|
|
| |||||||||||||||||||||||||||||||||||||||||
Background The respective roles of generalist and specialist physicians in the care of patients are currently a matter of debate. Information is limited about the knowledge and practices of generalist and specialist physicians regarding conditions that both groups treat, such as myocardial infarction.
Methods We therefore surveyed 1211 cardiologists, internists, and family practitioners in the states of New York and Texas about four treatments demonstrated by randomized clinical trials to be associated with improved survival after myocardial infarction (thrombolytic therapy, immediate and long-term use of aspirin, and long-term use of beta-blockers) and two treatments for which such evidence is lacking (diltiazem for patients with pulmonary congestion and prophylactic lidocaine). We asked physicians about the effect of each treatment on survival and the likelihood that they would prescribe each class of drugs.
Results For the four beneficial treatments, the cardiologists believed more strongly than the internists and family physicians that survival was improved by the treatment, and they were more likely to prescribe these drugs (P<0.001). For example, 94.1 percent of cardiologists said they were very likely to prescribe thrombolytic agents to treat an acute myocardial infarction, as compared with 82.0 percent of internists and 77.3 percent of family practitioners. Conversely, for the two treatments for which trials showed no evidence of a survival benefit, cardiologists were less likely than internists and family practitioners to think there was such a benefit and less likely to prescribe the drugs (P<0.001). For example, 4.7 percent of cardiologists reported that they were very likely to use prophylactic lidocaine, as compared with 13.1 percent of internists, and 16.5 percent of family practitioners. When we used logistic regression to adjust for potential confounders, all the differences between the cardiologists and the internists and family practitioners remained significant (P<0.02).
Conclusions Internists and family practitioners are less aware of or less certain about key advances in the treatment of myocardial infarction than are cardiologists. This finding underscores the need to improve the dissemination of information from clinical trials to generalist physicians, particularly if they are to have an enlarged role in the evolving health care system.
Myocardial infarction is an illness for which both generalists and specialists deliver care. It is the leading cause of death in the United States9 and accounts for nearly 700,000 hospital discharges annually10. Therefore, physicians' treatment of myocardial infarction is an important model for assessing the knowledge and practices of generalist physicians in the fields of family medicine and internal medicine as compared with those of specialists in the field of cardiology.
Over the past 30 years, randomized, controlled trials of drug therapy have produced valuable new insights into the treatment of myocardial infarction11,12. Such trials have demonstrated clear improvements in survival associated with the use of thrombolytic agents,13,14 aspirin,14,15 and beta-blockers16,17,18 in patients with myocardial infarction. Conversely, randomized trials have not shown improvements in survival with the prophylactic use of lidocaine19,20 or the use of calcium-channel blockers21,22,23,24 for this condition. Physicians at hospitals that participate in clinical trials may rapidly apply the results of such trials in their practices,25,26 but data from broader samples of hospitals have shown that thrombolytic agents are still used in a relatively small proportion of patients after myocardial infarction and that calcium-channel blockers are used more frequently than beta-blockers27,28.
Previous surveys of physicians have suggested that the prescription practices of internists and family practitioners differ from those of cardiologists,29,30 but this finding has not been uniform31. We surveyed physicians in the states of New York and Texas who had recently treated patients with acute myocardial infarction to assess their knowledge and practices regarding drug therapy for this condition.
Methods
Study Population
We selected New York and Texas because they are large states (together they account for 14 percent of the U.S. population32), they are in different regions of the country, and they are the focus of an extensive analysis of the costs and outcomes of acute myocardial infarction by us and others. These two states also have different patterns of care for patients with myocardial infarction; for example, coronary angiography is used much more frequently to assess patients with myocardial infarction in Texas than in New York (unpublished data).
We obtained the names and addresses of all licensed cardiologists, internists, and family practitioners from the Medical Society of the State of New York (including both members and nonmembers of the society) and the Texas State Board of Medical Examiners as of January 1993. These organizations update their records through regular contact with physicians and other professional organizations. Before sampling, we excluded physicians who were listed as retired or in training or who practiced outside the two states. The remaining physicians, available for sampling, included 1458 cardiologists, 6863 internists, and 1976 family practitioners in New York and 619 cardiologists, 3068 internists, and 3346 family practitioners in Texas.
