a 10 percent morbidity rate could have a major impact on cardiac surgery centres in terms of costs and productivity. The demand for rehabilitation services and home support services, among others, will also be increased by post-CABS patients who suffer complications.  Other Risks: The risks of waiting for cardiac surgery were determined by Rachlis, Olak and Naylor (1991) using outcomes from the major RCTs. In most subgroups the maximum vital risk was less than 0.33 percent per month of delay; a risk about one-seventh the risk of elective CABS itself. In patients with left main disease, unstable angina or LAD stenosis and left ventricular dysfunction, the risk rose to a maximum of 1.05 percent per month. Rachlis and colleagues notes that while older age may be associated with greater risk for medical care, this must be weighed against the markedly increased mortality from CABS. The risk of not undergoing surgery after being advised to do so, does not appear to have been studied as such. However, Graboys et al (1987), describe the  83  outcomes for 91 consecutive patients referred for a second opinion following cardiologists' recommendation for CABS. Following application of guidelines for deciding on a medical or surgical option, medical therapy was recommended for 74 patients. Of the 60 patients who did not have surgery, none died during a mean follow-up time of 28 months (range 3 to 63 months) but two had a MI (Figure 1). These patients were treated by an aggressive integrated treatment involving risk factor modification, an exercise program and attention to psychological issues. It is difficult to tell from this uncontrolled study whether the outcomes were a result of the aggressive treatment provided. However, Graboys estimated that, according to the indications for surgery that derive from the RCTs on CABS, only nine percent of the referred patients actually required surgery. Conservative medical management was as appropriate, or more appropriate, than surgery for 90 percent of these patients who were told by their original cardiologists that they required surgery. Medically treated CAD patients undergoing non-cardiac operation appear to have a higher risk of operative mortality than those who have had prior CABS. A CASS Registry study found operative mortality for non-cardiac major surgery, between 1978 and 1981, in patients without significant CAD to be 0.5 percent; in those with significant CAD who had undergone prior CABS, mortality was 0.9 percent. In CAD patients who had not received CABS, mortality was 2.4 percent. The authors recommend that patients with significant CAD receive prophylactic CABS before major non-cardiac surgery. However, they appear to be ignoring the fact that the operative mortality for CABS was itself around 2-3 percent during the years under study. Consequently patients undergoing CABS purely for prophylactic reasons prior to other surgery would likely be at increased risk of overall mortality rather than less.  84 FIGURE 1 OUTCOME FOR PATIENTS REFERRED FOR SECOND OPINION  Patients Referred 88 I Treatment Recommendation  Medicine^ Surgery 74^  14  Medicine^Surgery^Medicine^Surgery I^I^I^I 60^14^3^11  / \ / \ / \^/ \  Died MI^Died MI^Died MI^Died MI I^I^I^I ^I^I^I 0^2^2^0^1^2^1^2 Data from Graboys et al 1987.  COSTS A number of Canadian studies have examined the costs of CABS. The most recent (Krueger, Goncalves, Caruth and Hayden 1991) examined costs for triple or quadruple bypass surgery at Vancouver General Hospital in January 1989. The mean inpatient cost, including professional fees and fully allocated hospital costs (but not depreciation costs) was $14,329 (range $10,982-$33,676). Earlier studies by Keon, Menzies and Lay (1983) and Laupacis et al (1985) found costs of $9,595 and $14,958 respectively. Both these earlier studies had younger patients than the Vancouver study (55.3 and 58 years compared to 63.4 years) and fewer grafted  85  vessels. Both Krueger and Keon and colleagues found that the age of the patient  and the number of grafted vessels have an impact on the costs of CABS. The problem with knowing the cost of a procedure is that it tells us nothing about value for money. Do the benefits justify the cost? Coles and Coles (1982), using a sample of 332 Ontario patients who received CABS between 1972 and 1977 found that, based on the reduction in hospitalization expenses in the years following surgery, the period for amortization of CABS expenses was 22.3 months for those with left ventricular dysfunction, 38.4 months with normal LVF and 35.9 months overall. The study likely underestimated the costs of CABS because the Ontario per diem rate was used to calculate costs; this proxy would not capture the high operating room charges for CABS, which Krueger et al found to be 33 percent of the total cost. Moreover, Coles and Coles sample was relatively young (mean 52.3 years, range 27-72 years). Given that current CABS patients are older and sicker the specific and overall costs of the procedure are likely to be higher. A 1982 U.S. study on the cost-effectiveness of CABS in 55 year old males used data from the literature on costs and outcomes with medical and surgical treatment (Weinstein and Stason 1982). The costs (in 1982 dollars) per quality adjusted life year (QALY) are shown below. It appears that CABS is excellent value for left main disease and three-vessel disease even with mild to moderate symptoms, and a reasonable value for two-vessel disease with severe angina uncontrolled by medical treatment. The costs in 1991 dollars would be much higher than these, as would the costs for older or sicker patients.  86 TABLE 11 COST EFFECTIVENESS OF CABS AS A FUNCTION OF THE SEVERITY OF ANGINA Extent of disease 1 vessel 2 vessel 3 vessel Left Main Disease  Cost Per Quality Adjusted Life Year Mild Angina Severe Angina $470,000 $30,000 $47,000 $17,500 $7,200 $$7,500 $3,500 $3,800  Data from Weinstein and Stason 1982.  CONCLUSION When compared to medical therapy, CABS has been shown to be effective in prolonging survival in male patients under 65 years of age with CSA and left ventricular dysfunction, in left main disease, and in patients with proximal stenosis of the LAD. There is some evidence to show that it is also more effective than medical treatment in three vessel disease, especially in the presence of severe angina. In addition CABS unquestionably provides greater symptomatic relief than medical therapy, though this difference decreases over time. In patients with unstable angina a long-term survival advantage has been shown with CABS for male patients under 65 years of age with three-vessel disease and left ventricular dysfunction. In addition the short-term relief of angina is better with surgery. In general, there is no advantage from immediate surgery but patients should be monitored and surgery provided if their condition deteriorates. The efficacy of CABS has not been proven in any other clinical condition although  there is initial evidence that it may provide a survival advantage in evolving MI. Given that the proven efficacy of CABS is limited to a relatively few clinical conditions, it is disturbing to find in the literature that the procedure is widely used for conditions in which its efficacy is not known. Many observational studies appear to take the efficacy of CABS in their patient population as a given, even though these populations may not have been tested in a randomized trial or at least in a controlled trial. Many of these studies involve large numbers of patients;  87  it is, therefore, likely that many institutions would have sufficient patient populations to enable them to carry out a RCT. Observational studies, while providing information on a wider patient population than found in a RCT, essentially provide outcome information on an uncontrolled, possibly highly selected group of patients and, therefore, cannot show efficacy and their results cannot be generalized. The increase in CABS among the elderly is of especial concern, both because of the numbers involved and because of the higher risks run by the elderly. Operative mortality and morbidity are significantly higher in the over-65 population with increasing risk associated with increasing age. The finding from the RCTs that those at higher risk receive greater comparative benefit has not been tested in the over-65 population and thus cannot be assumed to apply to them. In addition the finding from several studies that the proportion of patients with severe angina symptoms decreases over time, raises questions about whether CABS may be appropriate treatment for symptom control in the elderly. Finally, the use of CABS in the elderly, especially in the over eighties, raises questions about the appropriate use of scarce resources. Sadly lacking in the literature is mention of the effect of CAD risk factors on the outcomes of treatment. Compliance with smoking cessation, dietary restrictions and exercise regimes have been relatively low in those studies that report them, but there is little information on whether compliance leads to a better outcome. This lack of information is disturbing because it appears that emphasis is not being given to risk factor reduction, and life-style changes, in cardiac surgery programmes. CAD is, after all, a life-style related disease and CABS is simply a palliative procedure. It is not a cure; the disease still exists, progresses, and may even be exacerbated by the very procedure that is designed to prolong life and reduce symptoms. Intuitively, it seems likely that risk factor reduction, especially  88  through smoking cessation, weight reduction and exercise regimen, may reduce the incidence of negative outcomes of CAD in both medical and surgical patients, but data are needed to confirm this. The reoperation rate for CABS has been increasing from approximately 9 to 11 percent in the early eighties to about 17 percent today 27 . Very little seems to be known about these patients and how they compare to patients who continue to function well post-operatively. In addition, little is known about the outcome of patients who are considered for reoperation and turned down. What the literature does tell us is that the risks of reoperation are considerably higher than those for primary CABS, and it is likely that the costs are higher too. Even if the incidence of reoperation should remain the same, the numbers will increase because of the increase in the overall incidence of CABS. It appears likely that the incidence of CABS will continue to rise as more patients return for second or subsequent reoperation. In a 1983 editorial, Braunwald predicted that in the future: "this operation [CABS] should and increasingly will be restricted to patients in whom intensive medical therapy has failed or in whom improved survival after surgery has been unambiguously demonstrated, rather than as a panacea for coronary artery disease" Braunwald evidently had faith that his fellow physicians would be guided by the scientific process, rather than by the results from non-randomized studies and their own intuitive beliefs about the efficacy of CABS. It may well be that these beliefs are well-founded and that CABS does indeed positively influence survival in all the conditions for which it is presently being performed. However, in the absence of unambiguous, or at least reasonable, evidence to the contrary, the possibility that CABS may have no effect on, or even decrease, survival and  27 The reasons for this do not appear to have been investigated. It may be a result of surgeons becoming more comfortable in performing re-operations.  89  increase morbidity to the extent that it negatively effects quality of life, must also be considered. In fact, if the principle of "Do no harm" is to be upheld, the possibility that CABS may contribute to a negative outcome in certain subsets of patients should be considered equally as likely as the present apparent belief that it is the "panacea of coronary artery disease".  90  REFERENCES Alderman, E.L. et al, 1990. Ten-year follow-up of survival and myocardial infarction in the randomized coronary artery study. Circulation 82 (November) :1629-1646. Anderson, G.M. and J. Lomas, 1989. Regionalization of coronary bypass surgery: Effects on access. Medical Care 27:288-296. Athanasuleas, C.L. et al, 1987. A reappraisal of surgical intervention for acute myocardial infarction. T Thorac Cardiovasc Surg 93:405-414. Berry, R.E. et al, 1981. Coronary artery bypass operation in septuagenarians. Ann Thorac Surg 31 (April):310-313. Bolooki, H. et al, 1975. Results of direct coronary artery surgery in women. I Thorac Cardiovasc Surg 69:271-277. Bonow, Robert 0. et al, 1984. Exercise-induced ischemia in mildly symptomatic patients with coronary-artery disease and preserved left ventricular function. N Engl T Med 311 (November):1339-1345. Booth, David C. et al, 1991. Quality of life after bypass surgery for unstable angina: 5-year follow-up results of a Veterans Affairs cooperative study. Circulation 83 (January):87-95. Booth, D.C. et al, 1991. Quality of life after bypass surgery for unstable angina: 5-year follow-up results of a Veteran's Affairs Cooperative Study. Circulation 83 (january):87-95. Braunwald, E., 1983. Effects of coronary-artery bypass grafting on survival: implications of the randomized Coronary-Artery Surgery Study. N Engl T Med 309 (November):1181-1184. Califf, R.M. et al, 1989. The evolution of medical and surgical therapy for coronary artery disease: A 15-year perspective. TAMA 261:2077-2086. CASS Principle Investigators and their Associates, 1983. Coronary Artery Surgery Study (CASS): a randomized trial of coronary artery bypass surgery. Survival data. Circulation 68 (November):939-950. CASS Principle Investigators and their Associates, 1981. The National Heart, Lung and Blood Institute Coronary Artery Surgery Study (CASS). Circulation 63 (June) Suppl II-1-4-41. Chassin, M.R. et al, 1986. Indications for selected medical and surgical procedures A literature review and ratings of appropriateness: coronary artery bypass  91  surgery. Document R-3204/2-CWF/HF/HCFA/PMT/RWJ. Santa Monica, CA: Rand Corporation. Christakis, George C. et al., 1989. The changing pattern of coronary artery bypass surgery. Circulation 80 (September) Supplement I: I-151-1-161. Cohn, Peter F., 1990. Prevalence and significance of asymptomatic coronary artery disease. Chapter 7 in Silent Myocardial ischemia and Angina: Prevalence, Prognostic and Therapeutic Significance, ed. B. N. Singh. New York: Pergamon Press:85-92. Coles, J.G. and J.C. Coles, 1982. The cost effectiveness of myocardial revascularization. Can J Surg 25 (March):123-6. Cosgrove D., F. Loop and W. Sheldon, 1982. Results of myocardial revascularization: A 12-year experience. Circulation 65:37-43. Cosgrove D., F. Loop and B. Lytle et al, 1984. Primary myocardial revascularization. J Thorac Cardiovasc Surg 88:673-684. Cosgrove D.M. et al, 1986. Predictors of reoperation after myocardial revascularization. T Thorac Cardiovasc Surg 92:811-821. Cromwell, Jerry, Janet B. Mitchell, and William B. Stason, 1990. Learning by doing in CABG surgery. Medical Care 28 (January):6-18. Deedwania, P.C., 1988. "Transient myocardial ischemia in survivors of acute myocardial infarction." In B.N. Singh, ed Silent Myocardial Iscemia and Angina. New York: Pergamon Press:157-173. DeWood, M.A. et al, 1983. Acute myocardial infarction: A decade of experience with surgical reperfusion in 701 patients. Circulation 68 (Suppl II):II-8--II-15. Douglas, J.S. et al, 1981. Reduced efficacy of coronary bypass surgery in women. Circulation 64 Supplement 2:1I-11--II-16. Eaker, E.D. et al, 1989. Comparison of the long-term postsurgical survival of women and men in the Coronary Artery Surgery Study (CASS). Am Heart L 117:71-81. Edmunds, H.L., et al, 1988. Open-heart surgery in octogenarians. N Engl T Med 319 (July):131-136. European Coronary Surgery Study Group, 1979. Coronary artery bypass surgery in stable angina pectoris: survival at two years. Lancet (April):879-893.  92  European Coronary Surgery Study Group, 1980. Prospective randomised study of coronary artery bypass surgery in stable angina pectoris. Lance t (September) :491-495. Feinleib, Manning et al., 1989. Coronary heart disease and related procedures. Circulation 79 Supplement LI-13--118. Ford, Earl et al., 1989. Coronary arteriography and coronary artery bypass surgery among whites and other racial groups relative to hospital-based incidence rates for coronary artery disease: Findings from NEDS. AJPH 79 (April):437439. Foster, Eric D. et al, 1984. Comparison of operative mortality and morbidity for initial and repeat coronary artery bypass grafting:The Coronary Artery Surgery Study (CASS) Registry experience. Ann Thorac Surg 38 (December): 563-570. Gardner, Timothy J., 1989. The risk of coronary bypass surgery for patients with postinfarction angina. Circulation 79 Supplement 1:1-79--I80. Gersh, B.J. et al, 1983. Long-term (5 year) results of coronary bypass surgery in patients 65 years old or older: a report from the Coronary Artery Surgery Study. Circulation 68 (September)(suppl 11):11-190--II-199. Gersh, B.J. et al, 1985. Comparison of coronary bypass surgery and medical therapy in patients 65 years of age or older: a non-randomized study from the Coronary Artery Surgery Study (CASS) Registry. N Engl T Med 313 (July):217-224. Gersh, B.J. et al, 1989. Coronary bypass surgery in chronic stable angina. Circulation 79 Suppl I:I-46--I-59. Gerstenblith, G. et al, 1982. Nifedipine in unstable angina: a double-blind randomized trial. N Engl T Med 306:885-889. Gillum, Richard F., 1987. Coronary artery bypass surgery and coronary angiography in the United States, 1979-1983. American Heart Journal 113 (May):1255-1260 Goldman, R.S. et al, 1987. Coronary artery bypass graft surgery in the elderly. Unpublished paper. Graboys, T.B. et al, 1987. Results of a second-opinion program for coronary artery bypass graft surgery. TAMA 258 (September):1611-1614. Grondin, Claude M. et al, 1990. Reoperations for coronary artery disease. In Advances in Cardiac Surgery, ed Robert B. Karp et al, Chicago: Year Book Medical Publishers, Inc: 93-110.  93  Hammermeister, K.E., T.A. DeRouen and H.T.Dodge, 1980. Effect of coronary surgery on survival in asymptomatic and minimally symptomatic patients. Circulation 62 (August) Suppl I:1-98--I-102. Hannan, E. et al, 1989. Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals. TAMA 262:503-510. Hlatky, Mark D.B., et al. 1987. Exercise treadmill score for predicting prognosis in coronary artery disease. Ann Intern Med 106:793-800. Hochberg, M.S. et al, 1984. Timing of coronary revascularization after acute myocardial infarction. T Thorac Cardiovasc Surg 88:914-921. Huddleston, C.B. et al, 1986. Internal mammary artery grafts: technical factors influencing patency. Ann Thorac Surg 42:543-549. Josephson, Martin A., 1990. Percutaneous transluminal angioplasty and myocardial revascularization in the reduction of silent myocardial ischemia in coronary artery disease. Chapter 25 in Silent Myocardial Ischemia and Angina: Prevalence, Prognostic and Therapeutic Significance, ed. B.N. Singh. New York:Pergamon Press:284-291. Kahn, S.S et al, 1990. Increased mortality of women in coronary artery bypass surgery: evidence for referral bias. Ann Intern Med 112:561-7. Kannel, W.B., 1990. Coronary Artery Surgery Study revisited: limitation of the intent-to-treat principle. Circulation 82 (November): 1859-1862. Keon, W.J., S.C. Menzies and C.M. Lay, 1985. Cost of coronary artery bypass surgery: A pilot study. Can T Surg 28:283-286. Killen, D.A. et al, 1982. Coronary artery bypass in women: long-term survival. Ann Thorac Surg 34:559-563. Killip, T., 1988. Twenty years of coronary bypass surgery. N Engl T Med 319 (August):366-368. Kirklin, John et al., 1991. ACC/ AHA guidelines and indications for coronary artery bypass graft surgery. Circulation 83 (March):1125-1173. Koshal, A. et al, 1988. Urgent surgical reperfusion in acute evolving myocardial infarction. A randomized controlled study. Circulation 78 (suppl I):1-171-1178. Kroncke, George M. et al., 1988. Five-year changes in coronary arteries of medical and surgical patients of the Veterans Administration randomized study of bypass surgery. Circulation 78 (September) Supplement 1:1-144--I-150.  94  Krueger, H., J.L. Goncalves, F.M. Caruth and R.I. Hayden, 1991. Coronary artery bypass grafting surgery: How much does it cost? Can T Med In press. Laupacis, A. et al, 1990. The cost-effectiveness of routine post-myocardial infarction exercise stress testing. Can J Cardiol 6(4):157-163. Loop, Floyd D. et al., 1986. Influence of the internal-mammary-artery graft on 10year survival and other cardiac events. N.Engl T Med 314 (january):1-6. Loop, F.D. et al, 1983. Coronary artery surgery in women compared with men: analyses of risks and long-term results. T Am Coll Cardiol 1:383-390. Luchi, Robert J. et al, 1987. Comparison of medical and surgical treatment for unstable angina pectoris: Results of a Veterans Administration Cooperative Study. N Engl T Med 316 (April):977-984. Meyer, J. et al, 1975. Coronary artery bypass in patients over 70 years of age: indications and results. Am J Cardiol 36 (September):342-345. Miller, Donald W., 1977. The Practice of Coronary Artery Bypass Surgery. New York: Plenum Publishing Corporation. Mullany, C.J. et al, 1990. Early and late results after isolated coronary bypass surgery in 159 patients aged 80 years and older. Circulation 82 (November) supplement IV:IV-229--IV-236. Murphy M.L. et al, 1977. Treatment of chronic stable angina: a preliminary report of survival data. N Engl T Med 297:621-627. Naunheim, Keith S. et al., 1988. The changing profile of the patient undergoing coronary artery bypass surgery. JACC 11 (March) 494- 498. Naunheim, Keith S. et al, 1988. Coronary bypass for recent infarction:Predictors of mortality. Circulation 78 Supplement I:1-1222--I-128. Naunheim, K.S. et al, 1988. Coronary artery bypass for recent infarction: predictors of mortality. Circulation 78 Suppl I:1-122--I-128. Naylor, C.D. and S.B. Jaglal, 1990 . Impact of intravenous thrombolysis on shortterm coronary revascualrization rates: a meta-analysis. JA MA 264 (August):697-702. Norris, R.M. et al, 1981. Coronary surgery after recurrent myocardial infarction: progress of a trial comparing surgical and nonsurgical management for asymptomatic patients with advanced coronary disease. Circulation 63 (April):785-92.  95  Parisi, Alfred F. et al, 1989. Medical compared with surgical management of unstable angina: 5-year mortality and morbidity in the Veterans Administration Study. Circulation 80 (November):1176-1189. Parsonnet, Victor, David Dean and Alan Bernstein, 1989. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation 79 (June) Supplement I:1-3--I-12. Peters, Steve et al., 1990. Coronary artery bypass surgery in Canada. Health Reports 2(1):9-26. Preston, Thomas A., 1989. Assessment of coronary artery bypass surgery and percutaneous transluminal angioplasty. Intl J of Technology Assessment in Health Care 5:431-442. Rachlis, M.M., J. Olak, and C.D. Naylor, 1991. The risk of waiting for coronary revascularization. Working paper of the Sunnybrook Clinical Epidemiology Unit. Richardson, J.V. and Cyrus, R.J., 1984. Elective coronary artery bypass in the elderly: experience in a community hospital. Southern Medical Journal 77 (January):30-32. Roos, L.L. and S.M Cageorge, 1987. Innovation, Centralization and Growth: Coronary Artery Bypass Graft Surgery in Manitoba. Winnipeg:University of Manitoba, Faculty of Management. Russell, Richard et al, 1980. Surgical versus medical therapy for treatment of unstable angina: changes in work status and family income. Am T Cardiol 45 (January):134-140. Russell, R.O. et al, 1978. Unstable angina pectoris: National Cooperative Study Group to compare surgical and medical therapy: II. In-hospital experience and initial follow-up results in patients with one, two and three vessel disease. Am T Cardiol 42 (November):839-848. Russell, R.O. et al, 1976. Unstable angina pectoris: National Cooperative Study Group to compare surgical and medical therapy: I. Report of protocol and patient population. Am T Cardiol 37 (May):896-902. Russell, R.O. et al, 1981. Unstable angina pectoris: National Cooperative Study Group to compare medical and surgical therapy. IV. Results in patients with left anterior descending coronary artery disease. Am T Cardiol 48 (September):517-24. Ryan,  T., 1990. Revascularization for acute myocardial infarction: Strategies in need of revision. Circulation 82 (September) Suppl 11:11-110--II-116.  96  Scott, Stewart M. et al, 1988. Veterans Administration Cooperative Study for treatment of patients with unstable angina: results for patients with abnormal left ventricular function. Circulation 78 (September) Supplement I:I-113--I-121. Showstack, J.A. et al, 1987. Association of volume with outcome of coronary artery bypass graft surgery: scheduled vs nonscheduled operations. TAMA 257 (February):785-789. Solymoss, B.C. et al, 1988. Late thrombosis of saphenous vein coronary bypass related to risk factors. Circulation 78(supplement 2):140-143. Takero, T. et al, 1982. The Veterans Administration Cooperative Study of stable angina: current status. Circulation 65 (June) Suppl II:II-60--II-67. Takero, T. et al, 1983. "Results of the VA randomized study of medical and surgical management of angina pectoris". In Coronary Surgery Bypass: The Late Results ed Hammermeister, K.E., New York:Praeger Publishers:17-38. Tobin, J.N. et al, 1987. Sex bias in considering coronary bypass surgery. Ann Intern Med 107:19-25. Tyras, D.H. et al, 1978. Myocardial revascularization in women. Ann Thorac Surg 25:449-453. Varnauskas, E. and the European Coronary Surgery Study Group, 1988. Twelve year follow-up of survival in the randomized European Coronary Surgery Study. N Engl T Med 319 (August):332-337. Veterans Administration Cooperative Group for the Study of Surgery for Coronary Arterial Occlusive Disease, 1977. Veterans Administration Cooperative Study of Surgery for coronary arterial occlusive disease: III. Methods and baseline characteristics, including experience with medical treatment. Am J Cardiol 40 (August):212-224 Veterans Administration Coronary Artery Bypass Surgery Cooperative Study Group, 1984. Eleven year survival in the Veterans Administration randomized trial of coronary artery bypass surgery for stable angina. N Engl T Med 311 (November):1333-1339. Weiner D.A. et al, 1987. Value of exercise testing in determining the risk classification and the response to coronary artery bypass surgery in threevessel coronary artery disease: A report from the Coronary Artery Surgery Study (CASS) Registry. Am J Cardiol 60:262-266. Weinstein, M.C. and W.B.Stason, 1982. Cost-effectiveness of coronary artery bypass surgery. Circulation 66 Suppl  97  Yeaton, W.H. and P.M. Wortman, 1985. The evaluation of coronary artery bypass surgery using data synthesis techniques. Intl I of Technology Assessment in Health Care 1:125-140.  98  CHAPTER TWO CORONARY REVASCULARIZATION TECHNIQUES PART II PERCUTANEOUS TRANSLUMINAL CORONARY ANGIOPLASTY  HISTORY Percutaneous transluminal coronary angioplasty (PTCA) was first performed in 1977 by Andreas Gruntzig who had built on the work of Dotter, Judkins and others who pioneered the recanalization of obstructed arteries by passing catheters across the obstructions. Data on trends in the dissemination of PTCA are difficult to find but data from the National Heart, Lung and Blood Institute Registry, started in 1979, indicate that approximately 400 new cases were registered in 1979, 1500 new cases in 1980 and 3000 new cases in 1981. Prior to March 1980 less than 30 institutions had enrolled in the registry; by September 1981 there were 105 contributing institutions. The steady upward rise in new cases in, and after, 1980 probably reflects the response to the U.S. Food and Drug Administration release of the Gruntzig balloon catheter for marketing in March 1980 (Mullin, Passamani and Mock 1984). Ryan et al (1988) indicate that in the U.S. an estimated 32,300 angioplasties were performed in 1983, rising to 133,000 in 1986. In Canada, there were 5,600 PTCAs in 1986 (Schwartz 1988).  Trends in Assessment: Assessment of PTCA has been even less rigorous than that of CABS. The first randomized trials to compare the efficacy of PTCA with that of CABS in multi-vessel disease are presently underway but have not yet issued their reports. There are no trials comparing PTCA to medical treatment. The three randomized trials to date have compared short-term outcomes in PTCA versus fibrinolytic  99  therapy and have studied the proper-timing of PTCA after fibrinolytic therapy (Preston 1989). Much of the present information on the outcomes of PTCA has come from the National Heart, Lung, and Blood Institute's (NHLBI) registry which was established in 1979. The initial protocol called for patients to have angina uncontrolled by medical therapy, lesions in one vessel that were proximal, accessible, discrete, concentric, segmental and high-grade and to be candidates for CABS. By September 1981, the results confirmed that PTCA in one-vessel disease could be carried out by an experienced operator with a relatively low rate of complications. The registry was then closed to new institutions and only patients with multi-vessel disease were enrolled. The registry was closed completely in 1982 but was re-opened at its previous sites in 1985 in order to document changes in angioplasty strategy and outcome (Detre et al 1988). With the exception of one retrospective matched case-controlled study (Hochberg et al 1989), all other reports on the outcomes of PTCA have been observational studies of case series. In a 1985 editorial Mock et al called for a RCT to provide reliable data on the true efficacy of PTCA, although these authors believed that the "current enthusiasm" for PTCA for single vessel disease would make an RCT for this population of patients "unrealistic". Preston (1988) attributed the lack of controlled studies to assess PTCA, to the lack of a physician group with no self-interest but with the experience necessary for assessment. In the early-70's, cardiologists rather than surgeons were the critics calling for controlled studies. Ironically, this same group are the ones now extolling the virtues of an inadequately assessed procedure!  100  Changes in PTCA: The 14 years since the introduction of PTCA have seen several technological advances in PTCA methodology and equipment. The introduction of catheters with guide wires and increased flexibility and steerability and balloons with lower deflated profiles has improved the immediate success rate (Schwartz 1988). More recently, the introduction of laser angioplasty allows angioplasty to reach lesions that cannot be crossed by a balloon catheter (Cumberland 1987) and may allow the vapourization or melting of coronary atheroma (Robischon 1987). Just as the patient population has changed in CABS over time, so too has the PTCA population changed. Analysis of NHLBI Registry data for 1977 to 1981 and for 1985 to 1986 showed that during the later period patients were significantly older (mean age 57.7 versus 53.5 years) and that there were significantly more patients over 65 years of age or with unstable angina, previous infarct, previous CABS, a low ejection fraction, or a history of CHF, diabetes or hypertension. The percentage of those patients who currently were smokers was, however, moderately but significantly decreased (30 percent versus 37 percent). In addition, there was a trend for patients to have more vessels affected by CAD, but this was not significant. Overall, patients coming for PTCA in the mid-eighties were sicker than those in the late seventies and early eighties (Detre et al 1988). King and Talley (1989), examining changes in revascularization therapy in two institutions between 1981 and 1986, found that in 1981, 11 percent of revascularization patients received PTCA compared to 44 percent in 1986. The incidence of multi-vessel disease among these patients increased from 11 percent in 1981 to 40 percent in 1986. Despite the worsening health of the PTCA population over the years, inhospital outcomes tended to be better. In the NHLBI registry, angiographic success rate per lesion increased from 67 to 88 percent between 1977-1981 and 1985-1986  101  and overall in-hospital success (reduction of at least 20 percent in all lesions attempted without death, MI or CABS) increased from 61 to 78 percent. In-hospital mortality and MI rates changed minimally, falling from 1.2 to 1.0 and 4.9 to 4.3 respectively.  MECHANISM OF ACTION The exact mechanism that causes enlargement of the arterial lumen during PTCA is not known. Suggested mechanisms include compression and redistribution of the plaque, aneurysm formation, and disruption of the plaque and arterial wall from overstretching of the wall. Embolization of plaque material is believed to be minor and neither a mechanism for enlarging the lumen nor a significant hazard during the procedure. It is likely that more than one of the mechanisms suggested above are involved in lumen enlargement during PTCA and that arterial size, characteristics of the plaque, balloon size, and the amount and duration of pressure applied, all contribute to successful angioplasty (Bresnahan 1987a).  EFFICACY OF PTCA There are no data to determine the efficacy of PTCA in comparison to medical treatment and few which help to determine the efficacy of the procedure in comparison to CABS. Consequently the discussion below will focus on the outcomes of PTCA in various clinical and angiographic conditions. In general, comparisons with outcomes in CABS for similar conditions will not be made unless the investigators in the study under discussion have made such a comparison. It is likely that differences between the CABS and PTCA populations would make such comparisons invalid unless statistical adjustment has been carried out. Also, as will be seen below, there is a difference in the way that  102  outcome incidence rates are calculated for PTCA and CABS which would also make comparison invalid. In many studies reviewed here the definition of "initial success" in enlarging the lumen, is a reduction of 20 percent or more of the stenosis. Therefore reduction of a 95 percent stenosis to 75 percent is viewed as being "successful", although clinically the patient would still be regarded as having a significant lesion. Another way in which the PTCA literature differs from the CABS literature is that the denominator for long-term outcome is generally not the original cohort who received the procedure but rather the patients in that cohort who realized initial success. The result is that in the long-term, PTCA appears to be more successful than it really is. A further difficulty in interpreting the literature is that many of the patient populations in different studies, overlap. The largest number of studies come from the data in the NHLBI Registry. This was a multi-institution registry and several of the institutions contributing data to the registry have also published reports of their own experience with PTCA. Where the dates of these studies overlap with the dates of the new or old NHLIB Registry, there is likely to be a patient overlap. There are, unfortunately, few U.S. studies from institutions which did not contribute data to the registry  PTCA in Single Vessel Disease: Although patients with single-vessel disease were shown in the RCTs comparing CABS to medical treatment to have a good prognosis and a better outcome with medical treatment, this group is the one for which PTCA is generally advocated. Initially the NHLBI PTCA Registry guidelines restricted PTCA to patients with single-vessel disease in an effort to maximize success and minimize complications during the era when operator experience was low.  103 Outcomes for patients in the Registry have been shown to differ by time and by operator experience. Total initial clinical success rates (defined as reduction of stenosis by at least 20 percent with no MI, death or emergency CABS) rose from 55.7 percent in and before 1979 to 65.7 percent in 1981. In the same time period, operators who had performed less than 50 procedures had a success rate of 55 percent compared to 77 percent for those with more than 150 proceduresl. Some of the difference by time may also be due to the introduction of new technology. For example, the introduction of the low profile catheter in 1981 likely accounts for some of the improvement in initial success for that year (Kelsey et al 1984). For single vessel disease prior to 1982, the registry initial clinical success rate was 63.6 percent; this figure rose to 84.3 percent between 1985 and 1986 (Detre et al 1988) For the same time periods, in-hospital mortality for single-vessel disease fell from 1.3 percent to 0.2 percent, non-fatal MIs from 5.0 to 3.5 percent, emergency CABS from 6.1 to 2.9 percent and elective CABS from 19.5 to 1.7 percent. Increased operative risk has been shown for PTCA in patients who have had previous bypass surgery and those with left main disease. Although these figures have not been released specifically for single vessel disease, the NHLBI Registry found an operative mortality of 0.9 percent for patients without these conditions, compared to eight percent for those with left main disease and 4.2 percent for those with previous CABS (Dorros et al 1983). Acute coronary closure following PTCA- was found to be more likely in women and in patients with stenoses that were longer, at a bend or branch point, in vessels with other stenoses  1 The assumption here is that increased operator experience leads directly to a better outcome. This pattern of outcomes could also be due to referral filter bias in which high-risk cases tend to be referred to certain operators who, because of specialization, perform more high risk cases than those who are referred the lower-risk cases. Outcome may be due to the patients' risk status rather than to the number of cases the operator has performed. This explanation appears unlikely in this case because the majority of patients in the PTCA Registry during the early years were low risk and because of the large number of operators involved in the analysis.  