UBC Faculty Research and Publications

Perceptions and realities : medical and surgical procedure variation : a literature review Sheps, Samuel Barry; Scrivens, S; Gait, Jennifer Mary 1990

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

Item Metadata


52383-Sheps_S_et_al_Medical_surgical_procedure.pdf [ 32.59MB ]
JSON: 52383-1.0048518.json
JSON-LD: 52383-1.0048518-ld.json
RDF/XML (Pretty): 52383-1.0048518-rdf.xml
RDF/JSON: 52383-1.0048518-rdf.json
Turtle: 52383-1.0048518-turtle.txt
N-Triples: 52383-1.0048518-rdf-ntriples.txt
Original Record: 52383-1.0048518-source.json
Full Text

Full Text

HEALTH POLICYRESEARCH UNITDiscussion Paper SeriesTHE UNIVERSITY OF BRITISH COLUMBIAPERCEPTIONS AND REALITIES:MEDICAL AND SURGICAL PROCEDUREVARIATIONA LITERATURE REVIEWS. ShepsS. ScrivensJ. GaitHPRU 90:24D December 1990‘“ ,-PERCEPTIONS AND REALITIES:MEDICAL AND SURGICAL PROCEDUREVARIATIONA LITERATURE REVIEWS. ShepsS. ScrivensJ. GaitHPRU 90:24D December 1990HEALTH POLICY RESEARCH UNITDivision of Health Services Research and DevelopmentThe University of British ColumbiaVancouver, B.C.V6T lZ6This research was supported by NHRDP Project No. 661O-1745-HT.PERCEPTIONS AND REALmES:MEDICAL AND SURGICAL PROCEDURE VARIATION­A LITERATURE REVIEWTABLE OF CONTENTSIntroductionBackgroundData Sources and MethodsResultsDiscussion137927BibliographYTable 1 Reported admission rates for selected procedures:Selected countries for which data were reported, 1980Table 2 Breakdown of variation literatureTable 3 Periodicals reviewedTable 4 Variation in rates of selected proceduresTable 5 Ten most frequently studied procedures in the 51"good papers"Table 6 Relationship of tested independent variables onprocedure ratesTable 7 Main shortcomings of 51 "good" papers.Appendix I Abstracts of 51 PapersAppendix IT An overview of papers which describe variationin procedure rates, but fail to test the effect ofindependent variables on this rate of variation.Appendix ill Other procedures included in the literatureINTRODUCflONAlthough variety is said to be the spice of life, in terms of health servicesprocedures, variations pose major conceptual and policy problems. At both levels,variations in health care utilization raise concerns because of the fundamentalassumption that individuals should have equal access to health care. In thiscontext, procedure or practice variations create the dilemma that either access to oravailability of health care services is uneven or that morbidity patterns are variable.Both possible violations of the assumption of equity pose political problems.While at a policy level health care utilization variations are a source of realconcern regarding the organization and cost of health services, at the conceptual andmethodological level the issue of the magnitude of utilization variations and therigor and consistency with which explanantions for such variations are undertakenpose equally substantial problems. For if there is no convincing evidence to explainobserved variations, assuming these observations to be correct, the solutions remainobscure. Moreover, it remains unclear if variations are a good thing or a bad thinggiven that normative criteria for the frequency of health care utilization in general,and procedure variations specifically, do not exist. An assumption is made,particularly in the "small area analysis" literature that the population groups beingstudied are sufficiently similar in terms of morbidity, that other strucutural andfunctional factors must be responsible for any variations observed. However, todate there have been limited data to support such an assumption. One reason forthis is the lack of a concerted effort to support rigorous epidemiological research todemonstrate substantial differences in health status across populations. Until such2information is available, the ongoing discussion of variations in health careutilization will remain largely speculative.Despite the lack of a clear indication that variations exist and notwithstandingthe limited effort to explain variations, there is no question that a literature hasdeveloped arguing that utilization variations do indeed exist, that they are ofsufficient magnitude to stimulate concern and that there is an urgent need for aclearer understanding of the factors repsonsible. This perception of significant andmeaningful variation, reported often in editorial and other non-empiricalstatements, probably represents a certain degree of truth, but the exact nature andmagnitude of that truth is unclear. Moreover, despite the identification thatvariation exists, there remains uncertainty regarding the factors producing it,whether these factors are consistent across procedures, and whether the studiesdocumenting variations or assessing associated (causal) factors are sufficientlyrigorous to provide useful information for planners and policy makers.We undertook an in-depth literature review with the overall goal of assessingthe current state of knowledge of procedure variations in medical and surgicalpractice with particular reference to research attempting to assess the impact ofspecific variables on the variation observed. Our explicit objectives were to reviewthe recent literature using explicit a priori criteria in order to: 1) identify papersusing sound methodology; 2) determine the magnitude of variation found to existand whether there is consistency in this estimate across studies; 3) determine, ingeneral and for specific procedures, which factors seem consistently related tovariations (if any); and 4) identify directions for future research.3Prior to discussing the methodology and results of our own study, it is useful toprovide an overview of the health care utilization literature. This is convenientlydone by considering two recent reviews.BACKGROUNDMcPherson (50) provides a broad overview of both the data (limited tointernational comparisons) and the issues surrounding the variation of surgicalprocedures. The data (Table 1) illustrate substantial variations across countries andalthough certain caveats must be kept in mind (eg. the lack of age standardization)several patterns emerge. Japan, for example, has generally very low rates for theprocedures listed while the United States has generally higher rates. Canadian ratesalso tend to be high. The magnitude of the variations observed clearly begexplanantion and McPherson provides a useful review of factors that may beresponsible. Even though his discussion is largly speculative (based on thefundamental limitations of the literature he is reviewing), he does point to criticalareas for future research.The factors discussed generally fall into two broad categories: strucutural andfunctional. Structural factors include physician and bed supply, and to a limitedextent differences in methods of payment across the countries studied. As will beseen later, this group of factors has been the most common focus in the procedurevariation literature, no doubt because of the relative simplicity of measurement andease of data availability. Perhaps more importantly, McPherson considers otherstructural differences such as true differences in morbidity (ie underlying variationsin coronary artery disease or cholecystitis), as well as the age and sex structure of the4population. These factors are critical since, as noted above, a common asssumptionin the variation literature is that these health status and demographic variables aresufficiently similar across study populations as to be discounted.Among the functional factors cited as important in understanding procedurerate variations, McPherson correctly identifies some as giving rise to artifactualdifferences. For example, the differential substitution of day care surgery forinpatient surgery in differing jurisdictions may have a substantial impact on thenumber of procedures counted and thus rates generated. While this factor isunlikely to effect coronary artery bypass rates, it could, without consideration ofother factors, account for a major proportion of the 16 fold variation in differencesin tonsillectomy rates or the 13 fold differences in cataract procedures presented inTable 1. Other sources of artifact include differential procedure coding andcompleteness, comparability of computer file definitions and formats, and protocolsregarding whether primary, secondary or tertiary diagnoses are counted.McPherson also stresses the importance of clinical uncertainty as a source ofvariations and distinguishes this factor from the more frequently cited, but no moreintensively studied, variables of prevailing local custom and what has been termed"practice style".Finally, the critical issue of basing rates on an accurate estimate of thepopulation at risk is discussed. McPherson believes this to be a source of artifact, butin our view it is a more fundamental methodological problem, which as will benoted below, is pervasive in the literature. It is noted here to highlight itsimportance for future research.The major limitation of McPerson's review, as indeed with the literaturegenerally, is the fact that in attempting to explain international differences one is5immediately confronted with the problem of the ecological fallacy. This issue iscompounded by a lack of normative expectation, thus although differences areobserved, it is not clear what to make of them: i.e. are they real. Moreover,McPherson fails to consider a host of methodological problems across studies (eg,inadequate definition of the independent variables or incomplete description ofanalytic techniques) which may cummulatively be responsible for at least some ofthe differences in observed rates.Paul-Shaheen et al (62) provide at once a more comprehensive discussion ofsmall area analysis and a more limited review since it is restricted to the NorthAmerican literature. Their complex paper cannot be easily summarized but severalgeneral points emerge which of are value to discuss at this point. The first is thatthey set forth a framework for categorizing papers which we found conceptuallyuseful but have adapted. Second,the striking feature of the data they present is therelatively small magnitude of the observed variation. Across the 59 "core" papersthey discuss in detail, the magnitude of variation was generally on the order of 1-3fold, rarely exceeding 5-6 fold . Whether this represents an effect of a restrictedsample of papers ( ie North American, thus reducing a large number of national andor cultural differences which may give rise to larger variations), a selection biasresulting from their methods of choosing and reviewing papers, or in fact is anaccurate reflection of procedure variations is unclear. However, if the latter be true,then the degree of variation that has stimulated so much concern, may not be asgreat as it is perceived to be. Of particular interest is that Paul-Shaheen et allimitedtheir review to papers which attempted in a formal way to asssess the impact ofspecific factors and found that, generally speaking, none of the usual variablesstudied was correlated with rate variations with a coefficient of greater than 0.7 (r2 =0.49). They concluded variables such as bed or physician supply, morbidity, socio-6economic status, etc., individually do not account for significant variation and that acombination of practitioner and community variables provides a better explanatorymodel. However, differing combination of variables accounted for huge differencesin the variations of utilization: 2-73% of the variation of discharge rates; 14-76% ofthe variation in length of stay; and 3-89% of the variation of patient day rates. Thismagnitude of variation explained by independent variables stongly suggests thatmethodological differences rather than true relationships are reflected in these data.While the variation in specific surgical procedures (eg Tonsillectomy andAdenoidectomy) seem to be particularly dependent on physician or bed supply, sucha clear relationship was not consistently seen across procedures. Unfortunately,methodological issues were not discussed in any detail thus, again, it is not possibleto evaluate the results in the light of the methods used in the studies reviewed.Moreover, many of the papers examined variations of multiple procedures ratherthan focusing on one procedure and exploring predictive factors in depth.Despite these limitations, Paul-Shaheen et al provide an excellent discussion offuture research approaches and issues. Among these are the need to assess theimpact of the unit of analysis under study, the need for normative estimates toalllow judgments of under- or over-utilization to be made, the need to providegreater detail in defining factors that may influence rate variations, the need to havebetter estimates of underlying mortality and morbidity patterns, and the need to beaware of policy changes (such as a move from inpatient to outpatient serviceprovision for a particular procedure) which may produce observed variations.Specific methodological considerations include the standardization of rates by ageand sex, an attempt to document the validity of underlying asssumptions regardingthe similarlity of the populations being compared, and the use of standard measuresacross studies such as procedure specific rates with appropriate at-risk denominators.7DATA SOURCES AND METHODSThe following sources were used to identify papers describing procedurevariations or examining factors associated with variation: 1) the 1985 and 1987bibliographies prepared by the Copenhagen Collaborating Centre (22,23); 2) therecent study by Paul-Sheehan, Clark and Williams which focused on NorthAmerican literature (62); 3) and several Medline searches utilizing different keyword combinations: geographic variation; small area analysis; small area variation;regional variation; utilization; rate; and physician practice patterns. We soughtpapers published (or abstracted) in English since 1975 (until May 1990), although afew papers published prior to 1975 are included if they were cited frequently.All papers identified were reviewed and categorized as follows, using thecategories adapted from Paul-Sheehan, Clark, and Williams:I. Utilization of Medical and Surgical ProceduresA. Studies which presented primary data on variations and examined factorsassociated with variations.B. Studies which presented primary data on variations and performed statisticalassessment on the rate variations but performed no analysis on causal factorsC. Studies which presented primary data without any analysisD. Studies which presented secondary data or editorials, letters, andcommentaries, etc.8n. Utilization of Health ServicesThere were several categories defining papers included in this group but thecritical feature of these papers was the lack of any primary data on medical orsurgical procedures (i.e. they discuss medical care utilization in general). Thesepapers focused on physician or hospital use, or discussed analytic approaches to theassessment of procedure variations.All papers categorized as IA were reviewed in detail and for each paper thefollowing information was abstracted: 1) focus of study; 2) procedure(s) studied; 3)relevant medical or surgical speciality; 4) level of comparison (ie national, regional,provincial); 5) study population; 6) unit of study definition (i.e. city, state, province,etc.): 7) data sources; 8) use measures; 9) area based statistical