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 - 1988ByJENNIFER MARY GAITB.S.N. The University of British Columbia, 1979A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTSFOR THE DEGREE OFMASTER OF SCIENCE(Health Services Planning and Administration)inTHE FACULTY OF GRADUATE STUDIESDepartment of Health Care and EpidemiologyWe accept this thesis as conforming to the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril 1992© Jennifer Mary Gait, 1992In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature) Department of -1-1t..oLtql Cal-k.... ex 2?‘ 6-Ur-I'M0 10t\- 1The University of British ColumbiaVancouver, CanadaDate M a t 1 1 l 0l g 2,DE-6 (2/88)iiABSTRACTSince the introduction of coronary artery bypass surgery (CABS) in the latesixties, the increase in the incidence rates has aroused controversy in the literature.Recent studies in the United States and Canada have documented both large rateincreases in the elderly and geographic variations in incidence rates. This studywas undertaken to discover whether similar patterns exist in British Columbia.Data from the British Columbia Hospital Morbidity Database, for fiscal years1979 through 1988, were used to calculate age-sex adjusted small-area incidencerates based on the school district of residence. Results showed a 1.2 fold increase inoverall 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, thepercentage of patients with either diabetes or chronic obstructive pulmonarydisease increased from three to twelve percent of annual cases.Over the study period, extreme variability in annual rates was seen bothwithin and among school districts. Within school districts, most variability wasseen in districts with populations below 10,000. Poisson regression (whichweighted school districts according to population size) showed that variationamong school districts was highly significant (p<0.0001).In an attempt to explain the variation in small area rates, the CABS rates foreach 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-orderinteractions were found to be important explanatory variables (R 2 = 0.21). Whileincome and distance from cardiologist had a negative effect on the CABS rate,111distance from centre had a surprising positive effect, which did not appear to beaccounted for by colinearity with distance from cardiologist.The model was then refitted, using the CABS rate adjusted for morbidity inthe school districts as the dependent variable. In this model distance fromcardiologist and income changed in relative importance, and distance from centreand the interaction between variables were no longer important (R 2 = 0.30).Refitting this model to account for mobility to Alberta showed that distance fromcardiologist and income explained more of the variation in rates (R 2 = 0.34).The presence of small-area variations in CABS rates, together withdifferences between centres in the number and type of procedures performed,suggest that there are inequities in cardiac care within B.C. These inequities appearto arise from complex relationships between distance from services and morbidityrates and average income in the school district of residence. In addition, it appearsthat the surgical centre referred to may also contribute to the variation in small-area rates, although this was not tested. It is clear that inequities in cardiac carecannot be redressed by simple solutions. Policy implications and suggestions forfurther research are discussed.ivTABLE OF CONTENTSABSTRACT^ IITABLE OF CONTENTS^ I VLIST OF MAPS^ IXLIST OF TABLES XLIST OF FIGURES^ XIIACKNOWLEDGMENTS X VINTRODUCTION^ 1CHAPTER 1^CORONARY ARTERY DISEASE^ 4INTRODUCTION^ 4THE PROBLEM OF CAD IN CANADA^4CAUSE AND EFFECT OF CORONARYARTERY DISEASE^ 5MANIFESTATIONS OF CORONARYARTERY DISEASE 8Myocardial Infarction^ 8Chronic Stable Angina 9Unstable Angina 10RISK FACTORS^ 21Stages of Disease 12Primary and Secondary Risk Factors^13Hypercholesterolemia^ 14Hypertension^ 14Tobacco Smoking 15Diabetes Mellitus 15High Density Lipoproteins^ 16Obesity^ 16Family History^ 16Physical Activity 17Age and Sex 17TRENDS IN CORONARY ARTERY DISEASE^18MORTALITYVDIAGNOSIS IN CAD^ 21Coronary Angiogram 21Exercise Stress Test 23PREVENTION AND TREATMENTOF HEART DISEASE^ 25Prevention 25Treatment^ 29CONCLUSIONS 30REFERENCES^ 32CHAPTER 2^CORONARY REVASCULARIZATION TECHNIQUES 37PART 1 CORONARY ARTERY BYPASS SURGERY 37HISTORY^ 38Growth and Utilization^ 38Assessment 40Changes in CABS Procedure 42Changes in Medical Treatment^ 43Changes in CABS Patients 43EFFICACY OF CABS^ 44The Intent-to-Treat Principle^ 45Chronic Stable Angina 46Unstable AnginaAsymptomatic Patients with "Silent"Myocardial Ischemia^ 63Post-Myocardial Infarction 64Evolving MI^ 66Emergency CABS after Failed PTCA^68Re-operations 69The Elderly^ 70Gender 76RISKS^ 78Operative Mortality^ 78Morbidity 81Other Risks^ 82COSTS^ 85CONCLUSION^ 86CORONARY REVASCULARIZATION TECHNIQUES 98REFERENCES^ 90viCHAPTER 2^PART II PERCUTANEOUS TRANSLUMINAL^98CORONARY ANGIOPLASTYHISTORY^ 98Trends in Assessment^ 98Changes in PTCA 100MECHANISM OF ACTION 101EFFICACY OF PTCA^ 101PTCA in Single Vessel Disease^102Multi-Vessel Disease 108PTCA in Unstable Angina 112PTCA in Left Ventricular Function^113Post-Myocardial Infarction^ 113Gender^ 114The Elderly 115Unsuccessful PTCA 117Summary 119RISKS^ 120Hospital Mortality^ 122Emergency Bypass Surgery^ 123Other Factors Affecting Risk 124COSTS^ 126CONCLUSION^ 128REFERENCES 131CHAPTER 3^REGIONAL VARIATIONS^ 135INTRODUCTION^ 135PROCEDURAL VARIATIONS IN GENERAL^135Factors Associated with Procedural Variations^138VARIATIONS IN CABS UTILIZATION RATES^140ISSUES IN VARIATIONS RESEARCH^146CONCLUSION^ 152REFERENCES 153viiCHAPTER 4^RATIONALE AND METHODOLOGY^156RATIONALE^ 156Questions 157METHODS^ 157Study Design 157Independent Variables^ 158Data Sources^ 159Study Population and Analysis^ 1601. Descriptive Analysis^ 1602. Regression Analysis 162CHAPTER 5^REVASCULARIZATION IN BRITISH COLUMBIA 1661979-1988: RESULTSCORONARY ARTERY BYPASS SURGERY^166Characteristics of the CABS population 166Re-operations^ 171Region of Residence of CABS Population^173Referral Patterns 180ANGIOPLASTY^ 184REGRESSION ANALYSIS^ 185SUMMARY^ 194REFERENCES 195CHAPTER 6^REVASCULARIZATION IN BRITISH COLUMBIA 1971979-1988 DISCUSSION OF RESULTSAND POLICY IMPLICATIONS OF STUDYDISCUSSION^ 196DESCRIPTIVE STUDY^ 196Incidence Rates 196Regression Analysis 203LIMITATIONS^ 207POLICY IMPLICATIONS^ 210Organization Of Health Care in B.C.^211Implications of Literature and Present Study^211Cost Effectiveness^ 213Inappropriate Treatment^ 214Other^ 217viiiRESEARCH^ 218REFERENCES 221APPENDIX A^222APPENDIX B 265ixLIST OF MAPSMap A^British Columbia School Districts^ 279Map B^School Districts Sending a Plurality of Cases to Centre^280Map C^School Districts Sending 90% or More of Cases to One Centre^281xLIST OF TABLESI Inclusion criteria for the three major RCTs 482 Characteristics of patients in RCTs for stable angina 503 Outcomes in RCTs for stable angina 534 Inclusion criteria for RCTs of unstable angina 585 Randomized trials for unstable angina 596 Outcomes in RCTs for unstable angina 617 Percentage of patients surviving five or more years afterreoperation718 Unadjusted cumulative 6-year survival in agesubgroups of the Over-65's729 Outcomes of CABS in CASS registry patients 7410 Predictors of operative mortality 7911 Cost effectiveness of CABS as a function of the severityof angina 8612 Comparison of outcomes in controlled trial of ptcaversus CABS 10713 Comparison of old- and new NHLBI registry characteristicsand outcomes by extent of disease 10914 Comparison of one-year outcomes in elderly and non-elderlypatients after successful and unsuccessful PTCA 11715 Regression models 16516 Annual CABS procedures for B.C. and out-ofprovince residents 22417 Annual CABS procedures by sex 22418 Annual mean age of CABS procedures per year 22519 Annual standardized incidence ratios and CABS rates 16720 Age-sex specific rates for coronary artery bypass surgery 16821 Distribution of cases by age-group 16822 Annual CABS procedures performed on sub-groupsof the over-65 population22623 Annual standardized incidence ratios andsex adjusted rates for subgroups of the over-65 populations22624 Annual numbers of patients with comorbidityreceiving isolated CABS22725 Distribution of comorbidity by age group isolatedcoronary artery bypass22726 Out-of-province revascularization services for B.C. Residents 228xi27 Numbers and mean age of B.C. Residents receiving CABS in Alberta 22828 Annual distribution of isolated CABS by school district^22929 Annual populations, rates and standardized incidence 231ratios by school district30 Variability of CABS and standardized incidence ratios^256within school districts31 Standardized incidence ratios for five and ten-year periods^25832 Effect of migration to Alberta for CABS procedure on 260standardized incidence ratios in B.C. school districts33 Percentage of CABS cases from each school district using B.C.^261centres34 Coronary artery bypass surgery per centre per year^ 26335 Percent of open heart surgery devoted to CABS annually by centre^26336 Distribution of CABS patients with comorbidity between centres^18238 Increase in non-CABS revascularization procedures^18239 Mean age by sex "no-CABS" and angioplasty populations^26440 Numbers and percentages of revascularization 183procedures by centre41 Growth and decline of the 4800 CCP code by centre^ 26541(a) Distribution of angioplasty (CCP codes 4801-4805) by centre^26542 Independent variables simple statistics^ 26743 Independent variables Pearson correlation coefficients^26844 Poisson regression variables explaining variation 269in CABS rate across all school districts45 Poisson regression variables explaining variation in^270CABS rate across school districts with adjustment for mobility46 Poisson regression saturated models^ 27147 Poisson regression variables explaining variation inmorbidity-adjusted CABS rate across all B.C. school districts^27248 Poisson regression variables explaining variation in^273morbidity-adjusted CABS rate across school districtswith adjustment for Alberta49 Regional divisions used in age-sex-year-regionregression analysis^ 27450 Poisson regression results interactions between age, sex, year^275and region for metropolitan/urban/rural/remote regions51' Poisson regression results interactions between age, sex,^277year and region for geographic regionsxiiLIST OF FIGURES1 Outcome for patients referrd for second opinion 842 Annual age-sex adjusted cabs rate per 10,000 population 1673 Age-use curves B.C. 1704 Scattergraph of coefficient of variation of observed 175CABS in school district by school district population5 Range of standardized incidence ratios for isolated 176CABS by school district6 Growth of revascularization procedures 183ACKNOWLEDGMENTSI would like to acknowledge the help of my committee members Dr. GeoffAnderson and Dr Charles Wright for the many hours they spent reviewing and fortheir helpful comments and suggestions during the planning and implementationof this thesis. Especial thanks go to my thesis supervisor, Dr Sam Sheps, whoselively interest and direction kept me going when I felt overwhelmed, and tocommittee member Dr Stephen Marion for his patient help and guidance in thestatistical analysis. Thanks also go to Sandi Wiggins who helped me through myinitial encounters with the mainframe computer and SPSS-X, and to Dr KenBenson 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 allmy friends, especially Jude and Ben Platzer and Linda Brown who shared theirhomes during my stays in Vancouver, and Judy Flagg who provided child-careduring my absences from home. My most heartfelt thanks go to my family, Carland Kate, without whose patient support, and ability to keep the home firesburning during my frequent absences from Victoria, this thesis would not havebeen completed.1INTRODUCTIONOver the past two decades there has been a vast number of studies in themedical literature on coronary revascularization techniques. The majority ofthese studies have been clinically based, comparing the outcomes ofrevascularization with those from medical treatment. A few studies havedocumented, and tried to account for, regional variations in the population-basedrates of revascularization procedures. A drawback to many of these latter studies isthat they have based their conclusions on a single years' data. Therefore, anyvariation found could be an artifact, peculiar to that year only. Also many studiesdid not account for the possibility of variation in the incidence of coronary arterydisease 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 verydifferent from those in B.C. where revascularization procedures are centralized,the numbers performed controlled by the Ministry of Health and where patientsmay be isolated without easy access to heart specialists or hospitals. The factorsassociated with procedural variations in the United States (US) might not berelevant in B.C.Over the past two years in B.C. a number of changes have either occurred orbeen proposed which will impact on future revascularization rates. In 1990,media attention to the length of waiting lists, prompted the government to start aregistry of patients waiting for cardiac surgery, and to make arrangements withhospitals in Seattle, Washington, for some B.C. patients to receive open-heartsurgery there. In the spring of 1991, a fourth hospital in B.C. started an open-heartprogram and the cardiac surgeons and cardiologists in the province set-up aclinical data-base to include all patients receiving coronary artery bypass surgery in2the province. At the University of British Columbia, preparations were started fora consensus process to develop indicators for appropriate use of coronary arterybypass surgery (CABS). The time seemed right to document the trends inrevascularization procedures in B.C.The purposes of this study are to:i. describe the Provincial and regional trends in CABS and percutaneoustransluminal coronary angioplasty (PTCA) in B.C. between 1979 and1988.ii. determine if there are variations in procedural rates among geographicregions, andiii. 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 andelsewhere, very briefly describes the causative theories and describes themanifestations of the disease, examines the role of risk factors, discusses thediagnosis of the disease and the use of the coronary angiogram and exercise "stresstest", discusses the primary and secondary prevention of the disease and toucheson the medical treatment.The purpose of Chapter 1 is to provide information that will help the layreader to understand the later chapters. Epidemiological and medical conceptsconsidered to be useful as background information are included, as footnotes orwithin the text itself.Chapter 2: Part I describes the history of CABS with respect to trends inutilization and the changes that have occurred in the procedure and in the patientpopulation receiving it. This section also reviews the findings of the clinicalstudies of CABS as they relate to the effectiveness of the procedure in chronic3stable angina, and in other conditions. Costs, complications and mortality arereviewed, both generally and as they relate to the above populations.Part II follows the same format for PTCA and reviews the comparisons ofPTCA and CABS.Chapter 3: describes regional variations in CABS and other procedures andexamines the factors which may account for such variation. Some of themethodological problems associated with studying regional variation arediscussed.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 thatneed to be done to clarify the issues raised.4CHAPTER ONECORONARY ARTERY DISEASEINTRODUCTIONCoronary artery disease (CAD) is the only condition for which coronaryrevascularization procedures are performed and, therefore, some knowledge aboutCAD is required in order to place the recent trends in coronary artery bypasssurgery (CABS) and percutaneous transluminal coronary angioplasty (PTCA) incontext. This chapter will define and describe CAD and will discuss the majorfindings in the literature in relation to the causes, risk factors, incidence, diagnosis,prevention and medical treatment of the disease. This discussion is not intendedto represent an exhaustive review of the extensive literature on this disease but isintended to simply set the stage for the later discussions on CABS, PTCA and onthe implications of the findings from this study.THE PROBLEM OF CAD IN CANADACoronary Artery Disease, also referred to as coronary heart disease (CHD) orischemic heart disease (IHD), ties with cancer as the leading cause of death inCanada. 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 thelowest age-standardized mortality rates for CAD, with rates of 162 per 100,000males and 77 per 100,000 females. Newfoundland, the province with the highestrates 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 percent5of all deaths) CAD accounted for 21 percent of all hospital days in Canada in 1985 ata cost, excluding doctors and surgery fees, of over $3 billion (Nicholls et al,1986).Despite the magnitude of the present problem, CAD mortality has beendeclining in Canada, and in much of the Western World, since the 1960's. Thesetrends will be described later in this chapter.CAUSE AND EFFECT OF CORONARY ARTERY DISEASEThe cause of coronary heart disease is artherosclerosis of the coronaryarteries. The right and left coronary arteries, the first two branches of the aorta,supply freshly oxygenated blood to the heart muscle (myocardium). The leftcoronary artery has two branches from the main stem - the left anterior descendingartery and the left circumflex artery. Thus, in effect, there are three coronaryarteries supplying the heart muscle, and the terms one- two- or three-vesseldisease indicates how many of these vessels are significantly affected by CAD 1 . Leftmain stem disease, stenosis of the left main stem before the bifurcation, affects theblood 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 majorbranches, the disease also frequently affects vein grafts such as those used incoronary artery bypass surgery. The build-up of collections of abnormal fats, cellsand debris (atherosclerotic plaque) under the endothelial layer of the arterygradually 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 inthe myocardial demand for oxygen (e.g., in exercise) cannot always be met andmyocardial ischemia results. An 80 percent reduction in the lumen may reduce1 In most studies an artery is considered to be significantly affected by CAD if the an atheroscleroticplaque occludes 70 percent, or more of the lumen. Some studies set this figure at 50 percent occlusion.6blood flow to the myocardium even when the body is at rest. The clinicalsymptoms of CAD, therefore, are those of vascular insufficiency which may resultfrom abrupt closure of the artery due to plaque rupture and thrombosis formation(usually resulting in myocardial infarction) as well as from the more gradualluminal narrowing by the atherosclerotic plaque. (Rolak and Rokey 1990).The mechanisms involved in the development of the atheroscleroticplaque are not clearly understood. Because CAD is associated withhypercholesterolemia, cholesterol is generally believed to be implicated in thedevelopment of atheroclerosis. The hypothesis of cholesterol as a causative agentaccounts for the increased incidence of CAD seen in societies having highsaturated fat consumption and for the higher relative risk of cardiovascular eventsin individuals with high cholesterol levels, but it does not account for theimportance of cardiovascular risk factors such as hypertension and smoking.There are several other theories of athero8._nesis but one of the mostpopular, though unproven, is the 'response to injury' theory which proposes thatatherosclerotic lesions form on injured areas of the epitheliun. As the lesiongrows 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 atvessel branch points where shearing forces, which could induce endothelialinjury, would be higher. Hypertension as a risk factor could also be mediated bythe higher stresses placed on the endothelium (Rolak and Rokey 1990).As indicated above, blood flow limitation due to plaque progression is themost direct mechanism by which atherosclerosis results in symptomatic disease.An increase in the myocardial demand for oxygen above the level that thecompromised coronary artery can provide, results in myocardial ischemia and thesymptoms of classic angina, i.e., visceral discomfort (located appropriately forcardiac origin and usually sub-sternal) which is precipitated by increased cardiac7work and which is relieved promptly by rest. In recent years it has been recognizedthat mental, as well as physical, stress can provoke myocardial ischemia andangina (Rozanski et al. 1988). It is also recognized that factors other than themechanical obstruction of the atheromatous plaque contribute to myocardialischemia in patients with CAD. Angiographic studies of coronary vasomotionhave demonstrated that while normal coronary arteries dilate in response tostimuli such as cold or exercise, arteries with atheromatous irregularitiesparadoxically constrict. The cellular mechanisms which lead to this response arenot, however, completely understood (Nabel et al, 1988).Because the symptoms of clinical coronary artery disease result frommyocardial ischemia, the terms coronary artery disease and ischemic heart diseaseare usually used synonymously, although they are not in fact synonymous. Thearchetypal patient with CAD will have angina resulting from myocardialischemia. However, myocardial ischemia may be caused by disease other thanCAD and it is also possible to have angina, or chest pain resembling it, withouthaving CAD or even myocardial ischemia.People who have 'silent myocardial ischemia' have CAD and myocardialischemia but no anginal symptoms. In recent years there has been considerableresearch on this condition as the potential hazards have become apparent (Epstein,Quyyumi and Bonow 1988). Patients with anginal symptoms may also haveepisodes of silent ischemia. Deanfield et al (1983) showed that in patients withstable angina pectoris up to 70 percent of ischemic episodes are silent.Finally, there are people who are found, on angiography or at autopsy, tohave CAD with no previous history of functional disturbances or symptoms. Insummary, while angina is an important warning of the probable presence ofmyocardial ischemia resulting from CAD, not all people with angina have8myocardial ischemia and/or CAD, nor does the absence of angina indicate theabsence of ischemia and/or CAD.MANIFESTATIONS OF CORONARY ARTERY DISEASEAlthough angina is the classical symptom of CAD it is, unfortunately, notthe most common 'cardiac event' heralding the onset of the disease in men.Rolak and Rokey (1990) state that one percent of previously asymptomatic malesaged 30 to 62 will develop clinical manifestations of CAD each year. Of these, 42percent will present with an acute myocardial infarction, 38 percent with stableangina 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, occurswhen 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 fromthrombus 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 tenyears, for men and women respectively 30 percent and 40 percent will developangina, 16 percent and 24 percent will suffer stroke, 27 percent and 31 percent willdevelop 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 figureapply to untreated or to "primitively" treated CAD. The survival rate withmodern medical treatment is likely to be better than this.Evidence for better survival in more recent times, comes from acommunity based investigation which was part of a World Health Organizationcollaborative study. Data from the Perth coronary register on patients who9suffered an MI between 1971 and 1979 and who survived the first 28 days, showedone, 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 tohave improved survival over these figure as wel1 2 .Chronic Stable Angina:Angina which is predictable in frequency and duration and which can berelieved by nitrates and rest is termed "chronic stable angina". Framingham studydata (Weiner and Kannel 1987) showed that 30 percent of men and 40 percent ofwomen died within ten years of developing angina; a mortality rate 1.6-1.9 foldthat of the general population of the same age. Again, modern treatment maywell improve the survival chances of angina patients.Stable angina has been graded, according to severity of symptoms, by theCanadian Cardiovascular Society as follows:1. Class I Angina occurs only with strenuous or prolonged exertion atwork or recreation and does not occur with ordinary physical activity.2. Class II Angina occurs with walking rapidly on level ground or a gradeand with rapidly walking upstairs. Ordinary walking for less than twoblocks on the level or climbing one flight of stairs does not causeangina except during the first few hours after awakening, after meals,under emotional stress, in the wind or in cold weather. This impliesslight limitation of ordinary activity.3. Class III Angina occurs when walking less than two blocks on levelground at normal pace, under normal conditions or when climbing2 By reducing the incidence of life-threatening cardiac arrythmias, beta-blockers probably reduce theincidence of sudden death which generally results from a cardiac arrythmia.1 0one flight of stairs. This implies marked limitation of ordinaryphysical activity.4. Class IV Angina occurs with even mild activity, and may occur at restbut must be of brief (less than 15 minutes) duration. This impliesinability to carry out even mild physical activity.Unstable Angina:The term "unstable angina" has been applied to several syndromes, whichare taken as clear evidence of important, but reversible, myocardial ischemia. TheAmerican College of Cardiology/American Heart Association Task Force onAssessment of Diagnostic and Therapeutic Cardiovascular Procedures in reportingtheir guidelines and indications for CABS (Kirklin et al, 1991) used the followingcriteria for defining unstable angina. Patients with severe and persistent anginaon presentation to the physician (or hospital) with electrocardiographic evidenceof myocardial ischemia and only minor enzyme evidence (available later) ofmyocardial infarction. Also patients with new onset angina (Canadian Class IV)within two months of presentation or recurring or prolonged (greater than 15minutes duration) angina within 10 days of presentation even if such angina wasnot new. They also applied the term unstable angina to those patients who hadsevere angina within two weeks of a myocardial infarction. There was also arequirement that in all groups there would be ECG evidence of myocardialischemia during the severe pain but no evidence of more than minimalmyocardial necrosis. Braunwald (1989) has devised a classification for unstableangina but it is uncertain whether this is in common use.1 1Atherosclerosis in Non-coronary Vessels: So far this discussion has centered on the effects of atherosclerosis on thecoronary vessels. However, it should be recognized that atherosclerosis is not justrestricted to the coronary vessels but may occur in other vessels of which thecerebral vessels and those of the lower extremities are the most common. It maybe expected that the presence of atherosclerotic disease in the arterial system of onepart of the body would herald or be associated with the same disease in anotherpart of the body. The Framingham study (Dawber 1980) found that the rate ofclaudication, a symptom of peripheral vascular disease, was seven times greater insubjects with angina than in those with no anginal symptoms. Also the risk ofmyocardial infarction was found to be four times greater in those who hadpreviously had a cerebrovascular accident (stroke) than in those who had not.Dawber concludes that the presence of one manifestation of atherosclerotic diseasegreatly increases the probability of developing another and that frequently thesame risk factors are involved in contributing to the different categories ofatherosclerotic disease.RISK FACTORSFactors whose presence is associated with an increased probability that aparticular disease will develop later are called "risk factors" for that disease3 . Suchan association does not necessarily imply that the risk factor always causes thedisease, that all patients with risk factors will develop the disease or thateliminating, or decreasing, the risk factor will prevent the disease from occurring.The "risk" measured in risk factors is the "relative risk" (the incidence rate amongthe exposed/the incidence rate among the unexposed) which is derived from3 The term "risk marker" is often used to refer to risk factors, such as age and sex, which cannot bechanged.12measurement of the presence of risk factors and disease in groups. Factors whichare powerful predictors for a group may be weak predictors for individuals.Stages of Disease:Understanding the concept of risk factors requires an understanding of thenatural history of disease. Most diseases do not arrive out of the blue but arerather a culmination of a long process, interactions between environmental andindividual factors, which eventually result in disease. The natural history, thecourse of the untreated disease over time, is different for each disease but may bemodeled as a series of stages (Mausner and Kramer 1985).1. Stage of Susceptibility. In this stage the groundwork for CAD is laid bythe 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 assmoking and dietary habits, can be changed.2. Stage of Presymptomatic disease. In this stage pathologic changes, suchas atherosclerotic changes in the coronary arteries, have started to occuralthough there are no clinical symptoms.3. Stage of Clinical Disease. This stage starts when the signs and symptomsof the disease become discernible and continues for as long as thepatient is clinically ill from the disease. The diverse manifestations andoutcomes of CAD mean that patients with the disease may have verydifferent symptoms, prognoses and requirements for treatment.Consequently, patients in this stage are usually placed in homogenoussubgroups for the purposes of therapeutic management or study. Forclinical study this may be done on the grade of angina experienced, thelocation of the atherosclerotic block, the number of vessels that areblocked, the functional status of the left ventricle, or on some13combination of these factors, all of which have been shown to influenceprognosis and the response to treatment. Subgroups in epidemiologictrials are generally categorized according to their risk factors or tosymptomatology.4. Stage of Disability. CAD may give rise to a residual defect which canreduce patients' activity to the state where they are functionallydisabled. However, because it is so often fatal, CAD may result in lessindividual disability and community disruption than othercardiovascular diseases .Risk factors for CAD may be of greater or lesser importance depending onthe stage of disease. Some may make the person more susceptible to thedevelopment of atheroma while others may cause the atheroma to form (Pearson1991). Yet others may increase the likelihood of outcomes such as myocardialinfarction or sudden death (Dawber 1980). Confounding factors, those associatedwith a disease only because of their association with a true risk factor for thedisease, may confuse the picture.Primary and Secondary Risk Factors:The results of numerous studies have shown that the major risk factors fordeveloping clinically detectable coronary artery disease are hypercholesterolemia,hypertension, cigarette smoking and diabetes mellitus. These "primary" riskfactors are independent of one another in that they make a separate contributionto the risk of developing CAD but they can also be additive, because the presence ofmore than one further increases the risk (Dawber 1980). Other variables, secondaryrisk factors, appear to accentuate the action of the primary risk factors, and includeobesity, family history, physical inactivity, age, male gender and decreased levels ofHDL (Rolak and Rokay 1990).14The major findings surrounding primary and secondary risk factors arebriefly outlined below.Hypercholesterolemia:The Framingham Study (Dawber 1980) showed that risk of developing CADin males appears to be continuous with a gradient of increased risk of the diseasewith increases in the total cholesterol level For men in their thirties, the relativerisk for those with cholesterol levels of 260 milligrams per decilitre (mg/dl) ormore, was over four times that for those with levels below 200. The trend wasconsistent, though not as great, for men in the forties and fifties. The MultipleRisk Factor Intervention Trial (MRFIT) demonstrated that with each 50mg/ dlincrease in cholesterol the coronary event risk doubled, implying a linearrelationship between the level of cholesterol and risk (Multiple Risk FactorIntervention Trial Research Group 1982). Other studies have suggested a non-linear relationship with a "threshold", between 200 and 220 mg/dl, at whichdisease will develop (Pooling Project Research Group 1978). In any event, in malesthe effect of elevated total cholesterol diminishes with age.In women, the picture is somewhat different. The Framingham studyshowed a gradient of risk only for certain outcomes in certain age-groups. Forexample, higher levels of total cholesterol were associated with increased risk ofMI in women age 40-49, and with an increased risk of CAD in general for womenage 50-59. In women, unlike men, the total cholesterol level tended to increasewith age.Hypertension:The risk of CAD increases progressively as both systolic and diastolic bloodpressure increase. (The Pooling Project Research Group 1978, Dawber, 1980). In15The Framingham Study men who were in their thirties at the time of entry to thestudy, had an incidence rate of total CAD six times greater in those with systolicblood 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 pressuresincreased, with an overall difference of 2 years in the age of onset. Hypertensionalso showed a significant correlation with the clinical manifestations of anginapectoris, myocardial infarction and sudden death (defined as death within onehour of being well).Tobacco Smoking:Tobacco smoking is an independent risk factor for CAD in both males andfemales but its contribution to the development of CAD is accentuated by thepresence of other risk factors (Dawber 1980, Pooling Project Research Group 1978,Doll et al 1980). The highest risk comes from smoking cigarettes but increased riskhas been shown for pipe and cigar smokers as well (Pooling Project ResearchGroup 1978).While the risk for smokers increases for all manifestations of CAD thegreatest risk is for sudden death and MI. The risk declines with advancing age sothat by the age of 65 there is no added risk for smokers, except from sudden deathwhere the risk is still present but diminished (Dawber 1980).Diabetes Mellitus:The Framingham results (Dawber 1980) showed that diabetes contributessignificantly to all manifestations of CAD. The risk for male diabetics was morethan twice that for non-diabetic men, while diabetic women had a risk five timesgreater than that of non-diabetics. Although other risk factors (hypertension,obesity and elevated cholesterol) were significantly related to diabetes, these factors16did not wholly explain the added risk observed in diabetics. CAD is the primarycause of death in overt diabetics and this correlates best with the age of onset andduration of the diabetes rather than with the severity of the disease (Rolak andRokay 1990).High Density Lipoproteins:Rolak and Rokey (1990) cite numerous studies which have shown anegative correlation between high density lipoproteins (HDL) and CAD. Theseauthors report that, because low density lipoproteins (LDL) and HDL have oppositeeffects on the development of CAD, the ratio of LDL to HDL in an individual caninfluence the development of disease. Patients with ratios of less than two are atlowest risk while those with ratios above five are at greatest risk of developingCAD.Obesity:The Framingham Study (Dawber 1980) showed a clear relationship betweenrelative weight (actual weight/median weight for that age-sex group) and theincidence of CAD. This finding has been confirmed in other studies (Lew andGarfinkel 1979) although a direct causative independent role for obesity in thedevelopment of CAD has not been established. Obesity is associated with severalother risk factors for CAD, notably hypertension and abnormal lipid profiles, andthese may contribute the majority of the increased risks seen in obese people(Rolak and Rokey 1990). More recent research suggests that the distribution ofbody fat may play an independent role in the development of CAD (Wingard1990).17Family History:Family history of heart disease below the age of 65 appears to predispose toCAD with a two to four-fold increased risk of MI for the immediate relatives of MIvictims. Because many other risk factors (including hypertension, hyper-chol '-erolemia, obesity, and physical inactivity) are found in families with CAD,an independent role for genetic factors cannot be established (Rolak and Rokey1990).Physical Activity:The evidence that physical activity plays a direct independent role in theprevention of CAD is difficult to establish; in part because of difficulties instandardizing physical activity and in part because of its effect on other cardiac riskfactors.. Both Gau (1985) and Rolak and Rokey (1990) cite many studies whichshow an inverse relationship between physical activity and the incidence of CAD,MI and mortality from heart disease. The Harvard Alumni Study (Paffenbergerand Hyde 1984) showed that the benefit in reduced CAD incidence and in reducedmortality obtained from exercise in excess of 2000Kcal per week, was independentof smoking, obesity, hypertension and family history of CAD. Of the alumni whodeveloped CAD during the 12 to 15 year follow-up, those who exercised in excessof 2000Kcal per week had 71 percent of the coronary artery disease mortality of theinactive alumni.Age and Sex:The incidence of CAD is greatest in males, particularly in the 35-44 year agegroup where the risk is three to ten times that for females. With increasing age,particularly past 60, the risk gradient between the genders diminishes. Post-menopausal women have three times the incidence of CAD as their age18counterparts who are still menstruating, but the role of hormones in protectingwomen is still unknown. The lower total cholesterol, lower LDL's and higherHDL'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 diseasein individuals should not detract from their importance in the development andprogression of CAD. The role of risk factors as "triggers" in the manifestation ofcardiac events such as MI or sudden death does not appear to be understood andshould be an avenue for further study.TRENDS IN CORONARY ARTERY DISEASE MORTALITYAlthough deaths with the characteristics of acute coronary artery diseasehave been described as early as 3000 B.C. (MacKinnon 1987) it was not until thebeginning of this century that they became more than a rare event. Numerousstudies in North America and Europe have shown that CAD mortality rosedramatically from the early 1900's and then started a decline which is stillcontinuing (Nair et al 1990, Mackenbach et al, 1989, Thom and Maurer 1988, Slateret al 1985, Gillum, Folsom and Blackburn 1984, Feinleib 1984,). These studies showthat the onset of the decline, between the mid-sixties and 1972, varied betweencountries and between geographic regions within countries. Several studiesshowed a gender variation with the slowing or decline in acute coronary deathsbeginning in women up to a decade before it was seen in males (Gillum, Folsumand Blackburn 1984, Nicholls, Jung and Davies,1981, Slater et al 1985).There are several problems associated with using cause-specific mortalityrates4 to estimate the rise or fall in incidence of a disease. One is that a decline or4 The cause-specific mortality rate is calculated by dividing the number of people dying of a specificcause in a certain time period by the number of people at risk. This rate is influenced by the agedistribution of the population, the proportion of people dying of various causes and their average ageat death by cause.19rise in other causes of death, such as tuberculosis or influenza, may cause anincrease or decrease in the mortality rate of the disease of interest without anychange occurring in the incidences or prevalence 6 of that disease. Cause-specificdeath rates are also vulnerable to conventions and fads in the classification ofcauses of death and to classification problems such as the distinction betweenprimary versus underlying cause of death and periodic nomenclature revisions inthe International Classification of Disease (ICD).These problems have led to concerns that the rise in coronary artery diseasemortality in the first half of this century was just "a paper epidemic" (Slater et al1985). However, the use of analytic techniques which account for the problemsoutlined above (Slater et al 1985, Thom and Maurer 1988) have shown that theepidemic 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 theincidence of CAD has not changed but that the incidence of the more lethalmanifestations, such as MI or sudden death, have declined. Some authors havereported 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 whilethe incidence of the disease remains unchanged, the result will be an increasing5 The incidence rate of a disease measures the probability that healthy people will develop thatdisease during a specified period of time. That is, it is the number of new cases of the disease withinthe given time period divided by the number of people at risk for that disease.6 The prevalence rate of a disease measures the number of people in a population who have thedisease at a given time. It is calculated by dividing the number of existing cases of the disease at agiven time by the population at that time.7 A detection bias may occur when improved diagnostic procedures, or the more frequent use ofexisting techniques, leads to the identification of cases which would previously not have beenidentified. Up to 40 percent of MI's are "silent" and are detected only by repeat ECG examinations ofasymptomatic individuals. More frequent examinations could result in a gradual shift of theseundetected MI's to clinical diagnosis and treatment (Kuller, 1988).20prevalence of the disease. Evidence for increasing prevalence of CAD in threeregions of the US is shown by Feinleib et al (1988)Studies and reviews which have attempted to account for the decliningCAD mortality rate have reported a declining incidence of out-of-hospital CHDdeath, a decline in the age-adjusted incidence of hospital admissions for MI and adecline 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 decliningmortality rate may be due to changes in medical care. Slater et al (1985) argue thatdietary-changes and medical care improvements, which would affect men andwomen equally, cannot account for their results, which show different patterns inthe rise and fall of acute CAD incidence for males and females. Instead they claimthat the changes in the male-female differential for acute CAD are compatible withchanges in smoking behaviour which has also been gender-related. The majorityof 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 assumptionsdrawn from the literature, conclude that over 50 percent of the decline in CADmortality is due to reductions in serum cholesterol levels and cigarette smoking.A further 40 percent of the decline is due to specific medical interventionsincluding coronary care units, medical treatment (especially beta-blockade therapy)of clinical CAD, and the treatment of hypertension. Killip (1988) estimates thatcoronary artery bypass surgery (CABS) accounts for only one to two percent of thedecline in CAD mortality in the US.In summary, a true epidemic of acute CAD began in the first half of thiscentury and reached a peak in mid-century. Mortality from the disease is now onthe decline for reasons unknown but thought to be related to changes in medicalcare, changes in lifestyle and treatment of hypertension. It is still not known if the21incidence of CAD has declined; if not the decline in mortality will lead to anincreased prevalence of the disease. Investigation into CAD incidence andmortality in the US. is hampered by the absence of valid and reliable nationaldatabases. It is likely that this problem also exists in Canada.DIAGNOSIS IN CADCoronary Angiogram:The gold standard 8 for the diagnosis of coronary atherosclerosis is thepathologist's examination. Since this is not possible in most instances, a morereadily available technique is required as a proxy. The coronary angiogram isusually accepted as such a proxy, since it has good, but not perfect, correlationswith anatomic findings (Hlatky et al 1989). Comparisons of the findings bycoronary angiography with those in subsequent post-mortem examinations haveshown that significant under-estimations of atherosclerosis can occur. Also,significant intra-observer variability may be seen with conventional coronaryangiography. Quantitative techniques (e.g. digital subtraction angiography andthe use of computer automated techniques) to reduce this variability have beendeveloped, 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 diseaseand 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 goldstandard. The first occurs when two tests measuring the same phenonomen are compared with oneanother. e.g the pathologists examination at autopsy and coronary angiogram are both measuringanatomic coronary disease. In this case a specific hierarchy can be established in which one test, thegold standard, is seen as "more correct" than the other and will overrule the other if there isdisagreement between them. In the second type of comparison an explicit hierarchy cannot beestablished because the tests are measuring inherently different phenomena. Thus, in a patient withatypical syptoms has a normal exercise test (functional test for ischemia) but an abnormal coronaryangiogram (anatomic test) the clinician cannot determine whether the exercise test is a falsenegative or a true negative for myocardial ischemia.22The problem with the use of coronary arteriogram as a diagnostic test is thatit is invasive, expensive and has several potentially life-threateningcomplications, including death, MI, cerebro-vascular accident (CVA), ventriculararrythmias, local vascular complications and contrast agent reactions. Stewart et al(1990) found a 0.0024 rate for major complications (requiring immediate CABS) ina 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 leftmain coronary stenosis, severe three-vessel disease, multi-vessel disease with leftventricular dysfunction, advanced age or unstable angina are at significantlyhigher risk than patients without these complications (Chassin et al 1986, Fisch etal 1987). However, it is patients with these conditions (with the exception ofadvanced age) who are most likely to require angiography.The primary purpose of coronary angiography is to define the anatomy ofthe coronary arteries when such definition is needed for patient management.The procedure may also be used to assess the results of therapy and to helpformulate prognosis in patients with CAD, although prognosis has several otherdeterminants which are not discernible by angiography. Guidelines for the use ofangiography have been developed by the American College of Cardiology inconjunction with the American Heart Association (Fisch et al 1987).As ii.aicated above, coronary angiogram is not necessarily required in allcases of CAD. The diagnosis of angina may be reliably determined by a physicalexamination and history taken by an experienced clinician (Hlatky 1989). Otherinitial examinations for patients with chest pain suggestive of angina wouldinclude resting electrocardiogram, blood lipids and routine blood chemistry9 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 fromangiography. The actual number of patients experiencing major complications would likely be highersince this figure was calculated from the rate for low risk patients.23determination. For patients believed to have chronic stable angina who continueto be minimally or moderately symptomatic after initial treatment to control riskfactors and the use of nitrates, beta-blockers or calcium antagonists, an exercisestress test should be performed (Silverman and Grossman 1984).Exercise Stress Test:The exercise electrocardiogram ("stress test") is a functional test ofmyocardial ischemia10 . The main purpose of the test is to identify those patientswho may at high risk for future cardiac events (and exclude those who are not) sothat patients may be appropriately referred for angiography. When compared tocoronary angiography findings, the sensitivity" of the stress test is between 60 and70 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 exerciseheart rate have been shown to have independent effects on the sensitivity of thestress test (Chassin et al 1986). Restriction of the test to those patients whosehistory and initial examination indicate a high pretest likelihood of untowardcardiac events, would likely reduce false positives and may minimize the number10 There is no universally accepted "gold standard" for measuring myocardial ischemia. Insteadfunctional tests, such as the exercise electrocardiogram, thallium-201 scintigam, radionucllideangiogram or positron emission tomogram, are referenced to the coronary angiogram, an anatomic goldstandard. Without a single gold standard for myocardial ischemia, physicians have to rely on theconcordance 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 inquestion. When sensitivity is high, the number of 'false-negatives' (i.e. those who have the diseasebut 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 inquestion. When specificity is high the number of "false-positives" (i.e. those who do not have thedisease 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 orspecificity but was not discussed in the literature reviewed here.24of patients who receive unnecessary coronary angiograms. Use of the stress test forscreening apparently healthy individuals is not recommended (Froelicher et al1988).Patients with a negative stress test should continue to receive therapy butrequire no further evaluation at that time. Patients with a positive response butwith no indications of left main or three-vessel disease, likely also require nofurther evaluation though a more aggressive approach may be favoured in theyounger more active patients. Patients whose stress test shows evidence of leftmain or three-vessel disease should generally undergo coronary angiography(Silverman and Grossman 1984).Patients who may be referred directly for angiography without a prior stresstest may include those with chronic stable angina resistant to maximal medicaltherapy (Silverman and Grossman 1984), unstable angina when pain does notrespond to medical management (Chassin et al 1986), candidates for intracoronarythrombolysis when less than six hours have elapsed since the onset of chest pain,convalescent MI patients in whom angina develops at rest or on minimalexertion, 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 goldstandard for the diagnosis of CAD and its prognostic indicators, the test is notappropriate for all patients with suspected CAD because of its invasiveness,potential for mortality or serious complications and its expense. Many patientscan be diagnosed on the basis of the history, resting ECG and simple laboratorytests. Those in whom the diagnosis is equivocal or whose symptoms are notcontrolled with medical therapy should be further evaluated by means of a stresstest to identify those at high risk. These high risk patients are those who willrequire definition of coronary anatomy by coronary angiogram.25PREVENTION AND TREATMENT OF HEART DISEASEThe increasing prevalence of CAD over the years with the consequent highcosts to the individual and the health care system have led to a plethora ofresearch into prevention and treatment of the disease. While the concepts ofprevention and treatment would appear to be separate, in the case of CAD, andother 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 ofoutcomes such as MI or sudden death. Moreover, many of the same methods areused in both prevention and treatmentPrevention:As indicated above, the term prevention may mean more than theinhibition of the