104  and in multi-vessel disease. There were in addition a number of procedural factors that were related to closure (Ellis et al 1988). Restenosis occurs in a substantial number of patients following an initial successful procedure. Overall restenosis rates range from 20 percent to 50 percent with the highest rates reported for patients with variant angina or for dilation of vein grafts. Comparison of restenosis rates between institutions is complicated by differing definitions for restenosis and by different follow-up periods. The majority of restenosis occur within eight months although some may develop after 12 months. Overall for single vessel disease, restenosis appears to occur in about one third of patients in whom successful dilatation was achieved (Ischinger 1986). About 75 percent of patients with restenosis will experience angina. (Holmes and Sugrue 1987). Holmes and Sugrue (1987) report on a number of studies that examined factors associated with risk of restenosis. Increased risk was found in men and those with variant angina, eccentric, calcified or severe irregular lesions, stenosis of the LAD, or insulin-dependent diabetes mellitus. Lower risk was found for females and for patients who had intimal dissection and a final trans-stenotic pressure gradient less than 15 mm mercury after angioplasty. Repeat angioplasty for restenosis has a higher success rate and lower complication rate than initial procedures. Meirer et al (1984) found a higher success rate (97 percent versus 85 percent, p<0.001) and lower complication rate (8 percent versus 15 percent, p<0.01) in 95 patients undergoing repeat angioplasty for restenosis compared to patients undergoing initial angioplasty in the same time period. Twenty-five percent of the 95 patients had a second restenosis. Similar improved outcomes were found in NHLBI registry patients undergoing repeat angioplasty for restenosis (Williams et al 1984). For these 203 patients repeat angioplasty was carried out a mean of 147 days (median 126 days) following the  105  first procedure. Thirty-four percent of the 173 patients in whom repeat angioplasty was successful, suffered a second recurrence. Long-term outcome of PTCA has not been extensively studied. The 169 patients who received the procedure from Gruentig prior to 1980 constitute the group with the longest follow-up; 40 percent of these patients had multi-vessel disease. By 1987 90 patients remained asymptomatic and there were only 5 cardiac deaths. Repeat angioplasty had been required in 27 patients and CABS in 19 patients. Up to five year follow-up in the NHLBI registry indicated that 70 percent of patients receiving initial successful angioplasty were pain free at 4 years. After hospital discharge the annual mortality rate was one percent per year and the MI rate was 2 percent (Kent et al 1984). Mabin et al (1985) reports up to 3 year (mean 14 months) follow-up on 229 patients who underwent PTCA, for symptoms unresponsive to medical therapy, at the Mayo clinic between 1979 and 1982. There was only one hospital death, and PTCA was successful in 153 patients overall (67 percent), but in only 61 percent of patients with single vessel disease. All successful patients with single vessel disease had complete revascularization. Of the 76 patients who did not have initial success, 59 (78 percent) underwent prompt CABS (surgical group) and 17 (22 percent) continued on medical therapy (medical group). By the most recent follow-up 90 percent of patients in the successful PTCA and surgical groups were subjectively improved; 74 percent of successful PTCA patients were asymptomatic compared to 85 percent of the surgical group and 88 percent of the medical group. However 35 percent of the medical group had required repeat PTCA or CABS compared to 22 percent of the successful PTCA group and none of the surgical group. There were no deaths in any group and seven patients (3 percent) had MI, five of these were in the successful PTCA group. Unfortunately we are not told what percent of each group comprised those with single vessel disease. However,  106  we are told that the unsuccessful patients as a whole were older (mean 56 years versus 54 years), had fewer patients with unstable or variant angina, and had a higher percentage of patients with single vessel disease (74 versus 57 percent) than the successful group. None of these differences were significant. These data appear to indicate that CABS, even after unsuccessful PTCA, is more effective in reducing symptoms and the need for further intervention than successful PTCA. The low short- and long-term serious outcomes are impressive and may be an indication of the low-risk status of this series. The major question that arises from this study is whether the medical group would have experienced the same outcomes if they had not had an attempted PTCA. Hochberg et al (1989) performed a controlled retrospective study of 125 consecutive angioplasty patients matched for single or double vessel disease with 125 CABS patients; 87 patients in each group had single-vessel disease. All procedures were performed at one institution in 1984, prior to the introduction of the steerable catheter at that institution. Patients were stated to be low-risk although the range of ejection fractions was 22 to 67 percent (54 + 11 percent) for angioplasty patients and 13 to 70 percent (mean 49 + 12 percent) for CABS patients. This was the only baseline variable which was significantly different between the two groups. Patients requiring an emergency intervention or those having had a recent MI (within 6 weeks) or thrombolysis were excluded. Results can be seen in Table 12. The authors conclude that, if follow-up is maintained for long enough, surgical therapy for single or double vessel disease may prove to be more effective than angioplsty. There were however, some problems with the statistical methods used in this study. Firstly, although the statistical method used to analyse followup data is not stated, it appears to be either chi-squared or Fisher's exact test rather than a method such as acturial event-free rates which would be more appropriate for time related data. Secondly, that authors considered that PTCA patients who  107  received CABS had "unsuccessful angioplasties" and removed them from the numerator for 3-year functional classification. The denominator remained as the number of patients originally assigned to angioplasty. When PTCA patients receiving CABS are included in the numerator the difference in functional classification is no longer significant (Atkins 1989). TABLE 12 COMPARISON OF OUTCOMES IN CONTROLLED TRIAL OF PTCA VERSUS CABS Outcome^ Length of stay Initial success hospital mortality perioperative MI 3-Year follow-up overall mortality late MI NYHA class I or II Repeat procedure PICA CABS  PTCA^CABS (n=125)^(n=125) 4.8 + 3.1 12.1 + 4.2 * % 88 3 4  % 99 1 5  7 6  2.5 3 92*  63** 18 19  2 2  *Significant at or below 0.00001 level ** Does not include patients in Class I or II following bypass; with these patients included, result is 78 percent - the difference between CABS and PTCA is then no longer significant.  There have been a number of reports of significant left main disease arising following PTCA of the left coronary system. The incidence in the literature ranges from 0.2 to 1.7 percent of angioplasties to the left system. Of the 18 patients reported in the literature, six had had no angiographic left main lesion prior to PTCA. The other 12 patients all had lesions that were less than or equal to 40 percent stenosis and which progressed rapidly, between six weeks and 14 months, post-PTCA. Vardhan et al (1991) postulate that trauma to the left main artery during passage of the guidewire or balloon is partly responsible . They conclude that the possibility of progression of left main disease should be considered prior to  108  selection of LAD or circumflex artery PTCA and that left main disease should be considered in any patient with recurring symptoms after angioplasty to these arteries. In summary, there are results which indicate that, over time, CABS is more effective for single vessel disease than PTCA. Given that the RCTs comparing CABS to medical treatment found that the latter was as effective, or more so, than surgery for single vessel disease, then it appears that, in general, patients with single vessel disease should not be submitted to a surgical intervention but should be treated medically. Of course patients with single vessel disease are a heterogenous group and some, those with left main disease for example, may be at high risk. Even so, the reports that angioplasty of the left coronary system may exacerbate left main disease means that these patients may have a better outcome with CABS. Unfortunately, it is unlikely that single-vessel PTCA versus medical therapy will ever be tested in an RCT. The low incidence of death in this group means that an extremely large number of patients would have to be randomized in order to gain sufficient power to show a difference in mortality. It might, however, be possible to gain some indication of the efficacy of PTCA in single-vessel disease by means of a meta-analysis should sufficient numbers of investigators a) conduct studies comparing PTCA and medical treatment in matched groups and b) publish their results in sufficient detail to enable a meta-analysis to be undertaken.  Multi-Vessel Disease: As operator experience and technical innovation have increased in PICA, the indications for its application have widened, with the result that more patients with multi-vessel disease are having the procedure Detre et al (1988) found an increase in patients with multi-vesel disease (53 versus 25 percent) in the 1985-86  109  (new) compared to the 1977-81 (old) NHLBI Registry. Baseline data and outcomes are shown in Table 13. The small and non-significant increases in mortality for both two and three vessel disease likely result from the worsening pre-procedural condition, and the increasing age, of the patients over time. Table 13 COMPARISON OF OLD- AND NEW NHLBI REGISTRY CHARACTERISTICS AND OUTCOMES BY EXTENT OF DISEASE  N Attempts Multi-lesion Left main Bypass graft Angio success per lesion per patient Clinical success Outcome ** operative death nonfatal MI emergency CABS elective CABS  One-vessel New Old 836 839 % %  Two-vessel Old New 568 203 % %  Three-vessel. Old New 89 395 % %  Total Old 1155 %  New 1802 %  1.0  21.6 1.0  17.7 4.4 5.4  53.2 1.3 3.7  21.3 10.1 20.2  59.2 1.8 14.9  8.5 1.6 3.3  39.8 0.8 4.9  68.6 67.3 63.6  89.0 86.8 84.3  60.7 55.2 51.2  86.4 78.9 74.6  66.1 60.7 58.4  88.0 77.5 70.9  66.8 64.7 61.0  87.3 82.2 78.3  1.3 5.0 6.1 19.9  0.2* 3.5 2.9* 1.7*  0.5 3.9 5.4 27.6  0.9 5.1 3.7 2.3 *  2.2 6.7 3.4 16.9  2.8 5.1 4.3 3.3*  1.2 4.9 5.8 20.7  1.0 4.3 3.4 2.2  5.0  Adapted from Detre et al 1988 *significant at or below 0.05 level **Outcome during hospitalization for PTCA  Vlietstra (1987) states that determination of whether to use PTCA is more complex in patients with multi-vessel disease. Factors to consider are: the way the multi-vessel disease is defined, the degree of revascularization that can be achieved, the amount of myocardium at risk and the development of a dilatation strategy. Definitions of multi-vessel disease vary from study to study. At the Mayo Clinic it is defined as a stenosis of 70 percent or more in at least one major coronary vessel and a stenosis of 50 percent or more in at least one other coronary vessel. In addition they have sub-divided multi-vessel disease into four types  110  depending on the number of vessels with proximal severe stenoses and whether any vessels are completely occluded.  The degree of revascularization that can be achieved is related to the subtype of multi-vessel disease. If only one vessel has a 70 percent or greater stenosis, a single-dilatation may achieve complete revascularization, whereas in subtypes with one or two occluded vessels complete revascularization may not be possible.  The amount of myocardium at risk may be much greater in patients with multi-vessel disease since the vessel being dilated may be supplying blood (either directly or through collaterals) to other diseased vessels. Consequently occlusion of the vessel being dilated may have the potential for jeopardizing a large amount of myocardium, resulting in severe hemodynamic and clinical effects.  A dialatation strategy needs to be developed for each patient, depending on the clinical importance of each stenosis. The two possible strategies are either dilation of all stenoses or dilatation of all functional stenoses. The worst stenosis is usually dilated first In other studies on patients having PTCA between 1979 and 1986, clinical success has ranged from 74-95 percent, operative mortality from 0-1.4 percent, perioperative MI from 2.5-6.9 percent and emergency CABS from 2.1 to 6.9 percent (Cowley et al 1985, Dorros, Lewin and Janke 1987, Talley et al 1988). Long term follow-up in these studies ranged from one to five years. At one year Cowley et al, looking at a subset of patients with initially successful angioplasty and at least a year follow-up, found the incidence of complications to be as follows: MI 2.3 percent, restenosis 34 percent, CABS 18 percent, and repeat PTCA 9 percent. Mortality was not reported. Event-free survival at one year was 64 percent and 46 percent of patients survived with no events and no symptoms. Data from Talley et al (1988), one of the few studies to examine long-term outcome in the whole cohort of PTCA recipients, shows 5-year event-free  111  survival 2 for patients with multi-vessel disease to be 72.4 percent. The postdischarge incidence of MI in these patients was 8.6 percent, mortality 5.1 percent, CABS 15.5 percent and repeat PTCA also 15.5 percent. The overall probability of 5year survival was 94.8 percent while that of survival with no CABS post-discharge was 84.5 percent. The five-year reported figures are somewhat misleading since they do not include initial outcomes. There were no in-hospital deaths but when other in-hospital events are include in the five-year outcomes the incidence of MI is 11 percent and that of CABS is 24 percent. Talley's data also showed a trend for worse initial and long-term outcomes in those with initial PTCA failure. These results will be discussed later. Restenosis is theoretically more likely for patients receiving multiple dilatations because of the greater number of lesions that could restenose. Vandermael et al (1987) studying restenosis following multiple dilatations, found restenosis at one dilatation site in 33 percent of 129 patients undergoing follow-up angioplasty, and at more than one dilatation site in an additional 19 percent of patients. Eighty-two percent of syptomatic patients had restenosis compared to 30 percent of asymptomatic patients. The restenosis rate per lesion was 29 percent. These authors report that the restenosis rate for multi-lesion angioplasty reported in the literature ranges from 26-68 percent but that rates are likely biased upwards by the greater number of symptomatic patients studied by angiography. Another concern with angioplasty for multi-vessel disease, is that all significant stenoses may not be dilated. Incomplete revascularization has shown to be associated with poorer long-term outcome following CABS (Jones et al 1983). Reeder et al (1988) investigated the role of revasculasrization on outcome in a study of 286 NHLBI Registry patients with multi-vessel disease and prior  2 Event-free survival was defined as no cardiac death, CABS or MI.  112  successful angioplasty who were followed-up for a mean of 26.2 months. Initial analysis showed that mortality, MI and CABS rates were all higher in the group who had incomplete revascularization. When differences were made for baseline differences between the two groups however, estimates of the risk of death, MI or presence of angina did not differ between the groups. The group with complete revascularization had more repeat PTCA procedures while those incompletely revascularized had more CABS during follow-up. From these results it appears that outcomes of PTCA at two years of follow-up are not affected by the incomplete revascularization. PTCA in Unstable Angina:  The NHLBI Registry documented an increase in the proportion of patients undergoing PTCA for treatment of unstable angina in the new versus the old registry (Detre et al 1988). Comparison of registry patients with stable and unstable angina showed no difference in immediate success rate; for successful patients there was no difference in the combined MI and mortality rate both in-hospital and at the 18-month follow-up. Other observational studies of patients with unstable angina found a rate of angiographic success from 84-93 percent, operative mortality from 0.2-0.9 percent, perioperative MI from 6.6-10.8 percent and late death from 1.7-2.6 percent (Myler et al 1990, Feyter et al 1985, Faxon et al 1983). Myler and colleagues found that outcomes were significantly worse when angioplasty was performed within one week of the onset of angina than when two or more weeks had elapsed before the procedure. The complication rate following PTCA for total occlusion has been shown to be significantly higher in patients with unstable angina compared to those with stable angina (Plante et al 1991). Complications occurred only among those with angina at rest or pre-infarction angina. The authors speculate that the presence of  113  intraluminal thrombus may increase the risk of acute vessel closure or of embolization.  PTCA in Left Ventricular Dysfunction: Serota et al (1991) studied 73 consecutive patients with low LVEF (range 1440 percent, mean 34 percent) who received PTCA between 1983 and 1989. Clinical success was achieved in 88 percent of patients, in-hospital mortality was 5 percent and the MI rate was 3.9 percent. Estimated survival, for successful patients, from one to four years was 79 percent, 74 percent, 66 percent and 57 percent respectively. Predictors of cardiac mortality were congestive heart failure and low ejection fraction. These survival figures appear low when compared to the five year outcomes for the medical group in the VA trial 3 but the patients in Serotoa et al's study were sicker than those in the VA trial.  Post-Myocardial Infarction: The efficacy of PTCA versus thrombolysis and the optimum timing of PTCA following MI has been tested in a number of RCTs and controlled trials which are reviewed in a chapter by Pitt (1990). The strategy of urgent PTCA following successful perfusion with thrombolyis was tested by the Thrombolysis and Angioplasty in Acute Myocardial Infarction (TAMI) trial, the European Cooperative Study Group and the TIMI-IIA trial. All these trials found a significantly increased risk, plus no demonstrated improvement in ventricular function, following urgent PTCA after thrombolysis. The strategy of delayed PTCA following thrombolysis was studied by the TIMI-IIB investigators who randomly  3 Survival for the medical group in the VA trial at five year follow-up was 73 percent for all patients with low ejection fraction and 66 percent for those with low ejection fraction and three-vessel disease. For the CABS group the corresponding survival rates were 80 and 83 percent.  114  assigned patients to an invasive strategy (angiography at 18 to 72 hours postthrombolysis with PTCA in vessels with significant residual stenosis) or to a noninvasive strategy in which patients received PTCA only if symptomatic. No benefit regarding LVF or survival was found in the group randomized to the invasive strategy. Pitt also reports on a small trial which randomized 56 patients to either PTCA without prior thrombolytic therapy or to thrombolysis. Reperfusion rates were similar but the PTCA group had significantly less residual stenosis in the infarct-related artery, better LVEF, less post-infarction angina and less exerciseinduced ischemia on stress testing. Numbers were too small to show the relative effects of these two strategies on survival. O'Keefe et al (1989) also report on 500 consecutive patients with MI who received PICA without antecedent thrombolysis. Successful angioplasty was achieved in 94 percent of patients but 15 percent of these reoccluded before discharge. The three strongest predictors of the 7.2 percent hospital mortality were cardiogenic shock, multi-vessel disease and failed angioplasty. It appears that PTCA following thrombolysis is not more effective and may carry a higher risk than thrombolysis alone. The strategy of direct PICA without thrombolysis appears promising but has not been adequately tested to assure efficacy.  Gender: Data from the 1977-1982 NHLBI Registry showed that PICA in women was associated with a significantly lower clinical success rate (56.6 versus 62.3 percent), and significantly higher hospital mortality (1.8 versus 0.7), in-hospital elective CABS (23.5 versus 17.6 percent) and post-emergency surgery mortality (17.4 versus 3.2 percent) rates than in men. Long-term results however were comparable or  115  better in women than in men. After 18-month mean follow-up, women with initially successful PTCA had lower incidence of mortality (0.3 versus 2.2), restenosis and additional revascularization procedures, but had comparable symptomatic improvement and higher event-free survival (79.7 versus 69.0 percent) than men. Although a higher percentage of women in the Registry had unstable angina, class III or IV angina and were older than men, the males had a higher incidence of multi-vessel disease, impaired LVF and prior bypass surgery. No analysis was performed that would show if the women's higher hospital mortality was due to higher risk (Cowley et al 1985b).  The Elderly: Because CABS results in a significantly higher morbidity and prolonged hospitalization among the over 65-year age group, PTCA would appear to be a better alternative for the elderly. Mock et al (1984), in an analysis of the NHLBI 1977-1982 Registry data, compared outcomes in 370 patients over the age of 65 and 2,709 patients under 65 years. The older group had a significantly greater proportion of females, patients with prior CABS, low ejection fraction and severe angina, but a significantly smaller proportion of patients with previous MI. The mean overall clinical success rate was lower in the elderly group (53 percent versus 62 percent) but hospital mortality (2.2 versus 0.7 percent) and elective CABS (25.4 versus 8.1 percent) were significantly higher. For the whole cohort, in-hospital mortality compares favourably with that reported for CABS in the CASS Registry over-65 population (2.2 versus 5.2 percent), but differences in patient selection may account for this difference, Mean length of stay was slightly greater for the elderly group. At one year follow-up the over-65's had significantly  116  higher mortality and CABS rate, but a lower incidence of repeat PTCA 4 . When one-year outcomes for only the successful patients in each group were analyzed, only the late PTCA rate was still significant; the other outcomes were remarkable similar between the two groups. This is one of the two studies reviewed here which gives sufficient information to allow the incidence of outcomes to be calculated for the unsuccessful PTCA recipients. These calculations (see Table 14) show that the mortality rates for the unsuccessful patients compared to the successful patients, are 2.5 times higher for the under-65's and almost four times higher for the elderly patients. MI and CABS rates are both approximately six times higher in the unsuccessful patients than they were in those, of whatever age, with initial success. Amazingly, over 75 percent of unsuccessful patients had had CABS by the first year follow-up; this included emergency and elective CABS both during the initial hospitalization and follow-up. Equally amazing, and extremely disquieting, is the fact that these results were not even mentioned by Mock and his colleagues. Although the incidence of CABS and MI were approximately the same for the under- and over-65's with failed PTCA, the overall effect on the older PTCA cohort was greater since almost 50 percent of them had initial angioplasty failure. These results are from PTCA's carried out prior to 1982; initial success rates in the elderly are likely higher today and complication rates are likely lower, although there appear to be no data to show this. Whether the incidence of complications seen in the unsuccessful patients will have changed is uncertain since these patients are rarely addressed in the literature.  is not clear whether this finding is due to an age bias in recommending treatment or whether the clinical condition of the older patients warrented the more invasive intervention. The higher mortality in the older group would support the second explanation but that mortality may have been due, in part, to the higher incidence of surgery. 4 It  117 TABLE 14 COMPARISON OF ONE-YEAR OUTCOMES IN ELDERLY AND NON-ELDERLY PATIENTS AFTER SUCCESSFUL AND UNSUCCESSFUL PTCA Outcome N Death MI CABS PTCA  All 2709 % 2.2 7.8 36.1 10.8  Under 65 years Success 1680 % 1.4 2.8 11.6 14.8  Failure 1029 % 3.7 15.9 76.1 4.3  All 370 % 4.5 8.3 41.5 6.4  65 years and over Failure Success 174 196 % % 7.4 1.9 2.5 14.9 11.4 75.2 2.9 9.9  Data from Mock et al 1984.  In summary, PTCA performed in the 1977-1982 'early era' had a significantly higher failure rate, operative and one-year mortality rates, and elective CABS rate for those over 65-years of age compared to younger patients. The markedly worse outcomes for patients with unsuccessful PTCA combined with the almost 50 percent failure rate in the elderly may indicate that PTCA is not the most efficacious and/or cost-effective treatment in the over-65 population. Welldesigned controlled studies are required to show efficacy. Unsuccessful PTCA:  The data shown above from Mock et al (1984) seems to indicate that patients who experience an immediately unsuccessful PTCA tend to have worse one-year outcomes than those in whom PTCA is initially successful. The other study reviewed here, in which the data showed worse outcomes in the initially unsuccessful PTCA patients was that of Talley et al (1988) who reported on 427 patients who underwent PTCA in 1981. Eighty-nine (20.8 percent) of the patients had clinically unsuccessful PTCA. For these patients the incidence of in-hospital outcomes were MI 25.8 percent and CABS 46 percent; there were no untoward outcomes in the successful patients. At five year follow-up the  118  incidence of adverse outcomes overall for both successful and unsuccessful patients respectively were: mortality 2.9 versus 5.6 percent; cardiac death 0.9 versus 4.4 percent; M.I. 5.6 versus 31.4 percent; CABS 12.4 versus 69.6 percent. Kent et al (1984) also reported worse long-term outcomes for those with initially unsuccessful PTCA, though they did not give results for these patients. It is not possible to tell from the reported data whether unsuccessful patients were at an initially higher risk than the successful patients, or whether attempted PTCA increased the risk for an adverse outcome. It appears that adverse outcomes are not due only to outcomes from emergency CABS because Kent et al reported worse outcomes in those with unsuccessful PTCA who were managed on medical therapy. Whatever the cause, the very fact that long-term outcomes are worse in these patients puts into question the common practice of reporting long-term outcomes only on patients with initial success 5 . This practice also prevents comparison of the results of angioplasty research with those of CABS research for similar patient populations. These results raise a number of questions about the efficacy of PTCA overall and in certain patient populations. The major question to be answered is whether the 'unsuccessful' patients in the above studies would have had the same outcomes whether or not they had had attempted PTCA. It appears possible that patients who have unsuccessful PTCA may be much worse off than if they had not had the procedure. These patients appear to constitute a high-risk group in angioplasty and, as such, they merit more attention than they are presently getting. It is possible that patients in whom PTCA is not initially successful have some 5 By reporting only on "successful" patients receiving PTCA it would appear that PTCA investigators  are not adhering to the "intent-to- treat" principal that was followed in the analysis of the CABS RCTs; rather they appear to be basing their analysis on successful treatment received. The assumption being made here is that an unsuccessful attempt at angioplasty has no effect on the patient; the above results show that this assumption is not valid.  119  common characteristics that would enable them to be recognised before they receive the procedure. If not, then other factors which contribute to immediate failure need to be investigated and, if possible, remedied. Ultimately, many of the above question can only be answered by a randomized trial or a well designed prospective controlled study comparing medical treatment with angioiplasty. Until such a study is carried out it is imperative that investigators report, for both short-term and long-term outcomes, on the whole cohort of patients receiving angioplasty. Analysis of patients having immediate success and those who are unsuccessful should be carried out for the group as a whole and separately. Analysis for the group as a whole will allow comparison with the results of investigations into the use of CABS or medical treatment in similar patient populations, and will give a more realistic estimate of the outcomes to be expected after PTCA. Separate analyses for successful and unsuccessful patients will allow comparison of outcomes between these two groups and may help to identify patients in whom failure is more likely to occur as well as factors which may contribute to the worse outcomes in the unsuccessful patients. Full reporting of results of angioplasty investigations will facilitate metaanalysis, which may be used to supplement the results from randomized trials.  Summary: To date, the efficacy of PTCA has been mainly evaluated using surrogate outcomes, i.e., blood flow in the treated artery immediately following the procedure, but there is no data to show how immediate increased blood flow relates to later morbidity and mortality. There are results which show that urgent or delayed PTCA following thrombolysis is not more effective, and carries an increased risk, than thrombolysis alone. In addition, results from one retrospective controlled trial indicate that CABS may result in a better long-term  120  outcome than PTCA, and results from the only two studies reporting data for patients having immediate PTCA failure indicate that these patients may have a markedly worse outcome than patients with initial success. RISKS Angioplasty carries a risk of significant and potentially fatal complications. Results on the 3079 patients from the 1977-1982 NHLBI Registry, reported by Dorros and Cowley (1986), show that a total of 1180 complications occurred in 652 patients (21 percent). These authors subdivided complications into two groups, acute coronary events and non-coronary events. The latter group include hospital death despite the fact that death often resulted from a coronary event. Eight hundred and thirty acute coronary events occurred in 418 patients while noncoronary events occurred in 234 patients. The most commonly occurring acute coronary events were prolonged angina, MI, coronary occlusion, coronary dissection and coronary spasm Prolonged Angina was the most frequent complication occurring in 6.8 percent of patients and was associated with major complications (death, MI, and emergency CABS) in over half these patients. Univariate analysis showed that prolonged angina was more likely to occur in patients with unstable angina, class IV angina, with eccentric lesions or those with stenoses greater than 90 percent. Myocardial Infarction occurred in 5.5 percent of patients overall, in 45 percent of patients who had emergency bypass surgery, in 3.6 percent who had elective CABS and in 2.6 percent who did not undergo CABS. Fifty percent of patients suffering an MI did so within 24 hours of PTCA. Death occurred in 9.4 percent of patients who had MI. The primary complications associated with occurrence of MI were coronary dissection, coronary occlusion, prolonged angina and coronary spasm.  121  Several authors have noted the occurrence of subendocardial MI following angioplasty on saphenous vein grafts, but this has not been implicated (and was possibly not tested) as a predictor for MI (Dorros and Cowley 1986). Coronary Dissection occurred in 12.9 percent of 3079 patients undergoing PTCA; of these, 70 percent had no adverse effects and 218 had clinically successful PTCA. Of the 34 percent of patients with a dissection who developed major complications, 15 percent had a non-fatal MI, 23 percent had emergency CABS and 1.5 percent of patients died. Univariate and multivariate analysis showed that coronary dissection was more likely to occur in women, in patients undergoing right artery PTCA, in those with multivessel disease, eccentric lesions or with nondiscrete lesions. Coronary occlusion occurred in 4.9 percent of PICA patients and major complications arose in 81 percent of these. Forty-one percent had an MI, 72 percent required emergency CABS and 5.3 percent died in hospital. Predictors of coronary occlusion by univariate analysis were onset of angina within 6 months, eccentric lesions, stenoses greater than 90 percent and tubular or non-discrete lesions. Coronary Spasm occurred in 4.2 percent of patients, 32 percent of whom had  a major complication. Non-fatal myocardial infarct arose in 12 percent while 24 percent required emergency CABS and three percent died in hospital. Spasm was more likely to occur in non-calcified lesions and in patients of a younger age. Multivariate analysis showed only younger age to be a significant independent predictor of coronary spasm. Overall, univariate analysis showed that acute coronary events were more frequent in women, in patients with unstable angina, with initial lesion severity  122 of more than 90 percent, with eccentric lesions 6 , nondiscrete or tubular lesions 7 . With multivariate analysis, unstable angina, severe stenosis, nondiscrete lesions and tubular lesions were associated with increased frequency of a coronary event (Cowley et al 1984).  Hospital Mortality: In Dorros and Cowley's study (1986), hospital mortality occurred in 29 patients but nine of these deaths were considered to be unrelated to the PTCA procedure because they occurred during or after CABS. This seems to be unnecessary hair-splitting because presumably the need for CABS arose because of complications from, or unsuccessful, PTCA. Clinical characteristics that had an effect on mortality were female gender (1.8 percent mortality versus 0.7 percent male mortality), age over 60 years (1.7 percent versus 0.7 percent mortality under 60 years), and duration of angina for more than one year (1.7 percent versus 0.5 percent mortality when angina present for less than 6 months). Angiographic characteristics also had an effect on mortality. There was a trend for lesion location to affect mortality with higher mortality in patients with PTCA to circumflex or left main artery lesions. The presence of left main disease was a significant risk factor but dilatation of a left main lesion was not. Thus, left main disease increases the risk of hospital death regardless of the site of dilatation. Factors found to have no influence on hospital mortality rates were lesion severity, history of diabetes, history of elevated cholesterol, unstable angina angina severity, lesion calcification and previous MI. Factors that were predictive of 6 Eccentric lesions are those which are not concentric around the lumen of the artery. 7 Tubular lesions are long concentric lesions. Generally coronary stenoses less than 10mm in length are  most suitable for PTCA (Ischinger 1986).  123  mortality were lesion location, sex, time since onset of angina, previous bypass surgery and number of vessels diseased; R2 figures were not reported. On multivariate analysis however, only female gender was significantly associated with PTCA-related deaths.  Emergency Bypass Surgery: Dorros and Cowley (1986) report that emergency bypass surgery was performed in 6.6 percent of NHLBI PTCA Registry patients; the most frequent problems necessitating this surgery were coronary dissection and coronary occlusion. Patients receiving emergency CABS had a high incidence of MI (45 percent) and mortality (6.4 percent). On univariate analysis, eccentric lesions, nondiscrete lesions and severe lesions were predictive of emergency CABS, although only lesion eccentricity was predictive on multivariate analysis. Procedural difficulties, e.g., inability to pass the stenosis or to dilate it once passed, were also associated with increased emergency CABS; possibly these difficulties result from the lesion characteristics that are also associated with increased surgery. Dorros and Cowley (1986) state that in the early days of PTCA emergency surgery was used whenever a significant complication occurred because it was believed that prompt CABS could prevent the evolution of an MI. However, when analysis of the Registry data showed that emergency surgery was associated with a significant mortality and almost half the patients still had an MI, they modified the treatment of myocardial ischemia or occlusion following PTCA. They reduced emergency surgery to 1.3 percent in Milwaukee during 1984, even though these patients had more extensive disease, a higher incidence of previous CABS and a higher incidence of previous MI than was found in the NHLBI patients. It seems likely that most large centres today will have a similar incidence of emergency surgery, although there does not seem to be data to back this up.  124  In summary, PTCA carries significant risks of complications which may lead to MI or death. How these risks compare to the risk of these outcomes during treatment with medical therapy, has not been tested.  