Correspondence

Knowledge of Drugs for Myocardial Infarction: Generalists versus Specialists

N Engl J Med 1995; 332:472-474February 16, 1995

Article

To the Editor:

The study of differences in knowledge between cardiologists and primary care physicians by Ayanian and colleagues (Oct. 27 issue)1 raises interesting questions. The authors assume that the results of the studies they cite are valid, that all physicians have read or have had access to these studies, and that physicians will draw similar conclusions and share the same degree of enthusiasm for the results. Reasonable, well-trained physicians may in fact read the same paper and draw different conclusions about its application because the study population is unlike their patients and the results are therefore not generalizable. An entire generation of physicians has learned to read the literature critically in the McMaster style,2,3 which among other things clearly emphasizes the issue of generalizability. Now, generalist physicians who apply these techniques are potentially labeled as having less knowledge than specialists when in fact they are applying knowledge differently. One wonders how the Journal would handle a survey comparing the knowledge of primary care physicians with that of cardiologists (or any other specialty) with regard to issues pertaining to “real” determinants of health (ancestry, diet, exercise, safety, and good fortune).

Of great concern, however, is the validity of the authors' conclusions. Their internal validity is threatened by the inconsistent classification of outcomes used to assess the relative likelihood that physicians would think favorably of the four beneficial treatments (for which only the “definitely improves survival” category was used in their Table 3) as compared with the two ineffective treatments (which could be rated as “definitely or probably” improving survival). If the latter, more liberal classification is used, the results look very different (Table 1Table 1Modification of Table 2 of Ayanian et Al., Showing the Percentages of Physicians Who Reported That Each Drug Therapy Definitely or Probably Improved the Survival of Patients under 75 Years of Age with Acute Myocardial Infarction.). No rationale is given for using these inconsistent categories of responses.

Henry C. Barry, M.D.
Michigan State University, East Lansing, MI 48824-1313

3 References

1
Ayanian JZ, Hauptman PJ, Guadagnoli E, Antman EM, Pashos CL, McNeil BJ. Knowledge and practices of generalist and specialist physicians regarding drug therapy for acute myocardial infarction. N Engl J Med 1994;331:1136-1142
Full Text | Web of Science | Medline

2
Department of Clinical Epidemiology and Biostatistics, McMaster University Health Sciences Centre. How to read clinical journals. I. Why to read them and how to start reading them critically. Can Med Assoc J 1981;124:555-558
Web of Science | Medline

3
Oxman AD, Sackett DL, Guyatt GH. Users' guides to the medical literature. I. How to get started. JAMA 1993;270:2093-2095
CrossRef | Web of Science | Medline

To the Editor:

Ayanian et al. asked physicians to rate the likelihood that they would use each of six drug treatments in patients under 75 years of age with acute myocardial infarction. After eliminating the “not sure” responses, the authors present a table (Table 4) that collapses the responses into three categories: “very likely to prescribe,” “somewhat likely to prescribe,” and “unlikely to prescribe or do not prescribe.” For the purposes of multivariate logistic regression, they further collapse the responses into two categories — “very likely” and all other responses. On the basis of the information in Table 4 and the aforementioned logistic regression, the authors conclude that generalists are less likely than specialists to prescribe effective cardiac medications and are more likely to prescribe ineffective (or potentially harmful) agents.

I collapsed the three response categories from Table 4 into two categories by combining the “very likely” responses with the “somewhat likely” responses. My Table 1Table 1Modification of Table 4 of Ayanian et Al., Showing Drug-Prescription Practices of Physicians in New York and Texas Treating Patients under 75 Years of Age with Acute Myocardial Infarction, According to Field of Practice. contains two categories — physicians who are likely to prescribe the medication and those who are unlikely to do so. This revised presentation shows that at least 95 percent of physicians in each group are likely to prescribe thrombolytic agents, aspirin, and beta-blockers and that there remain only small differences in prescription practices among groups. I suspect that a revised multivariate logistic regression in which the results were dichotomized in this manner would not demonstrate a significant difference between cardiologists and generalists with regard to prescribing practices for these medications.

In contrast, there remain differences between cardiologists and generalists when the “ineffective” agents are considered (lidocaine and calcium-channel blockers). Although the majority of physicians in each group were unlikely to prescribe lidocaine, a greater proportion of generalists would use that agent. Furthermore, the majority of physicians in each group were likely to prescribe calcium-channel blockers, but a greater proportion of generalists than of specialists would do so.

On the basis of this revised presentation of the data, I conclude that internists and family practitioners assimilate new treatments reasonably well. They differ from cardiologists in their ability to abandon old treatments that have been shown to be ineffective. Instead of demonstrating that generalists are out of touch, the data suggest that it is harder to extinguish practices that were once accepted (such as the use of lidocaine or calcium-channel blockers) than it is to accept new practices (such as the use of aspirin, thrombolytic agents, or beta-blockers). In this light, the results emphasize the importance of publishing the results of negative post-marketing drug trials and of incorporating this information into programs of continuing medical education.