calculations; 10)independent variables analysed; 11) results; 12) conclusions; 13) limitationsrecognized by the authors; and 14) shortcomings (i.e. problems not recognized by theauthors but noted by us).Prior to reviewing the 51 core papers, the two reviewers independently assessedthree papers to determine the consistency with which information was abstracted.The results were assessed by a third reader and for all three papers the data abstractedwas identical between the two readers. Although a small sample, the abstractionprocess was judged to be of high quality.In the following discussion and in the Tables, rates are quoted per 10,000general population unless other specificed.9RESULTS:Over 360 papers from 56 journals were reviewed and categorized (Tables 2 and3). Of these, S1 papers met the criteria of Category IA and were reviewed in depth.In general, most of these papers, which we would define as having reasonably goodmethodology, were published since 1980. Of these S1 papers, only 17 utilizedCanadian data and of the 106 papers presenting primary data of some sort, only 28utilized Canadian data.The discussion that follows refers only to those papersreviewed in depth. Details regarding the 13 characteristics examined in depth foreach paper are provided in Appendix I, A brief review of the findings of the other5S papers which provided primary data on procedure variations but did notformally examine factors influencing the observed variations is provided inAppendix IT. All papers abstracted (Appendix I), annotated (Appendix II), listed inTable 4, or cited in the discussion are presented alphabetically in the Bibliographywith appropriate cross -referencing by bibliography number in Table 4.With regard to the 51 papers reviewed in depth, we will first provide asummary of the the findings of six of the characteristics abstracted for each paper.This will be followed by a more detailed presentation of the results of those paperswhich examined the 10 most frequently studied procedures (Table 5). A list of thefrequency with which other procedures were reported in the literature is providedin Appendix ITl. Factors associated with rate variations for the 10 most frequentlydiscussed procedures and an indication regarding whether they were found to besignificantly associated are presented in Table 6.Procedures studied: The papers were roughly split between those examiningvariations in a single procedure and those examining several or many procedures.In the mid 1980's a few high technology procedures were examined (eg Coronary10artery bypass), but overwhelmingly the literature has assessed variations incommon procedures. Recently, appropriateness defined by panels of physicians hasbeen added as a variable in the assessment of variations, thus the distinctionbetween area variation and quality assurance as study objectives is becomingblurred.Levels of comparison: Almost half of the papers focus on within state orprovince variations. More recently there has been an increase in the number ofpapers using physicians as the unit of analysis.Area definitions: Canadian studies tended to use well defined geographicboundaries (i.e. counties) while US studies used hospital service areas or StandardMetropolitan Statistical Areas.Data sources: North American and UK studies used government oradministrative databases and calculated population based rates of utilization whileEuropean studies tended to focus on individual hospital data or surveys.Independent variables: The primary focus for most papers assessing therelationship between procedure variations and other factors was either supplyfactors (number of physicians or beds) or economic factors (measured either byinsurance coverage or socio-economic status) since these are generally the easiest tomeasure. Physician characteristics such as specialty, or place or time sincegraduation were examined in only a few studies. A major problem was the lack ofrigorous analysis, either because appropriate methods were not applied, or the datawere sufficiently limited to make such analysis impossible. A second problem wasthe limited assessment of patient demand factors (either defined as need on the basisof morbidity patterns across communities, or expressed demand as measured byrequests for procedures). This limitation has arisen primarily because the physician11is generally thought of as the "gate keeper" or initiator of service demand thus it hasbeen of interest to determine if the supply characteristics of physicians influenceutilization patterns.Results: All studies found variations in utilization rates and although thisvariation was somewhat greater for elective than non-elective procedures, themagnitude of the variation was generally speaking modest, on the order of 1-3 fold.Interestingly, variation within states or provinces was generally higher than thatbetween states or provinces. While this may reflect the instability of smallgeographic area rates (especially if the data are for single years), inter-provincial orinter-state comparisons may obscure real variations since one is examining averagerates. There has not been an adequate consideration of these statistical issues in theliterature to date. International comparisons tend to reveal wide variations. Therelationship between procedure variation and physician or bed supply was unclear,papers presented conflicting results both across procedures and for the sameprocedure in different jurisdictions. In general, however, no one factor could besaid to account for a substantial proportion of variation in procedures and moststudies concluded that several factors were operative.An important observation emerging from this review is that procedures mustbe examined independently. The range of variation and extent to which variousindependent factors influence utilization patterns varies from procedure toprocedure. In addition, the relationship between underlying patterns of morbidityand procedure utilization rates has been inadequately addressed.With regard to specific procedures, we will report here only those proceduresfor which we could identify at least ten papers meeting the criteria for Category IAand which presented data on a single procedure. An exception was made for12coronary artery bypass which although examined in only three papers is ofconsiderable interest to policy makers because of increasing frequency, high cost andits political impact in terms of claims regarding long waiting times. Table 4 providessome estimates of rates from recently published papers in the IA category. Table 6provides a tabulation of the frequency with which independent variables have beenassessed for each of the ten most commonly studied procedures and the number ofpapers which identified a statistically significant relationship between anindependent variable and procedure variation.CHOLECYSTECTOMY: Although this procedure was included in 29 papers..examining several procedures (21 of which analysed the effect of independentvariables and 8 simply describing differences in rates), 18 papers provided specificcomparative rates and only 3 papers focused on this procedure alone and assessedfactors influencing rates. Seven papers provided rates for Canada. As can be seen inTable 4, the rate variations seen in the Canadian studies are generally of the order of1-2 fold. However, a number of exceptions are of interest. The slightly higher ratevariations reported by Cageorge et al. (15) probably reflects that fact that they arereporting on a larger number of procedures (ie differing types of gall bladder surgeryrather than simply on cholecystectomy. The 5 fold variation reported by Stockwelland Vayda (89) is the highest reported in the literature within a single country, stateor province, but only reflects the experience of one year, thus how consistent thishigh variation is remains unclear. The almost 3 fold variation reported by Roosand Roos (72) may reflect the fact that they studied rate variations in the 65 +population which may indeed have greater variations in procedure rates than thegeneral population.Comparing the Canadian data with other countries it is striking that thereported rate variations (and rate levels) are quite similar (although slightly higher13in the United States) with the exception of the 5 fold variation across countriesreported by Vayda et al (96). Interestingly, the rates reported in this paper (6.3-33.2)are at the low end of the range reported generally.Only three papers could be found which focused on this procedure: McPhersonet al (52); Fowkes (31); and Cageorge et al(15). The results of these papers areconflicting. Although all 3 note a significant increase in rates over the past 15 yearsMcPherson et al and Cageorge et al found no relationship between rate andphysician or bed supply. The Scottish study (31) however, did find a strongassociation between rate and surgeon supply and a moderate association with bedsupply. McPherson et al did note a correlation between the prevalence of gallstonesacross the seven districts studied and the rate of cholecystectomy. These authorscaution however, that due to small numbers and crude measures the meaning ofthis correlation is uncertain, and they also note that their data source does notinclude procedures undertaken outside the National Health Service.Overall, the studies investigating the effect of patient and supply variables onrate variations provide a mixed picture. Bed supply was assessed most often but wasonly found to be significantly related to procedure rates in 25% of the studies; and inonly three studies utilizing multivariate techniques. Surgeon supply was examinedin 6 studies but found significant in only two. General supply of physicians wasfound not to be related to procedures rates although physician characteristics such astype of practice was in one of the three studies examining this factor. Clark (19)found that the weighted proportion of board certified specialists was inverselyrelated to procedure rates for this and several other procedures (hernia repair,hysterectomy, and hemmorhoidectomy). Type of payment was found significant inone of the two studies looking at this variable while country of graduation frommedical school and specialty were both found to be related to procedure rates in the14single study assessing these variables. As might be expected from some of the datareported above, proportion of the population over age 65 and comorbidity were themain patient characteristics found to be significant.TONSILLECTOMY AND ADENOIDECTOMY: There were 17 papers whichexamined both rate variations and assessed factors thought to be associated with theobserved variations; of these two focused on T &A and 15 included this procedurewith other procedures. Overall, across the 17 studies the range of rates is enormous4.18-643 although the variation within studies in generally on the order of 2-4 fold(Table 4). Among the 7 Canadian studies, the rate variations are usually about 2-3fold with two exceptions: Stockwell and Vayda (89) observed an 8 fold variationacross counties in Ontario for a single year and Vayda et al (94) found a 4 foldvariation across Ontario counties over a four year period. Neither of these studiesexamined T&A rates alone. The extremely high rates reported by Chassin et al (17)(267-643) for the United States is a bizarre finding since the population under studywas over 65 years of age. However, all of the rates reported by these authors, withthe exception of appendectomy rates) tend to be considerably higher than the ratesreported by others. The lowest reported rates, again from the United States (4.18­11.1) were based on data from Medicaid (le indigent) patients in Massachusetts andthe objective of the study was to assess the effect of cuts in the fee schedule forMedicaid patients (Shwartz et al (84». Although the types of regions studied, thesources of data, age range, and other methodological details varied across studies,several authors commented that the variations in T&A rates tended to be higherthan any other procedure. Despite the general perception that T&A rates havefallen over the last ten to fifteen years, there is no clear trend in the rates reported inthe literature we reviewed.15In terms of factors examined that might explain the observed rate variations,almost half of the papers used some form of multivariate analysis and in generalsupply factors seemed to account for most of the rate variations. Bed supply, thenumber of general practitioners specifically and physicians generally, and teachinghospital status were all found to be positively related to differences in rates.Surprisingly, the supply of ENT specialists was negatively correlated to T&A rates aswas surgeon age (younger surgeons had lower rates) and those surgeons trained inEngland. As might be expected from the data reported above from Chassin et al andShwartz et al the significant patient characteristics were income and percentage ofthe regional population with insurance coverage. These findings are consistentwith the fairly uniform rates in differing Canadian studies where ability to pay is notan issue.Two studies examined T&A rates specifically. In an English study (Black (10»looked at both tonsillectomy and surgery for glue ear and found a 2 fold regionaldifference in rates of surgery for glue ear, but no regional differences fortonsillectomy; hospital bed supply was not related to the variation in surgery forglue ear, but at the district level ENT supply was related to rates of surgery for glueear but not for tonsillectomy. In general, inter-district rate variations were largerthan inter-regional variations suggesting an effect of the instability of rates insmaller jurisdictions. A Canadian study (Roos et al (75» found a 2 fold variation inT&A rates across 9 regions in Manitoba and only 25% of the procedures werecompatible with predetermined standards for the procedure. There was nodifference in the standards of physicians performing a high proportion of T&As inany region and there was no correlation between T&A rates and respiratorymorbidity (as one might expect), or supply of surgeons.16APPENDECTOMY: This procedure was assessed in 15 papers in two of which itwas the focus of study. Four Canadian studies provided data on this procedure.Overall, despite the assumption that appendectomy is not a discretionary procedure,the rate variations are fairly high, on the order of 3-4 fold in most studies, althoughsome authors found somewhat higher rates. Stockwell and Vayda (89) for exampleobserved a 5 fold difference (11.8-56.6) across regions in Ontario for a single year,while Lichtner and Pflanz (47) found a 4.5 fold difference (26.1-118.8) amongdifferent insurance groups in Germany. Lichtner and Pflanz also reported thehighest rate, (227.5) for women aged 15-20. Interestingly, Bunker's internationalstudy (13) found virtually no variation (21.7-22 males, 18-22.3 females). Chassin (17),as might be expected given the population he studied (Medicare patients) foundvery low rates (2-5). In terms of the Canadian studies, Halliday and Le Riche (37)found only a two fold difference in rates in Alberta (16.3-29 .3), results which areessentially the same in both magnitude and degree of variation to those of Vayda etal (97) for Ontario.Of the independent variables tested and found significantly related to ratevariations, most were