We read with interest the report by Gnanaraj et al. comparing extended pharmacoinvasive percutaneous coronary intervention (PCI) (ePIPCI; 3 to 48 hours after fibrinolysis) with primary PCI (PPCI) in ST-segment elevation myocardial infarction (STEMI) patients from the Madras Medical College Registry. The authors address a clinically important question, particularly in resource-constrained settings where immediate PPCI may not be consistently available, and we commend them for presenting detailed real-world data while transparently acknowledging the observational nature of their analysis. In this context, the interpretation of “comparable” outcomes between PPCI and ePIPCI warrants further clarification considering the reported data and the inherent limitations of the study design.
In-hospital mortality occurred in 11 of 248 PPCI patients (4.4%) and 3 of 210 ePIPCI patients (1.4%), yielding a relative risk of 3.11 (95% confidence interval [CI] 0.88 to 10.98; p = 0.07). While the difference was not statistically significant, the wide confidence interval indicates substantial imprecision and is compatible with both clinically meaningful benefit and harm. As emphasized by Altman and Bland, absence of statistical significance does not constitute evidence of equivalence, particularly when a study is neither designed nor powered to establish noninferiority. In the absence of a prespecified noninferiority margin or formal equivalence framework, describing outcomes as “comparable” may overstate the inferential strength of statistically nonsignificant findings.
Moreover, the PPCI versus ePIPCI comparison presented in Table 2 is univariable. Notably, pre-PCI TIMI 2/3 flow differed markedly between groups (21.4% in PPCI vs 70% in ePIPCI; p <0.001), a variable that is strongly prognostic in STEMI. Such imbalance underscores the potential for confounding by indication. In observational studies, treatment allocation driven by clinical presentation, timing, and system-level constraints may influence both exposure and outcome. Although the authors performed multivariable modeling for the overall cohort, this analysis evaluated PCI versus non-PCI and did not provide an adjusted effect estimate directly comparing PPCI with ePIPCI. Consequently, residual confounding in the PPCI–ePIPCI comparison cannot be excluded.
Even advanced analytic approaches, including propensity score or instrumental variable methods, do not reliably eliminate confounding by indication when important prognostic factors remain unmeasured. As Bosco et al. have demonstrated, no adjustment method fully resolves this bias in nonrandomized settings. Accordingly, comparative effectiveness interpretations derived from such analyzes should be framed cautiously and regarded as hypothesis-generating rather than confirmatory.
The Madras Medical College STEMI Registry provides valuable insight into revascularization strategies within a resource-constrained healthcare system. We believe the data are most accurately interpreted as demonstrating no statistically significant difference in observed outcomes, while recognizing the limited precision and inherent constraints of the study design. Future adequately powered prospective studies will be essential to clarify whether extended pharmacoinvasive PCI yields outcomes equivalent to primary PCI in similar contexts.
Declaration of Generative AI and AI-Assisted Technologies in the Writing Process
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