Physicians were considered eligible if they reported serving as the attending or consulting physician for at least one patient hospitalized with acute myocardial infarction during the three months before the survey. Our objective was to obtain completed surveys from at least 1000 eligible physicians, roughly half from each state and one third from each of the three fields. This sample would provide the study with a statistical power of 80 percent to detect an absolute difference of 15 percent in responses between any two fields of practice within one state.
To estimate eligibility and response rates before the full sample was surveyed, we initially distributed the questionnaire to a randomly selected subsample of 389 physicians. The total random sample of 4386 physicians (including the initial 389 physicians) comprised 348 cardiologists, 848 internists, and 974 family practitioners in New York and 316 cardiologists, 788 internists, and 1112 family practitioners in Texas.
Data Collection
We designed a four-page questionnaire to assess physicians' knowledge about and practices regarding drug therapy for myocardial infarction. We inquired about four treatments in association with which randomized trials published in major clinical journals have demonstrated clear improvements in overall survival; they were thrombolytic agents administered within 6 hours of the onset of myocardial infarction,13,14 aspirin initiated within 24 hours14 and continued as long-term therapy,15 and the long-term use of beta-blockers16,17,18. We also inquired about two treatments that have not been shown to improve survival in randomized trials and that may actually reduce survival: prophylactic administration of lidocaine within 24 hours after the onset of myocardial infarction19,20 and diltiazem for patients with Q-wave myocardial infarctions and pulmonary congestion23,24.
We elicited each physician's beliefs about the effect of specific treatments on overall survival among patients under 75 years of age, based on reported studies. We chose this age limit because published data are less extensive for older patients than for younger ones. We asked the physicians to choose one of five terms to describe the effect of the four treatments: "definitely improve survival," "probably improve survival," "no clear effect on survival," "probably worsen survival," and "definitely worsen survival"; or to answer "not sure." We also asked physicians to rate the likelihood that they would use each class of drug in patients under 75 years of age with acute myocardial infarction. The choices were "very likely," "somewhat likely," "unlikely," "do not use after myocardial infarction," and "not sure."
We also collected demographic and professional data from the physicians, including their age and sex, whether they had board certification in their primary field of practice, the number of patients they had treated for acute myocardial infarction during the preceding three months, and the number of beds and teaching status of the hospital where they chiefly practiced.
The survey was distributed to the physicians by Northeast Research, a survey firm in Orono, Maine, along with a letter from one of us describing the purpose of the survey, the federal funding, and the voluntary and confidential nature of the survey. Physicians who did not respond to the initial mailing within three weeks received a second questionnaire by certified mail. The staff of Northeast Research called physicians who did not respond to the mailings, making at least eight telephone calls over a 15-day period to determine each physician's eligibility and willingness to complete the survey. Finally, a third mailing was sent to physicians who did not respond within four weeks of the last attempted call.
Eligibility and Response Rates
Of the 4386 physicians to whom we sent questionnaires, 12 had died. The remaining group of 4374 physicians were in the following categories: (a) 1211 were eligible and completed the survey; (b) 319 were eligible but declined to respond or did not return the survey; (c) 1818 returned the survey but were ineligible; and (d) 1026 had unknown eligibility and did not return the survey. On the basis of the 3348 physicians whose eligibility was known (categories a + b + c), we estimate that 45.7 percent of sampled physicians were eligible for the study [(a + b) divided by (a + b + c)]. When this eligibility rate was applied to the physicians with unknown eligibility (category d), the overall response rate was 60.6 percent33.
The eligibility rates according to field of practice in New York and Texas were 79.5 percent and 89.7 percent, respectively, for cardiologists; 40.3 percent and 41.8 percent for internists; and 40.9 percent and 32.9 percent for family practitioners. Among the eligible physicians, the response rates according to field in New York and Texas were 64.7 percent and 65.0 percent, respectively, for cardiologists, 54.1 percent and 61.1 percent for internists, and 64.2 percent and 57.3 percent for family practitioners.