development of a disease before it occurs but may includemeasures that interrupt or slow the progression of the disease. For this reason,prevention may be divided into levels which have different expected outcomesand 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 atpreventing the development of atherosclerosis and CAD or, at least, delaying itsonset. There are two major approaches which may be used here. The first is thepopulation approach; the universal adoption of health education measures toencourage and motivate individuals to modify their lifestyle in the hope that suchchanges will decrease the incidence of CAD. Oliver (1987) states that the problemwith this approach is that there is no conclusive evidence that universal lifestylechanges would affect the incidence of CAD. However, because lifestyle changeshave been shown to affect the progression of atheromatous lesions, both bycausing regression of the lesions and by slowing progression (Ornish et al, 1990), it26appears likely that the population approach will delay the onset of CAD in manyindividuals 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 factorswhere present. Such screening measures include estimations of total cholesterolfrom blood samples and blood pressure measurement but could also includeweight, dietary and smoking history and family history of heart disease. Severalproblems are associated with this approach, the first being that the specificity 14 ofrisk factors is low. This raises certain ethical questions abo ,t how to deal withthose whose results show borderline risk of CAD, especially because treatment ofhypertension and of hypercholesterolemia may requirethe use of drugs whichthemselves 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 highrisk individuals has not been shown to significantly reduce their incidence ofeither heart disease or mortality from cardiac events (Mitchell, 1987), although thismay be due to the negative effects of the commonly used hypertensive drugs (beta-blockers and diuretics) on other risk factors (Kaplan 1991). There is little use inscreening if there is no effective intervention for those found to be at risk.However, the most recent report on mortality for hypertensive participants inMRFIT (Multiple Risk Factor Intervention Trial Research Group 1990) has shownthat CAD mortality for the treatment group, after 10.5 years of follow-up, was 15percent lower than for the control group. This reversal of unfavourable trends for14 As a "test" of who will develop CAD, risk factors are poor predictors in individuals. Forexample, two-thirds of a group of healthy men age 40-55 years who are at the highest risk (abovethe 80th percentile) as a result of hypercholesterolemia and hypertension can be estimated to remainwell over the next 25 years (Oliver, 1987). While the relative risk for these individuals is hightheir absolute risk is quite low.27the experimental group during the trial is attributed, in part, to a mid-trial changein the diuretic treatment for the experimental group.The use of drugs to reduce blood-lipids does appear to be effective accordingto a review by Blankenhorn (1989). The results of two randomized double-blindplacebo-controlled studies (The Lipid Research Clinics Coronary PrimaryPrevention Trial and the Helsinki Heart Study) showed that reducing blood lipidsresulted in reduced mortality and morbidity from heart disease. Turnstall-Pedoe(1987) reports that the cost per coronary death prevented in the Lipid ResearchClinics Trial was £1,000,000 in drug costs alone. The four randomized controlledtrials, reviewed by Blankenhorn, that used diet alone to reduce serum cholesterolin post-MI patients produced conflicting results. The trials were small, however,and the trial which achieved the largest reduction in cholesterol also found asignificant reduction (33 percent) in CAD events over 11 years of follow-up. Itshould be noted that these latter trials were conducted on people who had alreadyshown manifestations of CAD. It is possible that similar dietary measures wouldhave an even greater effect on those in the pre-clinical or susceptible stages of thedisease.The final problem with screening lies in the cost of the screening itself.While screening for hypertension can be done relatively inexpensively by GeneralPractitioners (GP's) on an opportunistic basis, screening for hypercholesterolemiais 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 riskfactors or to screen family members of people with known CAD.Secondary Prevention is applied in the early clinical stages of disease andrefers to the early detection and prompt treatment of disease. The goal is to halt orslow the progression of the disease, to prevent complications and to limitdisability. In the case of coronary artery disease the term "secondary prevention" is28also used to describe the strategy aimed at reduction of recurrence of MI (althoughthis is actually tertiary prevention since the disease has already occurred and leftresidual damage). In this latter case, prevention includes reduction of risk factors,treatment by medication or surgery and may include attempts at psychologicalchange (Mathes 1985).In secondary prevention, screening for early ischemia may result in a betterpay-off than risk-factor screening. Rose (1987), reviewing data from the WhitehallStudy, concluded that minor disease is a better predictor of major disease inindividuals than are risk factors. This conclusion is not surprising since whenminor disease is present the disease process has already started, but risk factors maybe present without disease. Again, the problems associated with screening, asoutlined above, apply.A recent randomized controlled US. study (Ornish 1990) has shown thatangiographically determined regression in atherosclerotic lesions can be achievedby means of a rigorous life-style change program involving diet, exercise and theuse of relaxation techniques. Eighty-two percent of the experimental groupachieved regressive changes in only one year while 53 percent of the control grouphad progressive changes. The authors state that the purpose of their study was todetermine what could be done, rather than what would be practicable in a largerpopulation of patients. While it seems likely that the regression of atheromawould reduce the incidence of CAD, this has yet to be proven.Tertiary Prevention consists of limitation of disability and rehabilitation inthose cases where disability has already occurred. Examples of tertiary preventionin CAD are rehabilitation following MI, treatment of angina, and treatment toreduce risk factors in patients who already show manifestations of CAD.29Treatment:Treatment of CAD may be divided into acute and chronic management ofthe disease.Acute management refers to treatment during and immediately after amyocardial infarct. Over the past few years the focus in acute management of MIhas changed from treating the mechanical and electrical complications tosalvaging jeopardized myocardium, thus decreasing infarct size. This may beachieved by increasing myocardial oxygen supply, utilizing pharmacologic ormechanical revascularization within four to six hours of the onset of symptoms(Rolak and Rokey 1990).Chronic medical management of patients who have experienced symptomsof CAD, involves risk factor control (including dietary changes, cessation ofsmoking and exercise) and may involve the use of drugs to prevent plateletaggregation15, myocardial arrhythmias 16 and angina 17 . Patients in certain clinicalsub-groups, or those whose angina is not adequately controlled by medicaltreatment, may benefit from revascularization procedures such as coronary artery15 Platelet adhesion and aggregation are the precursors to the development of coronary thrombiwhich may result in MI. The use of asprin as an antiplatelet agent has been shown to significantlyreduce the incidence rate of MI in patients with unstable angina (Theroux et al, 1988) and in elderlypatients with a previous MI (Sixty-Plus Reinfarction Study Research Group, 1980). It is notreccommended that asprin be used in the primary prevention of death or MI because of the risk ofheamoorhagic 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 toprotect against both MI and unstable angina.16 Hjalmarson (1985) reports that several randomized double-blind trials have demonstrated theinfluence of certain beta-blockers on the mortality and reinfarction rate of post-MI patients. Heattributes these outcomes, in part, to the anti-arrythmic properties of beta blockers. Calciumantagonists also have anti-arrythmic effects and are increasingly being used in secondaryprevention in CAD.17 Coronary vasodilating drugs (usually nitrates or nitrites) are used to prevent or relieve anginapectoris. A retrospective study by Bussman (1985) has indicated that the regular use of nitrates inCAD patients may decrease mortality.30bypass 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, presentsethical questions about the use of potentially harmful and possibly ineffectivetreatments in asymptomatic patients, and about screening costs. On the positiveside, there is evidence that control of risk factors by means of diet or drugs isefficacious in reducing CAD mortality and that rigorous lifestyle change canproduce a reversal of atheroma. In the absence of a "cure" for CAD, treatment isaimed at halting the progression of atherosclerosis through risk factor control andat prevention of "cardiac events" such as MI and sudden death.CONCLUSIONSCoronary artery disease, a chronic, progressive, "lifestyle" disease with arelatively high mortality rate, is a major problem in Canada in both human andeconomic terms. The declining mortality, unless accompanied by an equal orgreater 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, hasidentified four primary risk factors which have both an independent and anadditive effect on the development and progression of the disease. Attempts toreduce CAD incidence and/or mortality by the control of these risk factors,through drugs or lifestyle change, have had variable results. Overall, it appearsthat it is possible to reduce risk factors in individuals, and that these changes canlead 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 beeffective. Furthermore, the risks from the drugs to reduce hypertension or31hyperlipidemia may be greater than the risks from CAD in asymptomaticindividuals.Treatment of CAD is aimed at the general reduction of risk factors and at theprevention of disease progression and cardiac "events" such as MI and suddendeath. It is interesting that research on risk factors has not focused to any extent onfactors which may trigger cardiac events or those which may protect high riskpeople from such events. In a society such as ours where there is a highprevalence of the classic risk factors it is perhaps surprising that more people donot show manifestations of CAD; the reasons why they do not should beinvestigated.32REFERENCESBlankenhorn, D.H., 1989. Prevention or reversal of atherosclerosis: review ofcurrent 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 withcoronary heart disease. Chapter 14 in Secondary Prevention in Coronary Heart Disease and Myocardial Infarction. ed. P. Mathes, Boston: MartinusNijhoff Publishers.Chassin, M.R. et al., 1986. Indications for Selected Medical and Surgical Procedures- A Literature Review and Ratings of Appropriateness: CoronaryAngiography. Santa Monica: Rand Corporation. R-3204 /1-CWF/HF/HCFA/PMT/RWJ. Dawber, T.R. 1980. The Framingham Study: The Epidemiology of Atherosclerotic Disease.  Cambridge, Mass.: HarvardUniversity Press.Deanfield, J.E. et al., 1983. Myocardial ischemia during daily life in patients withstable angina: its relation to symptoms and heart rate changes. Lancet 2:753-758 quoted in Stephen Epstein, Arshed Quyyumi and Robert Bonow, 1988.Myocardial ischemia - silent or symptomatic. N Engl T Med 318 (April):1038-1043.Doll R., et al. 1980. Mortality in relation to smoking: 22 years of observation onfemale British doctors. Br Med T 280:967.Epstein, S.E., A.A. Quyyumi, and R.O. Bonow, 1988. Myocardial ischemia - silentor symptomatic. N Engl I Med 318 (April):1038-1043.  Feinleib, M., 1984. Changes in cardiovascular epidemiology since 1950. Bulletin of the NewYork Academy of Medicine 60 (June):449-464.Feinleib, M. et al. Regional Variations in Coronary Heart Disease Mortality andMorbidity. In Trends in Coronary Heart Disease Mortality: The Influence ofMedical 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-a-Month XXXIV(11):727.Gau, G.T., 1985. The role of physical exercise in secondary prevention. evidence todate. In Secondary Prevention in Coronary Heart Disease and MyocardialInfarction. ed. Peter Mathes, Boston: Martinus Nijhoff Publishers: 143-150Gillum, R.F., A.R. Folsom, and H. Blackburn, 1984. Decline in coronaryheart disease mortality. Old questions and new facts. The AmericanJournal of Medicine 76:1055-1065.33Goldberg, R.J. et al., 1986. Recent changes in attack and survival rates of acutemyocardial infarction. The Worcester heart attack study. TAMA 255(May):2774-2779.Goldman, L., and E.F. Cook, 1984. The decline in ischemic heart disease mortalityrates: An analysis of the comparative effects of medical interventions andchanges in lifestyle. Annals of Internal Medicine 101 (December):825-836.Hjalmarson A., 1985. Beta-blocking agents in secondary prevention. Chapter 9 ofSecondary 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: goldstandards, operational definitions and correlations. -1 Clin Epidemiol 42:381-384.Killip, T., 1988. Has coronary artery bypass surgery influenced mortality fromcardiovascular disease in the United States? Chapter 27 of Trends inCoronary 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 CoronaryHeart Disease and Related Diseases. In Trends in Coronary Heart DiseaseMortality: 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 750,000men 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 maleischemic heart disease in the Netherlands. ATPH 79 (December):1621-1627.MacKinnon, A.U., 1987. The origin of the modern epidemic of coronary arterydisease 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 myocardialinfarction: a nine year follow-up of the Perth Coronary Register. Circulation 65 (November):961-969.34Mausner, 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. Chapter2 (Session III) of Screening for Risk of Coronary Heart Disease: Proceedingsof a Workshop on Strategies for Screening for Risk of Coronary HeartDisease, 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 factorintervention trial: Risk factor changes in mortality results. TAMA 248:1465Multiple Risk Factor Intervention Trial Research Group, 1990. Mortality after 10years for hypertensive participants in the multiple risk factor interventiontrial. Circulation 82 (November):1616-1628.Nabel E.G. et al. 1987. Asymptomatic ischemia in patients with coronary arterydisease. TAMA 257:1923-1928.Nabel E.G., P. Ganz et al. 1988. Dilatation of normal and constriction ofatherosclerotic 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 inCanada. CMA Journal 125 (November): 981-992.Oliver, M.F., 1987. Problems and limitations. Chapter 1 of Session I Screening forRisk of Coronary Heart Disease: Proceedings of a Workshop on Strategies forScreening for Risk of Coronary Heart Disease,  ed. M. Oliver, M. Ashley-Miller and D. Wood. 3-10. Chichester: John Wiley and Sons Ltd.Ornish, Dean et al. 1990. Can lifestyle reverse coronary heart disease? The lifestyle heart trial. Lancet 336 (July):129-133.Paffenberger, R.S. and R.T. Hyde. 1984. Exercise in the prevention of coronarydisease. Preventive Medicine 13:3-22.Paffenberger R.S., W.E. Hale , R.J. Brand, and R.T. Hyde. 1977. Work energylevel, 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 andpractical 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 maledoctors. Br Med T 296:313, quoted in Loren A. Rolak and Roxann Rokey,35Coronary and Cerebral Vascular Disease: A Practical Guide. New York:Futura Publishing Company, Inc.: 291.Pooling Project Research Group. 1978. Relationship of blood pressure, serumcholesterol, smoking habit, relative weight, and ECG abnormalities toincidence of major coronary events. Final Report of the Pooling project. LChronic Dis 31:201.Report of the Sixty Plus Reinfarction Study Research Group, 1980. A double blindtrial to asses the long-term oral anticoagulant therapy in elderly patientsafter myocardial infarction. Lancet 2:989, quoted in L.A. Rolak and R.Rokey, Coronary and Cerebral Vascular Disease: A Practical Guide. NewYork: Futura Publishing Company, Inc.: 293.Rolak, L.A. and R. Rokey. 1990. Coronary and Cerebral Vascular Disease: APractical Guide. New York: Futura Publishing Company, Inc.Rose, G.A., 1987. CHD risk factors as a basis for screening. Chapter 2 (Session I) ofScreening 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 andSons Ltd.Ross, John Jr. et al., 1987. Guidelines for coronary angiography. A report of theAmerican College of Cardiology/American Heart Association task force onthe 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 myocardialischemia in patients with coronary artery disease. N Engl T Med 318:1005-1012Silverman, K.J. and Grossman, W., 1984. Angina pectoris: natural history andstrategies for evaluation and management. N Engl T Med 310 (June):1712-1717.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 ofcardiac 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 ongoingphysician's health study. N Engl T Med 318:262, quoted in Loren A. Rolakand Roxann Rokey, Coronary and Cerebral Vascular Disease: A PracticalGuide. New York: Futura Publishing Company, Inc.: 291.36Theroux, 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 PublishingCompany, Inc.: 292.Thom, T.J. and J. Maurer, 1988. Time trends for coronary heart disease mortalityand morbidity. In Trends in Coronary Heart Disease Mortality: TheInfluence 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) ofScreening 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 andSons Ltd.Weiner, D.A. and Kannel W.B., 1987. A comparison of surgical and medicaltherapy for coronary artery disease. CUR & R 37-46 quoted in Kannel, W.B.,1990. Coronary artery surgery revisited: Limitation of the intent-to-treatprinciple. Circulation 82(5):1859-1862.Wingard, Deborah L., 1990. Sex differences and coronary heart disease. A case ofcomparing apples and pears? Circulation 80 (May):1710-1712.37CHAPTER TWOCORONARY REVASCULARIZATION TECHNIQUESPART ICORONARY ARTERY BYPASS SURGERYAs indicated in Chapter One, the treatment of coronary artery disease isgenerally aimed at the prevention of the serious sequelae of the disease, e.g., MI orsudden death, as well as at control or reduction of angina. Although these aimsmay often be attained through risk factor reduction and the use of medication,increasingly surgical revascularization techniques are being used. Theseprocedures revascularize the heart muscle by either bypassing or removing thestenosis 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 thisoperation has become one of the most investigated surgical procedures in themedical literature. Twelve randomized controlled trials and numerousprospective and retrospective studies compare short and long-term outcomes ofCABS versus medical treatment for a variety of patient conditions and ages. Thereare a multitude of other studies on almost all aspects of the procedure.Following a brief history of the growth and utilization of CABS and thechanges in the procedure over the years, this chapter will present the majorfindings from the literature on the effectiveness in various clinical conditions andpatient characteristics, the risks and monetary costs of CABS.38HISTORYGrowth and Utilization:Although the earliest CABS procedure was reported (in 1974) as beingperformed in 1962, it was not until 1968 that the first clinical trials began at fourcentres in the U.S. The initial reports from these trials appeared in the literaturein 1969 (Miller 1977) and by 1970, when 3000 procedures were performed in theU.S., bypass grafting was by far the most widely practiced type of direct coronaryartery surgery (Preston 1989).In describing the early growth of CABS, Miller (1977) states: "Such a rapidand large-scale application of a new operation is without parallel in the history ofsurgery". The rapid growth in the U.S has continued, rising from 57,000procedures (26 per 100,000 population) in 1975 to 284,000 procedures (122 per100,000 population) in 1986 (Preston 1989)In Canada the trend is also to increased growth of CABS but at a slower pacethan in the U.S. Peters et al (1990) report that between 1981 and 1986 the number ofCABS performed in Canada rose from 7,825 (31 per 100,000) to 10,865 (43.2 per100,000), while in B.C. they rose from 1,042 (37.7 per 100,000) to 1,140 (39.0 per100,000). In the six years reported by Peters, B.C. fell from being the province withthe highest unadjusted CABS per population rate to the third lowest.These aggregate rates obscure different trends in time and among the agegroups. In the Canadian data Peters et al., found three distinct phases in thegrowth of numbers CABS. From 1981 to 1983 there was an increase of 29.4 percentfollowed by a 2.2 percent decline over the next two years. The third phase, 1985 to1987, showed a moderate increase around 7 percent. Gillum (1987), who comparedpatterns of CABS utilization across the U.S. from 1979 to 1983, found similarphases though they were different for women and men. He also found that blackshad substantially lower utilization rates for CABS than did whites and that these39lower rates were not consistent with mortality and prevalence data, which, overall,are not substantially different for the two races. This discrepancy between blacksand whites is also borne out by Ford et al (1989).Both Peters and Gillum found that rates were highest for males in the 55 to64 age group. In the Canadian data Peters found that the numbers of CABSperformed in the youngest age group, 34 to 55-year old's, showed a sustaineddecline over the time period for both sexes. The 55 to 64 year olds show no growthafter 1983 while growth for both male and female over-65-year olds increasedsteadily throughout the times studied.Gillum did not report on trends over time for different age groups but notesthat the patterns of CABS by age and sex correspond more closely with the patternsof coronary heart disease prevalence than to coronary heart disease mortality orincidence. He reports that the peak rates for mortality and incidence occur in theover-65's while those for prevalence are in the 55 to 64-year old group for men andin the 65 to 74-year old group for women; these age-sex groups correspond to thosewith the peak CABS rates although the male:female ratio for CABS is substantiallyhigher than similar sex ratios for mortality or for prevalence. Feinleib et al (1989),in their analysis of CABS and angioplasty discharges in National HospitalDischarge Survey Data, also indicate that CABS is performed for chronic, ratherthan acute, ischemic heart disease.In both Canada and the U.S., the rates for the elderly have increaseddramatically. Gillum (1987) reports a 138 percent increase in rates for the over-65'sin the U.S. between 1979 and 1983. In Canada, Roos and Cageorge (1987) found thatCABS rates for the over-75 population in Manitoba had more than doubledbetween 1980 and 1983. In Ontario, the rates for the over-70 population increasedfive-fold between 1979 and 1985 although the overall CABS rate in the provinceincreased by only 39 percent during this period (Anderson and Lomas 1988).40In summary, after its inception in 1968 CABS showed a phenomenalgrowth, unparalleled by any other procedure in the history of medicine. Theinitial rapid growth during the seventies and early eighties started to plateau in themid-eighties as the procedure became diffused. The moderate increases seen afterthe plateau period appear to be due to the increase in the numbers of proceduresdone on the elderly. According to Peters et al (1990) this may indicate an increasingacceptance of the procedure for elderly patients with the result that the proceduremay not be completely diffused in this group.Assessment:In the more than twenty years which have passed since CABS was firstintroduced, the methods used to assess the efficacy of new procedures haveincreased in sophistication. The early reports on the efficacy of CABS were basedon observation of, and anecdotal reports from, patients and comparison ofoutcomes after surgery with the natural history of coronary artery disease. Patients,serving as their own controls, were assessed before and after surgery and weregenerally found to be improved. The assumption was made that thisimprovement was due to specific effects of CABS. Preston (1990) points out thatthis assumption ignored earlier reports that angina may naturally abate in somepatients over time. Yeaton and Wortman (1985), in an analysis of 90 studies todetermine changes over time in CABS assessment and outcome, found thatauthors of observational studies expressed significantly more enthusiasm forsurgery as favourably influencing survival, than did the authors reporting RCTs.Preston states that, by the early seventies some institutions had accumulatedenough cases to make retrospective analyses based on comparison of the surgicalgroup with a non-surgical control group. Many of these control groups had beendiagnosed and treated at a different time from the surgical group, and so were not a41valid comparison, though many physicians did not recognize that at the time.Later studies attempted to create more valid control groups by matching treatmentand control patients on important parameters. However, differences in importantbut unmatched prognostic indicators remained in some studies (Preston 1990).The first randomized controlled trials (RCTs) comparing CABS withmedical treatment, were started in 1971 and 1972 (see Table I). By randomlyassigning patients to either medical or surgical treatment, they eliminated the biasthat occurs when physicians choose medical or surgical treatment for the patient,and also maximized the likelihood that baseline characteristics would be the samein 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 toshow differences in mortality (Chassin et al 1986), were quickly followed by severallarger 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 intosubgroups for analysis and the first study to find increased survival in anysubgroup. Retrospective placement of patients into sub-groups showed improvedsurvival in patients with left main disease who received surgery. This studyshowed the importance of sub-group analysis, a technique which was incorporatedinto all subsequent RCTs (Preston 1990).The last RCT comparing CABS with medical treatment (The CoronaryArtery Surgery Study - CASS) randomized patients between 1975 and 1979 andissued its first reports in 1983. All studies on the outcomes of CABS since thenhave been either retrospective analyses or non-randomized prospective studieswith or without controls. Preston (1990) states that, unfortunately, there appears to42be a consensus among clinicians that the definitive studies on outcomes have beendone and there is no need for further RCTs, he goes on to say: "If anything we areregressing. We are like natives in a jungle hospital abandoning the newermethods after the Westerners have left".Changes in CABS Procedure:When interpreting the literature on the efficacy of CABS, it is important toknow the changes which have occurred in the procedure over time. Theprocedure now is very different from that in 1979 when the last randomizedcontrolled trial finished. These differences have resulted in decreased intra-operative and post-operative mortality and morbidity and improved long-termoutcomes from the procedure.Intra-operative changes which have contributed to better short-termoutcomes are cold cardioplegia, anaesthetic techniques which help to preserve themyocardium, and shorter operative time. The use of medications such as post-operative aspirin, may also contribute to short-term outcome.Improved long-term outcomes have been obtained by the use of the internalmammary arteries (IMA). These conduits were always favoured by somesurgeons (Preston 1990) but did not become routinely used until the early-to-mideighties when reports of the increased long-term patency of these grafts comparedto vein grafts, began to appear (Barner et al. 1982, Loop et al. 1986). The highestpatency rates, approximately 95 percent after 10 years, are achieved by the use of theIMA to bypass stenosis of the left descending artery. IMA grafts to other coronaryvessels appear to have lower patency rates and these may be no greater than thoseof vein grafts (Huddleston et al 1986).In contrast to the IMA patency rate, the long-term patency rates forsaphenous vein grafts are highly variable; some studies report only 50 to 60 percent43of grafts still patent at 10 years. Kroncke et al (1988), analyzing data from theVeteran's Administration randomized trial, found that progression occurred in 74percent of arteries with patent grafts at five years. Progression occurred in 38percent of ungrafted arteries in surgical patients and in 36 percent of arteries inmedical patients. After adjustment for baseline disease severity the risk ofprogression was three to six times higher in grafted arteries than in ungraftedarteries. It appears as though CABS, while relieving the patient of one arterialobstruction, may hasten the onset of others.Changes in Medical Treatment: The majority of studies looking at the efficacy of CABS, compare the short-and long-term outcomes following the procedure with those in patients receivingmedical treatment. Advances in the medical treatment of CAD over the last decadehave likely reduced difference between these outcomes. Consequently, some ofthe findings of early studies (including the RCTs) may not be relevant today. Forexample, all the randomized trials pre-date the wide-spread use of beta blockerspost MI, and the use of calcium channel antagonists (only introduced into NorthAmerica in the late-seventies), long-acting nitrates and platelet-inhibiting agentswhich have allowed for better control of CAD so that it is now seldom impossibleto relieve angina (Kannel 1990).Changes in CABS Patients:Changes in the type of patients receiving the procedure over time are ofinterest, both because they indicate continued diffusion of the procedure andbecause they indicate how relevant the findings of earlier studies are to laterpatient populations. Advances in cardiology and in the CABS procedure appear tohave resulted in a change in the profile of patients undergoing CABS. More44patients today are over the age of 65, have three-vessel disease, left ventriculardysfunction, associated medical diseases, emergency surgery or are receiving re-operations. A U.S. study (Naunheim et al 1988) found an increase in operativemortality 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 nosignificant change in mortality between 1982 and 1986, although in-hospitalmorbidity increased from 10.1 percent 13.3 percent.Neither of the above studies explored the reasons for the increase in high-risk patients undergoing CABS but both suggest that it is, at least in part, due to theincrease in the use of percutaneous transluminal coronary angioplasty (PTCA).Other factors will likely also play a part. Advances in medical therapy may meanthat patients come later to surgery than in earlier years. Also the increased comfortand proficiency with CABS that cardiac surgeons gain with experience, togetherwith the technical advances in the procedure, may make them more likely toaccept sicker patients.EFFICACY OF CABSThe efficacy of a procedure is a measure of whether it can do what it issupposed to do. Efficacy is related to, and is required for, effectiveness, a measureof how well the procedure can achieve the desired end under real worldconditions 14 . Efficacy of CABS is assessed by randomised controlled trials (RCTs).That is, by comparing outcomes in patients randomly selected, from the samepopulation, to receive either medical treatment (the conventional treatment) or14 The distinction between efficacy and effectiveness becomes important in evaluating the usefulnessof a medical treatment. If a treatment is not efficacious then it is of no use. It is efficacious but noteffective 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 generallyrequired for the effective treatment of CAD. Without these lifestyle changes, efficacious medicationmay not be effective.45CABS. Ideally, the treatments for each group would be standardized as far aspossible so that physicians would follow a protocol in, for example, prescribingmedication 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 dependto a great extent on their clinical condition. Consequently, proven efficacy for onecondition cannot be assumed for another. The second condition should also besubjected to rigorous evaluation in a randomized controlled trial.This section will review the literature on the efficacy of CABS for a numberof clinical conditions which result from CAD. In situations where there are noRCTs testing efficacy in a clinical population, or where the findings of RCTsconflict, evidence from non-randomized controlled studies (prospective orretrospective) will be presented. In addition, observational studies will be used tolook at outcomes in surgical populations who have not been studied in acontrolled 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. Thatmeans that patient data were analyzed according to the group to which the patientwas originally randomized. Thus, patients randomized to surgical treatment whonever received surgery were analyzed in the surgical group, and medical patientswho later received surgery, "crossovers", were analyzed as part of the medicalgroup. 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 over15 The Veteran's Administration Trial, the European Coronary Surgery Study and the CoronaryArtery Surgery Study.46while, in the VA study, 47 percent of medical patients with left main disease hadcrossed over by the 11 year follow-up.The result of analysis according to the intent-to-treat principle is that thecomparison is not surgical therapy versus medical therapy but rather the clinicalstrategy of immediate surgery (followed by medical therapy as required) versusmedical therapy with the option of later surgery (Gersh 1989). This is more anevaluation of effectiveness than of efficacy. In this situation, analysis by the intent-to-treat principle minimizes the difference in long term outcome. Kannel (1990)points out that with both groups (medically and surgically assigned) of patients inCASS receiving medical treatment and a 40 percent crossover to surgery from themedical group, the 10-year summary finding of no difference in survival orinfarction rate may be due to the fact that at 10 years there was little difference intreatment between groups. Analysis according to treatment received would give abetter indication of long-term efficacy, as would censoring 16 patients crossing overfrom medical to surgical therapy (providing that these patients had an acceptablesurgical mortality).Chronic Stable Angina:All three major RCTs showed that CABS was extremely effective inameliorating angina pain in patients with chronic stable angina (CSA), and thisfinding was also supported by the results of the small RCTs. Therefore, thediscussion on the effectiveness of CABS in patients with CSA will focus on thefindings related to survival; less focus was put on the incidence of MI in thesestudies. Only the three major RCTs will be discussed in detail because, as indicated16 Patients who are "censored" are analysed as though they were lost to follow-up.47earlier, the smaller RCTs had populations which were too small to detectdifferences in survival between treatment and control groups.The earliest randomized trial, the Veterans Administration cooperativestudy of coronary artery surgery (VA trial), was initially started in 1970 as anevaluation of surgical treatment for CAD and included the Vineberg implant.Because the majority of the patients were randomized in the CABS protoco1 17, thestudy was restarted in 1972 and the Vineberg procedure was dropped. Patientswere randomized from January 1972 to December 1974. Patient criteria forinclusion are presented in Table 1; excluded were patients with MI less than sixmonths previously, those with other serious cardiac diseases, other major diseaseswhich 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 patientswith significant left main disease (Takero et al 1976). These results showed asignificant advantage for the surgical group at 30 months; survival for the 60surgically treated patients was 85 percent compared to 65 percent for the 53medically treated patients. These patients were, however, assigned to subgroupsretrospectively rather than being prospectively randomized to them. The resultswere still significant at five years of follow-up but were not reported at ten yearsbecause by then 91 percent of patients in the medical subgroup had either died orcrossed over to surgery. Unfortunately, insufficient data were presented in thereport to allow calculation of the ten-year outcomes according to the intent-to treatprinciple.17 Obviously the procedure for randomizing patients was not effective!48TABLE 1INCLUSION CRITERIA FOR THE THREE MAJOR RCTsStudy VA Trial ECSS CASSDates 1/72-12/74 9/73-3/76 8/75-5/79No of Patients 686 768 620Age and Sex Male Veterans* Male under 65 Under 66Angina Present^at^least^6 Present^at^least^3 CCS Class I or II only or post-months with medicaltreatment at least 3monthsmonths MI (at lest 3 weeks prior torandomization) with noangina.Disease extent Reduction^at^least Obstruction 50% or Narrowing of 70% or more in50% in at^least one more in at least two at least one major artery withartery^withgraftable^distalsegmentmajor arteries. operable distal segment.LVEF** 25% or more 50% or more 35% or more*All ages included. The oldest patient was 67.**Left ventricular ejection fractionl 8Results for all patients at 30 months showed no differences in survival inpatients with one, two or three-vessel disease as a whole (Murphy et al 1977). Laterretrospective analysis, however, showed that when data from three hospitals withan aggregate mortality of 23 percent (and performing only 13 percent of thesurgery) were excluded 19 , the 6-year survival rate of surgically treated patients withthree-vessel disease was significantly better than that of medically treated patients(Takero et al 1982). Retrospective analysis at six years also showed that patients18 The left ventricular ejection fraction, the percentage of blood that the left ventricle expels perheartbeat, is a measure of left ventricular function.19 We are not given any information about these patients, hospitals or surgeons so cannot determinewhether 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 whencompared to mortality rates in the other hospitals used in the RCT. Consequently, it cannot beestablished whether Takero et al were justified in excluding patients in these hospital from theanalysis.49without left main disease, but considered to be at high clinical and angiographicrisk20 had significantly better survival with surgery than with medical treatment;surgical patients at high risk angiographically but not clinically had a slight, non-significant, survival advantage. Patients in low risk groups, however, had amarginally 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 inhigh-risk patients (measured both clinically and angiographically, i.e., with threevessel disease and impaired LVF). Patients with normal LVF and low clinical riskshowed a survival disadvantage with surgical treatment although this was onlysignificant for patients with two-vessel disease. The investigators concluded thatamong patients with stable ischemic heart disease those with a high risk of dyingbenefit from surgical treatment but that this benefit gradually diminishes afterseven years (The Veterans Administration Coronary Artery Bypass Surgery StudyGroup 1984).The VA trial has been criticized for its high operative mortality rate, 5.8percent, which likely handicapped the surgical group. In response to these critics,Takero et al (1983) cite several studies which show similar mortality rates duringthe same time period, an era in which CABS was just becoming established. Theaverage annual mortality rate for patients without left main disease during thefirst seven years of follow-up was 3.3 percent for the surgical group and 4.0 percentfor the medical group. Between seven and eleven years annual mortality for thesegroups were 4.8 and 3.5 percent respectively. The increasing mortality for the20 Clinical risk, high medium or low, was determined on the basis of a multivariate risk functionusing four clinical variables which were measured at base-line. These variables were: amount ofangina according to the New York Heart Association classification, history of hypertension, historyof myocardial infarction and an ST-segment depression on the resting ECG. Angiographic risk wasdetermined by left ventricular function and by the number of diseased vessels (The VeteransAdministration Coronary Artery Bypass Surgery Study Group 1984)50surgical group after seven years is likely due to progression of the disease leadingto late graft closures (Campeau et al 1983).TABLE 2CHARACTERISTICS OF PATIENTS IN RCTs FOR STABLE ANGINADatesStudyCASS(N=780)VA(N=686)ECSS(N=768)1972-74 1973-76 1975-79Characteristic % % %Male 90 100 100Anginanone 26 0 0Class I, II 74 42 57Class III, IV 0 58 43Prior MI 60 61 46LVEF < 0.5 21 26 0Data from Alderman et al (1990)The European Coronary Surgery Study (ECSS) randomized 768 males underthe age of 65 years between September 1973 and March 1976. Inclusion criteria areshown in Table I; no exclusion criteria were reported. Characteristics of patientsincluded in the study are shown in Table 2. Survival data reported at two years offollow-up showed that there was no significant treatment advantage overallthough there was a significantly higher survival rate with surgery for the sub-group 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 significantimprovement of survival was found overall for the group randomized to surgeryand for the surgical sub-group with left main disease (The European CoronarySurgery Study Group 1980). By 10 to 12 years of follow-up the benefit of surgery forthe overall group was still significant at the .05 level but the difference in survivalbetween medical and surgical groups was much less than it had been at five years.51In addition there was a significant survival advantage overall for surgical patientswith stenosis of the left anterior descending (LAD) artery. However, when thesepatients were stratified by extent of disease, those with two-vessel disease did notbenefit significantly from surgery (Varnuskas et al 1988).The differences between the overall results from the VA trial and ECSS maybe due to the differences in operative mortality between the two studies. Operativemortality in the ECSS was 3.6 percent overall (1.5 percent in the last third ofpatients) compared to 5.8 percent in the VA trial. Six of the ECSS surgical patientsdied before surgery. These studies also differed on the effect of surgery in patientswith normal LVF and three vessel disease. Chassin et al (1986) report that the ECSSresults prompted the reanalysis of the VA data, excluding the hospitals withextremely high surgical mortality. These authors comment: "This reanalysis doesnot alter the original study's conclusion that CABS failed to improve survival inthe study population with three-vessel disease. It is, however, a useful exercise inthe effort to resolve differences among RCTs on this important issue."The National Heart, Lung and Blood Institute's coronary artery surgerystudy (CASS) randomized 780 patients between 1975 and 1979. The populationfrom which these patients were drawn consisted of 94 percent of all patients at 15participating centres patients who had coronary angiography for suspected CAD.Of the 16,626 eligible and consenting patients, 2,099 were eligible for randomizationbut only 37 percent of these were randomized 21 . The non-randomized patientswere enrolled in the CASS registry and were followed-up by essentially the samemethods used to follow the randomized patients (Principle Investigators of CASSand their Associates 1981).21 It is not clear why more of the eligible patients were not randomized.52Criteria for inclusion in CASS are given in Table I. Excluded were thosewith previous CABS, the presence of another illness which would reduce the 5-year life expectancy, unstable angina or obstruction of the left coronary artery ofmore than 70 percent. There were no randomized patients with left main disease.Patients in the medical and surgical groups had similar clinical and anatomicfeatures. Follow-up at five, seven and ten years did not demonstrate anysignificant treatment advantage overall or for sub-groups stratified either on thebasis of one, two or three-vessel disease or according to LAD stenosis (CASSPrinciple Investigators and their Associates 1983).For patients with a normal ejection fraction, long-term survival was best forpatients randomized to medical therapy. Surgical patients in this cohortexperienced a greater frequency of coronary events (death or MI) after 4-5 years. Inpatients with an impaired ejection fraction and three-vessel disease, however,survival was significantly improved in the surgical group. At ten years of follow-up surgical patients in this sub-group showed 80 percent survival compared toonly 59 percent in medical patients. This difference in outcome started to becomeapparent 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-risksubset reached a maximum at seven years and was no longer apparent by elevenyears.All of the above RCTs were multicentre trials and, with the exception ofCASS, none of them standardized treatment. Consequently it is impossible to saywith certainty how CABS compares with maximum medical treatment.Multicentres also posed problems for standardization of surgery. There was a widerange in the operative mortality rate among hospitals in the VA trial and in theCASS. Finally, as the authors point out, the sample size of the CASS was likely too53TABLE 3OUTCOMES IN RCTs FOR STABLE ANGINAOutcomeCASSMed^Surg(N=390)^(N=390)VA TRIALMed^Surg(N=354)^(N=332)ECSSMed(N=373)Surg(N=394)% % % % % %Operative death - 1.4 - 5.8 - 3.2Perioperative MI 8 6.4 9.9 - <8.0Survival5 year 92 95 85 90 83 92*10 year 79 82 57 58 70 76*LMD (n=6) (n=8) (n=53) (n=60) (n=31) (n=28)5/6 year - 70 95* 65 79*10 year - - - 61 64LAD stenosis (n=275) (n=277) (n=240) (n=262)5/6 year 92 94 - 79 90*10 year 78 82 - 65 76Impaired LVF (n=82) (n=78) (n=72) (n=75) (n=0) (n=0)5/6 year 85 96 73 80 N/A N/A10/11 year 59 80 49 53 N/A N/A3-vess dis.5/6 year 78 91 66 83 N/A N/A10/11 year 57 75* 38 50 N/A N/A3-vessel disease (n=135) (n=123) (n=156) (n=135) (n=186) (n=220)5/6 year 90 93 75 81 82 94*10/11 year 75 76 50 56 68 78*Reoperations - 9.7 11 - 9.9Crossovers 38 7** 38 6** 36 5.8**Graft patency18 month - 90 70 - 7560 month - 82 - 67 69LMD left main disease, LAD left anterior descending artery, LVF left ventricular dysfunctionData 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.54small to detect a difference in mortality, given the minimal difference in outcomebetween the medical and surgical groups.Taken as a whole the data from the three randomized trials indicates adefinite surgical advantage for patients with left main disease and for those with amoderately impaired ejection fraction with three-vessel disease. In addition theECSS showed a significant surgical advantage for patients with a normal ejectionfraction and three vessel disease. The VA results can be reconciled with this butthe CASS findings are in conflict.In an editorial, Killip (1988), states that the outcome differences betweenCASS and the two earlier trials may be reconciled by comparison of their patientpopulations. The CASS patients likely had less severe disease than patients ineither of the other trials; they were specifically selected because they were at lowrisk, on the grounds that high-risk patients should not be denied the chance ofimmediate surgery. The medical mortality of only 2 percent per year is testimonyto the low-risk status of these patients. Killip states that low-risk patients will dowell 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 to1981, tentatively propose that the surgical survival benefit seen in the earlier trialswas 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 percentof the VA trial and ECSS respectively.The results from other studies also appear to conflict on this point. TheSeattle Heart Watch, a community-wide observational study of patients whounderwent angiography between 1969 and 1974, showed improved surgicalsurvival in patients who had three-vessel disease and impaired LVF, but not inthose with normal LVF (Hammermeister, DeRouen and Dodge 1980). Analysis ofdata from the 5,809 patients in the Duke registry indicated that in 1970 surgery for55one- or two-vessel disease was associated with poorer survival, and no significanttreatment difference was found for three-vessel disease. By 1984, however,patients with two- or three-vessel disease had significantly longer survival withsurgery (Califf et al 1989). The surgical advantage for three-vessel disease in thisstudy was already noticeable by 1977; it may be that the differences in time betweenthe VA and ECSS trials accounts for the different outcomes for patients withnormal LVF and three-vessel disease. It should be noted though, that Califf andcolleagues did not look at the extent of disease according to degree of LVF (i.e.,subgroup analysis) but instead adjusted for baseline differences in ventricularfunction. This may have affected the results.Despite the diverse findings from the literature on CABS for patients withCSA and three-vessel disease with normal LVF, many studies advocate it as anindication for CABS (Kirklin et al 1991, Gersh et al 1989). The likelihood is that, astime passes and the RCTs recede further in to the past, three-vessel disease willbecome an accepted indicator for elective CABS.For patients with stable angina a number of other variables besides thenumber of vessels involved, affect survival. Of these the most important are leftventricular function and severity of angina.Left ventricular function alone has been shown to be the most powerfuldeterminant of long-term survival in chronic stable angina. Gersh et al (1989)report that the substantial benefit of surgery demonstrated by medical registrystudies for patients with impaired LVF, have changed this characteristic from arelative 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 inpatients with left ventricular dysfunction, the comparative benefits (i.e., betweenmedical and surgical treatment) are better. These comparative benefits are56especially great when other risk factors, such as extensive coronary disease andsevere ischemia, are also present.Left ventricular dysfunction is also a significant predictor of operativemortality in CABS. Parsonnet, Dean and Bernstein (1989) found that operativemortality increased four percent when the ejection fraction was less than 30percent. In a prospective, multi-institution study in Toronto between 1982 and1986, Christakis et al (1989), using multivariate analysis, found that a LVEF of lessthan 40 percent was the second most important independent predictor of operativemortality overall, but that this changed over time. In 1982 LVEF was the mostimportant predictor of mortality but it had fallen to fourth place by 1986. Overalloperative mortality for these patients was not reported, but for those with LVEFless than 20 percent, operative mortality was 11.7 percent compared to 2.5 percentfor those with LVEF greater than 40 percent.Kirklin et al (1991) state that extreme left ventricular dysfunction probablyresults from myocardial scarring and consequent irreversible ischemia. In such acase the prognosis is limited whatever the treatment. The level of LVEF whichindicates lack of surgical benefit is not known but is probably below two percent.Severity of angina appears to be a powerful prognostic factor in patientstreated medically but not in those receiving CABS. The benefit of surgery onsurvival is greatest in patients with severe symptoms (Kirklin et al 1991, Gersh etal 1989). This is likely because the severity of angina is generally a reflection of theseverity of myocardial ischemia. Ischemia on exercise testing has been associatedwith lower survival and a higher rate of infarction in medical patients (Hlatky et al1987, Weiner et al 1987) even in the absence of severe symptoms (Bonow et al1984).In summary, for men under 65 years of age, the major issues to be taken intoaccount when considering surgery are the degree of LVF, the location and severity57of the stenoses, the extent of disease and the severity of symptoms. Patients withleft ventricular dysfunction and/or left main disease will receive a significantsurvival benefit from initial surgery while those with three-vessel disease, severesymptoms or with stenosis of the LAD artery may experience a survival benefit.Patients with one- or two-vessel disease, normal ventricular function, mild tomoderate symptoms and no exercise-induced ischemia have an excellentprognosis with medical treatment and initial surgery is probably not indicated.Unstable Angina:There have been four randomized trials which compared medical andsurgical treatment for unstable angina, but only two of them, the NationalCooperative Study and the Veterans Administration trial will be discussed as theothers were too small to detect significant differences in survival.The National Heart, Lung, and Blood Institute (NIH) conducted a trial innine cooperating centres between 1972 and 1976. Inclusion criteria are given inTable 4. Excluded were patients who had recent MI, left main disease, LVEF of lessthan 30 percent or a comorbid disease which would reduce life-expectancy to lessthan five years (Russell et al 1976).Two hundred and eighty-eight patients were randomized to medical orsurgical treatment. In-hospital (operative) mortality was 3 and 5 percent for themedical and surgical groups respectively. The greatest operative mortality wasseen for surgical patients with three-vessel disease, although this was notsignificant. Non-fatal in-hospital MI's were experienced by eight percent of themedical group and 17 percent of the surgical group; a significant difference (Russellet al 1978).At 30 months of follow-up, (Russell et al 1978), 36 percent of the patientsassigned to medical treatment had crossed over. Overall survival and MI rates58were similar for medical and surgical groups (see table 5). Significantly morepatients in the medical group had class III or IV angina, by the New York HeartTABLE 4INCLUSION CRITERIA FOR RCTs OF UNSTABLE ANGINACriteriaDatesNIH Study1972-1976VA Trial1976-1982# of patients 288 468Sex and age <70 male <70LVEF >30% >30%Signs and Symptoms One or more attacks of Type I angina: present > 2severe coronary pain at rest,associated with S-T or Twave changes or both.monthsIncrease^in^severity,frequency or angina at restAdmission to a coronary within 8 weeks prior tocare^unit^for^suspected entry. Pain within 10 daysimpending infarction. of admission. ECG changeswith pain at rest or onexercise stress test.Type II angina: recurringepisodes^of^chest^painresistant^to^nitrates.