supply factors; however these data are either not consistentacross studies or represent the results of single analyses. Bed supply was assessedeach time but found to be correlated with rate variations in three studies whileteaching status of the hospital and surgeon supply, although assessed three timeswere found significant only once. Physician supply generally, specialty and countryof graduation were found significant in a single study. The German study alsofound both month and day of the week significantly related to rate variations, whileWest's study in Wales (l08) found only day of admission to be significant. Withregard to patient characteristics, type of insurance, economic status, and income (allhighly inter-correlated) were found significant in one study. Pathological17confirmation was not related to rate variations suggesting that diagnosticuncertainty or inaccuracy may be a major contributor to the observed variations..CESAREAN SECTION: Primary and repeat C-sections are the procedureswhich have received the most attention in terms .of papers examining a singleprocedure. Thirteen papers (Table 4) provide rates and it is surprising to note howmodest the variations are: they rarely exceed 2 fold differences and are oftenconsiderably less. The four Canadian studies present interesting contrasts sinceAnderson and Lomas (4) found a very small variation across four regions in Ontario(17.1-20.2/100 deliveries), while Vayda et al (94) using a crude rate found almost afour fold difference. Wadhera (98) found a small variation across provinces (10.4­14.8/100 deliveries) for the year 1977, but a three fold rise in rates from 1968 to 1977.This particular finding highlights the critically different impressions that can begained from looking at cross-sectional variations as opposed to longitudinalvariations. The high rate found by Halliday and Le Riche (37) (37.2-66.1) is probablydue to the fact that the denominator is "per 10,000" women instead of per 100deliveries.Based on the eight papers which focus on this procedure, it is clear that neitherphysician convenience (2 papers) nor fear of litigation (l paper) were factorsassociated with variation in rates. Two studies found that C-section rates inteaching hospitals were lower than in community hospitals despite the increasedrisk status of the patients. This effect may be independent of hospital size sinceanother study noted that ability to perform timely procedures reduces the rate.Physician characteristics have a variable impact. Year since graduation forobstetricians does not influence rates, while the ratio of obstetricians per femalepopulation 15-44 years of age does influence rates but only among those with privateinsurance. In the US, differential fees for this procedure and vaginal delivery did18not correlate with ,different rates (3 studies), while type of health coverage did (4studies); thus, individuals with private insurance had the highest rates, whileindividuals covered by large HMOs, had lower rates. Conversely, one study foundthat non-profit hospitals had higher rates than proprietary hospitals while anotherstudy observed lower rates among both self-payers and the indigent, than among theinsured. For this procedure not only are rate variations relatively small, butmonetary factors appear to have greater impact on rate variations than non­monetary factors. In Canada, where monetary factors are irrelevant, the ratedifferences are very small and insufficient research has been carried out on thisspecific procedure to clarify why these small variations occur. The only Canadianstudy to focus on this procedure did so from the perspective of indications for theprocedure (eg breech presentation) rather than physician or patient characteristics,thus variation in clinical policy was identified as the main reason for ratevariations.HYSTERECTOMY: Fifteen papers assessed the effect of independent variableson this procedure and 15 papers provided some descriptive data on rates; only onepaper examined this procedure on its own. The data in Table 4 reveal a modestvariation in rates ranging between 2 and 3 fold although the magnitude of the ratesis quite variable. The extremely low rates reported by Shwartz et. al. (84) (1.51-2.94)are not age standardized, represent before/after data rather than geographicvariations and were collected to assess the effect of cuts in surgical fees for Medicaidpatients. At the other extreme, the data from Cohen (20),and Stockwell and Vayda(89) reveal not only fairly wide rate variations (of the order of 5-6 fold) but the upperrates are quite high. The relatively high variation reported by Stockwell and Vaydamay be due to the fact that only one year was studied. Interestingly, both Cohen, andRoos (69) present data from the same province for essentially the same time period19yet find quite different rate variations (34.5-198/10,000 women at risk over age 25,and 47-128/10,000 women over 25 years of age) respectively. The degree to whichthe difference in denominators has affected the rates is not clear since the lowerbounds are quite similar. There is no strong trend toward higher rates, or ratevariations in Canada compared either to the United States or other countriesalthough the three papers comparing international rates provide virtually identicalrate variations ,which might be considered to be slightly lower than Canadian ratestaken as a whole.With regard to factors responsible for rate variations, bed supply was foundsignificant in only one of the eight studies that looked at this factor, while physiciansupply seemed to be more influential. The number of physicians overall was foundsignificant in the single study that examined this variable, number of generalpractitioners and number of specialists were found significant in three of the fourstudies assessing these factors, while number of surgeons was found significant inonly one of the three papers assessing this factor. Both type of practice, and type ofincome were found to be related to rate variations (in the US) as were hospital type,country of training, specialty, and the number of hysterectomies performed in theprevious six months. Interestingly, patient characteristics have not been examinedextensively, although patient's age, mother tongue (other than English), number ofvisits to a physician for gynecological problems and having had a dilatation andcurretage in the previous two years were significantly related to rate variations inthe single (Canadian) study that examined these factors Roos (69). While it seemsclear that physician factors as opposed to bed capacity are more closely related to ratevariations, the overall importance of physicians characteristics has yet to beadequately assessed relative to patient characteristics, thus the impression gainedfrom many studies that there are hysterectomy-prone physicians is premature.20Moreover, although the rates do vary, the differences as noted above are modest,particularly given the rather discretionary nature of the procedure.PROSTATECTOMY: Eleven papers examined the effect of independentvariables on rate variations in this procedure, and seven papers simply describedrates. No paper examined this procedure alone and five Canadian studies providedboth rates and analysed the reasons for rate variations.Table 4 presents some interesting contrasts with regard to rate variations forthis procedure. While state and provincial variations seem to be modest, on theorder of 2 fold,they are also of the same magnitude (20-40). The Roos' data (72) onpatients over 65 years of age is again of a greater magnitude (125-282) but the ratevariation is not much higher than that reported by others. The striking variationreported by Vayda et al (94) represents rate variation over time rather than acrossgeographic regions. The two papers presenting international comparisons areslightly higher in both rate variation and magnitude. Chassin et al (17) using 1981data for a medicare population found found lower rates (57-98) than the Roos' and aconsiderably smaller rate variation.Although this procedure was included in quite a few papers, little specificmention was made of it and thus information on the relationship between ratevariations and independent variables is limited. Only five of the many factorsstudied were found to be significantly related to rate variations. Teaching status ofthe hospital and the supply of general practitioners and surgeons while explainingsome of the variation had modest effects in multivariate analysis, as didurban/rural residence and, interestingly, having surgery outside the hospital districtof residence. The meaning of this last factor was not explored in depth.21CATARACT SURGERY: Eleven papers assessed rate variations in thisprocedure, including 4 Canadian studies and three studies in which this procedurewas the focus of interest.As can be seen in Table 4, there was a 2-3 fold variation in rates for generalpopulations while those over 65 years of age had both higher rates and a greater ratevariations. Both Vayda et al (97) and Halliday and Le Riche (37) found relativelylow ra tes for each of the provinces they studied. The two studies examininginternational comparisons found higher rates, Bunker (13) found, for example, forrate variations of 47.2-65.3 (males) and 69.1-82.5 (female), while Schact andPemberton (82) found a variation of 29.0-65.5. In the three studies lookingexclusively at this procedure, the two by Bernth-Peterson and associate (8,9) usingdata from Denmark found higher rates than those reported from Canada but therate variation was similar. As can be seen, the Danish data also reveals higher ratesand rate variations for females, probably a function of exposure: females live longerand thus are at higher risk for cataract, and thus for surgery. Sanderson (78) did notreport rates for cataract removal, but assessed variations in terms of percentage ofthe English national rate: the variations was from 73-142%. Chassin et al (17)reported very high rates for the Medicare population, as did Roos for the Manitobapopulation over 65 years of age.Two-thirds of the studies finding significant variables to explain rate variationsused multivariate methods and overwhelmingly, the main factors related to ratevariations were supply factors . However, there was a lack of consistency acrossstudies. For example, bed supply which was assessed four times was found to besignificant only once. Physician supply (whether overall or for specific types ofphysician) and physician characteristics such as country of graduation, specialty,22practice type and source of income, were never found to be significant in more thanone study. Patient characteristics found to be significant were urban/rural residenceand whether or not the surgery took place in the hospital district of the patientsresidence. The fact that these variables are found to be significant for a number ofprocedures is primarily due to the fact that they come from studies examining manyprocedures, using the same independent variables and methodology, thus the extentto which these findings represent true independent relationships for each procedureis unclear. Another consideration of particular relevance to cataract removal is theshift from inpatient to outpatient surgery, a change that has not been adequatelystudied in terms of its effect of rate variations.HERNIA REPAIR: Ten studies examined the effect of independent variables onthis procedure, primarily inquinal repair, and an additional 7 studies reported ratevariations. No paper focused on this procedure by itself and there were threeCanadian papers.Referring to Table 4, it can be seen that, generally speaking, the rate variationsare on the order of 1-2 fold, although Roos and Roos (72) found almost a three foldvariation among patients over the age of 65, and Lewis (46) found a similar ratedifference across 11 regions in Kansas in a single year. While the magnitude of therates varied somewhat, from a low of 14.5-18.9 across three hospital regions inAlberta to the high rates reported by Roos, there were no striking differences ineither magnitude or rate variation between studies reporting American, Canadianor international data. Among the Canadian studies, both Halliday and Le Riche (37)and Vayda et al (97) found lower rates than Roos but this is most likely due to thepopulations from which the rates were derived, in particular the fact that RODSlimited his examination to the 65+ age group which we have already seen tends to23have higher rates. Wennberg et al (101), even without age standardization, foundrates well within those reported by others.Essentially all the studies assessing the impact of independent variablesexamined supply factors rather than population or patient characteristics. Bedsupply was found to be significant in a multivariate analysis in only one of the twostudies examining this factor, while surgeon supply was found significant in onlyone of three studies which assessed this factor. Other factors found to be related torate variations, but tested only once were physician supply in general, generalpractitioner supply, teaching status of the hospital, and hospital location vis a viswhere the patient lived. In terms of patients, characteristics, income and urbanversus rural residence were found significant in the single studies assessing thesefactors.HEMORRHOIDECTOMY: Nine papers analyse the effect of independentvariables on rate variations, and eight papers provide data on rates. No paperfocused on this procedure on its own, and there were two Canadian studies.Three papers provide relatively recent rates. Chassin et al (17) found just overa two fold difference in rates (13-31) in his study of procedure variations in thoseover 65 years of age, while Schact and Pemberton (82) found a slightly lowervariation (13.7-24.5) in their study. reporting data from 1979-81. Griffith et al (36)found much lower rates (6-8) for the general population in Michigan in 1980. Thetwo Canadian studies Halliday and Le Riche (37) and Vayda et al (97) found ratessimilar to Griffiths eba1G (4.7-9.6 and 5-10.9 respectively). A very low rate (2.6-3.2)was found by Shwartz et al (84) in their paper on the effect of a reduction in fees for aMedicaid population Massachusetts, while Wennberg and Gittelsohn (106) found a 6fold difference (3 -19). Lewis (46) found a three fold difference in rates (11.4-34.6)24across eleven regions in Kansas. Overall, recent papers tend to report lower rates,and Canadian studies have reported not only similar rate variations, but lower ratesthan those reported from the United States.In terms of factors assessed, supply variables were most often studied. Bed andsurgeon supply were found significant in only one of three studies examining thesefactors, while teaching status of the hospital, hospital location, general practitionersupply and regional rates of other procedures were each assessed in single studies,but found to be significant. Income and rural versus urban residence were thepatient characteristics found to be significantly related to rate variations. AlthoughLewis found a fairly large rate variation none of the supply variables he assessedwere related to the observed variation. Interestingly, Clark's recent study of ratevariation in Michigan (19) found that rates for this procedure, as for some othersactually increased as