In Texas, the 593 physicians who completed the survey were similar to the 637 physicians who did not respond in terms of mean age (both 45.9 years, P = 0.90) and the proportions who were men (92.2 percent vs. 90.1 percent, P = 0.19), who had graduated from medical schools in the United States and Canada (81.4 percent vs. 78.5 percent, P = 0.20), and who were practicing in metropolitan areas (83.3 percent vs. 85.4 percent, P = 0.31). Respondents were slightly more likely than nonrespondents to be non-Hispanic whites (79.4 percent vs. 74.1 percent, P = 0.05). In New York, only the name, address, and specialty of each physician were available to us, so we could not perform similar comparisons of respondents and nonrespondents.
Statistical Analysis
All analyses were performed on data from the 1211 completed surveys from eligible physicians. The demographic and professional characteristics of the physicians in the three fields are presented according to state. For the substantive questions relating to knowledge and practices, the patterns of response according to field of practice were similar in the two states; therefore, data have been combined for the two states in the tables, and results for each state are summarized in the text. Pairwise comparisons between fields were performed with the Wilcoxon rank-sum test. "Not sure" or missing responses to individual items were excluded from these analyses. Such responses were received from fewer than 5 percent of the respondents for each of the substantive questions, except for that related to the effect of diltiazem on survival among patients with a Q-wave myocardial infarction and pulmonary congestion (14.6 percent).
Multiple logistic regression was used to assess the relative likelihood that physicians in each field would believe that the four beneficial treatments definitely improved survival. For the two ineffective treatments, we calculated the relative likelihood that physicians would believe that the treatment probably or definitely improved survival (neither of these responses is supported by published data). We also assessed the odds that physicians in different specialties would be very likely to use each class of drugs to treat patients with myocardial infarction. All models controlled for the age of the physician (<40, 40 through 49, or 
50 years), board certification, the number of patients treated for myocardial infarction in the past three months (1 through 4, 5 through 9, or 10), and the number of beds (<200, 200 through 499, and 500) in the principal hospital in which the physician treated patients for myocardial infarction and its teaching status (university, university-affiliated, or community hospital).
Analyses were conducted with SAS statistical software34. Individual physicians were identified only by a confidential study number. Two-tailed P values are reported for descriptive statistics and 95 percent confidence intervals for adjusted odds ratios.
Results
Characteristics of the Physicians
The characteristics of the responding physicians in each field are presented in Table 1. In New York, but not in Texas, family practitioners were younger, on average, than physicians in the other two fields and more likely to be board certified in their specialty. As compared with internists and family practitioners, cardiologists in each state had recently treated more patients with acute myocardial infarction. In both states, family practitioners were more likely than cardiologists to work in community hospitals and smaller hospitals; internists occupied an intermediate position between the other two groups.
|
For the four drug therapies that are known to increase survival, cardiologists believed more strongly than internists and family practitioners that treatment improved overall survival after myocardial infarction (Table 2). With respect to thrombolytic agents, 93.1 percent of cardiologists believed therapy definitely improved survival, as compared with 85.4 percent of internists and 80.7 percent of family practitioners; for the immediate use of aspirin, these figures were 75.9 percent, 52.4 percent, and 44.7 percent; for the long-term use of aspirin, 71.6 percent, 54.6 percent, and 50.0 percent; and for the long-term use of beta-blockers, 75.3 percent, 54.2 percent, and 51.6 percent. Conversely, for the prophylactic use of lidocaine and the use of diltiazem for patients with Q-wave myocardial infarction and pulmonary congestion, cardiologists were less likely than internists and family practitioners to believe these treatments improved survival. Among cardiologists, 10.9 percent thought that prophylactic lidocaine definitely or probably improved survival, as compared with 34.0 percent of internists and 40.2 percent of family practitioners; for the use of diltiazem, these figures were 2.3 percent, 5.8 percent, and 16.5 percent. As shown in Table 2, beliefs about the efficacy of drugs did not differ significantly between internists and family practitioners, except for the immediate use of aspirin (P<0.03) and the use of diltiazem for patients with Q-wave myocardial infarctions and pulmonary congestion (P<0.001). In both cases, internists' beliefs fell between those of cardiologists and family practitioners.
