Background This study was conducted to assess the overall mortality rate and the amount of interhospital variation in risk-adjusted mortality rates after coronary artery bypass graft (CABG) surgery in Ontario, Canada. CABG outcomes data are not publicly disseminated in Ontario.
Methods and Results Clinical risk factors and surgical outcomes were collected on 15 608 patients undergoing isolated CABG surgery between April 1, 1991, and March 31, 1994, at the nine hospitals performing adult cardiac surgery in Ontario. The data were analyzed on the basis of a fiscal year. The overall mortality rate was 3.01%, and the risk-adjusted mortality rate declined from 3.17% in 1991 to 2.93% in 1993. In 1991, one of the nine hospitals had a risk-adjusted mortality rate significantly lower than the provincial average. Otherwise, the hospitals all had risk-adjusted mortality rates within the expected range during the time period of the study. All hospitals performed >300 CABG procedures in 1992 and 1993, and only 2 of 42 cardiac surgeons performed <50 CABG procedures in 1993.
Conclusions The in-hospital mortality rate after CABG surgery in Ontario is low, and the amount of interhospital variation in risk-adjusted mortality rates is no greater than that expected by chance alone. These outcomes are probably attributable to regionalization of CABG surgery and a very low prevalence of low-volume cardiac surgeons in Ontario.
The assessment of variations in patient outcomes is of great interest to health services researchers and policy makers in both Canada and the United States. Recently, a number of investigators in the United States have documented wide intersurgeon, interhospital, and interregional variations in patient mortality rates after coronary artery bypass graft (CABG) surgery that exist in spite of statistical adjustment for differences in patient case mix.1 2 3 4 5 Public dissemination of hospital- and surgeon-specific mortality data is now occurring in the United States with the release of annual CABG surgical report cards to the public in both New York and Pennsylvania.3 4 5 6 These surgical report cards have been advocated by some as a method of improving healthcare quality and enhancing consumer choice, although critics have suggested that they may cause some cardiac surgeons to avoid operating on high-risk patients.4 5 6 7 Although wide variations in CABG mortality rates have been documented in the US setting, it was not known whether similar variations existed within Canada's universal healthcare system, in which a comparable public reporting requirement does not exist.
An alternative strategy for improving CABG outcomes is the requirement of minimum surgical volumes annually at both the hospital and surgeon level. Numerous observational studies have shown a relationship between lower hospital and surgeon volumes and higher mortality rates after CABG surgery.8 9 10 11 As a result, policy makers have recommended that the delivery of cardiac surgery be regionalized and that minimum volume standards be set.10 11 12 13 An American College of Cardiology/American Heart Association task force has issued guidelines recommending that all cardiac surgery hospitals perform a yearly minimum of 200 to 300 open heart operations a year and that all cardiac surgeons perform a yearly minimum of 100 to 150 open heart operations, the majority of which should be CABG operations.12
In Ontario, Canada's largest province, the delivery of cardiac surgery has been regionalized since its introduction in the early 1970s, with the government restricting cardiac surgery to several high-volume sites.10 13 Limits on the number of cardiac surgeons trained and the existence of queues for surgery have also helped to ensure that most cardiac surgeons in Ontario perform a high volume of procedures.14 15 The current study was designed to determine whether these policies have an impact on the overall mortality rate and to ascertain the amount of interhospital variation in risk-adjusted mortality rates after CABG surgery in Ontario.
The Provincial Adult Cardiac Care Network (PACCN) of Ontario database was used as a source of clinical information for the current study.15 16 The PACCN database was established in 1991 by the Ontario Ministry of Health in cooperation with hospitals and cardiovascular specialists to provide data on cardiac surgery waiting lists and regional practice patterns in the province. It is a new clinical database that contains prospectively specified and concurrently gathered clinical information on all adult patients undergoing cardiac surgery in Ontario. For the present study, data were collected on all patients having isolated CABG surgery (without concomitant valve surgery) at the nine adult cardiac surgery institutions in Ontario from April 1, 1991, to March 31, 1994. Patient data were collected from the referring cardiologists, and cardiac surgeons were not directly involved. The data were analyzed on the basis of fiscal year: fiscal 1991 (April 1, 1991, to March 31, 1992), fiscal 1992 (April 1, 1992, to March 31, 1993), and fiscal 1993 (April 1, 1993, to March 31, 1994), hereinafter referred to as 1991, 1992, and 1993, respectively.