Other Factors Affecting Risk: Ryan et al (1988) in their guidelines for PTCA classify lesions into three different types which may be used to estimate the likelihood of successful angioplasty and the likelihood of development of complications. Lesions are classified according to morphology and location. Type A lesions are those in which the anticipated success rate should be 85 percent or greater and the risk of abrupt vessel closure is low. Type B lesions have an anticipated success rate from 60 to 85 percent and a moderate risk of abrupt closure, while type C lesions have an anticipated success rate less than 60 percent and a high risk of abrupt closure. Ryan and colleagues caution that attempts to dilate type C lesions should not be undertaken when they are present in vessels supplying large or moderate areas of viable myocardium. Bresnahan (1987b) notes that risk profiles should also include other risk factors, such as age, gender, and duration of angina and experience of the operator who will perform the PTCA. He states that young men with angina of recent onset and concentric non-calcified lesions of 70 to 90 percent severity are excellent subjects for angioplasty. Older patients, especially women, with long-standing angina and high-grade or calcified stenoses will likely have a much poorer outcome. He also states that because current technology limits the present ability to assess the pathologic characteristics of stenoses, steps cannot be taken to avoid coronary dissection or occlusion. It seems unlikely, therefore, that the complication rate will improve before there are significant improvements in coronary angiography.  125  As in CABS, operator experience has been shown to affect the outcome of PTCA. An NHLBI Registry report (Kelsey et al 1984) involving 3000 patients receiving PTCA from 105 centres between 1977 and 1981, showed that the success rate improved as the number of procedures per operator increased although the learning process was relatively long; success rates continued to improve beyond 150 cases. As operator experience increased there was a corresponding decrease in the incidence of elective CABS and in-hospital mortality. The incidence of emergency CABS, however, did not decrease significantly. Bresnahan (1987b) speculates that this may be because coronary dissection and occlusion are independent of operator experience. He also states that the effect of training and technologic improvements in balloon catheters has shortened the learning curve. As a result, "angiographers" today, who are trained by those experienced in angioplasty with steerable balloon catheters, may achieve a 90 percent success rate from the outset. As discussed earlier, it appears that failed angioplasty may carry greater risk of complications and death than successful angioplasty. The lack of coverage of this problem in the literature seems to imply that clinicians do not consider it to be a problem. There does not appear to be any research that looks specifically at either predictors of failure or at optimum treatment of patients following failed angioplasty. Such studies are obviously needed. Finally, the place that angioplasty is carried out may constitute a risk. The guidelines for PTCA developed by Ryan et al (1988) state that the minimal facilities for any hospital where PICA is carried out, should include a surgical operating suite that is equipped to provide cardiac surgery (they failed to mention the need for a cardiac surgeon). In Europe PTCA may be carried out at hospitals without these facilities; patients who require surgery are transferred to a nearby surgical unit. Shaw (1990) suggests that these circumstances have not resulted in clinical  126  deterioration in any patient requiring emergency surgery after PTCA. Parker (1990), giving the surgeons view, states that reports on patients who have been transferred for surgery show an unacceptably long mean time to revascularization. He points out that the earlier the patient comes to surgery the greater the chance that the surgeon will use the internal mammary artery, which is associated with significantly better long-term outcome. He concludes that angioplasty supported by on site surgical facilities is the correct policy.  COSTS Holmes et al (1984) using data from the NHLBI Registry, measured employment status and time of return to employment (representing an indirect societal cost) in 2,250 patients who underwent PTCA before 1981. Patients were divided into three groups: Group A, those with successful PTCA; Group B, those with immediate failure followed by CABS; and Group C, those with immediate failure and medical therapy. At follow-up (mean 1.5 years) the decrease in the percentage of patients who were employed was similar in all three groups. Further analysis was done on a sub-set of 1,150 patients who were working at baseline and were 60 years old or younger. At follow-up 86 percent of group A, 81 percent of group B and 83 percent of group C were employed. However, for group A the mean time to return to work was 7 days, compared to 73 days for group B and 13 days for group C. Reeder (1987) comments that successful PTCA has the potential for allowing a more rapid return to work than the 6 weeks to 3 months disability which generally occurs after CABS. Reeder (1987) in a chapter on costs of PTCA, reports on a number of observational studies which all use different costing methodology. All studies found substantially lower charges for the initial PTCA compared to initial charges for CABS. Two studies which examined expenditures for angioplasty versus  127 CABS patients during the first year of follow-up had different results. Kelly et al (1985), using average hospital charges and mean number of hospital days and services to estimate costs, found one-year expenditures of $7,689 per angioplasty patient and $13,559 per CABS patient. The primary success rate was 74 percent and restenosis occurred in 18 percent of these patients. These authors found that the average hospital stay for patients who had initially unsuccessful angioplasty was twice as long as for those who had CABS 8 ; this obviously increased the average first-year cost for angioplasty patients but the low restenosis rate may have off-set this. The rate of return to work was between 86 and 100 percent and did not differ significantly between patients having primary successful angioplasty, primary successful CABS or failed angioiplasty with subsequent surgery. The time of return to work was, however, significantly less in the successful angioplasty group. Reeder et al (1984), using actual patient costs, found that initial procedural costs were $5,493 for angioplasty (including a small charge for surgical standby for successful angioplasty) and $12,065 for CABS. The primary success rate was 70 percent and restenosis occurred in 33 percent of cases. For PTCA patients who had no restenosis the average first year costs were 56 percent of the CABS costs; for those with restenosis the average cost was $2,700 more than the average CABS cost. In a later report of the same study, Reeder states that the mean first-year expenditures for the treatment of restenosis were $10,002 for medical management, $11,285 for a second angioplasty and $20,421 for those having CABS (Reeder 1987). The above studies comparing PTCA and CABS costs are based on the premise that angioplasty may be substituted for CABS in certain patients, and cost savings realized. In fact the whole reasoning behind the development of PTCA was, presumably, to develop a procedure which would achieve the same effect as 8 A further indication that unsuccessful angioplasty may increase risk of an adverse outcome.  128  CABS but with a lower mortality, morbidity and cost. Reeder (1987) points out that if the use of angioplasty increases to include patients who would not normally be candidates for CABS, the concept of substituting a low-cost procedure for a highcost one no longer obtains. In this situation, revascularization costs would increase rather than decrease. Reeder presents data to show that at the MAYO Clinic, and in the U.S. as a whole, the use of PTCA has increased dramatically without a reduction in the number of bypass procedures. There is, therefore, no evidence that PTCA is substituting for CABS, and the economic argument falls flat; PTCA is not a cost-saver but is adding to the costs of coronary revascularization. It is interesting that no studies have been done which compare the cost of PTCA to that of medical treatment. Given that about 50 percent of PTCAs are performed for single vessel disease, for which CABS is very rarely done, the alternative treatment for comparison should be medical therapy. In summary, although initial PTCA, as a procedure, is substantially cheaper than initial CABS , patients who require further treatment because of initial failure or restenosis may have expenditures in the first year which are only slightly less, or even more, than those for CABS. In addition it appears that PTCA may not be substituting for CABS but may, instead, be additive. Therefore it appears likely that increased use of PTCA in an area will increase, rather than decrease, revascularization costs.  CONCLUSION Given that the efficacy of coronary angioplasty in the treatment of coronary artery disease has never been rigorously evaluated, its continued phenomenal growth over a 14-year span, is disquieting.  129  PTCA is reputed to be a substitute for CABS but is used frequently in singlevessel disease, which is usually managed medically. The coronary artery surgery RCTs showed that medical treatment is as effective as CABS and in fact resulted in better long-term survival. Although this finding was likely due to the operative mortality found with CABS, there is no reason to believe that PTCA would perform better than CABS in comparison to medical treatment. Although the incidence of hospital mortality is relatively low with PTCA, the approximately 30 percent incidence of restenosis means that patients may be subjected to two or more angioplasties. The risk of mortality with each may be low but, for the patient, these risks are additive and may ultimately exceed the risk of CABS. Once restenosis has occurred, the average costs of an initial strategy of PTCA exceed those of initial CABS. With a 30 percent restenosis rate there are likely to be relatively few cost savings with a strategy of initial PTCA. The costs to the patient includes the risks of death, morbidity, and emergency or elective CABS. For patients with failed angioplasty and those requiring emergency surgery, the risks are much higher than those with elective CABS. There appears to be no research identifying the characteristics of those patients in whom PTCA is not successful. In fact, these patients appear to be lost souls in the literature, generally mentioned only as a "rate" and left out of longterm analyses. The implication appears to be that PTCA failure has no effect on these patients, whereas it has been shown that, in at least one study, outcomes at one year were markedly worse. An issue that is rarely mentioned in the literature is that of "self-referral". Because cardiologists who care for patients with CAD are frequently also those who perform angioplasty, it is possible for a patient to be "referred" for PTCA by the same professional who will carry out the procedure. In this case the advantage of the second-opinion, which is an integral part of a referral to another physician,  130  is lost. In addition, there is the potential for a physician to refer patients motivated as much by self-interest as by patient need. Ryan et al (1988) suggest that in situations where the 'referring' cardiologist is also the one who will carry out the PTCA, the responsible physician should arrange for a consulting opinion from another specialist. At the present time, the place for PTCA in the treatment of CAD is unclear. Randomized trials comparing angioplasty to CABS are now underway. However, the efficacy of PICA in comparison to medical treatment, a non-invasive treatment, should have been the first step and needs to be established as soon as possible by means of prospective controlled studies, in which patients are prospectively matched for risk, or by randomized controlled trials.  131  REFERENCES Atkins, C.W. 1989. Discussion following Hochberg, M.S. et al, Coronary angioplasty versus coronary bypass. T Thorac Cardiovasc Surg 97 (April):502503. Bresnahan, D.R., 1987a. Mechanism of action. Chapter 2 in PICA: Percutaneous Transluminal Coronary Angioplasty, ed. R.E. Vlietstra and D.R. Holmes, 19-33. Chicago and London: The University of Chicago Press. Bresnahan, J.F., 1987b. Use in single-vessel disease: Factors that influence immediate success and clinical applications. Chapter 5 in PTC A : Percutaneous Transluminal Coronary Angioplasty, ed. R.E. Vlietstra and D.R. Holmes, 63-72. Chicago and London: The University of Chicago Press. Cowley, M.J. et al, 1984. Acute coronary events associated with percutaneous transluminal coronary angioplasty. Am T Cardiol 53:12c-16C. Cowley, M.J. et al, 1985a. Coronary angioplsty of multiple vessels: short-term outcome and long-term results. Circulation 72 (December):1314-1320. Cowley, M.J. et al, 1985b. Sex differences in early and long-term results of coronary angioplasty in the NHLBI PTCA Registry. Circulation 72 (January):90-97. Cumberland, D.C., 1987. Laser coronary angioplsty. Br T Hosp Med (April):281 Cumberland, D.C., 1987. Laser coronary angioplasty. British Journal of Hospital Medicine (April):281. Cumberland D.C. et al, 1986. Percutanoeus laser-assisted coronary angioplasty. The Lancet (July):214. Detre, K.M. et al, 1984. Baseline characteristics of patients in the National Heart, Lung, and Blood Institute Percutaneous Transluminal Coronary Angioplasty Registry. Am T Cardiol 54 (January):7C-11C. Detre, K. et al, 1988. Percutaneous transluminal coronary angioplasty in 1985-1986 and 1977-1981: The National Heart, Lung and Blood Institute Registry. N Engl T Med 318 (February):265-270. Dorros, G. et al, 1983. Percutaneous transluminal coronary angioplasty: report of complications from the National Heart, Lung and Blood Institute PICA Registry. Circulation 67 (April):723-730.  132  Dorros, G. and M. Cowley, 1986. Complications associated with PTCA. Chapter 13 in Practice of Coronary Angioplasty, ed T. Ischinger, 223-249. Berlin: Springer-Verlag. Dorros, G., R.F. Lewin and L. Janke, 1987. Multiple lesion transluminal coronary angioplasty in single and multivessel coronary artery disease: acute outcome and long-term effect. T Am Coll Cardiol 10 (November):1007-1013. Ellis, S.G. et al, 1988. Angiographic and clinical predictors of acute closure after native vessel coronary angioplasty. Circulation 77 (February):372-379. Faxon, D.P. et al, 1983. Role of percutaneous coronary angioplasty in the treatment of unstable angina. Am J Cardiol 53:131C-135C. Feyter, P.J. et al, 1985. Emergency coronary angioplasty in refractory unstable angina. N Engl T Med 313 (August):342-346. Hochberg, M.S. et al, 1989. Coronary angioplasty versus coronary bypass. T Thorac Cardiovasc Surg 97 (April):496-503. Holmes, D.R. et al, 1984. Return to work after coronary angioplasty: A report from the National Heart, Lung and Blood Institute Percutaneous Transluminal Coronary Angioplasty Registry. Am T Cardiol 53:48C-51C. Holmes, D.R. and D.D. Sugrue, 1987. Restenosis. Chapter 12 in PTCA: Percutaneous Transluminal Coronary Angioplasty, ed. R.E. Vlietstra and D.R. Holmes, 161-187. Chicago and London: The University of Chicago Press. Ischinger, T. and B.Meirer, 1986. Outcome of coronary angioplasty. Chapter 11 in Practice of Coronary Angioplasty ed T. Ischinger, 194-210. Berlin: SpringerVerlag. Jones, E.L. et al, 1983. Importance of complete revascularization in performance of the coronary bypass operation. Am T Cardiol 51:7. Kelly, M.E. et al, 1985. Comparative costs of myocardial revascularization: Percutaneous transluminal angioplasty and coronary artery bypass surgery. T Am Coll Cardiol 5:16. Kelsey, S.F. et al, 1984. Effect of investigator experience on percutaneous transluminal coronary angioplasty. Am T Cardiol 53 (June):56C-64C. Kent, K.M. et al, 1984. Long-term efficacy of percutaneous transluminal coronary angioplasty (PICA): Report from the National Heart, Lung and Blood Institute PTCA Registry. Am J Cardiol 53 (June):27C-31C.  133  King, S.B. and J.D. Talley, 1989. Coronary arteriography and percutaneous transluminal coronary angioplasty: changing patterns of use and results. Circulation 79 (suppl I):I-19-4-23 Meier, B. et al, 1984. Repeat coronary angioplasty (abstract). T Am Coll Cardiol 4:463. Mabin, T.A. et al, 1985. Follow-up clinical results in patients undergoing percutaneous transluminal coronary angioplasty. Circulation 71 (April):754760. Mock, M.B. et al, 1984. Percutaneous transluminal coronary angioplasty in the elderly patient: experience in the National Heart, Lung, and Blood Institute PTCA Registry. Am T Cardiol 53 (June):89C-91C. Mock M.B. et al, 1985. Percutaneous transluminal coronary angioplasty versus coronary artery bypass. Isn't it time for a randomized trial? N Engl T Med 312 (April):916-918. Mullin, S.M., Passamani, E.R. and Mock, M.B., 1984. Historical Background of the National Heart, Lung and Blood Institute Registry for Percutaneous Transluminal Coronary Angioplasty. Am I Cardiol 54 (January):3C-6C Myler, R.K. et al, 1990. Unstable angina and coronary angioplasty. Circulation 82 (Suppl II):II-88--II-95. O'Keefe, J.H. et al, 1989. Early and late results of coronary angioplasty without antecedent thrombolytic therapy for acute myocardial infarction. Am T Cardiol 64 (December):1221-1230. Parker, D.J., 1990. Does angioplasty need on site surgical cover? A surgeon's view. Br Heart T 64:1-2 Pitt, B., 1990.^Percutaneous Transluminal coronary angioplasty in acute myocardial infarction. Chapter 23 in Modern Coronary Care, ed. G.S. Francis and J.S. Alpert, 403-410. Boston: Little, Brown and Company. Plante, S. et al, 1991. Acute complications of percutaneous transluminal coronary angioplasty for total occlusion. American Heart Journal 121 (February) Part 1:417-426. Preston, T.A., 1989. Assessment of coronary bypass surgery and percutaneous transluminal angioplasty. Intl T of Technology in Health Care 5:431-442.  134  Reeder, G. et al, 1984. Is percutaneous coronary angioplasty less expensive than bypass surgery? N Engl I Med 311 (November):1157-1162. Reeder, G.S., 1987. Socioeconomic aspects. Chapter 16 in PTCA: Percutaneous Transluminal Coronary Angioplasty, ed. R.E. Vlietstra and D.R. Holmes, 215-221. Chicago and London: The University of Chicago Press. Reeder, G.S. et al, 1988. Degree of revascularization in patients with multivessel coronary disease: a report from the National Heart, Lung, and Blood Institute Percutaneous Transluminal Coronary Angioplasty Registry. Circulation 77 (March):638-644. Robischon, T., 1987. No ideal replacements yet for bypass, angioplsty. The Medical Post (October):30. Ryan, T.j. et al, 1988. Guidelines for percutaneous transluminal coronary angioplasty. Circulation 78 (August):486-502. Schwartz, L., 1988. Focus on PTCA. Perspectives in Cardiology 4 (June/July):23-43. Shaw, T.R.D., 1990. Does angioplasty need on site surgical cover? A physician's view. Br Heart Journal 64:3-4. Serota, H. et al, 1991. Predictors of cardiac survival after percutaneous transluminal coronary angioplasty in patients with severe left ventricular dysfunction. Am I Cardiol 67 (February):367-372. Talley, J.D. et al, 1988. Clinical outcome 5 years after attempted percutaneous transluminal coronary angioplasty in 427 patients. Circulation 77 (April):820-829. Vandormael, M. et al, 1987. Multilesion coronary angioplasty: clinical and angiographic follow-up. I Am Coll Cardiol 10 (August):246-252. Vandormael, M. et al, 1991. Predictors of long-term cardiac survival in patients with multivesel coronary artery disease undergoing percutaneous transluminal coronary angioplasty. The American journal of Cardiology 67 (January):1-6. Vardhan, I.N., Aharonian, V.J. and Maher, P.R., 1991. A rare complication of perceutaneous transluminal coronary angioplasty - left main disease. Am Heart 1 121 (March):902-905. Williams, D.O. et al, 1984. Efficacy of repeat percutaneous transluminal coronary angioplasty for coronary restenosis. Am I Cardiol 53:32C-35C.  135  CHAPTER 3 REGIONAL VARIATIONS  INTRODUCTION For over 20 years, health utilization researchers have been documenting regional differences in per capita rates of medical service use including hospital utilization and surgical procedure rates. These variations have been found within and between countries and for different sizes of region studied. Many investigators have tried to account for these variations by using multiple regression procedures to identify factors which explain a portion of the variance. Despite a large body of work in this area there is still very little known about how and why such variations occur, whether they are of concern and, if so, what should be done about them. This chapter will briefly describe the results from the literature on procedural variations in general, discuss the few studies that looked at variations in CABS and related procedures and describe some of the methodological problems associated with regional variations studies.  PROCEDURAL VARIATIONS IN GENERAL Most studies in regional variations of medical and surgical procedure rates or hospital use rates use a similar methodology. The number of occurrences of the health care event of interest in the population of a given region, is divided by the population within the region who are at risk for that event, thus creating a regional incidence rate for the event. Ideally, these incidence rates are standardized for age and sex to allow consistent comparison across geographic areas. The region may be defined at various geographic levels but usually represents an area in which the majority of the population use the same medical  136  care resources. Larger areas, such as provinces or states, usually have the best match between population and medical resources but may be so large that variations seen at lower levels of aggregation are lost. Smaller areas, such as counties, are more likely to have a relatively homogenous population and a relatively similar procedure rate within their boundaries, but may be suspect as a basis on which to match population and medical care resources. Early investigations into variations in the utilization of health care services examined variations between countries (Pearson et al 1968, Bunker 1970). These studies did not, however, take into account the differences between countries in recording proceduresl; these differences may have accounted for at least part of the variations found. Later investigations began to look at variations within countries using pre-existing administrative units such as provinces (Mindell, Vayda and Cardillo, 1982) and counties (Stockwell and Vayda 1979). Other investigators have created regions, generally termed hospital service areas, based on population size and service by one hospital where the majority of residents receive their care (Barnes et al 1985, Wennberg and Gittelsohn 1982). More recently, researchers have studied variations between "micro areas", areas within hospital service areas, (Tedeschi, Wolfe and Griffith 1990) and in subgroups of the general population, e.g., medicare recipients (Shwartz et al 1981) blacks (Wilson, Griffith and Tedeschi 1985), and the elderly (Roos et al 1984). Without exception all these investigators have found marked variation in hospital utilization or in procedure rates. Two recent comprehensive reviews have provided an overview of the literature on medical service use variations and of the issues surrounding them. Paul-Shaheen, Clark and Williams (1990) reviewed the North American small-  1 For example, in the United Kingdom primary procedures are reported while in the United States, primary, secondary and tertiary procedures are reported. Theoretically these differences should result in higher observed rates in the U.S., which is what McPherson found.  137  area analysis literature, restricting their review to those studies (59 in all) which used the area as the unit of investigation. Sheps, Scrivens and Gait (1991) restricted their review to procedural variations studies which attempted to assess the effect of specific factors on the variation observed. At the same time they broadened the scope of the review to include investigations outside North America. Both these reviews noted that the magnitude of observed variations within studies were relatively small, in general between one and three fold, rarely exceeding five to six fold, although between studies there were often wide variations which may have been due to methodological differences. Variations tended to be greater for elective procedures than for procedures in which the surgeon had less discretion, for smaller areas than for larger areas such as province or state, and for single years than for multiple years. These two latter findings likely reflect the unstable annual rates which arise in small areas with small populations but may also show real variations which are hidden at the larger level of aggregation. Of interest in many studies was the finding that different procedures may, and usually do, have widely different ranges of variation between regions but that for a given procedure, a high-rate or low-rate area tends to persist as such for several years. Wennberg (1979) termed these persistent patterns "surgical signatures" and attributed them to the practice-pattern of the surgeon serving that area. In another paper, he and Gittelsohn (1982) note that the greatest variations occur for procedures performed for conditions for which there is no consensus in the medical community as to treatment and, therefore, for which the individual physician has considerable discretion as to which treatment he chooses. Support for the belief that physician practice-patterns affect resource use comes from recent research into the financial effect of practice style on hospital resource use. Feinglass, Martin and Sen (1991), controlling for patient health  138  status, used linear and logistic regression to analyse the relationships between physicians clinical decisions and hospital charges and length of stay (LOS). They found that attending physicians were significant predictors of the log of total charges and the log of LOS but not as significant predictors of untransformed total charges. Many of the studies in the variations literature have methodological flaws which may affect their results. These flaws and other methodological issues will be discussed later.  Factors Associated with Procedural Variations: Both Paul-Shaheen, Clark and Williams (1990) and Sheps, Scrivens and Gait (1991) report on the factors identified and tested as variables which may explain a portion of the variation in rates. A large number of these explanatory variables have been investigated but between studies results have been inconsistent and often conflicting. The major factors believed to contribute to geographic variations may be classified under two main categories, supply factors and community factors. Supply factors include such variables as bed supply, physician supply and medical services available to the region. Several studies have extended these variables to include those related to physician characteristics such as age, sex and place and/or year of graduation. Community factors attempt to describe the region in terms of morbidity and mortality, socioeconomic status, or unemployment. For these variables, information about the individuals in the community is aggregated. Intuitively both supply and community variables would seem to affect procedural use rates. Evans (1984) convincingly argues that, in health care, supply induces demand, and that Roemer's Law (a built bed is a full bed) applies, at the aggregate level, to surgeons as well as to beds. When more are available, more are  139  used. Similarly a large sociological literature on indicators of health status shows relationships between morbidity and mortality and socio-economic status, education level, poverty, overcrowding, and race. Hay (1988) using data from the 1978 Canada Health Survey found a direct positive relationship between SES (income. education and occupation) and health status. Of the three SES measures used, income was consistently the best correlate of health status and occupational status was the most inconsistent. In the variations literature, supply factors have been the most frequently studied as explanatory variables, although results were inconsistent both between studies and according to the form of analysis. For example bed supply was tested was tested for its effect on the cholecystectomy rate in 12 studies but was found to be significant only four times (Sheps, Scrivens and Gait 1990). Of the studies looking at community explanatory variables, very few have included mortality or morbidity rates for the conditions for which the procedures are performed. Of the large number of community variables tested across studies, few have been tested more than once or twice and few were found to be significant. Paul-Shaheen, Clark and Williams (1989) note that when correlations between explanatory variables and use rates were run, the most highly correlated variable was some measure of bed supply. When multivariate regression models were used, the majority of studies found that a combination of community and supply variables provided the best explanation of the observed variation. However, for surgical procedure variation the supply of beds, doctors and surgeons explained the greatest amount of variation observed. Sheps, Scrivens and Gait (1990) suggest that the large differences between studies in the magnitude of variation explained by independent variables, results from methodological differences rather than true relationships between the variables.  140  VARIATIONS IN CABS UTILIZATION RATES As has been shown in Chapter Two, it is likely that physicians are using considerable discretion in the treatment of CAD. Although CABS has been found effective only for a few conditions in coronary artery disease, and the efficacy of PTCA has not been evaluated, the rates of both these procedures have risen exponentially sine they were first introduced. Therefore, it is to be expected that both CABS and PTCA rates will show some regional variation. In fact PTCA does not appear at all in the variations literature and CABS, addressed in six papers, shows a two to three-fold regional variation within studies. Whether this amount of variation is acceptable is unclear 2 . Roos and Cageorge (1987) studied the growth and regional variation in CABS in Manitoba between 1978 and 1984, and compared trends in CABS to those in cardiac valve surgery (a low growth procedure performed by the same surgeons who perform CABS) and in total hip and total knee replacements (high growth procedures which are less centralized than CABS). Seven regions, approximating those used by the Manitoba Department of Health for health services planning, were used. Regional mean annual rates for CABS ranged from 2.63 to 5.61 per 10,000 population; areas with high rates for CABS tended to have high rates for valve surgery. This last finding would support Wennberg's concept of the "surgical signature" since the same surgeons perform both operations. The orthopedic procedures showed less variation than the cardiac procedures and the authors attributed this finding to the fact that hip and knee surgery, although centralized, is more decentralized than CABS, thus providing more equal access to  Most regional variation studies appear to assume that any variation between regions is unacceptable even though there may be legitimate (but usually unevaluated) reasons for such variations, e.g., regional variations in morbidity. The range of variation that can occur and still be consonent with high quality care for all patients in a region has never been established. 2  141  all Manitoba residents 3 . Surprisingly there were no trends or consistent relationships between the regional rates of hospitalization for, and death from, acute myocardial infarction (AMI) and the regional CABS rates, although there were some minor regional differences in the AMI rate. Roos and Cageorge conclude that organizational factors, i.e., centralized versus decentralized surgery, and referral networks are critical in the development and maintenance of regional variations. While this is very likely true, this conclusion is an assumption since it was not tested in the study. Despite this shortcoming this study has contributed positively to the variations literature since it is one of the few studies to measure regional morbidity. The absence of any association between AMI and CAD is extremely interesting. Either it indicates that the incidence of CABS is not related to need, or that AMI is not a good measure of the need for CABS. In Chapter Two it was noted that the incidence of CABS is highest in the age group which has the highest incidence of chronic CAD, and that CABS is performed for chronic, rather than acute, ischemic heart disease. It appears likely that ischemic heart disease diagnoses other than AMI may prove a better measure of CAD morbidity. Other studies which have looked at regional variations in CABS rates include one from Ontario. Anderson and Lomas (1989) studied the effects of regionalization on the age-adjusted surgery rates in 38 counties in southern Ontario. These counties were served by eight referral centres in five geographically separate metropolitan areas. There was a three-fold difference in rates between the counties with the highest and lowest rates (9.4 versus 2.8 per 10,000 population over age 20). There was no significant relationship between the  3 One could also argue that decentralization would increase regional variation because more  physicians, and therefore more "practice styles", would be involved.  142 county rates and distance of the county from a referral centre'', but there was a significant relationship between the county rate and the referral centre serving the county 5 . In other words, the referral centres themselves explain more of the variance in county rates than does distance from the centre. In order to exclude other county variables besides distance from the centre it would be useful to know how the counties were distributed geographically around their referral centres. Contiguous counties could share some potentially relevant factors such as ethnicity, income and socio-economic status. Differences between institutions in recommending CABS to patients with similar clinical and angiographic characteristics, were found by Maynard et al (1986) in an analysis of the CASS Registry data. These authors found that across 15 institutions the percentages of patients in the CASS Registry who were recommended for CABS ranged from 35.2 percent to 73.2 percent. Even after adjustment for patient characteristics there was a significant difference between sites. These authors note that in most of the hospitals that they studied, the cardiologist was the dominant decision maker. The only other study which looked at regional variations in CABS was done by Chassin et al (1986) and studied variations in the use of several medical and surgical procedures in the medicare population in a number of selected areas in the U.S. The variations in the rates of CABS between areas ranged from 7 to 13 per  4 This was true whether distance was measured catagorically (i.e., county contains referral  centre, county borders referral centre, county does not border referral centre), or in miles.  5 Counties were assigned to referral centres by three different rules and the analysis was repeated  for each rule. The strict rule assigned to each centre the rates for counties for which the centre supplied at least 90 percent of the procedures done. Under this rule the referral centre explained about three-quarters of the variance. The majority rule assigned the rates for counties to the centre which supplied 50 percent or more of the procedures in the county. Under the plurality rule the county rate was assigned to the centre which provided the plurality of procedures in the county.  143 10,000 population. The generally higher CABS rates found in this study are likely due to the older population studied. Chassin and colleagues have also been instrumental in carrying out research which crosses the boundary between studies into geographic variations and those into quality of care. The methodology consists of using physician panels to rate indicators for CABS as appropriate, inappropriate or equivocal 6 . The subjectivity present in any rating procedure was likely reduced in Chassin's studies by the fact that panels of doctors were used and that the panels were provided with a review of the CABS literature prior to making their ratings. Retrospective samples of patients are then taken from geographic regions where per capita use rates for the procedure have already been determined. Chart review is then carried out and the appropriateness of the procedure is determined for each patient. The relationship between the proportion of appropriate procedures in a region and the region's procedural incidence rate is then determined. Chassin et al (1987) studied the appropriateness of use of coronary angiography and two other procedures provided to Medicare beneficiaries in three geographic regions (covering several states) in the U.S. They found small but statistically significant differences in appropriateness among the sites. For angiography the site with the highest per capita rate had the lowest rate of appropriateness, and the site with the lowest procedural rate had the highest rate of appropriateness. Across all sites inappropriate use accounted for 17 percent of angiographies. The authors concluded that although the differences were in the direction supporting their hypothesis, i.e., that there would be more inappropriate  An indication was deemed appropriate for a procedure if the expected health benefits of the procedure for that indication exceeded the negative consequences by a sufficiently wide margin that the procedure was worth doing. An indication was deemed inappropriate when the risks exceeded the expected benefit. Equivocal ratings were those which fell in the middle of the 9point rating score or those on which there was disagreement by the panel. 