|
In the logistic-regression models, the field of practice remained a significant predictor of beliefs (Table 3). After we adjusted for age, the number of patients treated recently who had acute myocardial infarction, the hospital's size and teaching status, and the state, cardiologists were more likely than both internists and family practitioners to respond that early thrombolytic therapy, the immediate and long-term use of aspirin, and the long-term use of beta-blockers definitely improved survival. Cardiologists were less likely than the physicians in the other two groups to believe that prophylactic lidocaine and diltiazem definitely or probably improved survival.
|
0.05). Board certification, the number of patients with myocardial infarction that each physician had treated recently, and the size and teaching status of each physician's principal hospital were not consistently significant predictors in the logistic-regression models. Prescription Practices
The physicians' reports of their prescription practices were consistent with their beliefs about the effects on survival of each class of drugs (Table 4). Cardiologists were more likely than internists and family practitioners to state that they would prescribe thrombolytic agents (94.1 percent said they were very likely to prescribe such agents, as compared with 82.0 percent and 77.3 percent, respectively), aspirin (98.7 percent, 91.3 percent, and 89.8 percent), and beta-blockers (77.5 percent, 63.0 percent, and 53.1 percent) for patients with acute myocardial infarction; cardiologists were least likely to report that they would use prophylactic lidocaine (4.7 percent, 13.1 percent, and 16.5 percent) and calcium-channel blockers (11.2 percent, 22.6 percent, and 26.0 percent). All these differences were significant (P 0.05) in stratified analyses in each state and in multivariate analyses (Table 5).
|
|
Use of Beta-Blockers
Beliefs and practices were generally similar among physicians in the two states, but within each field of practice more physicians in New York than in Texas believed that beta-blockers definitely improved survival (P<0.05). Accordingly, for each specialty, a larger proportion of physicians in New York than in Texas reported that they were very likely to prescribe beta-blockers (P<0.001); the proportions ranged widely, from 88 percent of cardiologists in New York to 37 percent of family practitioners in Texas (Figure 1). The state was a significant predictor of beliefs and practices regarding beta-blockers in the multivariate models (P<0.001).
|
Our survey of physicians in New York and Texas demonstrates that internists and family practitioners are less aware than cardiologists of the results of studies of the effect of important drugs on survival after myocardial infarction, or less certain about such results. Most striking was the finding that only about half of internists and family practitioners believed that the immediate and long-term use of aspirin and the long-term use of beta-blockers definitely improved survival, as compared with roughly three quarters of cardiologists. These variations in beliefs were associated with significant differences among the fields in the prescription of each class of drugs, and the patterns were consistent in each state.
Three factors may explain the differences in beliefs and practices between generalist and specialist physicians. First, internists and family practitioners may have less knowledge than cardiologists of the results of key clinical trials. Physicians maintain and refine their clinical knowledge by reading journals and textbooks, through discussions with colleagues, and through participation in conferences and continuing-education courses35,36,37. Cardiologists are better able than primary care physicians to focus their learning by these methods on myocardial infarction, since cardiologists treat patients for a narrower range of clinical problems than do internists or family practitioners. The major published guidelines for the treatment of myocardial infarction in the United States have appeared in journals that target cardiologists38,39; primary care physicians may also benefit from such guidelines.
A second explanation could be that physicians in generalist fields have greater concern about therapeutic complications, such as bleeding after thrombolytic therapy or congestive heart failure due to beta-blockers. Probably because aspirin is viewed as the most benign of the drugs in this study, 9 out of 10 internists and family practitioners who responded said they were very likely to prescribe aspirin after myocardial infarction -- even though only half believed that it definitely improved survival. However, this factor does not explain the greater willingness of internists and family practitioners to use prophylactic lidocaine and calcium-channel blockers, both of which can have important and occasionally lethal side effects.
A third explanation could be that generalist physicians may exercise greater caution than specialists in accepting new data or changing established patterns of treatment40. Some cardiologists may share this caution, as demonstrated by the one quarter who said they did not believe that aspirin and beta-blockers definitely improved survival after myocardial infarction. Even clinical experts can be slow to incorporate the results of clinical trials into their published recommendations41.