Clinical information from the PACCN database was linked to outcomes (ie, in-hospital mortality) and comorbidity information contained in the Canadian Institute for Health Information database, an administrative database based on hospital discharge abstracts, with the use of unique patient identifiers. Analysis of data from a large Canadian study has shown that >95% of deaths occur in the hospital within 30 days of CABG surgery.17 Complete information was available on 15 608 patients undergoing isolated CABG surgery, representing 98% of the patients identified in the PACCN database as having had isolated CABG surgery during the 3-year study period. The quality of the data in the PACCN database is maintained through a series of built-in logic and range checks, by cross-validation checks with clinical and administrative databases collected separately at each hospital, and by random chart audits by the local data coordinators at each site. A recent external audit of 20 fields from 539 patient records across the nine centers revealed an overall agreement rate of 97.5% between the data in the PACCN database and the information recorded in the patients' charts. The presence of a dedicated data coordinator at each hospital has resulted in very little missing data since the inception of the database.
After we reviewed the most commonly identified surgical risk factors found in other studies,1 3 16 we chose nine patient characteristics to study as potential risk factors for in-hospital mortality after CABG surgery. The nine characteristics studied were patient age, sex, left ventricular function, urgency of surgery, history of previous CABG, Canadian Cardiovascular Society (CCS) angina class, recent (within 7 days) myocardial infarction, coronary anatomy, and Charlson comorbidity index score.18 19 Age was initially subdivided into 5-year intervals and then grouped into three categories (<65, 65 to 74, and ≥75 years) based on similar mortality rates. Left ventricular function was categorized into four groups based on the ejection fraction (grade 1: >50%; grade 2: 35% to 50%; grade 3: 20% to 34%; grade 4: <20%) as assessed by echocardiography or angiography. Urgency of surgery was divided into three groups (emergency, urgent, and elective) based on a modification of a CABG urgency rating scale formulated by a consensus panel in Ontario.20 Emergency surgery was defined as surgery required within 24 hours, urgent surgery was that which was required within the same hospital admission, and any other surgery constituted elective surgery. CCS angina class was obtained at the time of referral for surgery and then updated before surgery if it had changed in the interim.18 Coronary anatomy was divided into five categories: left main disease (≥50% stenosis), two- or three-vessel disease with proximal left anterior descending artery involvement, three-vessel disease without proximal left anterior descending artery involvement, one-vessel disease with proximal left anterior descending artery involvement, and one- or two-vessel disease without proximal left anterior descending artery involvement.20 A Charlson comorbidity index score19 was constructed from the Canadian Institute for Health Information database to account for noncardiac comorbidity by use of methods previously described by Deyo et al21 and O'Connor et al.1
The method of risk-adjustment used in the current study was similar to that used in a recent report by Hannan et al3 from the New York State Cardiac Surgery Reporting System.3 The Stata statistical package was used for statistical analysis.22 To adjust for differences in patient case mix among the nine hospitals, a logistic regression model was developed to predict in-hospital mortality after isolated CABG surgery for the 15 608 patients in the combined Ontario database. Multivariate backward stepwise regression was used to determine the significant risk factors for in-hospital mortality, with variables retained in the model if they were significant at the P<.05 level. We evaluated the discrimination of the model by calculating the area under the receiver operating characteristic curve, and we used the Hosmer-Lemeshow statistic to evaluate the calibration of the model.23 24 We cross-validated the logistic regression model by estimating the model on half of the database and testing it on the other half before using the whole database to estimate the model when the model coefficients proved to be stable.
Risk-Adjusted Mortality Rates
To determine the overall change in risk-adjusted mortality rates during the 1991 to 1993 study period, the 3-year logistic regression model was used to determine an expected probability of mortality for each patient in the database, and then the expected probabilities were summed to determine the overall expected mortality rate for each fiscal year. The actual mortality rate for each year was divided by the expected mortality rate for that year, and then the ratio was multiplied by the average mortality rate during the 3 years to determine a risk-adjusted mortality rate for each fiscal year.3 The annual risk-adjusted mortality rate for each hospital was determined in a similar manner. The risk-adjusted mortality rate for a hospital can be interpreted as the mortality rate a hospital would have if the patient case mix at that hospital were similar to the average case mix in the province. Ninety-five percent CIs were constructed around each risk-adjusted mortality rate to determine whether the risk-adjusted mortality rate for a particular hospital was statistically significantly lower (ie, a low outlier) or higher (ie, a high outlier) than the provincial average for that year.25 The year-to-year correlation of risk-adjusted mortality rates for each hospital was determined.