the weighted proportion of Board certified surgeons decreased.CORONARY ARTERY PROCEDURES: Three papers describe rate variationsand three papers examine factors related to rate variations. Overall, both themagnitude and rate variations for coronary artery bypass surgery specifically are verysimilar. Roos and Cageorge (70) found about a 2 fold difference in rates over the sixyear period 1978-84 (2.63-5.61) in Manitoba, while Anderson and Lomas (6) reportedrate variations from 1979-85 of 4.28-7.78. Chassin et al (17) in their study of patientover 65 years of age found higher rates (7-13), but the same order of rate variations.In contrast to the other procedures reviewed, most of the data for this procedure isrelatively recent (late seventies to mid-eighties), which may explain the similarity inestimates of rate variations.In terms of factors related to rate variations, only one paper assessed factorsrelated to coronary artery bypass specifically. One paper studied coronary25arteriography by itself, and one paper examined this procedure along with carotidendarterectomy. Two papers assess the effect of regionalization of care and onepaper examined appropriateness..The appropriateness study,Winslow et al (112)found that overall, 56% of CABS were appropriate (ie met the criteria developed by apanel of physicians for the study), 30% were equivocal, and 14% were clearlyinappropriate, although the variation in appropriateness across the three hospitalsstudies was wide: 37 to 78%. A slightly higher proportion of CABS was"appropriate" in the elderly, although another study in the same jurisdiction(Ontario) noted that while CABS rates increased 5 fold among the elderly from 1979to 1986, the overall rate increased only 39%. The key factor influencing rates acrossregions (in Ontario) was the referral centre. Leape et al (43) in a study of coronaryangiography across 23 counties among patients over 65 years of age, usingappropriateness criteria, found a 12 fold variation from lowest to highest usecounties and that inappropriate use accounted for 28% of the variance in rates acrossthe counties while adding equivocally appropriate procedures increased the amountof variance in rate variation to 40%. Specific physician or patient factors were notexamined. It is worth noting that the 12 fold variation across counties obscuresmuch smaller variations in rates since the "high" county is an outlier with the nextlowest rate being almost half the "high" county rate (158 per 10,000 Medicareenrollees for the "high" country, 89 for the next lowest). Moreover, whenaggregated to larger areas the variation in rates from low to high fell from 12 to 2.3.Overall, Leape et al concluded that the variations in rates could not be explained byinappropriate use. Young et al (114) assessed coronary arteriography use amongfamily practitioners and cardiologists in two states using hypothetical patientscenarios and found that family practitioners tended to recommend this proceduremore often than cardiologists. In addition, younger physicians tended torecommend the procedure more often than older physicians. It is not clear,26however, to what extent hypothetical practice reflects actual practice althoughconsidering the variation noted for angiography the hypothetical results probablydo reflect actual practice to some degree.SHORTCOMINGS: The foregoing description of the available information onthe ten most commonly reported procedures has not provided any extensivediscussion of methodological problems. Although we point out particular featuresof specific papers which may account for either very high or low rate variations (egthe relative high rate variations found by Roos et al (72) in their study ofprocedures in the 65 + population), a host of other methodological issues areapparent. As noted above, in the papers we abstracted (Appendix 1), these problemare of two sorts: Limitations and Shortcomings. Limitations were those problemsnoted by the authors and these are presented in Appendix 1. It is useful however, toprovide an overview of what we felt were shortcomings in the literature reviewed,since each of these methodological problems may contribute to the highlyinconsistent nature of the results presented above, and to the general impressionthat dramatic rate variations exist. More general issues will be presented in thediscussion that follows.Table 7 presents a list of the major methodological limitations which werefound in the papers reviewed in depth (Appendix I). The most frequentlyencountered problems were failure to describe the methods of analysis in sufficientdetail so that the nature of the analysis being undertaken was clear and the failure tospecify whether patients were counted more than once, particularly in papersdescribing rate variation for more than one procedure. Surprisingly, the third mostcommon problem was a failure to test for statistical significance when differences inrates were observed. This is particularly critical since, as will be discussed in moredetail below, small area analysis may involve small numbers of procedures within27each area studied, thus producing unstable rates specifically, and the problem ofadequate power to detect a significant differences in rates in general. This latterproblem is highlighted by the fact that a small number of papers utilized very smallstudy populations. Other problems worth commenting on include the failure toutilize a denominator for rate calculation relevant to the objective of the study,failure to include a comment (if not data) on all the independent variables stated tobe included in the analysis and failure to define clearly the unit of analysis. Many ofthese problems are easy to overcome and every effort should be made to do so infu ture research.DISCUSSIONSeveral impressions emerge from this review. In the first place, although itappears large, the literature contains relatively few "good" papers and thus theavailable information on specific procedures is sparse. Moreover, there are fewrecent "good" papers, thus it is not clear if the rate variations reported for theseventies and early eighties still exist, particularly in the United States after theintroduction of prospective payment in the mid-80's. Shwartz et al (84) data on theeffect of reductions in Medicaid fees for surgical procedures in Massachusetts wouldsuggest that prospective payment may have a dramatic effect. In addition, given theconsiderable inconsistency across papers, even for those which focus on a singleprocedure, it is apparent that no clear pattern of factors emerges which would guideefforts to reduce substantially the variations observed. This finding is not surprisingwith respect to patient characteristics since these were tested relatively infrequentlyin the literature, however it is surprising with regard to supply factors (beds,number of physicians, etc.) since these have have generally been the focus of most28papers assessing the effect of independent variables on rate variations. For example,although bed supply has often been assumed to be a major contributor to procedurerate variations, the maximum number of times this variable was examined for anyone procedure was 14 and of these studies it was found significant in only 4. Theinconsistency we found agrees with the observations of Paul-Sheehan et al (62)who concluded that results are inconsistent and conflicting, and that a combinationof community and provider variables provided a better explanation of the observedvariation than either type of variable when analyzed separately. We found datafrom the US to be primarily concerned with insurance issues and thus of somewhatlimited relevance to Canada.Turning to a consideration of some general conceptual and methodologicalissues, few researchers have addressed the possibility that the observed variations inprocedure rates may result from the non-comparability of data sources (anassumption often made in small area analysis that has not been tested) (SauterHughes (80», the clarity with which the jurisdictions are actually defined (Paul­Saheen et al (62) and Diehr et al (28», the effect of the statistical analyses undertaken(Diehr (27), and Diehr et al (28» or even the impact of random fluctuations of ratesand their magnitude relative to the observed rates (Diehr et al (28». The issue ofcomparability arises most dramatically in international comparisons where a host ofdiffering disease definitions, reporting procedures, and data sources may beinvolved. For example, in the United Kingdom primary procedures are reportedwhereas in the United States primary, secondary and tertiary procedures arereported. The degree to which this differential reporting has resulted in higherobserved rates in the United States has not been adequately addressed. Even injurisdictions which may report more than one procedure the protocols for whichprocedure to report as primary may vary which will given rise to differential rates.29The latter problem has been examined by Sauter and Hughes and found to have asignificant impact on reported rates (almost 6% of the ten most common proceduresthey describe with about half of these procedures decreasing in frequency). Despitethis fairly obvious potential for differences in methodology, it is surprising howlittle attention is paid to their implications, not to mention to assessing their impactin the studies we reviewed.With regard to small area analysis, which experienced considerable popularityin the late 1970's and early 1980's, the difficulties of small populations, noted above,raise questions of stability of rates (and thus rate variations), the problem of power,and the fact extreme variations may result from comparing high and low ratejurisdictions without adequate consideration of the overall distribution of rates andthe degree to which the high and low rate jurisdictions are representative of thedata. A similar problem is noted in specific subpopulations, like the elderly, whererelatively small numbers may make rates unstable. Data from Roos et al (72) whichconsistently reveals quite high rates and rate variations across procedures may inpart reflect both the fact that they studied the 65 + population and the fact that someof the health regions in Manitoba have small populations overall. Mobility isanother factor which creates rate differentials and although this factor has beenassessed in a number of studies we would agree with Joffe (40) who commented onthis phenomenon some time ago, that insufficient attention has been paid to itseffect.The issue of the appropriate denominator is critical and we noted above thatthis was not an uncommon shortcoming in the papers we reviewed. A rate inepidemiology is defined as the number of events occurring in a population at risk ofthe event. Most studies use general population estimates as the denominator andwhile this will not distort rate estimates too severely, particularly in large30populations, it will in small populations. The lack of major distortion arisesbecause relative to the population generally, the number of individuals who haveprocedures is small. However, for those procedures in which an organ is removed(eg hysterectomy), obviously the population at risk varies with the rate of theprocedure since a woman cannot have two hysterectomies. Using generalpopulations as denominators may in fact underestimate the rate variations since thedenominator is actually inappropriately large.Where rates and rate variations for jurisdictions are compared, it is importantto undertake some form of age standardization since it is clear from the literaturethat the rate for some procedures is highly related to age (cataract surgery). Thisepidemiological methodology is widely recognised yet many of the early studies byWennberg failed to do this which may account for the relatively high rate variationshe observed which began to create the impression that significant rate variationsexist for which there was no obvious explanation.It cannot be assumed that large rate variations are in fact significant. Allvariations should be tested. As noted above, rate comparisons across small areasmay be unstable both across jurisdictions and over time. Diehr et al (28) haverecently observed, using computer simulations, that relatively large statisticallysignificant rate variations can occur by chance.The issue of the ecologic fallacy is important in this field. The ecologic fallacyarises when data on two variables are derived from different units of analysis orsources of data. For example, appendectomy rates and socioeconomic status (SES). Ifone observes that high rates occur in jurisdictions of low SES and that low ratesoccur in jurisdictions of high SES, and although there may be a statisticallysignificant association negative association between these two variables, causal31inferences are questionable because one does not know if it is the individuals withlow SES who have a higher frequency of appendectomy; the unit of analysis is notthe individual. Wilson (111) for example, found a high correlation between mediancommunity income and overall community mortality rates and general surgeryrates for patients over 65 years of age in Michigan, yet it remains unclear if thiscorrelation, particularly with regard to income, would be observed at the level of theindividual. Thus analyses seeking to associate various factors with rate variationsmust always be cognizant of this problem. Unfortunately, this is rarely the case aswe and Paul-Saheen et al (62) have observed.Finally, as noted at the outset, McPherson (50) raised the issue of clinicaluncertainty as a factor in producing rate variations. He did not define this conceptvery precisely thus it is not immediately apparent what is meant by it. However,clinical uncertainty has an intuitive appeal because no physician can be absolutelycertain about any diagnosis or the efficacy of any procedure in any patient. The datafrom Young et al (114) on the differences between family practitioners andcardiologists in recommending coronary arteriography may reflect differences incertainty. A related concept is practice style. This concept was first invoked byWennberg and Gittelsohn (106) to explain small area variations in his early studies.In a sense it is an explanation of exclusion but other than the recent paper by Stano(88) (which did not really clarify what the concept means) no operationalization ofthis concept has been attempted.Given these conceptual and methodological issues, how then is one to judgethe degree to which procedure rate variations are an important phenomenonworthy of attention and, if possible, action? In addition how is one to view themagnitude of rate variations? Is a two fold variation (if valid) of sufficientimportance to warrant action or should one concentrate on rate variations of greater32magnitude? For the ten procedures we have reviewed in depth, it is clear that thereis considerable variability in rate variations, some of which is immediatelyexplainable (at least on a speculative basis since precise or relevant data are notavailable) and some of which remains obscure. It is claimed for nearly all theprocedures examined in the literature, that the United States has higher rates thanthe United Kingdom (with Canada having rates somewhere in between), that ratesfor discretionary (elective) procedures are higher than rates for