^Atleast one episode (>15 min)within 10 days of entry.ECG changes during at leastone episode.Association classification (NYHA), in the first year but during the second year thisfinding was only significant for those patients with one-vessel disease; possiblybec I use the patients with more severe angina had crossed over and receivedsurgery which reduced their symptoms. Four-year follow-up on patients with 70percent or greater obstruction of the LAD (Russell et al 1981), showed no significantdifference in incidence of MI or mortality despite a large operative mortality in thesurgical group (9 percent versus 2 percent, non-significant).In a later analysis of a sample of the patients, Russell et al (1980) found that alarger percentage of medical than surgical patients had returned to work. Thisoverall difference was not significant but when this was expressed as a ratio of the59percent employed at time of follow-up to the percent employed 3 months prior toentry into the study, the difference was significant at the 0.01 level.TABLE 5RANDOMIZED TRIALS FOR UNSTABLE ANGINAPATIENT CHARACTERISTICSTrial NIH Trial VA TrialMed Surg Med Surg(N=147) (N=141) (N=237) (N=231)Mean Age 52.7 53 56.3 55.7Previous MI 27 35 42.6 41.7CAD severity1-vessel 24 24 18.6 18.82-vessel 37 33 33.1 36.73-vessel 39 43 48.3 44.5LVEF50+ 62 65 71.3 71.430-49 23 23 28.7 28.615 12 - -LAD 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 1989The Veterans Administration Cooperative Study on Unstable Angina (VATrial) randomized patients between June 1976 and June 1982, from eligible patientsadmitted to one of the participating hospital with chest pain. Inclusion criteria areshown in Table 4; excluded were patients with a diagnosis of acute MI, left maindisease, a LVEF below 30 percent, a history of previous surgery for angina, or thosewho were participants in another study, Of the 3,159 patients screened, 2,433 wereexcluded by virtue of the criteria desceibed above and 468 patients wererandomized. Prior to randomization, patients were stratified according toventricular function and type of angina, and were booked for surgery. Therapy formedical patients was assumed to start on the date that surgery had been scheduled.Patient characteristics are shown in Table 5. The mean interval betweenrandomization and surgery was 9.3 days in the surgical group and 3.6 days in the60medical group22 ; no deaths occurred in this interval. Operative mortality was 4.3in patients with Type I angina and 2.1 in those with Type 2. Eleven patients didnot have surgery and were classified as crossovers to medical therapy. In themedical group five patients died within 30 days of randomization; a mortality rateof 1.6 percent for type I and 4.3 percent for Type II angina. The crossover rate tosurgery in the first 30 days was 19.1 percent for Type II and 5.8 percent for Type Idisease; thereafter the crossover rates were the same. By the 2 year follow-up, thecumulative crossover rate was 34 percent. Operative mortality in patients whocrossed-over to surgery was much higher than for the assigned surgical group butthe perioperative MI rate was the same (Luchi et al 1987).At two years of follow-up (Luchi et al 1987), no treatment advantage wasseen overa1123 but a significant surgical survival benefit was found for the subset ofpatients who had LVEF between 0.3 and 0.59 percent. This benefit was onlysignificant when LVEF was treated as a continuous variable, and so was not foundon analysis using the life-table method where LVEF was entered as a categoricalvariable24 . At five year follow-up the surgical benefit was still present but was nolonger significant. However, significant surgical advantage was seen for othersubgroups at five years. Surgical patients with three-vessel disease hadsignificantly lower mortality than medical patients (11 versus 24 percent); thissurvival advantage was increased for those surgical patients with three-vessel22 No reason was given for this difference in randomization to surgery time between groups. Possiblythe 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 werenoted.24 Presumably the LVEF range for each categorical variable had clinical significance but we are nottold what they were.61disease and abnormal LVF (9 percent mortality versus 29 percent medicalmortality) (Parisi et al 1989).TABLE 6OUTCOMES IN RCTs FOR UNSTABLE ANGINAOutcomesNIH STUDYMed(N=147)Surg(N=141)VA TRIALMed(N=237)Surg(N=231)% % % %Mortality30-day 3 5 2.1 4.12-year 9 10 9.3 7.85-year - - 19 16Non-fatal MI30-day 8 17 4.6 11.72-year 11 13 12.2 11.75-year - - 17.7 15.6Angina**1-year 36 11.3 - -2-year 19 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 ofmortality for medically treated patients. For those with ejection fractions between0.3 and 0.49 survival was significantly better for the surgical group. There was nosignificant treatment difference for LVEF between 0.5 and 0.69, while for thesubgroup with LVEF above 0.7, survival was significantly better in the medicalgroup (Scott et al 1988).Relief of angina was significantly better in the surgical group. At five yearsof follow-up, approximately 33 percent of patients in the medical group were free62supported by the fewer medications used by the surgical group. At one and threeyears the number of patients not taking propanalol was significantly greater in thesurgical group, but this difference was no longer significant at five years (Booth etal 1991).In summary, the findings of the two randomized trials are congruent.There is no significant treatment advantage overall with respect to survival or theincidence of MI, but the short-term relief of angina is significantly better withinitial surgery. In addition the VA trial found long-term survival advantage forsurgical patients with three-vessel disease and those with three-vessel disease andabnormal left ventricular function. The NIH study was terminated after 30 monthsand 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 areinterpreted today. Firstly, in view of the relatively high survival of the medicalgroup in both trials, sample sizes may not have been sufficient to detect adifference between the treatment groups, especially in the smaller sub-group ofType II patients (Parisi et al 1989). In other words, there may have been a treatmentdifference which the study did not have the power to detect. The second group offactors relate to the advances in non-surgical treatment of angina since thesestudies 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, togetherwith beta-blockers and long-acting nitrates have reduced the number of failures inmedical therapy (Gerstenblith et al 1982). Also the use of aspirin for plateletinhibition, the introduction of thrombolysis and the growth in the use of coronaryangioplasty for unstable angina have likely increased the effectiveness of medicaltherapy.63Asymptomatic Patients with "Silent" Myocardial Ischemia:There have been no RCTs which looked at the effectiveness of CABS inasymptomatic patients, although the clinical significance of silent ischemia hasbeen the subject of much debate in the literature. One problem with interpretingthe literature on silent ischemia is that few, if any, authors have studied thephenomenon in completely asymptomatic patients. Most studies look at patientswho have exercise-induced or holter-monitor detected silent myocardial ischemiain addition to angina. Intuitively, it appears likely that completely asymptomaticmyocardial ischemia will have a different prognostic significance than that inpatients in whom angina is also present.It appears that silent myocardial ischemia does have prognostic significancein patients with documented CAD. Weiner et al (1987) analyzed stress test andmortality data from 2,982 patients with proven CAD in the CASS registry. Thosewith silent ischemia (S-T depression but no chest pain) on exercise testing (Group1), had a 7-year survival rate (76 percent) similar to those with angina but noischemia (77 percent) and those with angina and ischemia (78 percent). Patientswith neither angina nor ischemia had a seven-year survival of 88 percent, while acontrol group of patients who had no CAD but who had ischemia without anginahad a survival of 95 percent. Among Group 1 patients, survival was related to theextent of CAD; survival was 86 percent, 73 percent and 57 percent respectively forpatients with one-, two-, and three-vessel disease.Because quality of life, at least as far as symptoms are concerned, cannot bealtered following CABS in completely asymptomatic patients, survival databecomes most important in assessing benefit. Unfortunately there appear to be nosurvival data for completely asymptomatic patients who have not had amyocardial infarction. Patients in this latter category will be discussed later.64Josphson (1990) believes that determining the effect of CABS on silentischemia can provide useful information on the effectiveness of revascularization.He cites several studies which show that CABS reduces, and occasionallyeliminates, silent myocardial ischemia. However, in the absence of controlledsurvival data, one cannot conclude that treatment of silent ischemia byrevascularization provides a survival benefit over patients who are treatedmedically.Post-Myocardial Infarction:Two randomized controlled trials have compared medical and surgicaltreatment in patients who have had at least one M.I. One hundred and sixtyasymptomatic post-MI patients were included in the CASS randomized trial.There was no significant difference seen in survival between treatment groups; atten years 69 percent of the medical group and 81 percent of the surgical group werestill alive. Those alive and free of MI at ten years were 62 and 68 percent of themedical and surgical groups respectively (Alderman et al 1990).Between 1972 and 1979 in New Zealand, Norris et al (1981) randomized 100patients, aged 60 or younger, who had survived two or three myocardial infarctswhich were at least one month apart. Excluded were men with congestive heartfailure, 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 meanfollow-up time of 4.5 years six patients in the surgical group and five in themedical group had died, and all but one of these were cardiac deaths. Thecrossover rate, from medical to surgical treatment, was 14 percent.The results of these two studies indicate that CABS is not more effectivethan medical treatment for asymptomatic MI survivors who are at least 30 days65post-MI. These studies, however, did not distinguish between asymptomaticpatients who have silent myocardial ischemia and those who did not. In a chapterreviewing the literature of silent ischemia in survivors of MI, Deedwania (1988)concludes:Uncontrolled observations indicate ... that the prognostic implications ofsilent and symptomatic episodes in the survivors of acute MI are essentiallythe same. Thus, treatment regimen designed to alter the adverseconsequences of persistent myocardial ischemia should not distinguishbetween symptomatic and asymptomatic ischemia. However, controlledclinical trials will be needed [Italics mine] to determine whether theelimination of persistent asymptomatic myocardial ischemia in thesurvivors of acute myocardial ischemia [sic] will reduce the short- and long-term morbidity and mortality. (p171) .For patients who develop angina after MI the situation is also unclear. Boththe VA trial and the ECSS include patients who had an MI but they did not reporton them as a separate sub-group. However, Chassin and colleagues (1986), indiscussing these trials, state that "it is reasonable to presume that their resultsapply to patients who develop CSA following an MI".Gardner et al (1989) studied 300 patients who received CABS after theydeveloped postinfarction unstable angina. Hospital mortality was five percent,survival was 96 percent and 88 percent at one and five years respectively; there wasno control group. However, the hospital mortality was higher than the twopercent generally accepted for patients with unstable angina (Scott et al 1988). Thismay reflect the fact that CABS was carried out early after the infarction; the meantime from onset of MI to surgery was only 11.7 days.A similar retrospective study carried out by Naunheim et al (1988) on 336patients who underwent surgery within 30 days of MI. Overall operative mortalitywas 7.7 percent but this differed markedly between subgroups. Operative mortalitywas 2.3 percent for those with stable or no angina, 6.1 percent for those with anginaat rest, 9.5 percent for patients with severe angina requiring intra-aortic balloon66counterpulsation and 47.8 percent for those in cardiogenic shock. The authors,somewhat optimistically, conclude that early postinfarction CABS may beperformed on the stable patient at any time with a mortality similar to that ofelective CABS. However, an operative mortality of 2.3 percent is higher than theless than one percent found in other series in the early to mid- eighties (Gersh et al1989).On multivariate analysis, Naunheim and his colleagues found that age, leftventricular function and the hemodynamic state of the patient were significantpredictors of operative mortality. Time between the onset of MI to CABS was apredictor 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 timebetween surgery and the MI; those operated on in the first post-MI week had amortality of 46 percent, compared to 7 percent in the seventh week. However,when patients were stratified according to ventricular function these timedependencies changed. Patients with LVEF at or above 50 percent survived atwhatever time they were revascularized. Those with LVEF below 50 percent hadbetter survival the more remote their surgery was from the infarction.The question of whether CABS is effective in either symptomatic post-infarction patients, asymptomatic patients postinfarction with myocardialischemia, 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 isrequired for all situations.Evolving MI:Revascularization of the myocardium within 4-6 hours of the onset of anMI can prevent irreversible myocardial ischemia and, consequently, infarction.67There is only one RCT comparing medical and surgical treatment of evolving MIreported in the literature. Koshal et al (1988) randomized 68 patients presentingwithin 4 hours of onset of chest pain, to receive CABS or medical treatment thatdid not include thrombolysis. Patients older than 65 years, those with seriouscomorbidity, cardiogenic shock, or prior MI were excluded. Thirty-day mortalitywas 2.9 percent for surgical patients compared to 8.8 percent in the medicallytreated group; 18-month mortality was unchanged for the surgical group but 20.6percent for those medically treated. Although the size of the study was too smallto provide statistical significance of the results, the study demonstrates that promptreperfusion can be carried out with a low operative mortality and encouraginglong-term survival (Ryan 1990).Results from a case-series in Spokane, where reperfusion early in acute MIhas been carried out since 1971, support Koshal's findings. Between 1972 and 1976,surgical patients receiving CABS within 24 hours of peak symptoms experienced a4.4 percent operative mortality. Other studies though have shown less optimisticfindings. An observational study of 83 patients who received CABS duringevolving MI showed an operative mortality of 15.6 percent, compared to mortalityof 13.5 percent of patients in the same institution who were treated bythrombolysis, angioplasty or both. Significant predictors of operative mortalitywere cardiogenic shock, age over 65 years, LVEF less than 0.3 and absence ofcollateral vessels (Athanasuleas et al 1987). Differences in patient populationsdoubtless account for some of the differences between these and Koshal's results.There is no evidence to show the efficacy of CABS after thrombolysisalthough approximately 20 percent of patients suffer a reocclusion within the firstweek of thrombolytic therapy (Ryan 1990). CABS is evidently being used early afterthrombolysis because Naylor and Jagdal (1990) found a significant short-term68increase in revascularization for patients treated with thrombolysis for acute MIcompared to those treated conventionally.Ryan (1990) speculates that it is unlikely that CABS or PTCA will everbecome the primary modes of therapy for MI, given the high cost and the logisticalproblems in getting the patient to surgery within the time before infarction takesplace. He notes that probably 50 percent of the population who undergo acute MIwill be ineligible for thrombolysis because of age or other contraindications, and,therefore, it will be imperative to clearly establish the role of primaryrevascularization procedures in the treatment of acute MI.In summary, there is evidence that CABS can reduce mortality if performedduring an evolving initial MI on patients under 65 who are not in cardiogenicshock, although this evidence comes from an RCT which did not have sufficientpower to measure a clearly significant difference and which excluded the highestrisk patients and those over 65. Further research is required to determine whichpatients will benefit, and when, from revascularization during MI.Emergency CABS after Failed PTCA:A number of observational studies have looked at outcomes of emergencyCABS for failed coronary angioplasty. The incidence of failed PTCA requiringsurgical revascularization ranges from 0.9 to 21 percent (Kahn et al 1990). Failureappears to be related to the year in which the procedure was performed and to thenumber of procedures performed in the institution (Parsonnet et al 1988).Outcomes for these patients appear to be worse than for those patients receivingelective CABS. Operative mortality ranges from 1.4 percent to 15 percent,perioperative MI from 20 percent to 51 percent. Parsonnet et al (1988) found acomplication rate that was significantly higher than for patients undergoingelective CABS (61 percent versus 45 percent). Major complications arose in 4569percent of the emergency group. These authors conclude that in view of thehigher mortality and morbidity rates, increased length of stay (15.3 versus 13.4days) and consequent increased use of hospital resources, emergency CABSfollowing failed PTCA cannot be equated with elective CABS.Re-operations:There are three periods during which reoperation may be considerednecessary. The first is early, during the first or second year, and is generally due toa technical fault or to compromised blood flow which eventually leads to graftocclusion. The latter problem may be averted through the use of plateletinhibitors. The second period occurs four or five years post-operatively and isgenerally due to progression of disease in ungrafted arteries or to incompleterevascularization at surgery. Late revascularization, which occurs eight to twelveyears postoperatively, is generally performed because of graft disease which is oftenassociated with disease in ungrafted vessels. Nearly 70 percent of reoperationsoccur in this later time period which is the more hazardous because of the greaterage and worse condition of the patient, anatomical changes resulting from the firstoperation and the presence of atherosclerotic grafts which have the potential toembolize during the operation (Grondin et al 1990).The incidence of reoperation is about 3 percent during the firstpostoperative year (Foster et al 1984) rising to about 17 percent by twelve years(Cosgrove et al 1986). Factors which have been identified as predictors ofreoperation are total cholesterol level, smoking, hypertension, incompleterevascularization and lack of an internal mammary artery graft (Solymoss, B.C. etal 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 that70the risk of death was highest on the operative day; in this series the incidence ofdeath for subsequent days was not significantly higher than that for the primaryprocedures. The incidence of perioperative MI is also higher than at initialoperation and Foster and colleagues found that the complications of perioperativeMI, while being no more frequent than those in the primary procedure, were morefrequently fatal.There have been no RCTs which have looked at the effectiveness of CABSversus medical treatment for second or subsequent reoperations. In fact there donot appear to be any studies of reoperation that used control groups. Evidence oflong-term effectiveness of reoperation must, therefore, come from uncontrolledseries.Table 7 shows that long term survival following reoperation is better thanthat reported for primary operations in the RCTs. However, these studies werecarried out later in time than most of the RCTs, and more than 50 percent ofpatients in the Cleveland study received IMA. While it appears that reoperationhas the potential for long-term benefit the lack of comparative data to show howpatients receiving medical treatment rather than reoperation fared in theseinstitutions over the same time period means that such a conclusion is onlytentative at best.The Elderly:This literature review has thus far looked at the effectiveness of CABS invarious clinical conditions. It is, however, important to note that the populationsstudied in the RCTs were mainly males under the age of 65 years 25; consequentlythe efficacy and effectiveness of CABS in females and in older patients is not25 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.71known. In fact, older age was found to be a significant predictor of mortality inalmost all studies which used multivariate regression to identify predictors ofmortality or morbidity post-CABS, although the relative importance of age differedbetween studies.TABLE 7PERCENTAGE OF PATIENTS SURVIVING FIVE OR MORE YEARS AFTERRE-OPERATIONResults^Mayo Clinic*^Cleveland Clinic **5 year^10 year^7 year^10 yearSurvival^94^89 90 7563***^63*"Event free 76^48Asymptomatic^28^-^52Years^ 1969-1980^ 1967-1984Adapted from Grondin et al (1990)*N=160 **N=1,500 ***Excludes Class I-II anginaThere are several reasons why the over-65's may have different outcomesthan younger patients. Older people with CAD generally have poorer leftventricular function, more extensive coronary artery disease, a higher incidence ofunstable angina, and more associated medical diseases, all factors which have beenassociated with higher mortality and morbidity (Gersh et al 1983). In addition,older people are physiologically less able than younger patients to respond to thestress 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 cannotnecessarily be applied to those over-65. The efficacy of CABS in people over-65 canonly be established by a RCT which draws its sample from this population. To datethere are no RCTs which do this.72In the absence of data from RCTs, information on the outcomes of CABS inthe elderly must come from less rigorous studies. The only controlled studycomparing surgical and medical therapy in CAD patients 65-years of age or older,analyzed data from the CASS Registry of patients receiving angiography between1974 and 1979 (Gersh et al 1985). Excluded from this analysis were patients with leftmain disease, those who did not have angina as the primary symptom and thosewith less than 70 percent stenosis of a major vessel. Patients who received CABSwere assigned to the surgical group if they received the procedure within the timethat 95 percent of all patients, in the first year of the registry, undergoing CABS inthe same institution received their surgery. All other patients were in the medicalgroup which included more women and significantly more patients withassociated medical diseases and impaired LVF.On long-term follow-up the cumulative unadjusted six-year survival wassignificantly better in the surgical group (80 percent compared to 63 percent in themedical group), despite an operative mortality of 4.3 percent. When survivalcurves were adjusted for the major prognostic variables, survival was still better inthe surgical group (79 percent versus 64 percent). Unadjusted six-year survivalrates for age subgroups are shown in table 8.TABLE 8UNADJUSTED CUMULATIVE 6-YEAR SURVIVAL IN AGE SUBGROUPS OF THE OVER-65'SAge MedicalN^% survivalSurgicalN^% survivalp65-69 459 67 625 81 0.000170-74 129 51 200 77 0.000175+ 42 56 36 75 0.1473Analysis of functional status showed that symptoms in both medical andsurgical groups at five years were improved over baseline. In part this may reflecta relatively higher death rate in patients with more severe symptoms, but it alsolikely indicates improvement in angina over time.Gersh et al (1983) had earlier reported on a series of patients over the age of65 from the CASS registry. This study compared outcomes in 1,086 patients over-65 to 7,827 CASS registry patients under-65. As can be seen in table 9, bothoperative mortality and long-term survival were significantly worse in the over-65patients and mortality increased with increasing age. The presence of associatedcomorbid conditions had a significant effect on 5-year survival in the older age-group, as did the presence of left ventricular dysfunction, and a pre-operativehistory of hypertension. These variables were not reported for the under-65s.Surprisingly, event-free survival was found to be higher in the older group butwas significantly poorer in women who had less symptomatic relief than men.The data indicated that recurrence of angina was lower in older patients but Gershand colleagues suggest that this could be due to lower levels of activity in olderpatients.Taken together these studies indicate that although CABS patients over-65have 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 treatedsurgically than in those on medical treatment. Increased risk for poorer long-termoutcome is seen in those with associated co-morbid diseases, impaired LVF, ahistory of hypertension and in older age-groups. However, it should beremembered that these patients were not randomized to treatment and soselection bias will preclude generalization of these results to all over-65s withCAD.74TABLE 9OUTCOMES OF CABS IN CASS REGISTRY PATIENTSOutcomes Under-65(N=7827)Over-65(N=1086)% %Operative Mortality 1.9 5.2Age 65-69 (N=803) 4.6Age 70-74 (N=241) 6.6Age 75+ (N=42) - 9.55-year survival 91 83Age 65-69 - 84Age 70-74 - 80Age 75+ 70LVEF -<35* - 4736-50 7951+ 88Comorbid conditionsNone (N=489) - 89One - 80Two or more - 71Event-free 39 47Cause of deathCardiac related 69 53Non-cardiac, CAD rel 5 14Other non-cardiac 26 33Data from Gersh et al 1983*4 percent of those testedThere are numerous observational studies in the literature which describeoutcomes of CABS in various age subgroups of the elderly. Those looking atseptuagenarians 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 yearsold 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 4percent in those with isolated coronary disease compared to 13.8 percent in thosewith associated diseases, Goldman et al (1987) report on a prospective75observational study of 3,327 patients undergoing CABS at the Toronto GeneralHospital between 1982 and 1986. Over the four years of the study patients over 70years of age increased from 4.5 to 14.2 percent of total bypass patients. Significantlymore of the 340 patients over the age of 70 had high risk characteristics than didthose under 70. One third of the older group had urgent surgery compared to 17percent of those under 70. Clearly the older patients as a group were sicker thanthe younger ones and this is reflected in their poorer outcomes; an operativemortality of 6.1 compared to 2.9 percent, a higher incidence of stroke (4.4 versus 1.3percent), and post-operative low-output state. Operative mortality was seen toincrease with increasing age over the whole population, rising from one percent inthose 30 to 39 years old to 7.3 percent in those aged 75 to 80 years. There were nooperative deaths in the six patients aged over 80.In the above study, Goldman and colleagues also retrospectively reviewed arandom sample of patients from the original cohort and found that the patientsover the age of 70 required longer stays in the intensive care unit, longerventilatory support and longer overall hospitalization (18.9 ± 11 versus 15.5 ± 10days). Edmunds et al (1988), looking at open-heart surgery in octogenarians, foundthat 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 thosewithout.These observational studies confirm the CASS registry findings that elderlypatients are at higher risk of operative death and morbidity. Differences inoperative mortality between studies likely reflect the era in which surgery wasperformed, patient selection or both. In addition these studies confirm that olderpatients undergoing CABS use more hospital resources than younger ones; anunsurprising finding given the higher morbidity in these patients.76Several authors comment that elderly patients should be investigatedsooner in order to allow revascularization before their condition deteriorates tothe 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 theelderly, the fact that efficacy is not proven means that caution in subjectingpatients 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 tosmaller body size (Loop et al 1983) and to smaller cardiac size leading to moredifficulties during anastomosis and earlier closure of vessels (Fisher et al 1982).Tobin et al (1987) found that, following a positive nuclear exercise test, men wereten times more likely than women to have cardiac catheterization and suggestedthat this disparity may be caused by a sex bias.Kahn et al (1990) were led by Tobin's work to investigate whetherdifferences in the pre-operative status of women could account for their higheroperative mortality. They investigated 2,297 consecutive patients (1,815 men and482 women) undergoing isolated CABS at one institution after 1982. They foundthat women were significantly older than the men, more often had a history ofhypertension and diabetes mellitus, and more frequently had unstable angina,post-MI angina and pre-operative symptoms of heart failure. There was a trend forwomen to be referred in cardiogenic shock more frequently but this was notsignificant. Women were referred significantly more often than men with severesymptoms of coronary artery disease (heart failure, unstable angina, post-MIangina, cardiac arrest or cardiogenic shock) while men were referred for surgerysignificantly more frequently with an abnormal exercise test.77Operative mortality was 2.6 percent for men and 4.6 percent for women.Operative mortality differed significantly by referral symptom; patients referred foreither previous MI or post-MI angina had significantly higher rates while thosereferred 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 ofwomen was explained entirely by their greater age and higher NYHA functionalclass. The authors conclude that women are referred for bypass surgery later intheir disease and for different reasons than men, and this later referral maysignificantly increase their chance for operative death.Despite a higher operative mortality in women, many studies have shownno 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). Unfortunatelythe number of women in CASS, the only randomized trial to include women, wastoo few to draw meaningful conclusions about the effectiveness of CABS in thissex. There is, however, evidence from less rigorous studies that they have lesssymptom relief after surgery than men (Bolooki et al 1975, Douglas et al 1981). Itwould be useful to know whether men and women treated medically for CABSalso show this difference, but such data do not appear to be available.Overall the evidence indicates that while male and female surgicalsurvivors show no difference in long-term survival, women face an increasedoperative risk and receive less symptomatic relief than men. There is, however,no evidence to show that CABS is more effective than medical treatment inwomen with CAD.78RISKSOperative Mortality:The major risk for patients undergoing CABS is operative mortality. As hasbeen shown above, operative mortality differs among different conditions andacross time. Chassin et al (1986) used meta-analysis to calculate average mortalityrates 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 anoperative 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 largeregistries but these rates are considerably lower than those reported in manyobservational studies. It is clear, however, that operative mortality has decreasedsubstantially over the years despite the worse pre-operative condition of present-day patients coming to surgery.Patient selection likely accounts for much of the difference seen in mortalityrates for the same era. A number of clinical, anatomic, and physiologic patientconditions and operative factors have been shown to be risk factors for operativemortality, although only a few factors have consistently been demonstrated asimportant predictors. Moreover, the importance of some predictors appears tohave changed over time. Christakis et al (1989) found that between 1982 and 1986,urgency of surgery, age and reoperation became more significant predictors ofmortality with respect to time, while female sex, LVEF and left main diseasebecame less significant. Cosgrove et al, in an analysis of CABS at the ClevelandClinic between 1970 and 1982, found that over time, incomplete revascularizationemerged as a new factor, and impaired LVF became less significant.Operative mortality rates for factors predictive of operative mortality at theCleveland Clinic and Toronto General Hospital are shown in table 10. In readingthis table it is important to compare rates, with and without factors, within79institutions only. The figures are not comparable between institutions because ofdifferences in defining the factors and different methods of calculating the rates.Parsonnet, Dean and Bernstein (1989) devised an additive model of 14 riskfactors to stratify patients into levels of predicted operative mortality followingCABS. They included patients receiving other procedures along with CABS so theresults 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 (allwith weight 10). The correlation coefficient of anticipated and observed operativemortality was 0.99.TABLE 10PREDICTORS OF OPERATIVE MORTALITYPredictiveFactorCleveland1980-82*Toronto1982-86**Operative Mortality Operative MortalityWith factor Without With factor WithoutLVEF <20 1.6 0.7 11.7 2.5Age >70 2.2 0.9 7.4 1.8Urgent/Emerg. 4.9 0.7 8.0 1.9SurgeryFemale sex 1.9 0.6 6.0 3.0Reoperation - - 9.0 3.2Left main CAD 0.8 0.8 5.8 3.2Incompleterevasculariz.1.3 0.6 -CHF 7.3 0.7 - -IMA graft - - 4.0 2.8* Cosgrove, Loop and Sheldon (1982)**Christakis et al (1989)80Operative mortality rates have also been shown to be related to theinstitution where the surgery is performed. Data from the CASS registry showed a22-fold difference in operative mortality between institutions (range 0.3 to 6.6percent). After controlling for risk factors for operative mortality, this increased toa 31.5 fold difference between the institutions with the highest and lowest rates.Other investigators have found that hospital mortality is lower in hospitalswhich perform more than 200 CABS procedures per year (Luft, Bunker andEnthoven 1979). Showstack et al (1987) found that in California in 1983, highervolume hospitals (more than 100 procedures per year) had a lower in-hospitalmortality (3.5 vs 5.3 percent in low-volume hospitals) adjusted for case mix. Thisassociation of volume with mortality was seen to the greatest extent in patientshaving 'non-scheduled' CABS. Higher-volume hospitals also had shorter LOS andfewer patients with stays over 15 days. The authors conclude that the greatestimprovement in average outcomes of CABS would come from closure of low-volume surgical units.A recent study from New York state, found that physician volume was alsoa factor (Hannan et al 1989). Physicians performing more than 116 procedures peryear had risk adjusted hospital mortality rates 22 percent lower than surgeons withlower volumes. Low-volume surgeons in high-volume hospitals had loweradjusted mortality rates than high-volume surgeons in low volume hospitals 26 .These results do not indicate whether high-volume surgeons have a lowermortality due to greater experience with the procedure or whether low-mortalitysurgeons attract more referrals.The above studies obtained their data from discharge abstracts and,therefore, measured hospital mortality rather than 30-day "operative" mortality.26 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.81Also the measures of risk would not be as accurate as risk factors measuredprospectively or taken retrospectively from the chart. Even so, the fact that severalstudies have found an association between institutional volume for CABS andmortality rate, lends credence to the result.Morbidity:Other potentially serious, and not uncommon, complications of CABS areperioperative MI, stroke, and neuro-psychological complications. Kirklin et al(1991) indicate that perioperative MI in patients with chronic stable angina hasdecreased over the years from about five to eight percent in the 1970's to about 2.5percent now, and they attribute this change to improved methods of myocardialmanagement. The rates appear to be higher in unstable angina, ranging from 3.8to 17 percent (Kaiser et al 1989). Perioperative MI appears to be associated withoperative 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 70compared to 4.4 percent for those over 70. Mullany (1990), reporting on a series ofthose aged 80 or over, found a 2.5 percent rate of perioperative stroke. Neitherstudy defined their interpretation of stroke. Kirklin et al (1990) report that theprevalence is about 0.5 percent in relatively young patients, and about 5 and 8percent in patients over 70 and 75 years of age respectively.Subtle neurological defects have been reported in up to 75 percent ofpatients at 8 days post-CABS but only 10 to 30 percent of patients still exhibit themat 3 months. These defects may not be apparent unless the patient is specificallytested for them, and most patients are not handicapped by them (Kirklin et al1991).82The overall incidence of perioperative morbidity after CABS is, again, notoften reported. Parsonnet et al (1988) found a 45 percent morbidity rate for patientshaving elective CABS between 1980 and 1986. Christakis et al (1989) found anoverall morbidity of 10.9 percent in his series of 7,334 patients. Morbidity washigher for those undergoing urgent surgery for unstable angina (19.3 percentversus 6.9 percent), for those with LVEF under 20 percent (20.4 percent), those over60 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 onthe patient, it is clear that CABS carries a significant risk of long-term disabilityover and above the risk of death. Also, given the relatively large numbers ofpatients who are receiving the procedure, the effect of even a 10 percent morbidityrate could have a major impact on cardiac surgery centres in terms of costs andproductivity. The demand for rehabilitation services and home support services,among others, will also be increased by post-CABS patients who suffercomplications.Other Risks:The risks of waiting for cardiac surgery were determined by Rachlis, Olakand Naylor (1991) using outcomes from the major RCTs. In most subgroups themaximum vital risk was less than 0.33 percent per month of delay; a risk aboutone-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 amaximum of 1.05 percent per month. Rachlis and colleagues notes that whileolder age may be associated with greater risk for medical care, this must be weighedagainst the markedly increased mortality from CABS.The risk of not undergoing surgery after being advised to do so, does notappear to have been studied as such. However, Graboys et al (1987), describe the83outcomes for 91 consecutive patients referred for a second opinion followingcardiologists' recommendation for CABS. Following application of guidelines fordeciding on a medical or surgical option, medical therapy was recommended for 74patients. Of the 60 patients who did not have surgery, none died during a meanfollow-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 riskfactor modification, an exercise program and attention to psychological issues.It is difficult to tell from this uncontrolled study whether the outcomes werea result of the aggressive treatment provided. However, Graboys estimated that,according to the indications for surgery that derive from the RCTs on CABS, onlynine percent of the referred patients actually required surgery. Conservativemedical management was as appropriate, or more appropriate, than surgery for 90percent of these patients who were told by their original cardiologists that theyrequired surgery.Medically treated CAD patients undergoing non-cardiac operation appear tohave a higher risk of operative mortality than those who have had prior CABS. ACASS Registry study found operative mortality for non-cardiac major surgery,between 1978 and 1981, in patients without significant CAD to be 0.5 percent; inthose with significant CAD who had undergone prior CABS, mortality was 0.9percent. In CAD patients who had not received CABS, mortality was 2.4 percent.The authors recommend that patients with significant CAD receive prophylacticCABS before major non-cardiac surgery. However, they appear to be ignoring thefact that the operative mortality for CABS was itself around 2-3 percent during theyears under study. Consequently patients undergoing CABS purely forprophylactic reasons prior to other surgery would likely be at increased risk ofoverall mortality rather than less.84FIGURE 1OUTCOME FOR PATIENTS REFERRED FOR SECOND OPINIONPatients Referred88ITreatment RecommendationMedicine^ Surgery74 14Medicine^Surgery^Medicine^SurgeryI^I I I60 14^3^11/ \ / \ / \^/ \Died MI^Died MI^Died MI^Died MII^I I^I ^I I^I0^2^2^0^1^2^1^2Data from Graboys et al 1987.COSTSA number of Canadian studies have examined the costs of CABS. The mostrecent (Krueger, Goncalves, Caruth and Hayden 1991) examined costs for triple orquadruple bypass surgery at Vancouver General Hospital in January 1989. Themean inpatient cost, including professional fees and fully allocated hospital costs(but not depreciation costs) was $14,329 (range $10,982-$33,676). Earlier studies byKeon, 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 theVancouver study (55.3 and 58 years compared to 63.4 years) and fewer grafted85vessels. Both Krueger and Keon and colleagues found that the age of the patientand 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 nothingabout 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 1977found that, based on the reduction in hospitalization expenses in the yearsfollowing surgery, the period for amortization of CABS expenses was 22.3 monthsfor those with left ventricular dysfunction, 38.4 months with normal LVF and 35.9months overall. The study likely underestimated the costs of CABS because theOntario per diem rate was used to calculate costs; this proxy would not capture thehigh operating room charges for CABS, which Krueger et al found to be 33 percentof the total cost. Moreover, Coles and Coles sample was relatively young (mean52.3 years, range 27-72 years). Given that current CABS patients are older and sickerthe 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 malesused data from the literature on costs and outcomes with medical and surgicaltreatment (Weinstein and Stason 1982). The costs (in 1982 dollars) per qualityadjusted life year (QALY) are shown below. It appears that CABS is excellent valuefor left main disease and three-vessel disease even with mild to moderatesymptoms, and a reasonable value for two-vessel disease with severe anginauncontrolled by medical treatment. The costs in 1991 dollars would be muchhigher than these, as would the costs for older or sicker patients.86TABLE 11COST EFFECTIVENESS OF CABS AS A FUNCTION OF THE SEVERITY OF ANGINAExtent of disease Cost Per Quality Adjusted Life Year1 vessel2 vessel3 vesselLeft Main DiseaseMild Angina Severe Angina$470,000$47,000$$7,500$3,500$30,000$17,500$7,200$3,800Data from Weinstein and Stason 1982.CONCLUSIONWhen compared to medical therapy, CABS has been shown to be effectivein prolonging survival in male patients under 65 years of age with CSA and leftventricular dysfunction, in left main disease, and in patients with proximalstenosis of the LAD. There is some evidence to show that it is also more effectivethan medical treatment in three vessel disease, especially in the presence of severeangina. In addition CABS unquestionably provides greater symptomatic relief thanmedical therapy, though this difference decreases over time.In patients with unstable angina a long-term survival advantage has beenshown with CABS for male patients under 65 years of age with three-vessel diseaseand left ventricular dysfunction. In addition the short-term relief of angina isbetter with surgery. In general, there is no advantage from immediate surgery butpatients should be monitored and surgery provided if their condition deteriorates.The efficacy of CABS has not been proven in any other clinical condition  althoughthere 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 clinicalconditions, it is disturbing to find in the literature that the procedure is widelyused for conditions in which its efficacy is not known. Many observational studiesappear to take the efficacy of CABS in their patient population as a given, eventhough these populations may not have been tested in a randomized trial or atleast in a controlled trial. Many of these studies involve large numbers of patients;87it is, therefore, likely that many institutions would have sufficient patientpopulations to enable them to carry out a RCT. Observational studies, whileproviding information on a wider patient population than found in a RCT,essentially provide outcome information on an uncontrolled, possibly highlyselected group of patients and, therefore, cannot show efficacy and their resultscannot be generalized.The increase in CABS among the elderly is of especial concern, both becauseof the numbers involved and because of the higher risks run by the elderly.Operative mortality and morbidity are significantly higher in the over-65population with increasing risk associated with increasing age. The finding fromthe RCTs that those at higher risk receive greater comparative benefit has not beentested in the over-65 population and thus cannot be assumed to apply to them. Inaddition the finding from several studies that the proportion of patients withsevere angina symptoms decreases over time, raises questions about whetherCABS may be appropriate treatment for symptom control in the elderly. Finally,the use of CABS in the elderly, especially in the over eighties, raises questionsabout the appropriate use of scarce resources.Sadly lacking in the literature is mention of the effect of CAD risk factors onthe outcomes of treatment. Compliance with smoking cessation, dietaryrestrictions and exercise regimes have been relatively low in those studies thatreport them, but there is little information on whether compliance leads to a betteroutcome. This lack of information is disturbing because it appears that emphasis isnot being given to risk factor reduction, and life-style changes, in cardiac surgeryprogrammes. CAD is, after all, a life-style related disease and CABS is simply apalliative procedure. It is not a cure; the disease still exists, progresses, and mayeven be exacerbated by the very procedure that is designed to prolong life andreduce symptoms. Intuitively, it seems likely that risk factor reduction, especially88through smoking cessation, weight reduction and exercise regimen, may reducethe 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 to11 percent in the early eighties to about 17 percent today27. Very little seems to beknown about these patients and how they compare to patients who continue tofunction well post-operatively. In addition, little is known about the outcome ofpatients who are considered for reoperation and turned down. What the literaturedoes tell us is that the risks of reoperation are considerably higher than those forprimary CABS, and it is likely that the costs are higher too. Even if the incidenceof reoperation should remain the same, the numbers will increase because of theincrease in the overall incidence of CABS. It appears likely that the incidence ofCABS will continue to rise as more patients return for second or subsequentreoperation.In a 1983 editorial, Braunwald predicted that in the future:"this operation [CABS] should and increasingly will be restricted to patientsin whom intensive medical therapy has failed or in whom improvedsurvival after surgery has been unambiguously demonstrated, rather thanas a panacea for coronary artery disease"Braunwald evidently had faith that his fellow physicians would be guidedby the scientific process, rather than by the results from non-randomized studiesand their own intuitive beliefs about the efficacy of CABS. It may well be that thesebeliefs are well-founded and that CABS does indeed positively influence survivalin all the conditions for which it is presently being performed. However, in theabsence of unambiguous, or at least reasonable, evidence to the contrary, thepossibility that CABS may have no effect on, or even decrease, survival and27 The reasons for this do not appear to have been investigated. It may be a result of surgeonsbecoming more comfortable in performing re-operations.89increase morbidity to the extent that it negatively effects quality of life, must also beconsidered. In fact, if the principle of "Do no harm" is to be upheld, the possibilitythat CABS may contribute to a negative outcome in certain subsets of patientsshould be considered equally as likely as the present apparent belief that it is the"panacea of coronary artery disease".90REFERENCESAlderman, E.L. et al, 1990. Ten-year follow-up of survival and myocardialinfarction 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 acutemyocardial infarction. T Thorac Cardiovasc Surg  93:405-414.Berry, R.E. et al, 1981. Coronary artery bypass operation in septuagenarians. AnnThorac Surg 31 (April):310-313.Bolooki, H. et al, 1975. Results of direct coronary artery surgery in women. IThorac Cardiovasc Surg 69:271-277.Bonow, Robert 0. et al, 1984. Exercise-induced ischemia in mildly symptomaticpatients with coronary-artery disease and preserved left ventricularfunction. 