6  144  procedures in areas of high use, the differences in appropriateness could not explain the large differences in overall rates. The above study was later repeated in one state using the county as the level of analysis (Leape et al 1990). When all counties were included in the analysis the ratio of high to low use was 12.1 and inappropriate use accounted for 28 percent of the variance in use rates. When the county with the highest use-rate (an outlier since its rate of 189 per 10,000 Medicare enrollees was substantially higher that the next highest rate of 89 per 10,000) was removed from the analysis the high-low use rate ratio was only 2.3 and inappropriate use accounted for only 12 percent of the variance. The authors conclude that differences in the rates of use are not due to more inappropriate use in high-use areas. In fact they rightly state that one of the major findings in the study is that inappropriate use occurs in all areas whatever their use rate. The same group of researchers investigated the appropriateness of CABS in a random sample of patients from three hospitals during 1979, 1980 and 1982 (Winslow et al 1988). Overall 56 percent of CABS were performed for appropriate reasons, 30 percent for equivocal reasons and 14 percent for inappropriate reasons. Across hospitals the range of appropriate use was 37 to 78 percent and that for inappropriate use was 6 to 23 percent. Although there were significant differences in appropriateness scores by age the results showed slightly more appropriate use in the elderly. These authors point out that if procedures in the inappropriate and equivocal class were eliminated it would be possible to double the number of appropriate CABS without raising health-care expenditures. Differences in attitude to treatment of CAD between physicians in the U.K. and the U.S. were explored by Brook et al (1988). Panels of physicians from each country rated indications for coronary angiography and CABS according to their appropriateness. These ratings were then used to rate a retrospective sample of  145  patients who had angiography or CABS in the U.S. Using U.K. ratings, 42 percent and 60 percent of Medicare patients and non-medicare patients who received angiography, were inappropriate. The U.S. ratings found only 17 percent and 27 percent respectively, inappropriate. For CABS patients 13 percent were deemed inappropriate by U.S. ratings and 35 percent by U.K ratings. The authors note that if CABS were performed in the U.S. and the U.K. at the rates judged appropriate by the panelists, then the ratio would be US:UK 1.5:1. In fact the ratio is US:UK 4.8:1. This ratio cannot be accounted for simply by the difference in physicians beliefs and, therefore must have some other explanation. In the above study Brook et al stated that differences in attitude to the indicators appeared to be a result of different interpretations of the literature. Differences between the groups occurred for those indicators for which there was no clear evidence of benefit. The U.S panel rated these as appropriate or equivocal, the U.K. panel as inappropriate or equivocal. U.K panelists also attached great importance to the degree of medical therapy and did not rate indicators as appropriate when the patient was not on maximal medical therapy. There were also differences between panelists according to specialty. Surgeons on both panels rated more indicators appropriate, while general practitioners (G.P.'s) tended to rate more indications inappropriate than the other panelists did. This finding is in conflict with that of Young et al (1987) who used hypothetical patients to analyze U.S. cardiologists and G.P.'s recommendations for angiography. In this study it was found that cardiologists required a higher probability of CAD before recommending invasive testing. Few other explanatory variables have been tested in trying to account for regional variations in CABS rates, and none have been found significant. With the exception of the study by Roos and Cageorge, morbidity has not been tested as an explanatory variable; nor have other community variables such as SES.  146 In summary, the utilization rates of CABS have been shown to vary between geographic regions and hospital sites by at least two-fold. Few explanatory variables have been tested; of those that were, the referral hospital used by the region was found to be significant. More recent attempts to explain geographical variations by correlating regional rates with the percentage of appropriate and inappropriate procedures in the region, have shown a weakly significant relationship which the authors believed was not strong enough to account for the variations observed.  ISSUES IN VARIATIONS RESEARCH All the recent major reviews on regional variations (McPherson 1989, PaulShaheen, Clark and Williams 1989, Sheps, Scrivens and Gait 1990) discuss the issues involved in research into the variation of surgical procedures. McPherson, whose review was restricted to papers documenting international variations, identifies several factors which may give rise to artifactual differences in rates. These sources of artifact include the substitution of day care surgery for inpatient surgery affecting the number of procedures counted, differential procedure coding, differences in protocols regarding whether primary, secondary or tertiary diagnoses or procedures are counted, differences in which procedure is coded as the primary procedure, and the use of a denominator which accurately estimates the population at risk 7 . Sheps, Scrivens and Gait (1990), believe that use of the incorrect denominator is a methodological flaw which is pervasive in the literature. Use of the general population as a denominator will not severely distort the rate estimates in large populations, because the number of individuals  Not all individuals in the population are at risk of receiving certain procedures. For example, when calculating the procedural rate for hysterectomy, the denominator should exclude men and those women who have alrady had a hysterectomy, otherwise the rate will be spuriously low. 7  147 receiving the procedure is relatively small compared to the numbers in the population. In small populations, however, where the relative difference is much larger, rate estimates may be severely distorted and will be inappropriately large. When the procedure under study involves organ removal, use of the general population as a denominator, will produce rate estimates which are too small. Other issues discussed by these authors and by Paul-Shaheen, Clark and Williams, include the need for age standardization of the populations studied, since the rate for some procedures (including CABS) is highly related to age. Agesex standardization is less often performed although many procedures (also including CABS) are related to both age and sex. Early studies by Wennberg, among others, failed to standardize rates, which may account for some of the large variations found. Many studies reported in the above reviews assumed that large variations were statistically significant, and failed to test for this. However, Diehr et al (1990) showed, through computer simulations, that chance variability in procedural rates among populations with the same underlying rate, was surprisingly high. There was more variability for low-incidence surgeries and for populations which were small, had readmissions or in which sub-groups were studied. One problem with small area analysis is that small populations may make rates unstable across time and across areas. A few cases occurring in one year in a small population may result in a high rate which is not representative of that area in other years. Comparison of high and low rates across areas, without due consideration of the degree to which these extreme rates are representative of the data, may show spuriously large variations 8 . This is especially likely in studies which estimate 8 An example of this may be seen in Bayne's (1991) study of health in the Greater Vancouver Regional Hospital District. Hospital caesarean section (C-section) rates (number of C-sections over number of deliveries per hospital) varied just over two-fold when all hospitals were included (range 13 to 27) but only 1.28 fold when an outlier with a low rate was excluded (range 21 to 27).  148  rates for a sub-section of a small population, as Roos et al (1981) did when studying surgical rates for the elderly in rural Manitoba. Diehr et al (1990) discuss the statistical methods generally used in variations research. They suggest that null hypothesis 9 has rarely been tested because there is no information available about the distribution of rates in the null situation. Frequently used statistics, including the extremal quotient, the coefficient of variation and the systematic component of variation have no tables and so cannot be used to test the null hypothesis. Diehr and colleagues suggest that the chisquare statistic may be appropriate to test the null hypothesis, provided that the expected number of surgeries per county is at least five, readmissions are not possible, and the surgery does not have a low incidence. They also note that chisquare may be underused because it does not apply directly to age-sex standardized rates, and suggest that for standardized rates the Mantel-Haenszel approach or logistic regression may be used. All the above exclusions apply to CABS, a surgery of low incidence in which readmissions are possible and which is age-sex related so that age-sex standardized rates should be used. Another approach to determining whether there is excess variability among rates is to regress the observed rates on some relevant covariates, such as the number of surgeons per capita. If all underlying rates are the same, there should be no significant association between the rates and the regressors. In the above paper, Diehr et al note several problems with this method. Firstly, because there is usually only a single data point per region, outliers will tend to have a large influence on the estimated regression coefficients and significance levels. Also because the regressors are generally adjusted for population size (e.g., number of  9 The null hypothesis in variations research , is that the underlying utilization rate is the same  in all areas, (i.e., has a normal distribution with a common mean and standard deviation) and that the observed differences between areas are simply due to random variation.  149  surgeons divided by the population of the region), the resulting variable is correlated with population size. There is, therefore, the strong probability that significant associations are really due to correlation of the variables under investigation with a third variable, population size. Despite these caveats, Diehr and colleagues note that the regression approach is the only approach we have to determine characteristics of counties with high rates. Another factor which may create spurious rate differentials is mobility; the movement of the population out of their area of residence to seek health care in another area or jurisdiction. The smaller the size of area used as the unit of analysis, the more important it is that some adjustment be made for movement in and out of the area for receipt of health care services 10 . Joffe, (1979) found that when he regressed utilization rates, both with and without mobility adjustment, against bed supply and physician supply, adjustment for mobility showed a stronger association between utilization and bed supply but a weaker association between utilization and physician supply. McPherson (1989) noted that differing reporting procedures may lead to rate differentials. Sauter and Hughes (1983) found that even in jurisdictions which report only primary procedures, differences in the protocol for ordering operations within the patient's record may have a significant impact on reported rates. They conclude that surgical utilization statistics, and the inferences that can be drawn from them, vary considerably depending on the recording protocol employed, and suggest that policy decisions based on utilization research should be tempered if recording protocols were not considered in the study. 10 This particularly applies to those studies in utilization variation which use "hospital service areas" (i.e.' geographic areas in which the majority of residents receive care from a single hospital) as the unit of analysis. However, it is also important to consider mobility in any situation in which utilization by some area residents may not be detected because they cross borders for medical care. For example, B.C. patients receiving CABS in Alberta would not show up in B.C. data bases.  150  Sheps, Scrivens and Gait (1990) note that few studies appear to recognize the issue of the ecologic fallacy. This error may arise if the data on procedural use rates and the data used to describe the community come from different sources. Relationships between use rates and the community variables may not be present at the level of the individual. For example if high use rates are seen in areas of high SES, it is not necessarily the people with high SES who have the higher frequency of the procedure. To infer that high SES causes the high use rate would be to make an ecological fallacy. The final issue that needs to be discussed is the issue of the clinical importance of variations in procedural rates, and the size of variation which is cause for concern. Many studies stress that, because there is no normative rate for each procedure, one cannot tell whether low rates are too low or high rates too high (Chassin et al 1986, Wennberg, Freeman and Culp 1987) 11 . It follows, therefore, that one cannot know whether the "correct" rate even lies in the observed high-low range. This uncertainty about whether under-utilization or over-utilization exist, together with the uncertainty about how utilization relates to community variables, makes policy decisions very difficult. The approach used by Winslow et al (1988), Chassin et al (1986) and others in assessing the appropriateness of procedures in individual patients, to some extent gets away from the need for a normative rate. As was shown by Leape et al (1990), inappropriate procedures my occur in low rate areas as well as in high rate areas, although the appropriateness of the procedure did not account for all the 11 Though this is not to say that a correct rate (or range) does not exist. Such a rate for a given population would be one at which there were no inappropriate CABS and at which all individuals for whom the procedure is appropriate (given a condition for which CABS is more effective than medical therapy or angioplasty), would receive it. Such a rate could be estimated for a given population from the incidence of conditions for which CABS is proven to be the most effective treatment, adjusted for the demography of the population. Use of present rates for "appropriate" CABS would likerly underestimate the "correct" rate since those patients who require the procedure but do not receive it would not be included.  151  variation seen. In this case it would seem logical to reduce the number of inappropriate procedures rather than simply trying to reduce rates that are higher than those in surrounding areas . That is not to say that rate variations per se are unimportant. If the rate is not related to the underlying "need" of the population 12 (i.e., if there is no relationship) then clearly something is wrong. Measuring "need" requires good morbidity estimates plus an estimate of the rates of other appropriate therapies for the condition for which the procedure is performed. Both these are generally unavailable. Another issue which is generally ignored in the literature, is that of legitimate unmet need. The individuals who require the procedure but do not receive it, are not addressed in the variations literature. Sheps, Scrivens and Gait (1990) note that variation in unmet need should be an integral part of future research and caution that altering supply factors without concern for appropriateness may do more harm than good. In summary, it appears that there are a number of artifactual and methodological issues which have only recently been recognized in regional variations research, and which may have accounted for some of the large variations, and some of the inconsistencies in explanatory variables, seen in the literature. Future research must include morbidity and/or mortality estimates, and should include consideration of patient mobility and the protocol for ordering operations, appropriate age-sex standardization of rates and testing of the null hypothesis.  12 It is not clear which way the relationship would go. A negative relationship between morbidity and a procedure may indicate that the procedure is either curing the disease in question (high procedure rate and low morbidity) or that not enough procedures are being done (low rate and high morbidity). On the other hand, a positive relationship may indicate that either there is little underlying need (low rate and low morbidity) or that a high degree of need is causing a high rate (high rate and high morbidity).  152  CONCLUSION Over twenty years of work into procedural variations has documented that differences in procedural rates do exist among areas and between sub-groups in the population, although some of the variations found likely result from artifactual and methodological sources. However, the cause of these variations and their importance is still unclear. Consequently, the implications for policy are uncertain. It is clear, however, that despite a documented relationship between supply variables and procedural rates, alteration of supply without consideration of other community factors may cause more harm than good. Future research should include combinations of supply and community factors, pay attention to methodological issues and include some consideration of procedures that should have been done but were not.  153  REFERENCES: Anderson, G.M. and J. Lomas, 1989. Regionalization of coronary artery bypass surgery: effects on access. Medical Care 27 (March):288-296. Barnes, B. et al, 1985. Report on variation in rates of surgical services in the Commonwealth of Massachusetts. TAMA 254 (July):371-375. Brook, R.H. et al, 1988. Diagnosis and treatment of coronary disease: comparison of doctors' attitudes in the USA and the UK. The Lancet (April):750-753. Bunker, J.P., 1970. Surgical manpower. A comparison of operations and surgeons in the United States and in England and Wales. N Eng T Med 282 (January):135-144. Chassin, M.R. et al, 1986. Variations in the use of medical and surgical services by the medicare population. N Eng T Med 314 (January):285-290. Chassin, M.R. et al, 1987. Does inappropriate use explain geographic variations in the use of health care services? A study of three procedures. TAMA 258 (November) :2533-2537. Diehr, P. et al, 1990. What is too much variation? The null hypothesis in smallarea analysis. Health Services Research 24 (February):741-771. Evans, R.G., 1984. Strained Mercy: The Economics of Canadian Health Care. Toronto: Butterworths. Feinglass, J., Martin, G.J., and Sen, A., 1991. The financial effect of physician practice style on hospital resource use. Health Services Research 26 (June):183-205. Hay, D.I., 1988. Socioeconomic status and health status: a study of males in the Canada Health Survey. Soc Sci Med 27 (12):1317-1325. Joffe, J., 1979. Mobility adjustment for small area utilization studies. Inquiry 16 (Winter):350-355. Leape, L.L. et al, 1990. Does inappropriate use explain small-area variations in the use of health care services. TAMA 263 (February):669-672. Maynard, C. et al, 1986. Institutional differences in therapeutic decision making in the Coronary Artery Surgery Study (CASS). Medical Decision Making 6 (July-September):127-135. McPherson, K., 1989. International differences in medical care practices. Health Care Financing Review (Annual Supplement):21-32.  154  Mindell, W.R., E. Vayda and B. Cardillo, 1982. Ten year trends in Canada for selected operations. CMA Journal 127:23-27. Paul-Shaheen, P., J. Clark, and D. Williams, 1987. Small area analysis: a review and analysis of the North American literature. T Health Politics, Policy and Law 12 (Winter):741-807. Pearson, R.J.C. et al, 1968. Hospital caseloads in Liverpool, New England and Uppsala. The Lancet 2:559-566. Roos N.P. and L.L. Roos, 1981. High and low surgical rates. Risk factors for area residents. ATPH 74(4):313-314. Roos, N.P., E. Shapiro and L.L. Roos, 1984. Aging and the demand for health services: which aging and whose demand? The Gerontologist 24:31-36. Roos, L.L. and S.M. Cageorge, 1987. Innovation, centralization and growth: coronary artery bypass surgery in Manitoba. Unpublished paper. Winnipeg: University of Manitoba. Sauter, V. and E. Hughes, 1983. Surgical utilization statistics: some methodologic considerations. Medical Care 11 (March):370-377. Sheps, S., S. Scrivens and J. Gait. Perceptions and realities: medical and surgical procedure variation - a literature review. Unpublished paper. Vancouver: The University of British Columbia. Shwartz, M. et al, 1981. The effect of a thirty per cent reduction in physician fees on medicaid surgery rates in Massachusetts. ATPH 71 (April):370-375. Stockwell, H. and E. Vayda, 1979. Variations in Surgery in Ontario. Medical Care XVII (April):390-396. Tedeschi, P.J., R.A. Wolfe and J.R. Griffith, 1990. Micro-area variation in hospital use. Health Services Research 24 (February):29. Wennberg, J.E., 1979. Factors governing utilization of hospital services. Hospital Practice (September):117-127. Wennberg, J. and A. Gittelsohn, 1982. Variations in medical care among small areas. Scientific American 246:120-131. Wennberg, J.E., J.L. Freeman and J. Culp, 1987. Are hospital services rationed in New Haven or over-utilized in Boston? The Lancet (May):1185-1188. Wilson, P.A., J.R Griffith and P.J. Tedeschi, 1985. Does race affect hospital use? ATPH 75(3):263-269.  155  Winslow, R.L. et al, 1988. The appropriateness of performing coronary artery bypass surgery. JAMA 260 (July):505-509. Young, M.J. et al, 1987. Do cardiologist have higher thresholds for recommending coronary arteriography than family physicians? Health Services Research 22 (December):623-635.  156  CHAPTER FOUR RATIONALE AND METHODOLOGY  RATIONALE The previous chapters show that the incidence of revascularization procedures has risen dramatically since CABS was first introduced in the late 1960's. This increase has occurred despite the fact that CABS has been proven efficacious in increasing survival in only a limited number of conditions and the efficacy of PTCA, in comparison to non-invasive treatment, has not been evaluated. Moreover, the biggest increase in the incidence of CABS has occurred in the elderly; a group that has never been evaluated in an RCT but which has been shown, in observational studies, to have a higher mortality and morbidity from CABS than occurs in younger patients. Despite recent government initiatives in B.C. to reduce waiting lists and provide greater access to CABS by opening another unit, there is no evidence to show that CABS is being used appropriately in this Province. Identification of ageand sex-specific rates of CABS in B.C. and its regions, can help to document the diffusion of the procedure in the Province and can provide the government with the data needed to plan, allocate resources to, and evaluate revascularization programs. Identification of regional variations in rates and their contributory factors will help determine whether access to these procedures may be a problem in some areas and the approaches that may help to improve access.  157  Questions: The questions addressed in this study are: i)  What are the age and sex specific trends in CABS in B.C. between 1979 and 1988?  ii)  What are the age and sex trends in PTCA in B.C. between 1987 and 1988?  iii)  What are the trends in the incidence of comorbid conditions in the CABS population between 1979 and 1988?  iv) v)  What are the regional trends in CABS between 1979 and 1988? Is there any significant variation in the age-sex adjusted CABS rates among the small areas (school districts) in B.C.?  vi)  Is there any significant variation in the age-sex and morbidity adjusted CABS rates among the small areas in B.C.?  vii) How much of the variation found in (iv) and (v) is explained by the independent variables (described below)?  METHODS  Study Design: Questions (i) - (iv) were addressed by means of a retrospective descriptive study. Revascularization procedures performed in B.C. during the relevant time period were identified and the characteristics of the patients described. Annual population-based rates were calculated for each region (school district). These rates were then used as the dependent variable in a poisson regression to answer question (v); was there more regional variation than would be expected by chance? The second part of the study, was an ecological analysis in which the annual age-sex adjusted CABS rates per school district were regressed on district socio-economic and health-services characteristics of each school district.  158  Independent Variables: The independent variables tested in the regression analysis were: Year - this was coded from 0 to 5, with 1983 as the reference year. Distance of school district from school district of the nearest cardiologist (DSCAR) - this was measured categorically as same school district, adjacent school district or far school district, and was coded from 0-2 respectively. In cases where a cardiologist was resident in an adjacent school district but there was no road access between school districts in that year, the distance was coded as far school district. Similarly, distances between school districts which involved a ferry crossing were labeled as 'far'. Distance of school district from school district of the nearest internist (DSINT) - measured as for cardiologists. Distance of school district from school district of the nearest cardiac surgery centre (DSCEN) - measured as for cardiologists. Income (INC) - average annual income per census family measured in tens of thousands of dollars and entered as a continuous variable with $30,000 as the reference level. 1981 census data was used for 1983-1985 and 1986 census data for 1986-1988. Employment rate (EMPRAT) - proportion of the workforce (aged 15 and over) who were employed. Entered as a continuous variable with 0.7 as the reference level. Census data was used as for income. Graduation rate (GRADRAT) - proportion of the 15-19 year-old population who graduated from high school in a given year. Entered as a continuous variable. Census data was used as for income.  159  Data Sources: Data for all patients who received coronary artery bypass surgery (CCP 1 codes 4811-4819) or percutaneous transluminal angioplasty (CCP codes 4801-4805), or who were discharged with a diagnosis of coronary artery disease (IHD-9 410-414, 429.2) in British Columbia for the fiscal years 1979-88, were extracted from archived records of the Hospital Morbidity Database. The data in this database, maintained by the Ministry of Health is obtained from the patient's hospital separation form. Up to, and including, 1982 the Ministry of Health took responsibility for entering and validating the data. After that year, the hospitals sent their data to commercial records institutes. Population figures for each Local Health Area (LHA) in B.C. were obtained from the Planning and Statistics Division in the Ministry of Finance and Corporate Affairs. Census data were obtained for the years 1981 and 1986 and population estimates for the other years for 1979 through 1988. Although it was originally intended to use LHA's as the unit of analysis, the patient's location on the database was entered as school district until 1982, thus school districts were used instead. After 1982, the patient's location was recorded as postal code only; these were translated to school districts using a table provided by the Planning and Statistics Division. The Planning and Statistics Division also provided 1981 and 1986 census data for employment and income in the LHA's. The annual numbers of students graduating from high school in each school district, between 1983 and 1988, were obtained from the B.C. Ministry of Education. The location of cardiologists and  1 The Canadian Classification of Diagnostic, Therapeutic and Surgical Procedures, Second Edition, is produced by Statistics Canada to meet Canadian needs for a procedural classification to be used in conjunction with the International Classification of Diseases (ICD-9). The Hospital Medical Records Institute (HMRI) re-codes coded ICD-9CM procedures into CCP codes with the result that procedures in the Hospital Morbidity Database are coded using CCP codes.  160  internists in B.C. between 1983 and 1988 was obtained from the annual editions of the Physicians and Surgeons directory published by the B.C. College of Physicians and Surgeons. Throughout most of the province, LHA's and school districts have identical boundaries. In cases where this was not so (for LHA's 5,6,20,78, 53,93 and 95), LHA's were re-coded into school districts using a translation table provided by the Health Planning Database, Policy, Planning and Legislation Division in the Ministry of Health. In addition data from school districts 92 (Nishga) and 94 (Telegraph Creek), which have very small populations, were aggregated with those from school district 88 (Terrace). Information on the patients receiving CABS or angioplasty in Alberta was obtained from the three hospitals involved: Holy Cross Hospital, Foothills Hospital and the University of Alberta Hospitals. The total numbers of B.C. residents receiving CABS or PTCA out-of-province between 1979 and 1989 were obtained from Data Systems, Health Information Division, Health and Welfare Canada. Data were analyzed using SPSS on the mainframe computer at U.B.C. Regression analysis was performed using SAS Proc Logistic procedure on an IBM personal computer.  Study Population and Analysis: 1. Descriptive Analysis: Included in the analysis were all patients over the age of 20 years who received isolated CABS (i.e., CABS not accompanied by valve replacement or other cardiac surgery) or angioplasty in B.C. between April 1979 and April 1989. Patients below the age of 20 years were dropped from analysis on the basis that they were probably not representative of the usual population of CABS recipients. Patients  161  who had surgery at hospitals other than the cardiac centres were included in the analysis although they likely represent miscoding. Cases with missing postal codes were assumed, on advice from Data Support, Hospital Programs, to have come from out of Province and were included in the analysis under that category. Prior to 1983, cases who had more than one CABS (4811-4815 or 4819) coded for the same admission were assumed to have had a single operation 2 . Diabetes and COPD were chosen as indicators of co-morbidity because they are relatively stable over time 3 . In 1983, the number of diagnostic fields in the database changed from two to 16 resulting in an apparent rise in co-morbidity in that year4 . Annual standardized incidence ratios (SIRs) for CABS were derived for each school district as follows. Patients were categorized into twelve 5-year age groups between 20 and 75plus, and the number of procedures performed on B.C. patients in each of the 24 age-sex categories was divided by the total 1979-1988 population in each age-sex category to give an age-sex specific ten year rate. These rates were then used to calculate the annual expected numbers of CABS per school district by  2 In 1983 the format of the Hospital Morbidity Database was changed to include, among others, the date of each procedure. Prior to 1983 only admission and discharge dates were given. 3 Discussions with epidemiologists and clinical practitioners prior to the study indicated that other diseases, e.g., hypertension, were subject to changes in diagnosis (e.g. what b.p. level indicates hypertension) or to fads in diagnosis which could spuriously alter the rate. 4 It seems reasonable to assume that an increase in the number of diagnostic fields available would  result in an apparent rise in co-morbidity. At the same time that the number of diagnostic fields increased the definition of the included diagnoses changed. Prior to 1983 the "primary diagnosis" was the one responsible for the patient's admission to hospital, and the "secondary diagnoses" was another important diagnosis. In 1983 the first diagnosis became the 'principle diagnosis' which is defined as "the diagnosis most responsible for the patient's stay in the institution." The next most important diagnosis the 'primary diagnosis' describes "another important condition of the patient which usually has a significant influence on the patient's length of stay". The 'secondary diagnosis' "describes a condition for which the patient may (or may not) have received treatment but did not significantly contribute to the patient's length of stay in hospital". Other diagnostic fields include admitting diagnosis, complications arising in hospital, E-codes, morphology codes and transfer diagnoses (Ministry of Health, 1985).  162  multiplying the annual population in each age-sex group per school district by the age-sex specific rate. Observed and expected CABS for each age-sex group per school district per year were then summed to give an annual observed and expected figure per school district. Standardized incidence ratios (observed CABS/expected CABS) were then calculated for each school district per year. The SIRs were then multiplied by the overall Provincial CABS rate (total number of procedures 1979-1988/total population 1979-1988) x 10,000 to give the annual agesex adjusted rate for each school district per 10,000 population. 5 It proved impossible to accurately determine the numbers of angioplasties per year because a CCP code for angioplasty was not introduced until April 1987. Prior to this date, angioplasties were coded as 4800 (removal of coronary artery obstruction), which includes endarterectomy and gas endarterectomy among other procedures. Following the introduction of angioplasty codes (CCP 4801-4805), one centre did not use the new code but continued to code angioplasty under the CCP 4800 code. Therefore, regional SIRs were not calculated for angioplasty. Patient residence in the Alberta data was identified by means of three or sixdigit postal code. Because six-digit codes were required in order to identify most school districts, and because 1988 was the first year in which all Alberta centres collected six-digit codes, B.C. patients having surgery in Alberta could not be included in the annual school district figures. To determine how mobility affected  the incidence rates in school districts sending patients to Alberta, the 1988 SIRs were calculated both excluding and including the Alberta data. 2. Regression Analysis: To determine whether significant regional variation existed and, if so, the relative contribution that each independent variable made to the variation in area  5 This rate is provided to allow comparison with other studies, most of which use procedure rates  rather than SIRs.  163  rates, Poisson regression was carried out using a number of different models and 1983 to 1988 data. Poisson regression was chosen because, like weighted least squares regression, it assigns weights according to population size but has the added advantage of assigning weight even when the observed count is zero. In most years covered by this study there were some school districts which did not have any CABS. Poisson regression also allows the calculation of a 'saturated' model, used to test if there is more variation than would occur by chance alone. In the first group of models, the age-sex adjusted CABS rates (Table 15A) were regressed on the independent variables (described above). This regression was repeated using the age-sex adjusted CABS rate also adjusted for the CAD morbidity in each school district (Table 15B) as the dependent variable. Because it was not known which diagnosis of CAD would best indicate those patients who form the "potential CABS population", several diagnoses were tried to see which gave the best fit. These diagnoses were: i.  Acute myocardial infarction (ICD-9 410 only). This diagnosis was used for those patients who had no other diagnosis of CAD and is an indicator of the incidence of CAD in the school districts. It is obviously not a true incidence because patients who die before reaching hospital, or those having a 'silent' MI, are not included.  ii.  Chronic ischemic heart disease or unstable angina but not with AMI (ICD9 411-414 or 4292, but not 410).  iii.  Chronic ischemic heart disease or unstable angina including those who also had a diagnosis of AMI but not those with AMI only (ICD-9 411-411 or 4292 plus 410). This category included all those with chronic CAD whether or not they had had an AMI.  iv.  