non-discretionaryprocedures, and that these variations arise primarily because of differences in supplyvariables. The main problem with this inference is that supply variables have beenmost often studied and have not often been studied along with other variables,particularly differences in morbidity. Moreover, even given the relatively highfrequency with which supply variables have been assessed, it is clear from Table 4that there is a lack of consistency in the frequency with which these variables havebeen found to be significant, and even if significant whether they really explainsufficient amounts of the observed variation to warrant attention. In terms ofpatient characteristics, age seem to be a fairly consistent determinant of rates in thatmost studies examining surgical rates among the elderly (those of 65 years of age)find both higher rates and wider rate variations than observed for generalpopulations. Unfortunately, single studies rarely if ever looked at procedure ratevariations in younger and older age groups using the same methodology, thusmethodological issues may be in part responsible for the perception that olderpatients have higher rates. Nevertheless, since this population is at greater risk ofuntoward reactions to surgery, is growing in number and percent of the population,tends to stay in hospital longer, and may represent a group in whom clinicaluncertainty is high, the general observation that this group has higher ratesdeserves attention.33Given the very large number of possible explanatory variables, it is perhaps notsurprising that in general, for most procedures, the rate variations are on the orderof 2-4 fold and thus we remain to be convinced that procedure rate variations havebeen clearly demonstrated to be a major problem requiring urgent attention.However, we must also state that the literature to date indeed suggests thatconsiderable rate variations may exist and that a concerted effort to document andanalyse such variation is a valid policy objective. Moreover, we believe that thereis an urgent need to assess the relationship between procedure rates and morbidityand health status at the community level. This can either be done directly whichwould require good morbidity estimates, which are often unavailable, or indirectly,by assessing the appropriateness of procedures in individual patients. The indirectapproach, as exemplified in the work of Winslow et al (112) provides in our view abetter index of the meaning of rate variations since it affords not only a method foraddressing the issue of clinical uncertainty or practice style, but allows for theassessment of supply and patient variables on appropriateness rather then simplyoverall rates. There is an urgent need to continue the development ofappropriateness criteria; one of the major outstanding issues in the rate variationsliterature being whether low or high rates are "good" or "bad". Wennberg (lOS)noted in his study of medical and surgical variations in New Haven and Boston thatmedical procedures had higher rates of variation than surgical procedures amongpatients 65 years of age an older. Thus the issue of low rates representingunderutilization and high rates representing overutilization may be more germaneto medical than surgical procedures, especially among the elderly in whom we havenoted a fairly consistent pattern of higher rates and greater rate variations.It is critical that research focus on individual procedures. As can be seen in theabstracts (Appendix 1) many papers assess variations in many procedures and is not34always clear in the analyses which variables are associated with rate variations forwhich procedure. Moreover, in papers reporting variations for more than oneprocedure, the relationships between variables and procedure variations may bemixed leading to complexity in reporting and confusion in interpretation.Assessing single procedures both simplifies the analysis and clarifies theconclusions.Such a research agenda is a departure from the supply focus of much of theresearch to date which has failed to yield any consistent patterns of supply (either ofphysician or beds) effect. Although Roemer's Law ("a built bed is a filled bed ") maybe valid, and despite the intuitive appeal of the relationship between number ofphysicians (or surgeons) and rate of procedures, these effects are neither large norconsistent, thus other explanations must be sought. Moreover, it is important froma policy perspective that assessment of procedure variations account for underlyingestimates of need in addition to supply side factors since altering supply, which maybe relatively easy to do, may in fact be inappropriate even though supply factors may(to some degree) be associated with the observed variations. For example,consideration needs to be given to estimating the variation in rates of proceduresthat should have been done, but were not. Since most commentators observe thatrate variations per se tell us nothing about appropriateness, variation in legitimateunmet need should be an integral part of future research. Altering supply factorswithout concern for the critically important issue of appropriateness may well domore harm than good.In summary, given the substantial Canadian contribution to the procedurevariations literature, it is our view that Canada is in an excellent position toestablish a significant research agenda in this area. Although we have noted manymethodological limitations and shortcomings in both the international and35Canadian literature to date, most of the these problems are relatively easy toovercome. We believe that the Canadian health care system is an ideal setting forthe continued development of procedure variation methodology, particularly theassessment of appropriateness. The principles of comprehensiveness, universality,portability, and accessibility enshrined in the Canada Health Act create the perfectlaboratory for the evaluation of those structural and functional factors which mayinfluence rate variations, other than individual ability to pay, and thus whichmaintain the focus on the main policy issues underlying our collective interest inan efficient and effective health care system.BIBLIOGRAPHY1. Acker, D.B., Haas, S., O'Brien, E., Donahue, C.L., McGuirk, P., Sachs, B.P., 1988.Cesarean birth rate: Small-geographic-area analysis. Am J Obstet Gynecol159:386-88. (A-II, p. 2) (T-4)2. Amirikia, H., Zarewych, B., Evans, T.N., 1981. Cesarean Section: A 15 yearreview of changing incidence, indications, and risks. Am J Obstet Gynecol140:81-86. (A-II, p. 3)3. Anderson, G.M., Lomas, J., 1984. Determinants of the Increasing CesareanBirth Rate; Ontario Data 1979 to 1982. N Engl J Med 311:887-92. (A-II, p. 4)4. Anderson, G.M., Lomas, J., 1985. Explaining variations in cesarean sectionrates: Patients, facilities, or policies? Can Med Assoc J 132:253-59. (A-I, p. 3)(T-4)5. Anderson, G.M., Lomas, J., 1988. Monitoring the Diffusion of a Technology:Coronary Artery Bypass Surgery in Ontario. Am J Public Health 78(3):251-254.(A-II, p. 14)6. Anderson, G.M., Lomas, J., 1989. Regionalization of Coronary Bypass Surgery:Effects on Access. Medical Care 27(3):288-296. (A-I, p.I)7. Barnes, B.A., O'Brian, E., Comstock, C., D'Arpa, D.G., Donahue, c.r.. 1985.Report on Variation in Rates of Utilization of Surgical Services in theCommonwealth of Massachusetts. JAMA 264(3):371-408. (A-II, p. 10)8. Bernth-Petersen, P., 1985. Regional Variations in Indications for CataractSurgery? An Analysis of the Pre-IOL period. Acta Opthalmologica 63:281-284.(A-I, p. 5)9. Bernth-Petersen, P., Bach.E., 1983. Epidemiologic Aspects of Cataract Surgery:II Regional Variations in Frequencies. Acta Opthalmologica 61:397-405.(A-I, p. 7) (T-4)10. Black, N., 1985. Geographical Variations in the use of Surgery for Glue Ear.Journal of the Royal Society of Medicine 78:641-48. (A-I, p. 9)11. Bloor, M.J., Venters, G.A., Samphier, M.L., 1978. Geographical variation inthe incidence of operations on the tonsils and adenoids; An epidemiologicaland sociological investigation, Part I. Journal of Laryngology and Otology92:791-801. (A-II, p. 19) (T-4)212. Bombardier, C., Fuchs, V.R, Lillard, L.R, Warner, K.E., 1977. SocioeconomicFactors Affecting the Utilization of Surgical Operations. N Eng J Med 297:699­70S. (A-I, p. 12)13. Bunker, J.P., 1970. Surgical Manpower: A Comparison of Operations andSurgeons in the United States and in England and Wales. N Eng J Med282(3):135-144. (A-TI, p. 6) (T-4)14. Cageorge, S.M., Roes, L.L., 1984. When Surgical Rates Change: Workload andTurnover in Manitoba, 1974-1978. Medical Care 22(10):890-900. (A-I, p.1S)15. Cageorge, S.M., Roos, L.L., Dazinger, R, 1981. Gallbladder Operations: APopulation-Based Analysis. Medical Care 19(5):510-25. (A-I, p.18) (T-4)16. Carpenter, M.W., Soule, D., Yates, W.T., Meeker, C.l., 1987. PracticeEnvironment is Associated with Obstetric Decision Making RegardingAbnormal Labour. Obstet Gynecol 70:657-61. (A-I, p. 21)17. Chassin, M.R, Brook, RH., Park, R.E., Keesey, J., Fink, A., Kosecoff, J., Kahn,K., Merrick, N., Solomon, D.H., 1986. Variations in the Use of Medical andSurgical Services by the Medicare Population. N Engl J Med 314:285-90.(A-I, p. 24) (T-4)18. Chassin, M.R, Kosecoff, J., Soloman, D.H., Brook, RH., 1987. How CoronaryAngiography is Used: Clinical Determinants of Appropriateness. JAMA258(18):2543-2547. (A-TI, p. 13)19. Clark, J.D., 1990. Variation in Michigan Hospital Use Rates: Do Physician andHospital Characteristics Provide the Explanation? Soc. Sci. Med. 30(1):67-82.(A-I, p. 27)20. Cohen, M.M., 1986. Tubal Sterilization in Manitoba. Canadian Journal ofPublic Health 77:114-118. (A-TI, p. 16) (T-4)21. Connell, F.A., Day, RW., Logerfo, J.P., 1981. Hospital of Medicaid Children:Analysis of Small Area Variations in Admission Rates. Am J Public Health71:606-613. (A-I, p. 31)22. Copenhagen Collaborating Centre for the Study of Regional Variations inHealth Care. 1985. CCC Bibliography on Regional Variations in Health Care.Geneva: World Health Organization.23. Copenhagen Collaborating Centre for the Study of Regional Variations inHealth Care. Number 2: 1987. Bibliography on Regional Variations in HealthCare. Geneva: World Health Organization.324. Coulter, A., McPherson, K, 1985. Socioeconomic variations in the use ofcommon surgical operations. Soc Sci Med 291:183-87. (A-II, p. 10)25. Coulter, A., McPherson, K., Vessesy, M., 1988. Do British Women UndergoToo Many or Too Few Hysterectomies? Soc Sci Med 27:987-94.(A-II, p. 17) (T-4)26. Detmer, D.E., Tyson, T.J. 1978: Regional Differences in Surgical Care BasedUpon Uniform Physician and Hospital Discharge Data. Ann Surg 187(2):166­169. (A-I, p. 34) (T-4)27. Diehr, P., 1984. Small Area Statistics: Large Statistical Problems. Am J PubHealth 74(4):313-314.28. Diehr, P., Cain, K., Connell, F., Volinn, E., 1990. What is Too MuchVariation? The Null Hypothesis in Small-Area Analysis. Health ServicesResearch 24(6):740-771.29. Dyck, F.J., Murphy, F.A., Murphy, J.K., Road, D.A., Boyd, M.S., Osborne, E., DeVlieger, D., Korchinski, B., Ripley, C., Bromley, A.T., Innes, P.B., 1977. Effectof Surveillance on the Number of Hysterectomies in the Province ofSaskatchewan. N Engl J Med 296:1326-28. (A-II, p. 15)30. Evans, M.L, Richardson, D.A., Sholl, J.S., Johnson, B.A., 1984. CesareanSection: Assessment of the Convenience Factor. The Journal ofReproductive Medicine 29(9):670-76. (A-I, p. 36) (T-4)31. Fowkes, F.G.R., 1980. Cholecystectomy and Surgical Resources in Scotland.Health Bulletin 38(3):126-32. (A-I, p. 38) (T-4)32. Fraser, W., Usher, R.H., McLean, F.H., Bosenberry, C., Thomson, M.E.,Kramer, M.S., Smith, L.P., Power, H., 1987. Temporal variation in rates ofcesarean section for dystocia: Does convenience playa role? Am J ObstetGynecol 156:300-04. (A-II, p. 5)33. Gilstrap, L.C., Hauth, J.e., Toussant, 5., 1984. Cesarean Section: ChangingIncidence and Indication. Obstetrics and Gynecology 63:205. (A-II, p. 4)34. Gittelsohn, A.M., Wennberg, J.E., 1977. On the Incidence of Tonsillectomyand Other Common Surgical Procedures. In Costs, Risks, and Benefits ofSurgery, ed. Bunker, J.P., Barnes, B.A., Mosteller, F.. New York: OxfordUniversity Press. (A-II, p. 18) (T-4)35. Gornick, M., 1982. Trends and Regional Variations in Hospital Use underMedicare. In Regional Variations in Hospital Use, ed. D.L. Rothberg.Lexington, Massachusetts: Lexington Books. (A-II, p. 8)436. Griffith, J., Wilson, P.A., Wolfe, RA., Bischak, D.P., 1985. Clinical Profiles ofHospital Discharge Rates in Local Communities. Health Services Research20(2):131-51. (A-I, p. 40) (T-4)37. Halliday, M.L., Le Riche, W.H., 1987. Regional Variations in Surgical Rates,Alberta 1978, and the Relationship to Characteristics of Patients, DoctorsPerforming Surgery and Hospitals Where the Surgery was Performed.Canadian Journal of Public Health 78:193-200. (A-I, p. 42) (T-4)38. Haynes de Regt, R, Minkoff, H.L., Feldman, J., Schwarz, RH., 1986. Relationof Private or Clinic Care to the Cesarean Birth Rate. N Eng J Med 315:619-24.(A-I, p. 45) (T-4)39. Holland, C., Heaton, K.W., 1972. Increasing Frequency of Gall BladderOperations in the Bristol Clinical Area. British Medical Journal 3:672-675.(A-IT, p. 14)40. Joffe, J., 1979. Mobility Adjustments for Small Area Utilization Studies.Inquiry 16:350-355.41. Jonsson, B., Silberberg, R, 1982. Variations between and within Countries inHospital Care for Peptic Ulcer: A Comparison between Denmark and Sweden.Scand J Soc Med 10:63-69. (A-1, p. 17)42. Lagoe, RJ., 1986. Differences in Hospital Discharge Rates, A CommunityBased Analysis. Medical Care 24:868-872. (A-IT, p. 11)43. Leape, L.L., Park, RE., Soloman, D.H., Chassin, M.R, Kosecoff,J., Brook, RH.,1990. Does Inappropriate Use Explain Small-Area Variations in the Use ofHealth Care Services. JAMA 263(5):669-672. (A-I, p. 47)44. Legnini, M.W., Thompson, J.D., DerSimonian, R., Lynch, J.T., 1990. Changesin Connecticut Hospital Use rates: Have Small-Area Variations BeenAffected. Inquiry 27:51-60. (A-IT, p. 13)45. Leopold. D.A., Lagoe, RJ., 1986. Patterns of Otolaryngologic SurgeryUtilization in Syracuse, NY. Arch Otolaryngol Head Neck Surg 112:623-627.(A-IT, p. 19)46. Lewis, C.E., 1969. Variations in the Incidence of Surgery. N Engl J Med281(160):880-84. (A-I, p. 50) (T-4)47. Lichtner, 5., Pflanz. M., 1971. Appendectomy in the Federal Republic ofGermany: Epidemiology and Medical Care Patterns. Medical Care 9(4):311-330.(A-I, p. 52) (T-4)548. LoGerfo, J.P., Efird, R.A., Diehr, P.K., Richardson, W.C., 1979. Rates of SurgicalCare in Prepaid Group Practices and the Independent Setting: What are theReasons for the Differences? Medical Care 17(1):1-7. (A-I, p. 55)49. McLaughlin, C.G., Normolle, D.P., Wolfe, RA., McMahon, L.F., Griffith, J.R,1989. Small-Area Variation in Hospital Discharge Rates: Do SocioeconomicVariables Matter. Medical Care 27(5):507-521. (A-I, p. 57)50. McPherson, K., 1989. International differences in medical care practices.Health Care Financing Review/Annual Supplement 21-32. (A-II, p. 12)51. McPherson, K., Epstein, A., Jones, L., 1981. Regional Variations in the Use ofCommon Surgical Procedures: Within and Between England and Wales,Canada, and the United States of America. Soc. Sci. Med. 15A:273-88.