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. Circulation83 (January):87-95.Booth, D.C. et al, 1991. Quality of life after bypass surgery for unstable angina: 5-yearfollow-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 TMed 309 (November):1181-1184.Califf, R.M. et al, 1989. The evolution of medical and surgical therapy for coronaryartery disease: A 15-year perspective. TAMA 261:2077-2086.CASS Principle Investigators and their Associates, 1983. Coronary Artery SurgeryStudy (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, Lungand 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 bypass91surgery. 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 bypasssurgery. Circulation 80 (September) Supplement I: I-151-1-161.Cohn, Peter F., 1990. Prevalence and significance of asymptomatic coronary arterydisease. 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 myocardialrevascularization. Can J Surg 25 (March):123-6.Cosgrove D., F. Loop and W. Sheldon, 1982. Results of myocardialrevascularization: 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 myocardialrevascularization. T Thorac Cardiovasc Surg 92:811-821.Cromwell, Jerry, Janet B. Mitchell, and William B. Stason, 1990. Learning by doingin CABG surgery. Medical Care 28 (January):6-18.Deedwania, P.C., 1988. "Transient myocardial ischemia in survivors of acutemyocardial infarction." In B.N. Singh, ed Silent Myocardial Iscemia andAngina. New York: Pergamon Press:157-173.DeWood, M.A. et al, 1983. Acute myocardial infarction: A decade of experiencewith 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 ofwomen and men in the Coronary Artery Surgery Study (CASS). Am HeartL 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 instable angina pectoris: survival at two years. Lancet (April):879-893.92European Coronary Surgery Study Group, 1980. Prospective randomised study ofcoronary 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 surgeryamong whites and other racial groups relative to hospital-based incidencerates for coronary artery disease: Findings from NEDS. AJPH 79 (April):437-439.Foster, Eric D. et al, 1984. Comparison of operative mortality and morbidity forinitial and repeat coronary artery bypass grafting:The Coronary ArterySurgery Study (CASS) Registry experience. Ann Thorac Surg 38 (December):563-570.Gardner, Timothy J., 1989. The risk of coronary bypass surgery for patients withpostinfarction angina. Circulation  79 Supplement 1:1-79--I80.Gersh, B.J. et al, 1983. Long-term (5 year) results of coronary bypass surgery inpatients 65 years old or older: a report from the Coronary Artery SurgeryStudy. Circulation 68 (September)(suppl 11):11-190--II-199.Gersh, B.J. et al, 1985. Comparison of coronary bypass surgery and medical therapyin patients 65 years of age or older: a non-randomized study from theCoronary 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. Circulation79 Suppl I:I-46--I-59.Gerstenblith, G. et al, 1982. Nifedipine in unstable angina: a double-blindrandomized trial. N Engl T Med 306:885-889.Gillum, Richard F., 1987. Coronary artery bypass surgery and coronary angiographyin the United States, 1979-1983. American Heart Journal  113 (May):1255-1260Goldman, 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 arterybypass graft surgery. TAMA 258 (September):1611-1614.Grondin, Claude M. et al, 1990. Reoperations for coronary artery disease. InAdvances in Cardiac Surgery,  ed Robert B. Karp et al, Chicago: Year BookMedical Publishers, Inc: 93-110.93Hammermeister, K.E., T.A. DeRouen and H.T.Dodge, 1980. Effect of coronarysurgery 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 andmortality 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 incoronary artery disease. Ann Intern Med 106:793-800.Hochberg, M.S. et al, 1984. Timing of coronary revascularization after acutemyocardial infarction. T Thorac Cardiovasc Surg  88:914-921.Huddleston, C.B. et al, 1986. Internal mammary artery grafts: technical factorsinfluencing patency. Ann Thorac Surg 42:543-549.Josephson, Martin A., 1990. Percutaneous transluminal angioplasty andmyocardial revascularization in the reduction of silent myocardial ischemiain 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 bypasssurgery: evidence for referral bias. Ann Intern Med 112:561-7.Kannel, W.B., 1990. Coronary Artery Surgery Study revisited: limitation of theintent-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 coronaryartery bypass graft surgery. Circulation 83 (March):1125-1173.Koshal, A. et al, 1988. Urgent surgical reperfusion in acute evolving myocardialinfarction. A randomized controlled study. Circulation 78 (suppl I):1-171--1178.Kroncke, George M. et al., 1988. Five-year changes in coronary arteries of medicaland surgical patients of the Veterans Administration randomized study ofbypass surgery. Circulation 78 (September) Supplement 1:1-144--I-150.94Krueger, H., J.L. Goncalves, F.M. Caruth and R.I. Hayden, 1991. Coronary arterybypass grafting surgery: How much does it cost? Can T Med In press.Laupacis, A. et al, 1990. The cost-effectiveness of routine post-myocardial infarctionexercise stress testing. Can J Cardiol 6(4):157-163.Loop, Floyd D. et al., 1986. Influence of the internal-mammary-artery graft on 10-year 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 forunstable angina pectoris: Results of a Veterans Administration CooperativeStudy. 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. NewYork: Plenum Publishing Corporation.Mullany, C.J. et al, 1990. Early and late results after isolated coronary bypass surgeryin 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 reportof survival data. N Engl T Med 297:621-627.Naunheim, Keith S. et al., 1988. The changing profile of the patient undergoingcoronary artery bypass surgery. JACC 11 (March) 494- 498.Naunheim, Keith S. et al, 1988. Coronary bypass for recent infarction:Predictors ofmortality. Circulation 78 Supplement I:1-1222--I-128.Naunheim, K.S. et al, 1988. Coronary artery bypass for recent infarction: predictorsof mortality. Circulation 78 Suppl I:1-122--I-128.Naylor, C.D. and S.B. Jaglal, 1990 . Impact of intravenous thrombolysis on short-term 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 forasymptomatic patients with advanced coronary disease. Circulation 63(April):785-92.95Parisi, Alfred F. et al, 1989. Medical compared with surgical management ofunstable angina: 5-year mortality and morbidity in the VeteransAdministration Study. Circulation 80 (November):1176-1189.Parsonnet, Victor, David Dean and Alan Bernstein, 1989. A method of uniformstratification of risk for evaluating the results of surgery in acquired adultheart disease. Circulation  79 (June) Supplement I:1-3--I-12.Peters, Steve et al., 1990. Coronary artery bypass surgery in Canada. Health Reports2(1):9-26.Preston, Thomas A., 1989. Assessment of coronary artery bypass surgery andpercutaneous transluminal angioplasty. Intl J of Technology Assessment inHealth Care 5:431-442.Rachlis, M.M., J. Olak, and C.D. Naylor, 1991. The risk of waiting for coronaryrevascularization. Working paper of the Sunnybrook Clinical EpidemiologyUnit.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 ofManitoba, Faculty of Management.Russell, Richard et al, 1980. Surgical versus medical therapy for treatment ofunstable 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 StudyGroup to compare surgical and medical therapy: II. In-hospital experienceand initial follow-up results in patients with one, two and three vesseldisease. Am T Cardiol 42 (November):839-848.Russell, R.O. et al, 1976. Unstable angina pectoris: National Cooperative StudyGroup to compare surgical and medical therapy: I. Report of protocol andpatient population. Am T Cardiol 37 (May):896-902.Russell, R.O. et al, 1981. Unstable angina pectoris: National Cooperative StudyGroup to compare medical and surgical therapy. IV. Results in patients withleft anterior descending coronary artery disease. Am T Cardiol 48(September):517-24.Ryan, T., 1990. Revascularization for acute myocardial infarction: Strategies inneed of revision. Circulation 82 (September) Suppl 11:11-110--II-116.96Scott, Stewart M. et al, 1988. Veterans Administration Cooperative Study fortreatment of patients with unstable angina: results for patients withabnormal left ventricular function. Circulation 78 (September) SupplementI:I-113--I-121.Showstack, J.A. et al, 1987. Association of volume with outcome of coronary arterybypass graft surgery: scheduled vs nonscheduled operations. TAMA 257(February):785-789.Solymoss, B.C. et al, 1988. Late thrombosis of saphenous vein coronary bypassrelated to risk factors. Circulation 78(supplement 2):140-143.Takero, T. et al, 1982. The Veterans Administration Cooperative Study of stableangina: 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 surgicalmanagement of angina pectoris". In Coronary Surgery Bypass: The LateResults ed Hammermeister, K.E., New York:Praeger Publishers:17-38.Tobin, J.N. et al, 1987. Sex bias in considering coronary bypass surgery. Ann InternMed 107:19-25.Tyras, D.H. et al, 1978. Myocardial revascularization in women. Ann Thorac Surg25:449-453.Varnauskas, E. and the European Coronary Surgery Study Group, 1988. Twelveyear follow-up of survival in the randomized European Coronary SurgeryStudy. N Engl T Med 319 (August):332-337.Veterans Administration Cooperative Group for the Study of Surgery for CoronaryArterial Occlusive Disease, 1977. Veterans Administration CooperativeStudy of Surgery for coronary arterial occlusive disease: III. Methods andbaseline characteristics, including experience with medical treatment. Am JCardiol 40 (August):212-224Veterans Administration Coronary Artery Bypass Surgery Cooperative StudyGroup, 1984. Eleven year survival in the Veterans Administrationrandomized 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 riskclassification and the response to coronary artery bypass surgery in three-vessel coronary artery disease: A report from the Coronary Artery SurgeryStudy (CASS) Registry. Am J Cardiol 60:262-266.Weinstein, M.C. and W.B.Stason, 1982. Cost-effectiveness of coronary artery bypasssurgery. Circulation 66 Suppl97Yeaton, W.H. and P.M. Wortman, 1985. The evaluation of coronary artery bypasssurgery using data synthesis techniques. Intl I of Technology Assessment inHealth Care  1:125-140.98CHAPTER TWOCORONARY REVASCULARIZATION TECHNIQUESPART IIPERCUTANEOUS TRANSLUMINAL CORONARY ANGIOPLASTYHISTORYPercutaneous transluminal coronary angioplasty (PTCA) was firstperformed in 1977 by Andreas Gruntzig who had built on the work of Dotter,Judkins and others who pioneered the recanalization of obstructed arteries bypassing catheters across the obstructions. Data on trends in the dissemination ofPTCA are difficult to find but data from the National Heart, Lung and BloodInstitute Registry, started in 1979, indicate that approximately 400 new cases wereregistered in 1979, 1500 new cases in 1980 and 3000 new cases in 1981. Prior toMarch 1980 less than 30 institutions had enrolled in the registry; by September 1981there 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 DrugAdministration release of the Gruntzig balloon catheter for marketing in March1980 (Mullin, Passamani and Mock 1984). Ryan et al (1988) indicate that in the 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. Thefirst randomized trials to compare the efficacy of PTCA with that of CABS inmulti-vessel disease are presently underway but have not yet issued their reports.There are no trials comparing PTCA to medical treatment. The three randomizedtrials to date have compared short-term outcomes in PTCA versus fibrinolytic99therapy 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 fromthe National Heart, Lung, and Blood Institute's (NHLBI) registry which wasestablished in 1979. The initial protocol called for patients to have anginauncontrolled by medical therapy, lesions in one vessel that were proximal,accessible, discrete, concentric, segmental and high-grade and to be candidates forCABS. By September 1981, the results confirmed that PTCA in one-vessel diseasecould be carried out by an experienced operator with a relatively low rate ofcomplications. The registry was then closed to new institutions and only patientswith multi-vessel disease were enrolled. The registry was closed completely in1982 but was re-opened at its previous sites in 1985 in order to document changesin 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 beenobservational studies of case series. In a 1985 editorial Mock et al called for a RCTto provide reliable data on the true efficacy of PTCA, although these authorsbelieved that the "current enthusiasm" for PTCA for single vessel disease wouldmake an RCT for this population of patients "unrealistic". Preston (1988)attributed the lack of controlled studies to assess PTCA, to the lack of a physiciangroup with no self-interest but with the experience necessary for assessment. Inthe early-70's, cardiologists rather than surgeons were the critics calling forcontrolled studies. Ironically, this same group are the ones now extolling thevirtues of an inadequately assessed procedure!100Changes in PTCA:The 14 years since the introduction of PTCA have seen several technologicaladvances in PTCA methodology and equipment. The introduction of catheterswith guide wires and increased flexibility and steerability and balloons with lowerdeflated profiles has improved the immediate success rate (Schwartz 1988). Morerecently, the introduction of laser angioplasty allows angioplasty to reach lesionsthat cannot be crossed by a balloon catheter (Cumberland 1987) and may allow thevapourization or melting of coronary atheroma (Robischon 1987).Just as the patient population has changed in CABS over time, so too hasthe PTCA population changed. Analysis of NHLBI Registry data for 1977 to 1981and for 1985 to 1986 showed that during the later period patients were significantlyolder (mean age 57.7 versus 53.5 years) and that there were significantly morepatients over 65 years of age or with unstable angina, previous infarct, previousCABS, a low ejection fraction, or a history of CHF, diabetes or hypertension. Thepercentage 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 wasnot significant. Overall, patients coming for PTCA in the mid-eighties were sickerthan those in the late seventies and early eighties (Detre et al 1988). King andTalley (1989), examining changes in revascularization therapy in two institutionsbetween 1981 and 1986, found that in 1981, 11 percent of revascularization patientsreceived PTCA compared to 44 percent in 1986. The incidence of multi-vesseldisease among these patients increased from 11 percent in 1981 to 40 percent in1986.Despite the worsening health of the PTCA population over the years, in-hospital outcomes tended to be better. In the NHLBI registry, angiographic successrate per lesion increased from 67 to 88 percent between 1977-1981 and 1985-1986101and overall in-hospital success (reduction of at least 20 percent in all lesionsattempted without death, MI or CABS) increased from 61 to 78 percent. In-hospitalmortality and MI rates changed minimally, falling from 1.2 to 1.0 and 4.9 to 4.3respectively.MECHANISM OF ACTIONThe exact mechanism that causes enlargement of the arterial lumen duringPTCA is not known. Suggested mechanisms include compression andredistribution of the plaque, aneurysm formation, and disruption of the plaqueand arterial wall from overstretching of the wall. Embolization of plaque materialis believed to be minor and neither a mechanism for enlarging the lumen nor asignificant hazard during the procedure. It is likely that more than one of themechanisms suggested above are involved in lumen enlargement during PTCAand that arterial size, characteristics of the plaque, balloon size, and the amountand duration of pressure applied, all contribute to successful angioplasty(Bresnahan 1987a).EFFICACY OF PTCAThere are no data to determine the efficacy of PTCA in comparison tomedical treatment and few which help to determine the efficacy of the procedurein comparison to CABS. Consequently the discussion below will focus on theoutcomes of PTCA in various clinical and angiographic conditions. In general,comparisons with outcomes in CABS for similar conditions will not be madeunless the investigators in the study under discussion have made such acomparison. It is likely that differences between the CABS and PTCA populationswould make such comparisons invalid unless statistical adjustment has beencarried out. Also, as will be seen below, there is a difference in the way that102outcome incidence rates are calculated for PTCA and CABS which would alsomake comparison invalid.In many studies reviewed here the definition of "initial success" inenlarging 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 asignificant lesion. Another way in which the PTCA literature differs from theCABS literature is that the denominator for long-term outcome is generally notthe original cohort who received the procedure but rather the patients in thatcohort who realized initial success. The result is that in the long-term, PTCAappears to be more successful than it really is.A further difficulty in interpreting the literature is that many of the patientpopulations in different studies, overlap. The largest number of studies comefrom the data in the NHLBI Registry. This was a multi-institution registry andseveral of the institutions contributing data to the registry have also publishedreports of their own experience with PTCA. Where the dates of these studiesoverlap with the dates of the new or old NHLIB Registry, there is likely to be apatient overlap. There are, unfortunately, few U.S. studies from institutionswhich did not contribute data to the registryPTCA in Single Vessel Disease:Although patients with single-vessel disease were shown in the RCTscomparing CABS to medical treatment to have a good prognosis and a betteroutcome with medical treatment, this group is the one for which PTCA isgenerally advocated. Initially the NHLBI PTCA Registry guidelines restrictedPTCA to patients with single-vessel disease in an effort to maximize success andminimize complications during the era when operator experience was low.103Outcomes for patients in the Registry have been shown to differ by time and byoperator experience. Total initial clinical success rates (defined as reduction ofstenosis by at least 20 percent with no MI, death or emergency CABS) rose from55.7 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 55percent compared to 77 percent for those with more than 150 proceduresl. Someof 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 forsome 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 ratewas 63.6 percent; this figure rose to 84.3 percent between 1985 and 1986 (Detre et al1988) For the same time periods, in-hospital mortality for single-vessel disease fellfrom 1.3 percent to 0.2 percent, non-fatal MIs from 5.0 to 3.5 percent, emergencyCABS 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 havehad previous bypass surgery and those with left main disease. Although thesefigures have not been released specifically for single vessel disease, the NHLBIRegistry found an operative mortality of 0.9 percent for patients without theseconditions, compared to eight percent for those with left main disease and 4.2percent for those with previous CABS (Dorros et al 1983). Acute coronary closurefollowing PTCA- was found to be more likely in women and in patients withstenoses that were longer, at a bend or branch point, in vessels with other stenoses1 The assumption here is that increased operator experience leads directly to a better outcome. Thispattern of outcomes could also be due to referral filter bias in which high-risk cases tend to be referredto certain operators who, because of specialization, perform more high risk cases than those who arereferred the lower-risk cases. Outcome may be due to the patients' risk status rather than to thenumber of cases the operator has performed. This explanation appears unlikely in this case becausethe majority of patients in the PTCA Registry during the early years were low risk and because of thelarge number of operators involved in the analysis.104and in multi-vessel disease. There were in addition a number of proceduralfactors that were related to closure (Ellis et al 1988).Restenosis occurs in a substantial number of patients following an initialsuccessful procedure. Overall restenosis rates range from 20 percent to 50 percentwith the highest rates reported for patients with variant angina or for dilation ofvein grafts. Comparison of restenosis rates between institutions is complicated bydiffering definitions for restenosis and by different follow-up periods. Themajority of restenosis occur within eight months although some may developafter 12 months. Overall for single vessel disease, restenosis appears to occur inabout one third of patients in whom successful dilatation was achieved (Ischinger1986). 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 examinedfactors associated with risk of restenosis. Increased risk was found in men andthose with variant angina, eccentric, calcified or severe irregular lesions, stenosisof the LAD, or insulin-dependent diabetes mellitus. Lower risk was found forfemales and for patients who had intimal dissection and a final trans-stenoticpressure gradient less than 15 mm mercury after angioplasty.Repeat angioplasty for restenosis has a higher success rate and lowercomplication rate than initial procedures. Meirer et al (1984) found a highersuccess rate (97 percent versus 85 percent, p<0.001) and lower complication rate (8percent versus 15 percent, p<0.01) in 95 patients undergoing repeat angioplasty forrestenosis compared to patients undergoing initial angioplasty in the same timeperiod. Twenty-five percent of the 95 patients had a second restenosis. Similarimproved outcomes were found in NHLBI registry patients undergoing repeatangioplasty for restenosis (Williams et al 1984). For these 203 patients repeatangioplasty was carried out a mean of 147 days (median 126 days) following the105first procedure. Thirty-four percent of the 173 patients in whom repeat angioplastywas successful, suffered a second recurrence.Long-term outcome of PTCA has not been extensively studied. The 169patients who received the procedure from Gruentig prior to 1980 constitute thegroup with the longest follow-up; 40 percent of these patients had multi-vesseldisease. By 1987 90 patients remained asymptomatic and there were only 5 cardiacdeaths. Repeat angioplasty had been required in 27 patients and CABS in 19patients. Up to five year follow-up in the NHLBI registry indicated that 70 percentof patients receiving initial successful angioplasty were pain free at 4 years. Afterhospital discharge the annual mortality rate was one percent per year and the MIrate was 2 percent (Kent et al 1984).Mabin et al (1985) reports up to 3 year (mean 14 months) follow-up on 229patients who underwent PTCA, for symptoms unresponsive to medical therapy,at the Mayo clinic between 1979 and 1982. There was only one hospital death, andPTCA was successful in 153 patients overall (67 percent), but in only 61 percent ofpatients with single vessel disease. All successful patients with single vesseldisease had complete revascularization. Of the 76 patients who did not haveinitial success, 59 (78 percent) underwent prompt CABS (surgical group) and 17 (22percent) continued on medical therapy (medical group). By the most recentfollow-up 90 percent of patients in the successful PTCA and surgical groups weresubjectively improved; 74 percent of successful PTCA patients were asymptomaticcompared 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 CABScompared to 22 percent of the successful PTCA group and none of the surgicalgroup. 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 toldwhat percent of each group comprised those with single vessel disease. However,106we are told that the unsuccessful patients as a whole were older (mean 56 yearsversus 54 years), had fewer patients with unstable or variant angina, and had ahigher percentage of patients with single vessel disease (74 versus 57 percent) thanthe successful group. None of these differences were significant. These dataappear to indicate that CABS, even after unsuccessful PTCA, is more effective inreducing symptoms and the need for further intervention than successful PTCA.The low short- and long-term serious outcomes are impressive and may be anindication of the low-risk status of this series. The major question that arises fromthis study is whether the medical group would have experienced the sameoutcomes if they had not had an attempted PTCA.Hochberg et al (1989) performed a controlled retrospective study of 125consecutive angioplasty patients matched for single or double vessel disease with125 CABS patients; 87 patients in each group had single-vessel disease. Allprocedures were performed at one institution in 1984, prior to the introduction ofthe steerable catheter at that institution. Patients were stated to be low-riskalthough the range of ejection fractions was 22 to 67 percent (54 + 11 percent) forangioplasty patients and 13 to 70 percent (mean 49 + 12 percent) for CABS patients.This was the only baseline variable which was significantly different between thetwo groups. Patients requiring an emergency intervention or those having had arecent MI (within 6 weeks) or thrombolysis were excluded. Results can be seen inTable 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 effectivethan angioplsty. There were however, some problems with the statistical methodsused in this study. Firstly, although the statistical method used to analyse follow-up data is not stated, it appears to be either chi-squared or Fisher's exact test ratherthan a method such as acturial event-free rates which would be more appropriatefor time related data. Secondly, that authors considered that PTCA patients who107received CABS had "unsuccessful angioplasties" and removed them from thenumerator for 3-year functional classification. The denominator remained as thenumber of patients originally assigned to angioplasty. When PTCA patientsreceiving CABS are included in the numerator the difference in functionalclassification is no longer significant (Atkins 1989).TABLE 12COMPARISON OF OUTCOMES IN CONTROLLED TRIAL OF PTCA VERSUS CABSOutcome^ PTCA^CABS(n=125) (n=125)Length of stay 4.8 + 3.1 12.1 + 4.2*% %Initial success 88 99hospital mortality 3 1perioperative MI 4 53-Year follow-upoverall mortality 7 2.5late MI 6 3NYHA class I or II 63** 92*Repeat procedurePICA 18 2CABS 19 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 78percent - the difference between CABS and PTCA is then no longer significant.There have been a number of reports of significant left main disease arisingfollowing PTCA of the left coronary system. The incidence in the literature rangesfrom 0.2 to 1.7 percent of angioplasties to the left system. Of the 18 patientsreported in the literature, six had had no angiographic left main lesion prior toPTCA. The other 12 patients all had lesions that were less than or equal to 40percent stenosis and which progressed rapidly, between six weeks and 14 months,post-PTCA. Vardhan et al (1991) postulate that trauma to the left main arteryduring passage of the guidewire or balloon is partly responsible . They concludethat the possibility of progression of left main disease should be considered prior to108selection of LAD or circumflex artery PTCA and that left main disease should beconsidered in any patient with recurring symptoms after angioplasty to thesearteries.In summary, there are results which indicate that, over time, CABS is moreeffective for single vessel disease than PTCA. Given that the RCTs comparingCABS to medical treatment found that the latter was as effective, or more so, thansurgery for single vessel disease, then it appears that, in general, patients withsingle vessel disease should not be submitted to a surgical intervention but shouldbe treated medically. Of course patients with single vessel disease are aheterogenous group and some, those with left main disease for example, may be athigh risk. Even so, the reports that angioplasty of the left coronary system mayexacerbate left main disease means that these patients may have a better outcomewith CABS.Unfortunately, it is unlikely that single-vessel PTCA versus medical therapywill ever be tested in an RCT. The low incidence of death in this group means thatan extremely large number of patients would have to be randomized in order togain sufficient power to show a difference in mortality. It might, however, bepossible to gain some indication of the efficacy of PTCA in single-vessel disease bymeans of a meta-analysis should sufficient numbers of investigators a) conductstudies comparing PTCA and medical treatment in matched groups and b) publishtheir 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 patientswith multi-vessel disease are having the procedure Detre et al (1988) found anincrease in patients with multi-vesel disease (53 versus 25 percent) in the 1985-86109(new) compared to the 1977-81 (old) NHLBI Registry. Baseline data and outcomesare shown in Table 13. The small and non-significant increases in mortality forboth two and three vessel disease likely result from the worsening pre-proceduralcondition, and the increasing age, of the patients over time.Table 13COMPARISON OF OLD- AND NEW NHLBI REGISTRY CHARACTERISTICS AND OUTCOMESBY EXTENT OF DISEASEOne-vessel Two-vessel Three-vessel. TotalOld New Old New Old New Old NewN 836 839 203 568 89 395 1155 1802% % % % % % % %AttemptsMulti-lesion 5.0 21.6 17.7 53.2 21.3 59.2 8.5 39.8Left main - 4.4 1.3 10.1 1.8 1.6 0.8Bypass graft 1.0 1.0 5.4 3.7 20.2 14.9 3.3 4.9Angio successper lesion 68.6 89.0 60.7 86.4 66.1 88.0 66.8 87.3per patient 67.3 86.8 55.2 78.9 60.7 77.5 64.7 82.2Clinical success 63.6 84.3 51.2 74.6 58.4 70.9 61.0 78.3Outcome **operative death 1.3 0.2* 0.5 0.9 2.2 2.8 1.2 1.0nonfatal MI 5.0 3.5 3.9 5.1 6.7 5.1 4.9 4.3emergency CABS 6.1 2.9* 5.4 3.7 3.4 4.3 5.8 3.4elective CABS 19.9 1.7* 27.6 2.3 * 16.9 3.3* 20.7 2.2Adapted from Detre et al 1988 *significant at or below 0.05 level**Outcome during hospitalization for PTCAVlietstra (1987) states that determination of whether to use PTCA is morecomplex in patients with multi-vessel disease. Factors to consider are: the way themulti-vessel disease is defined, the degree of revascularization that can beachieved, the amount of myocardium at risk and the development of a dilatationstrategy.Definitions of multi-vessel disease vary from study to study. At the MayoClinic it is defined as a stenosis of 70 percent or more in at least one majorcoronary vessel and a stenosis of 50 percent or more in at least one other coronaryvessel. In addition they have sub-divided multi-vessel disease into four types110depending on the number of vessels with proximal severe stenoses and whetherany vessels are completely occluded.The degree of revascularization that can be achieved is related to the subtypeof multi-vessel disease. If only one vessel has a 70 percent or greater stenosis, asingle-dilatation may achieve complete revascularization, whereas in subtypeswith one or two occluded vessels complete revascularization may not be possible.The amount of myocardium at risk may be much greater in patients withmulti-vessel disease since the vessel being dilated may be supplying blood (eitherdirectly or through collaterals) to other diseased vessels. Consequently occlusionof the vessel being dilated may have the potential for jeopardizing a large amountof myocardium, resulting in severe hemodynamic and clinical effects.A dialatation strategy needs to be developed for each patient, depending onthe clinical importance of each stenosis. The two possible strategies are eitherdilation of all stenoses or dilatation of all functional stenoses. The worst stenosisis usually dilated firstIn other studies on patients having PTCA between 1979 and 1986, clinicalsuccess 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 termfollow-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 ayear follow-up, found the incidence of complications to be as follows: MI 2.3percent, 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 46percent of patients survived with no events and no symptoms.Data from Talley et al (1988), one of the few studies to examine long-termoutcome in the whole cohort of PTCA recipients, shows 5-year event-free111survival2 for patients with multi-vessel disease to be 72.4 percent. The post-discharge 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 5-year survival was 94.8 percent while that of survival with no CABS post-dischargewas 84.5 percent. The five-year reported figures are somewhat misleading sincethey do not include initial outcomes. There were no in-hospital deaths but whenother in-hospital events are include in the five-year outcomes the incidence of MIis 11 percent and that of CABS is 24 percent. Talley's data also showed a trend forworse initial and long-term outcomes in those with initial PTCA failure. Theseresults will be discussed later.Restenosis is theoretically more likely for patients receiving multipledilatations because of the greater number of lesions that could restenose.Vandermael et al (1987) studying restenosis following multiple dilatations, foundrestenosis at one dilatation site in 33 percent of 129 patients undergoing follow-upangioplasty, and at more than one dilatation site in an additional 19 percent ofpatients. Eighty-two percent of syptomatic patients had restenosis compared to 30percent of asymptomatic patients. The restenosis rate per lesion was 29 percent.These authors report that the restenosis rate for multi-lesion angioplasty reportedin the literature ranges from 26-68 percent but that rates are likely biased upwardsby the greater number of symptomatic patients studied by angiography.Another concern with angioplasty for multi-vessel disease, is that allsignificant stenoses may not be dilated. Incomplete revascularization has shownto 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 astudy of 286 NHLBI Registry patients with multi-vessel disease and prior2 Event-free survival was defined as no cardiac death, CABS or MI.112successful angioplasty who were followed-up for a mean of 26.2 months. Initialanalysis showed that mortality, MI and CABS rates were all higher in the groupwho had incomplete revascularization. When differences were made for baselinedifferences between the two groups however, estimates of the risk of death, MI orpresence of angina did not differ between the groups. The group with completerevascularization had more repeat PTCA procedures while those incompletelyrevascularized had more CABS during follow-up. From these results it appearsthat outcomes of PTCA at two years of follow-up are not affected by the incompleterevascularization.PTCA in Unstable Angina:The NHLBI Registry documented an increase in the proportion of patientsundergoing PTCA for treatment of unstable angina in the new versus the oldregistry (Detre et al 1988). Comparison of registry patients with stable and unstableangina showed no difference in immediate success rate; for successful patientsthere was no difference in the combined MI and mortality rate both in-hospitaland at the 18-month follow-up. Other observational studies of patients withunstable angina found a rate of angiographic success from 84-93 percent, operativemortality from 0.2-0.9 percent, perioperative MI from 6.6-10.8 percent and latedeath 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 whenangioplasty was performed within one week of the onset of angina than when twoor more weeks had elapsed before the procedure.The complication rate following PTCA for total occlusion has been shownto be significantly higher in patients with unstable angina compared to those withstable angina (Plante et al 1991). Complications occurred only among those withangina at rest or pre-infarction angina. The authors speculate that the presence of113intraluminal thrombus may increase the risk of acute vessel closure or ofembolization.PTCA in Left Ventricular Dysfunction:Serota et al (1991) studied 73 consecutive patients with low LVEF (range 14-40 percent, mean 34 percent) who received PTCA between 1983 and 1989. Clinicalsuccess was achieved in 88 percent of patients, in-hospital mortality was 5 percentand the MI rate was 3.9 percent. Estimated survival, for successful patients, fromone to four years was 79 percent, 74 percent, 66 percent and 57 percent respectively.Predictors of cardiac mortality were congestive heart failure and low ejectionfraction. These survival figures appear low when compared to the five yearoutcomes for the medical group in the VA trial3 but the patients in Serotoa et al'sstudy were sicker than those in the VA trial.Post-Myocardial Infarction:The efficacy of PTCA versus thrombolysis and the optimum timing ofPTCA following MI has been tested in a number of RCTs and controlled trialswhich are reviewed in a chapter by Pitt (1990). The strategy of urgent PTCAfollowing successful perfusion with thrombolyis was tested by the Thrombolysisand Angioplasty in Acute Myocardial Infarction (TAMI) trial, the EuropeanCooperative Study Group and the TIMI-IIA trial. All these trials found asignificantly increased risk, plus no demonstrated improvement in ventricularfunction, following urgent PTCA after thrombolysis. The strategy of delayed PTCAfollowing thrombolysis was studied by the TIMI-IIB investigators who randomly3 Survival for the medical group in the VA trial at five year follow-up was 73 percent for all patientswith 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.114assigned patients to an invasive strategy (angiography at 18 to 72 hours post-thrombolysis with PTCA in vessels with significant residual stenosis) or to a non-invasive strategy in which patients received PTCA only if symptomatic. Nobenefit regarding LVF or survival was found in the group randomized to theinvasive strategy.Pitt also reports on a small trial which randomized 56 patients to eitherPTCA without prior thrombolytic therapy or to thrombolysis. Reperfusion rateswere similar but the PTCA group had significantly less residual stenosis in theinfarct-related artery, better LVEF, less post-infarction angina and less exercise-induced ischemia on stress testing. Numbers were too small to show the relativeeffects of these two strategies on survival. O'Keefe et al (1989) also report on 500consecutive patients with MI who received PICA without antecedentthrombolysis. Successful angioplasty was achieved in 94 percent of patients but 15percent of these reoccluded before discharge. The three strongest predictors of the7.2 percent hospital mortality were cardiogenic shock, multi-vessel disease andfailed angioplasty.It appears that PTCA following thrombolysis is not more effective and maycarry a higher risk than thrombolysis alone. The strategy of direct PICA withoutthrombolysis appears promising but has not been adequately tested to assureefficacy.Gender:Data from the 1977-1982 NHLBI Registry showed that PICA in women wasassociated 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 electiveCABS (23.5 versus 17.6 percent) and post-emergency surgery mortality (17.4 versus3.2 percent) rates than in men. Long-term results however were comparable or115better in women than in men. After 18-month mean follow-up, women withinitially successful PTCA had lower incidence of mortality (0.3 versus 2.2),restenosis and additional revascularization procedures, but had comparablesymptomatic improvement and higher event-free survival (79.7 versus 69.0percent) than men. Although a higher percentage of women in the Registry hadunstable angina, class III or IV angina and were older than men, the males had ahigher incidence of multi-vessel disease, impaired LVF and prior bypass surgery.No analysis was performed that would show if the women's higher hospitalmortality was due to higher risk (Cowley et al 1985b).The Elderly:Because CABS results in a significantly higher morbidity and prolongedhospitalization among the over 65-year age group, PTCA would appear to be abetter alternative for the elderly. Mock et al (1984), in an analysis of the NHLBI1977-1982 Registry data, compared outcomes in 370 patients over the age of 65 and2,709 patients under 65 years. The older group had a significantly greaterproportion of females, patients with prior CABS, low ejection fraction and severeangina, but a significantly smaller proportion of patients with previous MI.The mean overall clinical success rate was lower in the elderly group (53percent versus 62 percent) but hospital mortality (2.2 versus 0.7 percent) andelective CABS (25.4 versus 8.1 percent) were significantly higher. For the wholecohort, in-hospital mortality compares favourably with that reported for CABS inthe CASS Registry over-65 population (2.2 versus 5.2 percent), but differences inpatient selection may account for this difference, Mean length of stay was slightlygreater for the elderly group. At one year follow-up the over-65's had significantly116higher mortality and CABS rate, but a lower incidence of repeat PTCA4 . Whenone-year outcomes for only the successful patients in each group were analyzed,only the late PTCA rate was still significant; the other outcomes were remarkablesimilar between the two groups.This is one of the two studies reviewed here which gives sufficientinformation to allow the incidence of outcomes to be calculated for theunsuccessful PTCA recipients. These calculations (see Table 14) show that themortality 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 elderlypatients. MI and CABS rates are both approximately six times higher in theunsuccessful patients than they were in those, of whatever age, with initialsuccess. Amazingly, over 75 percent of unsuccessful patients had had CABS by thefirst year follow-up; this included emergency and elective CABS both during theinitial 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 forthe under- and over-65's with failed PTCA, the overall effect on the older PTCAcohort 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 theelderly are likely higher today and complication rates are likely lower, althoughthere appear to be no data to show this. Whether the incidence of complicationsseen in the unsuccessful patients will have changed is uncertain since thesepatients are rarely addressed in the literature.4 It is not clear whether this finding is due to an age bias in recommending treatment or whether theclinical condition of the older patients warrented the more invasive intervention. The highermortality in the older group would support the second explanation but that mortality may have beendue, in part, to the higher incidence of surgery.117TABLE 14COMPARISON OF ONE-YEAR OUTCOMES IN ELDERLY AND NON-ELDERLY PATIENTSAFTER SUCCESSFUL AND UNSUCCESSFUL PTCAOutcome Under 65 years 65 years and overAll Success Failure All Success FailureN 2709 1680 1029 370 196 174% % % % % %Death 2.2 1.4 3.7 4.5 1.9 7.4MI 7.8 2.8 15.9 8.3 2.5 14.9CABS 36.1 11.6 76.1 41.5 11.4 75.2PTCA 10.8 14.8 4.3 6.4 9.9 2.9Data from Mock et al 1984.In summary, PTCA performed in the 1977-1982 'early era' had a significantlyhigher failure rate, operative and one-year mortality rates, and elective CABS ratefor those over 65-years of age compared to younger patients. The markedly worseoutcomes for patients with unsuccessful PTCA combined with the almost 50percent failure rate in the elderly may indicate that PTCA is not the mostefficacious and/or cost-effective treatment in the over-65 population. Well-designed controlled studies are required to show efficacy.Unsuccessful PTCA:The data shown above from Mock et al (1984) seems to indicate that patientswho experience an immediately unsuccessful PTCA tend to have worse one-yearoutcomes than those in whom PTCA is initially successful.The other study reviewed here, in which the data showed worse outcomesin the initially unsuccessful PTCA patients was that of Talley et al (1988) whoreported on 427 patients who underwent PTCA in 1981. Eighty-nine (20.8 percent)of the patients had clinically unsuccessful PTCA. For these patients the incidenceof in-hospital outcomes were MI 25.8 percent and CABS 46 percent; there were nountoward outcomes in the successful patients. At five year follow-up the118incidence of adverse outcomes overall for both successful and unsuccessfulpatients respectively were: mortality 2.9 versus 5.6 percent; cardiac death 0.9 versus4.4 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 initiallyunsuccessful PTCA, though they did not give results for these patients. It is notpossible to tell from the reported data whether unsuccessful patients were at aninitially higher risk than the successful patients, or whether attempted PTCAincreased the risk for an adverse outcome. It appears that adverse outcomes arenot due only to outcomes from emergency CABS because Kent et al reported worseoutcomes in those with unsuccessful PTCA who were managed on medicaltherapy. Whatever the cause, the very fact that long-term outcomes are worse inthese patients puts into question the common practice of reporting long-termoutcomes only on patients with initial success 5 . This practice also preventscomparison of the results of angioplasty research with those of CABS research forsimilar patient populations.These results raise a number of questions about the efficacy of PTCA overalland in certain patient populations. The major question to be answered is whetherthe 'unsuccessful' patients in the above studies would have had the sameoutcomes whether or not they had had attempted PTCA. It appears possible thatpatients who have unsuccessful PTCA may be much worse off than if they had nothad the procedure. These patients appear to constitute a high-risk group inangioplasty 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 some5 By reporting only on "successful" patients receiving PTCA it would appear that PTCA investigatorsare not adhering to the "intent-to- treat" principal that was followed in the analysis of the CABSRCTs; rather they appear to be basing their analysis on successful treatment received. The assumptionbeing made here is that an unsuccessful attempt at angioplasty has no effect on the patient; the aboveresults show that this assumption is not valid.119common characteristics that would enable them to be recognised before theyreceive the procedure. If not, then other factors which contribute to immediatefailure need to be investigated and, if possible, remedied.Ultimately, many of the above question can only be answered by arandomized trial or a well designed prospective controlled study comparingmedical treatment with angioiplasty. Until such a study is carried out it isimperative that investigators report, for both short-term and long-term outcomes,on the whole cohort of patients receiving angioplasty. Analysis of patients havingimmediate success and those who are unsuccessful should be carried out for thegroup as a whole and separately. Analysis for the group as a whole will allowcomparison with the results of investigations into the use of CABS or medicaltreatment in similar patient populations, and will give a more realistic estimate ofthe outcomes to be expected after PTCA. Separate analyses for successful andunsuccessful patients will allow comparison of outcomes between these twogroups and may help to identify patients in whom failure is more likely to occur aswell as factors which may contribute to the worse outcomes in the unsuccessfulpatients. Full reporting of results of angioplasty investigations will facilitate meta-analysis, which may be used to supplement the results from randomized trials.Summary:To date, the efficacy of PTCA has been mainly evaluated using surrogateoutcomes, i.e., blood flow in the treated artery immediately following theprocedure, but there is no data to show how immediate increased blood flowrelates to later morbidity and mortality. There are results which show that urgentor delayed PTCA following thrombolysis is not more effective, and carries anincreased risk, than thrombolysis alone. In addition, results from oneretrospective controlled trial indicate that CABS may result in a better long-term120outcome than PTCA, and results from the only two studies reporting data forpatients having immediate PTCA failure indicate that these patients may have amarkedly worse outcome than patients with initial success.RISKSAngioplasty carries a risk of significant and potentially fatal complications.Results on the 3079 patients from the 1977-1982 NHLBI Registry, reported byDorros and Cowley (1986), show that a total of 1180 complications occurred in 652patients (21 percent). These authors subdivided complications into two groups,acute coronary events and non-coronary events. The latter group include hospitaldeath despite the fact that death often resulted from a coronary event. Eighthundred and thirty acute coronary events occurred in 418 patients while non-coronary events occurred in 234 patients.The most commonly occurring acute coronary events were prolongedangina, MI, coronary occlusion, coronary dissection and coronary spasmProlonged Angina was the most frequent complication occurring in 6.8percent of patients and was associated with major complications (death, MI, andemergency CABS) in over half these patients. Univariate analysis showed thatprolonged angina was more likely to occur in patients with unstable angina, classIV angina, with eccentric lesions or those with stenoses greater than 90 percent.Myocardial Infarction occurred in 5.5 percent of patients overall, in 45percent of patients who had emergency bypass surgery, in 3.6 percent who hadelective CABS and in 2.6 percent who did not undergo CABS. Fifty percent ofpatients suffering an MI did so within 24 hours of PTCA. Death occurred in 9.4percent of patients who had MI. The primary complications associated withoccurrence of MI were coronary dissection, coronary occlusion, prolonged anginaand coronary spasm.121Several authors have noted the occurrence of subendocardial MI followingangioplasty on saphenous vein grafts, but this has not been implicated (and waspossibly not tested) as a predictor for MI (Dorros and Cowley 1986).Coronary Dissection occurred in 12.9 percent of 3079 patients undergoingPTCA; of these, 70 percent had no adverse effects and 218 had clinically successfulPTCA. Of the 34 percent of patients with a dissection who developed majorcomplications, 15 percent had a non-fatal MI, 23 percent had emergency CABS and1.5 percent of patients died. Univariate and multivariate analysis showed thatcoronary dissection was more likely to occur in women, in patients undergoingright artery PTCA, in those with multivessel disease, eccentric lesions or with non-discrete lesions.Coronary occlusion occurred in 4.9 percent of PICA patients and majorcomplications arose in 81 percent of these. Forty-one percent had an MI, 72 percentrequired emergency CABS and 5.3 percent died in hospital. Predictors of coronaryocclusion by univariate analysis were onset of angina within 6 months, eccentriclesions, stenoses greater than 90 percent and tubular or non-discrete lesions.Coronary Spasm occurred in 4.2 percent of patients, 32 percent of whom hada major complication. Non-fatal myocardial infarct arose in 12 percent while 24percent required emergency CABS and three percent died in hospital. Spasm wasmore likely to occur in non-calcified lesions and in patients of a younger age.Multivariate analysis showed only younger age to be a significant independentpredictor of coronary spasm.Overall, univariate analysis showed that acute coronary events were morefrequent in women, in patients with unstable angina, with initial lesion severity122of more than 90 