Amy C. Justice, M.D
University of Pennsylvania, Philadelphia, PA 19104-6095

To the Editor:

The timely study by Ayanian et al. dramatically illustrates the problems with the ill-conceived notion of placing increased emphasis on the primary care provider. Primary care physicians, although vital in their role as orchestra leaders and generalists, are in fact not the best-qualified physicians to care for patients with acute myocardial infarctions. This study demonstrates that well-studied treatments critical to improved survival are not understood or used by noncardiologists. The treatment of patients presenting with an acute myocardial infarction has both immediate and long-term consequences that directly affect the patients' survival. The conclusion this study reaches about the need for improved knowledge, although true, falls far short of the real finding — that providing the best possible care to any patient requires an appropriately trained physician. The obligation of the primary care provider, whether generalist or specialist, should be to provide the patient with the best possible care. In the case of an acute myocardial infarction, the cardiologist has the training, skills, and judgment to treat the patient best. . . .

Terrence C. Hack, M.D
Deaconess–Nashoba Hospital, Ayer, MA 01432

To the Editor:

The article by Ayanian and colleagues provides disturbing survey evidence that generalists are less certain than cardiologists about making use of selected advances in the treatment of myocardial infarction. The findings of these authors are similar to those we have reported about the use of zidovudine by AIDS specialists and generalists to treat patients with AIDS during the first three years after Food and Drug Administration approval of the drug.1 Using claims data, we found a three-year delay before the patients of generalists used zidovudine at the same rate as patients who received most of their care from AIDS specialists. However, when the patients of generalists had at least one consultation with an AIDS specialist within six months of the AIDS diagnosis, their probability of receiving zidovudine was similar to that of patients who received the majority of their care from specialists. We focused on the use of zidovudine in patients with AIDS, who have consistently been found to have a survival advantage when the therapy is begun at this point in the disease course.2,3 Our findings offer empirical support for the view that consultation, rather than primary care by specialists, may suffice for new therapeutic approaches to receive timely adoption.

Leona E. Markson, Sc.D.
Barbara J. Turner, M.D.
Jefferson Medical College, Philadelphia, PA 19107-5083

3 References

1
Markson LE, Cosler LE, Turner BJ. Implications of generalists' slow adoption of zidovudine in clinical practice. Arch Intern Med 1994;154:1497-1504
CrossRef | Web of Science | Medline

2
Fischl MA, Richman DD, Grieco MH, et al. The efficacy of azidothymidine (AZT) in the treatment of patients with AIDS and AIDS-related complex: a double-blind, placebo-controlled trial. N Engl J Med 1987;317:185-191
Full Text | Web of Science | Medline

3
Lundgren JD, Phillips AN, Pedersen C, et al. Comparison of long-term prognosis of patients with AIDS treated and not treated with zidovudine. JAMA 1994;271:1088-1092
CrossRef | Web of Science | Medline

Author/Editor Response

Dr. Ayanian replies:

To the Editor: The way in which physicians learn, interpret, and apply the results of clinical trials is an important subject that merits further study. Although Barry speculates that generalist physicians who are less convinced of the benefits of thrombolytic therapy, aspirin, and beta-blockers may not consider the results of clinical trials to be generalizable to their patients with myocardial infarction, we are not aware of any evidence to support his view. In fact, major textbooks of internal medicine and family practice have strongly recommended each of these three beneficial therapies.1,2

Barry also questions the rationale for the classification that we present in Table 3 of our report. As we specified in our Methods section, published data do not support the belief that there is either a definite or a probable improvement in survival with diltiazem or lidocaine. In contrast, clinical studies do provide clear evidence of improved survival with thrombolytic therapy, aspirin, and beta-blockers, so we focused on whether physicians viewed these treatments as definitely improving survival.

We agree with Justice about the importance of publicizing evidence when established treatments are shown to be ineffective. However, her revision of Table 4 of our report may overestimate the use of effective drugs. For example, among patients with a prior myocardial infarction who received routine care from 1991 through 1993, preliminary data demonstrate that the use of aspirin and beta-blockers in routine clinical practice (in 53 percent and 34 percent of patients, respectively) was substantially lower than the nearly universal use that Justice's interpretation of our data would imply.3 We are collecting additional data from medical records with which to compare the actual prescribing practices of generalist and specialist physicians.

Although we concur with Hack that cardiologists in our survey were more likely than internists and family practitioners to recognize and apply the results of clinical trials, we would add two notes of caution. First, a sizable proportion of generalist physicians reported knowledge and practices that were similar to those of the majority of cardiologists. Second, only three quarters of cardiologists believed that aspirin and beta-blockers definitely improve survival. These findings underscore the fact that the field of training is one of numerous factors influencing clinical practice.

Finally, the recent report by Markson et al.4 supports our view that effective collaboration between generalists and specialists is essential to ensuring that important results of clinical trials are applied widely to medical care in an appropriate and timely manner.

John Z. Ayanian, M.D., M.P.P.
Harvard Medical School

for the Authors

4 References

1
Pasternak RC, Braunwald E. Acute myocardial infarction. In: Wilson JD, Braunwald E, Isselbacher KJ, et al., eds. Harrison's principles of internal medicine. 12th ed. Vol. 1. New York: McGraw-Hill, 1991:953-64.

2
Castle H. Chest pain. In: Rakel RE, ed. Textbook of family practice. 4th ed. Philadelphia: W.B. Saunders, 1990:878-82.

3
Antman EM, Cannon CP, Mueller H, et al. Do clinical trial results influence physician drug use in myocardial infarction? Circulation 1994;90:Suppl I:I-167 abstract.

4
Markson LE, Cosler LE, Turner BJ. Implications of generalists' slow adoption of zidovudine in clinical practice. Arch Intern Med 1994;154:1497-1504
CrossRef | Web of Science | Medline

Citing Articles (1)