All cases (ICD-9 410-414, 4292), i.e., (i) plus (iii).  164  For each of the above diagnosis groups, a fully saturated "perfect model" was run and was compared to the null model; the model with the least unexplained variance was assumed to be the one for which the CAD diagnosis best indicated the potential CABS population. Pearson-product-moment correlations between the SIRs for CABS and those for the CAD diagnoses were also calculated, and supported this finding. The adjustment for CAD was accomplished by using the following outcome variable: CABS/(Expected CABS x CAD/Expected CAD x 100,000). In calculating the annual morbidity rates per school district all patients admitted to hospital in B.C., for each fiscal year from 1983 to 1988, with a diagnosis of ischemic heart disease (ICD-9 410-414, 4292) were identified. Within each year all non-B.C. residents, readmissions and patients receiving CABS in that year were removed. For each diagnosis group (outlined above) expected numbers of admissions per school district per year were calculated, using the same method as was used to calculate the numbers of expected CABS. Finally, in order to determine if there were interactions between age, sex, time (year) and region, the crude CABS rates were regressed on these variables were regressed on the crude CABS rate using three different models (see Table 15C). For these regressions, school districts were aggregated into four regions to reduce the otherwise unmanageable number of possible combinations of age-sexyear-region factors; Region was entered into the model as a categorical variable.  165  TABLE 14 REGRESSION MODELS A.  Models used to estimate effect of independent variables on CABS rate before adjustment for morbidity. i)^CABS/EXPC* = B0 + Bitime + B2distance from cardiologist + B3distance from centre + B4distance from internist + B5income + B6employment rate + B7graduation rate + interactions  B.  Model used to determine effect of independent variables on CABS rate after adjustment of expected rate for morbidity in school district. i)^CABS/ (EXP ** x CAD/EXPCAD *** x 100,000) = B0 + Bitime + B2distance from cardiologist + B3distance from centre + B4distance from internist + B5income + B6employment rate + B7graduation rate + interactions  C.^Models to determine age-sex-time-region interactions i)  CABS/POP = B0 + Biage + B2sex + B3time + interactions  ii)  CABS/POP = B0 + Bi age + B2sex + B3time + interactions + B4Regionl + B5Region2 + B6Region3  iii) CABS/POP = B0 + Bi age + B2sex + B3time + interactions + B4Regionl + B5Region2 + B6Region3 + interactions of age, sex and time with regions *Expected CABS x 100,000. The factor of 100,000 was used to force Proc Logistic into a poisson regression. **Expected CABS ***Expected CAD  166  CHAPTER FIVE REVASCULARIZATION IN BRITISH COLUMBIA 1979-1988 RESULTS  CORONARY ARTERY BYPASS SURGERY  Characteristics of the CABS population: From 1979 to 1988 the number of CABS performed annually in B.C. increased from 931 to 1496 , an overall increase of 60 percent. The trend in the growth of CABS follows that described by Peters et al (1990) for all of Canada; there was an increase between 1981 and 1983 followed by a decline over the next two years and then by a more moderate increase. The percent of annual cases performed on non-residents of B.C. increased from 0.9 to 1.7 percent over the ten year period, while the percentage of women receiving CABS remained around 20 percent. The mean age of the women was consistently 2-3 years higher than that for the men although the mean age for both groups rose by approximately five years (from 57.4 to 62.9 for the whole population) over the study period. These results are shown in Tables 16 - 18 in Appendix A. The Provincial annual crude and age-sex adjusted rates and the corresponding SIR's are shown in Table 19. The steady rise in rate over the 10year period, depicted in Figure 2, indicates that the rate of procedures is increasing faster than population growth. Despite annual fluctuations in rate there is an upward trend which becomes more marked after 1985. Table 20 (Appendix A) shows the overall age- and sex-specific 10-year rates for CABS. The female rates are markedly lower than those for males but the highest rates for both males and females are seen in the 65-69 age group. This result differs from other studies in the literature (Peters et al 1990, Gillum 1987) where the highest rates were seen in the 55-64 age group.  167 FIGURE 2 ANNUAL AGE-SEX ADJUSTED CABS RATE PER 10,000 POPULATION B.C. 1979-1988  7.00 6.50 6.00 5.50 5.00 4.50  1^1^i^i^1^i^i^1^1 79^80^81^82^83^84^85^86^87^88 YEAR  TABLE 19 ANNUAL STANDARDIZED INCIDENCE RATIOS AND CABS RATES B.C. 1979-1988 Year  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 Overall rate  Crude Rate per 10,000*  Age-sex adjusted rate per 10,000 **  Standardized Incidence Ratio  5.20 5.01 4.96 5.29 5.96 5.78 5.63 5.96 6.14 6.79 5.70  5.19 5.03 5.01 5.35 6.02 5.83 5.65 5.91 6.06 6.65  0.91 0.88 0.88 0.94 1.06 1.02 0.99 1.04 1.06 1.17 -  *All rates are per population > 20 years of age **SIR x overall ten-year rate  TABLE 21 ANNUAL DISTRIBUTION OF ISOLATED CABS BY AGEGROUP B.C. 1979-1988 Annual Numbers of Cabs  Age Group 1979  1980^1981  20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90+  0^0 0 2^0 0 4^6 8 17^18 16 43^40 47 100^89 100 140^157 156 209---,_ 203^202 '---205 -217 179 151^157 160 57^62 47 7^11 6 1^2 2 1^1 1 0^0 0  Total  931  931^962  *Line indicates modal age group  1983  1984  1985^1986  1987  1988  1 0 3 0 7 4 24 31 49 51 71 83 179 161 193 225 219 -279 227 207 118 77 27 28 3 3 0 0 0 0  0 0 2 16 38 94 144 203 267 241 148 39 2 0 0  0^0 1^2 4^3 19^16 40^43 69^72 117^117 190^198 277264 232 -275 166^203 52^70 1^8 0^0 0^0  0 1 1 11 36 72 148 198 255 320 211 64 7 1 0  0 1 10 14 32 95 154 212 260 365 236 100 17 0 0  1220  1194  1181^1271  1325  1496  1982  1050  ---------  169  The distribution of cases by age group (Table 20) shows changes in the modal age-group, followed by plateaus, between 1979 - 1980 and 1985-1986. However, when these figures are expressed as population rates the mode fluctuates between age groups 60-64 and 65-69 until 1984 when it remains in the latter group. It appears that demographic change does not have an immediate effect on the CABS rate but is likely mediated by established practice patterns. Figure 3 depicts the annual age-specific use rates (unadjusted for sex) for three time periods -1979, 1984 and 1988. Over the ten year period, utilization decreased slightly for patients under the age of 60, but increased markedly for those over 60, with the greatest increase seen in the over-65's. These age-use curves depict a pattern similar to that seen when there is a cohort effect. This possibility would be worth exploring in future research. When expressed as a percentage of total CABS, procedures in the over-65 population have increased from just under 25 percent of the procedures in 1979 to almost 50 percent in 1988. Their rate increased almost two-fold, from 3.32 to 6.38 per 10,000. However, when the over-65 population is broken down into subgroups, the 65-69 year olds, although doubling in number over the 10 years, have actually declined by about 25 percent as a percentage of over-65 cases (Table 21, Appendix A). When the age-specific sex-adjusted rates for subgroups of the over 65 population are considered (Table 22, Appendix A) it is seen that although all sub-groups have increased their incidence rate, the biggest increase has occurred in the 75+ year olds, where the rate has increased almost nine-fold, although starting at a very low rate. The 70-74 year old rate has increased almost three-fold. The 1988 CABS population is not only older than that in 1979, but it also appears to be sicker; at least as far as co-morbid conditions are concerned. Over the 10-year period the number of patients with other diseases (diabetes and COPD) who  170  FIGURE 3 AGE-USE CURVES  B.C. 1979, 1984, 1988.  30 25 20 15 -  RATE 10 5 0  I^  I  I  I  I  20-49^50-54^55-59 60-64 65-69^70-74^75+  AGE GROUP 10--  1979  -43-  1984 -*-- 1988  171  underwent CABS rose from 30 to 247 (Table 24, Appendix A). This represents a rise from 3.2 percent to 16.51 percent of the annual cases. Some of this apparent rise can be attributed to the 1983 increase of the number of diagnostic fields (from 2 to 16) in the morbidity database. The majority of these patients with co-morbidity had diabetes. Between 1979 and 1988, COPD increased from 1 to 4.3 percent, and diabetes from 3.2 to 12 percent, of annual CABS. No patients were found to have both diseases. Table 25 (Appendix A) shows the distribution of co-morbidity by age group for CABS. Although the number of patients with co-morbidity in the older age groups is small, when expressed as a percentage of total cases they indicate a higher percentage of co-morbid conditions in the over-65 age group with the highest rate in the 80-84 year olds. It appears likely that the increase in co-morbidity has occurred because of the increase in the age of the CABS population. The above data do not include B.C. residents who received their CABS procedure outside British Columbia. Figures supplied by the Federal Ministry of Health and Welfare (Table 26, Appendix A) indicate that between 1979 and 1988, 258 B.C. residents obtained CABS in other provinces or countries (2.2 percent of the total number of B.C. residents receiving CABS). Data obtained from Alberta Hospitals performing CABS show that between 1983 and 1988, 66 men and 10 women (15 percent of all cases) received CABS in Alberta although the 1988 figure provided by the Federal Ministry does not agree with the Alberta data. The sudden increase (from 3 to 28) in the number of cases leaving for Alberta in 1987 is likely due to the start-up of the CABS program in the Calgary hospitals in that year. Table 27 (Appendix A) shows the numbers and mean ages of B.C. residents receiving CABS in Alberta.  172  Re-operations: Re-operations after 1983 were identified using the scrambled MSP number and the sex and birthdate as described in Chapter 4. However, when these figures (82 patients receiving 83 re-operations) were compared with the re-operations at Vancouver General hospital in 1991 (121 out of 1100 CABS) it was evident that the numbers of re-operations had been grossly underestimated, even allowing for the fact that patients receiving their first operation prior to 1983 were not included. Consequently it was not possible to assess re-operations over time. The underestimation of re-operations appears to arise from the unreliability of the eight-digit MSP numbers which were used to identify individuals who were admitted more than once for CABS or with a diagnosis of ischemic heart disease. According to personnel in the Medical Services Plan Registration Section, individuals may change their MSP numbers over their life-time. In addition once a number becomes dormant it may be used again for another individual after a two-year period. Because sex and birthdate were also used to identify re-operations and re-admissions, it is unlikely that either was overestimated but both reoperations and re-admissions were likely underestimated. In the case of readmissions it is unlikely that there was any bias by school district so that the morbidity figuresl, although likely somewhat higher than they should have been, did not introduce any systematic bias. Because of coding differences between hospitals it proved impossible to develop reliable figures for the numbers of patients receiving mammary artery implants or for the number of vessels bypassed. For example, if two coronary arteries were bypassed, one with a saphenous graft and one with an internal mammary implant, Centres 2 and 3 would code the procedure as bypass of one I Approximately 3000-4000 CAD cases annually were dropped from the analysis becasuse they represented readmissions in that year.  173  coronary artery plus a single internal mammary implant (CCP codes 4812 and 4816); Centre 1 would code it as a double coronary bypass only (4813). In this latter centre, internal mammary implants (CCP codes 4816 or 4817) were only coded in cases in which a saphenous vein graft was not also used. To add to the confusion, Centre 2 personnel were not sure if they had always coded their implants the same way or when the changeover might have occurred. The next section addresses the issue of "place"; i.e., where the patients receiving CABS came from and where they went to receive their surgery. This section addresses the question: What are the regional trends in CABS in B.C. from 1979-1988?  Region of Residence of CABS Population: Table 28 (Appendix A), shows the distribution of CABS cases by year and school district. Visual inspection of this table shows areas of greater activity around Nelson (SD 7), the Prince George area (SD 57), the Okanagan (SD 14-15, 2224), the Victoria and mid-to-north Island area (SD 61,62 63, 65, 68, 69,70), and Vancouver and the Lower Mainland (SD 36 - 44). These areas are, of course, more densely populated than the rest of the Province. Even so, the regularity with which substantial numbers of cases occur in these areas is striking. The annual number of cases within the above areas do not show any discernible pattern overall. There is a tendency for the number of cases per school district to increase over time in Vancouver and the Lower mainland, while the reverse is true in Victoria and some of the near-by school districts. Sudden and sustained rises in annual numbers in a school district could not be related to the introduction of a specialist into the area except in Richmond (SD 38) where the arrival of a cardiologist in 1986 was followed in the subsequent year by a 56 percent increase in the annual cases. It is possible that increases in other school districts  174  may have coincided with a change in personnel (as opposed to the introduction of a specialist) but this was not studied. The annual numbers of CABS procedures performed on the residents of each school district, were used to calculate annual standardized incidence ratios (SIRs) for each school district. The most striking aspect of the these SIRs, shown in Table 29 (Appendix A), are their extreme variability within school districts. Although there are school districts which maintain a fairly steady ratio across the years, e.g., those in the Lower Mainland and the suburbs of Vancouver (SD 36-43), there are others which jump from low ratios to SIR's of 2 or above. A tendency to consistently high ratios, particularly in later years, can be seen in southern Vancouver Island (SD 61-66), in Kelowna (SD 23) and school districts to the south (SD 14-16), and in Kitimat (SD 80). Kelowna and Victoria (South Vancouver Island) are centres for cardiology and it could be expected that closer proximity to a cardiologist may lead to a higher CABS rate. The variability in observed CABS within school districts over time, is in part due to population size. When the coefficient of variation (the ratio of the standard deviation of observed CABS to the mean) for each school district over the ten year period is graphed against mean population size (Figure 4), those school districts with populations under 10,000 show the greatest variability, and those with populations over 20,000 show the least. When the range of SIRs within each school district are depicted by school district ordered by population size (Figure 5), the trend to decreasing variability as population size increases, can be seen. However, there are some school districts with populations over 50,000 which show variations of over two- to four-fold (SD 68 and 23) so there are other factors besides population size which are exerting an influence on rate variations.  175 FIGURE 4 SCAT^ IhRGRAPH OF COEFFICIENT OF VARIATION OF OBSERVED CABS IN SCHOOL DISTRICT BY SCHOOL DISTRICT POPULATION  2.00 —^■ 1.80 —  •  1.60 — 2 1.40 — 2 0 1.20—  w  •  ■  or  III •^• m^ 6 0.80 4-^  n 1.00 z  -^  ■  % NE^ ■ 1—^ IN ■^■ • co 0.60 — ■^• •^• Um ^■ ^■ IN is 0.40 —  .NO • II •^OM  ^  •  •  ■^■ ■  0.20 —  ■  0.00  1 5000  0  I 10000  I 20000  I 15000  POPULATION  2.00 — 1.80 — 1.60 — 1.40 — 1.20 — 1.00 — 0.80 — 0.60 — 0.40 — ■  VI •  ^■  0.20 —^o IP■^ ■ "^■ 0.00 ^  ■  0^50000 100000 150000 200000 250000 300000 350000  POPULATION  176 FIGURE 5 RANGE OF STANDARDIZED INCIDENCE RATIOS FOR ISOLATED CABS BY SCHOOL DISTRICT ORDERED BY POPULATION SIZE B.C. 1979-1988  3.50 —  ■  3.00 —  ■ ■  2.50 — •  (5 cc  2.00 —^  ■  ■  ■  1.50 — 1.00 — 0.50 —  •  ■  cc (i) Lij  ■  ■  ■  I I^I •  •  0.00  •  •  87 49 13 16 84 29 26 10 81 17 76 66 50 18  4  21 55 32 12  SCHOOL DISTRICT  3.50 —  ■  3.00 — 2.50 —  ■  Co  cc 2.00 — c7) w (5 4 1.50 1.00 —  ■  ■ ■  ■  0.50 — 0.00  • *tii  •  ■ ■  •  • I  I^I  111•11+1  11  30 77 31 19 3 64 9 80 54 85 56 86 48 14 1 52 46 47 2 SCHOOL DISTRICT  177  FIGURE 5 (contd.)  3.50 3.00 2.50 2.00 1.50 1.00  —— —— — —  ■  0.50 —• 0.00  Ii^1 .^1 i  7^28 11  I  i  1^I^I^I^1^I^I^I^I^I^I^I  69 60 59 75 88 72 89 70 15 27 65 71 62 63 42 SCHOOL DISTRICT  3.50 — 3.00 — 2.50 —  —1  2.00 — 1.50  1.00 —  0 .50 — 0.00 22 45 33 40 34 35 37 68 24 57 23 38 43 44 41 36 61 39 SCHOOL DISTRICT  178  Table 30 (Appendix A) shows the observed and expected numbers of CABS, the coefficient of variation and the overall external quotient (largest SIR/smallest SIR) for the study period. It is interesting that school districts showing a wide variation in SIRs across the study period may, overall, have either approximately equal observed and expected cases (e.g., SD 48 and 75) or a wide disparity between observed and expected cases (e.g., SD 11 and 60). The reverse is also true; school districts with a relatively small variation over time may have widely disparate observed and expected cases overall (e.g., SD 61) or very similar ones (e.g., SD 34). There are one or two regions which appear to have markedly fewer cases than expected over the ten-year period. For example, Golden on the Alberta border had only two cases receiving CABS in B.C. and none in Alberta, unless some of the Alberta cases which could not be assigned to a region came from Golden. Among school districts the annul external quotient ranged from nine- to thirty-fold. When school districts with populations under ten thousand plus those sending ten percent, or more, of overall cases to Alberta were excluded, the range was two- to twenty-two-fold. When data were aggregated into two time periods, 1979-1984 and 1985-1988 (Table 31, Appendix A ), there was an over twelve-fold variation for the first period and a thirteen-fold variation for the second period when all school districts were included. When school districts with small populations or mobility to Alberta were excluded as above, the annual variations among school districts fell to three-fold and two-fold for the first and second periods respectively. Aggregation of the data over the whole study period showed an eighteen-fold variation across school districts overall and a two-fold difference when districts were excluded as above, with Trail (SD 11) having the lowest SIR at 0.69 and Victoria (SD 61) and Sooke (SD 62) sharing the highest at 1.62.  179  The above data show that, even when estimated conservatively, there is a two- to nine-fold variation within school districts over the ten years of the study and at least a two-fold variation across districts. It also appears that the size of the variations in SIRs within school districts over time are not necessarily indicative of more, or less, CABS being performed then are expected over that period. Consequently, no clear pattern emerges to help clarify the issue of the importance of geographical variations in procedural rates. Despite the variability within and across school districts there is a clear trend to increased utilization of CABS across the Province during the study period. Over the years there are fewer areas with extremely low ratios (or no cases) and the procedure appears to have spread more into the north. The southeastern border of the Province consistently shares low ratios but these school districts appear to receive the procedure mainly in Alberta. Calculation of the 1988 SIRs both including and excluding patients leaving the Province for Alberta, shows marked differences in some SIRs for the southeastern and northeastern school districts (Table 32 Appendix A). In 1988, 47.6 percent of all CABS patients living in the seven school districts on the Alberta border went to Alberta for their surgery as did 37.5 percent of patients living in the East Kootenays and 57 percent of patients in the Peace River country. The number of people from the Central Okanagan school district (SD 23) who left B.C. for their procedure, three in 1988 and nine (out of 504 cases between 1983 and 1988) overall, is of interest because this school district is closer to Vancouver than to any of the Alberta centres. It may be that these cases were original Alberta residents who have retired to the Okanagan or that the cardiologists in Kelowna have professional links to Alberta.  180  Referral Patterns The geographic distribution of school districts is shown in Map A. The centres used by cases from each school district for the period are shown in Table 33 (Appendix A) and are mapped in Maps B and C. For 1979-88, Centre 3 has the most clearly defined catchment area receiving 79 percent of its cases from 5 school districts, all on Vancouver Island. Centre 1 receives 50 percent of its cases from Vancouver and suburbs (SD 38, 39, 40, 41, 42, and 43) which represents 68 percent of the cases from the area. It receives a further 33 percent of cases from the Fraser Valley (SD 34,35, 36,37) which represents 77 percent of the cases from that area. Thus, Centre 1 receives 83 percent of its cases from the immediate area. Centre 2, on the other hand, has a much wider and more diversified catchment area. One quarter of its cases come from Vancouver and suburbs and a further quarter from the Interior (53 percent of cases from this area). Seventeen percent of cases come from the North Vancouver (SD 44), Sechelt (SD 46), and Powell River (SD 47) areas (82 percent of these areas' cases) with a further 9 percent from the north (83 percent of cases from the area) and 8 percent from the Kootenays (75 percent of cases from the area who received surgery in B.C.). There are definite referral patterns governing the utilization of centres by school districts and these patterns can be clearly seen in Map B. Within each centre, differences over time in the numbers and type of population served may indicate changes in referral patterns, or may reflect changing attitudes towards the procedure within that centre. The number of CABS cases served by each centre have diverged over the years (Table 34). In 1979 all centres were serving approximately the same number of cases. Centre 3 has not changed while Centre 1 has doubled the number of cases and Centre 2 has increased the number 1.5 fold. Centre 3 has fallen from serving approximately 33 percent of all cases in 1979 to only 20 percent in 1988, while Centre 1 has increased  181  its share from 34 to 45 percent. Comparison of the annual CABS cases per centre with the annual total open heart cases per centre 2 (Table 35) shows that since 1983, the percentage of open heart procedures that are CABS cases has increased in Centre 2, decreased in Centre 3 and stayed the same in Centre 1. These changes, though relatively small, show a definite trend. Because all open-heart surgery is funded as one program, the change in proportion of CABS to other open-heart surgery in Centres 2 and 3 must either represent a change in referral patterns or a change in policy within the hospitals. The annual mean ages by centre (Table 36, Appendix A) show that Centre 3 consistently cares for a slightly older population than the other centres. Although older, this population does not appear to have more concomitant diseases than the population served by the other centres; 9.1 percent of all cases in Centre 3 had comorbidity compared with 9.6 percent in Centre 2 and 8.9 percent in Centre 1 (Table 37 Appendix A). TABLE 34 CORONARY ARTERY BYPASS SURGERY PER CENTRE PER YEAR B.C. 1979-1988 Year  Centre 1  Centre 2 Total  Centre 3  Annual  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  320 330 338 413 528 463 544 547 555 678  307 310 324 334 351 434 387 420 455 516  304 291 279 301 341 296 250 302 313 302  931 931 961 1048 1220 1193 1181 1269 1323 1496  Total  4736  3838  2979  11553  2 This data was obtained from the Medical Consultation Branch, Institutional Services Division in the B.C. Ministry of Health.  182  TABLE 35 PERCENT OF OPEN HEART SURGERY DEVOTED TO CABS ANNUALLY BY CENTRE B.C. 1983-1988 Year Open Heart  Centre 1 CABS  % Total  Open Heart  Centre 2 % CABS Total  Open Heart  Centre 3 CABS  % Total  1983 1984 1985 1986 1987 1988  778 744 820 811 825 998  528 463 544 547 555 678  67.87 62.23 66.34 67.45 67.27 67.93  509 623 584 596 628 705  351 434 387 420 455 516  68.96 69.66 66.26 70.47 72.45 73.19  423 369 336 384 414 392  341 296 250 302 313 302  80.61 80.21 74.40 78.64 75.60 77.04  Total  4976  3315  66.6  3645  2563  70.3  2318  1804  77.8  TABLE 38 INCREASE IN NON-CABS REVASCULARIZATION PROCEDURES B.C. 1979-1988 Year  Total Revasc.  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  933 935 975 1077 1286 1479 1768 1982 2136 2427  N 2 4 13 27 66 285 587 711 811 931  % total 0.21 0.43 1.33 2.50 5.13 19.27 33.20 35.87 37.97 38.36  N 931 931 962 1050 1220 1194 1181 1271 1325 1496  % total 99.79 99.57 98.67 97.50 94.87 80.73 66.80 64.13 62.03 61.64  Total  14998  3437  22.91  11561  77.08  Non-CABS  (CCP 4800-4805)  CABS (CCP 4810-4819)  183 FIGURE 6 GROWTH OF REVASCULARIZATION PROCEDURES B.C. 1979-1988  1979^1980^1981^1982^1983^1984^1985^1986 Year CABS  El^Other  ^  1987  ^  1988  ^•^ Total  TABLE 40 NUMBERS AND PERCENTAGES OF REVASCULARIZATION PROCEDURES BY CENTRE B.C. 1983-1988 Year  Total*  Centre 2 Other **  %^of total  Total*  355 466 464 534 619 714  4 32 77 114 164 198  1.1 6.9 16.6 21.4 26.5 27.7  387 443 551 650 632 602  46 147 301 348 319 300  11.9 33.1 54.6 53.5 50.5 50.0  Total 3829 1304 34.0 3152 589 18.1 * Total revascularization procedures (CCP 4800-4819) in centre. ** Non-CABS revascularization procedures (CCP 4800-4805) in centre.  3265  1461  44.7  Total * 1983 1984 1985 1986 1987 1988  542 561 750 790 876 1100  Centre 1 Other %^of ** total 14 98 206 243 321 422  2.6 17.5 27.5 30.7 36.6 38.4  Centre 3 Other %^of ** total  184  ANGIOPLASTY It was impossible to determine the number of angioplasties performed in the Province because, following the introduction of the CCP code for angioplasty in 1987, one hospital continued to use the old "catch-all" code (CCP 4800) for removal of coronary artery obstruction. Consequently, a detailed analysis was not performed for angioplasty but a general overview will be given. If all procedures for removal of coronary obstruction are considered together (Table 38), they increased from 2 in 1979 to 931 in 1988. This represents a 464-fold increase overall and an increase from 0.2 percent to 38.3 percent of all revascularization procedures. When the annual increase in both CABS and other revascularization procedures are graphed it can be clearly seen that between 1983 and 1985, when utilization of non-CABS procedures was increasing rapidly, the growth of CABS slightly decreased. Since 1985 both procedures have increased (Figure 6). As with CABS, some B.C. residents receive angioplasty in Alberta. Between 1983 and 1988, 100 patients went to Alberta centres for angioplasty and eleven of these patients received more than one angioplasty in Alberta. Although this reoperation rate appears lower than one would expect if restenosis is around 30 percent, it is possible that some patients with restenosis received another angioplasty in B.C. Examination of the age and sex distribution of the non-CABS revascularization procedures show some differences from the CABS population. A larger proportion of women (0.32) receive non-CABS revascularization and the overall average age tends to be somewhat lower (60.48 years in 1988). Numbers by sex and overall mean ages are shown in Table 39 (Appendix A).  185 When the annual number of "other revascularization procedures" is compared with the number of CABS procedures in each centre, there are marked differences between the centres. In 1988 Centre 3 had 50 percent of revascularization procedures as non-CABS procedures, compared to 38 percent in Centre 1 and 28 percent in Centre 2. In fact, the percent of revascularization procedures devoted to non-CABS has declined in Centre 3 since 1985 but has been steadily growing in the other two centres (Table 40). It is of interest that 38 non-CABS revascularization procedures were performed in other centres between 1982 and 1988; thirteen of these procedures were coded as angioplasties. It appears unlikely that so many cases would be due to miscoding and it is possible that cardiac centres without cardiacsurgical facilities may be performing angioplasties. A breakdown of annual numbers of angioplasties and other non-CABS revascularization by centre, is given in Appendix A (Tables 41 and 41A.  REGRESSION ANALYSIS This section describes the results of the regression analyses, which were used to address the questions:  i)  Is there any significant variation in the age-sex adjusted CABS rates among the small areas (school districts) in B.C.?  ii)  Is there any significant variation in the age-sex and morbidity adjusted CABS rates among the small areas (school districts) in B.C.?  iii) How much of the variation in (i) and (ii) above is explained by the independent variables 3 ? The independent variables examined, described in Chapter 4, were year, distance from a cardiologist (DSCAR), distance from an internist (DSINT), distance from a surgical centre (DSCENT), average family income in the school district (INC), employment rate (EMPRAT) and graduation rate (GRADRAT). 3  186  In order to determine if there was any small area variation in the small area annual age-sex adjusted CABS rates beyond that which could be ascribed to chance, the null model (intercept only) was compared to the saturated ("perfect") model by the likelihood ratio test. This gave a chi-square of 1082.80 with 74 degrees of freedom (p<0.0001). There was, therefore, significant variation in CABS rates among school districts and so the null hypothesis, i.e., that there was no small area variation other than that which could be ascribed to chance, was rejected. The second analysis estimated the amount of the variation in the annual age-sex adjusted CABS rate across school districts that could be accounted for by the independent variables. Information on these variables is shown in Tables 42 and 43 (Appendix B). Year was first entered as a categorical variable but a graph of the coefficients showed a quadratic form and it was found that year plus year-squared provided as good a fit. The model was then fitted for the main effects using a stepwise regression, with year and year-squared forced in, to determine the order in which the explanatory variables should be added. The model was then re-fitted adding each variable followed by its interactions with the variables already in the model. In order of importance, the variables included in the model were income, distance from a cardiologist and distance from a surgical centre. There were also first order interactions between distance from a cardiologist and year, yearsquared, income and centre; and between distance from a centre, and year, yearsquared and income. Distance from an internist, employment rate and graduation rate were found not to be important explanatory variables. Using the likelihood ratio test, no term in the final model could be omitted. Comparison of the final model with the null model gave a chi-square of 191.656 with 12 degrees of freedom  187  (df); p=0.0001. Comparison of the final model with the saturated model, gave a chi-square of 891.14 with 62 df (p=0.0001). R 2 was 0.21 4 . The parameter estimates for each important variable are shown in Table 43. The intercept of -11.5034 is simply the log of 1/100,000, the figure used to force the model into a poisson regression, plus the true intercept. The estimate for yearsquared is small compared to that for year so the effect of time is likely almost a linear one. The parameter for income (INC) is relatively large compared to the other parameter estimates; its negative sign indicates that people from regions with lower average incomes were more likely to have CABS. Similarly the negative sign of the parameter for distance from cardiologist (DSCAR) indicates that people in a region that contained a cardiologist were more likely to receive CABS. The positive sign for distance from centre (DSCEN) appears to indicate that people living in an area far from a surgical centre were more likely to receive the procedure. Intuitively it would appear that this surprising result likely arose because of the significant correlation between DSCAR and DSCEN. However, when all terms containing DSCAR were removed from the model the parameter estimate for DSCEN was still positive. It appears that the paradoxical effect of DSCEN is real. The relationship between both DSCAR and DSCEN and the CABS rates were influenced by their interactions with one another and with income. Of these three interaction terms the most influential was that between DSCAR and income; the parameter estimate is larger than that for the main DSCAR effect. To study the effect of time on the relative risk of receiving CABS, the parameter estimates were substituted into the model and all variables except year 4 R2 = (null model - final model)/(null model - saturated model).  188 and year-squared were held at zero. Compared to the reference year (1983) the relative risk5 (RR) of having the procedure rose from 1.09 in 1984 to 1.59 in 1988. Similarly when the parameter estimates for all those terms which include DSCAR were substituted into the model, and all other terms except year were held constant at reference levels, the relative risk for CABS was 0.93 in 1983 and 0.98 in 1988 for those people in school districts adjacent to cardiologists and 0.87 and 0.97 respectively for school districts far from cardiologists. Clearly the importance of the cardiologist in influencing the CABS rate decreased over the years, as did the importance of distance from centre. When all other variables except DSCEN and year were held constant at reference levels, the relative risk of DSCEN, compared to the reference of school districts which contained a centre, fell from 1.18 in 1983 to 0.99 in 1988 for those adjacent to a centre, and from 1.4 to 0.99 respectively for those far from a centre. For income the reference group was the school districts which had an average census family income of $30,000. When average family income was changed in increments of $10,000 the relative risks are 0.68, 0.47. 0.32 and 0.22 at $40,000 to $70,000 6 respectively. For incomes below the reference the relative risk was 1.45 for regions with an income of $20,000. A person from a school district, in a region with an average family income of $20,000 and a cardiologist, had 2.1 times the probability of receiving CABS than one from a district with an income of $40,000. Regional income clearly had a dramatic effect on the CABS rate. However, because of the interaction between DSCAR and income, the risk for  5 The relative risk is the ratio of the incidence rate for people exposed to a factor to the incidence rate for  those not exposed. In the reference year, when all other variables are zero, the RR is one.  6 There was one school district, West Vancouver (SD 45), which had an average 1986 census family income of $70,000. This income was obviously an outlier as the next highest income was $46,000. Repeating the stepwise regression without district 45 showed only minimal differences in the parameter estimates, so the school district was retained in subsequent analyses.  189  those with an income of $20,000 decreased with increased distance from a cardiologist, while the risk for those with an income of $70,000 increased with distance. In order to demonstrate the effect of mobility on the model, it was refitted using an indicator variable (ALBERTA) to represent those school districts which sent at least ten percent of their total CABS cases to Alberta (SD 1-4, 59, 60, 81, 86). The model was fitted in the same way as before with year, year-squared and ALBERTA forced in. Results are shown in Table 45 (Appendix B). In this model the interactions of DSCEN with year, year-squared, income and DSCAR were no longer significant. The Alberta indicator had the greatest influence on the CABS rate; the negative coefficient shows that people in the eight school districts in question had a lower probability of receiving CABS (in B.C.) than did those in the rest of the province. Their relative risk was only 0.47, when all other variables are at reference. In this model also, both income and distance from cardiologist had a negative effect. Those from low income areas had the greatest probability of receiving CABS in B.C., especially when their area of residence contained a cardiologist. In this model chi-square was 266.636 with 9 degrees of freedom (p<0.0001) and the independent variables explained 24 percent of the variance. The above models were then repeated using the CABS rate adjusted for morbidity in the school districts (Model C, Table 15). First, fully-saturated "perfect" models were run using the formula, described in Chapter 4. This model was repeated four times using different combinations of coronary artery disease diagnoses to estimate morbidity. The model with the least unexplained variance (the smallest chi-squares at 1259.050) was that in which all patients with a diagnosis of chronic IHD, including those who also had a diagnosis of AMI (ICD-9 411-414, 4292, with or without 410), were included; this diagnosis was assumed to  190  be the best estimate of morbidity. Results of the saturated models are shown in Table 46 (Appendix B). In order to check the above result, Pearson product-moment correlation coefficients were calculated between the CABS SIRs and the SIRs for the morbidity diagnoses. The highest correlation coefficient was found between CABS and the diagnosis of chronic IHD including AMI, and so this diagnosis was used to adjust the expected CABS rate. This correlation although seemingly low, at 0.20, was highly significant (p=0.0001). However the correlation between the SIRs for CABS and those for AMI (-0.05) was not significant. The model using the CABS rate adjusted for morbidity was tested in the same way as before. Stepwise regression, with year and year-squared forced in, was first carried out and then the model was refitted by entering the variables, and then their interactions, in the order suggested by the stepwise regression. In this model the first variable to be added after year and year-squared was distance from cardiologist, followed by income. No other variables were found to be significant and there were no interactions between variables. Chi-squared was 383 with 4 df (p=0.0001) so the significant difference in unadjusted rates cannot be explained by differences in morbidity between regions. In this model the independent variables explained 30 percent of the variance. Parameter estimates and chi-square results for the above morbidity adjusted model are shown in Table 47 (Appendix B). The effect of DSCAR is increased in that the relative risks are considerably lower than in the unadjusted model for those adjacent to (RR=0.77) and far from (RR=0.59) a cardiologist. Income is, however, not as important in this model. Those in an area with an average income of $20,000 had 1.24 times the probability of receiving CABS than those in an area with an average family income of $40,000.  