Annual volumes of isolated CABG surgery were also determined for each hospital in each fiscal year. Complete data on annual surgeon volume were available only for 1993. The number of surgeons in four volume categories was determined (≤50, 51 to 100, 101 to 150, >150 CABG procedures per year).
Table 1⇓ shows the frequency distribution of patient characteristics over the 1991 to 1993 study period along with the associated in-hospital mortality rate for patients with that characteristic. The risk factors with the highest mortality rates were recent (within 7 days) myocardial infarction, grade 4 left ventricular function, emergency surgery, previous CABG surgery, and patient age ≥75 years. The 3-year logistic regression model for in-hospital mortality is shown in Table 2⇓. All nine variables were statistically significant independent predictors of mortality in this model. The model predicted mortality well, as demonstrated by its area under the receiver operating characteristic curve of 0.78 and its Hosmer-Lemeshow χ2(8 df) statistic of 10.34, P=.24.23 24
The overall mortality rate for isolated CABG surgery in Ontario was 3.01% over the 3-year study period, as shown in Table 3⇓. The annual volume of surgery performed in Ontario increased from 4782 cases in 1991 to 5517 cases in 1993. The unadjusted mortality rate declined from 3.01% in 1991 to 2.88% in 1992 before increasing to 3.14% in 1993. However, patient severity increased even more during this time period, with the expected mortality rate increasing from 2.86% in 1991 to 3.23% in 1993. Thus, the overall risk-adjusted mortality rate decreased from 3.17% in 1991 to 2.93% in 1993.
Table 4⇓ shows the actual, expected, and risk-adjusted mortality rates for each of the nine hospitals performing CABG surgery in Ontario. The range of variation in risk-adjusted mortality rates was relatively narrow across all 3 years, ranging from 1.02% to 4.92% in 1991, 1.46% to 4.18% in 1992, and 1.55% to 4.67% in 1993. In 1991, one hospital had a risk-adjusted mortality rate significantly lower than the provincial average, while no hospital had a risk-adjusted mortality rate significantly greater. In 1992 and 1993, all hospitals had risk-adjusted mortality rates within the expected range. However, there was a trend toward some hospitals having consistently better risk-adjusted outcomes than other hospitals during the 3-year study period. Risk-adjusted mortality rates for hospitals were correlated from year to year, with correlations of 0.79 and 0.59 in 1991 through 1992 and 1992 through 1993, respectively.
Annual hospital volumes of CABG surgery are not included in Table 4⇑ to maintain the confidentiality of hospital identities. In 1992 and 1993, all hospitals in Ontario performed >300 CABG operations per year (range, 317 to 1353), whereas in 1991, eight of the nine hospitals performed >300 CABG operations per year. Table 5⇓ shows the number of surgeons performing CABG surgery in each of four different surgeon volume groups in 1993. A low-volume surgeon has been defined in a recent study from New York as one who performs ≤50 CABG operations per year.9 Ontario had relatively few low-volume surgeons in 1993, with only 2 of 42 surgeons (4.8%) performing <50 CABG operations a year. Complete data were not available on surgeon volume in 1991 and 1992, although the available data show a similar distribution of surgeon volume in those years (available on request from the authors).
