Quality of Care of International and Canadian Medical Graduates in Acute Myocardial Infarction
Total Page:16
File Type:pdf, Size:1020Kb
ORIGINAL INVESTIGATION Quality of Care of International and Canadian Medical Graduates in Acute Myocardial Infarction Dennis T. Ko, MD; Peter C. Austin, PhD; Benjamin T. B. Chan, MD, MPA; Jack V. Tu, MD, PhD Background: International medical graduates (IMGs) Results: Of the 127275 admitted AMI patients, 28061 make up a substantial proportion of the physician work- (22.0%) were treated by IMGs and 99214 (78.0%) by Ca- force and play an important role in the care of patients nadian medical graduates. The risk-adjusted mortality rates with acute myocardial infarction (AMI). There are con- of IMG- and Canadian medical graduate–treated patients cerns that IMGs may provide inferior medical care com- were not significantly different at 30 days (13.3% vs 13.4%, pared with locally trained medical graduates, but that has P=.57) and at 1 year (21.8% vs 21.9%, P=.63). Further- not been established. more, AMI patients treated by both groups had similar ad- justed likelihood of receiving secondary prevention medi- Methods: We performed a retrospective cohort study cations at 90 days and cardiac invasive procedures at 1 year. of linked administrative databases containing health care claims of physicians’ service payments, hospital dis- Conclusions: The use of secondary prevention medications charge abstracts, and patients’ vital status. We included and cardiac procedures and the mortality of AMI patients 127275 AMI patients admitted between April 1, 1992, were similar, regardless of the origin of medical education and March 31, 2000, to acute care hospitals in Ontario. of the admitting physician. This information places the care We then compared the risk-adjusted mortality rates and provided by IMGs into perspective and supports the abil- adjusted use of secondary prevention medications and ity of well-selected IMGs in caring for AMI patients. cardiac invasive procedures in patients treated by IMGs vs Canadian medical graduates. Arch Intern Med. 2005;165:458-463 NTERNATIONAL MEDICAL GRADU- evidence-based therapy for secondary ates (IMGs) in the United States prevention in patients with acute myocar- and Canada are usually referred dial infarction (AMI),9,10 and the appli- to as physicians who have com- cation of these therapies has been used pleted their undergraduate medi- as quality indicators of care.11 This well- Ical education outside of these countries. established performance framework al- They may have entered practice directly lowed a unique opportunity to compare or after completing a period of postgradu- the quality of medical care provided by Author Affiliations: Division of ate training in North America. Both coun- IMGs with the care provided by Cana- Cardiology and Schulich Heart tries use licensing requirements to screen dian medical graduates (CMGs). In addi- Centre (Dr Ko), Institute for the competency of IMGs,1,2 but whether tion, we compared the risk-adjusted mor- Clinical Evaluative Sciences these efforts translate into selecting those tality rates of AMI patients treated by IMGs (Drs Ko, Austin, Chan, and Tu), who can provide high-quality medical care and CMGs. and Division of General Internal Medicine and Clinical is uncertain. Many suspect that the qual- Epidemiology and Health Care ity of medical care provided by IMGs may METHODS 3-6 Research Program (Dr Tu), be inferior, but this hypothesis has not Sunnybrook and Women’s been fully evaluated.3 International medi- College Health Sciences Centre; cal graduates play a substantial role in the DATA SOURCES and Departments of Public delivery of health care, as they make up Health Sciences (Drs Austin, approximately one quarter of the physi- The Ontario Myocardial Infarction Database has 12 Chan, and Tu), Family and cian workforce in both countries.7,8 Un- previously been described. Briefly, the infor- Community Medicine derstanding the potential discrepancies of mation in this database is obtained by linking (Dr Chan), Health Policy, care provided by IMGs is important in the several health care administrative databases in Management and Evaluation Ontario. All 190354 patients admitted to On- (Drs Chan and Tu), and interests of many health care consumers tario hospitals with an AMI between April 1, Medicine (Dr Tu), Faculty and policy makers. 