191  It is notable that in this model the paradoxical effect of distance from centre disappears and that the model is considerably simpler. Much of the complexity of the variation in CABS rate is apparently mediated through complex relationships between diagnosis of CAD and the other variables. This model was also repeated with the indicator variable (ALBERTA) used to represent those school districts sending more than ten percent of cases to Alberta. Year, year-squared and ALBERTA were forced in as before. Again, ALBERTA had the greatest effect upon the CABS rate, followed by distance from cardiologist and income (Table 48, Appendix B). No other variables were found to be significant. The relative risk of CABS for residents of the 'ALBERTA' school districts was 0.54 in 1983, when all other variables were at reference. For residents in areas with a census family average income of $20,000 the relative risk of CABS in B.C. was 1.12. This relative risk fell as either income or distance from a cardiologist, or both, increased so that for residents of ALBERTA areas far from a cardiologist and with a regional income of $40,000 the relative risk was only 0.29. In this model the independent variables explained 34 percent of the variance. Chisquare was 437 with 5 degrees of freedom (p<0.0001). For all these the models the independent variables explain 34 percent or less of the variation in CABS rates among school districts over time. Sources of the unexplained residual variation may include interactions of age and sex with region and with the other variables in the model, the possibility of a more complex dependence on income, misclassification occurring because the ecological variables may not apply to the individuals having CABS within the region, other factors which may influence the CABS rate but which were not included in the model and non-independence of observations at the individual level. This latter problem may arise if one person's decision to undergo, or not undergo, the procedure influences the decisions of others in the same vicinity.  192  The above analyses based on age-sex adjusted rates identify factors which influence these rates, but cannot address the question of whether the age-sex effect itself varies from area to area, nor whether the influence of other factors, such as secular time, is different for different age-sex groups. To examine these questions a regression of age-sex-region-year specific rates against age, sex, year and region, and interactions among them, was performed. School districts were aggregated into four regions in two different ways, see Table 49 (Appendix B), for this analysis. This aggregation was done to reduce the otherwise unmanageable number of possible combinations of age-sex-year-region factors. This regression was carried out using models as shown in Chapter 4 (Table 15). Again, year was first entered as a categorical variable and a graph of the coefficients showed a quadratic form; year plus year-squared were found to provide as good a fit. Age (by five-year age groups) was entered as a categorical variable with the 65-69 group as the reference. Region (classified into geographical regions, and into remote, rural, urban or metropolitan areas, according to population density and distance from a referral hospital) 7 was entered as an indicator variable with the metropolitan area as the reference. First year, age and sex were entered, followed by their first and second-order interactions; the latter were not significant. Then region was entered and this was followed by the first order interactions with region. An attempt was then made to withdraw some of the variables which no longer appeared to be significant. The final models contained age, sex, year and region and all second order interactions between these variables (Tables 50 and 51, Appendix B). Chi-square for the remote/rural/urban/metropolitan model was 15232.299 with 55 degrees of freedom (p<0.0001). For the geographical model, chi-square was 14799.910 with 55 This classification was developed by the B.C. Royal Commission on Health Care and Costs. School district 89 straddled two regions and so was excluded from this analysis.  7  193  df (p<0.0001) but, for this model, age-groups one to five had to be aggregated because many of the age-sex-region groups in this age range had no cases. There are, therefore, significant differences in the time trends for CABS rates for different age-sex groups and for different regions when school districts are categorized into regions both geographically and on a population density and proximity to health services basis. The previous age-sex adjusted analysis produces effect estimates (e.g., for year) which are averages of the different effects for different age-sex-region groups. Such averaging is not problematic if the interactions are relatively small. However, some of the interactions are not negligible in magnitude and these indicate important departures from the assumption of homogeneity. In addition, because other factors such as income are only available as ecological variables, the interaction analysis could not be extended to include them. There may well be interactions between these factors and age and sex. This would cause additional unexplained variation. The number of factors in the final model makes interpretation of the effects of the age, sex, year or region on the CABS rate very tedious. However, a few examples will be given for the effect of sex in the remote/rural/urban/ metropolitan model. In 1983 in the metropolitan region, the relative risk for females age 65-69 compared to that for males is 0.2; in 1988 the relative risk is only 0.13. For females age 55-60 in remote regions in 1983, the relative risk of CABS is 0.32 compared to 0.06 in 1988. It appears that, at least in these two regions, the relative risk of CABS for females as compared to males has decreased over the years.  194  SUMMARY The sixty percent increase in the number of CABS performed in B.C. between 1979 and 1988 has exceeded the growth in the population. The increase in mean age and of co-morbid conditions points to an older and sicker population receiving the procedure in 1988 than in 1979. The annual standardized incidence ratios in many school districts show great variability over the ten-year period. Even so some school districts are characterized by consistent ratios; two of the three areas with the highest overall ratio have been consistently high throughout the ten years. Examination of the centres used by school districts for CABS procedures show definite referral areas. Between the centres there are differences in the proportion of open-heart surgery and revascularization overall devoted to CABS. Poisson regression analysis showed significant variations in CABS rates among the school districts of B.C. between 1983 and 1988. Some of this variation can be explained by interactions between age, sex, year and region, by regional health service factors (distance from a cardiologist and from a B.C. surgical centre) and by average family income within the region. The importance of income and distance from a cardiologist remain whether or not the CABS rate is adjusted for morbidity within the region (although they switched positions, with income becoming less important after adjustment) but following adjustment, distance from a surgical centre is no longer significant.  195 REFERENCES  Gillum, R.F., 1987. Coronary artery bypass surgery and coronary angiography in the United States, 1979-1983. American Heart Journal 113 (May):1255-1260. Peters, S. et al, 1990. Coronary artery bypass surgery in Canada. Health Reports 2(1):9-26. Preston, T.A., 1989. Assessment of coronary artery bypass surgery and percutaneous transluminal angioplasty. Intl I of Technology Assessment in Health Care 5:431-442.  196  CHAPTER 6 REVASCULARIZATION IN BRITISH COLUMBIA 1979-1988 DISCUSSION OF RESULTS AND POLICY IMPLICATIONS OF STUDY DISCUSSION DESCRIPTIVE STUDY Incidence Rates An increase in the numbers, and incidence rate, of CABS between the late-seventies and mid-eighties, has been well documented in the literature. This study shows a similar increase in B.C. although the size of the rate increase (1.28-fold) is markedly smaller than the four-fold increase shown by Preston (1989) in the United States, and somewhat smaller than the 1.39-fold increase in all Canadian provinces between 1981 and 1986 (Peters et al 1990). The relatively conservative increase in B.C., compared to other jurisdictions, likely reflects the provincial government's ability to restrict growth of this expensive procedure through budgetary controls. To compare the B.C. increase to that in other areas gives the impression that this province may be providing CABS at too low a rate; it is not keeping up with the population need. This supposition may, or may not, be true but it cannot be demonstrated by comparing incidence rates which are a reflection of 'what is' rather than of 'what ought to be'. Relating the present CABS rate to population need either requires population outcome studies or an estimate of the proportion of the provincial population who have the conditions for which CABS has been shown to be effective. The two-fold increase in incidence rates for the over-65 population found in this study is similar to the 2.38-fold increase found by Gillum in the United States between 1979 and 1983. Canadian authors have not reported the over-65  197  rate alone, but Anderson and Lomas (1989) reported a five-fold increase in the rate for the population of 70 and over between 1979 and 1985; in the present study there was over a four-fold increase for this age-group over a ten year period. It appears, therefore, that although the rates in the B.C. elderly have increased in the direction shown in the literature, the increases are more modest than in Ontario. One explanation for the differential increase within Canada, could be that the rates for the elderly in B.C. were higher at the beginning of the period than they were in Ontario. There is some evidence for this in the data on rates in the census metropolitan areas, reported by Peters et al (1990). The mean 1981-1986 rate for the 75 and over population in Vancouver was considerably higher than in either Toronto or London and only slightly lower than the rate in Hamiliton. However, because Peters et al do not report age-specific rates by year this explanation must remain only a supposition. However the rates in B.C. compare to those in other areas, the increase in the rates in the elderly are of concern because benefit of CABS to the over-65 population has never been demonstrated through either randomized trials or prospective controlled studies. As shown in the literature, the elderly have a higher operative mortality (Gersh et al 1983, Goldman et al 1987) and are at a greater risk of stroke (Goldman et al 1987), and of neuro-psychological problems post-operatively (Townes et al 1989) than are those under the age of 65. The risks are increased for elderly patients with co-morbid diseases and with increasing age (Goldman et al 1987). In this study almost 12 percent of patients over the age of 65 had either diabetes or COPD; if other disease such as renal failure and hypertension were also included the population with comorbidity would be much higher. The risks are, of course, additive and so for many elderly patients the risks of having CABS may well exceed the potential benefits  198  of the surgery. In addition, the increased use of resources post-CABS by the elderly (Goldman et al 1987) restricts the resources available for other patients and raises questions about the cost-effectiveness of the procedure in this age group. Because many of the above risks have been shown to increase with increasing age, the number of procedures performed on the "old" old, i.e., those over the age of 80, is of especial concern. The numbers of CABS performed on those over the age of 80 more than doubled in the second five years of the study; in 1988 those over 80 comprised one percent of the CABS population compared to 0.1 percent in 1979. This trend to increasing numbers of very old patients at high risk for a poor outcome undergoing a costly procedure raises the issues of quality of care and of appropriate use of resources.  Variations in Utilization by Region and by Centre The extreme variability of standardized incidence ratios for CABS within school districts over the ten years of the study shows, if nothing else, that conclusions about geographical variations in the B.C. CABS rate (and likely all rare procedures) cannot be drawn on the basis of one year's data. The extreme variability of the SIR's over time and across school districts, especially in areas with small populations, raises methodological questions about appropriate population sizes for small areas, about the number of years of data that should be studied in order to gain stable rates and about how best such variable data should be reported and analyzed. The two-fold variation (at the most conservative estimate) in SIRs among school districts when the data are aggregated over the ten year period, shows that geographic variations in the rate of CABS do exist over the long term and are not simply an artifact caused by transient changes in annual rates. These  199  variations are important both because of their clinical implications and because they may indicate inequities in care across the Province. These issues will be discussed later in this chapter. The differences between centres in the amount of resources committed to revascularization, type of patients and in referral patterns are of interest. Centre 3 shows a very different picture from the other two centres, in that it performs fewer CABS but devotes a greater percentage of its total open-heart surgeries to CABS and a greater percentage of its revascularization procedures to non-CABS procedures. This centre also draws its patients from a much more circumscribed area than do the other centres. It is not certain whether these differences are a result of the population served, referral patterns, policy within the hospital and "cardiac" community, to some other factors or to a combination of all of these. Because the areas served by Centre 3 tend to have SIRs that are consistently over one, and because Centre 3 has a higher proportion of total revascularization procedures devoted to non-CABs procedures, it appears likely that the school districts served by this centre utilize revascularization procedures as a whole at a much higher rate than do most other school districts in the provinces. This possible high level of revascularization appears to be at odds with the morbidity levels in some of the areas served by Centre 3. For school districts 61, 62 and 63 the annual SIRS for chronic CAD are consistently below one after 1983 while school districts 65, 66, 69 and 70 have chronic CAD SIRs that are consistently above one. There appear to be two opposing patterns here. High morbidity areas with high CABS rates and low morbidity areas with high CABS rates. It is possible that either the increased morbidity in an area results in a higher CABS rate or that the high CABS rate in an area results in 1 This is assuming that Centre 3 has the same referral areas for non-CABS procedures as it does for CABS.  200  lower morbidity but it appears unlikely that both these patterns would occur in school districts served by the same surgical centre. The fact that school districts 63 and 62 are adjacent to the Greater Victoria school district, which contains the surgical centre, likely influences the different pattern shown in these districts. It should be emphasized that the differences between morbidity rates and surgical rates in the school districts served by centre 3 were not tested for significance and therefore may be likely to have occurred by chance. However, the existence of these differences is puzzling and warrants further research. The differences noted above for centre 3, together with the differences between the hospitals, raise the question of whether the variations in small area CABS rates are related as much to the referral hospital itself as to the characteristics of the area from which the patient comes. Anderson and Lomas (1989) found a significant relationship between the referral centre providing care to Ontario counties and the rates of surgery on residents of those counties. Examination of the overall SIRs for the ten year period and the map showing those school districts which receive a plurality of CABS services from one centre, show that the majority of school districts served by Centre 2 have fewer CABS than expected over the ten years, all school districts served by Centre 3 have more CABS than expected while half the school districts served by Centre 1 have fewer CABS than expected and half have more. While it cannot be ruled out that these differences are due to referral patterns, it appears likely that physicians attached to different hospitals are using different indications when recommending CABS. If this is so, the issue of appropriateness of the procedure, and of under- or over-servicing, in different institutions is raised. Both Bunker (1988) and Anderson and Lomas (1989) point out that quality of care is based on the appropriateness of the procedure (the match between the patient's clinical condition and the correct indications for the procedure) as  201  much as on the technical skill demonstrated in the performance of the procedure. Consistent differences in rates between school districts served by different institutions likely indicate that the quality of care is being compromised. It is not possible to gauge the clinical implications of these results. In order to evaluate the clinical success of the revascularization program it would be important to know i) that all revascularization procedures performed were appropriate given the patient's condition and response to medical treatment; ii) that all patients whose condition required revascularization were offered it; iii) that all patients accepting surgery received it in a timely fashion as required by their condition and iv) that short and long-term outcomes for both medical and surgical patients were equivalent to, or better than those reported in the literature. The presence of significant variations in small area rates even after adjustment for morbidity and the presence of the differences, noted above, between surgical centres implies that either inappropriate procedures are being performed or patients requiring surgery are not receiving it, or both. However, it is also possible that patients in areas of high CAD morbidity but low CABS rates, may be receiving intensive medical treatment that mitigates their need for surgery or may be receiving relatively higher rates of angioplasty. Research to establish the clinical effect of variations in CABS rates is required. Comparisons of appropriateness of treatment and of outcomes in areas with different morbidity-CABS patterns would be one starting point 2 . Another route would be to compare similarities and differences in adjoining  2 School districts which typify the four possible morbidity-CABS patterns are: Golden (SD 18) with a high morbiditry rate and an extremly low CABS rate; Kitimat (SD 80) with a high morbidity rate and a high CABS rate; Kamloops (SD 24) which has a low morbidity rate and a low CABS rate and Lake Cowichan (SD 66) which has a low morbidity rate and a high CABS rate.  202  school districts which have the same sources of care but different CABS rates. For example, both the Central Okanagan (SD 23) and Vernon (SD 22) residents have access to the cardiologists in Kelowna and both receive about 50 percent of CABS procedures from Centre 2, but Central Okanagan has a consistently higher rate than Vernon. Comparisons of appropriateness of CABS between the different centres will also provide information about whether differences in appropriateness of care may be contributing to variations in area rates. Ultimately though it seems likely that, in order to fully understand how variations arise, research will have to be carried out at the level of the individual patient and physician.  Other Issues: Two other issues deserve comment. The first is the finding that a number of angioplasties appear to be performed in centres without surgical facilities. The guidelines for PTCA developed by Ryan et al (1988) state that PTCA should be carried out in a hospital with facilities for providing cardiac surgery. Certainly they should at the very least be carried out in a hospital which is not more than half-an-hour's drive away from cardiac surgical facilities; not all B.C. hospitals with angiography units are this close to the surgical centres. This issue likely requires discussion, and development of guidelines, by the cardiologists and cardiac surgeons in the province. If, on the other hand, these angioplasties simply represent miscoding, this situation also needs to be redressed. The second issue relates to the differences in coding between the cardiac centres. These differences reduce the validity of the hospital morbidity data-base and thereby reduce its usefulness. While adjustments can be made in the analysis to accommodate for differences in coding, many hospitals do not appear  203  to keep records of changes that they have made to coding practices so that information about coding changes within a particular hospital, and when they occurred, is often impossible to obtain. It is suggested that both the Hospital Medical Records Institute and the Health Records Association take responsibility for ensuring that the same procedures are coded in the same way across the Province. It is also suggested that medical records departments keep records of coding practices and changes within their department.  Regression Analysis: This study is the first to show that variation in small area CABS rates remain even after adjustment for small-area morbidity. It is also the only study into small-area variations in coronary artery bypass surgery which has examined the effect of area socio-economic characteristics on the CABS rate. The results indicate that, both before and after adjustment for morbidity, regional income and proximity to a cardiologist are the most important explanatory variables, while distance from centre was important in the models unadjusted for morbidity.  Income: The importance of income in explaining the variation is CABS rates is interesting. There is likely a stronger association than is shown here because income was only available for census years so only two values were entered for each school district. This would tend to dampen the income effect. The interaction of income with distance from cardiologist likely occurred because cardiologists in B.C. generally practice in the urban, higher income, areas. The finding that people in lower-income areas are more likely to receive CABS the nearer they are to a cardiologist while the reverse is true for those in high-  204  income areas, is difficult to understand and requires further study. The fact that income became relatively less important after the CABS rate was adjusted for morbidity, likely reflects the association between CAD and income reported in the literature. However, the higher incidence of CAD in lower income groups obviously does not explain all the income effect because income is still important in explaining the CABS rate even after adjustment for morbidity. The reasons for this are obscure but may relate to differences in the populations of low- and high-income areas with respect to life-style, willingness to comply with a medical regimen, or willingness to accept surgery. This area also requires further study. The negative association of income with CABS is a positive accolade for the Canadian health care system because it indicates that lower income is not a barrier to receipt of this expensive procedure. It would be interesting to know if there is a similar relationship in the United States. Given the importance of income in explaining the CABS rate, the unimportance of the employment rate is surprising. However, given the highly significant correlation between income and employment rate it is likely that employment rate did not appear to be important because the variables are colinear. Distance from Services The increasing importance of the cardiologist in affecting the CABS rate is not surprising given the progress made in diagnostic cardiology in recent years. The fact that the importance of distance from cardiologist increased after adjustment for morbidity may indicate that the proximity of a cardiologist has a greater impact after CAD has been diagnosed. As a whole, the results on the importance of proximity to a cardiologist appear to show that cardiologists make  205  more liberal decisions about the surgical treatment of CAD than do either internists or general practitioners. This finding is similar to that found by Brook et al (1988) but in the opposite direction to that found by Young et al (1987), who showed that cardiologists were more conservative than general practitioners in referring hypothetical patients with angina for angiography. It may be that decisions about hypothetical patients do not reflect real world practice, or that cardiologists are more risk averse prior to diagnosis of CAD than they are in patients known to have the disease. This area too needs further study. The fact that distance from an internist was not an important explanatory variable is surprising because, they are more available than cardiologists in the hinterland of British Columbia. Therefore, they are likely to be the specialists responsible for the care of patients with CAD in small urban. rural and remote areas. However, the colinearity of DSCAR and DSINT likely explains the seeming unimportance of distance from internist. The finding that distance from a surgical centre had a positive correlation with the CABS rate is very surprising; especially because on Vancouver Island, the school districts adjacent to Victoria have consistently high rates. However, the large positive interaction between DSCEN and DSCAR may mean that the positive effect of distance from centre is only active in proximity to a cardiologist. It is also possible that cardiologists close to a centre may be more willing to "nurse" sick cardiac patients along because they have access to cardiac facilities if required, while physicians far from cardiac centre may transfer patients while they are still in relatively good condition. Once these patients are admitted to a cardiac centre surgeons may be more likely to operate on them, again because of reluctance to return a sick patient to an area without cardiac facilities. It would be interesting to see whether the positive affect of DSCEN  206  remains over time as cardiologists and cardiac facilities become more widely distributed in the province.  Other Variables The effect of secular time in increasing the relative risk of CABS, implies that the Ministry of Health has been increasing the numbers of CABS funded at a rate faster than population growth. Even in the morbidity adjusted model the relative risk of CABS increases over time which indicates that the procedure is increasing faster than the increase in morbidity rates. The unimportance of the graduation rate in all models is not surprising because the variable used was not an indicator of the actual educational levels in the region but a rather uncertain proxy for it. Level of education has been shown to be negatively associated with the prevalence of risk factors for CAD (Millar and Wigle 1986) and to be a moderately important indicator of health status (Hay 1988). The lack of individual data on patients' socio-economic status, both in the hospital morbidity data base and in hospital records, hampers research into the relationship between socio-economic status and the utilization of hospital services. At best the independent variables tested explained only 34 percent of the variation in CABS rates across school districts. While some of the unexplained variance was likely due to interactions between the age-sex distribution in the school districts and the independent variables, there are obviously other factors at work which were not considered in this study. These factors likely include cultural factors and knowledge about, and attitude to, CABS of the general practitioners in the school districts. It should be emphasized that manipulation of health services factors (such as the supply of cardiologists) in order to alter small area rates, without consideration of the other factors which may also be  207  affecting that rate is unlikely to redress the inequities in the treatment of CAD which appear to exist in B.C.. Small area CABS rates appear to be a result of a complex relationship between area age-sex distribution, income, morbidity rates, and proximity to a cardiologist as well as other factors not yet identified. This study has avoided some of the methodological pitfalls found in other regional variations studies. The use of six years of data for the regression analysis, has reduced the effect of outliers, while the use of independent variables which were not directly related to total population size should have reduced the confounding effect of population. However, there are also a number of limitations to the study, which may affect its validity, reliability and the generalizability of the results, and these are described below.  LIMITATIONS Perhaps the most significant limitation in this study is that the validity and reliability of the hospital morbidity data base are not known. Within the data support section of the B.C. Ministry of Health it is believed that until April 1st, 1983, when the Ministry of Health ceased the management of the data base, the data were valid representations of the information on the patients' hospital charts. Ministry staff would meticulously check any incongruent or abnormal entries with the hospitals concerned. After 1983, when data management institutions took over data entry, the quality of the data likely declined. Errors in the database which would affect the validity of the results would be those relating to the coding of diagnosis, procedure3 , age, sex or postal code of the patient. While it seems unlikely that CABS would be missed as a procedure it is possible that it could be miscoded. Twenty-one patients over the ten-year period 3 There are difference between the hospitals in the way that CABS procedures are coded but these all relate to the number of vessels coded when the internal mammary artery is used.  208  were coded as receiving the procedure in a hospital which was not a cardiac surgical centre. These patients were kept in the study, but either the procedure or the hospital was likely miscoded. Miscodings for diagnosis, age or sex would be more difficult to spot if they fell within the approved alpha - numeric code for that field. It is likely though that any miscodings would be randomly distributed around the province and should, therefore, not systematically bias the results of the study. Studies to assess the reliability and validity of the hospital morbidity database are presently underway. The second major limitation is that the population used (all adults aged 20 and over) is not the true population at risk because a relatively small proportion of this population would have CAD and only a small proportion of those would be eligible for CABS. Therefore the incidence rates given are an underestimate of the incidence in those at risk. It is not known how well hospital discharges with a diagnosis of CAD reflect the actual morbidity rate for severe atherosclerosis and myocardial ischemia in an area. Systematic differences between areas in the tendency to give a CAD diagnosis or to admit CAD patients to hospital could bias the results. It seems likely that the morbidity is overestimated rather than underestimated (Kircher et al 1985) although people who have severe atherosclerosis and are not hospitalized, and those with silent myocardial ischemia who have never been diagnosed will not be included in the morbidity rates. If, as seems intuitively likely, general practitioners in rural and remote areas, who do not have access to cardiologists for referral, both diagnose CAD in more borderline patients and admit more CAD patients to hospital there will be a bias to higher morbidity rates in rural and remote school districts. The division of the morbidity diagnoses placed patients with unstable angina into the "chronic CAD" category; this was done in order to avoid putting  209  the AMI and unstable angina patients into one class. In retrospect it would have been better to have placed the unstable angina patients into a class of their own so that the rate for this category could be correlated with the CABS rate both alone and in combination with other CAD diagnoses. This may have produced a different combination of diagnoses to use in adjusting the CABS rate for morbidity. The presence of many interactions posed difficulties in interpreting the model and these interpretations may not be valid reflections of reality. While the greater simplicity of the morbidity-adjusted model appears to indicate that this is the more valid model, the potential limitations of the morbidity data cannot be ignored. Similarly, the method used in the regression analysis to account for the mobility to Alberta appears to be valid but may not be capturing this effect fully. It is possible that more school districts should have been included in the mobility variable because 15 percent of patients going to Alberta could not be assigned to a school district. The size of the data set and the number of analyses done increase the probability of significant results arising by chance. The 'p values' were not adjusted to account for the number of analyses, but had they been adjusted the results would still have been significant. The limitations posed by the data used to estimate income, employment rate and graduation rate has already been discussed above. The likely effect of the limited data for income and employment rate would be to reduce the relationship between the variable and the CABS rate. Also, because income is not linear, but was assumed to be so the amount of variation explained by the model is likely reduced. The use of ecological variables, rather than independent variables which relate directly to the individual receiving CABS, means that the relationship  210  between the CABS rate and the explanatory variables cannot be generalized to individuals. It is not known whether individuals from a low income area who receive CABS, themselves have a low income, nor whether patients living in the same school district as a cardiologist ever visited that cardiologist. To assume that the results of this study apply to individuals would be an ecological fallacy. The inability to identify patients having reoperations has likely weakened the relationship between the dependent and independent variables because there are likely different factors associated with having a reoperation than there are with having the initial CABS. Finally, the inability to identify the angioplasties in the province as a whole and in the school districts, meant that the angioplasty rate could not be included as either an independent or a dependent variable in the regression equation. It is highly likely that the small-area CABS rates, at least in the last few years, have been mediated by the angioplasty rates in the areas. The exclusion of the angioplasty rate as an independent variable has likely increased the amount of unexplained variance in the model. The inclusion of the angioplasty rate as a dependent variable would have allowed comparison of the factors which explain the rates of both types of revascularization procedures.  POLICY IMPLICATIONS This section will discuss the major policy implications of both the results from the analysis and from the literature review contained in chapters one through three. In order to put these policy implications into context, it is first necessary to understand how health care is organized in British Columbia.  