percent, with eccentric lesions 6 , nondiscrete or tubular lesions7 .With multivariate analysis, unstable angina, severe stenosis, nondiscrete lesionsand 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 29patients but nine of these deaths were considered to be unrelated to the PTCAprocedure because they occurred during or after CABS. This seems to beunnecessary hair-splitting because presumably the need for CABS arose because ofcomplications 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 (1.7percent versus 0.7 percent mortality under 60 years), and duration of angina formore than one year (1.7 percent versus 0.5 percent mortality when angina presentfor less than 6 months). Angiographic characteristics also had an effect onmortality. There was a trend for lesion location to affect mortality with highermortality in patients with PTCA to circumflex or left main artery lesions. Thepresence of left main disease was a significant risk factor but dilatation of a leftmain lesion was not. Thus, left main disease increases the risk of hospital deathregardless of the site of dilatation.Factors found to have no influence on hospital mortality rates were lesionseverity, history of diabetes, history of elevated cholesterol, unstable angina anginaseverity, lesion calcification and previous MI. Factors that were predictive of6 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 aremost suitable for PTCA (Ischinger 1986).123mortality were lesion location, sex, time since onset of angina, previous bypasssurgery and number of vessels diseased; R2 figures were not reported. Onmultivariate analysis however, only female gender was significantly associatedwith PTCA-related deaths.Emergency Bypass Surgery:Dorros and Cowley (1986) report that emergency bypass surgery wasperformed in 6.6 percent of NHLBI PTCA Registry patients; the most frequentproblems necessitating this surgery were coronary dissection and coronaryocclusion. Patients receiving emergency CABS had a high incidence of MI (45percent) and mortality (6.4 percent). On univariate analysis, eccentric lesions, non-discrete lesions and severe lesions were predictive of emergency CABS, althoughonly lesion eccentricity was predictive on multivariate analysis. Proceduraldifficulties, e.g., inability to pass the stenosis or to dilate it once passed, were alsoassociated with increased emergency CABS; possibly these difficulties result fromthe lesion characteristics that are also associated with increased surgery.Dorros and Cowley (1986) state that in the early days of PTCA emergencysurgery was used whenever a significant complication occurred because it wasbelieved that prompt CABS could prevent the evolution of an MI. However,when analysis of the Registry data showed that emergency surgery was associatedwith a significant mortality and almost half the patients still had an MI, theymodified the treatment of myocardial ischemia or occlusion following PTCA.They reduced emergency surgery to 1.3 percent in Milwaukee during 1984, eventhough these patients had more extensive disease, a higher incidence of previousCABS and a higher incidence of previous MI than was found in the NHLBIpatients. It seems likely that most large centres today will have a similar incidenceof emergency surgery, although there does not seem to be data to back this up.124In summary, PTCA carries significant risks of complications which may leadto MI or death. How these risks compare to the risk of these outcomes duringtreatment with medical therapy, has not been tested.Other Factors Affecting Risk:Ryan et al (1988) in their guidelines for PTCA classify lesions into threedifferent types which may be used to estimate the likelihood of successfulangioplasty and the likelihood of development of complications. Lesions areclassified according to morphology and location. Type A lesions are those inwhich the anticipated success rate should be 85 percent or greater and the risk ofabrupt vessel closure is low. Type B lesions have an anticipated success rate from60 to 85 percent and a moderate risk of abrupt closure, while type C lesions have ananticipated success rate less than 60 percent and a high risk of abrupt closure. Ryanand colleagues caution that attempts to dilate type C lesions should not beundertaken when they are present in vessels supplying large or moderate areas ofviable myocardium.Bresnahan (1987b) notes that risk profiles should also include other riskfactors, such as age, gender, and duration of angina and experience of the operatorwho will perform the PTCA. He states that young men with angina of recentonset and concentric non-calcified lesions of 70 to 90 percent severity are excellentsubjects for angioplasty. Older patients, especially women, with long-standingangina and high-grade or calcified stenoses will likely have a much pooreroutcome. He also states that because current technology limits the present abilityto assess the pathologic characteristics of stenoses, steps cannot be taken to avoidcoronary dissection or occlusion. It seems unlikely, therefore, that thecomplication rate will improve before there are significant improvements incoronary angiography.125As in CABS, operator experience has been shown to affect the outcome ofPTCA. An NHLBI Registry report (Kelsey et al 1984) involving 3000 patientsreceiving PTCA from 105 centres between 1977 and 1981, showed that the successrate improved as the number of procedures per operator increased although thelearning process was relatively long; success rates continued to improve beyond150 cases. As operator experience increased there was a corresponding decrease inthe incidence of elective CABS and in-hospital mortality. The incidence ofemergency CABS, however, did not decrease significantly. Bresnahan (1987b)speculates that this may be because coronary dissection and occlusion areindependent of operator experience. He also states that the effect of training andtechnologic improvements in balloon catheters has shortened the learning curve.As a result, "angiographers" today, who are trained by those experienced inangioplasty with steerable balloon catheters, may achieve a 90 percent success ratefrom the outset.As discussed earlier, it appears that failed angioplasty may carry greater riskof complications and death than successful angioplasty. The lack of coverage ofthis problem in the literature seems to imply that clinicians do not consider it to bea problem. There does not appear to be any research that looks specifically at eitherpredictors of failure or at optimum treatment of patients following failedangioplasty. Such studies are obviously needed.Finally, the place that angioplasty is carried out may constitute a risk. Theguidelines for PTCA developed by Ryan et al (1988) state that the minimal facilitiesfor any hospital where PICA is carried out, should include a surgical operatingsuite that is equipped to provide cardiac surgery (they failed to mention the needfor a cardiac surgeon). In Europe PTCA may be carried out at hospitals withoutthese facilities; patients who require surgery are transferred to a nearby surgicalunit. Shaw (1990) suggests that these circumstances have not resulted in clinical126deterioration in any patient requiring emergency surgery after PTCA. Parker(1990), giving the surgeons view, states that reports on patients who have beentransferred for surgery show an unacceptably long mean time to revascularization.He points out that the earlier the patient comes to surgery the greater the chancethat the surgeon will use the internal mammary artery, which is associated withsignificantly better long-term outcome. He concludes that angioplasty supportedby on site surgical facilities is the correct policy.COSTSHolmes et al (1984) using data from the NHLBI Registry, measuredemployment status and time of return to employment (representing an indirectsocietal cost) in 2,250 patients who underwent PTCA before 1981. Patients weredivided into three groups: Group A, those with successful PTCA; Group B, thosewith immediate failure followed by CABS; and Group C, those with immediatefailure and medical therapy. At follow-up (mean 1.5 years) the decrease in thepercentage 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 base-line and were 60 years old or younger. At follow-up 86 percent of group A, 81percent of group B and 83 percent of group C were employed. However, for groupA the mean time to return to work was 7 days, compared to 73 days for group Band 13 days for group C. Reeder (1987) comments that successful PTCA has thepotential for allowing a more rapid return to work than the 6 weeks to 3 monthsdisability which generally occurs after CABS.Reeder (1987) in a chapter on costs of PTCA, reports on a number ofobservational studies which all use different costing methodology. All studiesfound substantially lower charges for the initial PTCA compared to initial chargesfor CABS. Two studies which examined expenditures for angioplasty versus127CABS 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 andservices to estimate costs, found one-year expenditures of $7,689 per angioplastypatient and $13,559 per CABS patient. The primary success rate was 74 percent andrestenosis occurred in 18 percent of these patients. These authors found that theaverage hospital stay for patients who had initially unsuccessful angioplasty wastwice as long as for those who had CABS 8; this obviously increased the averagefirst-year cost for angioplasty patients but the low restenosis rate may have off-setthis. The rate of return to work was between 86 and 100 percent and did not differsignificantly between patients having primary successful angioplasty, primarysuccessful CABS or failed angioiplasty with subsequent surgery. The time ofreturn to work was, however, significantly less in the successful angioplasty group.Reeder et al (1984), using actual patient costs, found that initial proceduralcosts were $5,493 for angioplasty (including a small charge for surgical standby forsuccessful angioplasty) and $12,065 for CABS. The primary success rate was 70percent and restenosis occurred in 33 percent of cases. For PTCA patients who hadno restenosis the average first year costs were 56 percent of the CABS costs; forthose with restenosis the average cost was $2,700 more than the average CABScost. In a later report of the same study, Reeder states that the mean first-yearexpenditures for the treatment of restenosis were $10,002 for medicalmanagement, $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 thepremise that angioplasty may be substituted for CABS in certain patients, and costsavings realized. In fact the whole reasoning behind the development of PTCAwas, presumably, to develop a procedure which would achieve the same effect as8 A further indication that unsuccessful angioplasty may increase risk of an adverse outcome.128CABS but with a lower mortality, morbidity and cost. Reeder (1987) points out thatif the use of angioplasty increases to include patients who would not normally becandidates for CABS, the concept of substituting a low-cost procedure for a high-cost one no longer obtains. In this situation, revascularization costs wouldincrease rather than decrease. Reeder presents data to show that at the MAYOClinic, and in the U.S. as a whole, the use of PTCA has increased dramaticallywithout a reduction in the number of bypass procedures. There is, therefore, noevidence that PTCA is substituting for CABS, and the economic argument fallsflat; PTCA is not a cost-saver but is adding to the costs of coronaryrevascularization.It is interesting that no studies have been done which compare the cost ofPTCA to that of medical treatment. Given that about 50 percent of PTCAs areperformed for single vessel disease, for which CABS is very rarely done, thealternative treatment for comparison should be medical therapy.In summary, although initial PTCA, as a procedure, is substantially cheaperthan initial CABS , patients who require further treatment because of initialfailure or restenosis may have expenditures in the first year which are onlyslightly less, or even more, than those for CABS. In addition it appears that PTCAmay not be substituting for CABS but may, instead, be additive. Therefore itappears likely that increased use of PTCA in an area will increase, rather thandecrease, revascularization costs.CONCLUSIONGiven that the efficacy of coronary angioplasty in the treatment of coronaryartery disease has never been rigorously evaluated, its continued phenomenalgrowth over a 14-year span, is disquieting.129PTCA is reputed to be a substitute for CABS but is used frequently in single-vessel disease, which is usually managed medically. The coronary artery surgeryRCTs showed that medical treatment is as effective as CABS and in fact resulted inbetter long-term survival. Although this finding was likely due to the operativemortality found with CABS, there is no reason to believe that PTCA wouldperform better than CABS in comparison to medical treatment. Although theincidence of hospital mortality is relatively low with PTCA, the approximately 30percent incidence of restenosis means that patients may be subjected to two ormore angioplasties. The risk of mortality with each may be low but, for thepatient, these risks are additive and may ultimately exceed the risk of CABS. Oncerestenosis has occurred, the average costs of an initial strategy of PTCA exceedthose of initial CABS. With a 30 percent restenosis rate there are likely to berelatively few cost savings with a strategy of initial PTCA.The costs to the patient includes the risks of death, morbidity, andemergency or elective CABS. For patients with failed angioplasty and thoserequiring emergency surgery, the risks are much higher than those with electiveCABS. There appears to be no research identifying the characteristics of thosepatients in whom PTCA is not successful. In fact, these patients appear to be lostsouls in the literature, generally mentioned only as a "rate" and left out of long-term analyses. The implication appears to be that PTCA failure has no effect onthese patients, whereas it has been shown that, in at least one study, outcomes atone 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 thosewho perform angioplasty, it is possible for a patient to be "referred" for PTCA bythe same professional who will carry out the procedure. In this case the advantageof the second-opinion, which is an integral part of a referral to another physician,130is lost. In addition, there is the potential for a physician to refer patients motivatedas much by self-interest as by patient need. Ryan et al (1988) suggest that insituations where the 'referring' cardiologist is also the one who will carry out thePTCA, the responsible physician should arrange for a consulting opinion fromanother 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-invasivetreatment, should have been the first step and needs to be established as soon aspossible by means of prospective controlled studies, in which patients areprospectively matched for risk, or by randomized controlled trials.131REFERENCESAtkins, C.W. 1989. Discussion following Hochberg, M.S. et al, Coronaryangioplasty versus coronary bypass. T Thorac Cardiovasc Surg 97 (April):502-503.Bresnahan, D.R., 1987a. Mechanism of action. Chapter 2 in PICA: PercutaneousTransluminal 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 influenceimmediate success and clinical applications. Chapter 5 in PTC A : Percutaneous Transluminal Coronary Angioplasty,  ed. R.E. Vlietstra andD.R. Holmes, 63-72. Chicago and London: The University of Chicago Press.Cowley, M.J. et al, 1984. Acute coronary events associated with percutaneoustransluminal coronary angioplasty. Am T Cardiol 53:12c-16C.Cowley, M.J. et al, 1985a. Coronary angioplsty of multiple vessels: short-termoutcome and long-term results. Circulation 72 (December):1314-1320.Cowley, M.J. et al, 1985b. Sex differences in early and long-term results of coronaryangioplasty in the NHLBI PTCA Registry. Circulation 72 (January):90-97.Cumberland, D.C., 1987. Laser coronary angioplsty. Br T Hosp Med (April):281Cumberland, D.C., 1987. Laser coronary angioplasty. British Journal of HospitalMedicine (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 CoronaryAngioplasty Registry. Am T Cardiol 54 (January):7C-11C.Detre, K. et al, 1988. Percutaneous transluminal coronary angioplasty in 1985-1986and 1977-1981: The National Heart, Lung and Blood Institute Registry. NEngl T Med 318 (February):265-270.Dorros, G. et al, 1983. Percutaneous transluminal coronary angioplasty: report ofcomplications from the National Heart, Lung and Blood Institute PICARegistry. Circulation 67 (April):723-730.132Dorros, G. and M. Cowley, 1986. Complications associated with PTCA. Chapter 13in Practice of Coronary Angioplasty, ed T. Ischinger, 223-249. Berlin:Springer-Verlag.Dorros, G., R.F. Lewin and L. Janke, 1987. Multiple lesion transluminal coronaryangioplasty in single and multivessel coronary artery disease: acute outcomeand long-term effect. T Am Coll Cardiol 10 (November):1007-1013.Ellis, S.G. et al, 1988. Angiographic and clinical predictors of acute closure afternative vessel coronary angioplasty. Circulation 77 (February):372-379.Faxon, D.P. et al, 1983. Role of percutaneous coronary angioplasty in thetreatment of unstable angina. Am J Cardiol 53:131C-135C.Feyter, P.J. et al, 1985. Emergency coronary angioplasty in refractory unstableangina. N Engl T Med 313 (August):342-346.Hochberg, M.S. et al, 1989. Coronary angioplasty versus coronary bypass. T ThoracCardiovasc Surg 97 (April):496-503.Holmes, D.R. et al, 1984. Return to work after coronary angioplasty: A report fromthe National Heart, Lung and Blood Institute Percutaneous TransluminalCoronary 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 andD.R. Holmes, 161-187. Chicago and London: The University of ChicagoPress.Ischinger, T. and B.Meirer, 1986. Outcome of coronary angioplasty. Chapter 11 inPractice of Coronary Angioplasty ed T. Ischinger, 194-210. Berlin: Springer-Verlag.Jones, E.L. et al, 1983. Importance of complete revascularization in performance ofthe 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 percutaneoustransluminal coronary angioplasty. Am T Cardiol 53 (June):56C-64C.Kent, K.M. et al, 1984. Long-term efficacy of percutaneous transluminal coronaryangioplasty (PICA): Report from the National Heart, Lung and BloodInstitute PTCA Registry. Am J Cardiol 53 (June):27C-31C.133King, S.B. and J.D. Talley, 1989. Coronary arteriography and percutaneoustransluminal coronary angioplasty: changing patterns of use and results.Circulation 79 (suppl I):I-19-4-23Meier, B. et al, 1984. Repeat coronary angioplasty (abstract). T Am Coll Cardiol4:463.Mabin, T.A. et al, 1985. Follow-up clinical results in patients undergoingpercutaneous transluminal coronary angioplasty. Circulation 71 (April):754-760.Mock, M.B. et al, 1984. Percutaneous transluminal coronary angioplasty in theelderly patient: experience in the National Heart, Lung, and Blood InstitutePTCA Registry. Am T Cardiol 53 (June):89C-91C.Mock M.B. et al, 1985. Percutaneous transluminal coronary angioplasty versuscoronary artery bypass. Isn't it time for a randomized trial? N Engl T Med312 (April):916-918.Mullin, S.M., Passamani, E.R. and Mock, M.B., 1984. Historical Background of theNational Heart, Lung and Blood Institute Registry for PercutaneousTransluminal Coronary Angioplasty. Am I Cardiol 54 (January):3C-6CMyler, 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 withoutantecedent thrombolytic therapy for acute myocardial infarction. Am TCardiol 64 (December):1221-1230.Parker, D.J., 1990. Does angioplasty need on site surgical cover? A surgeon's view.Br Heart T 64:1-2Pitt, B., 1990.^Percutaneous Transluminal coronary angioplasty in acutemyocardial 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 coronaryangioplasty for total occlusion. American Heart Journal 121 (February) Part1:417-426.Preston, T.A., 1989. Assessment of coronary bypass surgery and percutaneoustransluminal angioplasty. Intl T of Technology in Health Care 5:431-442.134Reeder, G. et al, 1984. Is percutaneous coronary angioplasty less expensive thanbypass surgery? N Engl I Med 311 (November):1157-1162.Reeder, G.S., 1987. Socioeconomic aspects. Chapter 16 in PTCA: PercutaneousTransluminal 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 multivesselcoronary disease: a report from the National Heart, Lung, and BloodInstitute Percutaneous Transluminal Coronary Angioplasty Registry.Circulation 77 (March):638-644.Robischon, T., 1987. No ideal replacements yet for bypass, angioplsty. The MedicalPost (October):30.Ryan, T.j. et al, 1988. Guidelines for percutaneous transluminal coronaryangioplasty. 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'sview. Br Heart Journal 64:3-4.Serota, H. et al, 1991. Predictors of cardiac survival after percutaneoustransluminal coronary angioplasty in patients with severe left ventriculardysfunction. Am I Cardiol 67 (February):367-372.Talley, J.D. et al, 1988. Clinical outcome 5 years after attempted percutaneoustransluminal coronary angioplasty in 427 patients. Circulation 77(April):820-829.Vandormael, M. et al, 1987. Multilesion coronary angioplasty: clinical andangiographic follow-up. I Am Coll Cardiol 10 (August):246-252.Vandormael, M. et al, 1991. Predictors of long-term cardiac survival in patientswith multivesel coronary artery disease undergoing percutaneoustransluminal 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 ofperceutaneous transluminal coronary angioplasty - left main disease. AmHeart 1 121 (March):902-905.Williams, D.O. et al, 1984. Efficacy of repeat percutaneous transluminal coronaryangioplasty for coronary restenosis. Am I Cardiol 53:32C-35C.135CHAPTER 3REGIONAL VARIATIONSINTRODUCTIONFor over 20 years, health utilization researchers have been documentingregional differences in per capita rates of medical service use including hospitalutilization and surgical procedure rates. These variations have been found withinand between countries and for different sizes of region studied. Manyinvestigators have tried to account for these variations by using multipleregression 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 howand why such variations occur, whether they are of concern and, if so, whatshould be done about them.This chapter will briefly describe the results from the literature onprocedural variations in general, discuss the few studies that looked at variationsin CABS and related procedures and describe some of the methodologicalproblems associated with regional variations studies.PROCEDURAL VARIATIONS IN GENERALMost studies in regional variations of medical and surgical procedure ratesor hospital use rates use a similar methodology. The number of occurrences of thehealth care event of interest in the population of a given region, is divided by thepopulation within the region who are at risk for that event, thus creating aregional incidence rate for the event. Ideally, these incidence rates arestandardized for age and sex to allow consistent comparison across geographicareas. The region may be defined at various geographic levels but usuallyrepresents an area in which the majority of the population use the same medical136care resources. Larger areas, such as provinces or states, usually have the bestmatch between population and medical resources but may be so large thatvariations seen at lower levels of aggregation are lost. Smaller areas, such ascounties, are more likely to have a relatively homogenous population and arelatively similar procedure rate within their boundaries, but may be suspect as abasis on which to match population and medical care resources.Early investigations into variations in the utilization of health care servicesexamined variations between countries (Pearson et al 1968, Bunker 1970). Thesestudies did not, however, take into account the differences between countries inrecording proceduresl; these differences may have accounted for at least part of thevariations found. Later investigations began to look at variations within countriesusing pre-existing administrative units such as provinces (Mindell, Vayda andCardillo, 1982) and counties (Stockwell and Vayda 1979). Other investigators havecreated regions, generally termed hospital service areas, based on population sizeand service by one hospital where the majority of residents receive their care(Barnes et al 1985, Wennberg and Gittelsohn 1982). More recently, researchershave 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 Tedeschi1985), and the elderly (Roos et al 1984). Without exception all these investigatorshave found marked variation in hospital utilization or in procedure rates.Two recent comprehensive reviews have provided an overview of theliterature 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 UnitedStates, primary, secondary and tertiary procedures are reported. Theoretically these differencesshould result in higher observed rates in the U.S., which is what McPherson found.137area analysis literature, restricting their review to those studies (59 in all) whichused the area as the unit of investigation. Sheps, Scrivens and Gait (1991)restricted their review to procedural variations studies which attempted to assessthe effect of specific factors on the variation observed. At the same time theybroadened the scope of the review to include investigations outside NorthAmerica. Both these reviews noted that the magnitude of observed variationswithin studies were relatively small, in general between one and three fold, rarelyexceeding five to six fold, although between studies there were often widevariations which may have been due to methodological differences. Variationstended to be greater for elective procedures than for procedures in which thesurgeon had less discretion, for smaller areas than for larger areas such as provinceor state, and for single years than for multiple years. These two latter findingslikely reflect the unstable annual rates which arise in small areas with smallpopulations but may also show real variations which are hidden at the larger levelof 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 thatfor a given procedure, a high-rate or low-rate area tends to persist as such forseveral years. Wennberg (1979) termed these persistent patterns "surgicalsignatures" and attributed them to the practice-pattern of the surgeon serving thatarea. In another paper, he and Gittelsohn (1982) note that the greatest variationsoccur for procedures performed for conditions for which there is no consensus inthe medical community as to treatment and, therefore, for which the individualphysician has considerable discretion as to which treatment he chooses.Support for the belief that physician practice-patterns affect resource usecomes from recent research into the financial effect of practice style on hospitalresource use. Feinglass, Martin and Sen (1991), controlling for patient health138status, used linear and logistic regression to analyse the relationships betweenphysicians clinical decisions and hospital charges and length of stay (LOS). Theyfound that attending physicians were significant predictors of the log of totalcharges and the log of LOS but not as significant predictors of untransformed totalcharges.Many of the studies in the variations literature have methodological flawswhich may affect their results. These flaws and other methodological issues willbe 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 aportion of the variation in rates. A large number of these explanatory variableshave been investigated but between studies results have been inconsistent andoften conflicting.The major factors believed to contribute to geographic variations may beclassified under two main categories, supply factors and community factors.Supply factors include such variables as bed supply, physician supply and medicalservices available to the region. Several studies have extended these variables toinclude those related to physician characteristics such as age, sex and place and/oryear of graduation. Community factors attempt to describe the region in terms ofmorbidity and mortality, socioeconomic status, or unemployment. For thesevariables, information about the individuals in the community is aggregated.Intuitively both supply and community variables would seem to affectprocedural use rates. Evans (1984) convincingly argues that, in health care, supplyinduces demand, and that Roemer's Law (a built bed is a full bed) applies, at theaggregate level, to surgeons as well as to beds. When more are available, more are139used. Similarly a large sociological literature on indicators of health status showsrelationships between morbidity and mortality and socio-economic status,education level, poverty, overcrowding, and race. Hay (1988) using data from the1978 Canada Health Survey found a direct positive relationship between SES(income. education and occupation) and health status. Of the three SES measuresused, income was consistently the best correlate of health status and occupationalstatus was the most inconsistent.In the variations literature, supply factors have been the most frequentlystudied as explanatory variables, although results were inconsistent both betweenstudies and according to the form of analysis. For example bed supply was testedwas tested for its effect on the cholecystectomy rate in 12 studies but was found tobe significant only four times (Sheps, Scrivens and Gait 1990). Of the studieslooking at community explanatory variables, very few have included mortality ormorbidity rates for the conditions for which the procedures are performed. Of thelarge number of community variables tested across studies, few have been testedmore than once or twice and few were found to be significant.Paul-Shaheen, Clark and Williams (1989) note that when correlationsbetween explanatory variables and use rates were run, the most highly correlatedvariable was some measure of bed supply. When multivariate regression modelswere used, the majority of studies found that a combination of community andsupply variables provided the best explanation of the observed variation.However, for surgical procedure variation the supply of beds, doctors and surgeonsexplained the greatest amount of variation observed. Sheps, Scrivens and Gait(1990) suggest that the large differences between studies in the magnitude ofvariation explained by independent variables, results from methodologicaldifferences rather than true relationships between the variables.140VARIATIONS IN CABS UTILIZATION RATESAs has been shown in Chapter Two, it is likely that physicians are usingconsiderable discretion in the treatment of CAD. Although CABS has been foundeffective only for a few conditions in coronary artery disease, and the efficacy ofPTCA has not been evaluated, the rates of both these procedures have risenexponentially sine they were first introduced. Therefore, it is to be expected thatboth CABS and PTCA rates will show some regional variation. In fact PTCA doesnot 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 amountof variation is acceptable is unclear 2 .Roos and Cageorge (1987) studied the growth and regional variation inCABS in Manitoba between 1978 and 1984, and compared trends in CABS to thosein cardiac valve surgery (a low growth procedure performed by the same surgeonswho perform CABS) and in total hip and total knee replacements (high growthprocedures which are less centralized than CABS). Seven regions, approximatingthose 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 per10,000 population; areas with high rates for CABS tended to have high rates forvalve surgery. This last finding would support Wennberg's concept of the"surgical signature" since the same surgeons perform both operations. Theorthopedic procedures showed less variation than the cardiac procedures and theauthors attributed this finding to the fact that hip and knee surgery, althoughcentralized, is more decentralized than CABS, thus providing more equal access to2 Most regional variation studies appear to assume that any variation between regions isunacceptable even though there may be legitimate (but usually unevaluated) reasons for suchvariations, e.g., regional variations in morbidity. The range of variation that can occur and stillbe consonent with high quality care for all patients in a region has never been established.141all Manitoba residents 3 . Surprisingly there were no trends or consistentrelationships between the regional rates of hospitalization for, and death from,acute myocardial infarction (AMI) and the regional CABS rates, although therewere some minor regional differences in the AMI rate.Roos and Cageorge conclude that organizational factors, i.e., centralizedversus decentralized surgery, and referral networks are critical in the developmentand maintenance of regional variations. While this is very likely true, thisconclusion is an assumption since it was not tested in the study. Despite thisshortcoming this study has contributed positively to the variations literature sinceit is one of the few studies to measure regional morbidity. The absence of anyassociation between AMI and CAD is extremely interesting. Either it indicates thatthe incidence of CABS is not related to need, or that AMI is not a good measure ofthe need for CABS. In Chapter Two it was noted that the incidence of CABS ishighest in the age group which has the highest incidence of chronic CAD, and thatCABS is performed for chronic, rather than acute, ischemic heart disease. Itappears likely that ischemic heart disease diagnoses other than AMI may prove abetter measure of CAD morbidity.Other studies which have looked at regional variations in CABS ratesinclude one from Ontario. Anderson and Lomas (1989) studied the effects ofregionalization on the age-adjusted surgery rates in 38 counties in southernOntario. These counties were served by eight referral centres in fivegeographically separate metropolitan areas. There was a three-fold difference inrates between the counties with the highest and lowest rates (9.4 versus 2.8 per10,000 population over age 20). There was no significant relationship between the3 One could also argue that decentralization would increase regional variation because morephysicians, and therefore more "practice styles", would be involved.142county rates and distance of the county from a referral centre'', but there was asignificant relationship between the county rate and the referral centre serving thecounty5 . In other words, the referral centres themselves explain more of thevariance in county rates than does distance from the centre. In order to excludeother county variables besides distance from the centre it would be useful to knowhow the counties were distributed geographically around their referral centres.Contiguous counties could share some potentially relevant factors such asethnicity, income and socio-economic status.Differences between institutions in recommending CABS to patients withsimilar clinical and angiographic characteristics, were found by Maynard et al(1986) in an analysis of the CASS Registry data. These authors found that across 15institutions the percentages of patients in the CASS Registry who wererecommended for CABS ranged from 35.2 percent to 73.2 percent. Even afteradjustment for patient characteristics there was a significant difference betweensites. These authors note that in most of the hospitals that they studied, thecardiologist was the dominant decision maker.The only other study which looked at regional variations in CABS was doneby Chassin et al (1986) and studied variations in the use of several medical andsurgical procedures in the medicare population in a number of selected areas inthe U.S. The variations in the rates of CABS between areas ranged from 7 to 13 per4 This was true whether distance was measured catagorically (i.e., county contains referralcentre, 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 repeatedfor each rule. The strict rule assigned to each centre the rates for counties for which the centresupplied at least 90 percent of the procedures done. Under this rule the referral centre explainedabout three-quarters of the variance. The majority rule assigned the rates for counties to thecentre which supplied 50 percent or more of the procedures in the county. Under the plurality rulethe county rate was assigned to the centre which provided the plurality of procedures in thecounty.14310,000 population. The generally higher CABS rates found in this study are likelydue to the older population studied.Chassin and colleagues have also been instrumental in carrying outresearch which crosses the boundary between studies into geographic variationsand those into quality of care. The methodology consists of using physician panelsto rate indicators for CABS as appropriate, inappropriate or equivocal 6 . Thesubjectivity present in any rating procedure was likely reduced in Chassin's studiesby the fact that panels of doctors were used and that the panels were provided witha review of the CABS literature prior to making their ratings. Retrospectivesamples of patients are then taken from geographic regions where per capita userates for the procedure have already been determined. Chart review is thencarried out and the appropriateness of the procedure is determined for eachpatient. The relationship between the proportion of appropriate procedures in aregion and the region's procedural incidence rate is then determined.Chassin et al (1987) studied the appropriateness of use of coronaryangiography and two other procedures provided to Medicare beneficiaries in threegeographic regions (covering several states) in the U.S. They found small butstatistically significant differences in appropriateness among the sites. Forangiography the site with the highest per capita rate had the lowest rate ofappropriateness, and the site with the lowest procedural rate had the highest rateof appropriateness. Across all sites inappropriate use accounted for 17 percent ofangiographies. The authors concluded that although the differences were in thedirection supporting their hypothesis, i.e., that there would be more inappropriate6 An indication was deemed appropriate for a procedure if the expected health benefits of theprocedure for that indication exceeded the negative consequences by a sufficiently wide marginthat the procedure was worth doing. An indication was deemed inappropriate when the risksexceeded the expected benefit. Equivocal ratings were those which fell in the middle of the 9-point rating score or those on which there was disagreement by the panel.144procedures in areas of high use, the differences in appropriateness could notexplain the large differences in overall rates.The above study was later repeated in one state using the county as the levelof analysis (Leape et al 1990). When all counties were included in the analysis theratio of high to low use was 12.1 and inappropriate use accounted for 28 percent ofthe variance in use rates. When the county with the highest use-rate (an outliersince its rate of 189 per 10,000 Medicare enrollees was substantially higher that thenext highest rate of 89 per 10,000) was removed from the analysis the high-low userate ratio was only 2.3 and inappropriate use accounted for only 12 percent of thevariance. The authors conclude that differences in the rates of use are not due tomore inappropriate use in high-use areas. In fact they rightly state that one of themajor findings in the study is that inappropriate use occurs in all areas whatevertheir use rate.The same group of researchers investigated the appropriateness of CABS ina random sample of patients from three hospitals during 1979, 1980 and 1982(Winslow et al 1988). Overall 56 percent of CABS were performed for appropriatereasons, 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 forinappropriate use was 6 to 23 percent. Although there were significant differencesin appropriateness scores by age the results showed slightly more appropriate usein the elderly. These authors point out that if procedures in the inappropriate andequivocal class were eliminated it would be possible to double the number ofappropriate 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 eachcountry rated indications for coronary angiography and CABS according to theirappropriateness. These ratings were then used to rate a retrospective sample of145patients who had angiography or CABS in the U.S. Using U.K. ratings, 42 percentand 60 percent of Medicare patients and non-medicare patients who receivedangiography, were inappropriate. The U.S. ratings found only 17 percent and 27percent respectively, inappropriate. For CABS patients 13 percent were deemedinappropriate by U.S. ratings and 35 percent by U.K ratings. The authors note thatif CABS were performed in the U.S. and the U.K. at the rates judged appropriateby the panelists, then the ratio would be US:UK 1.5:1. In fact the ratio is US:UK4.8:1. This ratio cannot be accounted for simply by the difference in physiciansbeliefs and, therefore must have some other explanation.In the above study Brook et al stated that differences in attitude to theindicators appeared to be a result of different interpretations of the literature.Differences between the groups occurred for those indicators for which there wasno 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 greatimportance to the degree of medical therapy and did not rate indicators asappropriate when the patient was not on maximal medical therapy. There werealso differences between panelists according to specialty. Surgeons on both panelsrated more indicators appropriate, while general practitioners (G.P.'s) tended torate more indications inappropriate than the other panelists did. This finding is inconflict with that of Young et al (1987) who used hypothetical patients to analyzeU.S. cardiologists and G.P.'s recommendations for angiography. In this study itwas found that cardiologists required a higher probability of CAD beforerecommending invasive testing.Few other explanatory variables have been tested in trying to account forregional variations in CABS rates, and none have been found significant. Withthe exception of the study by Roos and Cageorge, morbidity has not been tested asan explanatory variable; nor have other community variables such as SES.146In summary, the utilization rates of CABS have been shown to varybetween geographic regions and hospital sites by at least two-fold. Few explanatoryvariables have been tested; of those that were, the referral hospital used by theregion was found to be significant. More recent attempts to explain geographicalvariations by correlating regional rates with the percentage of appropriate andinappropriate procedures in the region, have shown a weakly significantrelationship which the authors believed was not strong enough to account for thevariations observed.ISSUES IN VARIATIONS RESEARCHAll the recent major reviews on regional variations (McPherson 1989, Paul-Shaheen, Clark and Williams 1989, Sheps, Scrivens and Gait 1990) discuss theissues 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 inpatientsurgery affecting the number of procedures counted, differential procedure coding,differences in protocols regarding whether primary, secondary or tertiary diagnosesor procedures are counted, differences in which procedure is coded as the primaryprocedure, and the use of a denominator which accurately estimates thepopulation at risk7 . Sheps, Scrivens and Gait (1990), believe that use of theincorrect denominator is a methodological flaw which is pervasive in theliterature. Use of the general population as a denominator will not severelydistort the rate estimates in large populations, because the number of individuals7 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 andthose women who have alrady had a hysterectomy, otherwise the rate will be spuriously low.147receiving the procedure is relatively small compared to the numbers in thepopulation. In small populations, however, where the relative difference is muchlarger, rate estimates may be severely distorted and will be inappropriately large.When the procedure under study involves organ removal, use of the generalpopulation as a denominator, will produce rate estimates which are too small.Other issues discussed by these authors and by Paul-Shaheen, Clark andWilliams, include the need for age standardization of the populations studied,since the rate for some procedures (including CABS) is highly related to age. Age-sex standardization is less often performed although many procedures (alsoincluding 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 largevariations found.Many studies reported in the above reviews assumed that large variationswere statistically significant, and failed to test for this. However, Diehr et al (1990)showed, through computer simulations, that chance variability in procedural ratesamong populations with the same underlying rate, was surprisingly high. Therewas more variability for low-incidence surgeries and for populations which weresmall, had readmissions or in which sub-groups were studied. One problem withsmall area analysis is that small populations may make rates unstable across timeand across areas. A few cases occurring in one year in a small population mayresult 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 thedegree to which these extreme rates are representative of the data, may showspuriously large variations 8 . This is especially likely in studies which estimate8 An example of this may be seen in Bayne's (1991) study of health in the Greater VancouverRegional Hospital District. Hospital caesarean section (C-section) rates (number of C-sectionsover number of deliveries per hospital) varied just over two-fold when all hospitals wereincluded (range 13 to 27) but only 1.28 fold when an outlier with a low rate was excluded (range 21to 27).148rates for a sub-section of a small population, as Roos et al (1981) did when studyingsurgical rates for the elderly in rural Manitoba.Diehr et al (1990) discuss the statistical methods generally used in variationsresearch. They suggest that null hypothesis 9 has rarely been tested because there isno information available about the distribution of rates in the null situation.Frequently used statistics, including the extremal quotient, the coefficient ofvariation and the systematic component of variation have no tables and so cannotbe used to test the null hypothesis. Diehr and colleagues suggest that the chi-square statistic may be appropriate to test the null hypothesis, provided that theexpected number of surgeries per county is at least five, readmissions are notpossible, and the surgery does not have a low incidence. They also note that chi-square may be underused because it does not apply directly to age-sex standardizedrates, and suggest that for standardized rates the Mantel-Haenszel approach orlogistic regression may be used. All the above exclusions apply to CABS, a surgeryof low incidence in which readmissions are possible and which is age-sex relatedso that age-sex standardized rates should be used.Another approach to determining whether there is excess variability amongrates is to regress the observed rates on some relevant covariates, such as thenumber of surgeons per capita. If all underlying rates are the same, there shouldbe no significant association between the rates and the regressors. In the abovepaper, Diehr et al note several problems with this method. Firstly, because there isusually only a single data point per region, outliers will tend to have a largeinfluence on the estimated regression coefficients and significance levels. Alsobecause the regressors are generally adjusted for population size (e.g., number of9 The null hypothesis in variations research , is that the underlying utilization rate is the samein all areas, (i.e., has a normal distribution with a common mean and standard deviation) andthat the observed differences between areas are simply due to random variation.149surgeons divided by the population of the region), the resulting variable iscorrelated with population size. There is, therefore, the strong probability thatsignificant associations are really due to correlation of the variables underinvestigation with a third variable, population size. Despite these caveats, Diehrand colleagues note that the regression approach is the only approach we have todetermine characteristics of counties with high rates.Another factor which may create spurious rate differentials is mobility; themovement of the population out of their area of residence to seek health care inanother area or jurisdiction. The smaller the size of area used as the unit ofanalysis, the more important it is that some adjustment be made for movement inand out of the area for receipt of health care services10 . Joffe, (1979) found thatwhen he regressed utilization rates, both with and without mobility adjustment,against bed supply and physician supply, adjustment for mobility showed astronger association between utilization and bed supply but a weaker associationbetween utilization and physician supply.McPherson (1989) noted that differing reporting procedures may lead to ratedifferentials. Sauter and Hughes (1983) found that even in jurisdictions whichreport only primary procedures, differences in the protocol for ordering operationswithin the patient's record may have a significant impact on reported rates. Theyconclude that surgical utilization statistics, and the inferences that can be drawnfrom them, vary considerably depending on the recording protocol employed, andsuggest that policy decisions based on utilization research should be tempered ifrecording protocols were not considered in the study.10 This particularly applies to those studies in utilization variation which use "hospital serviceareas" (i.e.' geographic areas in which the majority of residents receive care from a singlehospital) as the unit of analysis. However, it is also important to consider mobility in anysituation in which utilization by some area residents may not be detected because they crossborders for medical care. For example, B.C. patients receiving CABS in Alberta would not showup in B.C. data bases.150Sheps, Scrivens and Gait (1990) note that few studies appear to recognize theissue of the ecologic fallacy. This error may arise if the data on procedural use ratesand the data used to describe the community come from different sources.Relationships between use rates and the community variables may not be presentat the level of the individual. For example if high use rates are seen in areas ofhigh SES, it is not necessarily the people with high SES who have the higherfrequency of the procedure. To infer that high SES causes the high use rate wouldbe to make an ecological fallacy.The final issue that needs to be discussed is the issue of the clinicalimportance of variations in procedural rates, and the size of variation which iscause for concern. Many studies stress that, because there is no normative rate foreach procedure, one cannot tell whether low rates are too low or high rates toohigh (Chassin et al 1986, Wennberg, Freeman and Culp 1987) 11 . It follows,therefore, that one cannot know whether the "correct" rate even lies in theobserved high-low range. This uncertainty about whether under-utilization orover-utilization exist, together with the uncertainty about how utilization relatesto community variables, makes policy decisions very difficult.The approach used by Winslow et al (1988), Chassin et al (1986) and othersin assessing the appropriateness of procedures in individual patients, to someextent 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 rateareas, although the appropriateness of the procedure did not account for all the11 Though this is not to say that a correct rate (or range) does not exist. Such a rate for a givenpopulation would be one at which there were no inappropriate CABS and at which allindividuals for whom the procedure is appropriate (given a condition for which CABS is moreeffective than medical therapy or angioplasty), would receive it. Such a rate could be estimatedfor a given population from the incidence of conditions for which CABS is proven to be the mosteffective treatment, adjusted for the demography of the population. Use of present rates for"appropriate" CABS would likerly underestimate the "correct" rate since those patients whorequire the procedure but do not receive it would not be included.151variation seen. In this case it would seem logical to reduce the number ofinappropriate procedures rather than simply trying to reduce rates that are higherthan those in surrounding areas . That is not to say that rate variations per se areunimportant. If the rate is not related to the underlying "need" of thepopulation 12 (i.e., if there is no relationship) then clearly something is wrong.Measuring "need" requires good morbidity estimates plus an estimate of the ratesof other appropriate therapies for the condition for which the procedure isperformed. Both these are generally unavailable.Another issue which is generally ignored in the literature, is that oflegitimate unmet need. The individuals who require the procedure but do notreceive 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 futureresearch and caution that altering supply factors without concern forappropriateness may do more harm than good.In summary, it appears that there are a number of artifactual andmethodological issues which have only recently been recognized in regionalvariations research, and which may have accounted for some of the largevariations, and some of the inconsistencies in explanatory variables, seen in theliterature. Future research must include morbidity and/or mortality estimates,and should include consideration of patient mobility and the protocol for orderingoperations, appropriate age-sex standardization of rates and testing of the nullhypothesis.12 It is not clear which way the relationship would go. A negative relationship betweenmorbidity and a procedure may indicate that the procedure is either curing the disease inquestion (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 thateither there is little underlying need (low rate and low morbidity) or that a high degree of needis causing a high rate (high rate and high morbidity).152CONCLUSIONOver twenty years of work into procedural variations has documented thatdifferences in procedural rates do exist among areas and between sub-groups in thepopulation, although some of the variations found likely result from artifactualand methodological sources. However, the cause of these variations and theirimportance is still unclear. Consequently, the implications for policy areuncertain. It is clear, however, that despite a documented relationship betweensupply variables and procedural rates, alteration of supply without considerationof other community factors may cause more harm than good. Future researchshould include combinations of supply and community factors, pay attention tomethodological issues and include some consideration of procedures that shouldhave been done but were not.153REFERENCES:Anderson, G.M. and J. Lomas, 1989. Regionalization of coronary artery bypasssurgery: effects on access. Medical Care 27 (March):288-296.Barnes, B. et al, 1985. Report on variation in rates of surgical services in theCommonwealth of Massachusetts. TAMA 254 (July):371-375.Brook, R.H. et al, 1988. Diagnosis and treatment of coronary disease: comparison ofdoctors' attitudes in the USA and the UK. The Lancet (April):750-753.Bunker, J.P., 1970. Surgical manpower. A comparison of operations and surgeonsin 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 bythe medicare population. N Eng T Med 314 (January):285-290.Chassin, M.R. et al, 1987. Does inappropriate use explain geographic variations inthe 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 small-area 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 physicianpractice 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 theCanada 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 theuse of health care services. TAMA 263 (February):669-672.Maynard, C. et al, 1986. Institutional differences in therapeutic decision making inthe Coronary Artery Surgery Study (CASS). Medical Decision Making  6(July-September):127-135.McPherson, K., 1989. International differences in medical care practices. HealthCare Financing Review (Annual Supplement):21-32.154Mindell, W.R., E. Vayda and B. Cardillo, 1982. Ten year trends in Canada forselected operations. CMA Journal 127:23-27.Paul-Shaheen, P., J. Clark, and D. Williams, 1987. Small area analysis: a review andanalysis of the North American literature. T Health Politics, Policy and Law12 (Winter):741-807.Pearson, R.J.C. et al, 1968. Hospital caseloads in Liverpool, New England andUppsala. The Lancet 2:559-566.Roos N.P. and L.L. Roos, 1981. High and low surgical rates. Risk factors for arearesidents. ATPH 74(4):313-314.Roos, N.P., E. Shapiro and L.L. Roos, 1984. Aging and the demand for healthservices: 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 methodologicconsiderations. Medical Care 11 (March):370-377.Sheps, S., S. Scrivens and J. Gait. Perceptions and realities: medical and surgicalprocedure 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 onmedicaid surgery rates in Massachusetts. ATPH 71 (April):370-375.Stockwell, H. and E. Vayda, 1979. Variations in Surgery in Ontario. Medical CareXVII (April):390-396.Tedeschi, P.J., R.A. Wolfe and J.R. Griffith, 1990. Micro-area variation in hospitaluse. 