We thank Dr. Bell (1) for his interest in and insightful comments about our study. Dr. Bell proposes that the use of lower doses of cardioselective β-blockers could have reduced the effectiveness of this class of agents after myocardial infarction (MI) in diabetic subjects, a reasonable hypothesis. Another hypothesis is that short-acting cardioselective β-blockers lead to periods of withdrawal and decreased efficacy (2).
As in previous studies (3, 4), we assumed a class effect among β-blockers. Unfortunately, we could not differentiate the type of β-blockers that were prescribed, although our sample size would have been precipitously reduced if we had done so. However, atenolol and metoprolol are the most commonly prescribed β-blockers in Canada (5).
Dr. Bell asserts that cardioselective β-blockers have never been shown to reduce mortality post-MI unless used intravenously. However, meta-analyses have demonstrated a 25% reduction in mortality with long-term β-blockade after MI (including cardioselective agents) (6, 7); cardioselectivity or membrane-stabilizing activity did not predict mortality. Rather, Yusuf et al. (6) demonstrated decreased effectiveness when trials using agents with intrinsic sympathomimetic activity (ISA) were pooled (odds ratio [OR] for agents with ISA 0.90 [0.77–1.05]; OR without ISA 0.69 [0.61–0.79]). Freemantle et al. (7) also failed to demonstrate a significant effect of cardioselectivity, although the OR for the protective effect of β-blockers with ISA was 1.19 (0.96–1.47). Moreover, older observational studies of β-blockers (as a class) post-MI, conducted in an era when metoprolol and atenolol were frequently prescribed, have found no effect of cardioselectivity (3, 4, 8). For example, in the Bezafibrate Infarction Prevention (BIP) study, mortality was similar in patients who received propranolol or a cardioselective β-blocker (4), and in the Diabetes Insulin-Glucose in Acute Myocardial Infarction (DIGAMI) study, metoprolol was an independent predictor of reduced mortality after MI (relative risk ratio 47%) (8). If Dr. Bell’s hypothesis is correct, it seems unlikely that these earlier studies would have shown mortality benefit.
Therefore, we believe that the putative benefits of β-blockade were less in our study because of cointerventions that also reduce mortality. Future studies should include the type of β-blocker as a covariate rather than assuming a class effect, as we and others have done. The answer to this question is clinically important, and we thank Dr. Bell for his interest in our work.
