Author + information
- Raj Ganeshan, MD,
- Brian Malm, MD and
- John Concato, MD, MS, MPH∗ ()
- ↵∗Yale University School of Medicine, Veterans Affairs Connecticut Healthcare System, Clinical Epidemiology Research Center, 950 Campbell Avenue (Mailcode 151B), West Haven, Connecticut 06516
In a post-hoc analysis, Yeh et al. (1) provide evidence that dual-antiplatelet therapy with thienopyridine and aspirin, for 30 (vs. 12) months after coronary stenting, had beneficial effects regardless of whether patients had acute myocardial infarction (MI) or less acute presentations. In particular, stent thrombosis was reduced—although bleeding was increased—similarly in both groups. Clinicians should, however, consider various elements of the study design when interpreting results and using the findings in clinical practice.
Importantly, the study classified patients with unstable angina as having less acute (“no MI”) presentations. This subset accounted for nearly one-fourth (22.6%) of the 8,072 patients without MI, or roughly 1,800 patients—representing more than one-half of the 3,576 patients with MI. Combining unstable angina patients within the non-MI group is of more utility in allowing for conclusions to be drawn on the MI group, given that the non-MI group represents a population of patients with a wide range of risk of subsequent events—and it is unclear whether the beneficial effects of prolonged treatment was driven by those at higher risk within this group. Perhaps a sensitivity analysis could be done, excluding patients with unstable angina from the non-MI group, to see if the overall results are affected.
From a more general perspective, the reader may not immediately recognize that the current analyses represent a post-hoc subgroup analysis of a previously completed randomized trial, with randomization 12 months after initial enrollment. We would defend the authors from reflexive criticisms that such analyses are always problematic, but we also note that whether the findings are valid, or due to chance or bias, is always a consideration. For example, lacking a formal power calculation linked to antecedent MI status, the discordant findings of similarly reduced stent thrombosis without an interaction, compared to differential reduction of major adverse cardiovascular and cerebrovascular events with an interaction, have uncertain implications.
Finally, as a specific methodological issue, the data in Figure 1 in the paper by Yeh et al. (1) indicate that (25,682 total minus 5,844 not eligible equals) 19,838 patients were eligible for randomization, but 8,190 were not randomized at 12 months for the reasons stated. Given that this group represents (8,190 divided by 19,838 equals) 41.3% of eligible patients, it would be useful to know whether pertinent baseline characteristics (including the distribution of patients in MI and non-MI groups) differ between analyzed and excluded patients, for better understanding the overall results.
Please note: The authors have reported that they have no relationships relevant to the contents of this paper to disclose. P.K. Shah, MD, served as Guest Editor for this paper.
- 2015 American College of Cardiology Foundation