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 smallareas. Scientific American 246:120-131.Wennberg, J.E., J.L. Freeman and J. Culp, 1987. Are hospital services rationed inNew 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):263-269.155Winslow, R.L. et al, 1988. The appropriateness of performing coronary arterybypass surgery. JAMA 260 (July):505-509.Young, M.J. et al, 1987. Do cardiologist have higher thresholds for recommendingcoronary arteriography than family physicians? Health Services Research 22(December):623-635.156CHAPTER FOURRATIONALE AND METHODOLOGYRATIONALEThe previous chapters show that the incidence of revascularizationprocedures has risen dramatically since CABS was first introduced in the late1960's. This increase has occurred despite the fact that CABS has been provenefficacious in increasing survival in only a limited number of conditions and theefficacy of PTCA, in comparison to non-invasive treatment, has not beenevaluated. Moreover, the biggest increase in the incidence of CABS has occurredin the elderly; a group that has never been evaluated in an RCT but which hasbeen shown, in observational studies, to have a higher mortality and morbidityfrom CABS than occurs in younger patients.Despite recent government initiatives in B.C. to reduce waiting lists andprovide greater access to CABS by opening another unit, there is no evidence toshow that CABS is being used appropriately in this Province. Identification of age-and sex-specific rates of CABS in B.C. and its regions, can help to document thediffusion of the procedure in the Province and can provide the government withthe data needed to plan, allocate resources to, and evaluate revascularizationprograms. Identification of regional variations in rates and their contributoryfactors will help determine whether access to these procedures may be a problemin some areas and the approaches that may help to improve access.157Questions:The questions addressed in this study are:i) What are the age and sex specific trends in CABS in B.C. between 1979and 1988?ii) What are the age and sex trends in PTCA in B.C. between 1987 and1988?iii) What are the trends in the incidence of comorbid conditions in theCABS population between 1979 and 1988?iv) What are the regional trends in CABS between 1979 and 1988?v) Is there any significant variation in the age-sex adjusted CABS ratesamong the small areas (school districts) in B.C.?vi) Is there any significant variation in the age-sex and morbidity adjustedCABS rates among the small areas in B.C.?vii) How much of the variation found in (iv) and (v) is explained by theindependent variables (described below)?METHODSStudy Design:Questions (i) - (iv) were addressed by means of a retrospective descriptivestudy. Revascularization procedures performed in B.C. during the relevant timeperiod were identified and the characteristics of the patients described. Annualpopulation-based rates were calculated for each region (school district). These rateswere then used as the dependent variable in a poisson regression to answerquestion (v); was there more regional variation than would be expected bychance? The second part of the study, was an ecological analysis in which theannual age-sex adjusted CABS rates per school district were regressed on districtsocio-economic and health-services characteristics of each school district.158Independent 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 schooldistrict or far school district, and was coded from 0-2 respectively. In cases where acardiologist was resident in an adjacent school district but there was no road accessbetween school districts in that year, the distance was coded as far school district.Similarly, distances between school districts which involved a ferry crossing werelabeled 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 surgerycentre (DSCEN) - measured as for cardiologists.Income (INC) - average annual income per census family measured in tensof thousands of dollars and entered as a continuous variable with $30,000 as thereference level. 1981 census data was used for 1983-1985 and 1986 census data for1986-1988.Employment rate (EMPRAT) - proportion of the workforce (aged 15 andover) who were employed. Entered as a continuous variable with 0.7 as thereference level. Census data was used as for income.Graduation rate (GRADRAT) - proportion of the 15-19 year-old populationwho graduated from high school in a given year. Entered as a continuousvariable. Census data was used as for income.159Data Sources:Data for all patients who received coronary artery bypass surgery (CCP 1codes 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 archivedrecords of the Hospital Morbidity Database. The data in this database, maintainedby 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 enteringand validating the data. After that year, the hospitals sent their data to commercialrecords institutes.Population figures for each Local Health Area (LHA) in B.C. were obtainedfrom the Planning and Statistics Division in the Ministry of Finance andCorporate Affairs. Census data were obtained for the years 1981 and 1986 andpopulation estimates for the other years for 1979 through 1988. Although it wasoriginally intended to use LHA's as the unit of analysis, the patient's location onthe database was entered as school district until 1982, thus school districts wereused 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 andStatistics Division.The Planning and Statistics Division also provided 1981 and 1986 censusdata for employment and income in the LHA's. The annual numbers of studentsgraduating from high school in each school district, between 1983 and 1988, wereobtained from the B.C. Ministry of Education. The location of cardiologists and1 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 usedin conjunction with the International Classification of Diseases (ICD-9). The Hospital MedicalRecords Institute (HMRI) re-codes coded ICD-9CM procedures into CCP codes with the result thatprocedures in the Hospital Morbidity Database are coded using CCP codes.160internists in B.C. between 1983 and 1988 was obtained from the annual editions ofthe Physicians and Surgeons directory published by the B.C. College of Physiciansand Surgeons.Throughout most of the province, LHA's and school districts have identicalboundaries. 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 theHealth Planning Database, Policy, Planning and Legislation Division in theMinistry of Health. In addition data from school districts 92 (Nishga) and 94(Telegraph Creek), which have very small populations, were aggregated withthose from school district 88 (Terrace).Information on the patients receiving CABS or angioplasty in Alberta wasobtained from the three hospitals involved: Holy Cross Hospital, FoothillsHospital and the University of Alberta Hospitals. The total numbers of B.C.residents receiving CABS or PTCA out-of-province between 1979 and 1989 wereobtained from Data Systems, Health Information Division, Health and WelfareCanada.Data were analyzed using SPSS on the mainframe computer at U.B.C.Regression analysis was performed using SAS Proc Logistic procedure on an IBMpersonal computer.Study Population and Analysis:1. Descriptive Analysis:Included in the analysis were all patients over the age of 20 years whoreceived isolated CABS (i.e., CABS not accompanied by valve replacement or othercardiac surgery) or angioplasty in B.C. between April 1979 and April 1989. Patientsbelow the age of 20 years were dropped from analysis on the basis that they wereprobably not representative of the usual population of CABS recipients. Patients161who had surgery at hospitals other than the cardiac centres were included in theanalysis although they likely represent miscoding. Cases with missing postal codeswere assumed, on advice from Data Support, Hospital Programs, to have comefrom out of Province and were included in the analysis under that category. Priorto 1983, cases who had more than one CABS (4811-4815 or 4819) coded for thesame admission were assumed to have had a single operation 2 .Diabetes and COPD were chosen as indicators of co-morbidity because theyare relatively stable over time3 . In 1983, the number of diagnostic fields in thedatabase changed from two to 16 resulting in an apparent rise in co-morbidity inthat year4 .Annual standardized incidence ratios (SIRs) for CABS were derived for eachschool district as follows. Patients were categorized into twelve 5-year age groupsbetween 20 and 75plus, and the number of procedures performed on B.C. patientsin each of the 24 age-sex categories was divided by the total 1979-1988 population ineach age-sex category to give an age-sex specific ten year rate. These rates were thenused to calculate the annual expected numbers of CABS per school district by2 In 1983 the format of the Hospital Morbidity Database was changed to include, among others, thedate 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 thatother diseases, e.g., hypertension, were subject to changes in diagnosis (e.g. what b.p. levelindicates 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 wouldresult in an apparent rise in co-morbidity. At the same time that the number of diagnostic fieldsincreased 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" wasanother important diagnosis. In 1983 the first diagnosis became the 'principle diagnosis' which isdefined as "the diagnosis most responsible for the patient's stay in the institution." The next mostimportant diagnosis the 'primary diagnosis' describes "another important condition of the patientwhich usually has a significant influence on the patient's length of stay". The 'secondarydiagnosis' "describes a condition for which the patient may (or may not) have received treatmentbut did not significantly contribute to the patient's length of stay in hospital". Other diagnosticfields include admitting diagnosis, complications arising in hospital, E-codes, morphology codesand transfer diagnoses (Ministry of Health, 1985).162multiplying the annual population in each age-sex group per school district by theage-sex specific rate. Observed and expected CABS for each age-sex group perschool district per year were then summed to give an annual observed andexpected figure per school district. Standardized incidence ratios (observedCABS/expected CABS) were then calculated for each school district per year. TheSIRs were then multiplied by the overall Provincial CABS rate (total number ofprocedures 1979-1988/total population 1979-1988) x 10,000 to give the annual age-sex adjusted rate for each school district per 10,000 population. 5It proved impossible to accurately determine the numbers of angioplastiesper 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 arteryobstruction), which includes endarterectomy and gas endarterectomy among otherprocedures. Following the introduction of angioplasty codes (CCP 4801-4805), onecentre did not use the new code but continued to code angioplasty under the CCP4800 code. Therefore, regional SIRs were not calculated for angioplasty.Patient residence in the Alberta data was identified by means of three or six-digit postal code. Because six-digit codes were required in order to identify mostschool districts, and because 1988 was the first year in which all Alberta centrescollected six-digit codes, B.C. patients having surgery in Alberta could not beincluded in the annual school district figures. To determine how mobility affectedthe incidence rates in school districts sending patients to Alberta, the 1988 SIRswere calculated both excluding and including the Alberta data.2. Regression Analysis:To determine whether significant regional variation existed and, if so, therelative contribution that each independent variable made to the variation in area5 This rate is provided to allow comparison with other studies, most of which use procedure ratesrather than SIRs.163rates, Poisson regression was carried out using a number of different models and1983 to 1988 data.Poisson regression was chosen because, like weighted least squaresregression, it assigns weights according to population size but has the addedadvantage of assigning weight even when the observed count is zero. In mostyears covered by this study there were some school districts which did not haveany 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 regressionwas repeated using the age-sex adjusted CABS rate also adjusted for the CADmorbidity in each school district (Table 15B) as the dependent variable. Because itwas not known which diagnosis of CAD would best indicate those patients whoform the "potential CABS population", several diagnoses were tried to see whichgave the best fit. These diagnoses were:i. Acute myocardial infarction (ICD-9 410 only). This diagnosis was usedfor those patients who had no other diagnosis of CAD and is an indicatorof the incidence of CAD in the school districts. It is obviously not a trueincidence because patients who die before reaching hospital, or thosehaving a 'silent' MI, are not included.ii. Chronic ischemic heart disease or unstable angina but not with AMI (ICD-9 411-414 or 4292, but not 410).iii. Chronic ischemic heart disease or unstable angina including those whoalso had a diagnosis of AMI but not those with AMI only (ICD-9 411-411or 4292 plus 410). This category included all those with chronic CADwhether or not they had had an AMI.iv. All cases (ICD-9 410-414, 4292), i.e., (i) plus (iii).164For each of the above diagnosis groups, a fully saturated "perfect model"was run and was compared to the null model; the model with the leastunexplained variance was assumed to be the one for which the CAD diagnosis bestindicated the potential CABS population. Pearson-product-moment correlationsbetween the SIRs for CABS and those for the CAD diagnoses were also calculated,and supported this finding. The adjustment for CAD was accomplished by usingthe following outcome variable:CABS/(Expected CABS x CAD/Expected CAD x 100,000).In calculating the annual morbidity rates per school district all patientsadmitted to hospital in B.C., for each fiscal year from 1983 to 1988, with a diagnosisof ischemic heart disease (ICD-9 410-414, 4292) were identified. Within each yearall non-B.C. residents, readmissions and patients receiving CABS in that year wereremoved. For each diagnosis group (outlined above) expected numbers ofadmissions per school district per year were calculated, using the same method aswas 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 variableswere regressed on the crude CABS rate using three different models (see Table15C). For these regressions, school districts were aggregated into four regions toreduce the otherwise unmanageable number of possible combinations of age-sex-year-region factors; Region was entered into the model as a categorical variable.165TABLE 14REGRESSION MODELSA. 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 + interactionsB. Model used to determine effect of independent variables on CABS rate afteradjustment 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 + B4distancefrom internist + B5income + B6employment rate + B7graduation rate+ interactionsC.^Models to determine age-sex-time-region interactions i) CABS/POP = B0 + Biage + B2sex + B3time + interactionsii) CABS/POP = B0 + Bi age + B2sex + B3time + interactions +B4Regionl + B5Region2 + B6Region3iii) CABS/POP = B0 + Bi age + B2sex + B3time + interactions +B4Regionl + B5Region2 + B6Region3 + interactions of age, sex andtime with regions*Expected CABS x 100,000. The factor of 100,000 was used to force Proc Logistic into a poissonregression.**Expected CABS***Expected CAD166CHAPTER FIVEREVASCULARIZATION IN BRITISH COLUMBIA 1979-1988 RESULTSCORONARY ARTERY BYPASS SURGERYCharacteristics 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 thegrowth of CABS follows that described by Peters et al (1990) for all of Canada; therewas an increase between 1981 and 1983 followed by a decline over the next twoyears and then by a more moderate increase. The percent of annual casesperformed on non-residents of B.C. increased from 0.9 to 1.7 percent over the tenyear period, while the percentage of women receiving CABS remained around 20percent. The mean age of the women was consistently 2-3 years higher than thatfor the men although the mean age for both groups rose by approximately fiveyears (from 57.4 to 62.9 for the whole population) over the study period. Theseresults are shown in Tables 16 - 18 in Appendix A.The Provincial annual crude and age-sex adjusted rates and thecorresponding SIR's are shown in Table 19. The steady rise in rate over the 10-year period, depicted in Figure 2, indicates that the rate of procedures is increasingfaster than population growth. Despite annual fluctuations in rate there is anupward 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 aremarkedly lower than those for males but the highest rates for both males andfemales are seen in the 65-69 age group. This result differs from other studies inthe literature (Peters et al 1990, Gillum 1987) where the highest rates were seen inthe 55-64 age group.167FIGURE 2ANNUAL AGE-SEX ADJUSTED CABS RATE PER 10,000 POPULATIONB.C. 1979-19887.006.506.005.505.00 4.50 1^1^i^i^1^i^i^1^179^80^81^82^83^84^85^86^87^88YEARTABLE 19ANNUAL STANDARDIZED INCIDENCE RATIOS AND CABS RATESB.C. 1979-1988Year Crude Rate per10,000*Age-sex adjustedrate per 10,000**StandardizedIncidence Ratio1979 5.20 5.19 0.911980 5.01 5.03 0.881981 4.96 5.01 0.881982 5.29 5.35 0.941983 5.96 6.02 1.061984 5.78 5.83 1.021985 5.63 5.65 0.991986 5.96 5.91 1.041987 6.14 6.06 1.061988 6.79 6.65 1.17Overall rate 5.70 -*All rates are per population > 20 years of age**SIR x overall ten-year rateTABLE 21ANNUAL DISTRIBUTION OF ISOLATED CABS BY AGEGROUPB.C. 1979-1988AgeGroupAnnual Numbers of Cabs1979 1980^1981 1982 1983 1984 1985^1986 1987 198820-24 0 0^0 0 1 0 0^0 0 025-29 0 2 0 0 3 0 1 2 1 130-34 8 4^6 4 7 2 4^3 1 1035-39 16 17 18 24 31 16 19 16 11 1440-44 47 43^40 51 49 38 40^43 36 3245-49 100 100 89 83 71 94 69 72 72 9550-54 156 140^157 161 179 144 117^117 148 15455-59 209---,_ 203 202 193 225 203 190 198 198 21260-64 179 '---205 -217 219 -279 267 277264 255 26065-69 160 151^157 207 227 241 --------- -275232 320 36570-74 47 57 62 77 118 148 166^203 211 23675-79 6 7^11 28 27 39 52 70 64 10080-84 2 1 2 3 3 2 1^8 7 1785-89 1 1^1 0 0 0 0 0 1 090+ 0 0 0 0 0 0 0^0 0 0Total 931 931^962 1050 1220 1194 1181^1271 1325 1496*Line indicates modal age group169The distribution of cases by age group (Table 20) shows changes in the modalage-group, followed by plateaus, between 1979 - 1980 and 1985-1986. However,when these figures are expressed as population rates the mode fluctuates betweenage groups 60-64 and 65-69 until 1984 when it remains in the latter group. Itappears that demographic change does not have an immediate effect on the CABSrate but is likely mediated by established practice patterns. Figure 3 depicts theannual age-specific use rates (unadjusted for sex) for three time periods -1979, 1984and 1988. Over the ten year period, utilization decreased slightly for patientsunder the age of 60, but increased markedly for those over 60, with the greatestincrease seen in the over-65's. These age-use curves depict a pattern similar tothat seen when there is a cohort effect. This possibility would be worth exploringin future research.When expressed as a percentage of total CABS, procedures in the over-65population have increased from just under 25 percent of the procedures in 1979 toalmost 50 percent in 1988. Their rate increased almost two-fold, from 3.32 to 6.38per 10,000. However, when the over-65 population is broken down intosubgroups, 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 (Table21, Appendix A). When the age-specific sex-adjusted rates for subgroups of theover 65 population are considered (Table 22, Appendix A) it is seen that althoughall sub-groups have increased their incidence rate, the biggest increase has occurredin the 75+ year olds, where the rate has increased almost nine-fold, althoughstarting 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 alsoappears to be sicker; at least as far as co-morbid conditions are concerned. Over the10-year period the number of patients with other diseases (diabetes and COPD)whoI I I I25 -20 -15 -RATE10 -50 I^170FIGURE 3AGE-USE CURVESB.C. 1979, 1984, 1988.30 -20-49^50-54^55-59 60-64 65-69^70-74^75+AGE GROUP10-- 1979 -43- 1984 -*-- 1988171underwent CABS rose from 30 to 247 (Table 24, Appendix A). This represents arise from 3.2 percent to 16.51 percent of the annual cases. Some of this apparentrise can be attributed to the 1983 increase of the number of diagnostic fields (from 2to 16) in the morbidity database. The majority of these patients with co-morbidityhad diabetes. Between 1979 and 1988, COPD increased from 1 to 4.3 percent, anddiabetes from 3.2 to 12 percent, of annual CABS. No patients were found to haveboth diseases.Table 25 (Appendix A) shows the distribution of co-morbidity by age groupfor CABS. Although the number of patients with co-morbidity in the older agegroups is small, when expressed as a percentage of total cases they indicate a higherpercentage of co-morbid conditions in the over-65 age group with the highest ratein the 80-84 year olds. It appears likely that the increase in co-morbidity hasoccurred because of the increase in the age of the CABS population.The above data do not include B.C. residents who received their CABSprocedure outside British Columbia. Figures supplied by the Federal Ministry ofHealth 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 ofthe total number of B.C. residents receiving CABS). Data obtained from AlbertaHospitals performing CABS show that between 1983 and 1988, 66 men and 10women (15 percent of all cases) received CABS in Alberta although the 1988 figureprovided by the Federal Ministry does not agree with the Alberta data. The suddenincrease (from 3 to 28) in the number of cases leaving for Alberta in 1987 is likelydue 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. residentsreceiving CABS in Alberta.172Re-operations: Re-operations after 1983 were identified using the scrambled MSP numberand 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 atVancouver General hospital in 1991 (121 out of 1100 CABS) it was evident that thenumbers of re-operations had been grossly underestimated, even allowing for thefact 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 unreliabilityof the eight-digit MSP numbers which were used to identify individuals who wereadmitted 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 oncea number becomes dormant it may be used again for another individual after atwo-year period. Because sex and birthdate were also used to identify re-operationsand re-admissions, it is unlikely that either was overestimated but both re-operations and re-admissions were likely underestimated. In the case of re-admissions it is unlikely that there was any bias by school district so that themorbidity 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 todevelop reliable figures for the numbers of patients receiving mammary arteryimplants or for the number of vessels bypassed. For example, if two coronaryarteries were bypassed, one with a saphenous graft and one with an internalmammary implant, Centres 2 and 3 would code the procedure as bypass of oneI Approximately 3000-4000 CAD cases annually were dropped from the analysis becasuse they representedreadmissions in that year.173coronary artery plus a single internal mammary implant (CCP codes 4812 and4816); Centre 1 would code it as a double coronary bypass only (4813). In this lattercentre, internal mammary implants (CCP codes 4816 or 4817) were only coded incases 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 sameway or when the changeover might have occurred.The next section addresses the issue of "place"; i.e., where the patientsreceiving CABS came from and where they went to receive their surgery. Thissection 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 andschool district. Visual inspection of this table shows areas of greater activityaround Nelson (SD 7), the Prince George area (SD 57), the Okanagan (SD 14-15, 22-24), the Victoria and mid-to-north Island area (SD 61,62 63, 65, 68, 69,70), andVancouver and the Lower Mainland (SD 36 - 44). These areas are, of course, moredensely populated than the rest of the Province. Even so, the regularity withwhich substantial numbers of cases occur in these areas is striking.The annual number of cases within the above areas do not show anydiscernible pattern overall. There is a tendency for the number of cases per schooldistrict to increase over time in Vancouver and the Lower mainland, while thereverse is true in Victoria and some of the near-by school districts. Sudden andsustained rises in annual numbers in a school district could not be related to theintroduction of a specialist into the area except in Richmond (SD 38) where thearrival of a cardiologist in 1986 was followed in the subsequent year by a 56 percentincrease in the annual cases. It is possible that increases in other school districts174may have coincided with a change in personnel (as opposed to the introduction ofa specialist) but this was not studied.The annual numbers of CABS procedures performed on the residents ofeach school district, were used to calculate annual standardized incidence ratios(SIRs) for each school district. The most striking aspect of the these SIRs, shownin Table 29 (Appendix A), are their extreme variability within school districts.Although there are school districts which maintain a fairly steady ratio across theyears, 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 toconsistently high ratios, particularly in later years, can be seen in southernVancouver 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 VancouverIsland) are centres for cardiology and it could be expected that closer proximity to acardiologist may lead to a higher CABS rate.The variability in observed CABS within school districts over time, is inpart due to population size. When the coefficient of variation (the ratio of thestandard deviation of observed CABS to the mean) for each school district over theten year period is graphed against mean population size (Figure 4), those schooldistricts with populations under 10,000 show the greatest variability, and thosewith populations over 20,000 show the least. When the range of SIRs within eachschool 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 whichshow variations of over two- to four-fold (SD 68 and 23) so there are other factorsbesides population size which are exerting an influence on rate variations.175FIGURE 4SCAT^ IhRGRAPH OF COEFFICIENT OF VARIATION OF OBSERVED CABS IN SCHOOLDISTRICT BY SCHOOL DISTRICT POPULATION2.00 —^■1.80 —• •1.60 —2 1.40 —20 1.20—■ orwzn 1.00 -^ IIIm^•^• ■6 0.80 4- %NE ■1— IN ■^■co 0.60 —. •^■ ^IN■• •^• UmNO • II •^OM • • is■^■ ■■I I I15000 10000 15000 20000POPULATION2.00 —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 350000POPULATION0.40 —0.20 —0.000•3.503.00——■ ■ ■cc(i)Lij2.502.00— •—^■(5cc 1.50 —1.00 —0.50 — ■ •0.0087 49 13 16 843.50 —3.00 —2.50 —Coccc7)w2.00 — ■(54 1.501.00 —0.50 —■•0.00 *tii I•■■■•■111•11+1 11■■■■29 26 10 81 17 76 66 50 18 4 21 55 32 12SCHOOL DISTRICT•I I^I••■■I^I•■176FIGURE 5RANGE OF STANDARDIZED INCIDENCE RATIOS FOR ISOLATED CABS BY SCHOOLDISTRICT ORDERED BY POPULATION SIZEB.C. 1979-198830 77 31 19 3 64 9 80 54 85 56 86 48 14 1 52 46 47 2SCHOOL DISTRICT3.50 —3.00 —2.50 —2.00 —1.50 —■1.00 —0.500.00—• 11i^.^i7^28 11FIGURE 5 (contd.)iI I 1^I^I^I^1^I^I^I^I^I^I^I69 60 59 75 88 72 89 70 15 27 65 71 62 63 42177SCHOOL DISTRICT3.50 —3.00 —2.50 —2.00 —01.00 —.50 —1.50 —10.0022 45 33 40 34 35 37 68 24 57 23 38 43 44 41 36 61 39SCHOOL DISTRICT178Table 30 (Appendix A) shows the observed and expected numbers of CABS, thecoefficient of variation and the overall external quotient (largest SIR/smallest SIR)for the study period. It is interesting that school districts showing a wide variationin SIRs across the study period may, overall, have either approximately equalobserved and expected cases (e.g., SD 48 and 75) or a wide disparity betweenobserved and expected cases (e.g., SD 11 and 60). The reverse is also true; schooldistricts with a relatively small variation over time may have widely disparateobserved 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 thanexpected over the ten-year period. For example, Golden on the Alberta border hadonly two cases receiving CABS in B.C. and none in Alberta, unless some of theAlberta cases which could not be assigned to a region came from Golden.Among school districts the annul external quotient ranged from nine- tothirty-fold. When school districts with populations under ten thousand plusthose sending ten percent, or more, of overall cases to Alberta were excluded, therange was two- to twenty-two-fold. When data were aggregated into two timeperiods, 1979-1984 and 1985-1988 (Table 31, Appendix A ), there was an overtwelve-fold variation for the first period and a thirteen-fold variation for thesecond period when all school districts were included. When school districts withsmall populations or mobility to Alberta were excluded as above, the annualvariations among school districts fell to three-fold and two-fold for the first andsecond periods respectively. Aggregation of the data over the whole study periodshowed an eighteen-fold variation across school districts overall and a two-folddifference when districts were excluded as above, with Trail (SD 11) having thelowest SIR at 0.69 and Victoria (SD 61) and Sooke (SD 62) sharing the highest at1.62.179The above data show that, even when estimated conservatively, there is atwo- to nine-fold variation within school districts over the ten years of the studyand at least a two-fold variation across districts. It also appears that the size of thevariations in SIRs within school districts over time are not necessarily indicativeof more, or less, CABS being performed then are expected over that period.Consequently, no clear pattern emerges to help clarify the issue of the importanceof geographical variations in procedural rates.Despite the variability within and across school districts there is a clear trendto increased utilization of CABS across the Province during the study period. Overthe years there are fewer areas with extremely low ratios (or no cases) and theprocedure appears to have spread more into the north. The southeastern border ofthe Province consistently shares low ratios but these school districts appear toreceive the procedure mainly in Alberta.Calculation of the 1988 SIRs both including and excluding patients leavingthe Province for Alberta, shows marked differences in some SIRs for thesoutheastern and northeastern school districts (Table 32 Appendix A). In 1988, 47.6percent of all CABS patients living in the seven school districts on the Albertaborder went to Alberta for their surgery as did 37.5 percent of patients living in theEast 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 between1983 and 1988) overall, is of interest because this school district is closer toVancouver than to any of the Alberta centres. It may be that these cases wereoriginal Alberta residents who have retired to the Okanagan or that thecardiologists in Kelowna have professional links to Alberta.180Referral PatternsThe geographic distribution of school districts is shown in Map A. Thecentres 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 mostclearly defined catchment area receiving 79 percent of its cases from 5 schooldistricts, all on Vancouver Island. Centre 1 receives 50 percent of its cases fromVancouver and suburbs (SD 38, 39, 40, 41, 42, and 43) which represents 68 percentof the cases from the area. It receives a further 33 percent of cases from the FraserValley (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 diversifiedcatchment area. One quarter of its cases come from Vancouver and suburbs and afurther quarter from the Interior (53 percent of cases from this area). Seventeenpercent of cases come from the North Vancouver (SD 44), Sechelt (SD 46), andPowell River (SD 47) areas (82 percent of these areas' cases) with a further 9 percentfrom the north (83 percent of cases from the area) and 8 percent from theKootenays (75 percent of cases from the area who received surgery in B.C.). Thereare definite referral patterns governing the utilization of centres by school districtsand these patterns can be clearly seen in Map B.Within each centre, differences over time in the numbers and type ofpopulation served may indicate changes in referral patterns, or may reflectchanging attitudes towards the procedure within that centre. The number ofCABS cases served by each centre have diverged over the years (Table 34). In 1979all centres were serving approximately the same number of cases. Centre 3 has notchanged while Centre 1 has doubled the number of cases and Centre 2 hasincreased the number 1.5 fold. Centre 3 has fallen from serving approximately 33percent of all cases in 1979 to only 20 percent in 1988, while Centre 1 has increased181its share from 34 to 45 percent. Comparison of the annual CABS cases per centrewith 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 inCentre 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 isfunded as one program, the change in proportion of CABS to other open-heartsurgery in Centres 2 and 3 must either represent a change in referral patterns or achange in policy within the hospitals.The annual mean ages by centre (Table 36, Appendix A) show that Centre 3consistently cares for a slightly older population than the other centres. Althougholder, this population does not appear to have more concomitant diseases than thepopulation served by the other centres; 9.1 percent of all cases in Centre 3 had co-morbidity compared with 9.6 percent in Centre 2 and 8.9 percent in Centre 1 (Table37 Appendix A).TABLE 34CORONARY ARTERY BYPASS SURGERY PER CENTRE PER YEARB.C. 1979-1988Year Centre 1 Centre 2TotalCentre 3 Annual1979 320 307 304 9311980 330 310 291 9311981 338 324 279 9611982 413 334 301 10481983 528 351 341 12201984 463 434 296 11931985 544 387 250 11811986 547 420 302 12691987 555 455 313 13231988 678 516 302 1496Total 4736 3838 2979 115532 This data was obtained from the Medical Consultation Branch, Institutional Services Division in the B.C.Ministry of Health.182TABLE 35PERCENT OF OPEN HEART SURGERY DEVOTED TO CABS ANNUALLY BY CENTREB.C. 1983-1988Year Centre 1 Centre 2 Centre 3Open CABS % Open CABS % Open CABS %Heart Total Heart Total Heart Total1983 778 528 67.87 509 351 68.96 423 341 80.611984 744 463 62.23 623 434 69.66 369 296 80.211985 820 544 66.34 584 387 66.26 336 250 74.401986 811 547 67.45 596 420 70.47 384 302 78.641987 825 555 67.27 628 455 72.45 414 313 75.601988 998 678 67.93 705 516 73.19 392 302 77.04Total 4976 3315 66.6 3645 2563 70.3 2318 1804 77.8TABLE 38INCREASE IN NON-CABS REVASCULARIZATION PROCEDURESB.C. 1979-1988Year TotalRevasc.Non-CABS(CCP 4800-4805)CABS(CCP 4810-4819)N % total N % total1979 933 2 0.21 931 99.791980 935 4 0.43 931 99.571981 975 13 1.33 962 98.671982 1077 27 2.50 1050 97.501983 1286 66 5.13 1220 94.871984 1479 285 19.27 1194 80.731985 1768 587 33.20 1181 66.801986 1982 711 35.87 1271 64.131987 2136 811 37.97 1325 62.031988 2427 931 38.36 1496 61.64Total 14998 3437 22.91 11561 77.08183FIGURE 6GROWTH OF REVASCULARIZATION PROCEDURESB.C. 1979-19881979^1980^1981^1982^1983^1984^1985^1986^1987^1988YearCABS El^Other ^•^ TotalTABLE 40NUMBERS AND PERCENTAGES OF REVASCULARIZATION PROCEDURES BY CENTREB.C. 1983-1988Year Centre 1 Centre 2 Centre 3Total * Other %^of Total* Other %^of Total* Other %^of** total ** total ** total1983 542 14 2.6 355 4 1.1 387 46 11.91984 561 98 17.5 466 32 6.9 443 147 33.11985 750 206 27.5 464 77 16.6 551 301 54.61986 790 243 30.7 534 114 21.4 650 348 53.51987 876 321 36.6 619 164 26.5 632 319 50.51988 1100 422 38.4 714 198 27.7 602 300 50.0Total 3829 1304 34.0 3152 589 18.1 3265 1461 44.7*Total revascularization procedures (CCP 4800-4819) in centre.**Non-CABS revascularization procedures (CCP 4800-4805) in centre.184ANGIOPLASTYIt was impossible to determine the number of angioplasties performed inthe Province because, following the introduction of the CCP code for angioplastyin 1987, one hospital continued to use the old "catch-all" code (CCP 4800) forremoval of coronary artery obstruction. Consequently, a detailed analysis was notperformed 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-foldincrease overall and an increase from 0.2 percent to 38.3 percent of allrevascularization procedures. When the annual increase in both CABS and otherrevascularization procedures are graphed it can be clearly seen that between 1983and 1985, when utilization of non-CABS procedures was increasing rapidly, thegrowth of CABS slightly decreased. Since 1985 both procedures have increased(Figure 6).As with CABS, some B.C. residents receive angioplasty in Alberta. Between1983 and 1988, 100 patients went to Alberta centres for angioplasty and eleven ofthese patients received more than one angioplasty in Alberta. Although this re-operation rate appears lower than one would expect if restenosis is around 30percent, it is possible that some patients with restenosis received anotherangioplasty in B.C.Examination of the age and sex distribution of the non-CABSrevascularization procedures show some differences from the CABS population.A larger proportion of women (0.32) receive non-CABS revascularization and theoverall average age tends to be somewhat lower (60.48 years in 1988). Numbers bysex and overall mean ages are shown in Table 39 (Appendix A).185When the annual number of "other revascularization procedures" iscompared with the number of CABS procedures in each centre, there are markeddifferences between the centres. In 1988 Centre 3 had 50 percent ofrevascularization procedures as non-CABS procedures, compared to 38 percent inCentre 1 and 28 percent in Centre 2. In fact, the percent of revascularizationprocedures devoted to non-CABS has declined in Centre 3 since 1985 but has beensteadily growing in the other two centres (Table 40).It is of interest that 38 non-CABS revascularization procedures wereperformed in other centres between 1982 and 1988; thirteen of these procedureswere coded as angioplasties. It appears unlikely that so many cases would be dueto miscoding and it is possible that cardiac centres without cardiacsurgical facilitiesmay be performing angioplasties. A breakdown of annual numbers ofangioplasties and other non-CABS revascularization by centre, is given inAppendix A (Tables 41 and 41A.REGRESSION ANALYSISThis section describes the results of the regression analyses, which wereused to address the questions:i) Is there any significant variation in the age-sex adjusted CABS ratesamong the small areas (school districts) in B.C.?ii) Is there any significant variation in the age-sex and morbidity adjustedCABS rates among the small areas (school districts) in B.C.?iii) How much of the variation in (i) and (ii) above is explained by theindependent variables 3?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 familyincome in the school district (INC), employment rate (EMPRAT) and graduation rate (GRADRAT).186In order to determine if there was any small area variation in the smallarea annual age-sex adjusted CABS rates beyond that which could be ascribed tochance, 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 74degrees of freedom (p<0.0001). There was, therefore, significant variation in CABSrates among school districts and so the null hypothesis, i.e., that there was nosmall area variation other than that which could be ascribed to chance, wasrejected.The second analysis estimated the amount of the variation in the annualage-sex adjusted CABS rate across school districts that could be accounted for by theindependent variables. Information on these variables is shown in Tables 42 and43 (Appendix B). Year was first entered as a categorical variable but a graph of thecoefficients showed a quadratic form and it was found that year plus year-squaredprovided as good a fit. The model was then fitted for the main effects using astepwise regression, with year and year-squared forced in, to determine the orderin which the explanatory variables should be added. The model was then re-fittedadding each variable followed by its interactions with the variables already in themodel.In order of importance, the variables included in the model were income,distance from a cardiologist and distance from a surgical centre. There were alsofirst order interactions between distance from a cardiologist and year, year-squared, income and centre; and between distance from a centre, and year, year-squared and income. Distance from an internist, employment rate and graduationrate were found not to be important explanatory variables. Using the likelihoodratio test, no term in the final model could be omitted. Comparison of the finalmodel with the null model gave a chi-square of 191.656 with 12 degrees of freedom187(df); p=0.0001. Comparison of the final model with the saturated model, gave achi-square of 891.14 with 62 df (p=0.0001). R2 was 0.214 .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 themodel into a poisson regression, plus the true intercept. The estimate for year-squared is small compared to that for year so the effect of time is likely almost alinear one. The parameter for income (INC) is relatively large compared to theother parameter estimates; its negative sign indicates that people from regionswith lower average incomes were more likely to have CABS. Similarly thenegative sign of the parameter for distance from cardiologist (DSCAR) indicatesthat people in a region that contained a cardiologist were more likely to receiveCABS.The positive sign for distance from centre (DSCEN) appears to indicate thatpeople living in an area far from a surgical centre were more likely to receive theprocedure. Intuitively it would appear that this surprising result likely arosebecause of the significant correlation between DSCAR and DSCEN. However,when all terms containing DSCAR were removed from the model the parameterestimate for DSCEN was still positive. It appears that the paradoxical effect ofDSCEN is real.The relationship between both DSCAR and DSCEN and the CABS rateswere influenced by their interactions with one another and with income. Of thesethree 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, theparameter estimates were substituted into the model and all variables except year4 R2= (null model - final model)/(null model - saturated model).188and year-squared were held at zero. Compared to the reference year (1983) therelative 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 includeDSCAR were substituted into the model, and all other terms except year were heldconstant at reference levels, the relative risk for CABS was 0.93 in 1983 and 0.98 in1988 for those people in school districts adjacent to cardiologists and 0.87 and 0.97respectively for school districts far from cardiologists. Clearly the importance ofthe cardiologist in influencing the CABS rate decreased over the years, as did theimportance of distance from centre. When all other variables except DSCEN andyear were held constant at reference levels, the relative risk of DSCEN, comparedto the reference of school districts which contained a centre, fell from 1.18 in 1983to 0.99 in 1988 for those adjacent to a centre, and from 1.4 to 0.99 respectively forthose far from a centre.For income the reference group was the school districts which had anaverage census family income of $30,000. When average family income waschanged in increments of $10,000 the relative risks are 0.68, 0.47. 0.32 and 0.22 at$40,000 to $70,0006 respectively. For incomes below the reference the relative riskwas 1.45 for regions with an income of $20,000. A person from a school district, ina region with an average family income of $20,000 and a cardiologist, had 2.1 timesthe 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 for5 The relative risk is the ratio of the incidence rate for people exposed to a factor to the incidence rate forthose 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 incomeof $70,000. This income was obviously an outlier as the next highest income was $46,000. Repeating thestepwise regression without district 45 showed only minimal differences in the parameter estimates, so theschool district was retained in subsequent analyses.189those with an income of $20,000 decreased with increased distance from acardiologist, while the risk for those with an income of $70,000 increased withdistance.In order to demonstrate the effect of mobility on the model, it was refittedusing an indicator variable (ALBERTA) to represent those school districts whichsent 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 andALBERTA forced in. Results are shown in Table 45 (Appendix B). In this modelthe interactions of DSCEN with year, year-squared, income and DSCAR were nolonger significant. The Alberta indicator had the greatest influence on the CABSrate; the negative coefficient shows that people in the eight school districts inquestion had a lower probability of receiving CABS (in B.C.) than did those in therest of the province. Their relative risk was only 0.47, when all other variables areat reference. In this model also, both income and distance from cardiologist had anegative effect. Those from low income areas had the greatest probability ofreceiving CABS in B.C., especially when their area of residence contained acardiologist. 