The relation of the age of physicians to their beliefs about drug therapy suggests that younger physicians have had more recent or more concentrated exposure to the results of relevant clinical trials than older physicians or that they may be more receptive to new standards of practice42. Physicians in New York were more likely than those in Texas to say they believed that beta-blockers improved survival after myocardial infarction and to use them for patients with this condition; this finding supports other studies that have demonstrated geographic variations in medical care43,44. The number of patients that each physician had recently treated for myocardial infarction was not a consistently significant predictor in the multivariate models. However, our ability to assess this variable was limited because it was closely correlated with field of practice.
Our study had several strengths. We found consistent results in a large sample of physicians from two regions of the United States. We focused on physicians who had recently treated patients for acute myocardial infarction, and the response rate was substantially higher than in other published surveys of physicians,30,45 thus reducing the possibility of bias related to the characteristics of nonrespondents.
This study has at least three limitations. First, the data on prescription practices are based on reports by physicians, not data from medical records. Second, some physicians may have based their answers on the limited clinical detail provided in the survey questions. For example, specifying the presence of ST-segment elevation might more accurately elicit beliefs about the effect of thrombolytic therapy on survival, and specifying the level of ventricular function might influence the responses to questions about the likelihood of prescribing calcium-channel blockers. Third, we expect that more appropriate use of drugs results in better clinical outcomes,46 but our data do not address this issue. In preliminary data from a study of patients with unstable angina, internists were less likely than cardiologists to use aspirin and beta-blockers in clinical practice, and mortality among patients treated by internists tended to be higher47.
Many internists and family practitioners are aware of the results of published clinical trials and meta-analyses of therapy for myocardial infarction and apply these results in their practices. On average, however, generalist physicians appear to be less aware of or less certain about these results than are specialists. A number of factors may improve the knowledge and practice patterns of physicians,48,49 including continuing-education programs, broader dissemination of guidelines, more rigorous recertification procedures, provision of feedback about clinical practices, and participation in community-based clinical trials. A multifaceted approach to increasing physicians' knowledge of trial results will help to ensure that important lessons from research are applied to medical care in an appropriate and timely manner.
Supported by a Patient Outcomes Research Team grant (HS06341) from the Agency for Health Care Policy and Research. Dr. Ayanian is a Generalist Physician Faculty Scholar of the Robert Wood Johnson Foundation.
We are indebted to Michael Elliott and David Kovenock of Northeast Research for administering the survey, to Alison Eastwood and Bud Davies for statistical programming, to Brian Egleston for assistance in the preparation of the manuscript, and to Anthony L. Komaroff, M.D., and Thomas J. Ryan, M.D., for helpful comments on an earlier draft of the manuscript.
Source Information
From the Division of General Medicine, Section on Health Services and Policy Research (J.Z.A.), and the Cardiovascular Division (P.J.H., E.M.A.) of the Department of Medicine, and the Department of Radiology (B.J.M.), Brigham and Women's Hospital and Harvard Medical School, Boston, and the Department of Health Care Policy, Harvard Medical School, Boston (J.Z.A., E.G., C.L.P., B.J.M.).
Address reprint requests to Dr. Ayanian at the Department of Health Care Policy, Harvard Medical School, 25 Shattuck St., Parcel B, 1st Fl., Boston, MA 02115.
References
-Blocker Heart Attack Trial Research Group. A randomized trial of propranolol in patients with acute myocardial infarction. I. Mortality results. JAMA 1982;247:1707-1714. [Abstract]
| |||||||||||||||||||||||||||||||||||||||||
Related Letters:
Knowledge of Drugs for Myocardial Infarction: Generalists versus Specialists
Barry H. C., Justice A. C., Hack T. C., Markson L. E., Turner B. J., Ayanian J. Z.
Extract |
Full Text
N Engl J Med 1995;
332:472-474, Feb 16, 1995.
Correspondence
Assessing the Appropriateness of Medical Care
Wassertheil-Smoller S., Tobin J., Steingart R., Hsu J., Black N., Ayanian J. Z., Shekelle P. G., Park R.E., Naylor C. D.
Extract |
Full Text
N Engl J Med 1998;
339:1478-1481, Nov 12, 1998.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