211  Organization Of Health Care in B.C.: British Columbia administers a universal health insurance program which provides comprehensive coverage to all residents. The five essential elements of this program are that it is equitable, universal, comprehensive, accessible and portable between Provinces. The program is funded through Federal grants, provincial tax revenues and monthly premiums paid by residents. All medically required care is free to the patient at the time of service. Although some physicians are salaried, most are paid by fee-for-service according to a fee-schedule which is negotiated annually between the British Columbia Medical Association (BCMA) and the Ministry of Health. Most physicians are, in effect, self-employed professionals contracting out their services to the government. Hospitals in B.C. are funded by a global budget, negotiated annually with the Ministry of Health, which covers all operating costs and services provided by the hospital. Capital equipment is funded separately. In addition specialized and expensive services, such as cardiac surgery, are funded as a program on a per case basis but the number of cases funded per hospital is limited. Most hospitals only provide acute care and/or rehabilitation, although in recent years some hospital-based prevention programs have emerged. Health promotion and preventive services are generally provided by the Community Health Units which are run by either the Community and Family Health Service Division in the Ministry of Health or, in the large municipalities, by the Local Board of Health.  Implications of Literature and Present Study: The discussion on policy implications will centre around the two essential aspects of the B.C. (and Canadian) health care system which appear to  212  be in question with regard to CABS in this province. These aspects are accessibility and equity. Accessibility may be defined as the relationship between available resources and patient needs and reasonable expectations for obtaining entry in terms of location, time, effort and cost. Accessibility is also affected by continuity of care. A patient may be able to access the system via his family doctor but not be referred to a specialist, or other resources, when necessary. Equity is more difficult to define because in an environment in which the need for services may be composed of many different elements (e.g., physical condition, lifestyle, cultural beliefs etc), equal treatment is impossible. For the purpose of this discussion, equity in health care in B.C. will mean that all residents have an equal right to the best health outcome that can be achieved given their condition and the resources available in the system. This best "health outcome" can likely be achieved in a number of different ways and how it is achieved will depend on the condition and preferences of the patient, the judgement of the physician and the services available. However, implicit in the notion of equity of outcome is the notion of efficiency, both technical and allocative 4 . All health care activities involve an opportunity cost - the loss of services which could have been provided with the money spent on those activities. When resources are scarce, care which is not efficient is not equitable because the money lost through inefficiency is not available for other services. Therefore, in order to promote equity the government has to provide for both allocative and technical efficiency in the health care system.  4 Health care is technically efficient if it uses the least costly quantity and mix of services consistent with the desired outcome. Foe example, CABS would be technically efficient care for a patient with significant left main disease but not for one with single vessel disease and Class II angina. Allocative efficiency relates to the mix of services provided by the health care system. The system has allocative efficiency if resources are distributed so that it would be impossible, by reallocating resources, to make someone better off without simulataneously making others worse off.  213  Cost Effectiveness: The potential for inequity in care arises from the resources allocated to revascularization and other types of cardiac care in the province. Extrapolating from the annual numbers of CABS identified by this study, and the average cost per patient reported by Krueger et al (1991), it appears that about $19 million dollars are spent annually on providing CABS. At a rough estimate a further five million dollars are spent on providing angioplasty 5 . The, approximately, $24 million spent annually on revascularization is only well spent if all procedures provided are appropriate, if all required procedures are provided and if the same outcome could not be achieved at a lower cost. The answers to the above questions are likely not known. Research into the appropriateness, outcomes and cost-effectiveness of cardiac care in B.C. is required. At the present time it appears that the major focus in cardiac care is on acute care and on revascularization in particular. Balance in the cardiac program may require the government to promote research into non-invasive treatment of severe atherosclerosis, similar to that described by Graboys et al (1989) and Ornish et al (1990), for patients who are eligible for CABS or angioplasty but who do not choose invasive treatment as an option. The cost per patient in such a program may be high but is unlikely to be as high as the cost of CABS. A program like this could, by halting progression or causing regression of CAD, also be more effective in prolonging life and reducing morbidity than is CABS.  5 This figure is arrived at by multiplying the total number of non-CABS revascualrization procedures in 1988 by $5,500, the cost per procedure found by Reeder et al (1984) in the United States. Because of inflation since 1984 five million dollars is likely an underestimation of the cost of angioplasty.  214  The role of the Medical Services Plan in paying for costly, high-risk procedures used in situations for which efficacy has not been shown, may need to be evaluated. It may be difficult to impose sanctions on the use of procedures which are commonly accepted as effective by the general public, but the government should not in the long term continue to fund therapies which may pose greater risks to patients than are posed by the patients' conditions. Public education about the causes and treatment of CAD may, by putting CABS into its proper perspective as the treatment of choice for only relatively few conditions, reduce some of the pressure on the government to make policy decisions based on public demands. It would not seem unreasonable to require physicians who continue to use procedures in situations in which efficacy has not been shown, to provide proof of efficacy. Such proof would have to be by means of a randomized trial or a controlled study with an adequate sample size. Proof of efficacy is required for the use of CABS in the elderly, particularly those over the age of 75, for CABS in comparison to aggressive medical treatment, and for angioplasty in comparison to medical treatment.  Inappropriate Treatment: If equity is defined as an equal opportunity to achieve an optimum outcome, then inappropriate treatment is an equity problem both because of the misuse of scarce resources and because it puts patients at unnecessary risk and likely increases morbidity and mortality. 6 Leape (1989) concluded that geographical variations in procedure rates are indirect indicators of inappropriate surgery. Both the literature review, which showed that CABS is frequently used for conditions in which efficacy has not been demonstrated, and "This is likely true whether inappropriate treatment of CAD occurs through overuse or underuse of CABS  (  215  the finding of significant geographical variation in this study, indicate that there are likely inappropriate CABS being performed in British Columbia. Leape believes that inappropriate surgery results from inadequate information, and suggests that the role of the policy-maker in preventing inappropriate surgery is that of stimulating the development and dissemination of practice guidelines which are derived from evidence of effectiveness. He recommends that guidelines be used for operations which have unusual potential to harm a patient, or involve extensive use of resources, or which are controversial or which are suspected of inappropriate use. Coronary artery bypass surgery fits into all of these categories and so is a good candidate for a procedure for which guidelines should be used. In the same paper, Leape discusses the responsibilities of professional societies, academic institutions and government in developing guidelines. He believes that professional medical societies are the logical groups to develop guidelines, while academic centres play a role in developing an appropriate methodology for deriving guidelines. The role of government is to encourage and fund both the development of both the methodology and the actual guidelines themselves, but not to play a direct role in deriving guidelines. Leape sees medical practice guidelines as a classic "public good" from which everyone should benefit. Funding their development is therefore a logical government function; closer involvement in the development process would give government unprecedented control over medical practice and would lead the public to question whether the purpose of the guidelines was to improve quality of care or to reduce utilization and costs. Indications for the use of CABS have recently been identified (but not yet published) by a consensus group working in cooperation with the Canadian Cardiovascular Society. These guidelines, if acceptable to the B.C.  216  cardiovascular-surgeons, could be used as, or as a basis for, provincial practice guidelines. The advantage of using national guidelines is that by reflecting a broader scope of practice they will have more credibility than locally developed guidelines, and it may be cheaper to develop, revise and update one set of national guidelines than ten provincial sets. National guidelines on indications for use will also facilitate valid comparisons between provinces on utilization rates and outcomes of CABS. However, it is often the case that physicians at the local level believe that national guidelines do not reflect their circumstances, and so they want input into the guidelines that they will be expected to use. Local physician input will likely increase the acceptability of the guidelines to those using them but will decrease their usefulness in comparison. Once developed and accepted by the medical community, how should such guidelines be used? Leape suggests three ways in which guidelines could be used to reduce unnecessary surgery; to educate practitioners, to provide explicit criteria for quality assurance programs within hospitals, and as standards for physician payment. Possibly the most effective use of guidelines would be to require prospective assessment of appropriateness of CABS before the patient is admitted to the waiting list, similar to the prospective assessment used in many Health Maintenance Organizations in the United States. This would likely not be well accepted by the physicians but may be more acceptable than requiring them to submit proof that a case meets the guidelines before they are paid for it. However guidelines are used, they are likely to have a positive impact in reducing inappropriate CABS in the province. Applying the use of guidelines in all of the ways mentioned above would have the most impact. Cost savings would depend on the number of inappropriate surgeries being done, the numbers of appropriate patients who are presently not considered candidates for surgery but who may become so under the guidelines, and the cost of the  217  development of the guidelines and the monitoring procedures that would need to be set up. The benefits to patients could be enormous. Elimination of inappropriate surgery is but one aspect of improving technical efficiency and equity in the cardiac program. The presence of regional variations in care also leads to the suspicion that some patients who require CABS may not be receiving the procedure. Identifying such patients, however, is more complex than identifying those receiving inappropriate CABS. Chart review, and follow-up, of all patients receiving coronary angiography who are shown to have significant CAD, may identify some of these patients. However, because angiography itself has regional variations and a high proportion of inappropriate use (Chassin et al 1987) such an investigation is likely to be biased. Another method of identifying areas where patients may not be receiving adequate care for CAD would be to carry out studies of comparative outcomes (cardiac mortality and morbidity) in areas known to have high and low rates of CABS. Stimulation, and funding, of such research would appear to be part of the role of government.  Other: The marked differences in CABS rates in the school districts served by Centre 2 and Centre 3 may indicate problems in access to CABS for patients referred to Centre 2, inappropriate use in Centre 3 or both. It is possible that the differences in the percent of cardiac surgery devoted to CABS makes access to the procedure easier in Centre 3 but access for Vancouver Island patients requiring other types of cardiac surgery could be reduced. These differences between hospitals need to be addressed especially as they may contribute to the variation in small-area rates.  218  The importance of the proximity to a cardiologist in determining the CABS rate may indicate that residents in those school districts far from cardiologists have difficulty in accessing CABS. However, before decisions about the dispersion of these specialists throughout the province are made, some evidence is required that cardiologists are increasing the number of appropriate referrals, rather than increasing the CABS rate through inappropriate or equivocal referrals. Should it be found that cardiologists do indeed increase the appropriate referrals, and that there are school districts far from cardiologists where patients who need CABS do not appear to be referred for the procedure, then either visiting cardiologists or education programs for the local physicians could be instituted. The impact of income on the CABS rate, even after adjustment for morbidity, indicates clearly that differences in other population characteristics besides demographics, can affect need. Policy which addresses health service supply factors without also taking the income of the area into account may result in inequities. The results on income also show that a marked decrease in both CAD and CABS could likely be achieved by the introduction of measures to stimulate the economy in low-income areas and by the narrowing of the gap between rich and poor throughout the province.  Research: This study has identified a number of areas in which further research is required; this section will briefly address those which impact on revascularization policy. Perhaps the most important item on the research agenda is to establish the efficacy of CABS in the elderly, especially in those over 75 who are the group with the fastest growing rate. Given the importance of this endeavour, the  219  Ministry of Health may wish to encourage, and contribute funds to, such a study. The optimum study would be a randomized trial but a controlled study in which cases are matched for risk would also be appropriate. Of equal importance to the above, is the need to establish the role that angioplasty should play in the treatment of CAD. Efficacy, and costeffectiveness, of the procedure in relation to medical treatment needs to be established. Given the present interest in a healthy lifestyle shown by the general public, alternative therapies for the treatment of CAD, along the lines of the program described by Ornish et al (1990), would likely be well received. The promotion and evaluation of alternatives could reduce what may be an inappropriate reliance on the biomedical approach. The different patterns in CABS rates between school districts served by different centres, leads to the suspicion that the variation in small-area CABS rates could be primarily related to the referral centre. A study similar to that carried out by Anderson and Lomas (1989) in Ontario would test this hypothesis. With the start of the new cardiac surgery centre at the Royal Columbian Hospital in 1989, referral patterns for CABS will likely have changed in the Lower Mainland and in parts of the Interior. Comparison of referral patterns in these areas before, during and after the start-up period may help to establish the effect that referral patterns have on the CABS rate. As mentioned above, the clinical implications of the variations in CABS rates need to be explored through outcome studies on CAD patients in areas with different patterns of CABS and morbidity rates. Outcome studies would likely require links between the hospital morbidity database and the vital statistics database.  220  Finally, the new clinical data base proposed by the cardiac surgeons in B.C. will be a wonderful resource for answering clinical questions but will only be able to be used to address questions of efficacy or effectiveness if medical patients are included. If the database were set up similar to the CASS Registry so that all patients receiving coronary angiograms were entered, then preliminary studies on efficacy and/or on cost-effectiveness would be facilitated, although prospective studies would still be required. If possible the data-base should also contain information on risk factors.  221 REFERENCES:  Anderson, G.M. and J. Lomas, 1989. Regionalization of coronary artery bypass surgery: Effects on access. Medical Care 27 (March):288-296. Brook, R.H. et al, 1988. Diagnosis and treatment of coronary disease: Comparison of doctors' attitudes in the USA and the UK. The Lancet (April):750-753. Bunker, J.P., 1988. Is efficacy the gold standard for quality assessment? Inquiry 25:51. Chassin, M.R. et al, 1987. Does inappropriate use explain geographic variations in the use of health care services? TAMA 258 (November):2533-2537. Gillum, R.F., 1987. Coronary artery bypass surgery and coronary angiography in the United States, 1979-1983. American Heart Journal 113 (May):1255-1260. Graboys, T.B. et al, 1987. Results of a second-opinion program for coronary artery bypass surgery. TAMA 258 (September):1611-1614. Hay, D.I., 1988. Socioeconomic status and health status: a study of males in the Canada Health Survey. Social Science and Medicine 27(12):1317-1325. Kircher, T., J. Nelson and H. Burdo, 1985. The autopsy as a measure of accuracy of the death certificate. N. Engl J Med 313:1263-1269. Leape, L.L., 1989. Unnecessary Surgery. Health Services Research 24 (August):351-407. Millar, W.J. and D.T. Wigle, 1986. Socioeconomic disparities in risk factors for cardiovascular disease. CMAT 134 (January):127-32. Ornish, D. et al, 1990. Can lifestyle changes reverse coronary heart disease? The Lancet 336 (July):129-133. Peters, S. et al, 1990. Coronary artery bypass surgery in Canada. Health Reports 2(1):9-26. Preston, T.A., 1989. Assessment of coronary artery bypass surgery and percutaneous transluminal angioplasty. Intl T of Technology Assessment in Health Care 5:431-442. Ryan, T.J. et al, 1988. Guidelines for percutaneous transluminal coronary angioplasty. Circulation 78 (August):486-502.  222  Young, M.J. et al, 1987. Do cardiologist have higher thresholds for recommending coronary arteriography than family physicians? Health Services Research 22 (December):623-635.  223  APPENDIX A  224  TABLE 16 ANNUAL CABS PROCEDURES FOR B.C. AND OUT-OF PROVINCE RESIDENTS B.C. 1979-1988 Year  Isolated Coronary Artery Bypass Surgery Out-of-Province B.C. Residents #^% #^%  Total  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  922^99.1 924^99.3 950^98.8 1039^99.0 1193^97.8 1178^98.7 1163^98.5 1240^97.6 1304^98.5 1470^98.3  9^0.9 7^0.7 12^1.2 11^1.0 27^2.2 16^1.3 18^1.5 31^2.4 21^1.5 26^1.7  931 931 962 1050 1220 1194 1181 1271 1325 1496  Total  11383  178  11561  TABLE 17 ANNUAL CABS PROCEDURES BY SEX B.C. 1979-1988 Year  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  Isolated Coronary Artery Bypass Surgery Male Female #^% #^% 751^81 749^80 778^81 863^82 966^79 943^79 952^81 1031^81 1038^78 1234^82  180^19 182^20 184^19 187^18 254^21 251^21 229^19 240^19 287^22 262^18  Total 931 931 %2 1050 1220 1194 1181 1271 1325 1496  225 TABLE 18 ANNUAL MEAN AGE OF CABS PROCEDURES PER YEAR B.C. 1979-1988  Year  Male  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  57.08 57.23 57.72 59.08 58.86 60.42 61.30 61.95 62.05 62.52  Female  Total  59.00 59.93 59.69 60.75 63.05 63.55 64.60 64.81 65.02 65.21  57.45 57.76 58.10 58.86 59.73 61.08 61.94 62.49 62.69 62.99  TABLE 20 AGE-SEX SPECIFIC RATES FOR CORONARY ARTERY BYPASS SURGERY B.C. 1979-1988  Age Group  Age-Sex Specific Rates per 10,000 population Male Female  20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+  0.008 0.06 0.34 1.47 4.08 9.39 17.68 25.13 31.00 33.70 24.89 6.44  *There were no female cases in this age-group  0.00* 0.02 0.06 0.18 0.53 1.70 2.70 4.37 7.60 9.28 6.57 1.75  226 TABLE 22 ANNUAL CABS PROCEDURES PERFORMED ON SUB-GROUPS OF THE OVER-65 POPULATION B.C. 1979-1988 Year  Age Group 70-74  65-69 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  #  %  #  %  #  160 151 157 207 227 241 232 275 320 365  74.1 69.6 67.4 65.7 60.5 56.0 50.0 49.5 53.0 50.8  47 57 62 77 118 148 166 203 211 236  21.7 26.3 26.6 24.4 31.5 34.4 35.8 36.5 35.0 32.9  9 9 14 31 30 41 66 78 72 117  +11.2  -23.3  % change  75+^i %  Total 216 217 233 315 375 430 464 556 603 718  4.2 4.1 6.0 0.9 8.0 9.6 14.2 14.0 12.0 16.3 +12.1  TABLE 23 ANNUAL STANDARDIZED INCIDENCE RATIOS AND SEX ADJUSTED RATES FOR SUBGROUPS OF THE OVER-65 POPULATION FOR ISOLATED CORONARY ARTERY BYPASS SURGERY B.C. 1979-1988  Year  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 10-yr rate  65-69 years  Age Group 70-74 years^75+ years  SIR  Rate*  SIR  Rate  SIR  Rate  0.78 0.70 0.70 0.91 0.98 1.07 1.00 1.11 1.25 1.36  16.04 14.42 14.33 18.68 20.14 21.94 20.48 22.79 25.62 27.94  0.44 0.50 0.52 0.63 0.90 1.08 1.17 1.38 1.40 1.56  6.47 7.49 7.73 9.36 13.41 16.08 17.34 20.51 20.87 23.15  0.24 0.23 0.27 0.73 0.69 0.90 1.38 1.42 1.37 2.14  0.87 0.84 0.98 2.67 2.50 3.29 5.04 5.18 5.01 7.78  20.51  14.88  * Rate per 10,000 population = SIR x 10-year rate for age-group  3.64  227  TABLE 24 ANNUAL NUMBERS OF PATIENTS WITH COMORBIDITY RECEIVING ISOLATED CABS B.C. 1979-1988  Year 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  Diabetes  COPD 0 0 0 0 45 29 31 44 55 65  30 18 25 38 65 93 90 113 144 182  Total Comorbidity 30 18 25 38 110 122 121 157 199 247  % of Total Cases 3.20 1.93 2.60 3.62 9.02 10.22 10.25 12.35 15.02 16.51  Total Cases 931 931 %2 1050 1220 1194 1181 1271 1325 14%  TABLE 25 DISTRIBUTION OF COMORBIDITY BY AGE GROUP ISOLATED CORONARY ARTERY BYPASS SURGERY B.C. 1979-1988  Age Group  Diabetes  COPD  Total Comorbidity  Total CABS Cases in Age Group  %^of^Cases With Comorbidity  20-24 25-29 30-34 35-39 40-44 45-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89  0 1 2 11 13 58 82 .^114 167 188 124 31 7  0 0 0 1 4 8 28 46 54 57 53 15 3  0 1 2 12 17 66 110 160 221 245 177 46 10  1 10 49 182 419 845 1473 2033 2422 2355 1325 404 59  0.00 10.00 4.08 6.59 4.05 7.81 7.46 7.87 9.12 10.40 13.35 11.38 16.99  Total  798  269  1067  11568  9.22  228  TABLE 26 OUT-OF-PROVINCE REVASCULARIZATION SERVICES FOR B.C. RESIDENTS 1979-1988 Year  CABS  PTCA  Total  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  23 23 24 29 36 32 31 38 Not known 22  0 0 0 0 0 0 4 7 Not known 43  23 23 24 29 36 32 35 45 65  Total  258  54  312  Table 27 NUMBERS AND MEAN AGE OF B.C. RESIDENTS RECEIVING CABS IN ALBERTA Total  Year 1983 1984 1985 1986 1987 1988  N 5 8 3 3 28 28  Mean Age 58.2 64.6 60.6 62.0 61.8 61.8  Std 8.92 6.06 2.08 13.4 9.44 9.44  229  School District 1 2 3 4 7 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 52 54 55  TABLE 28 ANNUAL DISTRIBUTION OF ISOLATED CABS BY SCHOOL DISTRICT B.C. 1979-1988 1979 1980 1981 1982 1983 1984 1988 1985 1986 1987 1 4 0 0 12 5 5 5 3 0 4 16 3 2 0 1 0 2 13 28 13 1 9 0 2 0 1 0 12 21 14 47 17 23 125 22 44 18 39 23 6 5 2 2 0 1 4 3 1  2 1 1 0 6 2 1 4 1 0 5 9 2 2 1 0 0 2 8 22 17 1 9 1 0 4 2 3 20 17 13 50 24 33 106 22 42 13 41 31 20 2 7 6 0 1 4 2 0  1 4 3 0 8 1 3 8 1 1 5 12 2 2 0 2 0 2 15 40 17 1 7 5 0 1 2 2 13 21 17 64 21 32 127 11 50 16 27 29 14 5 5 1 0 0 2 0 6  1 4 4 00 5 1 1 6 3 1 10 13 1 2 1 1 0 1 10 74 17 0 6 6 1 0 1 2 18 26 19 59 32 28 123 35 42 23 43 18 15 5 3 3 0 1 3 3 0  1 5 2 0 9 2 0 5 2 1 7 21 3 1 0 4 0 3 16 71 17 1 6 6 0 1 3 1 12 24 23 83 32 31 151 21 54 26 47 37 9 5 10 4 2 1 3 4 0  1 5 2 2 12 2 5 5 4 0 15 27 3 5 0 4 0 4 23 81 22 1 4 5 1 5 5 4 14 19 20 70 20 29 135 28 57 16 40 39 24 12 6 5 1 2 3 4 2  0 5 4 1 11 5 3 1 4 0 13 22 3 2 0 2 0 2 15 99 17 1 9 10 0 0 3 6 13 24 22 74 27 34 182 17 46 18 52 38 12 8 6 5 0 1 4 2 0  1 5 1 0 10 8 4 10 0 1 15 23 3 5 0 2 0 1 18 83 18 1 11 11 1 3 3 2 16 28 33 111 33 32 152 17 58 17 45 33 22 6 5 5 0 2 5 5 0  0 3 4 0 19 6 5 9 6 3 19 32 7 2 0 1 0 5 19 76 23 0 16 5 0 2 3 3 32 26 25 94 33 50 167 18 60 16 40 37 14 10 12 5 0 2 12 6 1  1 3 4 1 13 8 0 14 6 5 13 32 4 3 0 1 0 4 32 85 29 1 12 10 1 2 1 5 23 33 29 113 32 50 206 23 71 23 50 46 23 12 14 9 1 1 7 2 1  Total 9 39 25 4 105 40 27 67 30 12 106 207 31 26 2 18 0 26 169 659 190 8 89 59 6 18 24 28 173 239 215 765 271 342 1474 214 524 186 424 331 159 70 70 45 4 12 47 31 11  230  TABLE 28 (contd.) ANNUAL DISTRIBUTION OF ISOLATED CABS BY SCHOOL DISTRICT B.C. 1979-1988  School^1979^1980^1981  1982  1983  1984  1985  1986  1987  1988  Tota 1  District 56^1^1^3 57^19^18^18 59^0^2^1 60^2^2^0 61^147^140^141 62^21^22^17 63^20^24^27 64^3^8^12 65^15^16^13 66^5^1^1 68^57^47^42 69^15^12^9 70^12^13^10 71^14^17^12 72^7^9^9 75^2^5^11 76^2^2^3 77^1^1^2 80^3^3^3 81^0^2^1 84^1^1^0 85^3^1^0 86^1^2^1 87^0^0^1 88^2^5^4 89^10^10^6 0 *^9^7^9  3 19 3 1 165 14 36 2 17 4 35 9 11 8 7 5 0 3 5 1 0 1 2 0 8 9 11  0 19 1 3 149 30 30 6 20 2 36 28 20 28 10 6 3 7 3 1 0 0 5 0 8 6 27  8 20 3 4 124 26 31 8 22 5 34 18 7 17 12 8 0 5 6 0 0 4 4 1 6 12 16  4 25 2 4 97 15 25 6 22 2 37 16 9 10 12 9 2 4 11 0 0 3 4 0 4 17 18  1 21 4 8 124 16 24 4 24 3 26 23 16 22 9 16 0 9 4 0 3 1 1 1 9 14 23  7 16 3 5 125 27 23 9 15 1 37 13 10 16 9 13 2 6 9 1 2 5 10 1 3 9 20  2 19 4 2 131 19 30 11 20 1 33 21 14 16 9 19 4 6 6 1 2 3 6 0 14 15 29  30 194 23 31 1343 207 270 69 184 25 384 164 122 160 93 94 18 44 53 7 9 21 36 4 63 108 169  1050  1220  1194  1181  1271  1325  1496  1155 6  Annual^931^931^962 Total *Represents non-residents of B.C.  231  TABLE 29 ANNUAL POPULATIONS, RATES AND STANDARDIZED INCIDENCE RATIOS PER 10,000 POPULATION BY SCHOOL DISTRICT B.C. 1979-1988 School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  1 Fernie 1 1 1 1 1 1 1 1 1  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  9347 9909 10570 11578 11791 11468 11259 10815 10324 9900  1 2 1 1 1 1 0 1 0 1  0.24 0.47 0.23 0.21 0.21 0.21 0.00 0.22 0.00 0.23  1.07 2.02 0.95 0.86 0.85 0.87 0.00 0.92 0.00 1.01  1.36 2.68 1.31 1.22 1.19 1.20 0.00 1.23 0.00 1.29  2 Cranbrook 2 2 2 2 2 2 2 2 2  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  12543 12943 13275 13949 14099 14284 14177 14090 11733 13522  4 1 4 4 5 5 5 5 3 3  0.66 0.16 0.65 0.60 0.73 0.71 0.69 0.70 0.45 0.41  3.19 0.77 3.01 2.87 3.55 3.50 3.53 3.55 2.56 2.22  3.79 0.93 3.70 3.43 4.16 4.03 3.94 3.97 2.56 2.33  3 Kimberley 3 3 3 3 3 3 3 3 3  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  5920 6109 6410 6368 6247 6266 6211 6035 9617 5415  0 1 3 4 2 2 4 1 4 4  0.00 0.25 0.74 0.98 0.49 0.49 0.97 0.25 0.93 1.05  0.00 1.64 4.68 6.28 3.20 3.19 6.44 1.66 4.16 7.39  0.00 1.44 4.20 5.60 2.79 2.79 5.53 1.42 5.30 5.97  232  TABLE 29 (contd.)  School District  Year  Population  Observed CABS  S .I.R.  Crude Rate  Age-Sex Adjusted Rate  4 Windermere 4 4 4 4 4 4 4 4 4  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  3738 4030 4380 4495 4599 4476 4539 4395 5069 4505  0 0 0 0 0 2 1 0 0 1  0.00 0.00 0.00 0.00 0.00 0.80 0.38 0.00 0.00 0.38  0.00 0.00 0.00 0.00 0.00 4.47 2.20 0.00 0.00 2.22  0.00 0.00 0.00 0.00 0.00 4.58 2.19 0.00 0.00 2.18  7 Nelson 7 7 7 7 7 7 7 7 7  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  13981 14412 14675 14850 15148 15063 14559 13885 13816 13878  12 6 8 5 9 12 11 10 19 13  1.42 0.7 0.93 0.57 1.00 1.33 1.23 1.13 2.11 1.42  8.58 4.16 5.45 3.37 5.94 7.97 7.56 7.20 13.75 9.37  8.08 3.96 5.33 3.26 5.69 7.56 7.00 6.43 12.03 8.09  9 Castlegar 9 9 9 9 9 9 9 9 9  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  7694 7956 8260 8209 8244 8165 8105 7985 7729 7711  5 2 1 1 2 2 5 8 6 8  1.07 0.42 0.21 0.21 0.41 0.40 1.00 1.63 1.24 1.62  6.50 2.51 1.21 1.22 2.43 2.45 6.17 10.02 7.76 1037  6.10 2.41 1.20 1.17 2.31 2.29 5.71 9.27 7.08 9.25  233  TABLE 29 (contd.) School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  10 Arrow Lakes 10 10 10 10 10 10 10 10 10  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2941 3027 3175 3244 3242 3256 3209 3100 3004 2942  5 1 3 1 0 5 3 4 5 0  2.67 0.52 1.58 0.49 0.00 2.45 1.46 1.94 2.55 0.00  17.00 3.30 9.45 3.08 0.00 15.36 9.35 12.90 16.64 0.00  15.24 2.94 9.00 2.78 0.00 13.97 8.34 11.07 14.54 0.00  11^Trail 11 11 11 11 11 11 11 11 11  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  15149 15367 15910 15741 15354 15171 14947 14370 13927 13788  5 4 8 6 5 5 1 10 9 14  0.50 0.40 0.79 0.60 0.51 0.51 0.10 1.05 0.97 1.50  3.30 2.60 5.03 3.81 3.26 3.30 0.67 6.96 6.46 10.15  2.84 2.28 4.52 3.44 2.90 2.94 0.59 6.01 5.52 8.56  12 Grand Forks 12 12 12 12 12 12 12 12 12  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  4618 4775 5080 5143 5221 5214 5213 5060 5068 5081  3 1 1 3 2 4 4 0 6 6  0.90 0.29 0.27 0.80 0.53 1.06 1.06 0.00 1.60 1.58  6.50 2.09 1.97 5.83 3.83 7.67 7.67 0.00 11.84 11.81  5.12 1.66 1.54 4.57 3.01 6.05 6.02 0.00 9.14 9.00  234  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  13 Kettle Valley 13 13 13 13 13 13 13 13 13  1979  2132  0  0.00  0.00  0.00  1980 1981 1982 1983 1984 1985 1986 1987 1988  2148 2170 2221 2197 2223 2174 2130 2099 2077  0 1 1 1 0 0 1 3 5  0.00 0.65 0.63 0.64 0.00 0.00 0.63 2.00 3.36  0.00 4.61 4.50 4.55 0.00 0.00 4.69 14.29 24.07  0.00 3.68 3.58 3.63 0.00 0.00 3.58 11.4 19.13  14 S. Okanagan 14 14 14 14 14 14 14 14 14  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  8793 9389 10115 10545 10735 10590 10467 10405 10131 10220  4 5 5 10 7 15 13 14 19 13  0.58 0.68 0.61 1.16 0.80 1.72 1.49 1.61 2.17 1.47  4.55 5.33 4.94 9.48 6.52 14.16 12.42 13.46 18.75 12.72  3.28 3.86 3.47 6.60 4.54 9.82 8.49 9.20 12.36 8.36  15 Penticton 15 15 15 15 15 15 15 15 15  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  18716 19144 19925 20618 21262 21783 21586 21935 21999 22341  16 9 12 13 21 27 22 23 32 32  1.23 0.67 0.83 0.87 1.37 1.72 1.40 1.43 1.97 1.93  8.55 4.70 6.02 6.31 9.88 12.39 10.19 10.49 14.55 14.32  7.02 3.81 4.73 4.97 7.80 9.79 7.99 8.17 11.22 11.01  235 TABLE 29 (contd.)  School District  Year  Population  Observed CABS  S .I.R.  Crude Rate  Age-Sex Adjusted Rate  16 Keremeos 16 16 16 16 16 16 16 16 16  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2333 2387 2585 2644 2623 2619 2617 2580 2563 2694  3 2 2 1 3 3 3 2 7 4  1.52 1.02 0.94 0.46 1.38 1.36 1.35 0.90 3.17 1.79  12.86 8.38 7.74 3.78 11.44 11.45 11.46 7.75 27.31 14.85  8.68 5.82 5.38 2.61 7.88 7.77 7.70 5.11 18.05 10.18  17 Princeton 17 17 17 17 17 17 17 17 17  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  3112 3142 3170 3265 3264 3348 3343 3305 3280 3317  2 2 2 2 1 5 2 5 2 3  1.07 1.05 1.04 0.99 0.48 2.34 0.91 2.39 0.96 1.41  6.43 6.37 6.31 6.13 3.06 14.93 5.98 15.13 6.10 9.04  6.10 5.97 5.94 5.62 2.71 13.32 5.18 13.64 5.45 8.03  18^Golden 18 18 18 18 18 18 18 18 18  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  3796 3935 4165 4109 4099 4230 4433 4500 4661 4870  0 1 0 1 0 0 0 0 0 0  0.00 0.56 0.00 0.53 0.00 0.00 0.00 0.00 0.00 0.00  0.00 2.54 0.00 2.43 0.00 0.00 0.00 0.00 0.00 0.00  0.00 3.20 0.00 3.05 0.00 0.00 0.00 0.00 0.00 0.00  236 TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  19 Revelstoke 19 19 19 19 19 19 19 19 19  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  6037 6265 6460 6588 6497 6397 6194 6010 5927 5774  1 0 2 1 4 4 2 2 1 1  0.36 0.00 0.70 0.34 1.34 1.36 0.69 0.69 0.35 0.35  1.66 0.00 3.10 1.52 6.16 6.25 3.23 3.33 1.69 1.73  2.07 0.00 4.00 1.91 7.63 7.76 3.94 3.92 2.00 2.00  21 ArmstrongSpallumcheen 21 21 21 21 21 21 21 21 21  1979  4072  2  0.78  4.91  4.47  1980 1981 1982 1983 1984 1985 1986 1987 1988  4287 4485 4687 4713 4758 4776 4720 4795 4794  2 2 1 3 4 2 1 5 4  0.75 0.69 0.33 0.98 1.27 0.63 0.31 1.59 1.25  4.67 4.46 2.13 6.37 8.41 4.19 2.12 10.43 8.34  4.30 3.96 1.89 5.61 7.26 3.61 1.76 9.05 7.10  22 Vernon 22 22 22 22 22 22 22 22 22  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  26410 27267 28370 29525 29809 30263 30504 30375 30398 30783  13 8 15 10 16 23 15 18 19 32  0.81 0.48 0.87 0.56 0.87 1.22 0.78 0.91 0.96 1.59  4.92 2.93 5.29 3.39 5.37 7.60 4.92 5.93 6.25 10.4  4.61 2.74 4.98 3.18 4.98 6.97 4.44 5.21 5.47 9.06  237 TABLE 29 (contd.) School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  23 Central Okanagan 23 23 23 23 23 23 23 23 23  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  52846 56303 59925 61691 63062 64309 64786 65880 66884 68525  28 22 39 74 71 81 99 83 76 87  0.78 0.57 0.97 1.77 1.66 1.85 2.23 1.81 1.63 1.81  5.30 3.91 6.51 12.00 11.26 12.6 15.28 12.60 11.36 12.70  4.47 3.25 5.51 10.11 9.47 10.56 12.69 10.33 9.28 10.34  24 Kamloops 24 24 24 24 24 24 24 24 24  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  46669 48512 51320 52425 52846 53019 52603 51170 52921 53119  13 17 17 17 17 22 17 18 23 29  0.57 0.72 0.69 0.67 0.65 0.82 0.62 0.65 0.81 0.99  2.79 3.50 3.31 3.24 3.22 4.15 3.23 3.52 4.35 5.46  3.26 4.11 3.95 3.79 3.69 4.65 3.53 3.69 4.60 5.63  26 .N.Thompson 26 26 26 26 26 26 26 26 26  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2967 3072 3150 3096 3045 3080 3117 2975 3042 2927  1 1 1 0 1 1 1 1 0 1  0.72 0.71 0.70 0.00 0.68 0.66 0.65 0.67 0.00 0.65  3.37 3.26 3.17 0.00 3.28 3.25 3.21 3.36 0.00 3.42  4.13 4.07 3.99 0.00 3.85 3.77 3.68 3.83 0.00 3.68  238 TABLE 29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  27 CaribooChilcotin 27 27 27 27 27 27 27 27 27  1979  19651  9  1.01  4.58  5.75  1980 1981 1982 1983 1984 1985 1986 1987 1988  20695 22250 23194 23589 24056 23894 23140 24227 23579  9 7 6 6 4 9 11 16 13  0.96 0.67 0.55 0.53 0.34 0.74 0.91 1.27 1.04  4.35 3.15 2.59 2.54 1.66 3.77 4.75 6.60 5.51  5.46 3.82 3.15 3.02 1.92 4.22 5.19 7.26 5.92  28 Quesnel 28 28 28 28 28 28 28 28 28  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  13032 13668 14175 14803 14912 15093 15198 15070 15570 15415  0 1 5 6 6 5 10 11 5 10  0.00 0.16 0.75 0.87 0.85 0.68 1.32 1.44 0.65 1.28  0.00 0.73 3.53 4.05 4.02 3.31 6.58 7.30 3.21 6.49  0.00 0.89 4.27 4.99 4.83 3.90 7.54 8.22 3.70 7.31  29 Lilooet 29 29 29 29 29 29 29 29 29.  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2425 2491 2775 2768 2792 2861 2916 2450 2926 2946.  2 0 0 1 0 1 0 1 0 1  1.49 0.00 0.00 0.65 0.00 0.62 0.00 0.71 0.00 0.61  8.25 0.00 0.00 3.61 0.00 3.50 0.00 4.08 0.00 3.39  8.51 0.00 0.00 3.73 0.00 3.54 0.00 4.07 0.00 3.48  239  TABLE 29 (contd.) School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  30 S.Cariboo 30 30 30 30 30 30 30 30 30.  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  5068 5187 5100 5395 5395 5503 5241 4880 4849 4677.  0 4 1 0 1 5 0 3 2 2.  0.00 1.34 0.34 0.00 0.33 1.61 0.00 1.02 0.70 0.70  0.00 7.71 1.96 0.00 1.85 9.09 0.00 6.15 4.12 4.28  0.00 7.65 1.97 0.00 1.88 9.16 0.00 5.8 4.01 4.01  31 Merritt 31. 31. 31. 31. 31. 31. 31. 31 31  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  5789. 5840. 6030. 6086. 6130. 6113. 6234. 6405. 6504 6463  1. 2. 2. 1. 3. 5. 3. 3. 3 1  0.32 0.63 0.60 0.31 0.89 1.47 0.86 0.81 0.83 0.27  1.73 3.42 3.32 1.64 4.89 8.18 4.81 4.68 4.61 1.55  1.82 3.58 3.44 1.74 5.07 8.36 4.93 4.63 4.75 1.56  32 Hope 32 32 32 32 32 32 32 32 32  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  4566 4674 4895 5036 5060 5029 5150 4795 4893 4994  0 3 1 2 1 4 6 2 3 5  0.00 0.98 0.31 0.62 0.31 1.23 1.79 0.64 0.93 1.53  0.00 6.42 2.04 3.97 1.98 7.95 11.65 4.17 6.13 10.01  0.00 5.57 1.79 3.51 1.75 7.04 10.21 3.67 5.29 8.74  240 TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  33 Chilliwack 33 33 33 33 33 33 33 33 33  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  28118 29322 30465 30872 30903 32833 31722 32090 32765 34036  12 20 13 18 12 14 13 16 32 22  0.66 1.05 0.66 0.90 0.60 0.67 0.63 0.77 1.50 0.99  4.27 6.82 4.27 5.83 3.88 4.26 4.10 4.99 9.77 6.46  3.76 5.96 3.74 5.15 3.41 3.80 3.60 4.39 858 5.65  34 Abbotsford 34 34 34 34 34 34 34 34 34  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  31688 34065 36860 40162 41950 43640 45070 45400 47535 49906  21 17 21 26 24 19. 24 28 26 33  1.13 0.86 0.99 1.16 1.02 0.77 0.94 1.07 0.97 1.18  6.63 4.99 5.70 6.47 5.72 4.35 5.33 6.17 5.47 6.61  6.46 4.90 5.66 6.58 5.79 4.41 5.37 6.12 5.50 6.73  35 Langley 35 35 35 35 35 35 35 35 35  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  35026 36850 38780 39504 41110 43224 44967 46855 49500 51395  14 13 17 19 23 20 22 32 25 29  0.77 0.68 0.84 0.92 1.07 0.88 0.93 1.28 0.97 1.07  4.00 3.53 4.38 4.81 5.59 4.63 4.89 6.83 5.05 5.64  4.39 3.89 4.82 5.22 6.07 5.04 5.29 7.31 551 6.12  241  TABLE 29 (contd.) School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  36 Surrey 36 36 36 36 36 36 36 36 36  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  96867 103734 109725 114954 120920 126917 131855 137170 145841 153070  47 50 64 59 83 70 74 110 94 113  0.88 0.88 1.07 0.94 1.27 1.02 1.03 1.46 1.19 1.37  4.85 4.82 5.83 5.13 6.86 5.52 5.61 8.02 6.45 7.38  4.99 5.01 6.09 5.38 7.23 5.82 5.89 8.33 6.80 7.79  37 Delta 37. 37. 37. 37. 37. 37. 37. 37. 37.  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  42638 45159. 46915. 48240. 49741. 50579. 51604. 52920. 54161. 55178.  17 24. 21. 32. 32. 20. 27. 33. 33. 