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 formorbidity in the school districts (Model C, Table 15). First, fully-saturated"perfect" models were run using the formula, described in Chapter 4. This modelwas repeated four times using different combinations of coronary artery diseasediagnoses to estimate morbidity. The model with the least unexplained variance(the smallest chi-squares at 1259.050) was that in which all patients with adiagnosis of chronic IHD, including those who also had a diagnosis of AMI (ICD-9411-414, 4292, with or without 410), were included; this diagnosis was assumed to190be the best estimate of morbidity. Results of the saturated models are shown inTable 46 (Appendix B).In order to check the above result, Pearson product-moment correlationcoefficients were calculated between the CABS SIRs and the SIRs for the morbiditydiagnoses. The highest correlation coefficient was found between CABS and thediagnosis of chronic IHD including AMI, and so this diagnosis was used to adjustthe expected CABS rate. This correlation although seemingly low, at 0.20, washighly significant (p=0.0001). However the correlation between the SIRs for CABSand those for AMI (-0.05) was not significant.The model using the CABS rate adjusted for morbidity was tested in thesame way as before. Stepwise regression, with year and year-squared forced in, wasfirst carried out and then the model was refitted by entering the variables, andthen their interactions, in the order suggested by the stepwise regression. In thismodel the first variable to be added after year and year-squared was distance fromcardiologist, followed by income. No other variables were found to be significantand 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 bydifferences in morbidity between regions. In this model the independent variablesexplained 30 percent of the variance.Parameter estimates and chi-square results for the above morbidity adjustedmodel are shown in Table 47 (Appendix B). The effect of DSCAR is increased inthat the relative risks are considerably lower than in the unadjusted model forthose 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 averageincome of $20,000 had 1.24 times the probability of receiving CABS than those inan area with an average family income of $40,000.191It is notable that in this model the paradoxical effect of distance from centredisappears and that the model is considerably simpler. Much of the complexity ofthe variation in CABS rate is apparently mediated through complex relationshipsbetween diagnosis of CAD and the other variables.This model was also repeated with the indicator variable (ALBERTA) usedto represent those school districts sending more than ten percent of cases toAlberta. Year, year-squared and ALBERTA were forced in as before. Again,ALBERTA had the greatest effect upon the CABS rate, followed by distance fromcardiologist and income (Table 48, Appendix B). No other variables were found tobe significant. The relative risk of CABS for residents of the 'ALBERTA' schooldistricts was 0.54 in 1983, when all other variables were at reference. For residentsin areas with a census family average income of $20,000 the relative risk of CABSin B.C. was 1.12. This relative risk fell as either income or distance from acardiologist, or both, increased so that for residents of ALBERTA areas far from acardiologist 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. Chi-square was 437 with 5 degrees of freedom (p<0.0001).For all these the models the independent variables explain 34 percent or lessof the variation in CABS rates among school districts over time. Sources of theunexplained residual variation may include interactions of age and sex withregion and with the other variables in the model, the possibility of a morecomplex dependence on income, misclassification occurring because the ecologicalvariables may not apply to the individuals having CABS within the region, otherfactors which may influence the CABS rate but which were not included in themodel and non-independence of observations at the individual level. This latterproblem may arise if one person's decision to undergo, or not undergo, theprocedure influences the decisions of others in the same vicinity.192The above analyses based on age-sex adjusted rates identify factors whichinfluence these rates, but cannot address the question of whether the age-sex effectitself varies from area to area, nor whether the influence of other factors, such assecular time, is different for different age-sex groups. To examine these questions aregression of age-sex-region-year specific rates against age, sex, year and region, andinteractions among them, was performed. School districts were aggregated intofour regions in two different ways, see Table 49 (Appendix B), for this analysis.This aggregation was done to reduce the otherwise unmanageable number ofpossible combinations of age-sex-year-region factors.This regression was carried out using models as shown in Chapter 4 (Table15). Again, year was first entered as a categorical variable and a graph of thecoefficients showed a quadratic form; year plus year-squared were found to provideas good a fit. Age (by five-year age groups) was entered as a categorical variablewith the 65-69 group as the reference. Region (classified into geographical regions,and into remote, rural, urban or metropolitan areas, according to populationdensity and distance from a referral hospital) 7 was entered as an indicator variablewith the metropolitan area as the reference.First year, age and sex were entered, followed by their first and second-orderinteractions; the latter were not significant. Then region was entered and this wasfollowed by the first order interactions with region. An attempt was then made towithdraw some of the variables which no longer appeared to be significant. Thefinal models contained age, sex, year and region and all second order interactionsbetween these variables (Tables 50 and 51, Appendix B). Chi-square for theremote/rural/urban/metropolitan model was 15232.299 with 55 degrees offreedom (p<0.0001). For the geographical model, chi-square was 14799.910 with 557 This classification was developed by the B.C. Royal Commission on Health Care and Costs. Schooldistrict 89 straddled two regions and so was excluded from this analysis.193df (p<0.0001) but, for this model, age-groups one to five had to be aggregatedbecause many of the age-sex-region groups in this age range had no cases.There are, therefore, significant differences in the time trends for CABSrates for different age-sex groups and for different regions when school districts arecategorized into regions both geographically and on a population density andproximity to health services basis.The previous age-sex adjusted analysis produces effect estimates (e.g., foryear) 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 theseindicate important departures from the assumption of homogeneity. In addition,because other factors such as income are only available as ecological variables, theinteraction analysis could not be extended to include them. There may well beinteractions between these factors and age and sex. This would cause additionalunexplained variation.The number of factors in the final model makes interpretation of the effectsof the age, sex, year or region on the CABS rate very tedious. However, a fewexamples will be given for the effect of sex in the remote/rural/urban/metropolitan model. In 1983 in the metropolitan region, the relative risk forfemales age 65-69 compared to that for males is 0.2; in 1988 the relative risk is only0.13. For females age 55-60 in remote regions in 1983, the relative risk of CABS is0.32 compared to 0.06 in 1988. It appears that, at least in these two regions, therelative risk of CABS for females as compared to males has decreased over theyears.194SUMMARYThe 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 inmean age and of co-morbid conditions points to an older and sicker populationreceiving the procedure in 1988 than in 1979.The annual standardized incidence ratios in many school districts showgreat variability over the ten-year period. Even so some school districts arecharacterized by consistent ratios; two of the three areas with the highest overallratio have been consistently high throughout the ten years. Examination of thecentres used by school districts for CABS procedures show definite referral areas.Between the centres there are differences in the proportion of open-heart surgeryand revascularization overall devoted to CABS.Poisson regression analysis showed significant variations in CABS ratesamong the school districts of B.C. between 1983 and 1988. Some of this variationcan be explained by interactions between age, sex, year and region, by regionalhealth 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 anddistance from a cardiologist remain whether or not the CABS rate is adjusted formorbidity within the region (although they switched positions, with incomebecoming less important after adjustment) but following adjustment, distancefrom a surgical centre is no longer significant.195REFERENCESGillum, R.F., 1987. Coronary artery bypass surgery and coronary angiography in theUnited States, 1979-1983. American Heart Journal 113 (May):1255-1260.Peters, S. et al, 1990. Coronary artery bypass surgery in Canada. Health Reports2(1):9-26.Preston, T.A., 1989. Assessment of coronary artery bypass surgery and percutaneoustransluminal angioplasty. Intl I of Technology Assessment in Health Care 5:431-442.196CHAPTER 6REVASCULARIZATION IN BRITISH COLUMBIA 1979-1988 DISCUSSION OFRESULTS AND POLICY IMPLICATIONS OF STUDYDISCUSSIONDESCRIPTIVE STUDYIncidence RatesAn increase in the numbers, and incidence rate, of CABS between thelate-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 inall Canadian provinces between 1981 and 1986 (Peters et al 1990). The relativelyconservative increase in B.C., compared to other jurisdictions, likely reflects theprovincial government's ability to restrict growth of this expensive procedurethrough budgetary controls.To compare the B.C. increase to that in other areas gives the impressionthat this province may be providing CABS at too low a rate; it is not keeping upwith the population need. This supposition may, or may not, be true but itcannot 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 topopulation need either requires population outcome studies or an estimate ofthe proportion of the provincial population who have the conditions for whichCABS has been shown to be effective.The two-fold increase in incidence rates for the over-65 population foundin this study is similar to the 2.38-fold increase found by Gillum in the UnitedStates between 1979 and 1983. Canadian authors have not reported the over-65197rate alone, but Anderson and Lomas (1989) reported a five-fold increase in therate for the population of 70 and over between 1979 and 1985; in the presentstudy there was over a four-fold increase for this age-group over a ten yearperiod. It appears, therefore, that although the rates in the B.C. elderly haveincreased in the direction shown in the literature, the increases are moremodest than in Ontario.One explanation for the differential increase within Canada, could be thatthe rates for the elderly in B.C. were higher at the beginning of the period thanthey were in Ontario. There is some evidence for this in the data on rates in thecensus metropolitan areas, reported by Peters et al (1990). The mean 1981-1986rate for the 75 and over population in Vancouver was considerably higher thanin 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 thisexplanation must remain only a supposition.However the rates in B.C. compare to those in other areas, the increase inthe rates in the elderly are of concern because benefit of CABS to the over-65population has never been demonstrated through either randomized trials orprospective controlled studies. As shown in the literature, the elderly have ahigher operative mortality (Gersh et al 1983, Goldman et al 1987) and are at agreater risk of stroke (Goldman et al 1987), and of neuro-psychological problemspost-operatively (Townes et al 1989) than are those under the age of 65. Therisks are increased for elderly patients with co-morbid diseases and withincreasing age (Goldman et al 1987). In this study almost 12 percent of patientsover the age of 65 had either diabetes or COPD; if other disease such as renalfailure and hypertension were also included the population with comorbiditywould be much higher. The risks are, of course, additive and so for manyelderly patients the risks of having CABS may well exceed the potential benefits198of the surgery. In addition, the increased use of resources post-CABS by theelderly (Goldman et al 1987) restricts the resources available for other patientsand raises questions about the cost-effectiveness of the procedure in this agegroup.Because many of the above risks have been shown to increase withincreasing age, the number of procedures performed on the "old" old, i.e., thoseover the age of 80, is of especial concern. The numbers of CABS performed onthose 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 comparedto 0.1 percent in 1979. This trend to increasing numbers of very old patients athigh risk for a poor outcome undergoing a costly procedure raises the issues ofquality of care and of appropriate use of resources.Variations in Utilization by Region and by CentreThe extreme variability of standardized incidence ratios for CABS withinschool districts over the ten years of the study shows, if nothing else, thatconclusions about geographical variations in the B.C. CABS rate (and likely allrare procedures) cannot be drawn on the basis of one year's data. The extremevariability of the SIR's over time and across school districts, especially in areaswith small populations, raises methodological questions about appropriatepopulation sizes for small areas, about the number of years of data that shouldbe studied in order to gain stable rates and about how best such variable datashould be reported and analyzed.The two-fold variation (at the most conservative estimate) in SIRs amongschool districts when the data are aggregated over the ten year period, showsthat geographic variations in the rate of CABS do exist over the long term andare not simply an artifact caused by transient changes in annual rates. These199variations are important both because of their clinical implications and becausethey may indicate inequities in care across the Province. These issues will bediscussed later in this chapter.The differences between centres in the amount of resources committed torevascularization, type of patients and in referral patterns are of interest. Centre3 shows a very different picture from the other two centres, in that it performsfewer CABS but devotes a greater percentage of its total open-heart surgeries toCABS and a greater percentage of its revascularization procedures to non-CABSprocedures. This centre also draws its patients from a much more circumscribedarea than do the other centres. It is not certain whether these differences are aresult 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 areconsistently over one, and because Centre 3 has a higher proportion of totalrevascularization procedures devoted to non-CABs procedures, it appears likelythat the school districts served by this centre utilize revascularizationprocedures as a whole at a much higher rate than do most other school districtsin the provinces. This possible high level of revascularization appears to be atodds with the morbidity levels in some of the areas served by Centre 3. Forschool districts 61, 62 and 63 the annual SIRS for chronic CAD are consistentlybelow one after 1983 while school districts 65, 66, 69 and 70 have chronic CADSIRs that are consistently above one. There appear to be two opposing patternshere. High morbidity areas with high CABS rates and low morbidity areas withhigh CABS rates. It is possible that either the increased morbidity in an arearesults in a higher CABS rate or that the high CABS rate in an area results in1 This is assuming that Centre 3 has the same referral areas for non-CABS procedures as it does forCABS.200lower morbidity but it appears unlikely that both these patterns would occur inschool districts served by the same surgical centre. The fact that school districts63 and 62 are adjacent to the Greater Victoria school district, which contains thesurgical centre, likely influences the different pattern shown in these districts. Itshould be emphasized that the differences between morbidity rates and surgicalrates in the school districts served by centre 3 were not tested for significance andtherefore may be likely to have occurred by chance. However, the existence ofthese differences is puzzling and warrants further research.The differences noted above for centre 3, together with the differencesbetween the hospitals, raise the question of whether the variations in smallarea CABS rates are related as much to the referral hospital itself as to thecharacteristics of the area from which the patient comes. Anderson and Lomas(1989) found a significant relationship between the referral centre providing careto Ontario counties and the rates of surgery on residents of those counties.Examination of the overall SIRs for the ten year period and the mapshowing those school districts which receive a plurality of CABS services fromone centre, show that the majority of school districts served by Centre 2 havefewer CABS than expected over the ten years, all school districts served byCentre 3 have more CABS than expected while half the school districts served byCentre 1 have fewer CABS than expected and half have more. While it cannotbe ruled out that these differences are due to referral patterns, it appears likelythat physicians attached to different hospitals are using different indicationswhen recommending CABS. If this is so, the issue of appropriateness of theprocedure, and of under- or over-servicing, in different institutions is raised.Both Bunker (1988) and Anderson and Lomas (1989) point out that quality ofcare is based on the appropriateness of the procedure (the match between thepatient's clinical condition and the correct indications for the procedure) as201much as on the technical skill demonstrated in the performance of theprocedure. Consistent differences in rates between school districts served bydifferent institutions likely indicate that the quality of care is beingcompromised.It is not possible to gauge the clinical implications of these results. Inorder to evaluate the clinical success of the revascularization program it wouldbe important to know i) that all revascularization procedures performed wereappropriate 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 bytheir condition and iv) that short and long-term outcomes for both medical andsurgical patients were equivalent to, or better than those reported in theliterature. The presence of significant variations in small area rates even afteradjustment for morbidity and the presence of the differences, noted above,between surgical centres implies that either inappropriate procedures are beingperformed 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 CABSrates, may be receiving intensive medical treatment that mitigates their need forsurgery or may be receiving relatively higher rates of angioplasty.Research to establish the clinical effect of variations in CABS rates isrequired. Comparisons of appropriateness of treatment and of outcomes inareas with different morbidity-CABS patterns would be one starting point 2 .Another route would be to compare similarities and differences in adjoining2 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 highmorbidity rate and a high CABS rate; Kamloops (SD 24) which has a low morbidity rate and alow CABS rate and Lake Cowichan (SD 66) which has a low morbidity rate and a high CABSrate.202school districts which have the same sources of care but different CABS rates.For example, both the Central Okanagan (SD 23) and Vernon (SD 22) residentshave access to the cardiologists in Kelowna and both receive about 50 percent ofCABS procedures from Centre 2, but Central Okanagan has a consistently higherrate than Vernon. Comparisons of appropriateness of CABS between thedifferent centres will also provide information about whether differences inappropriateness of care may be contributing to variations in area rates.Ultimately though it seems likely that, in order to fully understand howvariations arise, research will have to be carried out at the level of theindividual patient and physician.Other Issues:Two other issues deserve comment. The first is the finding that anumber of angioplasties appear to be performed in centres without surgicalfacilities. The guidelines for PTCA developed by Ryan et al (1988) state thatPTCA should be carried out in a hospital with facilities for providing cardiacsurgery. Certainly they should at the very least be carried out in a hospitalwhich is not more than half-an-hour's drive away from cardiac surgicalfacilities; not all B.C. hospitals with angiography units are this close to thesurgical centres. This issue likely requires discussion, and development ofguidelines, by the cardiologists and cardiac surgeons in the province. If, on theother hand, these angioplasties simply represent miscoding, this situation alsoneeds to be redressed.The second issue relates to the differences in coding between the cardiaccentres. These differences reduce the validity of the hospital morbidity data-baseand thereby reduce its usefulness. While adjustments can be made in theanalysis to accommodate for differences in coding, many hospitals do not appear203to keep records of changes that they have made to coding practices so thatinformation about coding changes within a particular hospital, and when theyoccurred, is often impossible to obtain. It is suggested that both the HospitalMedical Records Institute and the Health Records Association take responsibilityfor ensuring that the same procedures are coded in the same way across theProvince. It is also suggested that medical records departments keep records ofcoding practices and changes within their department.Regression Analysis:This study is the first to show that variation in small area CABS ratesremain even after adjustment for small-area morbidity. It is also the only studyinto small-area variations in coronary artery bypass surgery which has examinedthe effect of area socio-economic characteristics on the CABS rate. The resultsindicate that, both before and after adjustment for morbidity, regional incomeand proximity to a cardiologist are the most important explanatory variables,while distance from centre was important in the models unadjusted formorbidity.Income:The importance of income in explaining the variation is CABS rates isinteresting. There is likely a stronger association than is shown here becauseincome was only available for census years so only two values were entered foreach school district. This would tend to dampen the income effect. Theinteraction of income with distance from cardiologist likely occurred becausecardiologists in B.C. generally practice in the urban, higher income, areas. Thefinding that people in lower-income areas are more likely to receive CABS thenearer they are to a cardiologist while the reverse is true for those in high-204income areas, is difficult to understand and requires further study. The fact thatincome became relatively less important after the CABS rate was adjusted formorbidity, likely reflects the association between CAD and income reported inthe literature. However, the higher incidence of CAD in lower income groupsobviously does not explain all the income effect because income is stillimportant in explaining the CABS rate even after adjustment for morbidity.The reasons for this are obscure but may relate to differences in the populationsof low- and high-income areas with respect to life-style, willingness to complywith a medical regimen, or willingness to accept surgery. This area also requiresfurther study.The negative association of income with CABS is a positive accolade forthe Canadian health care system because it indicates that lower income is not abarrier to receipt of this expensive procedure. It would be interesting to know ifthere is a similar relationship in the United States.Given the importance of income in explaining the CABS rate, theunimportance of the employment rate is surprising. However, given the highlysignificant correlation between income and employment rate it is likely thatemployment rate did not appear to be important because the variables are co-linear.Distance from ServicesThe increasing importance of the cardiologist in affecting the CABS rateis not surprising given the progress made in diagnostic cardiology in recentyears. The fact that the importance of distance from cardiologist increased afteradjustment for morbidity may indicate that the proximity of a cardiologist has agreater impact after CAD has been diagnosed. As a whole, the results on theimportance of proximity to a cardiologist appear to show that cardiologists make205more liberal decisions about the surgical treatment of CAD than do eitherinternists or general practitioners. This finding is similar to that found by Brooket al (1988) but in the opposite direction to that found by Young et al (1987), whoshowed that cardiologists were more conservative than general practitioners inreferring hypothetical patients with angina for angiography. It may be thatdecisions about hypothetical patients do not reflect real world practice, or thatcardiologists are more risk averse prior to diagnosis of CAD than they are inpatients known to have the disease. This area too needs further study.The fact that distance from an internist was not an important explanatoryvariable is surprising because, they are more available than cardiologists in thehinterland of British Columbia. Therefore, they are likely to be the specialistsresponsible for the care of patients with CAD in small urban. rural and remoteareas. However, the colinearity of DSCAR and DSINT likely explains theseeming unimportance of distance from internist.The finding that distance from a surgical centre had a positive correlationwith 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 thepositive effect of distance from centre is only active in proximity to acardiologist. It is also possible that cardiologists close to a centre may be morewilling to "nurse" sick cardiac patients along because they have access to cardiacfacilities if required, while physicians far from cardiac centre may transferpatients while they are still in relatively good condition. Once these patients areadmitted 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 cardiacfacilities. It would be interesting to see whether the positive affect of DSCEN206remains over time as cardiologists and cardiac facilities become more widelydistributed in the province.Other VariablesThe effect of secular time in increasing the relative risk of CABS, impliesthat the Ministry of Health has been increasing the numbers of CABS funded ata rate faster than population growth. Even in the morbidity adjusted model therelative risk of CABS increases over time which indicates that the procedure isincreasing faster than the increase in morbidity rates.The unimportance of the graduation rate in all models is not surprisingbecause the variable used was not an indicator of the actual educational levels inthe region but a rather uncertain proxy for it. Level of education has beenshown to be negatively associated with the prevalence of risk factors for CAD(Millar and Wigle 1986) and to be a moderately important indicator of healthstatus (Hay 1988). The lack of individual data on patients' socio-economicstatus, both in the hospital morbidity data base and in hospital records, hampersresearch into the relationship between socio-economic status and the utilizationof hospital services.At best the independent variables tested explained only 34 percent of thevariation in CABS rates across school districts. While some of the unexplainedvariance was likely due to interactions between the age-sex distribution in theschool districts and the independent variables, there are obviously other factorsat work which were not considered in this study. These factors likely includecultural factors and knowledge about, and attitude to, CABS of the generalpractitioners in the school districts. It should be emphasized that manipulationof health services factors (such as the supply of cardiologists) in order to altersmall area rates, without consideration of the other factors which may also be207affecting that rate is unlikely to redress the inequities in the treatment of CADwhich appear to exist in B.C.. Small area CABS rates appear to be a result of acomplex relationship between area age-sex distribution, income, morbidityrates, and proximity to a cardiologist as well as other factors not yet identified.This study has avoided some of the methodological pitfalls found inother regional variations studies. The use of six years of data for the regressionanalysis, has reduced the effect of outliers, while the use of independentvariables which were not directly related to total population size should havereduced the confounding effect of population. However, there are also anumber of limitations to the study, which may affect its validity, reliability andthe generalizability of the results, and these are described below.LIMITATIONSPerhaps the most significant limitation in this study is that the validityand reliability of the hospital morbidity data base are not known. Within thedata support section of the B.C. Ministry of Health it is believed that until April1st, 1983, when the Ministry of Health ceased the management of the data base,the data were valid representations of the information on the patients' hospitalcharts. Ministry staff would meticulously check any incongruent or abnormalentries with the hospitals concerned. After 1983, when data managementinstitutions took over data entry, the quality of the data likely declined. Errorsin the database which would affect the validity of the results would be thoserelating to the coding of diagnosis, procedure3, age, sex or postal code of thepatient. While it seems unlikely that CABS would be missed as a procedure it ispossible that it could be miscoded. Twenty-one patients over the ten-year period3 There are difference between the hospitals in the way that CABS procedures are coded but theseall relate to the number of vessels coded when the internal mammary artery is used.208were coded as receiving the procedure in a hospital which was not a cardiacsurgical centre. These patients were kept in the study, but either the procedureor the hospital was likely miscoded. Miscodings for diagnosis, age or sex wouldbe more difficult to spot if they fell within the approved alpha - numeric codefor that field. It is likely though that any miscodings would be randomlydistributed around the province and should, therefore, not systematically biasthe results of the study. Studies to assess the reliability and validity of thehospital morbidity database are presently underway.The second major limitation is that the population used (all adults aged20 and over) is not the true population at risk because a relatively smallproportion of this population would have CAD and only a small proportion ofthose would be eligible for CABS. Therefore the incidence rates given are anunderestimate of the incidence in those at risk.It is not known how well hospital discharges with a diagnosis of CADreflect the actual morbidity rate for severe atherosclerosis and myocardialischemia in an area. Systematic differences between areas in the tendency togive 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 arenot hospitalized, and those with silent myocardial ischemia who have neverbeen diagnosed will not be included in the morbidity rates. If, as seemsintuitively likely, general practitioners in rural and remote areas, who do nothave access to cardiologists for referral, both diagnose CAD in more borderlinepatients and admit more CAD patients to hospital there will be a bias to highermorbidity rates in rural and remote school districts.The division of the morbidity diagnoses placed patients with unstableangina into the "chronic CAD" category; this was done in order to avoid putting209the AMI and unstable angina patients into one class. In retrospect it would havebeen better to have placed the unstable angina patients into a class of their ownso that the rate for this category could be correlated with the CABS rate bothalone and in combination with other CAD diagnoses. This may have produceda different combination of diagnoses to use in adjusting the CABS rate formorbidity.The presence of many interactions posed difficulties in interpreting themodel and these interpretations may not be valid reflections of reality. Whilethe greater simplicity of the morbidity-adjusted model appears to indicate thatthis is the more valid model, the potential limitations of the morbidity datacannot be ignored. Similarly, the method used in the regression analysis toaccount for the mobility to Alberta appears to be valid but may not be capturingthis effect fully. It is possible that more school districts should have beenincluded in the mobility variable because 15 percent of patients going to Albertacould not be assigned to a school district.The size of the data set and the number of analyses done increase theprobability of significant results arising by chance. The 'p values' were notadjusted to account for the number of analyses, but had they been adjusted theresults would still have been significant.The limitations posed by the data used to estimate income, employmentrate and graduation rate has already been discussed above. The likely effect ofthe limited data for income and employment rate would be to reduce therelationship between the variable and the CABS rate. Also, because income isnot linear, but was assumed to be so the amount of variation explained by themodel is likely reduced.The use of ecological variables, rather than independent variables whichrelate directly to the individual receiving CABS, means that the relationship210between the CABS rate and the explanatory variables cannot be generalized toindividuals. It is not known whether individuals from a low income area whoreceive CABS, themselves have a low income, nor whether patients living inthe same school district as a cardiologist ever visited that cardiologist. Toassume that the results of this study apply to individuals would be an ecologicalfallacy.The inability to identify patients having reoperations has likely weakenedthe relationship between the dependent and independent variables becausethere are likely different factors associated with having a reoperation than thereare with having the initial CABS.Finally, the inability to identify the angioplasties in the province as awhole and in the school districts, meant that the angioplasty rate could not beincluded as either an independent or a dependent variable in the regressionequation. It is highly likely that the small-area CABS rates, at least in the lastfew years, have been mediated by the angioplasty rates in the areas. Theexclusion of the angioplasty rate as an independent variable has likely increasedthe amount of unexplained variance in the model. The inclusion of theangioplasty rate as a dependent variable would have allowed comparison of thefactors which explain the rates of both types of revascularization procedures.POLICY IMPLICATIONSThis section will discuss the major policy implications of both the resultsfrom the analysis and from the literature review contained in chapters onethrough three. In order to put these policy implications into context, it is firstnecessary to understand how health care is organized in British Columbia.211Organization Of Health Care in B.C.:British Columbia administers a universal health insurance programwhich provides comprehensive coverage to all residents. The five essentialelements of this program are that it is equitable, universal, comprehensive,accessible and portable between Provinces. The program is funded throughFederal grants, provincial tax revenues and monthly premiums paid byresidents. 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-serviceaccording to a fee-schedule which is negotiated annually between the BritishColumbia Medical Association (BCMA) and the Ministry of Health. Mostphysicians are, in effect, self-employed professionals contracting out theirservices to the government.Hospitals in B.C. are funded by a global budget, negotiated annually withthe Ministry of Health, which covers all operating costs and services provided bythe hospital. Capital equipment is funded separately. In addition specializedand expensive services, such as cardiac surgery, are funded as a program on a percase basis but the number of cases funded per hospital is limited. Most hospitalsonly provide acute care and/or rehabilitation, although in recent years somehospital-based prevention programs have emerged.Health promotion and preventive services are generally provided by theCommunity Health Units which are run by either the Community and FamilyHealth Service Division in the Ministry of Health or, in the largemunicipalities, by the Local Board of Health.Implications of Literature and Present Study:The discussion on policy implications will centre around the twoessential aspects of the B.C. (and Canadian) health care system which appear to212be in question with regard to CABS in this province. These aspects areaccessibility and equity. Accessibility may be defined as the relationship betweenavailable resources and patient needs and reasonable expectations for obtainingentry in terms of location, time, effort and cost. Accessibility is also affected bycontinuity of care. A patient may be able to access the system via his familydoctor but not be referred to a specialist, or other resources, when necessary.Equity is more difficult to define because in an environment in which theneed for services may be composed of many different elements (e.g., physicalcondition, lifestyle, cultural beliefs etc), equal treatment is impossible. For thepurpose of this discussion, equity in health care in B.C. will mean that allresidents have an equal right to the best health outcome that can be achievedgiven their condition and the resources available in the system. This best"health outcome" can likely be achieved in a number of different ways and howit is achieved will depend on the condition and preferences of the patient, thejudgement of the physician and the services available. However, implicit in thenotion of equity of outcome is the notion of efficiency, both technical andallocative4 . All health care activities involve an opportunity cost - the loss ofservices which could have been provided with the money spent on thoseactivities. When resources are scarce, care which is not efficient is not equitablebecause the money lost through inefficiency is not available for other services.Therefore, in order to promote equity the government has to provide for bothallocative and technical efficiency in the health care system.4 Health care is technically efficient if it uses the least costly quantity and mix of servicesconsistent with the desired outcome. Foe example, CABS would be technically efficient care for apatient with significant left main disease but not for one with single vessel disease and Class IIangina. 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, byreallocating resources, to make someone better off without simulataneously making others worseoff.213Cost Effectiveness:The potential for inequity in care arises from the resources allocated torevascularization and other types of cardiac care in the province. Extrapolatingfrom the annual numbers of CABS identified by this study, and the average costper patient reported by Krueger et al (1991), it appears that about $19 milliondollars are spent annually on providing CABS. At a rough estimate a furtherfive million dollars are spent on providing angioplasty 5. The, approximately,$24 million spent annually on revascularization is only well spent if allprocedures provided are appropriate, if all required procedures are providedand if the same outcome could not be achieved at a lower cost. The answers tothe 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 onacute care and on revascularization in particular. Balance in the cardiacprogram may require the government to promote research into non-invasivetreatment of severe atherosclerosis, similar to that described by Graboys et al(1989) and Ornish et al (1990), for patients who are eligible for CABS orangioplasty but who do not choose invasive treatment as an option. The costper patient in such a program may be high but is unlikely to be as high as thecost of CABS. A program like this could, by halting progression or causingregression of CAD, also be more effective in prolonging life and reducingmorbidity than is CABS.5 This figure is arrived at by multiplying the total number of non-CABS revascualrizationprocedures in 1988 by $5,500, the cost per procedure found by Reeder et al (1984) in the UnitedStates. Because of inflation since 1984 five million dollars is likely an underestimation of the costof angioplasty.214The role of the Medical Services Plan in paying for costly, high-riskprocedures used in situations for which efficacy has not been shown, may needto be evaluated. It may be difficult to impose sanctions on the use of procedureswhich are commonly accepted as effective by the general public, but thegovernment should not in the long term continue to fund therapies which maypose greater risks to patients than are posed by the patients' conditions. Publiceducation about the causes and treatment of CAD may, by putting CABS into itsproper perspective as the treatment of choice for only relatively few conditions,reduce some of the pressure on the government to make policy decisions basedon public demands.It would not seem unreasonable to require physicians who continue touse procedures in situations in which efficacy has not been shown, to provideproof of efficacy. Such proof would have to be by means of a randomized trial ora controlled study with an adequate sample size. Proof of efficacy is required forthe use of CABS in the elderly, particularly those over the age of 75, for CABS incomparison to aggressive medical treatment, and for angioplasty in comparisonto medical treatment.Inappropriate Treatment:If equity is defined as an equal opportunity to achieve an optimumoutcome, then inappropriate treatment is an equity problem both because of themisuse of scarce resources and because it puts patients at unnecessary risk andlikely increases morbidity and mortality. 6 Leape (1989) concluded thatgeographical variations in procedure rates are indirect indicators ofinappropriate surgery. Both the literature review, which showed that CABS isfrequently 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 CABS215the finding of significant geographical variation in this study, indicate that thereare likely inappropriate CABS being performed in British Columbia.Leape believes that inappropriate surgery results from inadequateinformation, and suggests that the role of the policy-maker in preventinginappropriate surgery is that of stimulating the development and disseminationof practice guidelines which are derived from evidence of effectiveness. Herecommends that guidelines be used for operations which have unusualpotential to harm a patient, or involve extensive use of resources, or which arecontroversial or which are suspected of inappropriate use. Coronary arterybypass surgery fits into all of these categories and so is a good candidate for aprocedure for which guidelines should be used.In the same paper, Leape discusses the responsibilities of professionalsocieties, academic institutions and government in developing guidelines. Hebelieves that professional medical societies are the logical groups to developguidelines, while academic centres play a role in developing an appropriatemethodology for deriving guidelines. The role of government is to encourageand fund both the development of both the methodology and the actualguidelines themselves, but not to play a direct role in deriving guidelines.Leape sees medical practice guidelines as a classic "public good" from whicheveryone should benefit. Funding their development is therefore a logicalgovernment function; closer involvement in the development process wouldgive government unprecedented control over medical practice and would leadthe public to question whether the purpose of the guidelines was to improvequality of care or to reduce utilization and costs.Indications for the use of CABS have recently been identified (but not yetpublished) by a consensus group working in cooperation with the CanadianCardiovascular Society. These guidelines, if acceptable to the B.C.216cardiovascular-surgeons, could be used as, or as a basis for, provincial practiceguidelines. The advantage of using national guidelines is that by reflecting abroader scope of practice they will have more credibility than locally developedguidelines, and it may be cheaper to develop, revise and update one set ofnational guidelines than ten provincial sets. National guidelines on indicationsfor use will also facilitate valid comparisons between provinces on utilizationrates and outcomes of CABS. However, it is often the case that physicians at thelocal 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 tothose using them but will decrease their usefulness in comparison.Once developed and accepted by the medical community, how shouldsuch guidelines be used? Leape suggests three ways in which guidelines could beused to reduce unnecessary surgery; to educate practitioners, to provide explicitcriteria for quality assurance programs within hospitals, and as standards forphysician payment. Possibly the most effective use of guidelines would be torequire prospective assessment of appropriateness of CABS before the patient isadmitted to the waiting list, similar to the prospective assessment used in manyHealth Maintenance Organizations in the United States. This would likely notbe well accepted by the physicians but may be more acceptable than requiringthem 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 inreducing inappropriate CABS in the province. Applying the use of guidelinesin all of the ways mentioned above would have the most impact. Cost savingswould depend on the number of inappropriate surgeries being done, thenumbers of appropriate patients who are presently not considered candidates forsurgery but who may become so under the guidelines, and the cost of the217development of the guidelines and the monitoring procedures that would needto be set up. The benefits to patients could be enormous.Elimination of inappropriate surgery is but one aspect of improvingtechnical efficiency and equity in the cardiac program. The presence of regionalvariations in care also leads to the suspicion that some patients who requireCABS may not be receiving the procedure. Identifying such patients, however,is more complex than identifying those receiving inappropriate CABS. Chartreview, and follow-up, of all patients receiving coronary angiography who areshown to have significant CAD, may identify some of these patients. However,because angiography itself has regional variations and a high proportion ofinappropriate use (Chassin et al 1987) such an investigation is likely to be biased.Another method of identifying areas where patients may not be receivingadequate 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 ofCABS. Stimulation, and funding, of such research would appear to be part ofthe role of government.Other:The marked differences in CABS rates in the school districts served byCentre 2 and Centre 3 may indicate problems in access to CABS for patientsreferred to Centre 2, inappropriate use in Centre 3 or both. It is possible that thedifferences in the percent of cardiac surgery devoted to CABS makes access to theprocedure easier in Centre 3 but access for Vancouver Island patients requiringother types of cardiac surgery could be reduced. These differences betweenhospitals need to be addressed especially as they may contribute to the variationin small-area rates.218The importance of the proximity to a cardiologist in determining theCABS rate may indicate that residents in those school districts far fromcardiologists have difficulty in accessing CABS. However, before decisions aboutthe dispersion of these specialists throughout the province are made, someevidence is required that cardiologists are increasing the number of appropriatereferrals, rather than increasing the CABS rate through inappropriate orequivocal referrals. Should it be found that cardiologists do indeed increase theappropriate referrals, and that there are school districts far from cardiologistswhere patients who need CABS do not appear to be referred for the procedure,then either visiting cardiologists or education programs for the local physicianscould be instituted.The impact of income on the CABS rate, even after adjustment formorbidity, indicates clearly that differences in other population characteristicsbesides demographics, can affect need. Policy which addresses health servicesupply factors without also taking the income of the area into account mayresult in inequities. The results on income also show that a marked decrease inboth CAD and CABS could likely be achieved by the introduction of measuresto stimulate the economy in low-income areas and by the narrowing of the gapbetween rich and poor throughout the province.Research:This study has identified a number of areas in which further research isrequired; this section will briefly address those which impact onrevascularization policy.Perhaps the most important item on the research agenda is to establishthe efficacy of CABS in the elderly, especially in those over 75 who are the groupwith the fastest growing rate. Given the importance of this endeavour, the219Ministry of Health may wish to encourage, and contribute funds to, such astudy. The optimum study would be a randomized trial but a controlled studyin which cases are matched for risk would also be appropriate.Of equal importance to the above, is the need to establish the role thatangioplasty should play in the treatment of CAD. Efficacy, and cost-effectiveness, of the procedure in relation to medical treatment needs to beestablished.Given the present interest in a healthy lifestyle shown by the generalpublic, alternative therapies for the treatment of CAD, along the lines of theprogram described by Ornish et al (1990), would likely be well received. Thepromotion and evaluation of alternatives could reduce what may be aninappropriate reliance on the biomedical approach.The different patterns in CABS rates between school districts served bydifferent centres, leads to the suspicion that the variation in small-area CABSrates could be primarily related to the referral centre. A study similar to thatcarried out by Anderson and Lomas (1989) in Ontario would test this hypothesis.With the start of the new cardiac surgery centre at the Royal ColumbianHospital in 1989, referral patterns for CABS will likely have changed in theLower Mainland and in parts of the Interior. Comparison of referral patterns inthese areas before, during and after the start-up period may help to establish theeffect that referral patterns have on the CABS rate.As mentioned above, the clinical implications of the variations in CABSrates need to be explored through outcome studies on CAD patients in areaswith different patterns of CABS and morbidity rates. Outcome studies wouldlikely require links between the hospital morbidity database and the vitalstatistics database.220Finally, 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 beable to be used to address questions of efficacy or effectiveness if medical patientsare included. If the database were set up similar to the CASS Registry so that allpatients receiving coronary angiograms were entered, then preliminary studieson efficacy and/or on cost-effectiveness would be facilitated, althoughprospective studies would still be required. If possible the data-base should alsocontain information on risk factors.221REFERENCES:Anderson, G.M. and J. Lomas, 1989. Regionalization of coronary artery bypasssurgery: Effects on access. Medical Care 27 (March):288-296.Brook, R.H. et al, 1988. Diagnosis and treatment of coronary disease: Comparisonof 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? Inquiry25:51.Chassin, M.R. et al, 1987. Does inappropriate use explain geographic variationsin the use of health care services? TAMA 258 (November):2533-2537.Gillum, R.F., 1987. Coronary artery bypass surgery and coronary angiography inthe 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 arterybypass surgery. TAMA 258 (September):1611-1614.Hay, D.I., 1988. Socioeconomic status and health status: a study of males in theCanada Health Survey. Social Science and Medicine 27(12):1317-1325.Kircher, T., J. Nelson and H. Burdo, 1985. The autopsy as a measure of accuracyof 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 forcardiovascular disease. CMAT 134 (January):127-32.Ornish, D. et al, 1990. Can lifestyle changes reverse coronary heart disease? TheLancet 336 (July):129-133.Peters, S. et al, 1990. Coronary artery bypass surgery in Canada. Health Reports2(1):9-26.Preston, T.A., 1989. Assessment of coronary artery bypass surgery andpercutaneous transluminal angioplasty. Intl T of Technology Assessmentin Health Care 5:431-442.Ryan, T.J. et al, 1988. Guidelines for percutaneous transluminal coronaryangioplasty. Circulation 78 (August):486-502.222Young, M.J. et al, 1987. Do cardiologist have higher thresholds forrecommending coronary arteriography than family physicians? HealthServices Research 22 (December):623-635.223APPENDIX A224TABLE 16ANNUAL CABS PROCEDURES FOR B.C. AND OUT-OF PROVINCE RESIDENTSB.C. 1979-1988Year Isolated Coronary Artery Bypass SurgeryB.C. Residents Out-of-Province Total#^% #^%1979 922^99.1 9^0.9 9311980 924 99.3 7 0.7 9311981 950^98.8 12^1.2 9621982 1039 99.0 11 1.0 10501983 1193^97.8 27^2.2 12201984 1178 98.7 16 1.3 11941985 1163^98.5 18^1.5 11811986 1240 97.6 31 2.4 12711987 1304^98.5 21^1.5 13251988 1470 98.3 26 1.7 1496Total 11383 178 11561TABLE 17ANNUAL CABS PROCEDURES BY SEXB.C. 1979-1988Year Isolated Coronary Artery Bypass SurgeryMale Female Total#^% #^%1979 751^81 180^19 9311980 749 80 182 20 9311981 778^81 184^19 %21982 863 82 187 18 10501983 966^79 254^21 12201984 943 79 251 21 11941985 952^81 229^19 11811986 1031 81 240 19 12711987 1038^78 287^22 13251988 1234 82 262 18 1496225TABLE 18ANNUAL MEAN AGE OF CABS PROCEDURES PER YEARB.C. 1979-1988Year Male Female Total1979 57.08 59.00 57.451980 57.23 59.93 57.761981 57.72 59.69 58.101982 59.08 60.75 58.861983 58.86 63.05 59.731984 60.42 63.55 61.081985 61.30 64.60 61.941986 61.95 64.81 62.491987 62.05 65.02 62.691988 62.52 65.21 62.99TABLE 20AGE-SEX SPECIFIC RATES FOR CORONARY ARTERY BYPASS SURGERYB.C. 1979-1988Age GroupAge-Sex Specific Ratesper 10,000 populationMale Female20-24 0.008 0.00*25-29 0.06 0.0230-34 0.34 0.0635-39 1.47 0.1840-44 4.08 0.5345-49 9.39 1.7050-54 17.68 2.7055-59 25.13 4.3760-64 31.00 7.6065-69 33.70 9.2870-74 24.89 6.5775+ 6.44 1.75*There were no female cases in this age-group226TABLE 22ANNUAL CABS PROCEDURES PERFORMED ON SUB-GROUPS OF THE OVER-65 POPULATIONB.C. 1979-1988Year Age Group65-69 70-74 75+^i Total# % # % # %1979 160 74.1 47 21.7 9 4.2 2161980 151 69.6 57 26.3 9 4.1 2171981 157 67.4 62 26.6 14 6.0 2331982 207 65.7 77 24.4 31 0.9 3151983 227 60.5 118 31.5 30 8.0 3751984 241 56.0 148 34.4 41 9.6 4301985 232 50.0 166 35.8 66 14.2 4641986 275 49.5 203 36.5 78 14.0 5561987 320 53.0 211 35.0 72 12.0 6031988 365 50.8 236 32.9 117 16.3 718% change -23.3 +11.2 +12.1TABLE 23ANNUAL STANDARDIZED INCIDENCE RATIOS AND SEX ADJUSTED RATES FORSUBGROUPS OF THE OVER-65 POPULATION FOR ISOLATED CORONARY ARTERYBYPASS SURGERYB.C. 1979-1988YearAge Group65-69 years 70-74 years^75+ yearsSIR Rate* SIR Rate SIR Rate1979 0.78 16.04 0.44 6.47 0.24 0.871980 0.70 14.42 0.50 7.49 0.23 0.841981 0.70 14.33 0.52 7.73 0.27 0.981982 0.91 18.68 0.63 9.36 0.73 2.671983 0.98 20.14 0.90 13.41 0.69 2.501984 1.07 21.94 1.08 16.08 0.90 3.291985 1.00 20.48 1.17 17.34 1.38 5.041986 1.11 22.79 1.38 20.51 1.42 5.181987 1.25 25.62 1.40 20.87 1.37 5.011988 1.36 27.94 1.56 23.15 2.14 7.7810-yr rate 20.51 14.88 3.64* Rate per 10,000 population = SIR x 10-year rate for age-group227TABLE 24ANNUAL NUMBERS OF PATIENTS WITH COMORBIDITY RECEIVING ISOLATED CABSB.C. 1979-1988Year Diabetes COPD TotalComorbidity% of TotalCasesTotal Cases1979 30 0 30 3.20 9311980 18 0 18 1.93 9311981 25 0 25 2.60 %21982 38 0 38 3.62 10501983 65 45 110 9.02 12201984 93 29 122 10.22 11941985 90 31 121 10.25 11811986 113 44 157 12.35 12711987 144 55 199 15.02 13251988 182 65 247 16.51 14%TABLE 25DISTRIBUTION OF COMORBIDITY BY AGE GROUP ISOLATED CORONARY ARTERY BYPASSSURGERYB.C. 1979-1988Age Group Diabetes COPD TotalComorbidityTotal CABSCases inAge Group%^of^CasesWithComorbidity20-24 0 0 0 1 0.0025-29 1 0 1 10 10.0030-34 2 0 2 49 4.0835-39 11 1 12 182 6.5940-44 13 4 17 419 4.0545-54 58 8 66 845 7.8155-59 82 28 110 1473 7.4660-64 .