This study has examined the overall mortality rate and the amount of interhospital variation in risk-adjusted mortality rates after isolated CABG surgery in Ontario, Canada, between 1991 and 1993. The overall mortality rate was 3.01%, and the risk-adjusted mortality rate declined from 3.17% in 1991 to 2.93% in 1993. No hospitals were found to be high-outlier hospitals (ie, risk-adjusted mortality rates significantly greater than the provincial average), although some hospitals did perform consistently better than other hospitals during the 3-year study period. All hospitals in Ontario performed >300 CABG procedures in 1992 and 1993, and only 2 of 42 cardiac surgeons performed <50 CABG procedures in 1993. Our study provides empirical evidence that is compatible with the hypothesis that regionalization of CABG surgery and a very low prevalence of low-volume cardiac surgeons have a positive impact on surgical outcomes, avoiding the phenomenon of low-volume hospitals and surgeons with higher-than-expected mortality rates that was previously noted in the United States.3 4 5 6 8 9 10 11
The overall mortality rate of 3.01% for isolated CABG surgery in Ontario between 1991 and 1993 is comparable to the best outcomes reported in other jurisdictions. Officials from New York have reported an average mortality rate of 3.11% and a 41% decrease in the risk-adjusted mortality rate from 4.17% to 2.45% between 1989 and 1992, which they have attributed to the public dissemination of outcomes data and resulting quality-improvement activities.3 Similarly, the average mortality rate in Pennsylvania declined from 3.89% in 1990 to 2.85% in 1993.5 The results of the present study show that risk-adjusted mortality rates are also declining in jurisdictions in which outcomes data are not being disseminated publicly and demonstrate that alternative approaches for achieving a low surgical mortality rate do exist.
The amount of interhospital variation in risk-adjusted mortality rates found in the present study is less than that which has been reported elsewhere.1 3 4 5 The lack of any high-outlier hospitals during the 3-year study period may be a reflection of the high surgical volumes of hospitals and cardiac surgeons in Ontario. In the Canadian single-payer healthcare system, the government determines the number of cardiac surgery centers and, through funding policies, the annual surgical volume at each institution, creating regional “centers of excellence.”10 13 All nine adult cardiac surgery centers in Ontario perform >300 CABG procedures per year, and eight of the nine hospitals are tertiary-care teaching facilities. Previous studies of CABG surgery in the United States have shown that mortality rates are often significantly higher than expected at low-volume hospitals when comparisons have been made across regions with a mixture of low- and high-volume hospitals.8 10 11
Data from the New York and Pennsylvania CABG registries have shown that in contrast to Ontario, there are many low-volume cardiac surgeons operating in the United States, and these surgeons consistently have the highest risk-adjusted mortality rates.4 5 9 The very low prevalence of low-volume surgeons in Ontario arises from government policies that restrict the number of cardiac surgery centers, and university training policies that limit the number of cardiac surgeons trained in the province.14 High surgeon volumes follow as a matter of course when the number of surgeons trained is restricted and impact analyses are conducted to ensure that there is truly a need for each additional surgeon hired at each hospital. One of the two low-volume surgeons in Ontario in 1993 was winding down his surgical career, while the other low-volume surgeon left Ontario midway through the year and would otherwise not be in that category.
Outcomes data (ie, risk-adjusted mortality and length of stay) from the Ontario cardiac surgery registry were first made available confidentially to hospitals in Ontario in the fall of 1993.16 Clinicians at these hospitals have found the data to be very helpful in assessing and improving their own relative performance. In addition to providing ongoing feedback on surgical outcomes, the Ontario CABG registry is also used to monitor surgical waiting lists so that patients are triaged for surgery in an efficient manner.15 20 Analyses of these data have shown that deaths of people who are on the waiting list are rare and that there was a median wait of 17 days for CABG surgery in Ontario during the period 1991 to 1993.15
We conclude that the mortality rate after CABG surgery in Ontario is low, and the amount of interhospital variation in risk-adjusted mortality rates is no greater than that expected by chance alone. Although a causal link is not proven by this analysis, the most plausible explanation for these findings is the policy of strict regionalization of CABG surgery in Ontario, with all centers and cardiac surgeons managing reasonably high surgical caseloads. Maintenance of excellence is reinforced by confidential mortality reports to each surgical center annually. We suggest that this Canadian approach to quality assurance in cardiac surgery represents a potential alternative to more controversial methods based on the public dissemination of surgical report cards.6 26 Policy makers in other jurisdictions may wish to consider a similar approach for minimizing interhospital variations in risk-adjusted CABG mortality rates.