(A-I, p. 62)52. McPherson, K., Strong, P.M., Jones, L., Britton, B.J., 1985. Do CholecystectomyRates correlate with Geographic Variations in the Prevalence of Gallstones.Journal of Epidemiology and Community Health 39:179-82. (A-I, p. 60)53. McPherson, K., Wennberg, J.E., Hovind, O.B., Clifford, P., 1982. Small-areaVariations in the Use of Common Surgical Procedures: An InternationalComparison of New England, England and Norway. N Eng1 J Med 307:1310­1314. (A-II, p. 8)54. Merrick, N.J., Brook, R.H., Fink, A., Soloman, D.H., 1986. Use of CarotidEndarterectomy in Five California Veterans Administration Medical Centers.JAMA 256(18):2531-2535. (A-I, p. 66)55. Mindell, W.R., Vayda, E., Cardillo, B., 1982. Ten-Year Trends in Canada forSelected Operations. CMA Journal 127:23-27. (A-I, p. 68) (T-4)56. Minkoff, H.L., Schwarz, RH., 1980. The Rising Cesarean Section Rate: Can ItSafely Be Reversed? Obstetrics and Gynecology 56(2)135-143. (A-II, p. 2)57. Mitchell, J.B., Cromwell, J., 1982. Variations in Surgery Rates and the Supplyof Surgeons. In Regional Variations in Hospital Use, ed. D.L. Rothberg.Lexington, Massachusetts:Lexington Books. (A-I, p. 70)58. Munoz, E., Boiardo, R, Mulloy, K., Goldstein, J., Tenenbaum, N., Wise, L.,1990. Economies of scale, Physican Volume for Neurosurgery Patients, andthe Diagnosis-related Group Prospective Hospital Payment System.Neurosurgery 26(1):156-161. (A-II, p. 17)659. Newton, E.R., Higgins, C.S., 1989. Factors Associated with Hospital-SpecificCesarean Birth Rates. Journal of Reproductive Medicine 34(6):407-11.(A-I, p. 73)60. Nilsen, S., Bergsjo, P., Lokling, A., Skarsten, K.W., Johannessen, K.H., Blaas,H., 1983. A Comparison of Cesarean Section Frequencies in Two NorwegianHospitals. Acta Obstet Gynecol Scand 62:555-561. (A-I, p. 76)61. Ohio Department of Human Services, 1988. Variations in Medicaid SurgicalRates AcrossCounties: An Ohio Department of Human services study-1982­1985. Ohio Medicine July 1988:571-575. (A-IT, p. 12)62. Paul-Shaheen, P., Clark, J., Williams, D., 1987. Small Area Analysis: AReview and Analysis of the North American Literature. J Health Politics,Policy and Law 12(4):741-807.63. Phillips, R.N., Thornton, J., Gleicher, N., 1982. Physician Bias in CesareanSections. JAMA 248:1082-1084. (A-IT, p. 3)64. Placek, P.J., Taffel, S.M., 1980. Trends in Cesarean Section Rates for the UnitedStates, 1970-78. Public Health Reports 95(6):540-48. (A-IT, p. 3) (T-4)65. Placek, P.J., Taffel, S., Moien, M., 1981. Cesarean Section Delivery Rates:United States, 1981. Am J Public Health 73:861-62. (A-IT, p. 3) (T-4)66. Reid, A.J., Carroll, J.C., Ruderman, J., Murray M.A., 1989. Differences inIntrapartum Obstetric Care Provided to Women at Low Risk by FamilyPhysicians and Obstetricians. Can Med Assoc J 140:625-33. (A-I, p. 79)67. Roos, L.L., 1983. Supply, Workload and Utilization: A Population-BasedAnalysis of Surgery in Rural Manitoba. Am J Public Health 73:414-421.(A-TI, p. 9)68. Roos, N.P., 1984. Hysterectomies in One Canadian Province: A New Look atRisks and Benefits. Am J Public Health 74:39-46. (A-IT, p. 16)69. Roos, N.P., 1984. Hysterectomy: Variations in Rates Across Small Areas andAcross Physicians' Practices. Am J Public Health 74(4):327-35. (A-I, p. 90) (T-4)70. Roos, L.R., Cageorge, S.M., 1987. Innovation, Centralization and Growth:Coronary Artery Bypass Graft Surgery in Manitoba. Unpublished paper.(A-I, p. 85) (T-4)71. Roos, N.P., Lyttle, D., 1985. The Centralization of Operations and Access toTreatment: Total Hip Replacement. Am J Public Health 75(2):130-133.(A-IT, p. 14)772. Roos, N.P., Roos, L.L., 1981. High and Low Surgical Rates: Risk Factors forArea Residents. Am J Public Health 71:591-600. (A-I, p. 82) (T-4)73. Roos, N.P., Roos, L.L., 1982. Surgical Rate Variations: Do They Reflect theHealth or Socioeconomic Characteristics of the Population? Medical Care20(9):945-58. (A-I, p. 93)74. Roos, N.P., Roes, L.L., Henteleff, P.D., 1977. Elective Surgical Rates-Do HighRates Mean Lower Standards?: Tonsillectomy and Adenoidectomy inManitoba. N Eng J Med 297:360-365. (A-I, p. 88)75. Roos, N.P., Shapiro, E., Roos, L.L., 1984. Aging and the Demand for HealthServices: Which Aging and Whose Demand? The Gerontologist 24:31-36.(A-II, p. 10) (T-4)76. Richmond, G.M., Prior, J.T., 1989. Variations in the Delivery of Medical Carein Central New York. New York State Journal of Medicine 89:63-68.(A-TI, p. 12)77. Rutkow, I.M., Starfield, a.H., 1984. Surgical Decision Making and OperativeRates. Arch Surg 119:899-905. (A-I, p. 96)78. Sanderson, H.F., 1980. Regional Variation in Cataract Extraction Rates andTheir Relationship with Resource Supply and Need. Journal of the RoyalSociety of Medicine 73:492-96. (A-I, p. 99) (T-4)79. Santos-Eggimann, B; Paccaud, F., Gutzwiller, F., 1989. Coronary AngiographyRates in Switzerland-How Do They Vary? Soc. Sci. Med. 28(2):115-120.(A-II, p. 13)80. Sauter, V., Hughes, E., 1983. Surgical Utilization Statistics: SomeMethodologic Considerations. Medical Care 11(3):370-377.81. Savage, W., 1983. Hysterectomy and sterilisation rates. The Practitioner227:839-845. (A-II, p. 15) (T-4)82. Schact, P.J., Pemberton, A., 1985. What is Unnecessary Surgery? Who ShallDecide? Issues of Consumer Sovereignty, Conflict, and Self-Regulation. Soc.Sci. Med. 20:199-206. (A-II, p. 11) (T-4)83. Scott, H.D., Mackie, A., 1975. Decisions ,to Hospitalize and to Operate: ASocioeconomic Perspective in an Urban State. Surgery 77(2):311-317.(A-II, p. 6)884. Shwartz, M., Martin, S.G., Cooper, D.D., Ljung, G.M., Whalen, B.J., Blackburn,J., 1981. The Effect of a Thirty Percent Reduction in Physician Fees onMedicaid Surgery in Massachussets. Am J Public Health 71(4):370-375.(A-I, p. 101) (T-4)85. Sofer, T., Roos, N.P., Nelson, N., 1983. Hysterectomy in Manitoba-1970-1978:Patterns of Practice and Changes Over Time. Canadian Journal of PublicHealth 74:100-05. (A-II, p. 16)86. Stafford, R.S, 1990. Cesarean Section Use and Source of Payment: AnAnalysis of California Hospital Discharge Abstracts. Am J Pub Health 80:313­315. (A-I, p. 104) (T-4)87. Stang, H., 1987. Tonsillectomy referrals in an HMO-based population.Minnesota Medicine 70:283-286. (A-II, p. 8)88. Stano, M. 1986. A Further Analysis of the "Variations in Practice Style"Phenomenon. Inquiry 23:176-182. (A-I, p. 106)89. Stockwell, H., Vayda, E., 1979. Variations in Surgery in Ontario. 1. MedicalCare 17(4):390-396. (A-I, p. 109) (T-4)90. Taffel, S.M., Placek, P.J., 1985. Letters to the Editor; One-Fifth of 1983 USBirths by Cesarean Section. Am J Public Health 75:190. (A-II, p. 5)91. Taffel, S.M., Placek, P.J., Moien, M., 1989. Cesarean Section Rate Levels Off in1987. Family Planning Perspectives 21(5):227-228. (A-II, p. 5)92. Vayda, E., 1973. A Comparison of Surgical Rates in Canada and in Englandand Wales. N Eng J Med 289(23):1224-1229. (A-II, p. 6) (T-4)93. Vayda, E., Anderson, G.D., 1975. Comparison of Provincial Surgical Rates in1968. Canadian Journal of Surgery 18:18-26. (A-I, p. 119)94. Vayda, E., Barnsley, J., Mindell, W., Cardillo, B., 1984. Five Year Study ofSurgical Rates in Ontario's Counties. Can Med Assoc J 131:111-115.(A-I, p. 111) (T-4)95. Vayda, E., Mindell, W.R., Mueller, C.B., Yaffe, B. 1982. Measuring SurgicalDecision Making with Hypothetical Cases. Can Med Assoc J 127:287-90.(A-I, p. 116)96. Vayda, E., Mindell, W. R., Rutkow, 1M., 1982. A Decade of Surgery in Canada,England and Wales, and the United States. Arch Surg 117:846-853.(A-II, p. 8) (T-4)997. Vayda, E., Morison, M., Anderson, G., 1976. Surgical Rates in the CanadianProvinces 1968 to 1972. The Canadian Journal of Surgery 19:235-42.(A-I, p. 113) (T-4)98. Wadhera, S., Nair, C., 1982. Trends in Cesarean Section Deliveries, Canada,1968-1977. Canadian Journal of Public Health 73:47-51. (A-II, p. 4) (T-4)99. Walker, A., [ick, H., 1979. Temporal and Regional Variation in HysterectomyRates in the United States, 1970-1975. Am. J. Epidemiolll0(1):41-46.(A-II, p. 15) (T-4)100. Wennberg, J E., 1979. Factors Governerning Utilization of Hospital Services.Hospital Practice September:115-127. (A-II, p. 7)101. Wennberg, J.E., Barnes, B.A., Zubkoff, M., 1982. Professional Uncertainty andthe Problem of Supplier-Induced Demand. Soc. Sci. Med. 16:811-24.(A-II, p. 9)102. Wennberg, J.E., Blowers, L., Parker, R., Gittelsohn, A.M., 1977. Changes inTonsillectomy Rates Associated With Feedback and Review. Pediatrics59(6):821-826. (A-II, p. 18) (T-4)103. Wennberg, J.E., Bunker, J.P., Barnes, B. 1980. The Need for Assessing TheOutcome of Common Medical Practices. Ann. Rev. Public Health 1:277-295.(A-II, p. 7)104. Wennberg, J.E., Freeman, J.L., Culp, J., 1987. Are Hospital services Rationedin New Haven or Over-Utilized in Boston? The Lancet, May 23, 1987:1185­1188. (A-II, p. 11)105. Wennberg, J.E., Freeman, J.L., Shelton, R.M., and Bubolz, T.A., 1989. HospitalUse and Mortality Among Medicare Beneficiaries in Boston and New Haven.New Eng J Med 321(17):1169-73106. Wennberg, J.E., Gittelsohn, A., 1975. Health Care Delivery in Maine I:Patterns of Use of Common Surgical Procedures. The Journal Of the MaineMedical Association 66(5):123-130,149. (A-II, p. 7) (T-4)107. Wennberg J., Gittelsohn, A., 1982. Variations in Medical Care among SmallAreas. Scientific American 246:120-131. (A-II, p. 9) (T-4)108. West, R.R., Carey, M.J., 1978. Variation in Hospital Admission forAppendicitis in Wales. British Medical Journal 1:1662-1664. (A-I, p. 122)109. Wilke, K., Storr-Paulsen, A. 1986. Indications of Trabeculectomy in Open­Angle Glaucoma. Acta Ophthalmomogica 64:258-262. (A-II, p. 20)10110. Williams, RL., 1983. Controlling the Rise in Cesarean Section Rates by theDissemination of Information from Vital Records. Am J Pub Health 73:863­867. (A-I, p. 124) (T-4)111. Wilson, P.A., 1981. Hospital Use by the Aging Population. Inquiry 18:332-44.(A-I, p. 126)112. Winslow, C.M., Kosecoff, J.B., Chassin, M., Kanouse, D.E., Brook, R.H., 1988.The Appropriateness of Performing Coronary Artery Bypass Surgery. JAMA260(4):505-509. (A-I, p. 128)113. Yoshida, Y., Yoshida, K., 1976. The High Rate of Appendectomy in Japan.Medical Care 14(11):950-957. (A-II, p. 2) (T-4)114. Young, M.J., Fried, L.S., Eisenberg, J., Hershey, J., Williams, S., 1987. DoCardiologists Have Higher Thresholds for Recommending CoronaryArteriography than Family Physicians? Health Services Research 22(5):623­635. (A-I, p. 130)TABLE 1REPORTED ADMISSION RATES FOR SELECTED PROCEDURES: SELECTED COUNTRIES FOR WHICH DATA WERE REPORTED,1980INGUINAL EXPLOR.COUNTRY TONSILL- CORONARY CHOLECYST- HERNIA LAPAR- PROSTAT- HYSTER- OPERATION APPEND-EcrOMY BYPASS EcrOMY REPAIR OTOMY EcrOMY EcrOMY ON LENS EcrOMYAUSTRALIA 115 32 145 202 99 183 405 101 340CANADA 89 26 219 224 105 229 479 139 143DENMARK 229 21 234 255 118 248IRElAND 256 4 91 100 52 124 123 64 245JAPAN 61 1 2 67 90 35 244NETHERLANDS 421 5 131 175 116 381 68 149NEWZEALAND 412 2 99 211 110 191 431 95 169~AY 45 13 30 78 238 71 64SW'ED8" 65 140 206 11 1 48 145 168SWITZERlAND 51 49 116 68 22 74UNITED KINGDOM 26 6 78 154 116 144 250 98 131UNITED STATES 205 61 203 238 41 308 557 294 130NOTES: These figures are not age standarized and assume equal proportions and assume equal proportions of men and women. Someare likely to be incomparable for artifactual reasons.SOURCE: Organization for Economic Cooperation and Development: Health Data File, 1989from MacPherson (50)TABLE 2BREAKDOWN OF VARIATION LITERATUREInternationalUtilization of Procedures Canadian American Comparisons Specific CountryData Presented & Analyzed 17 23 4 7 51re Independent VariablesData Presented & Analyzed 5 13 5 4 27in Utilization TrendsData Presented on Trends 6 15 4 3 28- no AnalysisSubtotal 106SpecificProcedures not AssessedEditorial, Thinkpiece, Letter, PolicyModels (Analytic Approaches)Heal th ServicesPhysician ServicesHospital ServicesIncidence of DiseaseTarAL624838255827364TABLE 3PERIODICALS REVIEWEDActa Obstetrica et Gynecologica ScandinavicaActa OphtamologicaAmerican Journal of Obstetrics & GynecologyArchives of Internal MedicineArchives of Otolaryngology - Head and Neck SurgeryAnnals of Internal MedicineAnnals of SurgeryAnnual Review of Public HealthArchives of SurgeryBritish Journal of Preventive and Social MedicineBritish Medical JournalCanadian Journal of Public HealthCanadian Journal of SurgeryCanadian Medical Association JournalCommunity HealthCommunity MedicineGerontologistHealth AffairsHealth BulletinHeal th Services ReportsTable 3 (Cont)Health Services ResearchHospital PracticeHospitalsInquiryInternational Journal of Health ServicesJournal of Chronic DiseasesJournal of Health and Social BehaviorJournal of Laryngology and OtologyJournal of Public HealthJournal of the American Board of Family PracticeJournal of the American Medical Association (JAMA)Journal of the Maine Medical AssociationJournal of the Royal Society of MedicineLancetMedical CareMilbank QuarterlyNew England Journal of MedicineOhio MedicinePediatricsPractitionerProceedings of the Royal Society of MedicineRhode Island Medical JournalTable 3 (Cont)Scandinavian Joumal of Social MedicineScienceScientific AmericanSocial Science and MedicineSocial Security BulletinSocio-Economic Planning SciencesSurgerySurgical Clinics of North AmericaSurgical ForumTABLE 4VARIATION IN RATES OF SELECfED PROCEDURESHemorrhol·YEAR Hemla Repair dectomy Hysterectomy Cholecystectomy Cesarean1965 17.8·42.5 11.4·34.6 12.1 - 42.3Regions in State Regions in State Regions in StateLewis (46) Lewis (46) Lewis (46)1965w19661967·1968 21.3·47.2 14.7 -17§International InternationalVayda (92) Vayda (92)1969-1971 38·52 30·60 18·53Regions in State Regions in State Regions in StateGittlesohn & Gittlesohn & Gittlesohn &Wennberg (34) Wennberg (34) Wennberg (34)1968-1972 13.5·24.6 5 ·10.9 43.9·70.9 18.4·40.9Province Province Province ProvinceVayda et al (97) Vayda et al (97) Vayda et al (97) Vayda et al (97)1971 ·1972 31·85.... 36·99....Other OtherRoos & Roos (72) Roos & Roos (72)1973 27.8·37.9 3 -19 20.4·33.6 21·29Regions in State Regions in State Regions in State Regions in StateDetmer & Tyson Wennberg & Detmer & Tyson Detmer & Tyson(26) Gittlesohn (106) (26) (26)35-60 39·93 27 - 55 FemaleRegions in State Regions in State Region in StateWennberg & Wennberg & Wennberg &Gittlesohn (l06) Gittlesohn (106) Gittlesohn (106)1974 41-202 20.6 -101.7Regions in Regions InProvince ProvinceStockwell & Stockwell &Vayda (89) Vayda (89)§ Per 10,000 women at risk + Insurancegroups Bold Canadian Study.. Not age standardized ++ Occupational groups.... 65+ or 66+years old +++ Mean annual rate...... 25+ years old ++++ Medicaid Patients....... < 15 years oldTable 4 2Hemorrhoi-YEAR Hernia Repair dectomy Hysterectomy Cholecystectomy Cesarean1975 19 - 35" 25 - 9Q1t 14 -34Other Other OtherWennberg et al Wennberg et al Wennberg et al(101) (101) (101)1970~1975 78 -114§OtherWalker &: lick(99)1976 15.3 - 26Regions in NationFowkes (31)1970-1976 14.2 - 24.3 23.5 - 50.1 6.3 - 33.2 5.4 - 8.3*International International International InternationalVayda, et al (96) Vayda et al (96) Vayda et al (96) Vayda et al (96)1974-1976 47 -128......Regions inProvinceRoos (68)1977 33- 68 41.5 - 52.2 10.4 -14.8Province Regions in ProvincesMindell et a1 (55) Province Wadhera & NalrCageorge et al (98)(15)21-30ProvincesMindell et a1 (55)1973·1977 9.19 - 26.9 24.3 - 51.4 3.6 -13.8*Regions in Regions in Regions inProvince Province ProvinceVayda et a1 (94) Vayda (94) Vayda et al (94)1974,1977 34.5 -198......Regions inProvinceCohen (20)§....................Per 10,000women at riskNot age standardized65+or 66+ years old25+ years old< 15 years old++++++++++Insurance groupsOccupational groupsMean annual rateMedicaid PatientsBold Canadian StudyYEAR Hernia RepairTable 4Hemorrhoi-dectomy Hysterectomy Cholecystectomy Cesarean319781975-197814.5 -18.9Regions inProvinceHalliday &: LeRiche (37)4.7 - 9.6Region inProvinceHalliday &: LeRiche (37)2.6 - 3.2'"OtherShwartz et al(84)42.2 -72.8Regions inProvinceHalliday &: LeRiche (37)27.1 - 47.3InternationalSavage (81)1.51 - 2.94'"OtherShwartz et al(84)21.4 -31.7Regions inProvinceHalliday &: LeRiche (37)0.91 - 1.76'"OtherShwartz et al(84)37.2 - 66.1Regions inProvinceHalliday &: LeRiche (37)13.9 -17.6Regions in NationPlacek & Taffel(64)19801978·19801979·19806 - 8 43 - 46 19 - 24Regions in State Regions in State Region in StateGriffith et al (36) Griffith et al (36) Griffith et al (36)8.6 -13.3OtherWilliams (110)9.97 -17.08OtherEvans (30)1981 38-53...... 