32.  0.87 1.13 0.94 1.37 1.31 0.79 1.02 1.19 1.16 1.07  3.99 5.31 4.48 6.63 6.43 3.95 5.23 6.24 6.09 5.80  4.95 6.45 5.34 7.79 7.46 4.50 5.83 6.80 6.62 6.12  38 Richmond 38 38 38 38 38 38 38 38 38  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  59641 63945 66440 68740 71182 73659 76352 78590 81541 84116  23 33 32 28 31 29 34 31 50 50  0.76 1.01 0.93 0.79 0.84 0.76 0.85 0.74 1.16 1.11  3.86 5.16 4.82 4.07 4.36 3.94 4.45 3.94 6.13 5.94  4.32 5.76 5.31 4.49 4.78 4.30 4.83 4.22 6.59 6.35  242  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  39 Vancouver 39 39 39 39 39 39 39 39 39  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  318610 325747 329515 334452 337823 341847 346142 349095 350030 356688  125 106 127 123 151 135 182 151 167 207  0.65 0.55 0.65 0.63 0.78 0.70 0.94 0.78 0.84 1.02  3.92 3.25 3.85 3.68 4.47 3.95 5.26 4.33 4.77 5.80  3.69 3.11 3.73 3.61 4.44 3.97 5.36 4.43 4.77 5.83  40 New Westminster 40 40 40 40 40 40 40 40 40  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  29722 30556 31270 31214 31645 31890 32619 32875 32535 33568  22 22 11 35 21 28 17 17 18 23  1.19 1.18 0.58 1.89 1.15 1.55 0.93 0.94 0.93 1.18  7.40 7.20 3.52 11.21 6.64 8.78 5.21 5.17 5.53 6.85  6.77 6.70 3.30 10.77 6.53 8.81 5.30 5.34 5.32 6.71  41 Burnaby 41 41 41 41 41 41 41 41 41  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  98365 100905 102800 104828 106972 109694 111604 113510 114094 116397  44 42 50 42 54 57 46 57 60 71  0.77 0.72 0.84 0.69 0.87 0.91 0.72 0.87 0.90 1.04  4.47 4.16 4.86 4.01 5.05 5.20 4.12 5.02 5.26 6.10  4.38 4.09 4.79 3.95 4.98 5.16 4.10 4.98 5.1 5.92  243  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  42 Maple Ridge 42 42 42 42 42 42 42 42 42  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  24232 25100 25650 25958 27192 28104 28951 30100 32563 34947  18 13 16 23 26 16 18 17 16 23  1.4 0.99 1.20 1.69 1.85 1.10 1.20 1.10 0.98 1.31  7.43 5.18 6.24 8.86 9.56 5.69 6.22 5.65 4.91 6.58  8.0 5.66 6.84 9.65 10.54 6.29 6.85 6.26 5.56 7.47  43 Coquitlam 43 43 43 43 43 43 43 43 43  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  62661 66616 69950 70887 73756 77171 78809 80700 84253 88087  39 41 27 43 47 40 52 45 40 50  1.36 1.34 0.85 1.32 1.39 1.14 1.42 1.17 0.99 1.17  6.22 6.15 3.86 6.07 6.37 5.18 6.60 5.58 4.75 5.68  7.76 7.63 4.87 7.50 7.90 6.50 8.08 6.65 5.64 6.69  44 N.Vancouver 44 44 44 44 44 44 44 44 44.  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  69154 71366 72910 73892 74701 75932 77402 78840 80291 82325  23 31 29 18 37 39 38 33 37 46  0.64 0.83 0.75 0.46 0.93 0.96 0.91 0.77 0.84 1.01  3.33 4.34 3.98 2.44 4.95 5.14 4.91 4.19 4.61 5.59  3.64 4.73 4.30 2.62 5.29 5.47 5.20 4.37 4.77 5.73  244  TABLE 29 (contd.) School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  45 W.Vancouver 45 45 45 45 45 45 45 45 45  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  28824 29456 29710 29994 30618 31163 31683 31915 31830 32495  6 20 14 15 9 24 12 22 14 23  0.29 0.95 0.65 0.69 0.41 1.06 0.52 0.94 0.58 0.94  2.08 6.79 4.71 5.00 2.94 7.70 3.79 6.89 4.40 7.08  1.65 5.39 3.71 3.93 2.31 6.05 2.95 5.34 3.31 5.36  46 Sunshine Coast 46 46 46 46 46 46 46 46 46  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  9604 10272 11275 11738 12026 12182 12389 12235 12291 12511  5 2 5 5 5 12 8 6 10 12  0.71 0.27 0.63 0.59 0.57 1.35 0.88 0.66 1.08 1.28  5.21 1.95 4.43 4.26 4.16 9.85 6.46 4.90 8.14 9.59  4.07 1.54 3.56 3.37 3.28 7.72 5.03 3.75 6.16 7.32  47 Powell River 47 47 47 47 47 47 47 47 47  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  12219 12427 12830 12839 12832 12777 12744 12835 12348 12531  2 7 5 3 10 6 6 5 12 14  0.27 0.93 0.64 0.38 1.25 0.75 0.74 0.60 1.51 1.72  1.64 5.63 3.90 2.34 7.79 4.70 4.71 3.90 9.72 11.17  1.54 5.29 3.63 2.16 7.14 4.26 4.21 3.42 8.58 9.83  245  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S .I.R.  Crude Rate  Age-Sex Adjusted Rate  48 Howe Sound 48 48 48 48 48 48 48 48 48  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  8390 8886 9485 9578 9354 9764 9687 9600 10825 11395  2 6 1 3 4 5 5 5 5 10  0.55 1.55 0.25 0.71 0.94 1.12 1.09 1.08 0.99 1.89  2.38 6.75 1.05 3.13 4.28 5.12 5.16 5.21 4.62 8.78  3.16 8.84 1.40 4.02 5.38 6.38 6.21 6.17 5.64 10.75  49 Central Coast 49 49 49 49 49 49 49 49 49  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2271 2222 1895 1889 1928 1963 2016 2010 2146 2102  0 0 0 0 2 1 0 0 0 1  0.00 0.00 0.00 0.00 2.06 1.01 0.00 0.00 0.00 0.98  0.00 0.00 0.00 0.00 10.37 5.09 0.00 0.00 0.00 4.76  0.00 0.00 0.00 0.00 11.75 5.76 0.00 0.00 0.00 5.59  50 Queen Charlotte 50 50 50 50 50 50 50 50 50  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  3266 3382 3520 3571 3557 3537 3588 3585 3690 3630  1 1 0 1 1 2 1 2 2 1  0.91 0.86 0.00 0.79 0.78 1.53 0.72 1.44 1.44 0.72  3.06 2.96 0.00 2.8 2.81 5.65 2.79 5.58 5.42 2.75  5.18 4.91 0.00 4.49 4.45 8.70 4.13 8.20 8.20 4.10  246  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  52 Prince Rupert 52 52 52 52 52 52 52 52 52  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  10709 11108 11635 11817 11941 11973 11830 11535 11799 11843  4 4 2 3 3 3 4 5 12 7  0.85 0.85 0.42 0.61 0.6 0.59 0.77 0.94 2.27 1.28  3.74 3.60 1.72 2.54 2.51 2.51 3.38 4.33 10.17 5.91  4.85 4.82 2.38 3.50 3.41 3.37 4.41 5.36 12.93 7.31  54 Smithers 54 54 54 54 54 54 54 54 54  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  7085 7641 8420 8959 9018 9168 9169 8870 9211 9349  3 2 0 3 4 4 2 5 6 2  1.03 0.68 0.00 0.84 1.10 1.06 0.52 1.36 1.56 0.50  4.23 2.62 0.00 3.35 4.44 4.36 2.18 5.64 6.51 2.14  5.90 3.86 0.00 4.8 6.26 6.05 2.96 7.77 8.88 2.84  55 Burns Lake 55 55 55 55 55 55 55 55 55  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  4443 4704 4780 4714 4704 4644 4708 4645 4829 5001  1 0 6 0 5 2 0 0 1 1  0.51 0.00 2.88 0.00 2.29 0.90 0.00 0.00 0.44 0.42  2.25 0.00 12.55 0.00 10.63 4.31 0.00 0.00 2.07 2.00  2.89 0.00 16.44 0.00 13.07 5.14 0.00 0.00 2.50 2.38  247  TABLE 29 (contd.) School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  56 Nechako 56 56 56 56 56 56 56 56 56  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  8713 9021 9250 9675 9902 9730 9650 9150 9380 9431  1 1 3 3 0 8 4 1 7 2  0.26 0.25 0.77 0.7 0.00 1.79 0.89 0.23 1.56 0.43  1.15 1.11 3.24 3.1 0.00 8.22 4.15 1.09 7.46 2.12  1.45 1.41 4.36 3.99 0.00 10.18 5.08 1.32 8.91 2.46  57 Prince George 57 57 57 57 57 57 57 57 57  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  51937 52369 54855 56306 56797 57284 57639 56945 58335 57734  19 18 18 19 19 20 25 21 16 19  1. 0.94 0.88 0.88 0.86 0.87 1.05 0.87 0.64 0.74  3.66 3.44 3.28 3.37 3.35 3.49 4.34 3.69 2.74 3.29  5.7 5.34 5.03 5.03 4.88 4.96 5.99 4.94 3.66 4.24  59 Peace River S. 59 59 59 59 59 59 59 59 59  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  12381 13414 14365 14621 15487 16980 17678 17660 17795 17374  0 2 1 3 1 3 2 4 3 4  0.00 0.31 0.15 0.44 0.14 0.4 0.26 0.5 0.38 0.5  0.00 1.49 0.7 2.05 0.65 1.77 1.13 2.27 1.69 2.3  0.00 1.76 0.86 2.5 0.81 2.26 1.47 2.83 2.16 2.87  248 TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  60 Peace River N.  1979  13225  2  0.39  1.51  2.2  60 60 60 60 60 60 60 60 60  1980 1981 1982 1983 1984 1985 1986 1987 1988  14427 16165 16325 16236 15934 15609 15600 15638 15384  2 0 1 3 4 4 8 5 2  0.36 0.00 0.16 0.48 0.62 0.61 1.18 0.72 0.28  1.39 0.00 0.61 1.85 2.51 2.56 5.13 3.2 1.3  2.04 0.00 0.93 2.72 3.53 3.47 6.71 4.08 1.6  61 Greater Victoria 61 61 61 61 61 61 61 61 61  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  127330 130019 132515 134125 135625 136786 138433 138880 138773 140271  147 140 141 165 149 124 124 123 125 131  1.79 1.69 1.68 1.98 1.79 1.49 1.48 1.47 1.4 1.45  11.54 10.77 10.64 12.3 10.99 9.07 8.96 8.86 9.01 9.34  10.21 9.62 11.29 10.21 8.52 8.44 8.37 7.99 8.27  62 Sooke 62 62 62 62 62 62 62 62 62  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  21818 22368 23035 23814 24765 25437 26240 27155 28145 28973  21 22 17 14 30 26 15 16 27 19  1.88 1.91 1.43 1.1 2.26 1.89 1.05 1.06 1.76 1.19  9.63 9.84 7.38 5.88 12.11 10.22 5.72 5.89 9.59 6.56  10.71 10.89 8.13 6.28 12.9 10.75 5.96 6.03 10.01 6.77  249 TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  63 Saanich 63 63 63 63 63 63 63 63 63  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  21549 23027 24765 25528 26525 27851 28744 29505 30454 31878  20 24 27 36 30 31 25 24 23 30  1.32 1.47 1.52 1.96 1.56 1.54 1.19 1.09 1.01 1.26  9.28 10.42 10.9 14.1 11.31 11.13 8.7 8.13 7.55 9.41  7.51 8.39 8.65 11.19 8.87 8.76 6.78 6.19 5.78 7.2  64 Gulf Islands 64 64 64 64 64 64 64 64 64  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  5391 5727 6295 6553 6723 6888 7059 7055 7002 7253  3 8 12 2 6 8 6 4 9 11  0.62 1.55 2.11 0.34 0.99 1.29 0.95 0.63 1.41 1.65  5.56 13.97 19.06 3.05 8.92 11.61 8.5 5.67 12.85 15.17  3.51 8.82 12.04 1.92 5.63 7.38 5.41 3.61 8.04 9.43  65 Cowichan 65 65 65 65 65 65 65 65 65  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  20826 21483 23010 23772 23978 24185 24451 24240 24677 25200  15 16 13 17 20 22 22 24 15 20  1.2 1.25 0.94 1.19 1.37 1.47 1.44 1.54 0.95 1.23  7.2 7.45 5.65 7.15 8.34 9.1 9. 9.9 6.08 7.94  6.82 7.15 5.33 6.78 7.78 8.39 8.18 8.78 5.43 7.04  250  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  66 Lake Cowichan 66 66 66 66 66 66 66 66 66  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  3537 3704 3695 3537 3520 3562 3578 3460 3380 3295  5 1 1 4 2 5 2 3 1 1  2.44 0.46 0.46 1.85 0.9 2.18 0.84 1.32 0.46 0.45  14.14 2.7 2.71 11.31 5.68 14.04 5.59 8.67 2.96 3.03  13.9 2.64 2.6 10.56 5.11 12.45 4.81 7.5 2.63 2.59  68 Nanaimo 68 68 68 68 68 68 68 68 68  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  43021 45571 48785 50327 50763 51572 51850 52120 52342 53638  57 47 42 35 36 34 37 26 37 33  2.11 1.67 1.41 1.14 1.14 1.06 1.13 0.77 1.09 0.95  13.25 10.31 8.61 6.95 7.09 6.59 7.14 4.99 7.07 6.15  12.03 9.52 8.01 6.49 6.52 6.03 6.42 4.38 6.22 5.41  69 Qualicum 69 69 69 69 69 69 69 69 69  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  12021 13215 14490 15283 15584 16018 16310 16675 16389 17016  15 12 9 9 28 18 16 23 13 21  1.48 1.12 0.77 0.73 2.19 1.37 1.18 1.63 0.92 1.44  12.48 9.08 6.21 5.89 17.97 11.24 9.81 13.79 7.93 12.34  8.43 6.37 4.37 4.14 12.47 7.8 6.74 9.28 5.25 8.23  251  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  70 Alberni 70 70 70 70 70 70 70 70 70  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  20062 20415 20745 21083 21062 20964 20796 20390 20608 20398  12 13 9 11 20 7 9 16 10 14  1.14 1.21 0.83 0.98 1.74 0.6 0.76 1.34 0.82 1.13  5.98 6.37 4.34 5.22 9.5 3.34 4.33 7.85 4.85 6.86  6.5 6.93 4.71 5.58 9.93 3.43 4.34 7.62 4.68 6.44  71 Courtney 71 71 71 71 71 71 71 71 71  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  21346 22231 23875 24903 25379 25943 26629 26735 26774 27631  14 17 12 8 28 17 10 22 16 16  1.11 1.3 0.85 0.54 1.81 1.07 0.61 1.31 0.95 0.92  6.56 7.65 5.03 3.21 11.03 6.55 3.76 8.23 5.98 5.79  6.34 7.4 4.82 3.05 10.34 6.11 3.46 7.45 5.42 5.24  72 Campbell River 72 72 72 72 72 72 72 72 72  1979  16586  7  0.83  4.22  4.73  1980 1981 1982 1983 1984 1985 1986 1987 1988  17493 18685 19255 19460 19755 19860 20050 20809 21517  9 9 7 10 12 12 9 9 9  1.03 0.99 0.74 1.04 1.21 1.17 0.86 0.83 0.8  5.14 4.82 3.64 5.14 6.07 6.04 4.49 4.33 4.18  5.87 5.64 4.2 5.91 6.9 6.67 4.89 4.72 4.55  252  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  75 Mission 75 75 75 75 75 75 75 75 75  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  13760 14386 15400 15674 15915 16363 16734 17000 17601 18324  2 5 11 5 6 8 9 16 13 19  0.25 0.6 1.24 0.56 0.66 0.86 0.94 1.66 1.3 1.83  1.45 3.48 7.14 3.19 3.77 4.89 5.38 9.41 7.39 10.37  1.41 3.41 7.04 3.21 3.76 4.92 5.37 9.46 7.43 10.42  76 AgassizHarrison 76 76 76 76 76 76 76 76 76  1979  2998  2  1.07  6.67  6.1  1980 1981 1982 1983 1984 1985 1986 1987 1988  3078 3220 3370 3509 3598 3704 3675 3838 3844  2 3 0 3 0 2 0 2 4  1.06 1.56 0.00 1.41 0.00 0.88 0.00 0.86 1.69  6.5 9.32 0.00 8.55 0.00 5.4 0.00 5.21 10.41  6.03 8.91 0.00 8.03 0.00 5.02 0.00 4.91 9.66  77 Summerland 77 77 77 77 77 77 77 77 77  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  5131 5370 5630 5789 5861 5913 5997 6035 6094 6257  1 1 2 3 7 5 4 9 5 6  0.24 0.23 0.44 0.65 1.48 1.04 0.82 1.86 0.97 1.15  1.95 1.86 3.55 5.18 11.94 8.46 6.67 14.91 8.2 9.59  1.37 1.33 2.51 3.68 8.42 5.95 4.69 10.6 5.56 6.55  253  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  80 Kitimat 80 80 80 80 80 80 80 80 80  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  8004 8265 8740 8989 8606 8397 8433 8080 8253 8117  3 3 3 5 3 6 11 4 9 5  1.09 1.02 0.97 1.54 0.9 1.74 3.13 1.11 2.39 1.29  3.75 3.63 3.43 5.56 3.49 7.15 13.04 4.95 10.91 6.16  6.2 5.84 5.53 8.77 5.14 9.91 17.86 6.33 13.61 7.36  81 Fort Nelson 81 81 81 81 81 81 81 81 81  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2889 3086 3070 3174 3175 3163 3197 3185 3381 3192  0 2 1 1 1 0 0 0 1 1  0.00 2. 0.97 1.04 1.03 0.00 0.00 0.00 0.93 0.97  0.00 6.48 3.26 3.15 3.15 0.00 0.00 0.00 2.96 3.13  0.00 11.4 5.53 5.94 5.88 0.00 0.00 0.00 5.33 5.53  84 Van Isl North 84 84 84 84 84 84 84 84 84  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  2871 2899 3000 2950 2803 2681 2594 2420 2435 2481  1 1 0 0 0 0 0 3 2 2  1. 1.06 0.00 0.00 0.00 0.00 0.00 3.19 2.17 2.02  3.48 3.45 0.00 0.00 0.00 0.00 0.00 12.4 8.21 8.06  5.7 6.06 0.00 0.00 0.00 0.00 0.00 18.19 12.39 11.52  254  TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  85 Van Isl West 85 85 85 85 85 85 85 85 85  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  8113 8606 9095 9601 9684 9780 9712 9385 9683 9658  3 1 0 1 0 4 3 1 5 3  1.04 0.33 0.00 0.31 0.00 1.17 0.85 0.28 1.35 0.8  3.7 1.16 0.00 1.04 0.00 4.09 3.09 1.07 5.16 3.11  5.92 1.86 0.00 1.79 0.00 6.67 4.87 1.61 7.68 4.57  86 Creston-Kaslo 86 86 86 86 86 86 86 86 86  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  9047 9293 9545 9521 9620 9634 9650 9680 10335 9289  1 2 1 2 5 4 4 1 10 6  0.14 0.27 0.13 0.26 0.65 0.52 0.53 0.13 1.32 0.8  1.11 2.15 1.05 2.1 5.2 4.15 4.15 1.03 9.68 6.46  0.8 1.55 0.75 1.48 3.7 2.96 3. 0.74 7.54 4.55  87 Stikine 87 87 87 87 87 87 87 87 87  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  1238 1176 1220 1238 1283 1295 1298 1260 1218 1142  0 0 1 0 0 1 0 1 1 0  0.00 0.00 2.5 0.00 0.00 2.22 0.00 1.96 2.27 0.00  0.00 0.00 8.2 0.00 0.00 7.72 0.00 7.94 8.21 0.00  0.00 0.00 14.25 0.00 0.00 12.67 0.00 11.18 12.95 0.00  255 TABLE  29 (contd.)  School District  Year  Population  Observed CABS  S.I.R.  Crude Rate  Age-Sex Adjusted Rate  88 Terrace 88 88 88 88 88 88 88 88 88  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  14332 15423 16180 16960 16911 16741 16524 16255 17347 17540  2 5 4 8 8 6 4 9 3 14  0.35 0.79 0.65 1.22 1.18 0.86 0.56 1.21 0.4 1.8  1.4 3.24 2.47 4.72 4.73 3.58 2.42 5.54 1.73 7.98  1.99 4.51 3.69 6.95 6.71 4.91 3.18 6.88 2.29 10.26  89 Shuswap 89 89 89 89 89 89 89 89 89  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  17988 18634 19770 20720 21176 21268 21290 20835 20650 20736  10 10 6 9 6 12 17 14 9 15  0.77 0.75 0.42 0.59 0.38 0.76 1.07 0.89 0.57 0.95  5.56 5.37 3.03 4.34 2.83 5.64 7.98 6.72 4.36 7.23  4.37 4.25 2.39 3.34 2.19 4.33 6.08 5.06 3.27 5.44  256 TABLE 30 VARIABILITY OF CABS AND STANDARDIZED INCIDENCE RATIOS WITHIN SCHOOL DISTRICTS B.C. 1979-1988 School District 1. 2. 3. 4. 7. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 21. 22. 23. 24. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48.  Fernie Cranbrook Kimberley Windermere Nelson Castlegar Arrow Lakes Trail Grand Forks Kettle Valley S.Okanagan Penticton Keremeos Princeton Golden Revelstoke Armstrong-Spall Vernon Central OkAnagan Kamloops N.Thompson Cariboo-Chilcotin Quesnel Lilooet S.Cariboo Merritt Hope Chilliwack Abbotsford Langley Surrey Delta Richmond Vancouver New Westminster Burnaby Maple Ridge Coquitlam N.Vancouver W.Vancouver Sunshine Coast Powell River Howe Sound  * Std/Mean CABS ** Largest SIR/smallest SIR  Observed CABS  STD  9 39 25 4 105 40 27 67 30 12 105 207 30 26 2 18 26 169 660 190 8 90 59 6 18 24 27 172 239 214 764 271 341 1474 214 523 186 424 331 159 70 70 46  0.54 1.22 1.43 0.66 3.77 2.61 1.95 3.47 1.95 154 4.82 7.72 1.55 1.28 0.40 1.25 1.28 6.49 25.18 4.27 0.40 3.41 3.48 0.66 1.66 1.20 1.79 5.93 4.44 5.82 22.11 5.84 8.47 29.22 6.25 8.84 3.85 6.64 7.79 5.82 3.19 3.66 2.33  Coefficient of Variation* 0.6 0.31 0.57 1.66 0.36 0.65 0.72 0.52 0.65 1.28 0.46 0.37 0.52 0.49 2.00 0.69 0.49 0.38 0.38 0.22 0.50 0.38 0.59 1.11 0.92 0.50 0.66 0.34 0.19 0.27 0.29 0.22 0.25 0.20 0.29 0.17 0.21 0.16 0.24 0.37 0.46 0.52 0.51  Expected CABS  Extremal Quotient**  45.83 68.14 40.93 25.03 89.53 49.21 20.15 98.59 37.35 15.71 84.76 153.26 21.77 20.74 19.83 29.27 30.41 186.21 432.98 265.98 14.97 113.55 72.13 15.48 29.95 34.54 32.38 205.41 239.52 226.59 683.67 251.78 383.24 1975.85 188.74 632.8 147.63 355.25 411.11 227.91 86.29 80.28 44.65  2.3 4.5 4.2 2.1 3.7 7.7 5.1 15 5.8 5.3 3.5 2.9 6.9 4.9 1.05 4.0 5.1 4.1 3.9 1.7 1.1 3.7 8.2 2.4 4.8 5.4 5.7 2.5 1.5 4.1 1.6 1.5 1.5 1.8 3.2 1.5 1.9 1.7 2.2 3.6 5.0 6.4 7.5  257 TABLE 30 (contd.)  School District 49. 50. 52. 54. 55. 56. 57. 59. 60. 61. 62. 63. 64. 65. 66. 68. 69. 70. 71. 72. 75. 76. 77. 80. 81. 84. 85. 86. 87. 88. 89.  Central Coast Queen Charlotte Prince Rupert Smithers Burns Lake Nechako Prince George Peace River S. Peace River N. Greater Victori Sooke Saanich Gulf Islands Cowichan Lake Cowichan Nanaimo Qualicum Alberni Courtney Cambell River Mission Agassiz-Harrison Summerland Kitimat Fort Nelson Van Isl North Van Isl West Creston-Kaslo Stikine Terrace Shuswap  Observed CABS  StD CABS  Coefficient of Variation  Expected CABS  Extremal Quotient**  4 12 47 31 16 30 194 23 31 1342 207 270 69 184 25 384 164 121 160 93 94 18 43 52 7 9 21 36 4 63 108  0.66 0.6 2.76 1.64 2.06 2.53 2.24 1.27 2.17 13.29 5.22 4.49 3.14 3.50 1.57 8.10 5.87 3.65 5.50 1.62 5.08 1.33 2.49 2.64 0.64 1.04 1.64 2.73 0.49 3.38 3.49  1.66 0.5 0.59 0.53 1.29 0.84 0.12 0.55 0.70 0.10 0.25 0.17 0.46 0.19 0.63 0.21 0.36 0.30 0.34 0.17 0.54 0.74 0.58 0.51 0.91 1.16 0.78 0.76 1.22 0.54 0.32  10.33 13.09 51.08 35.9 22.18 43.55 226.6 73.29 63.64 855.45 136.85 199.06 60.3 147.77 22.38 318.6 128.87 116.27 155.64 99.44 93.29 21.56 47.79 33.99 10.19 9.61 34.01 76.53 4.49 69.32 152.53  2.1 2.1 5.4 3.0 6.8 7.1 1.6 3.7 7.3 1.4 2.1 1.9 3.4 1.6 5.4 2.7 3.0 2.9 3.3 1.7 7.3 1.9 5.0 3.2 2.1 3.1 4.4 9.4 1.3 5.1 2.8  258  TABLE 31 STANDARDIZED INCIDENCE RATIOS FOR FIVE AND TEN-YEAR PERIODS Standardized Incidence Ratios  School District  1979-1983^1984-1988^1979-1988 1 Fernie 2 Cranbrook 3 Kimberley 4 Windermere 7 Nelson 9 Castlegar 10 Arrow Lakes 11 Trail 12 Grand Forks 13 Kettle Valley 14 S.Okonagan 15 Penticton 16 Keremeos 17 Princeton 18 Golden 19 Revelstoke 21 ArmstrongSpallumcheen 22 Vernon 23 Central Okonagan 24 Kamloops 26 N.Thompson 27 Cariboo-Chilcotin 28 Quesnel 29 Lilooet 30 S.Cariboo 31 Merrit 32 Hope 33 Chilliwack 34 Abbotsford 35 Langley 36 Surrey 37 Delta 38 Richmond 39 Vancouver 40 New Westminster 41 Burnaby 42 Maple Ridge 43 Coquitlam 44 N.Vancouver  0.26 0.59 0.50 0.14 0.99 0.45 1.26 0.55 0.64 0.32 0.94 1.13 1.11 1.17 0.18 0.69 0.81  0.14 0.61 0.70 0.25 1.37 1.05 1.40 0.77 0.97 1.07 1.54 1.64 1.66 1.41 0.00 0.80 1.00  0.20 0.60 0.60 0.20 1.18 0.75 1.33 0.66 0.81 0.70 1.25 1.40 1.39 1.29 0.09 0.75 0.91  0.81 1.30 0.69 0.58 0.65 0.57 0.45 0.61 0.71 0.58 0.75 0.98 0.87 1.02 1.07 0.84 0.66 1.25 0.80 1.38 1.23 0.76  1.06 1.83 0.76 0.55 0.81 1.04 0.31 0.73 0.85 1.08 0.87 1.00 1.04 1.23 1.09 0.92 0.84 1.11 0.89 1.25 1.21 0.90  0.94 1.58 0.72 0.56 0.74 0.82 0.38 0.67 0.78 0.83 0.81 0.99 0.96 1.13 1.08 0.88 0.75 1.18 0.85 1.31 1.22 0.84  259 TABLE 31 (contd.) Standardized Incidence Ratios  School District  1979-1983^1984-1988^1979-1988 45 W.Vancouver 46 Sunshine Coast 47 Powell River 48 Howe Sound 49 Central Coast 50 Queen Charlotte 52 Prince Rupert 54 Smithers 55 Burns Lake 56 Nechako 57 Prince George 59 Peace River S. 60 Peace River N. 61 Greater Victoria 62 Sooke 63 Saanich 64 Gulf Islands 65 Cowichan 66 Lake Cowichan 68 Nanaimo 69 Qualicum 70 Alberni 71 Courtney 72 Cambell River 75 Mission 76 Agassiz-Harrison 77 Summerland 80 Kitimat 81 Fort Nelson 84 Van Isl North 85 Van Isl West 86 Creston-Kaslo 87 Stikine 88 Terrace 89 Shuswap  0.68 0.70 0.71 0.86 0.49 0.82 0.65 0.80 1.10 0.64 0.90 0.25 0.34 1.74 1.75 1.57 1.15 1.24 1.38 1.40 1.29 1.08 1.11 0.98 0.70 0.83 0.70 1.22 0.85 0.35 0.47 0.33 0.77 0.84 0.61  0.74 0.97 1.09 1.20 0.67 1.11 1.08 1.01 0.66 0.82 0.84 0.37 0.65 1.51 1.51 1.26 1.16 1.33 1.03 1.02 1.45 1.06 1.10 0.98 1.23 0.81 1.22 1.76 0.49 1.25 0.75 0.65 1.09 1.01 0.77  0.71 0.85 0.90 1.04 0.58 0.97 0.88 0.91 0.87 0.73 0.87 0.31 0.50 1.62 1.62 1.40 1.16 1.29 1.20 1.20 1.37 1.07 1.11 0.98 0.98 0.82 0.97 1.51 0.67 0.80 0.62 0.49 0.94 0.94 0.69  260  TABLE 32 EFFECT OF MIGRATION TO ALBERTA FOR CABS PROCEDURE ON STANDARDIZED INCIDENCE RATIOS IN B.C. SCHOOL DISTRICTS 1988  School District* 1 Fernie 2 Cranbrook 3 Kimberley 7 Nelson 12 Grand Forks 15 Penticton 16 Keremeos 22 Vernon 23 Central Okanagan 57 Prince George 59 Peace River S 60 Peace River N 86 Creston-Kaslo  B.C. CABS  Alberta CABS  SIR B.C. Only  SIR All Cases  1 3 4 13 6 32 4 32 87  1 6 1 2 1 1 1 1 3  0.23 0.41 1.05 1.42 1.58 1.93 1.79 1.59 1.81  0.45 1.22 1.30 1.63 1.84 1.99 2.26 1.63 1.87  19 4 2 6  1 4 4 1  0.74 0.50 0.28 0.78  0.78 1.00 0.84 0.93  *All school districts sending patients to Alberta in 1988  261  TABLE 33 PERCENTAGE OF CABS CASES FROM EACH SCHOOL DISTRICT USING B.C. CENTRES 1979-1988 School District  1 Fernie 2 Cranbrook 3 Kimberley 4 Windermere 7 Nelson 9 Castlegar 10 Arrow Lakes 11 Trail 12 Grand Forks 13 Kettle Valley 14 S.Okonagan 15 Penticton 16 Keremeos 17 Princeton 18 Golden 19 Revelstoke 21 ArmstrongSpallumcheen 22 Vernon 23 Central Okanagan 24 Kamloops 26 N.Thompson 27 Cariboo-Chilcotin 28 Quesnel 29 Lilooet 30 S.Cariboo 31 Merrit 32 Hope 33 Chilliwack 34 Abbotsford 35 Langley 36 Surrey 37 Delta 38 Richmond 39 Vancouver 40 New Westminster 41 Burnaby 42 Maple Ridge  Centre 1  Centre 2  Centre 3  Number of Cases  33.3 7.3 12.0 50.0 13.0 20.5 44.8 27.8 60.6 50.0 58.9 56.3 50.0 46.7 50.0 44.4 43.3  66.7 92.7 84.0 0.0 85.2 77.3 51.7 70.8 36.4 50.0 37.1 41.6 47.1 53.3 50.0 55.6 56.7  0.0 0.0 4.0 50.0 1.7 2.3 3.4 1.4 3.0 0.0 4.0 2.1 2.9 0.0 0.0 0.0 0.0  9 41 25 4 115 44 29 72 33 16 124 238 34 30 2 18 30  39.8 49.5 18.5 25.0 72.5 8.1 28.6 27.8 33.3 84.8 61.7 67.5 80.7 80.6 75.3 57.8 69.8 71.3 66.3 72.2  57.7 48.0 78.7 62.5 25.5 88.7 71.4 66.7 66.7 15.2 37.2 31.4 18.1 19.0 22.7 41.7 30.0 28.3 32.7 27.3  2.5 2.6 2.8 12.5 2.0 3.2 0.0 5.6 0.0 0.0 1.0 1.1 1.2 0.4 1.9 0.5 0.2 0.4 1.0 0.5  201 784 216 8 102 62 7 18 24 33 1% 283 254 851 308 386 1665 230 578 205  262  TABLE 33 (contd.)  School District  43 Coquitlam 44 N.Vancouver 45 W.Vancouver 46 Sunshine Coast 47 Powell River 48 Howe Sound 49 Central Coast 50 Queen Charlotte 52 Prince Rupert 54 Smithers 55 Burns Lake 56 Nechako 57 Prince George 59 Peace River S. 60 Peace River N. 61 Greater Victoria 62 Sooke 63 Saanich 64 Gulf Islands 65 Cowichan 66 Lake Cowichan 68 Nanaimo 69 Qualicum 70 Alberni 71 Courtney 72 Cambell River 75 Mission 76 Agassiz-Harrison 77 Summerland 80 Kitimat 81 Fort Nelson 84 Van Isl North 85 Van Isl West 86 Creston-Kaslo 87 Stikine 88 Terrace  Centre 1  Centre 2  Centre 3  Number of Cases  73.2 18.0 17.0 39.7 21.2 13.5 40.0 15.4 36.0 21.9 33.3 5.9 10.6 40.0 11.8 0.4 0.5 0.4 11.6 1.1 4.0 4.7 5.5 10.6 43.6 31.3 65.5 52.2 34.0 26.4 77.8 22.2 22.7 13.6 42.9 3.3  25.9 81.2 82.4 60.3 71.8 86.5 60.0 84.6 60.0 78.1 66.7 94.1 88.5 48.0 85.3 0.3 0.5 0.0 7,2 .0 0.0 7.0 6.7 4.9 3.0 6.1 33.6 47.8 66.0 73.6 22.2 11.1 18.2 84.1 57.1 95.1  0.9 0.8 0.5 0.0 7.0 0.0 0.0 0.0 4.0 0.0 0.0 0.0 0.9 12.0 2.9 99.3 99.0 99.6 81.2 98.9 96.0 88.3 87.9 84.6 53.3 62.6 0.9 0.0 0.0 0.0 0.0 66.7 59.1 2.3 0.0 1.6  456 377 182 73 71 52 5 13 50 32 18 34 218 25 34 1340 207 269 69 184 25 385 165 123 165 99 113 23 50 53 9 9 22 44 7 61  263  TABLE 36 CORONARY ARTERY BYPASS SURGERY ANNUAL MEAN AGE BY CENTRE B.C. 1979-1988  Year  Centre 1 Mean^Std  Centre 2 Mean^Std  Centre3 Mean^Std  1979 1980 1981 1982 1983 1984 1985 1986 1987 1988  58.23^9.25 57.60^8.82 58.56^8.49 58.60^8.78 59.45^9.49 61.20^8.59 61.65^9.45 62.55^9.13 62.64^8.70 62.75^9.20  56.41^9.30 56.50^9.07 56.87^8.27 58.10^9.56 56.09^8.49 60.39^8.91 61.42^8.97 61.59^9.72 62.51^9.30 62.30^9.82  57.57^8.36 65.60^7.75 58.94^8.67 60.16^9.05 60.87^9.11 61.90^9.16 63.39^8.88 63.59^8.64 63.02^8.40 64.72^9.08  TABLE 37 DISTRIBUTION OF CABS PATIENTS WITH COMORBIDITY BETWEEN CENTRES B.C. 1979-1988 Centre  Diabetes  COPD  1 2 3  327 281 188  96 90 83  423 371 271  4736 3838 2979  8.93 9.66 9.10  Total  796*  269  1065  11553  9.21  Total Comorbidity  Total CABS  * 2 cases of diabetes occurred in patients coded as receiving CABS in other centres.  Comorbidity as % of Total  264  TABLE 39 MEAN AGE BY SEX 'NO-CABS" AND ANGIOPLASTY POPULATIONS Year^Male  Female  Total  Mean Age  2 4 6 24 43 222 439 547 233 223  0 0 7 3 23 63 148 164 91 78  2 4 13 27 66 285 587 711 324 301  57.0 51.5 55.92 52.77 58.01 57.37 60.04 59.34 61.08 62.01  378 477  109 144  487 621  59.46 59.75  No-CABS Revascularization 1979 1980 1981 1982 1863 1984 1985 1986 1987 1988 Angioplasty 1987 1988  265  TABLE 41 GROWTH AND DECLINE OF THE 4800 CCP CODE BY CENTRE B.C. 1979-1988 Year  Centre 1  Total  I^2  J^9  3  79 80 81 82 83 84 85 86 87 88  1 3 8 10 14 98 206 243 1 0  1 0 1 2 4 32 77 114 0 0  0 1 4 14 46 147 301 348 319 300  0 0 0 1 2 8 3 6 4 1  2 4 13 27 66 285 587 711 324 301  Total  584  231  1480  25  2320  TABLE 41(A) DISTRIBUTION OF ANGIOPLASTY (CCP CODES 4801-4805) BY CENTRE B.C. 1987-1988  Year  Centre  Total  1  2  3  9  1987 1988  320 422  164 189  0 0  3 10  487 621  Total  742  353  0  13  1108  266  APPENDIX B  267  TABLE 42 INDEPENDENT VARIABLES SIMPLE STATISTICS Variable  Reference Level  Min  Max  Mean  Standard Deviation  Year  0  0  5  2.50  1.70  Year-squared  0  0  25  9.16  8.90  Income (INC)  3  -1.15  4.01  0.26  0.68  Distance from Cardiolgist (DSCAR)  0  0  2  1.71  0.60  Distance from centre (DSCEN)  0  0  2  1.86  0.14  Distance from Internist (DSINT)  0  0  2  0.60  0.70  Employment Rate (EMPRAT)  0.7  0.04  0.26  0.17  0.04  Graduation Rate (GRADRAT)  0  0.03  0.76  0.12  0.04  268  TABLE 43 INDEPENDENT VARIABLES PEARSON CORRELATION COEFFICIENTS  Variable  INC  DSCAR  DSINT  DSCEN  GRADRAT  EMPRAT  INC P  1.000010.0001  0.28860.00001  0.20750.0 0.0001  -0.3288 0.0609  0.0890 0.0609  -0.0692 0.1451  DSCAR P  -0.2886 0.0001  1.0000  0.3339 0.0001  0.6846 0.0001  -0.1338 0.0047  -0.2499 0.0001  DSINT P  -0.2075 0.0001  0.3339 0.0001  1.0000  0.2360 0.0001  -0.1527 0.0012  -0.1876 0.0001  DSCEN P  -0.3288 0.0001  0.6846 0.0001  0.23604 0.0001  1,0000  -0.0760 0.1094  -0.2668 0.0001  GRADRAT P  0.0890 0.0609  -0.1338 0.0047  -0.1527 0.0012  -0.0760 0.1094  1.0000  0.0031 0.9479  EMPRAT P  -0.0692 0.1451  -0.2499 0.0001  -0.1876 0.0001  -0.2668 0.0001  0.0031 0.9479  1.0000  269  TABLE 44 POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN CABS RATE ACROSS ALL SCHOOL DISTRICTS Variable*  Parameter Estimate  Standard Error  Intercept  -11.5034  0.0515  Year  0.0891  0.0588  Year-squared  -0.000839  0.0099  INC  -0.3725  0.0650  DSCAR  -0.0641  0.0879  DSCEN  0.1710  0.0490  Year*DSCAR  -0.0481  0.0458  Year*DSCEN  -0.00174  0.0496  Year-squared *DSCAR  0.0117  0.00809  Year-squared *DSCEN  -0.00660  0.00856  DSCAR*INC  0.0789  0.0454  DSCEN*INC  0.0751  0.0531  DSCEN*DSCAR  -0.0720  0.0415  -2 log Likelihood  Intercept  Intercept & Covariates  188050.26  187858.60  chi-squared  df  p  191.656  12  0.0001  Difference from saturated model chi-squared  df  p  891.146  66  0.0001  * Only significant variables are incided  270 TABLE 45 POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN CABS RATE ACROSS SCHOOL DISTRICTS WITH ADJUSTMENT FOR MOBILITY Variable*  Parameter Estimate  Standard Error  Intercept  -11.4765  0.0426  Year  0.0783  0.0444  Year-squared  -0.00172  0.00764  ALBERTA  -0.7363  0.0903  INC  -0.3145  0.0442  DSCAR  -0.1458  0.0379  DSCEN  0.1240  0.0215  Year*DSCAR  -0.0454  0.0332  Year-squared *DSCAR  0.00588  0.00574  DSCAR*INC  0.1346  0.0358  -2 log Likelihood  Intercept  Intercept & Covariates  188050.26  187783.62  chi-squared  df  p  266.636  9  0.0001  Difference from saturated model chi-squared  df  p  816.16  66  0.001  * Only significant variables are included  271 TABLE 46 POISSON REGRESSION SATURATED MODELS Variable  Parameter Estimate  Standard Error  CABS Unadjusted for Morbidity Intercept  0.000145  0.3843  Lograte  1.00  0.0337  -2 log Likelihood  Intercept  Intercept & Covariates  188050.26  186967.46  chi-squared  df  p  1082.802  74  0.0001  CABS Adjusted for Mo rbidity Intercept  0.00014  0.3683  Lograte  1.00  0.0323  -2 log Likelihood  Intercept  Intercept & Covariates  188208.49  186949.44  chi-squared  df  P  1259.05  74  0.0001  272  TABLE 47 POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN MORBIDITY-ADJUSTED CABS RATE ACROSS ALL B.C. SCHOOL DISTRICTS Variable  Parameter Estimate  Standard Error  Intercept  -11.241  0.0295  Year  -0.00733  0.0269  Year-squared  0.0131  0.0048  DSCAR  -0.2597  0.0141  INC  -0.1108  0.0237  -2 log Likelihood  Intercept  Intercept & Covariates  188208.49  187825.01  chi-squared  df  p  383.482  4  0.0001  Difference from saturated model chi-squared  df  p  875.568  70  0.0001  273 TABLE 48 POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN MORBIDITY-ADJUSTED CABS RATE ACROSS SCHOOL DISTRICTS WITH ADJUSTMENT FOR ALBERTA Variable  Parameter Estimate  Standard Error  Intercept  -11.2429  0.0295  Year  -0.00681  0.0269  Year-squared  -0.0132  0.00480  ALBERTA  -0.6020  0.0902  INC  -0.1133  0.0237  DSCAR  -0.2451  0.0143  -2 log Likelihood  Intercept  Intercept & Covariates  188208.49  187720.92  chi-squared  df  p  437.570  5  0.0001  Difference from saturated model chi-squared 821.48  df 69  p 0.0001  274 TABLE 49 REGIONAL DIVISIONS USED IN AGE-SEX-YEAR-REGION REGRESSION ANALYSIS REMOTE/RURAL/URBAN/METROPOLITAN DIVISION Region  School Districts  Remote (Reg3)  27, 28, 46, 49, 50, 52, 54 55, 56, 57, 59, 60, 80, 81 85, 87, 88  Rural (Reg2)  1-10, 18, 19, 21, 22, 29, 47 48, 65, 66, 70, 86  Urban (Regl)  11-17, 23-26, 30-34, 42, 68 69, 71-77, 84  Metropolitan (Reference)  35-41, 43-45, 61-64  *School district 89 straddled two regions and was excluded from this analysis. GEOGRAPHICAL DIVISION Region^  School District  Vancouver & S.W. (Reference)  32-35, 48, 75, 76  S.E. & Okanagan (Gregg)  1-26, 30, 31, 77, 78  Vancouver Island (Gregl) & Central Coast  47, 49, 61-72, 84, 85  North (Greg3)  27, 28, 50-60, 80, 81, 87, 88  275 TABLE 50 POISSON REGRESSION RESULTS INTERACTIONS BETWEEN AGE, SEX, YEAR AND REGION FOR METROPOLITAN/URBAN/RURAL/REMOTE REGIONS B.C. 1983-1988 Variable Intercept Age Year Sex Agesex Agesq Agesqsex Yearsex Yearsq Yearsqsex Yearsqage Agesqyear Yearsqagesq Yearage Reg2 Reg3 Regl Reg2Age Reg3Age ReglAge RegiSex Reg2Sex Reg3Sex ReglYear Reg2Year Reg3Year Reg2Agesq Reg3Agesq Regl Agesq Reg3Yearsq Reg2Yearsq ReglYearsq Reg2Agesex Reg3Agesex Regl Agesex Reg2Sexyear Reg3Sexyear ReglSexyear Reg2Ageyear Reg3Ageyear Regl Ageyear Reg2Agesqsex  Parameter Estimate  Standard Error  -5.8572 -0.2672 0.0199 -1.3271 0.0675 -0.1188 -0.0277 -0.0688 0.00738 0.01 -0.00251 -0.00905 0.00135 0.039 -0.1307 -0.5054 0.00159 -0.1254 -0.0628 -0.0603 0.2957 0.2378 -0.1854 0.0579 -0.0151 0.0924 -0.0418 0.0123 -0.015 -0.00598 0.00887 -0.00587 0.1165 0.0188 -0.0476 0.1161 0.3468 0.0732 0.0209 -0.0583 0.0988 0.028  0.0492 0.0312 0.0444 0.0931 0.0304 0.00721 0.0102 0.0820 0.00832 0.0156 0.00518 0.00700 0.00136 0.0274 0.1401 0.0298 0.0874 0.0957 0.1397 0.0569 0.1578 0.2498 0.4050 0.0773 0.1238 0.1784 0.0238 0.0281 0.0134 0.0325 0.0229 0.0144 0.0836 0.1223 0.0535 0.2214 0.3159 0.1393 0.0799 0.1167 0.0482 0.0274  276 TABLE 50 (contd.) Variable  Parameter Estimate  Standard Error  Reg3Agesqsex ReglAgesqsex Reg2Agesqyear Reg3Agesqyear ReglAgesqyear Reg2Yearsqage Reg3Yearsqage ReglYearsqage Reg2Yearsqsex Reg3Yearsqsex ReglYearsqsex Reg2Yearsqagesq Reg3Yearsqagesq ReglYearsqasq  -0.00166 -0.0172 0.007 0.00425 0.0226 0.00518 0.0162 -0.0142 -0.042 -0.0426 -0.0347 0.00146 0.00039 -0.00325  0.0333 0.0179 0.0211 0.0246 0.0124 0.0145 0.0210 0.00889 0.0424 0.0564 0.0267 0.00392 0.00456 0.002326  Intercept 125720.66  Intercept and covariates 110488.36  chi-square  df  p  15232.299  55  <0.0001  -2loglikelihood  277  TABLE 51 POISSON REGRESSION RESULTS INTERACTIONS BETWEEN AGE, SEX, YEAR AND REGION FOR GEOGRAPHIC REGIONS B.C. 1983-1988  -  Variable  Parameter Estimate  Standard Error  Intercept Age Year Sex Agesex Agesq Agesqsex Yearsex Yearsq Yearsqsex Yearsqage Agesqyear Yearsqagesq Yearage GRegl GReg2 GReg3 GReglAge GReg2Age GReg3Age GReglSex GReg2Sex GReg3Sex GReglYear GReg2Year GReg3Year GReglAgesq GReg2Agesq GReg3Agesq GReglYearsq GReg2Yearsq GReg3Yearsq GReglAgesex GReg2Agesex GReg3Agesex GReglSexyear GReg2Sexyear GReg3Sexyear GReglAgeyear GReg2Ageyear GReg3Ageyear GReglAgesqsex^_  -5.7095 -0.0108 -0.0377 -1.4783 0.1580 -0.1913 -0.0246 0.0292 0.0165 -0.0108 -0.00502 0.00471 -0.0006 0.0626 0.4884 0.0923 -0.2622 0.0433 -0.0280 -0.1156 0.1224 0.4306 -0.0967 -0.1509 0.0900 0.1994 0.00332 -0.0241 -0.0407 0.0147 -0.0209 -0.0431 -0.0499 -0.0421 -0.0283 -0.0728 -0.1869 -0.0390 0.0128 0.0800 0.0826 0.0113  0.0517 0.0264 0.0471 0.1031 0.0260 0.0098 0.0113 0.0898 0.00885 0.0169 0.00411 0.00872 0.00163 0.0224 0.0835 0.1015 0.1568 0.0415 0.0519 0.1033 0.1614 0.1843 0.3489 0.0794 0.0920 0.1393 0.0159 0.0201 0.0338 0.0152 0.0173 0.0261 0.0440 0.0482 0.0860 0.1471 0.1641 0.2920 0.0367 0.0425 0.0824 0.0191  278 TABLE 51 (contd.) Variable  Parameter Estimate  Standard Error  GReg2Agesqsex GReg3Agesqsex GReglAgesqyear GReg2Agesqyear GReg3Agesqyear GReglYearsqage GReg2Yearsqage GReg3Yearsqage GReglYearsqsex GReg2Yearsqsex GReg3Yearsqsex GReglYearsqagesq GReg2Yearsqagesq GReg3Yearsqasq  0.00941 0.00873 0.00138 0.0131 0.0288 -0.00324 -0.0115 -0.0161 0.00781 0.0191 0.0241 0.000296 0.000027 -0.00507  0.0211 0.0334 0.0148 0.0173 0.0279 0.00691 0.00769 0.0147 0.0283 0.0311 0.0535 0.00284 0.00319 0.00507  Intercept 126954.92  Intercept and covariates 112155.01  chi-square  df  p  14799.910  55  <0.0001  -2loglikelihood  279  MAP A  BRITISH COLUMBIA — COLOMBIE-BRITANNIOUE School Districts — Oistncts scotaires  SCNOOL OISTPICTS. CttST•HCTS SCOLAIRES 1 Berme ^ 30 S Can000^57 Prw+Ct George 2 Cranorook^3, 164,,,rm^59 *tact/ m•ver S 3 KImovity^32 moot^60 Peace M•ver N • wsnottnnet•^33 critlineraC4 ^61 Greater vtctorta 7 Nelson^34 A000tstorc^62 Soo•e S Casnegar^35 Langtey^63 Saan•en '0 Arrow Lakes^36 S u rre y^64 Cult tstancs 11 rraa^ 27 Dena^65 COvntrtan '2 GIAnCI Fors^38 P.CrirnOnd^36 La'. Covncrian '3 Kerue valley^39 Vat", COuvO ,^6! Niaftanno 69 OLtakcurn 14 S Okanagan^40 Nev.^ 15 Penncton^westmnster^73 Alcorn. 16 Keremeos^Al Surriaor^71 COunenay 17 Princeton^42 Nacre P.Oge^'2 CarTlooeit After 18 Gowen^a3 Cooutuam^75 mtsstort 19 Revelstoke^44 N Vancouver^76 aO3SSZ 21 Armstrong •^45 w Vancouver^mamsgn Scatturncneen^46 SketWWI* Coast ^77 Summe. , 4net :2 Vernon^47 Pov.ett Awe'^8C Kii•Mat 23 ctmv a i^A! Havre Soun0^8' =on NetSOn Okanagan^4 Central Coast^iki Van IsJ WeSt 2• KarrO000s^SO Queen Cnartorte ^85 van 1311.40nm 26 N Tnonloson^52 Ponce Puce ,'^A6 Creston•nasto 2 7 CattOoo•Crurconn^54 S./vine's^8' Simone 28 Ouesneo ^55 Burns Lake^6a '.,race 29 Ltsooet^56 Necna•o ^89  280  MAP B SCHOOL DISTRICTS SENDING A PLURALITY OF CASES TO CENTRE BRITISH COLUMBIA - COLOMBIE-BRITANNIOUE School Districts - Distncts scolaires  Centre 1  Centre 2  Centre 3  %171k1 ■114. t,":132  -twm11.  1 ....-14 6  1  ..0 ik  ln:47740.^ -  ilacri ,  :2 Vernon^47 Powell r;',ue•  23 Central^a8 wowe Sound  Okanagan^.19 Central Coast 24 Ka/114000s^50 Queen CNarinne 26 N ThOrnoson^52 Prot! Rupert 27 Card000•Crurcobn^SNUMWS 26 Ouesne•^55 Burns Lake 29 Lwooet^56 Nechako  b,  40.^NTAww,„  1. 7°  SCHOOL. DISTRICTS - DISTPtCTS SCOT AIRES Fern,.^ 30 S Cane= 2 Cranorook^3, Merritt 3 K■moerley^32 looe a winaermer•^33 CNretru,aci, 7 NelSOn^34 Alsootstora Castiegar^35 Langley '0 Arrow Lakes^36 Surrey 11 Trani^ 37 Delta '2 Grand :was^38 RIcnrridnd '3 Kent, vasey^39 Vancouver 14 S Okanagan^.40 New 15 Penucton^wrestrrunster 16 Keremeos^4 I Sur rtaCIV 17 Princeton^12 Wide R•oge 18 Gorden^43 Coctuntam 19 Revelstoke^44 N VanCOwver 21 Armstrong •^45 W vancouyer Scallumcneen^46 SurlSnme Coast  '".1%  4  ga  mtwas.,  57 Proce George 59 Peace R , ver S  60 Peace R.yer 61 Greater \Actona 62 Sooke 63 Saarkcn 54 G u ll islanos 65 Cow.cnan 56 Lake  62 Nama,, ,o  DO  `q4kk  WC  -^. 43  r °Pr ae  ‘  AO  5  39.  ,  69 Cluarcum 70 sitserru 71 COunenae  72 CamPOelliqnrer 75 kossoy, 76 Agass.: • ►iarnson 77 Summer ano 8C Kdonat  ,  e n son Ne.son 84 van 1st west as van 1st NOrtn Rs C•,SIOn•KASIO SIAme  si  Terrace  as Snuswars  36  C.34  281 MAP C  SCHOOL DISTRICTS SENDING 90 PERCENT, OR MORE, OF CASES TO ONE CENTRE BRITISH COLUMBIA - COLOMBIE-BRITANNIOUE School Districts - Districts scolaires  51 57  60  59  4 57 55 25  27 85 29 . 85  72  47  gt. SCHOOL DISTRICTS - Of ST111tCTS SCOLAIIRES  Maai.  1 germ* 2 Cranoroor 3 1cmoefloy WIrtgermere 7 kie•SOn S C•Stlegar '0 Attaw LaktS  7,•11 *2 Grano Fors '3 hero, Valley 14 S Okanagan 15^ u .cton 16 '<eremites 17 Princeton 10 Gowen 19 Revelstoke 21 Armstrong • Seallurncrteen 22 Vernon 23 Central Okanagan  24 KanktOODS 26 N Thompson 27 Catmoo•Crutoottn 23 (Due snet 29 I. teeetet  30 5 Cahoot, 31 k.44p•ell 32 meet Cruotntiso■ 34 Apootsfore 35 Lam•e ,/ 36 Stotev 2 7 Oena  38 Gtl.crtrnona  39 V1 ,, C*ukt , se tie. Strnalste , at Btor130V  42 macto a•:3ge 4 3 COCI‘onarn  ai N Var■COuve ,  43 w vamcouver se Sumsrune CCASI •7 Poweli 35 hto4.4 Sounc 49 Central Coast 50 Queen Crlariorte Pnnee Queen  SA S•rtarte•s 55 Burns Lame 56 Nechako  57 Ot.nce George 55 °one Ll.ver S 60 Peace q.v., N 61 Greater V.ctor.a 62 Sooke 63 Saamoh 6.4 Gun islanes 65 Comortan 56 .6.2keCo..Crt4/1 55 Naha•mto 69 Okrattcurrt  70 AMperfu  7 1 COUneMaY 72 Carhacte4 1:1 fleet 75 #.4tss.oh 76 4 9aSS•: • rtiarrrspn  7 7 S.rnme ,, anCI SC Kawnat 8' :on Ne•son 5.4 Van 1st West as Van Iv Norm  /46 Cresion•has•o 87 St.*.". 81 'create 85 Shus.4ap  ■111 ^V...1aU -'111,■"  do5  69  75  36  68  64  5  34  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items