Steering Committee of the Provincial Adult Cardiac Care Network of Ontario
Alnoor Abdulla, MD, Sudbury Memorial Hospital, Ont (1991-1994)
Glenn Bartlett, MD, London, Ont (1991-1993)
Donald S. Beanlands, MD, University of Ottawa Heart Institute, Ont
Lorna Bickerton, BScN, University of Ottawa Heart Institute, Ont
Robert Chisholm, MD, St Michael's Hospital, Toronto, Ont
Martin Goldbach, MD, Victoria Hospital, London, Ont
Vicki Kaminski, BScN, Sudbury Memorial Hospital, Ont
Jeffrey Lozon, MHA, St Michael's Hospital, Toronto, Ont
Neil McKenzie, MB BCh, University Hospital, London, Ont
Barry J. Monaghan, BComm DHA, West Park Hospital, Toronto, Ont
Christopher D. Morgan, MD, Sunnybrook Health Science Centre, Toronto, Ont
John Pym, MB BCh, Kingston General Hospital, Ont
Hugh Scully, MD, Toronto Hospital, Ont
B. William Shragge, MD, Hamilton General Hospital, Ont
James Swan, MD, Scarborough Centenary Health Centre, Ont
Dr Tu is supported by a Health Research Personnel Development Program Fellowship (No. 04544) from the Ontario Ministry of Health. Dr Naylor is supported by a Career Scientist Award (No. 02377) from the Ontario Ministry of Health. This project was supported by grant No. R03 HS08464 from the Agency for Health Care Policy and Research. The results and conclusions are those of the authors, and no official endorsement by the sponsoring agencies is intended or should be inferred. The authors would like to thank the many cardiologists and cardiac surgeons of the Provincial Adult Cardiac Care Network of Ontario, the institutional and regional coordinators who are responsible for data collection and patient follow-up, and the staff of the PACCN coordinating office at Victoria Hospital, London, Ont, particularly Steve Jefferson.
Reprint requests to Dr C. David Naylor, Institute for Clinical Evaluative Sciences in Ontario, G-106, 2075 Bayview Ave, North York, Ontario, Canada M4N 3M5. E-mail firstname.lastname@example.org.
*A complete list of Steering Committee members is provided in the “Appendix.”
- Received February 28, 1996.
- Revision received May 8, 1996.
- Accepted June 7, 1996.
- Copyright © 1996 by American Heart Association
O'Connor GT, Plume SK, Olmstead EM, Coffin LH, Morton JR, Maloney CT, Nowicki ER, Tryzelaar JF, Hernandez F, Adrian L, Casey KJ, Soule DN, Marrin CAS, Nugent WC, Charlesworth DC, Clough R, Katz S, Leavitt BJ, Wennberg JE, for the Northern New England Cardiovascular Disease Study Group. A regional prospective study of in-hospital mortality associated with coronary artery bypass grafting. JAMA. 1991;266:803-809.
New York State Department of Health. Cardiac Surgery in New York State 1990-1992. Albany, NY: New York State Department of Health; December 1993.
Pennsylvania Health Care Cost Containment Council. A Consumer Guide to Coronary Artery Bypass Graft Surgery: Volume IV 1993. Harrisburg, Pa: Pennsylvania Health Care Cost Containment Council; June 1995.
Omoigui NA, Miller DP, Brown KJ, Annan K, Cosgrove D, Lytle B, Loop F, Topol EJ. Out-migration for coronary bypass surgery in an era of public dissemination of clinical outcomes. Circulation. 1996;93:27-33.
Hannan EL, Siu AL, Kumar D, Kilburn H Jr, Chassin MR. The decline in coronary artery bypass graft surgery mortality in New York State: the role of surgeon volume. JAMA. 1995;273:200-213.
Luft HS, Garnick DW, Mark DH, McPhee SJ. Hospital Volume, Physician Volume, and Patient Outcomes: Assessing the Evidence. Ann Arbor, Mich: Health Administration Press; 1990.
ACC/AHA Task Force Report. Guidelines and indications for coronary artery bypass graft surgery: a report of the American College of Cardiology/American Heart Association Task Force on assessment of diagnostic and therapeutic cardiovascular procedures (Subcommittee on Coronary Artery Bypass Graft Surgery). J Am Coll Cardiol. 1991;17:543-589.
Tu JV, Jaglal SB, Naylor CD, and the Steering Committee of the Provincial Adult Cardiac Care Network of Ontario. Multicenter validation of a risk index for mortality, intensive care unit stay, and overall hospital length of stay after cardiac surgery. Circulation. 1995;91:677-684.
Stata Release 4.0. College Station, Tex: Stata; 1995.
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons Inc; 1989.
Breslow NE, Day NE. Statistical Methods in Cancer Research: Volume II—The Design and Analysis of Cohort Studies. Lyon, France: International Agency for Research on Cancer; 1987.