1992, and March 31, 2000, were identified of Medicine, University The American College of Cardiology based on a “most responsible diagnosis” of AMI of Toronto; Toronto, Ontario. and American Heart Association practice (International Classification of Diseases, Ninth Financial Disclosure: None. guidelines have highlighted the use of Revision code 410) in the Canadian Institute (REPRINTED) ARCH INTERN MED/ VOL 165, FEB 28, 2005 WWW.ARCHINTERNMED.COM 458 ©2005 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/27/2021 for Health Information hospital discharge database. We ex- STATISTICAL ANALYSIS cluded patients younger than 20 years or older than 105 years, as well as those who were not Ontario residents, had an in- We first compared the physician characteristics of IMGs and valid Ontario health card number, were admitted as transfers CMGs. Then, we compared the demographic and clinical char- from another acute care institution or to a noncardiac surgical acteristics of patients treated by IMGs with those of patients service, had an AMI coded as an in-hospital complication, were treated by CMGs. Categorical variables were compared be- discharged alive with a length of stay of less than 3 days, or tween the 2 groups using 2 tests, while continuous variables were admitted with an AMI in the year before the index ad- were compared using t tests or nonparametric tests. 2 Tests mission. The rationale for these inclusion and exclusion crite- compared the use of medical therapy, cardiac procedure use, ria is detailed elsewhere.12 A total of 146382 patients met these and mortality between IMG- and CMG-treated patients. This criteria, and the coding accuracy of AMI in this cohort is greater was repeated comparing IMGs from English-speaking devel- than 94%.12 The Canadian Institute for Health Information da- oped countries (England, Ireland, Australia, New Zealand, and tabase was then linked to the Ontario Registered Persons Da- South Africa) with IMGs originating from all other countries tabase, which contains information on the vital status of all On- where English was not the predominant language. We ex- tario residents. cluded US medical graduates in all our analyses because of their similarities with CMGs, which was consistent with previous studies.14,15 CLASSIFICATION OF ADMITTING PHYSICIANS We used the Ontario AMI mortality prediction rules to cal- culate risk-adjusted mortality rates.16 These rules are based on The admitting physician for each AMI patient was determined logistic regression models that predict 30-day and 1-year mor- by linking the Ontario Myocardial Infarction Database cohort tality after an AMI. International Classification of Diseases, Ninth to the Ontario Health Insurance Plan database, which con- Revision codes were used to identify the prevalence of 9 clinical tains information on physician claims for all fee-for-service bill- risk factors in the 15 secondary diagnostic fields of the Cana- ings in Ontario. The billing codes for each patient were ana- dian Institute for Health Information database and the age and lyzed, and the admitting physician was identified as the first sex of the patients. These variables included severity of cardiac physician who submitted a claim rendered on or after the ad- disease (shock, congestive heart failure, pulmonary edema, and mission date. Billing codes for emergency department physi- arrhythmia) and comorbid conditions (cerebrovascular dis- cians were not included in determining the admitting physi- ease, cancer, diabetes mellitus, and acute and chronic renal fail- cian. In the event that 2 or more physicians submitted claims ure). This model has good predictive power, with areas under on the admission date, the admitting physician was defined as the receiver operating characteristic curve of 0.78 for 30-day mor- the physician who submitted the most claims for follow-up care tality and 0.79 for 1-year mortality. These rules were validated during that hospitalization. A unique admitting physician was in 2 separate independent data sets in California and Manitoba, identified for 127275 patients, representing the final study co- with receiver operating characteristic curves of 0.77 and 0.78 for hort. Additional characteristics of the admitting physician were 30-day and 1-year mortality, respectively. The model develop- identified by linkage to the Corporate Provider Database of the ment and validation are described in detail elsewhere.16 Ontario Ministry of Health and verified against information in Multivariate analyses of the use of secondary prevention the Ontario Physician Human Resource Data Centre Data- medications and cardiac procedures provided by IMGs were base. These data sources provided information on the age, sex, also conducted using random-effects hierarchical logistic re- self-reported medical specialty, and medical school education gression models. A multilevel analysis allows one to correctly of the physicians. All patient identifiers and physician billing incorporate variables measured at different levels of the hier- numbers were encrypted to maintain patient and physician con- archy and to take into account the fact that the outcomes of fidentiality. patients under the care of a single physician or within the