13-31.........Other OtherChassin et al (17) Chassin et al (17)34 - 52'" 17.1 - 20.0Other Regions in NationChassin et al (17) Placek (65)1979·1981 45.8 - 65.9OtherSchact (8213.7 - 24.5OtherSchact (82)81.4 -151.6OtherSchact (82)34.3 - 58.6OtherSchact &:Pemberton (82)1982 17.1- 20.2Regions inProvinceAnderson &:Lomas (4)1977 ·1982 16.1-21.31OtherHaynes De Rl.!glet al (38)§ Per 10,000women at risk + Insurance groups... Not age standardized ++ Ckcupationalgroups...... 65+ or 66+ years old +++ Mean annual rate......... 25+ years old ++++ Medicaid Patients............ < 15 years oldBold Canadian Study198,';1983~19851986Hemia RepairTable 4Hemorrhoi-dectomy Hysterectomy Cholecystectomy45.6 ~ 114.3OtherCoulter et al (25)Cesarean13.5 - 30.4OtherAcker et al (1)15.6 - 29.1OtherStafford (86)4§ Per 10,000 women at risk + Insurancegroups Bold Canadian Study.. Not age standardized ++ Occupational groups.... 65+ or 66+years old +++ Mean annual rate...... 25+ years old ++++ Medicaid Patients........ < 15 years oldYear AppendectomyTable 4Prostatectomy T & A Cataract CABS196.'; 14.6 - 61.8Regions in StateLewis (46)1965-1966 21.7 - 22 male 32.2 - 63.7 male 47.2 - 65.3 male18 - 22.3 female 32.1 - 64.1 female 69.1 - 82.5 femaleIn ternational International InternationalBunker (13) Bunker (13) Bunker (13)1966-1967 26.1 -118.8 +29.9 - 94.6 ++OtherLichtner &Pflanz(47)1967-1968 72 -183InternationalVayda (92)1968 14-26 53 -120Province ProvinceMindell et al (55) Mindell et al (55)1961 -1970 167 - 195............OtherBloor et al (11)1947-1971 8.1 - 26.5OtherYoshida &Yoshida (113)1969·1971 14-31 15-32 23-122Regions in State Regions in State Regions in StateGittelsohn Gittelsohn Gittelsohn& Wennberg (34) & Wennberg (34) & Wennberg (34)1968-1971 17.3 - 30.3 42.7 -97.9 4.36 - 9.56Provinces Provinces ProvincesVayda et al (97) Vayda et al (97) Vayda et al (97)1971-1972 "'125 - 282...... 31-131.....Other OtherRoos &: Roos (72) Roos &: Roos (73)1969·1973 102 - 126.6..........OtherWennberg et al(102)§ Per 10,000 women at risk + Insurance groups Bold Canadian Study... Not age standardized ++ Occupational groups...... 65+ or 66+ years old +++ Mean annual rate......... 25+ years old ++++ Medicaid Patients.......... < 15 years old6Table 4Year Appendectomy Prostatectomy T&:A Cataract CABS1973 11.9 - 25.6 18-40 24.4 - 39.9Regions in State Regions in State Regions In StateDehner &: Tyson Wennberg &: Dehner &: Tyson(25) Glttlesohn (106) (25)11- 22 23 -122Regions in State Regions in StateWennberg&: Wennberg&:Glttlesohn (106) Glttlesohn (106)1972·1974 80.8 -163.6Regions inProvinceRoos et al (74)1974 11.8 - 56.6 23.2 -191.1Regions in Regions in OtherProvince ProvinceStockwell & Stockwell &Vayda (89) Vayda (89)1974·1975 73 - 142% of thenational rateOtherSanderson (78)1975 10 -28" 13 - 42" 11 - 61"Other Other OtherWennberg et al Wennberg et al Wennberg et al(101) (101) (101)1973-1977 11.6 - 42.8+++ 4.8 - 36 +++ 21.0 - 89.2 +++Regions in Regions in Regions inProvince Province ProvinceVayda et al (94) Vayda et al (94) Vayda et al (94)1975-1978 4.18 -11.1 +++Regions in StateShwartz et al(84)1977 16-25 37- 68Provinces ProvincesMindell et al (55) Mindell et al (55)1978 16.3 -29.3 17.4 - 25.7 32 - 57.4 6.5 -14.4Regions in Regions in Regions in Regions InProvince Province Province ProvinceHalliday & Halliday & Halliday & Halliday &LaRiche (37) LaRiche (37) LaRiche (37) LaRiche (37)§ Per 10,000 women at risk + Insurance groups Bold Canadian Study.. Not age standardized ++ Occupational groups.... 65+ or 66+ years old +++ Mean annual rate...... 25+ years old ++++ Medicaid Patients....... < 15 years old7Table 4Year Appendectomy Prostatectomy T&:A Cataract CABS1977·1979 15 - 26 male22 - 43 femaleBernth-Peterson&: Bach (9)1979-1981 50 - 98.3 38.7 - 61.1 52.8 -108.5 29.0 - 65.5Other Other Other OtherSchact &: Schact &: Schact &: Schact &:Pemberton (82) Pemberton (82) Pemberton (82) Pemberton (82)1980 12 -17 20-24 24-34Regions in State Regions in State Regions in StateGriffith et al (36) Griffith et al (36) Griffith et al (36)1981 2 - 5.... 57 - 9~· 267 - 643"· 120 - 180.... 7 - 13··Other Other Other Other OtherChassin et al (17) Chassin et al (17) Chassin et al (17) Chassin et al (17) Chassin et al (17)1978·1984 2.63 - 5.61+++Regions inProvinceRoos &: Cageorge(70)1979·1985 30 -130" 4.28 -7.7876 -161"· Regions inOther ProvinceRoos et al (75) Anderson&:Lomas (6)§ Per 10,000women at risk + Insurance groups Bold Canadian Study• Not age standardized ++ Occupational groups.... 65+ or 66+ years old +++ Mean annual rate..... 25+ years old ++++ Medicaid Patients....... < 15 years oldTABLE 5TEN MOST FREQUENTLY STUDIEDPROCEDURES IN THE 51 "GOOD PAPERS"Cholecystectomy 21Tonsillectomy/ Adenoidectomy 17Appendectomy 15Cesarean 15Hysterectomy 15Prostatectomy 11Lens Extraction (Cataract Removal) 11Hernia Repair 10Hemorrhoidectomy 9Coronary' Artery Procedures 4TABLE 6Relationship of Tested Independent Variables on Procedure RatesCholecys- Tonsillectomy Itectomy Adenoidectomy Appendectomy Cesarean HysterectomyNumber of studies which testedindependent variables on procedure 21 17 15 15 15T' S" T S T S T S T SSUPPLY OF MEDICAL RESOURCESSupply of Hospital ResourcesBed SupplyNumber of corporate owned beds 5 2 1 4 5 1Number of hospital beds 5 1 1 6 1Number of specialty beds 1 1 1 1Number of surgical bedsOccupancyThroughput (number of admissions) 1 1 1 1Hospital TypeLevel of neonatal intensive carePeer grouping 1OwnershipSizeTeaching Status 3 1 1 1 3 1,. T = Number of studies in which effect of variable was testedS = Number of times in which variable was tested and found to be significantA number printed in outline style signifies that the test involved multiple regression analysisA number printed in normal style signifies that the test involved some other type of analysis such as t-test, or a correlational analysis1T ABLE 6 <Con tinued )Otolecys- Tonsillectomy Itectomy Adenoidectomy Appendectomy Cesarean HysterectomyT S T S T S T S T SHospital StructureAllocation of funding minustarget funding of Resource 1Allocation Working PartyHospital district codeNumber of full-time equivalents 1 1 1 1Number of physicians sharingnight call with respondent 1 1Number of registered pharmacists 1 1 1 1Payment differential betweenvaginal and cesarean delivery 1Ratio of number of registered nurses 1to hospital bed supply 1 1 1Subject to peerreview 1 1 1 1 1Hospital ServicesAvailability of pediatricianat delivery 1Centralized surgeryFetal scalp pH sampling performedon delivery floor 124 hour ultrasound imaging ondelivery floor 1Availability of 24 hour blood bankservice 1 1Mean number of hospital services 1 1 12TABLE 6 (contin ued )Cholecys- Tonsillectomy Itectomy Adenoidectomy Appendectomy Cesarean HysterectomyT S T S T S T S T SAnesthesia ServicesAvailability of epidural anesthesia 1 1Perceived number of anesthesiologists ] 1Presence of 24 hour in-hospitalanesthesia coverage 1 1Hospital ProcessAlternative treatment combinationsAnnual number of deliveries 2 1 1Average length of stay 1 2 1 1 1Day of admission 2 2Deliveries in second hospital 1Month of admission 2 1 ]Number of hospital days 1 1 1 ] 1Number of outpatients per capita 1 1 1 1Number of outpatients seenNumber of outpatient sessionsNumber of sight testsProportion of simple orcomplex procedures 1Rate of organ removal 1 1Time of deliveryday of the week 1time of the day 1 1Time required to begin immediatecesarean section 1 1Waiting listmedian waiting time 1number of people on list 13TABLE 6 (continued)Cholecys- Tonsillectomy Ilectomy Adenoidectomy Appendectomy Cesarean HysterectomyT S T S T S T S T STechnologic sophistication (percentage of deliveries)Fallopian tube surgey 1 1Intrau terine procedures 1 1(including fetal monitoring)Voluntary sterilization 1Supply of Physician ResourcesAll physicians 2 1 2All physicians performing surgery 4 1 5 1 4 1 3 1 4 1Board certified surgeons 2 1 1Consultants 2 1General practitioners 3 2 1 2 1 3 3 1Medical StaffSalaried physicians per capita 1 1 1 1Specialists 4 2 1 2 1 3 1 4 1Physician CharacteristicsAge 1 1 1 1 1Gti.zenship 1 1 1 1Employment Status 1 1 1 1Graduating year 2 1 1 2Location of internship 2 1 1 1 1Location of medical school 3 1 4 1 1 1 2 1 3 1Medical degree 1 1 1 1Medical school attended 1 1Number of hysterectomies performed 1 2 1Number of years since training4TABLE 6 (continued)Cholecys- Tonsillectomy Ilectomy Adenoidectomy Appendectomy Cesarean HysterectomyT S T S T S T S T SPerceived liability of litigationestimated 1977 malpractice premium 1recall of intrapartum death withintwo years 1recall of threat of malpractice 1suit within five yearsrecall of fetal death at greater 1than or equal to 35 weekswithin two yearsPlace of birth 1 1 1Post-graduate degree 1 1 1 1Residence of physician's primary patients 1Sex 1 1 1 1 1Specialty 2 1 3 1 2 1 1 1Status of surgical tra ining 2Type of hospital of practice 1 1 1 1Type of income (fee-for-service, 2 1 3 1 1 2 2 2salaried etc.)Type of practice (solo, group etc.) 3 1 3 1 1 2 2 2Type of registration 1 1 1 1Years of practice in same country 1 1 1Year of registration in Province 1 1 1 1Patient CharacteristicsAgeall 1 1 1 4 1 1elderly (65+) 1 1 1 15TABLE 6 (continued)Cholecys- Tonsillectomy Itectomy Adenoidectomy Appendectomy Cesarean HysterectomyT S T S T S T S T SEthnicity 1 1 1 1Native American 1Black 1 1 1Non-white 2 1 1Language 1 1Mother tongue other than English 1 1 1Comorbidity 2 2Education 2 2 1 1 1Income 1 1 1 1 1 2 1Method of Payment 2 1 2 1 1indigent services 1 1Kaiser Permanente 1 1Medi-Cal 1 1other HMOs 1 1other payors 1 1private insurance 1 1Blue Crossself-pay 1 1Population size 1Reduction in Medicare payment 1 1 1ResidenceRural/urban 1 1 1 1Hospital district of with 1 1 1 1 1 1respect to hospitaldistrict of surgeryProximity to referral centreSocioeconomic status 1 1Sex 1 1 16O1olecys­tectomyT STABLE 6 (continued)Tonsillectomy IAdenoidectomyT SAppendectomyT SCesareanT SHysterectomyT SDiagnosis/angiographyDilation and currettage in previous 2 yearsPlace of carePrivate physicianHospital clinicPre-hospital diagnosisSymptomsNumber of different physicians seenNumber of operations in last 6 monthsNumber of pre-operativehospitalizationsNumber of physician visitsfor menstrual disordersfor gynelogic problemsfor vague psychological problemsPercent reporting eye problemsFetal presentationabnormal cephalic presentationbreechmultiple factorsoblique or transverseother malpresentationFetal asphyxiagreen amniotic fluidpathological fetal heart raleby auscultation141111121111111111117pathological fetal heart rateby cardiotocographypathological stress or non-stress testOther fetal indicationsbad obstetric historybirthweightdiabetesdystociafailed inductionfetal distressgrowth retardationmacrosomiapast term pregnancypathological placental function testspreterm births (less than 37 weeks)prolapse of umbilical cordrhesus immunizationsuspected intrauterine infectionother diagnosesCholecys­tectomyT STABLE 6 (continued>Tonsillectomy IAdenoidectomyT SAppendectomyT SCesareanT S11112 12 12 112 211 11111111HysterectomyT SMaternalconcomitant sterilizationhigh maternal agematernal diseaseparitypreeclampsia/eclampsiasubfertilitypelvic tumour111311118Cholecys­tectomyT STABLE 6 (continued)Tonsillectomy IAdenoidectomyT SAppendectomyT SCesareanT SHysterectomyT SMechanicaldisproportionsprolonged labourPlacental, acuteplacental abruptionplacenta previaplacental separationUterinecervical dystociathreatening rupture and overt ruptureprevious cesarean sectionother previous uterine surgerypelvic tumourREGIONAL OIARAcrERISTICSMorbidityAMIURTIOther procedural ratesOverall surgical ratesPercent of adult population whoare disabledPercent of adult population whoare disabled and unable to workPercent of population who areunemployed1 1112222112311111211111111119TABLE 6 (continued)Cholecys- Tonsillectomy Itectomy Adenoidectomy Appendectomy Cesarean HysterectomyT S T S T S T S T SPercent of population with college 2 2 1educationPercent of population with medical 2 2 2coveragePresence of hospitalPresence of teaching centre 3 1 2 3 1 2Referral patterns 2 2Socioeconomic status 1 1 1 1 1 1Standards of selection 1Urban/rural mix 1 1 1 110TABLE 6 (con tinued)Lens Extraction Hernia Hemorrhoi-Prostatectomy Cataract Removal Repair dectomy CABSNumber of studies which testedindependent variables on procedure 11 11 10 9 4T S T S T S T S T SSUPPLY OF MEDICAL RESOURCESSupply of Hospital ResourcesBed SupplyNumber of corporate owned beds 3 2 3 2 4 2Number of hospital beds 1Number of specialty beds 1 1 1Number of surgical bedsOccupancyThroughput (number of admissions) 1 1 1 1Hospital TypeLevel of neonatal intensive carePeer groupingOwnershipSizeTeaching Status 1 1 1 1 1Hospital StructureAllocation of funding minus 1target funding of ResourceAllocation Working PartyHospi tal district codeNumber of full-time equivalents 1 1 1 111ProstatectomyT STA5LE 6 <Continued)Lens ExtractionCataract RemovalT SHerniaRepairT SHemorrhoi­dectomyT SCABST SNumber of physicians sharingnight call with respondentNumber of registered pharmacists 1 1 1Payment differential betweenvaginal and cesarean deliveryRatio of number of registered nurses 1 1 1to hospital bed supply 1Subject to peerreview 1 1 1 1Hospital ServicesAvailability of pediatricianat deliveryCentralized surgery 1 1Fetal scalp pH sampling performedon delivery floor24 hour ultrasound imaging ondelivery floorAvailability of 24 hour blood bankserviceMean number of hospital services 1 1 1Anesthesia ServicesAvailability of epidural anesthesiaPerceived number of anesthesiologistsPresence of 24 hour in-hospitalanesthesia coverage12TABLE 6 (continued)Hospital ProcessAlternative treatment combinationsAnnual number of deliveriesAverage length of stayDay of admissionDeliveries in second hospitalMonth of admissionNumber of hospital daysNumber of outpatients per capitaNumber of outpatients seenNumber of outpatient sessionsNumber of sight testsProportion of simple orcomplex proceduresRate of organ removalTIme of deliveryday of the weektime of the dayTIme required to begin immediatecesarean sectionWaiting listmedian waiting timenumber of people on listProstatectomyT S111Lens ExtractionCataract RemovalT S1HerniaRepairT111111S1Hemorrhoi­dectomyT S1111111CABST STechnologic sophistication (percentage of deliveries>Fallopian tube surgeyIntrauterine procedures(including fetal monitoring)Voluntary sterilization13Supply of Physician ResourcesAll physiciansAll physicians performing surgeryBoard certified surgeonsConsultantsGeneral practitionersMedical StaffSalaried physicians per capitaSpecialistsTABLE 6 (continued)Lens Extraction Hernia Hem orrhoi-Prostatectomy Cataract Removal Repair dectomy CABST S T S T S T S T S1 22 1 2 1 3 1 2 11 11 2 1 2 11 1 11 2 1 1Physician CharacteristicsAgeCitizenshipEmployment StatusGraduating yearLocation of internshipLocation of medical schoolMedical degreeMedical school attendedNumber of hysterectomies performedNumber of years since trainingPerceived liability of litigationestimated 1977 malpractice premiumrecall of intrapartum death withintwo yearsrecall of threat of malpracticesuit within five years11112211111121111 211111114TABLE 6 (continued)Lens Extraction Hernia Hemorrhoi-Prostatectomy Cataract Removal Repair dectomy CABST S T S T S T S T Srecall of fetal death at greaterthan or equal to 35 weekswithin two yearsPlace of birth 1 1Post-graduate degree 1 1 1Residence of physician's primary patientsSex 1 1 1 1Specialty 1 1 1 1 1 1Status of surgical training 2 2Type of hospital of practice 1 1Type of income (fee-for-service, 1 2 1 1 1salaried etc)Type of practice (solo, group etc.) 2 2 1 1 1Type of registration 1 1 1Years of practice in same country 1 1Year of registration in Province 1 1 1Patient CharacteristicsAgeall 1 1 1 1 1elderly (65+)Ethnicity 1 1 1 1Native AmericanBlackNon-whiteLanguageMother tongue other than English15TABLE 6 (continued)Lens Extraction Hernia Hemorrhoi-Prostatectomy Cataract Removal Repair dectomy CABST S T S T S T S T SComorbidityEducation 1 1 1 1 1Income 1 1 1 1 1 1 1Method of Payment 1 1indigent servicesKaiser PermanenteMedi-Calother HMOsother payorsprivate insuranceBlue Crossself-payPopulation sizeReduction in Medicare payment 1ResidenceRural/urban 1 1 1 1 1 1 1 1Hospital distric t of with 1 1respect to hospitaldistrict of surgeryProximity to referral centre 2 1Socioeconomic statusSex 1 1 1 1Diagnosis by tissue sample 1Dilation and currettage in previous 2 yearsPlace of carePrivate physicianHospital clinic16TABLE 6 €con tinued)Pre-hospital diagnosisSymptomsNumber of different physicians seenNumber of operations in last 6 monthsNumber of pre-operativehospitalizationsNumber of physician visitsfor menstrual disordersfor gynelogic problemsfor vague psychological problemsPercent reporting eye problemsFetal presentationabnormal cephalic presentationbreechmultiple factorsoblique or transverseother malpresentationFetal asphyxiagreen amniotic fluidpathological fetal heart rateby auscultationpathological fetal heart rateby cardiotocographypathological stress or non-stress testProstatectomyT 5Lens ExtractionCataract RemovalT 522HerniaRepairT 5Hemorrhoi­dectomyT 5CABST11517TABLE 6 (continued)Other fetal indicationsbad obstetric historybirlhweightdiabetesdystociafailed inductionfetal distressgrowth retardationmacrosomiapast term pregnancypathological placental function testspreteJIII births Oess than 37 weeks)prolapse of umbilical cordrhesus immunizationsuspected intrauterine infectionother diagnosesMaternalconcomitant sterilizationhigh maternal agematernal diqoac;pparitypreeclampsia/eclampsiasubfertilitypelvic tumourProstatectomyT SLens ExtractionCataract RemovalT SHerniaRepairT SHemorrhoi­dectomyT SCABST S18TABLE 6 (continued)Lens Extraction Hernia Hemorrhoi-Prostatectomy Cataract Removal Repair dectomy CABST S T S T S T S T SMechanicaldisproportionsprolonged labourPlacental, acuteplacental abruptionplacenta previaplacental separationUterinecervical dystociathreatening rupture and overt ruptureprevious cesarean sectionother previous uterine surgerypelvic tumourREGIONAL CHARACTERISTICSMorbidityAMI 1URTIOther procedural rates 1 1 1 1 1Overall surgical rates 1 1Percent of adult population who 1 1 1are disabledPercent of adult population who 1 1 1are disabled and unable to work19TABLE 6 (continued)Lens Extraction Hernia HemOIIboi-Prostatectomy Cataract Removal Repair dectomy CABST S T S T S T S T SPercent of population who are 1 1 1unemployedPercent of population with college 1 1 1educationPercent of population with medical 1 2 2 2coveragePresence of hospitalPresence of teaching centre 1 2 1Referral patterns 1 1Socioeconomic status 1 1 1Standards of selection 1 1UrbanIrural mix 1 120TABLE 7MAJOR SHORTCOMINGS OF 51 "GOOD" PAPERStSHORTCOMINGFailure to specify or describe methods of analysisFailure to describe whether patients counted more than onceFailure to test differences in variationFailure to include all variables of interest in the analysisDenominator was not relevant to study questionSmall study populationFailure to clearly describe unit of analysisFailure to clearly describe procedures being assessedFREQUENCY121210554441 Major =Foundin more than 3 papers.HEALTH POLICY RESEARCH UNIT429 - 2194 Health Sciences MallUniversity of British ColumbiaVancouver, B.C. CANADAV6T 1Z6DISCUSSION PAPER SERIESHPRU 88:1DHPRU 88:1RAccommodating Rapid Growth in Physician Supply: Lessons fromIsrael, WarnIngs for Canada. February 1988. (M.L. Barer, A.Gafni, J. Lomas)Barer, M.L., Gafni, A. and Lomas, J. (1989), "AccommodatingRapid Growth in Physician Supply: Lessons from Israel,Warnings for Canada", International Journal of Health Services19(1):95-115HPRU 88:2DHPRU 88:3DHPRU 88:2RHPRU 88:3RHPRO 88:4RThe Long Goodbye: The Great Transformation of the BritishColumbia Hospital System. March 1988. (R.G. Evans, M.L.Barer, C. Hertzman, G.M. Anderson, I.R. Pulcins, J. Lomas)Evans, R.G., Barer, M.L., Hertzman, C., Anderson, G.M.,Pulcins, I.R. and Lomas, J . (1989), "The Long Goodbye: TheGreat Transformation of the British Columbia Hospital System ll ,Health Services ~esesrch 24(4):435-459Reading the Menu With Better Glasses: Aging and Health PolicyResearch. April 1988. (R.G. Evans)Evans, R.G. (1989), IIReading the Menu With Better Glasses:Aging and Health Policy Research", in S.J. Lewis (ed.), Agingand Health: Linking Research and Public Polley, LewisPublishers Inc., Chelsea, 145-167Barer, M.L. (1988), llRegulating Physician Supply: TheEvolution of British Columbia's Bill 41", Journal of HealthPolItics, Policy and Law 13(1):1 -25HPRU 88: 5D Regional1zation of Coronary Artery Bypass Surgery: Effects onAccess. May 1988. (G.M. Anderson, J. Lomas) .HPau 88:Sa Anderson, G.M. and tomas, J. (1989), "Regionalhation ofCoronary Artery Bypass surgery: Effects on Access", MedicalCare 27(3):2a8-296HPaU 88:60 Diagnosing Senescence: The Medicalization of B.C.'s Elderly.July 1988. (M.L. Barer, I.R. Pulcins, R.G. Evans, C.Hertzman, J. Lomas, G.M. Anderson)HPRU 88:6R ~arer, M.L., Pulcins, t.R., Evans, a.G., Hertzman, C., Lomas ,J. and Anderson, G.M. (1989), IITrends in Use of MedicalServic:::eS by the Elderly in British Columbia", Canadian MedicalAssociation Journal 141:39 -45HPRU 88:70 The Development of Utilization Analysis: How, Why, and WhereIt's Going. August 1988 . (G.M. Anderson, J. Lomas)HPRU 88:8D Squaring the Circle: Reconciling Fee-for-Service with GlobalExpenditure Control. September 1988. (R.G. Evans)o c Discussion Paper R ... ReprintHPRU 89:2DHPRU 89 : 2aHEALTH POLICY RESEARCH UNITDISCUSSION PAPER SERIESBPRU 88:9D Practice Pat.terns of Physicians with Two Year Residency VersusOne Year Internship Training: Do Both Roads Lead to Rome?September 1988. (M.T. Schechter, S.B. Sheps, P. Grantham, N.Finlayson, R. Sizto)BfRU 88:10R Anderson, G.M. and tomas, J. (1988), "Monitoring the Diffusionof a '1'eohnology: Coronary Artery Bypass SU'I:'g'ery in OntEltio",AlJ'Je.ri¢Sl'1 Jou.rt141 "<rt PublJ.o Health 78(3) :251-·254HfRO 88:1ltl J!:V'imS, a.G. (1988),''''We'll Take Care of it For You": HealthCare in the Canadian Community", Daedalus 117(4):155-189BPRO 88:121\ Barer, M.L., Evans, a.G. and Labelle, a.J. (1988), "FeeControls as Cost Control: Tales From the Frozen North ll , TheMilbsnk QuarterlY 66(1):1-64HPRU 88:13D Tension, Compression and Shear: Directions, Stresses andOutcomes of Health Care Cost Control. December 1988. (R.G.Evans)BPRU 88:1S:a Jl:vans, a.G . (1990), IItension, Compression, and Shear:birections, Stresses and Outcomes of Health Care Costcontrol", Journal ()f Heal th Poll tics, Policy snd Law15(1) : 101 -128HfRO 88:14R J!:V'IU1S, a .G., Robinson, G.C. and Baret', M.L. (1988), "WhereHave All the Children Gone? Accounting for the PaediatricHoSpital Implosion", in R.S . Tonkin and J.R. wright (ede.),Redesigning Relationships in Child Health Care, B.C.Children's Hospital; 63-76HPRU 89:1D Physician Utilization Before and After Entering a Long TermCare Program: An Application of Markov Modelling. January1989. (H. Krueger, A.Y. Ellencweig, D. Uyeno, B. McCashin, N.Pagliccia)Flat on Your Back or Back to Your Flat? Sources of IncreasedHospital Services Utilization Among the Elderly in BritishColumbia. January 1989. (C. Hertzman, l.R. Pulcins, M.L.Barer, R.G. Evans, G.M. Anderson, J. Lomas)Hertzm~, C., Pulcine, I .R., Barer, M.L ., Evans, R.G.,Arldersqri, G.M . and Lomas , J . (1990) "Flat on YO\1r Back or Backto your Flat? sotirces Qf Irtcreased Hospital S~rvicesUtiliz~tion Among ~he £ldeily in British Columbia", Soc!~lScience and Medicine 30(7):819-828 "aPRU S9:3R auhler, t., Gl~ok, N. and Sheps, S.B. (1988), "P~enatal Care :A comparative E~aluation of Nurse-Midwives arid Fami11Physicians", Canadian Medical Association Journal 139;3~7-403HPRU 89:4DBPRU 89:4ilRecent Trends in Cesarean Section Rates: Ontario Data 1983 to1987. February 1989. (G.M. Anderson, J. Lomas)Anderson, q.~. $nd Lomas, J . (19a9), ~Recent T~ends inCesarean Seotion aates in Ontario", Canadian M$dicalAssociation Joutnal 141 :1049-1053HPRU 89:5D The Canadian Health Care System: A King's Fund Interrogatory.March 1989. (R.G. Evans)D = Discussion Paper R = Reprint'aPRU 9011RHEALTH POLICY RESEARCH UNITDISCUSSION PAPER SERIESHPRU 89:6D BenefLts, RIsks and Costs of Prescrlptlon Drugs 112 Ontarlo: ASclentlflc Bssls to Evaluate Pollcy Optlons. April 1989.(W.O. Spitzer, G.M. Anderson, U. Bergman, J.L. Blackburn, E.Wang, M.C. Weinstein) .HPRU 89:6R Anderson, a.M., Spitzer, W.O., Weinstein~ M.C., Wang( B.,Blackburn, J.L. and Ber9lllan, U. (1990), 'Benefits, lusks, and~ . Costo of Preocription Drugs: A'Scientific Basis for Bvaluatingpolicy Options", CHnicsl Phsrmscology snd Therspeutics48 (2) : 111-119HPRU 89:7D The Dog 112 the Nlght Tlme: Medlcal Practlce Varlatlons andHealth Pollcy. June 1989. (R.G. Evans)HPRU 89:8D Llfe and Death, Money and Power: The Polltlcs of Health CareFlnance. June 1989. (R.G. Evans)HPRU 89:9D From Medlbank to Medlcare: Trends 112 Australlan Medlcal CareCosts and Use, 1976-1986. August 1989. (M. Barer, M. Nicoll,M. Diesendorf, R. Harvey)HPRU 89:98. Barer, M.L., Nicoll, M., Oiesendorf( M. and Harvey, R. (1990),"prom Medibank to Medicare: Trends 1n Australian Medical CareCosts and Use Prom 1976 to 1986~~, CommunIty Heslth StudJ.esXIV( 1) : 8-18HPRU 89:100 Cholesterol Screenlng: Evaluatlng Alternatlve Strategles.August 1989. (G. Anderson, S. Brinkworth, T. Ng)HflU 89:11R EVans, R.a., Lomas, J., Barer, M.L.; Labelle, R.J., Fooks, C. ,Stoddart, G.L., Anderson, a .M., reeny, D., Gafni, A.,Torrance, G.W. and Tholl, W.G. (1989), "Controlling HealthExpenditures - The Canadian Reality", New England Journs.l ofMeoloine 320(9):571-577HPRU 89: 12D The Effect of Admlsslon to Long Term Care Program onUtl1lzatlon of Health Servlces by the Elderly ln BrltlshColumbla. November 1989. ·(A.Y. Ellencweig, A.J. Stark, N.pagliccia, B. McCashin, A. Tourigny)HPRU 89:130 Utl1lzatlon Patterns of Cllents Admltted or Assessed but notAdmltted to a Long Term Care Program - Characterlstlcs andDlfferences. November 1989. (A.Y. Ellencweig, N. Pagliccia,B. McCashin, A. Tourigny, A.J. Stark)HPRU 89:l4D Acute Care Hospltal Utlllzatlon Under Canadlan Natlon~l HealthInsurance: The Brltlsh Columbla Experlence from 1969 to 1988.December 1989. (G.M. Anderson, I.R. Pulcins, M.L. Barer, R.G.Evans, C. Hertzman)Anderson, G.M., NeWhousei J.P. and Roos, L.L. (1989),"tfospital care for JJ:ll;1$t 'i patienh with Oi.Sea$es of theCiroulatory System. A Comparison of Hospital Use in theUnitEid st.ates and Canada", New Englmtd Journal 6f Medlc1tie321:1443-1448HPRU 90:2D Poland: Health and Envlronment ln the Context of SocioeconomlcDecllne. January 1990. (C. Hertzman)D = Discussion Paper R = ReprintHPRU 90:3DHPRU 90:4DHPRU 90:5DHPRU 90:6DHl'lU 90:71Hl'lU 90:81HEALTH POLICY RESEARCH UNITDISCUSSION PAPER SERIESThe Appropriate Use of Intrapartum Electronic Fetal Heart RateMonitoring. January 1990. (G.M. Anderson, D.J. Allison)A Comparison of Cost Sharing Versus Free Care in Children:Effects on the Demand for OffLee-based Medical Care. January1990. (G.M. Anderson)Does Familr Practice CertLfication Affect Practice Style? An&Jalysi, of Office~bBsed Carat February 1990. (G.M.Anderson)An Assessment of the Value of Routine Prenatal UltrasoundScreening. February 1990. (G.M. Anderson, D. Allison)Nemetz, P.N., Ballard, D.J., Beard, C.M., Ludwig, J.,Tangalos, I.G., Kokmen, E' L Weigel, R.M., Belau, P.G., Bourne,W.M. and Kurland, L.T. (19ts9) "An Anatomy of the Autopsy,Olmsted County, 1935 through 198511, Mityo ClJ.nJ.c ProceedJ.ngs64:1055...1064Nemetz, P.N., Beard, C.M., Ballard, D.J., Ludwig, J.,Tangalos, I.G., Kokmen, B., weigel, R.M., Belau, P.G., Bourne,W.M. and Kurland, L.T. (1989l "Resurrecting the Autopsy:Benefits and Recommendations', Mayo Cllnlc Proceedings64:1065-1076HPRUHPRUHPRU 90:9D Technology Diffusion: The Troll Under the Bridge. A PilotStudy of Low and High Technology in British Columbia. March1990. (A. Kazanjian, K. Friesen)HP1U 90:101 Sapphires in the Mud? The Export Potential of American HealthCare Financing. Enthoven, A.C. (1989), "What Can EuropeansLearn from Americans?", Evans, R.G., Barer, M.L. (1989),Comment. Health Care Financing Review, Annual SupplemGnc 1989HPRU 90:11D Healthy Community Indicators: The Perils of the Search and thePaucity of the Find. March 1990. (M. Hayes, S. Willms)HPRU 90:12D Use of HMRI Data in Nineteen British Columbia Hospitals andFuture Directions for Case Mix Groups. April 1990. (K.M.Antioch)HPRU 90:13D Producing Health, Consuming Health Care. April 1990. (R.G.Evans, G.L. Stoddart)90:14D Automated Blood Sample-Handling in the Clinical Laboratory.June 1990. (W. Godo1phin, K. Bodtker, D. Uyeno, L.-a,' Goh)90:15D Hospital-Based Utilization Management: A Cross Canada Survey.June 1990. (G.M. Anderson, S. Sheps, K. Cardiff)'HPRU 90:16D Hospital-Based Utilization Management: A Literature Review.June 1990. (S. Sheps, G.M. Anderson, K. Cardiff)HPRU 90:17D Reflections on the Financing of Hospital Capital: A CanadianPerspective. June 1990. (M.L. Barer, R.G. Evans)HPRU 90:18D The Twenty Year Experiment: Accounting for, Explaining, andEvaluating Health Care Cost Containment in Canada and theUnited States. September 1990. (R.G. Evans, M.L. Barer, C.Hertzman)D = Discussion Paper R = ReprintHPRU 90:21DHEALTH POLICY RESEARCH UNITDISCUSSION PAPBa SERIBSHPRU 90:19D AccessIble, Acceptable and Affordable: FinancIng Health Carein Cnnada. Septomber 1990. (R.G. Evans)HPRU 90:20D Hungary Report. October 1990. (C. Hertzman)ThtJ Great Transforllmt:ion of the British ColumbIa Hc>spitalSector: Polley Design or PolItIcal AccIdent for the Elderly?October 1990. (J. Lomas, C. Hertzman, M.L. Barer, I.R.Pulcins, R.G. Ivana, G.M. Anderson)HPRU 20:22D Recent Trends In Acute Care Hospital UtI11zatlon for D1seasesand D1sorders of the C1rculatory System: Ontar10 Data from1979 to 1989. October 1990. (G.M. Anderson, I.R. Pulcins)HPRU 20:23D Env1ronment and HtJalth 1n Czechoslovak1a. December 1990. (C.Hertzman)D - Discussion Paper R ... Reprint


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