^114 46 160 2033 7.8765-69 167 54 221 2422 9.1270-74 188 57 245 2355 10.4075-79 124 53 177 1325 13.3580-84 31 15 46 404 11.3885-89 7 3 10 59 16.99Total 798 269 1067 11568 9.22228TABLE 26OUT-OF-PROVINCE REVASCULARIZATION SERVICES FOR B.C. RESIDENTS1979-1988Year CABS PTCA Total1979 23 0 231980 23 0 231981 24 0 241982 29 0 291983 36 0 361984 32 0 321985 31 4 351986 38 7 451987 Not known Not known -1988 22 43 65Total 258 54 312Table 27NUMBERS AND MEAN AGE OF B.C. RESIDENTS RECEIVING CABS IN ALBERTAYear TotalN Mean Age Std1983 5 58.2 8.921984 8 64.6 6.061985 3 60.6 2.081986 3 62.0 13.41987 28 61.8 9.441988 28 61.8 9.44229TABLE 28ANNUAL DISTRIBUTION OF ISOLATED CABS BY SCHOOL DISTRICTB.C. 1979-1988School 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 TotalDistrict1 1 2 1 1 1 1 0 1 0 1 92 4 1 4 4 5 5 5 5 3 3 393 0 1 3 4 2 2 4 1 4 4 254 0 0 0 00 0 2 1 0 0 1 47 12 6 8 5 9 12 11 10 19 13 1059 5 2 1 1 2 2 5 8 6 8 4010 5 1 3 1 0 5 3 4 5 0 2711 5 4 8 6 5 5 1 10 9 14 6712 3 1 1 3 2 4 4 0 6 6 3013 0 0 1 1 1 0 0 1 3 5 1214 4 5 5 10 7 15 13 15 19 13 10615 16 9 12 13 21 27 22 23 32 32 20716 3 2 2 1 3 3 3 3 7 4 3117 2 2 2 2 1 5 2 5 2 3 2618 0 1 0 1 0 0 0 0 0 0 219 1 0 2 1 4 4 2 2 1 1 1820 0 0 0 0 0 0 0 0 0 0 021 2 2 2 1 3 4 2 1 5 4 2622 13 8 15 10 16 23 15 18 19 32 16923 28 22 40 74 71 81 99 83 76 85 65924 13 17 17 17 17 22 17 18 23 29 19026 1 1 1 0 1 1 1 1 0 1 827 9 9 7 6 6 4 9 11 16 12 8928 0 1 5 6 6 5 10 11 5 10 5929 2 0 0 1 0 1 0 1 0 1 630 0 4 1 0 1 5 0 3 2 2 1831 1 2 2 1 3 5 3 3 3 1 2432 0 3 2 2 1 4 6 2 3 5 2833 12 20 13 18 12 14 13 16 32 23 17334 21 17 21 26 24 19 24 28 26 33 23935 14 13 17 19 23 20 22 33 25 29 21536 47 50 64 59 83 70 74 111 94 113 76537 17 24 21 32 32 20 27 33 33 32 27138 23 33 32 28 31 29 34 32 50 50 34239 125 106 127 123 151 135 182 152 167 206 147440 22 22 11 35 21 28 17 17 18 23 21441 44 42 50 42 54 57 46 58 60 71 52442 18 13 16 23 26 16 18 17 16 23 18643 39 41 27 43 47 40 52 45 40 50 42444 23 31 29 18 37 39 38 33 37 46 33145 6 20 14 15 9 24 12 22 14 23 15946 5 2 5 5 5 12 8 6 10 12 7047 2 7 5 3 10 6 6 5 12 14 7048 2 6 1 3 4 5 5 5 5 9 4549 0 0 0 0 2 1 0 0 0 1 450 1 1 0 1 1 2 1 2 2 1 1252 4 4 2 3 3 3 4 5 12 7 4754 3 2 0 3 4 4 2 5 6 2 3155 1 0 6 0 0 2 0 0 1 1 11230TABLE 28 (contd.)ANNUAL DISTRIBUTION OF ISOLATED CABS BY SCHOOL DISTRICTB.C. 1979-1988School^1979^1980^1981 1982 1983 1984 1985 1986 1987 1988 Tota1District56^1^1^3 3 0 8 4 1 7 2 3057 19^18^18 19 19 20 25 21 16 19 19459^0^2^1 3 1 3 2 4 3 4 2360 2^2^0 1 3 4 4 8 5 2 3161^147^140^141 165 149 124 97 124 125 131 134362 21^22^17 14 30 26 15 16 27 19 20763^20^24^27 36 30 31 25 24 23 30 27064 3^8^12 2 6 8 6 4 9 11 6965^15^16^13 17 20 22 22 24 15 20 18466 5^1^1 4 2 5 2 3 1 1 2568^57^47^42 35 36 34 37 26 37 33 38469 15^12^9 9 28 18 16 23 13 21 16470^12^13^10 11 20 7 9 16 10 14 12271 14^17^12 8 28 17 10 22 16 16 16072^7^9^9 7 10 12 12 9 9 9 9375 2^5^11 5 6 8 9 16 13 19 9476^2^2^3 0 3 0 2 0 2 4 1877 1^1^2 3 7 5 4 9 6 6 4480^3^3^3 5 3 6 11 4 9 6 5381 0^2^1 1 1 0 0 0 1 1 784^1^1^0 0 0 0 0 3 2 2 985 3^1^0 1 0 4 3 1 5 3 2186^1^2^1 2 5 4 4 1 10 6 3687 0^0^1 0 0 1 0 1 1 0 488^2^5^4 8 8 6 4 9 3 14 6389 10^10^6 9 6 12 17 14 9 15 1080 *^9^7^9 11 27 16 18 23 20 29 169Annual^931^931^962 1050 1220 1194 1181 1271 1325 1496 1155Total 6*Represents non-residents of B.C.231TABLE 29ANNUAL POPULATIONS, RATES AND STANDARDIZED INCIDENCE RATIOS PER 10,000POPULATION BY SCHOOL DISTRICT B.C. 1979-1988School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate1 Fernie 1979 9347 1 0.24 1.07 1.361 1980 9909 2 0.47 2.02 2.681 1981 10570 1 0.23 0.95 1.311 1982 11578 1 0.21 0.86 1.221 1983 11791 1 0.21 0.85 1.191 1984 11468 1 0.21 0.87 1.201 1985 11259 0 0.00 0.00 0.001 1986 10815 1 0.22 0.92 1.231 1987 10324 0 0.00 0.00 0.001 1988 9900 1 0.23 1.01 1.292 Cranbrook 1979 12543 4 0.66 3.19 3.792 1980 12943 1 0.16 0.77 0.932 1981 13275 4 0.65 3.01 3.702 1982 13949 4 0.60 2.87 3.432 1983 14099 5 0.73 3.55 4.162 1984 14284 5 0.71 3.50 4.032 1985 14177 5 0.69 3.53 3.942 1986 14090 5 0.70 3.55 3.972 1987 11733 3 0.45 2.56 2.562 1988 13522 3 0.41 2.22 2.333 Kimberley 1979 5920 0 0.00 0.00 0.003 1980 6109 1 0.25 1.64 1.443 1981 6410 3 0.74 4.68 4.203 1982 6368 4 0.98 6.28 5.603 1983 6247 2 0.49 3.20 2.793 1984 6266 2 0.49 3.19 2.793 1985 6211 4 0.97 6.44 5.533 1986 6035 1 0.25 1.66 1.423 1987 9617 4 0.93 4.16 5.303 1988 5415 4 1.05 7.39 5.97232TABLE 29 (contd.)School District Year Population ObservedCABSS .I.R. CrudeRateAge-SexAdjustedRate4 Windermere 1979 3738 0 0.00 0.00 0.004 1980 4030 0 0.00 0.00 0.004 1981 4380 0 0.00 0.00 0.004 1982 4495 0 0.00 0.00 0.004 1983 4599 0 0.00 0.00 0.004 1984 4476 2 0.80 4.47 4.584 1985 4539 1 0.38 2.20 2.194 1986 4395 0 0.00 0.00 0.004 1987 5069 0 0.00 0.00 0.004 1988 4505 1 0.38 2.22 2.187 Nelson 1979 13981 12 1.42 8.58 8.087 1980 14412 6 0.7 4.16 3.967 1981 14675 8 0.93 5.45 5.337 1982 14850 5 0.57 3.37 3.267 1983 15148 9 1.00 5.94 5.697 1984 15063 12 1.33 7.97 7.567 1985 14559 11 1.23 7.56 7.007 1986 13885 10 1.13 7.20 6.437 1987 13816 19 2.11 13.75 12.037 1988 13878 13 1.42 9.37 8.099 Castlegar 1979 7694 5 1.07 6.50 6.109 1980 7956 2 0.42 2.51 2.419 1981 8260 1 0.21 1.21 1.209 1982 8209 1 0.21 1.22 1.179 1983 8244 2 0.41 2.43 2.319 1984 8165 2 0.40 2.45 2.299 1985 8105 5 1.00 6.17 5.719 1986 7985 8 1.63 10.02 9.279 1987 7729 6 1.24 7.76 7.089 1988 7711 8 1.62 1037 9.25233TABLE 29 (contd.)School District Year Population ObservedCABSS.I.R. CrudeRateAge-SexAdjustedRate10 Arrow Lakes 1979 2941 5 2.67 17.00 15.2410 1980 3027 1 0.52 3.30 2.9410 1981 3175 3 1.58 9.45 9.0010 1982 3244 1 0.49 3.08 2.7810 1983 3242 0 0.00 0.00 0.0010 1984 3256 5 2.45 15.36 13.9710 1985 3209 3 1.46 9.35 8.3410 1986 3100 4 1.94 12.90 11.0710 1987 3004 5 2.55 16.64 14.5410 1988 2942 0 0.00 0.00 0.0011^Trail 1979 15149 5 0.50 3.30 2.8411 1980 15367 4 0.40 2.60 2.2811 1981 15910 8 0.79 5.03 4.5211 1982 15741 6 0.60 3.81 3.4411 1983 15354 5 0.51 3.26 2.9011 1984 15171 5 0.51 3.30 2.9411 1985 14947 1 0.10 0.67 0.5911 1986 14370 10 1.05 6.96 6.0111 1987 13927 9 0.97 6.46 5.5211 1988 13788 14 1.50 10.15 8.5612 Grand Forks 1979 4618 3 0.90 6.50 5.1212 1980 4775 1 0.29 2.09 1.6612 1981 5080 1 0.27 1.97 1.5412 1982 5143 3 0.80 5.83 4.5712 1983 5221 2 0.53 3.83 3.0112 1984 5214 4 1.06 7.67 6.0512 1985 5213 4 1.06 7.67 6.0212 1986 5060 0 0.00 0.00 0.0012 1987 5068 6 1.60 11.84 9.1412 1988 5081 6 1.58 11.81 9.00TABLE23429 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate13 Kettle 1979 2132 0 0.00 0.00 0.00Valley13 1980 2148 0 0.00 0.00 0.0013 1981 2170 1 0.65 4.61 3.6813 1982 2221 1 0.63 4.50 3.5813 1983 2197 1 0.64 4.55 3.6313 1984 2223 0 0.00 0.00 0.0013 1985 2174 0 0.00 0.00 0.0013 1986 2130 1 0.63 4.69 3.5813 1987 2099 3 2.00 14.29 11.413 1988 2077 5 3.36 24.07 19.1314 S. Okanagan 1979 8793 4 0.58 4.55 3.2814 1980 9389 5 0.68 5.33 3.8614 1981 10115 5 0.61 4.94 3.4714 1982 10545 10 1.16 9.48 6.6014 1983 10735 7 0.80 6.52 4.5414 1984 10590 15 1.72 14.16 9.8214 1985 10467 13 1.49 12.42 8.4914 1986 10405 14 1.61 13.46 9.2014 1987 10131 19 2.17 18.75 12.3614 1988 10220 13 1.47 12.72 8.3615 Penticton 1979 18716 16 1.23 8.55 7.0215 1980 19144 9 0.67 4.70 3.8115 1981 19925 12 0.83 6.02 4.7315 1982 20618 13 0.87 6.31 4.9715 1983 21262 21 1.37 9.88 7.8015 1984 21783 27 1.72 12.39 9.7915 1985 21586 22 1.40 10.19 7.9915 1986 21935 23 1.43 10.49 8.1715 1987 21999 32 1.97 14.55 11.2215 1988 22341 32 1.93 14.32 11.01235TABLE 29 (contd.)School District Year Population ObservedCABSS .I.R. CrudeRateAge-SexAdjustedRate16 Keremeos 1979 2333 3 1.52 12.86 8.6816 1980 2387 2 1.02 8.38 5.8216 1981 2585 2 0.94 7.74 5.3816 1982 2644 1 0.46 3.78 2.6116 1983 2623 3 1.38 11.44 7.8816 1984 2619 3 1.36 11.45 7.7716 1985 2617 3 1.35 11.46 7.7016 1986 2580 2 0.90 7.75 5.1116 1987 2563 7 3.17 27.31 18.0516 1988 2694 4 1.79 14.85 10.1817 Princeton 1979 3112 2 1.07 6.43 6.1017 1980 3142 2 1.05 6.37 5.9717 1981 3170 2 1.04 6.31 5.9417 1982 3265 2 0.99 6.13 5.6217 1983 3264 1 0.48 3.06 2.7117 1984 3348 5 2.34 14.93 13.3217 1985 3343 2 0.91 5.98 5.1817 1986 3305 5 2.39 15.13 13.6417 1987 3280 2 0.96 6.10 5.4517 1988 3317 3 1.41 9.04 8.0318^Golden 1979 3796 0 0.00 0.00 0.0018 1980 3935 1 0.56 2.54 3.2018 1981 4165 0 0.00 0.00 0.0018 1982 4109 1 0.53 2.43 3.0518 1983 4099 0 0.00 0.00 0.0018 1984 4230 0 0.00 0.00 0.0018 1985 4433 0 0.00 0.00 0.0018 1986 4500 0 0.00 0.00 0.0018 1987 4661 0 0.00 0.00 0.0018 1988 4870 0 0.00 0.00 0.00TABLE23629 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate19 Revelstoke 1979 6037 1 0.36 1.66 2.0719 1980 6265 0 0.00 0.00 0.0019 1981 6460 2 0.70 3.10 4.0019 1982 6588 1 0.34 1.52 1.9119 1983 6497 4 1.34 6.16 7.6319 1984 6397 4 1.36 6.25 7.7619 1985 6194 2 0.69 3.23 3.9419 1986 6010 2 0.69 3.33 3.9219 1987 5927 1 0.35 1.69 2.0019 1988 5774 1 0.35 1.73 2.0021 Armstrong- 1979 4072 2 0.78 4.91 4.47Spallumcheen21 1980 4287 2 0.75 4.67 4.3021 1981 4485 2 0.69 4.46 3.9621 1982 4687 1 0.33 2.13 1.8921 1983 4713 3 0.98 6.37 5.6121 1984 4758 4 1.27 8.41 7.2621 1985 4776 2 0.63 4.19 3.6121 1986 4720 1 0.31 2.12 1.7621 1987 4795 5 1.59 10.43 9.0521 1988 4794 4 1.25 8.34 7.1022 Vernon 1979 26410 13 0.81 4.92 4.6122 1980 27267 8 0.48 2.93 2.7422 1981 28370 15 0.87 5.29 4.9822 1982 29525 10 0.56 3.39 3.1822 1983 29809 16 0.87 5.37 4.9822 1984 30263 23 1.22 7.60 6.9722 1985 30504 15 0.78 4.92 4.4422 1986 30375 18 0.91 5.93 5.2122 1987 30398 19 0.96 6.25 5.4722 1988 30783 32 1.59 10.4 9.06237TABLE 29 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate23 Central Okanagan 1979 52846 28 0.78 5.30 4.4723 1980 56303 22 0.57 3.91 3.2523 1981 59925 39 0.97 6.51 5.5123 1982 61691 74 1.77 12.00 10.1123 1983 63062 71 1.66 11.26 9.4723 1984 64309 81 1.85 12.6 10.5623 1985 64786 99 2.23 15.28 12.6923 1986 65880 83 1.81 12.60 10.3323 1987 66884 76 1.63 11.36 9.2823 1988 68525 87 1.81 12.70 10.3424 Kamloops 1979 46669 13 0.57 2.79 3.2624 1980 48512 17 0.72 3.50 4.1124 1981 51320 17 0.69 3.31 3.9524 1982 52425 17 0.67 3.24 3.7924 1983 52846 17 0.65 3.22 3.6924 1984 53019 22 0.82 4.15 4.6524 1985 52603 17 0.62 3.23 3.5324 1986 51170 18 0.65 3.52 3.6924 1987 52921 23 0.81 4.35 4.6024 1988 53119 29 0.99 5.46 5.6326 .N.Thompson 1979 2967 1 0.72 3.37 4.1326 1980 3072 1 0.71 3.26 4.0726 1981 3150 1 0.70 3.17 3.9926 1982 3096 0 0.00 0.00 0.0026 1983 3045 1 0.68 3.28 3.8526 1984 3080 1 0.66 3.25 3.7726 1985 3117 1 0.65 3.21 3.6826 1986 2975 1 0.67 3.36 3.8326 1987 3042 0 0.00 0.00 0.0026 1988 2927 1 0.65 3.42 3.68238TABLE 29 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate27 Cariboo- 1979 19651 9 1.01 4.58 5.75Chilcotin27 1980 20695 9 0.96 4.35 5.4627 1981 22250 7 0.67 3.15 3.8227 1982 23194 6 0.55 2.59 3.1527 1983 23589 6 0.53 2.54 3.0227 1984 24056 4 0.34 1.66 1.9227 1985 23894 9 0.74 3.77 4.2227 1986 23140 11 0.91 4.75 5.1927 1987 24227 16 1.27 6.60 7.2627 1988 23579 13 1.04 5.51 5.9228 Quesnel 1979 13032 0 0.00 0.00 0.0028 1980 13668 1 0.16 0.73 0.8928 1981 14175 5 0.75 3.53 4.2728 1982 14803 6 0.87 4.05 4.9928 1983 14912 6 0.85 4.02 4.8328 1984 15093 5 0.68 3.31 3.9028 1985 15198 10 1.32 6.58 7.5428 1986 15070 11 1.44 7.30 8.2228 1987 15570 5 0.65 3.21 3.7028 1988 15415 10 1.28 6.49 7.3129 Lilooet 1979 2425 2 1.49 8.25 8.5129 1980 2491 0 0.00 0.00 0.0029 1981 2775 0 0.00 0.00 0.0029 1982 2768 1 0.65 3.61 3.7329 1983 2792 0 0.00 0.00 0.0029 1984 2861 1 0.62 3.50 3.5429 1985 2916 0 0.00 0.00 0.0029 1986 2450 1 0.71 4.08 4.0729 1987 2926 0 0.00 0.00 0.0029. 1988 2946. 1 0.61 3.39 3.48239TABLE 29 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate30 S.Cariboo 1979 5068 0 0.00 0.00 0.0030 1980 5187 4 1.34 7.71 7.6530 1981 5100 1 0.34 1.96 1.9730 1982 5395 0 0.00 0.00 0.0030 1983 5395 1 0.33 1.85 1.8830 1984 5503 5 1.61 9.09 9.1630 1985 5241 0 0.00 0.00 0.0030 1986 4880 3 1.02 6.15 5.830 1987 4849 2 0.70 4.12 4.0130. 1988 4677. 2. 0.70 4.28 4.0131 Merritt 1979 5789. 1. 0.32 1.73 1.8231. 1980 5840. 2. 0.63 3.42 3.5831. 1981 6030. 2. 0.60 3.32 3.4431. 1982 6086. 1. 0.31 1.64 1.7431. 1983 6130. 3. 0.89 4.89 5.0731. 1984 6113. 5. 1.47 8.18 8.3631. 1985 6234. 3. 0.86 4.81 4.9331. 1986 6405. 3. 0.81 4.68 4.6331 1987 6504 3 0.83 4.61 4.7531 1988 6463 1 0.27 1.55 1.5632 Hope 1979 4566 0 0.00 0.00 0.0032 1980 4674 3 0.98 6.42 5.5732 1981 4895 1 0.31 2.04 1.7932 1982 5036 2 0.62 3.97 3.5132 1983 5060 1 0.31 1.98 1.7532 1984 5029 4 1.23 7.95 7.0432 1985 5150 6 1.79 11.65 10.2132 1986 4795 2 0.64 4.17 3.6732 1987 4893 3 0.93 6.13 5.2932 1988 4994 5 1.53 10.01 8.74TABLE24029 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate33 Chilliwack 1979 28118 12 0.66 4.27 3.7633 1980 29322 20 1.05 6.82 5.9633 1981 30465 13 0.66 4.27 3.7433 1982 30872 18 0.90 5.83 5.1533 1983 30903 12 0.60 3.88 3.4133 1984 32833 14 0.67 4.26 3.8033 1985 31722 13 0.63 4.10 3.6033 1986 32090 16 0.77 4.99 4.3933 1987 32765 32 1.50 9.77 85833 1988 34036 22 0.99 6.46 5.6534 Abbotsford 1979 31688 21 1.13 6.63 6.4634 1980 34065 17 0.86 4.99 4.9034 1981 36860 21 0.99 5.70 5.6634 1982 40162 26 1.16 6.47 6.5834 1983 41950 24 1.02 5.72 5.7934 1984 43640 19. 0.77 4.35 4.4134 1985 45070 24 0.94 5.33 5.3734 1986 45400 28 1.07 6.17 6.1234 1987 47535 26 0.97 5.47 5.5034 1988 49906 33 1.18 6.61 6.7335 Langley 1979 35026 14 0.77 4.00 4.3935 1980 36850 13 0.68 3.53 3.8935 1981 38780 17 0.84 4.38 4.8235 1982 39504 19 0.92 4.81 5.2235 1983 41110 23 1.07 5.59 6.0735 1984 43224 20 0.88 4.63 5.0435 1985 44967 22 0.93 4.89 5.2935 1986 46855 32 1.28 6.83 7.3135 1987 49500 25 0.97 5.05 55135 1988 51395 29 1.07 5.64 6.12241TABLE 29 (contd.)School District Year Population ObservedCABSS.I.R. CrudeRateAge-SexAdjustedRate36 Surrey 1979 96867 47 0.88 4.85 4.9936 1980 103734 50 0.88 4.82 5.0136 1981 109725 64 1.07 5.83 6.0936 1982 114954 59 0.94 5.13 5.3836 1983 120920 83 1.27 6.86 7.2336 1984 126917 70 1.02 5.52 5.8236 1985 131855 74 1.03 5.61 5.8936 1986 137170 110 1.46 8.02 8.3336 1987 145841 94 1.19 6.45 6.8036 1988 153070 113 1.37 7.38 7.7937 Delta 1979 42638 17 0.87 3.99 4.9537. 1980 45159. 24. 1.13 5.31 6.4537. 1981 46915. 21. 0.94 4.48 5.3437. 1982 48240. 32. 1.37 6.63 7.7937. 1983 49741. 32. 1.31 6.43 7.4637. 1984 50579. 20. 0.79 3.95 4.5037. 1985 51604. 27. 1.02 5.23 5.8337. 1986 52920. 33. 1.19 6.24 6.8037. 1987 54161. 33. 1.16 6.09 6.6237. 1988 55178. 32. 1.07 5.80 6.1238 Richmond 1979 59641 23 0.76 3.86 4.3238 1980 63945 33 1.01 5.16 5.7638 1981 66440 32 0.93 4.82 5.3138 1982 68740 28 0.79 4.07 4.4938 1983 71182 31 0.84 4.36 4.7838 1984 73659 29 0.76 3.94 4.3038 1985 76352 34 0.85 4.45 4.8338 1986 78590 31 0.74 3.94 4.2238 1987 81541 50 1.16 6.13 6.5938 1988 84116 50 1.11 5.94 6.35TABLE24229 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate39 Vancouver 1979 318610 125 0.65 3.92 3.6939 1980 325747 106 0.55 3.25 3.1139 1981 329515 127 0.65 3.85 3.7339 1982 334452 123 0.63 3.68 3.6139 1983 337823 151 0.78 4.47 4.4439 1984 341847 135 0.70 3.95 3.9739 1985 346142 182 0.94 5.26 5.3639 1986 349095 151 0.78 4.33 4.4339 1987 350030 167 0.84 4.77 4.7739 1988 356688 207 1.02 5.80 5.8340 New Westminster 1979 29722 22 1.19 7.40 6.7740 1980 30556 22 1.18 7.20 6.7040 1981 31270 11 0.58 3.52 3.3040 1982 31214 35 1.89 11.21 10.7740 1983 31645 21 1.15 6.64 6.5340 1984 31890 28 1.55 8.78 8.8140 1985 32619 17 0.93 5.21 5.3040 1986 32875 17 0.94 5.17 5.3440 1987 32535 18 0.93 5.53 5.3240 1988 33568 23 1.18 6.85 6.7141 Burnaby 1979 98365 44 0.77 4.47 4.3841 1980 100905 42 0.72 4.16 4.0941 1981 102800 50 0.84 4.86 4.7941 1982 104828 42 0.69 4.01 3.9541 1983 106972 54 0.87 5.05 4.9841 1984 109694 57 0.91 5.20 5.1641 1985 111604 46 0.72 4.12 4.1041 1986 113510 57 0.87 5.02 4.9841 1987 114094 60 0.90 5.26 5.141 1988 116397 71 1.04 6.10 5.92TABLE24329 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate42 Maple Ridge 1979 24232 18 1.4 7.43 8.042 1980 25100 13 0.99 5.18 5.6642 1981 25650 16 1.20 6.24 6.8442 1982 25958 23 1.69 8.86 9.6542 1983 27192 26 1.85 9.56 10.5442 1984 28104 16 1.10 5.69 6.2942 1985 28951 18 1.20 6.22 6.8542 1986 30100 17 1.10 5.65 6.2642 1987 32563 16 0.98 4.91 5.5642 1988 34947 23 1.31 6.58 7.4743 Coquitlam 1979 62661 39 1.36 6.22 7.7643 1980 66616 41 1.34 6.15 7.6343 1981 69950 27 0.85 3.86 4.8743 1982 70887 43 1.32 6.07 7.5043 1983 73756 47 1.39 6.37 7.9043 1984 77171 40 1.14 5.18 6.5043 1985 78809 52 1.42 6.60 8.0843 1986 80700 45 1.17 5.58 6.6543 1987 84253 40 0.99 4.75 5.6443 1988 88087 50 1.17 5.68 6.6944 N.Vancouver 1979 69154 23 0.64 3.33 3.6444 1980 71366 31 0.83 4.34 4.7344 1981 72910 29 0.75 3.98 4.3044 1982 73892 18 0.46 2.44 2.6244 1983 74701 37 0.93 4.95 5.2944 1984 75932 39 0.96 5.14 5.4744 1985 77402 38 0.91 4.91 5.2044 1986 78840 33 0.77 4.19 4.3744 1987 80291 37 0.84 4.61 4.7744. 1988 82325 46 1.01 5.59 5.73TABLE24429 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate45 W.Vancouver 1979 28824 6 0.29 2.08 1.6545 1980 29456 20 0.95 6.79 5.3945 1981 29710 14 0.65 4.71 3.7145 1982 29994 15 0.69 5.00 3.9345 1983 30618 9 0.41 2.94 2.3145 1984 31163 24 1.06 7.70 6.0545 1985 31683 12 0.52 3.79 2.9545 1986 31915 22 0.94 6.89 5.3445 1987 31830 14 0.58 4.40 3.3145 1988 32495 23 0.94 7.08 5.3646 Sunshine Coast 1979 9604 5 0.71 5.21 4.0746 1980 10272 2 0.27 1.95 1.5446 1981 11275 5 0.63 4.43 3.5646 1982 11738 5 0.59 4.26 3.3746 1983 12026 5 0.57 4.16 3.2846 1984 12182 12 1.35 9.85 7.7246 1985 12389 8 0.88 6.46 5.0346 1986 12235 6 0.66 4.90 3.7546 1987 12291 10 1.08 8.14 6.1646 1988 12511 12 1.28 9.59 7.3247 Powell River 1979 12219 2 0.27 1.64 1.5447 1980 12427 7 0.93 5.63 5.2947 1981 12830 5 0.64 3.90 3.6347 1982 12839 3 0.38 2.34 2.1647 1983 12832 10 1.25 7.79 7.1447 1984 12777 6 0.75 4.70 4.2647 1985 12744 6 0.74 4.71 4.2147 1986 12835 5 0.60 3.90 3.4247 1987 12348 12 1.51 9.72 8.5847 1988 12531 14 1.72 11.17 9.83TABLE24529 (contd.)School District Year Population Observed S .I.R. Crude Age-SexCABS Rate AdjustedRate48 Howe Sound 1979 8390 2 0.55 2.38 3.1648 1980 8886 6 1.55 6.75 8.8448 1981 9485 1 0.25 1.05 1.4048 1982 9578 3 0.71 3.13 4.0248 1983 9354 4 0.94 4.28 5.3848 1984 9764 5 1.12 5.12 6.3848 1985 9687 5 1.09 5.16 6.2148 1986 9600 5 1.08 5.21 6.1748 1987 10825 5 0.99 4.62 5.6448 1988 11395 10 1.89 8.78 10.7549 Central Coast 1979 2271 0 0.00 0.00 0.0049 1980 2222 0 0.00 0.00 0.0049 1981 1895 0 0.00 0.00 0.0049 1982 1889 0 0.00 0.00 0.0049 1983 1928 2 2.06 10.37 11.7549 1984 1963 1 1.01 5.09 5.7649 1985 2016 0 0.00 0.00 0.0049 1986 2010 0 0.00 0.00 0.0049 1987 2146 0 0.00 0.00 0.0049 1988 2102 1 0.98 4.76 5.5950 Queen Charlotte 1979 3266 1 0.91 3.06 5.1850 1980 3382 1 0.86 2.96 4.9150 1981 3520 0 0.00 0.00 0.0050 1982 3571 1 0.79 2.8 4.4950 1983 3557 1 0.78 2.81 4.4550 1984 3537 2 1.53 5.65 8.7050 1985 3588 1 0.72 2.79 4.1350 1986 3585 2 1.44 5.58 8.2050 1987 3690 2 1.44 5.42 8.2050 1988 3630 1 0.72 2.75 4.10TABLE24629 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate52 Prince Rupert 1979 10709 4 0.85 3.74 4.8552 1980 11108 4 0.85 3.60 4.8252 1981 11635 2 0.42 1.72 2.3852 1982 11817 3 0.61 2.54 3.5052 1983 11941 3 0.6 2.51 3.4152 1984 11973 3 0.59 2.51 3.3752 1985 11830 4 0.77 3.38 4.4152 1986 11535 5 0.94 4.33 5.3652 1987 11799 12 2.27 10.17 12.9352 1988 11843 7 1.28 5.91 7.3154 Smithers 1979 7085 3 1.03 4.23 5.9054 1980 7641 2 0.68 2.62 3.8654 1981 8420 0 0.00 0.00 0.0054 1982 8959 3 0.84 3.35 4.854 1983 9018 4 1.10 4.44 6.2654 1984 9168 4 1.06 4.36 6.0554 1985 9169 2 0.52 2.18 2.9654 1986 8870 5 1.36 5.64 7.7754 1987 9211 6 1.56 6.51 8.8854 1988 9349 2 0.50 2.14 2.8455 Burns Lake 1979 4443 1 0.51 2.25 2.8955 1980 4704 0 0.00 0.00 0.0055 1981 4780 6 2.88 12.55 16.4455 1982 4714 0 0.00 0.00 0.0055 1983 4704 5 2.29 10.63 13.0755 1984 4644 2 0.90 4.31 5.1455 1985 4708 0 0.00 0.00 0.0055 1986 4645 0 0.00 0.00 0.0055 1987 4829 1 0.44 2.07 2.5055 1988 5001 1 0.42 2.00 2.38TABLE24729 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate56 Nechako 1979 8713 1 0.26 1.15 1.4556 1980 9021 1 0.25 1.11 1.4156 1981 9250 3 0.77 3.24 4.3656 1982 9675 3 0.7 3.1 3.9956 1983 9902 0 0.00 0.00 0.0056 1984 9730 8 1.79 8.22 10.1856 1985 9650 4 0.89 4.15 5.0856 1986 9150 1 0.23 1.09 1.3256 1987 9380 7 1.56 7.46 8.9156 1988 9431 2 0.43 2.12 2.4657 Prince George 1979 51937 19 1. 3.66 5.757 1980 52369 18 0.94 3.44 5.3457 1981 54855 18 0.88 3.28 5.0357 1982 56306 19 0.88 3.37 5.0357 1983 56797 19 0.86 3.35 4.8857 1984 57284 20 0.87 3.49 4.9657 1985 57639 25 1.05 4.34 5.9957 1986 56945 21 0.87 3.69 4.9457 1987 58335 16 0.64 2.74 3.6657 1988 57734 19 0.74 3.29 4.2459 Peace River S. 1979 12381 0 0.00 0.00 0.0059 1980 13414 2 0.31 1.49 1.7659 1981 14365 1 0.15 0.7 0.8659 1982 14621 3 0.44 2.05 2.559 1983 15487 1 0.14 0.65 0.8159 1984 16980 3 0.4 1.77 2.2659 1985 17678 2 0.26 1.13 1.4759 1986 17660 4 0.5 2.27 2.8359 1987 17795 3 0.38 1.69 2.1659 1988 17374 4 0.5 2.3 2.87TABLE24829 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate60 Peace River N. 1979 13225 2 0.39 1.51 2.260 1980 14427 2 0.36 1.39 2.0460 1981 16165 0 0.00 0.00 0.0060 1982 16325 1 0.16 0.61 0.9360 1983 16236 3 0.48 1.85 2.7260 1984 15934 4 0.62 2.51 3.5360 1985 15609 4 0.61 2.56 3.4760 1986 15600 8 1.18 5.13 6.7160 1987 15638 5 0.72 3.2 4.0860 1988 15384 2 0.28 1.3 1.661 Greater Victoria 1979 127330 147 1.79 11.54 10.2161 1980 130019 140 1.69 10.77 9.6261 1981 132515 141 1.68 10.6461 1982 134125 165 1.98 12.3 11.2961 1983 135625 149 1.79 10.99 10.2161 1984 136786 124 1.49 9.07 8.5261 1985 138433 124 1.48 8.96 8.4461 1986 138880 123 1.47 8.86 8.3761 1987 138773 125 1.4 9.01 7.9961 1988 140271 131 1.45 9.34 8.2762 Sooke 1979 21818 21 1.88 9.63 10.7162 1980 22368 22 1.91 9.84 10.8962 1981 23035 17 1.43 7.38 8.1362 1982 23814 14 1.1 5.88 6.2862 1983 24765 30 2.26 12.11 12.962 1984 25437 26 1.89 10.22 10.7562 1985 26240 15 1.05 5.72 5.9662 1986 27155 16 1.06 5.89 6.0362 1987 28145 27 1.76 9.59 10.0162 1988 28973 19 1.19 6.56 6.77TABLE24929 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate63 Saanich 1979 21549 20 1.32 9.28 7.5163 1980 23027 24 1.47 10.42 8.3963 1981 24765 27 1.52 10.9 8.6563 1982 25528 36 1.96 14.1 11.1963 1983 26525 30 1.56 11.31 8.8763 1984 27851 31 1.54 11.13 8.7663 1985 28744 25 1.19 8.7 6.7863 1986 29505 24 1.09 8.13 6.1963 1987 30454 23 1.01 7.55 5.7863 1988 31878 30 1.26 9.41 7.264 Gulf Islands 1979 5391 3 0.62 5.56 3.5164 1980 5727 8 1.55 13.97 8.8264 1981 6295 12 2.11 19.06 12.0464 1982 6553 2 0.34 3.05 1.9264 1983 6723 6 0.99 8.92 5.6364 1984 6888 8 1.29 11.61 7.3864 1985 7059 6 0.95 8.5 5.4164 1986 7055 4 0.63 5.67 3.6164 1987 7002 9 1.41 12.85 8.0464 1988 7253 11 1.65 15.17 9.4365 Cowichan 1979 20826 15 1.2 7.2 6.8265 1980 21483 16 1.25 7.45 7.1565 1981 23010 13 0.94 5.65 5.3365 1982 23772 17 1.19 7.15 6.7865 1983 23978 20 1.37 8.34 7.7865 1984 24185 22 1.47 9.1 8.3965 1985 24451 22 1.44 9. 8.1865 1986 24240 24 1.54 9.9 8.7865 1987 24677 15 0.95 6.08 5.4365 1988 25200 20 1.23 7.94 7.04TABLE25029 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate66 Lake Cowichan 1979 3537 5 2.44 14.14 13.966 1980 3704 1 0.46 2.7 2.6466 1981 3695 1 0.46 2.71 2.666 1982 3537 4 1.85 11.31 10.5666 1983 3520 2 0.9 5.68 5.1166 1984 3562 5 2.18 14.04 12.4566 1985 3578 2 0.84 5.59 4.8166 1986 3460 3 1.32 8.67 7.566 1987 3380 1 0.46 2.96 2.6366 1988 3295 1 0.45 3.03 2.5968 Nanaimo 1979 43021 57 2.11 13.25 12.0368 1980 45571 47 1.67 10.31 9.5268 1981 48785 42 1.41 8.61 8.0168 1982 50327 35 1.14 6.95 6.4968 1983 50763 36 1.14 7.09 6.5268 1984 51572 34 1.06 6.59 6.0368 1985 51850 37 1.13 7.14 6.4268 1986 52120 26 0.77 4.99 4.3868 1987 52342 37 1.09 7.07 6.2268 1988 53638 33 0.95 6.15 5.4169 Qualicum 1979 12021 15 1.48 12.48 8.4369 1980 13215 12 1.12 9.08 6.3769 1981 14490 9 0.77 6.21 4.3769 1982 15283 9 0.73 5.89 4.1469 1983 15584 28 2.19 17.97 12.4769 1984 16018 18 1.37 11.24 7.869 1985 16310 16 1.18 9.81 6.7469 1986 16675 23 1.63 13.79 9.2869 1987 16389 13 0.92 7.93 5.2569 1988 17016 21 1.44 12.34 8.23TABLE25129 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate70 Alberni 1979 20062 12 1.14 5.98 6.570 1980 20415 13 1.21 6.37 6.9370 1981 20745 9 0.83 4.34 4.7170 1982 21083 11 0.98 5.22 5.5870 1983 21062 20 1.74 9.5 9.9370 1984 20964 7 0.6 3.34 3.4370 1985 20796 9 0.76 4.33 4.3470 1986 20390 16 1.34 7.85 7.6270 1987 20608 10 0.82 4.85 4.6870 1988 20398 14 1.13 6.86 6.4471 Courtney 1979 21346 14 1.11 6.56 6.3471 1980 22231 17 1.3 7.65 7.471 1981 23875 12 0.85 5.03 4.8271 1982 24903 8 0.54 3.21 3.0571 1983 25379 28 1.81 11.03 10.3471 1984 25943 17 1.07 6.55 6.1171 1985 26629 10 0.61 3.76 3.4671 1986 26735 22 1.31 8.23 7.4571 1987 26774 16 0.95 5.98 5.4271 1988 27631 16 0.92 5.79 5.2472 Campbell 1979 16586 7 0.83 4.22 4.73River72 1980 17493 9 1.03 5.14 5.8772 1981 18685 9 0.99 4.82 5.6472 1982 19255 7 0.74 3.64 4.272 1983 19460 10 1.04 5.14 5.9172 1984 19755 12 1.21 6.07 6.972 1985 19860 12 1.17 6.04 6.6772 1986 20050 9 0.86 4.49 4.8972 1987 20809 9 0.83 4.33 4.7272 1988 21517 9 0.8 4.18 4.55TABLE25229 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate75 Mission 1979 13760 2 0.25 1.45 1.4175 1980 14386 5 0.6 3.48 3.4175 1981 15400 11 1.24 7.14 7.0475 1982 15674 5 0.56 3.19 3.2175 1983 15915 6 0.66 3.77 3.7675 1984 16363 8 0.86 4.89 4.9275 1985 16734 9 0.94 5.38 5.3775 1986 17000 16 1.66 9.41 9.4675 1987 17601 13 1.3 7.39 7.4375 1988 18324 19 1.83 10.37 10.4276 Agassiz- 1979 2998 2 1.07 6.67 6.1Harrison76 1980 3078 2 1.06 6.5 6.0376 1981 3220 3 1.56 9.32 8.9176 1982 3370 0 0.00 0.00 0.0076 1983 3509 3 1.41 8.55 8.0376 1984 3598 0 0.00 0.00 0.0076 1985 3704 2 0.88 5.4 5.0276 1986 3675 0 0.00 0.00 0.0076 1987 3838 2 0.86 5.21 4.9176 1988 3844 4 1.69 10.41 9.6677 Summerland 1979 5131 1 0.24 1.95 1.3777 1980 5370 1 0.23 1.86 1.3377 1981 5630 2 0.44 3.55 2.5177 1982 5789 3 0.65 5.18 3.6877 1983 5861 7 1.48 11.94 8.4277 1984 5913 5 1.04 8.46 5.9577 1985 5997 4 0.82 6.67 4.6977 1986 6035 9 1.86 14.91 10.677 1987 6094 5 0.97 8.2 5.5677 1988 6257 6 1.15 9.59 6.55TABLE25329 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate80 Kitimat 1979 8004 3 1.09 3.75 6.280 1980 8265 3 1.02 3.63 5.8480 1981 8740 3 0.97 3.43 5.5380 1982 8989 5 1.54 5.56 8.7780 1983 8606 3 0.9 3.49 5.1480 1984 8397 6 1.74 7.15 9.9180 1985 8433 11 3.13 13.04 17.8680 1986 8080 4 1.11 4.95 6.3380 1987 8253 9 2.39 10.91 13.6180 1988 8117 5 1.29 6.16 7.3681 Fort Nelson 1979 2889 0 0.00 0.00 0.0081 1980 3086 2 2. 6.48 11.481 1981 3070 1 0.97 3.26 5.5381 1982 3174 1 1.04 3.15 5.9481 1983 3175 1 1.03 3.15 5.8881 1984 3163 0 0.00 0.00 0.0081 1985 3197 0 0.00 0.00 0.0081 1986 3185 0 0.00 0.00 0.0081 1987 3381 1 0.93 2.96 5.3381 1988 3192 1 0.97 3.13 5.5384 Van Isl North 1979 2871 1 1. 3.48 5.784 1980 2899 1 1.06 3.45 6.0684 1981 3000 0 0.00 0.00 0.0084 1982 2950 0 0.00 0.00 0.0084 1983 2803 0 0.00 0.00 0.0084 1984 2681 0 0.00 0.00 0.0084 1985 2594 0 0.00 0.00 0.0084 1986 2420 3 3.19 12.4 18.1984 1987 2435 2 2.17 8.21 12.3984 1988 2481 2 2.02 8.06 11.52TABLE25429 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate85 Van Isl West 1979 8113 3 1.04 3.7 5.9285 1980 8606 1 0.33 1.16 1.8685 1981 9095 0 0.00 0.00 0.0085 1982 9601 1 0.31 1.04 1.7985 1983 9684 0 0.00 0.00 0.0085 1984 9780 4 1.17 4.09 6.6785 1985 9712 3 0.85 3.09 4.8785 1986 9385 1 0.28 1.07 1.6185 1987 9683 5 1.35 5.16 7.6885 1988 9658 3 0.8 3.11 4.5786 Creston-Kaslo 1979 9047 1 0.14 1.11 0.886 1980 9293 2 0.27 2.15 1.5586 1981 9545 1 0.13 1.05 0.7586 1982 9521 2 0.26 2.1 1.4886 1983 9620 5 0.65 5.2 3.786 1984 9634 4 0.52 4.15 2.9686 1985 9650 4 0.53 4.15 3.86 1986 9680 1 0.13 1.03 0.7486 1987 10335 10 1.32 9.68 7.5486 1988 9289 6 0.8 6.46 4.5587 Stikine 1979 1238 0 0.00 0.00 0.0087 1980 1176 0 0.00 0.00 0.0087 1981 1220 1 2.5 8.2 14.2587 1982 1238 0 0.00 0.00 0.0087 1983 1283 0 0.00 0.00 0.0087 1984 1295 1 2.22 7.72 12.6787 1985 1298 0 0.00 0.00 0.0087 1986 1260 1 1.96 7.94 11.1887 1987 1218 1 2.27 8.21 12.9587 1988 1142 0 0.00 0.00 0.00TABLE25529 (contd.)School District Year Population Observed S.I.R. Crude Age-SexCABS Rate AdjustedRate88 Terrace 1979 14332 2 0.35 1.4 1.9988 1980 15423 5 0.79 3.24 4.5188 1981 16180 4 0.65 2.47 3.6988 1982 16960 8 1.22 4.72 6.9588 1983 16911 8 1.18 4.73 6.7188 1984 16741 6 0.86 3.58 4.9188 1985 16524 4 0.56 2.42 3.1888 1986 16255 9 1.21 5.54 6.8888 1987 17347 3 0.4 1.73 2.2988 1988 17540 14 1.8 7.98 10.2689 Shuswap 1979 17988 10 0.77 5.56 4.3789 1980 18634 10 0.75 5.37 4.2589 1981 19770 6 0.42 3.03 2.3989 1982 20720 9 0.59 4.34 3.3489 1983 21176 6 0.38 2.83 2.1989 1984 21268 12 0.76 5.64 4.3389 1985 21290 17 1.07 7.98 6.0889 1986 20835 14 0.89 6.72 5.0689 1987 20650 9 0.57 4.36 3.2789 1988 20736 15 0.95 7.23 5.44256TABLE 30VARIABILITY OF CABS AND STANDARDIZED INCIDENCE RATIOS WITHIN SCHOOLDISTRICTSB.C. 1979-1988School DistrictObservedCABSSTD CoefficientofVariation*ExpectedCABSExtremalQuotient**1. Fernie 9 0.54 0.6 45.83 2.32. Cranbrook 39 1.22 0.31 68.14 4.53. Kimberley 25 1.43 0.57 40.93 4.24. Windermere 4 0.66 1.66 25.03 2.17. Nelson 105 3.77 0.36 89.53 3.79. Castlegar 40 2.61 0.65 49.21 7.710. Arrow Lakes 27 1.95 0.72 20.15 5.111. Trail 67 3.47 0.52 98.59 1512. Grand Forks 30 1.95 0.65 37.35 5.813. Kettle Valley 12 154 1.28 15.71 5.314. S.Okanagan 105 4.82 0.46 84.76 3.515. Penticton 207 7.72 0.37 153.26 2.916. Keremeos 30 1.55 0.52 21.77 6.917. Princeton 26 1.28 0.49 20.74 4.918. Golden 2 0.40 2.00 19.83 1.0519. Revelstoke 18 1.25 0.69 29.27 4.021. Armstrong-Spall 26 1.28 0.49 30.41 5.122. Vernon 169 6.49 0.38 186.21 4.123. Central OkAnagan 660 25.18 0.38 432.98 3.924. Kamloops 190 4.27 0.22 265.98 1.726. N.Thompson 8 0.40 0.50 14.97 1.127. Cariboo-Chilcotin 90 3.41 0.38 113.55 3.728. Quesnel 59 3.48 0.59 72.13 8.229. Lilooet 6 0.66 1.11 15.48 2.430. S.Cariboo 18 1.66 0.92 29.95 4.831. Merritt 24 1.20 0.50 34.54 5.432. Hope 27 1.79 0.66 32.38 5.733. Chilliwack 172 5.93 0.34 205.41 2.534. Abbotsford 239 4.44 0.19 239.52 1.535. Langley 214 5.82 0.27 226.59 4.136. Surrey 764 22.11 0.29 683.67 1.637. Delta 271 5.84 0.22 251.78 1.538. Richmond 341 8.47 0.25 383.24 1.539. Vancouver 1474 29.22 0.20 1975.85 1.840. New Westminster 214 6.25 0.29 188.74 3.241. Burnaby 523 8.84 0.17 632.8 1.542. Maple Ridge 186 3.85 0.21 147.63 1.943. Coquitlam 424 6.64 0.16 355.25 1.744. N.Vancouver 331 7.79 0.24 411.11 2.245. W.Vancouver 159 5.82 0.37 227.91 3.646. Sunshine Coast 70 3.19 0.46 86.29 5.047. Powell River 70 3.66 0.52 80.28 6.448. Howe Sound 46 2.33 0.51 44.65 7.5* Std/Mean CABS** Largest SIR/smallest SIR257TABLE 30 (contd.)School DistrictObservedCABSStDCABSCoefficientofVariationExpectedCABSExtremalQuotient**49. Central Coast 4 0.66 1.66 10.33 2.150. Queen Charlotte 12 0.6 0.5 13.09 2.152. Prince Rupert 47 2.76 0.59 51.08 5.454. Smithers 31 1.64 0.53 35.9 3.055. Burns Lake 16 2.06 1.29 22.18 6.856. Nechako 30 2.53 0.84 43.55 7.157. Prince George 194 2.24 0.12 226.6 1.659. Peace River S. 23 1.27 0.55 73.29 3.760. Peace River N. 31 2.17 0.70 63.64 7.361. Greater Victori 1342 13.29 0.10 855.45 1.462. Sooke 207 5.22 0.25 136.85 2.163. Saanich 270 4.49 0.17 199.06 1.964. Gulf Islands 69 3.14 0.46 60.3 3.465. Cowichan 184 3.50 0.19 147.77 1.666. Lake Cowichan 25 1.57 0.63 22.38 5.468. Nanaimo 384 8.10 0.21 318.6 2.769. Qualicum 164 5.87 0.36 128.87 3.070. Alberni 121 3.65 0.30 116.27 2.971. Courtney 160 5.50 0.34 155.64 3.372. Cambell River 93 1.62 0.17 99.44 1.775. Mission 94 5.08 0.54 93.29 7.376. Agassiz-Harrison 18 1.33 0.74 21.56 1.977. Summerland 43 2.49 0.58 47.79 5.080. Kitimat 52 2.64 0.51 33.99 3.281. Fort Nelson 7 0.64 0.91 10.19 2.184. Van Isl North 9 1.04 1.16 9.61 3.185. Van Isl West 21 1.64 0.78 34.01 4.486. Creston-Kaslo 36 2.73 0.76 76.53 9.487. Stikine 4 0.49 1.22 4.49 1.388. Terrace 63 3.38 0.54 69.32 5.189. Shuswap 108 3.49 0.32 152.53 2.8258TABLE 31STANDARDIZED INCIDENCE RATIOS FOR FIVE AND TEN-YEAR PERIODSSchool District Standardized Incidence Ratios1979-1983^1984-1988^1979-19881 Fernie 0.26 0.14 0.202 Cranbrook 0.59 0.61 0.603 Kimberley 0.50 0.70 0.604 Windermere 0.14 0.25 0.207 Nelson 0.99 1.37 1.189 Castlegar 0.45 1.05 0.7510 Arrow Lakes 1.26 1.40 1.3311 Trail 0.55 0.77 0.6612 Grand Forks 0.64 0.97 0.8113 Kettle Valley 0.32 1.07 0.7014 S.Okonagan 0.94 1.54 1.2515 Penticton 1.13 1.64 1.4016 Keremeos 1.11 1.66 1.3917 Princeton 1.17 1.41 1.2918 Golden 0.18 0.00 0.0919 Revelstoke 0.69 0.80 0.7521 Armstrong- 0.81 1.00 0.91Spallumcheen22 Vernon 0.81 1.06 0.9423 Central Okonagan 1.30 1.83 1.5824 Kamloops 0.69 0.76 0.7226 N.Thompson 0.58 0.55 0.5627 Cariboo-Chilcotin 0.65 0.81 0.7428 Quesnel 0.57 1.04 0.8229 Lilooet 0.45 0.31 0.3830 S.Cariboo 0.61 0.73 0.6731 Merrit 0.71 0.85 0.7832 Hope 0.58 1.08 0.8333 Chilliwack 0.75 0.87 0.8134 Abbotsford 0.98 1.00 0.9935 Langley 0.87 1.04 0.9636 Surrey 1.02 1.23 1.1337 Delta 1.07 1.09 1.0838 Richmond 0.84 0.92 0.8839 Vancouver 0.66 0.84 0.7540 New Westminster 1.25 1.11 1.1841 Burnaby 0.80 0.89 0.8542 Maple Ridge 1.38 1.25 1.3143 Coquitlam 1.23 1.21 1.2244 N.Vancouver 0.76 0.90 0.84259TABLE 31 (contd.)School District Standardized Incidence Ratios1979-1983^1984-1988^1979-198845 W.Vancouver 0.68 0.74 0.7146 Sunshine Coast 0.70 0.97 0.8547 Powell River 0.71 1.09 0.9048 Howe Sound 0.86 1.20 1.0449 Central Coast 0.49 0.67 0.5850 Queen Charlotte 0.82 1.11 0.9752 Prince Rupert 0.65 1.08 0.8854 Smithers 0.80 1.01 0.9155 Burns Lake 1.10 0.66 0.8756 Nechako 0.64 0.82 0.7357 Prince George 0.90 0.84 0.8759 Peace River S. 0.25 0.37 0.3160 Peace River N. 0.34 0.65 0.5061 Greater Victoria 1.74 1.51 1.6262 Sooke 1.75 1.51 1.6263 Saanich 1.57 1.26 1.4064 Gulf Islands 1.15 1.16 1.1665 Cowichan 1.24 1.33 1.2966 Lake Cowichan 1.38 1.03 1.2068 Nanaimo 1.40 1.02 1.2069 Qualicum 1.29 1.45 1.3770 Alberni 1.08 1.06 1.0771 Courtney 1.11 1.10 1.1172 Cambell River 0.98 0.98 0.9875 Mission 0.70 1.23 0.9876 Agassiz-Harrison 0.83 0.81 0.8277 Summerland 0.70 1.22 0.9780 Kitimat 1.22 1.76 1.5181 Fort Nelson 0.85 0.49 0.6784 Van Isl North 0.35 1.25 0.8085 Van Isl West 0.47 0.75 0.6286 Creston-Kaslo 0.33 0.65 0.4987 Stikine 0.77 1.09 0.9488 Terrace 0.84 1.01 0.9489 Shuswap 0.61 0.77 0.69260TABLE 32EFFECT OF MIGRATION TO ALBERTA FOR CABS PROCEDURE ON STANDARDIZEDINCIDENCE RATIOS IN B.C. SCHOOL DISTRICTS1988SchoolDistrict*B.C.CABSAlbertaCABSSIRB.C. OnlySIRAll Cases1 Fernie 1 1 0.23 0.452 Cranbrook 3 6 0.41 1.223 Kimberley 4 1 1.05 1.307 Nelson 13 2 1.42 1.6312 Grand Forks 6 1 1.58 1.8415 Penticton 32 1 1.93 1.9916 Keremeos 4 1 1.79 2.2622 Vernon 32 1 1.59 1.6323 Central 87 3 1.81 1.87Okanagan57 Prince George 19 1 0.74 0.7859 Peace River S 4 4 0.50 1.0060 Peace River N 2 4 0.28 0.8486 Creston-Kaslo 6 1 0.78 0.93*All school districts sending patients to Alberta in 1988261TABLE 33PERCENTAGE OF CABS CASES FROM EACH SCHOOL DISTRICT USING B.C. CENTRES1979-1988School District Centre 1 Centre 2 Centre 3 Numberof Cases1 Fernie 33.3 66.7 0.0 92 Cranbrook 7.3 92.7 0.0 413 Kimberley 12.0 84.0 4.0 254 Windermere 50.0 0.0 50.0 47 Nelson 13.0 85.2 1.7 1159 Castlegar 20.5 77.3 2.3 4410 Arrow Lakes 44.8 51.7 3.4 2911 Trail 27.8 70.8 1.4 7212 Grand Forks 60.6 36.4 3.0 3313 Kettle Valley 50.0 50.0 0.0 1614 S.Okonagan 58.9 37.1 4.0 12415 Penticton 56.3 41.6 2.1 23816 Keremeos 50.0 47.1 2.9 3417 Princeton 46.7 53.3 0.0 3018 Golden 50.0 50.0 0.0 219 Revelstoke 44.4 55.6 0.0 1821 Armstrong- 43.3 56.7 0.0 30Spallumcheen22 Vernon 39.8 57.7 2.5 20123 Central Okanagan 49.5 48.0 2.6 78424 Kamloops 18.5 78.7 2.8 21626 N.Thompson 25.0 62.5 12.5 827 Cariboo-Chilcotin 72.5 25.5 2.0 10228 Quesnel 8.1 88.7 3.2 6229 Lilooet 28.6 71.4 0.0 730 S.Cariboo 27.8 66.7 5.6 1831 Merrit 33.3 66.7 0.0 2432 Hope 84.8 15.2 0.0 3333 Chilliwack 61.7 37.2 1.0 1%34 Abbotsford 67.5 31.4 1.1 28335 Langley 80.7 18.1 1.2 25436 Surrey 80.6 19.0 0.4 85137 Delta 75.3 22.7 1.9 30838 Richmond 57.8 41.7 0.5 38639 Vancouver 69.8 30.0 0.2 166540 New Westminster 71.3 28.3 0.4 23041 Burnaby 66.3 32.7 1.0 57842 Maple Ridge 72.2 27.3 0.5 205262TABLE 33 (contd.)School District Centre 1 Centre 2 Centre 3 Numberof Cases43 Coquitlam 73.2 25.9 0.9 45644 N.Vancouver 18.0 81.2 0.8 37745 W.Vancouver 17.0 82.4 0.5 18246 Sunshine Coast 39.7 60.3 0.0 7347 Powell River 21.2 71.8 7.0 7148 Howe Sound 13.5 86.5 0.0 5249 Central Coast 40.0 60.0 0.0 550 Queen Charlotte 15.4 84.6 0.0 1352 Prince Rupert 36.0 60.0 4.0 5054 Smithers 21.9 78.1 0.0 3255 Burns Lake 33.3 66.7 0.0 1856 Nechako 5.9 94.1 0.0 3457 Prince George 10.6 88.5 0.9 21859 Peace River S. 40.0 48.0 12.0 2560 Peace River N. 11.8 85.3 2.9 3461 Greater Victoria 0.4 0.3 99.3 134062 Sooke 0.5 0.5 99.0 20763 Saanich 0.4 0.0 99.6 26964 Gulf Islands 11.6 7,2 81.2 6965 Cowichan 1.1 .0 98.9 18466 Lake Cowichan 4.0 0.0 96.0 2568 Nanaimo 4.7 7.0 88.3 38569 Qualicum 5.5 6.7 87.9 16570 Alberni 10.6 4.9 84.6 12371 Courtney 43.6 3.0 53.3 16572 Cambell River 31.3 6.1 62.6 9975 Mission 65.5 33.6 0.9 11376 Agassiz-Harrison 52.2 47.8 0.0 2377 Summerland 34.0 66.0 0.0 5080 Kitimat 26.4 73.6 0.0 5381 Fort Nelson - 77.8 22.2 0.0 984 Van Isl North 22.2 11.1 66.7 985 Van Isl West 22.7 18.2 59.1 2286 Creston-Kaslo 13.6 84.1 2.3 4487 Stikine 42.9 57.1 0.0 788 Terrace 3.3 95.1 1.6 61263TABLE 36CORONARY ARTERY BYPASS SURGERY ANNUAL MEAN AGE BY CENTREB.C. 1979-1988Year Centre 1 Centre 2 Centre3Mean^Std Mean^Std Mean^Std1979 58.23^9.25 56.41^9.30 57.57^8.361980 57.60 8.82 56.50 9.07 65.60 7.751981 58.56^8.49 56.87^8.27 58.94^8.671982 58.60 8.78 58.10 9.56 60.16 9.051983 59.45^9.49 56.09^8.49 60.87^9.111984 61.20 8.59 60.39 8.91 61.90 9.161985 61.65^9.45 61.42^8.97 63.39^8.881986 62.55 9.13 61.59 9.72 63.59 8.641987 62.64^8.70 62.51^9.30 63.02^8.401988 62.75 9.20 62.30 9.82 64.72 9.08TABLE 37DISTRIBUTION OF CABS PATIENTS WITH COMORBIDITY BETWEEN CENTRESB.C. 1979-1988Centre Diabetes COPD TotalComorbidityTotalCABSComorbidityas % of Total1 327 96 423 4736 8.932 281 90 371 3838 9.663 188 83 271 2979 9.10Total 796* 269 1065 11553 9.21* 2 cases of diabetes occurred in patients coded as receiving CABS in other centres.264TABLE 39MEAN AGE BY SEX 'NO-CABS" AND ANGIOPLASTY POPULATIONSYear^Male Female Total Mean AgeNo-CABS Revascularization1979 2 0 2 57.01980 4 0 4 51.51981 6 7 13 55.921982 24 3 27 52.771863 43 23 66 58.011984 222 63 285 57.371985 439 148 587 60.041986 547 164 711 59.341987 233 91 324 61.081988 223 78 301 62.01Angioplasty1987 378 109 487 59.461988 477 144 621 59.75265TABLE 41GROWTH AND DECLINE OF THE 4800 CCP CODE BY CENTREB.C. 1979-1988Year Centre Total1 I^2 3 J^979 1 1 0 0 280 3 0 1 0 481 8 1 4 0 1382 10 2 14 1 2783 14 4 46 2 6684 98 32 147 8 28585 206 77 301 3 58786 243 114 348 6 71187 1 0 319 4 32488 0 0 300 1 301Total 584 231 1480 25 2320TABLE 41(A)DISTRIBUTION OF ANGIOPLASTY (CCP CODES 4801-4805) BY CENTREB.C. 1987-1988Year Centre Total1 2 3 91987 320 164 0 3 4871988 422 189 0 10 621Total 742 353 0 13 1108266APPENDIX B267TABLE 42INDEPENDENT VARIABLES SIMPLE STATISTICSVariable ReferenceLevelMin Max Mean StandardDeviationYear 0 0 5 2.50 1.70Year-squared 0 0 25 9.16 8.90Income (INC) 3 -1.15 4.01 0.26 0.68Distance from 0 0 2 1.71 0.60Cardiolgist(DSCAR)Distance from centre 0 0 2 1.86 0.14(DSCEN)Distance from 0 0 2 0.60 0.70Internist(DSINT)Employment Rate 0.7 0.04 0.26 0.17 0.04(EMPRAT)Graduation Rate 0 0.03 0.76 0.12 0.04(GRADRAT)268TABLE 43INDEPENDENT VARIABLES PEARSON CORRELATION COEFFICIENTSVariable INC DSCAR DSINT DSCEN GRADRAT EMPRATINC 1.00001- 0.2886- 0.20750.0 -0.3288 0.0890 -0.0692P 0.0001 0.00001 0.0001 0.0609 0.0609 0.1451DSCAR -0.2886 1.0000 0.3339 0.6846 -0.1338 -0.2499P 0.0001 0.0001 0.0001 0.0047 0.0001DSINT -0.2075 0.3339 1.0000 0.2360 -0.1527 -0.1876P 0.0001 0.0001 0.0001 0.0012 0.0001DSCEN -0.3288 0.6846 0.23604 1,0000 -0.0760 -0.2668P 0.0001 0.0001 0.0001 0.1094 0.0001GRADRAT 0.0890 -0.1338 -0.1527 -0.0760 1.0000 0.0031P 0.0609 0.0047 0.0012 0.1094 0.9479EMPRAT -0.0692 -0.2499 -0.1876 -0.2668 0.0031 1.0000P 0.1451 0.0001 0.0001 0.0001 0.9479269TABLE 44POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN CABS RATE ACROSS ALLSCHOOL DISTRICTSVariable* ParameterEstimateStandardErrorIntercept -11.5034 0.0515Year 0.0891 0.0588Year-squared -0.000839 0.0099INC -0.3725 0.0650DSCAR -0.0641 0.0879DSCEN 0.1710 0.0490Year*DSCAR -0.0481 0.0458Year*DSCEN -0.00174 0.0496Year-squared 0.0117 0.00809*DSCARYear-squared -0.00660 0.00856*DSCENDSCAR*INC 0.0789 0.0454DSCEN*INC 0.0751 0.0531DSCEN*DSCAR -0.0720 0.0415-2 log Intercept Intercept &Likelihood Covariates188050.26 187858.60chi-squared df p191.656 12 0.0001Difference from saturated modelchi-squared df p891.146 66 0.0001* Only significant variables are incided270TABLE 45POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN CABS RATE ACROSSSCHOOL DISTRICTS WITH ADJUSTMENT FOR MOBILITYVariable* ParameterEstimateStandardErrorIntercept -11.4765 0.0426Year 0.0783 0.0444Year-squared -0.00172 0.00764ALBERTA -0.7363 0.0903INC -0.3145 0.0442DSCAR -0.1458 0.0379DSCEN 0.1240 0.0215Year*DSCAR -0.0454 0.0332Year-squared 0.00588 0.00574*DSCARDSCAR*INC 0.1346 0.0358-2 log Intercept Intercept &Likelihood Covariates188050.26 187783.62chi-squared df p266.636 9 0.0001Difference from saturated modelchi-squared df p816.16 66 0.001* Only significant variables are included271TABLE 46POISSON REGRESSION SATURATED MODELSVariable ParameterEstimateStandardErrorCABS Unadjusted forInterceptLograte-2 logLikelihoodchi-squared1082.802Morbidity0.0001451.00Intercept188050.26df740.38430.0337Intercept &Covariates186967.46p0.0001CABS Adjusted for MoInterceptLograte-2 logLikelihoodchi-squared1259.05rbidity0.000141.00Intercept188208.49df740.36830.0323Intercept &Covariates186949.44P0.0001272TABLE 47POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN MORBIDITY-ADJUSTEDCABS RATE ACROSS ALL B.C. SCHOOL DISTRICTSVariable ParameterEstimateStandardErrorIntercept -11.241 0.0295Year -0.00733 0.0269Year-squared 0.0131 0.0048DSCAR -0.2597 0.0141INC -0.1108 0.0237-2 log Intercept Intercept &Likelihood Covariates188208.49 187825.01chi-squared df p383.482 4 0.0001Difference from saturated modelchi-squared df p875.568 70 0.0001273TABLE 48POISSON REGRESSION VARIABLES EXPLAINING VARIATION IN MORBIDITY-ADJUSTEDCABS RATE ACROSS SCHOOL DISTRICTS WITH ADJUSTMENT FOR ALBERTAVariable ParameterEstimateStandardErrorIntercept -11.2429 0.0295Year -0.00681 0.0269Year-squared -0.0132 0.00480ALBERTA -0.6020 0.0902INC -0.1133 0.0237DSCAR -0.2451 0.0143-2 log Intercept Intercept &Likelihood Covariates188208.49 187720.92chi-squared df p437.570 5 0.0001Difference from saturated modelchi-squared df p821.48 69 0.0001274TABLE 49REGIONAL DIVISIONS USED IN AGE-SEX-YEAR-REGION REGRESSION ANALYSISREMOTE/RURAL/URBAN/METROPOLITAN DIVISIONRegion School DistrictsRemote (Reg3) 27, 28, 46, 49, 50, 52, 5455, 56, 57, 59, 60, 80, 8185, 87, 88Rural (Reg2) 1-10, 18, 19, 21, 22, 29, 4748, 65, 66, 70, 86Urban (Regl) 11-17, 23-26, 30-34, 42, 6869, 71-77, 84Metropolitan (Reference) 35-41, 43-45, 61-64*School district 89 straddled two regions and was excludedfrom this analysis.GEOGRAPHICAL DIVISIONRegion^ School DistrictVancouver & S.W. 32-35, 48, 75, 76(Reference)S.E. & Okanagan (Gregg) 1-26, 30, 31, 77, 78Vancouver Island (Gregl) 47, 49, 61-72, 84, 85& Central CoastNorth (Greg3) 27, 28, 50-60, 80, 81, 87, 88275TABLE 50POISSON REGRESSION RESULTS INTERACTIONS BETWEEN AGE, SEX, YEAR AND REGIONFOR METROPOLITAN/URBAN/RURAL/REMOTE REGIONSB.C. 1983-1988Variable ParameterEstimateStandardErrorIntercept -5.8572 0.0492Age -0.2672 0.0312Year 0.0199 0.0444Sex -1.3271 0.0931Agesex 0.0675 0.0304Agesq -0.1188 0.00721Agesqsex -0.0277 0.0102Yearsex -0.0688 0.0820Yearsq 0.00738 0.00832Yearsqsex 0.01 0.0156Yearsqage -0.00251 0.00518Agesqyear -0.00905 0.00700Yearsqagesq 0.00135 0.00136Yearage 0.039 0.0274Reg2 -0.1307 0.1401Reg3 -0.5054 0.0298Regl 0.00159 0.0874Reg2Age -0.1254 0.0957Reg3Age -0.0628 0.1397ReglAge -0.0603 0.0569RegiSex 0.2957 0.1578Reg2Sex 0.2378 0.2498Reg3Sex -0.1854 0.4050ReglYear 0.0579 0.0773Reg2Year -0.0151 0.1238Reg3Year 0.0924 0.1784Reg2Agesq -0.0418 0.0238Reg3Agesq 0.0123 0.0281Regl Agesq -0.015 0.0134Reg3Yearsq -0.00598 0.0325Reg2Yearsq 0.00887 0.0229ReglYearsq -0.00587 0.0144Reg2Agesex 0.1165 0.0836Reg3Agesex 0.0188 0.1223Regl Agesex -0.0476 0.0535Reg2Sexyear 0.1161 0.2214Reg3Sexyear 0.3468 0.3159ReglSexyear 0.0732 0.1393Reg2Ageyear 0.0209 0.0799Reg3Ageyear -0.0583 0.1167Regl Ageyear 0.0988 0.0482Reg2Agesqsex 0.028 0.0274276TABLE 50 (contd.)Variable ParameterEstimateStandardErrorReg3Agesqsex -0.00166 0.0333ReglAgesqsex -0.0172 0.0179Reg2Agesqyear 0.007 0.0211Reg3Agesqyear 0.00425 0.0246ReglAgesqyear 0.0226 0.0124Reg2Yearsqage 0.00518 0.0145Reg3Yearsqage 0.0162 0.0210ReglYearsqage -0.0142 0.00889Reg2Yearsqsex -0.042 0.0424Reg3Yearsqsex -0.0426 0.0564ReglYearsqsex -0.0347 0.0267Reg2Yearsqagesq 0.00146 0.00392Reg3Yearsqagesq 0.00039 0.00456ReglYearsqasq -0.00325 0.002326-2loglikelihoodIntercept Intercept andcovariates125720.66 110488.36chi-square df p15232.299 55 <0.0001277TABLE 51POISSON REGRESSION RESULTS INTERACTIONS BETWEEN AGE, SEX, YEAR AND REGIONFOR GEOGRAPHIC REGIONSB.C. 1983-1988Variable ParameterEstimateStandardErrorIntercept -5.7095 0.0517Age -0.0108 0.0264Year -0.0377 0.0471Sex -1.4783 0.1031Agesex 0.1580 0.0260Agesq -0.1913 0.0098Agesqsex -0.0246 0.0113Yearsex 0.0292 0.0898Yearsq 0.0165 0.00885Yearsqsex -0.0108 0.0169Yearsqage -0.00502 0.00411Agesqyear 0.00471 0.00872Yearsqagesq -0.0006 0.00163Yearage 0.0626 0.0224GRegl 0.4884 0.0835GReg2 0.0923 0.1015GReg3 -0.2622 0.1568GReglAge 0.0433 0.0415GReg2Age -0.0280 0.0519GReg3Age -0.1156 0.1033GReglSex 0.1224 0.1614GReg2Sex 0.4306 0.1843GReg3Sex -0.0967 0.3489GReglYear -0.1509 0.0794GReg2Year 0.0900 0.0920GReg3Year 0.1994 0.1393GReglAgesq 0.00332 0.0159GReg2Agesq -0.0241 0.0201GReg3Agesq -0.0407 0.0338GReglYearsq 0.0147 0.0152GReg2Yearsq -0.0209 0.0173GReg3Yearsq -0.0431 0.0261- GReglAgesex -0.0499 0.0440GReg2Agesex -0.0421 0.0482GReg3Agesex -0.0283 0.0860GReglSexyear -0.0728 0.1471GReg2Sexyear -0.1869 0.1641GReg3Sexyear -0.0390 0.2920GReglAgeyear 0.0128 0.0367GReg2Ageyear 0.0800 0.0425GReg3Ageyear 0.0826 0.0824GReglAgesqsex^_ 0.0113 0.0191278TABLE 51 (contd.)Variable ParameterEstimateStandardErrorGReg2Agesqsex 0.00941 0.0211GReg3Agesqsex 0.00873 0.0334GReglAgesqyear 0.00138 0.0148GReg2Agesqyear 0.0131 0.0173GReg3Agesqyear 0.0288 0.0279GReglYearsqage -0.00324 0.00691GReg2Yearsqage -0.0115 0.00769GReg3Yearsqage -0.0161 0.0147GReglYearsqsex 0.00781 0.0283GReg2Yearsqsex 0.0191 0.0311GReg3Yearsqsex 0.0241 0.0535GReglYearsqagesq 0.000296 0.00284GReg2Yearsqagesq 0.000027 0.00319GReg3Yearsqasq -0.00507 0.00507-2loglikelihoodIntercept Intercept andcovariates126954.92 112155.01chi-square df p14799.910 55 <0.0001279MAP ABRITISH COLUMBIA — COLOMBIE-BRITANNIOUESchool Districts — Oistncts scotairesSCNOOL OISTPICTS. CttST•HCTS SCOLAIRES1 Berme^ 30 S Can000^57 Prw+Ct George2 Cranorook^3, 164,,,rm 59 *tact/ m•ver S3 KImovity 32 moot 60 Peace M•ver N• wsnottnnet•^33 critlineraC4^61 Greater vtctorta7 Nelson^34 A000tstorc 62 Soo•eS Casnegar 35 Langtey 63 Saan•en'0 Arrow Lakes^36 Surrey^64 Cult tstancs11 rraa^ 27 Dena 65 COvntrtan'2 GIAnCI Fors^38 P.CrirnOnd^36 La'. Covncrian'3 Kerue valley 39 Vat",COuvO ,^6! Niaftanno14 S Okanagan^40 Nev.^ 69 OLtakcurn15 Penncton westmnster^73 Alcorn.16 Keremeos^Al Surriaor 71 COunenay17 Princeton 42 Nacre P.Oge^'2 CarTlooeit After18 Gowen a3 Cooutuam 75 mtsstort19 Revelstoke^44 N Vancouver^76 aO3SSZ21 Armstrong • 45 w Vancouver mamsgnScatturncneen^46 SketWWI* Coast^77 Summe. , 4net:2 Vernon^47 Pov.ett Awe'^8C Kii•Mat23 ctmv ai A! Havre Soun0 8' =on NetSOnOkanagan^4 Central Coast^iki Van IsJ WeSt2• KarrO000s SO Queen Cnartorte^85 van 1311.40nm26 N Tnonloson^52 Ponce Puce ,'^A6 Creston•nasto2 7 CattOoo•Crurconn^54 S./vine's^8' Simone28 Ouesneo^55 Burns Lake 6a '.,race29 Ltsooet 56 Necna•o 89 Stnusv.ao5C.34DO `q4kk WC‘r °Pr-^.43ae AO3639.280MAP BSCHOOL DISTRICTS SENDING A PLURALITY OF CASES TO CENTREBRITISH COLUMBIA - COLOMBIE-BRITANNIOUESchool Districts - Distncts scolairesCentre 1Centre 2Centre 3SCHOOL. DISTRICTS - DISTPtCTS SCOTAIRES-twm11.1:132%176....11k1-14■114.0..ikt,"-ln:47740.^'".1%-ilacri b,,ga40.^NTAww,„41. 7°  mtwas.,778Ce n84asRssiasFern,.^30 S Cane=2 Cranorook^3, Merritt3 K■moerley 32 looea winaermer•^33 CNretru,aci,7 NelSOn^34 AlsootstoraCastiegar 35 Langley'0 Arrow Lakes^36 Surrey11 Trani^ 37 Delta'2 Grand :was^38 RIcnrridnd'3 Kent, vasey 39 Vancouver14 S Okanagan^.40 New15 Penucton wrestrrunster16 Keremeos^4 I Sur rtaCIV17 Princeton 12 Wide R•oge18 Gorden 43 Coctuntam19 Revelstoke^44 N VanCOwver21 Armstrong • 45 W vancouyerScallumcneen^46 SurlSnme Coast:2 Vernon^47 Powell r;',ue•23 Central a8 wowe SoundOkanagan^.19 Central Coast24 Ka/114000s 50 Queen CNarinne26 N ThOrnoson^52 Prot! Rupert27 Card000•Crurcobn^SNUMWS26 Ouesne•^55 Burns Lake29 Lwooet 56 Nechako57 Proce George59 Peace R , ver S60 Peace R.yer61 Greater \Actona62 Sooke63 Saarkcn54 Gull islanos65 Cow.cnan56 Lake62 Nama,,, ,o69 Cluarcum70 sitserru71 COunenae72 CamPOelliqnrer75 kossoy,76 Agass.: •►iarnsonSummer , anoKdonatson Ne.sonvan 1st westvan 1st NOrtnC•,SIOn•KASIOSIAmeTerraceSnuswarsBRITISH COLUMBIA - COLOMBIE-BRITANNIOUESchool Districts - Districts scolaires516045 75525278529 .72 47gt.SCHOOL DISTRICTS - Of ST11 tCTS SCOLAIIRES595 7857534do553669686 4281MAP CSCHOOL DISTRICTS SENDING 90 PERCENT, OR MORE, OF CASESTO ONE CENTRE1 germ*2 Cranoroor3 1cmoefloyWIrtgermere7 kie•SOnS C•Stlegar'0 Attaw LaktS7,•11*2 Grano Fors'3 hero, Valley14 S Okanagan15^ u.cton16 '<eremites17 Princeton10 Gowen19 Revelstoke21 Armstrong •Seallurncrteen22 Vernon23 CentralOkanagan24 KanktOODS26 N Thompson27 Catmoo•Crutoottn23 (Due snet29 I. teeetet30 5 Cahoot,31 k.44p•ell32 meetCruotntiso■34 Apootsfore35 Lam•e ,/36 Stotev2 7 Oena38 Gtl.crtrnona39 V1 ,,C*ukt ,se tie.Strnalste ,at Btor130V42 macto a•:3ge4 3 COCI‘onarnai N Var■COuve ,43 w vamcouverse Sumsrune CCASI•7 Poweli35 hto4.4 Sounc49 Central Coast50 Queen CrlariortePnnee QueenSA S•rtarte•s55 Burns Lame56 Nechako57 Ot.nce George55 °one Ll.ver S60 Peace q.v., N61 Greater V.ctor.a62 Sooke63 Saamoh6.4 Gun islanes65 Comortan56 .6.2keCo..Crt4/155 Naha•mto69 Okrattcurrt70 AMperfu7 1 COUneMaY72 Carhacte4 1:1fleet75 #.4tss.oh76 49aSS•: •rtiarrrspn7 7 S.rnme ,, anCISC Kawnat8' :on Ne•son5.4 Van 1st Westas Van Iv Norm/46 Cresion•has•o87 St.*.".81 'create85 Shus.4apMaai.■111^V...1aU-'111,■"


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