Liberal versus restrictive transfusion strategies in acute myocardial infarction: a systematic review and comparative frequentist and Bayesian meta-analysis of randomized controlled trials.

Abstract

Background: The transfusion strategy in the acute phase of myocardial infarction (AMI) remains a debated topic with non-standardized guidelines. This study aimed to evaluate the impact of liberal versus restrictive transfusion strategies on mortality during AMI.

Methods: A systematic search was conducted across MEDLINE, EMBASE, and the COCHRANE library databases, focusing on randomized controlled trials (RCTs). The primary endpoint was the latest measured mortality within 90 days following myocardial infarction (MI). Secondary endpoints included recurrence of MI, cardiovascular mortality, stroke occurrence, unplanned revascularization, and a composite endpoint of death or recurrent MI. Mixed and random-effects models were employed to estimate relative risks. Sensitivity analyses were conducted using two approaches: one incorporating only studies assessed as low risk of bias according to the Rob2 tool, and another employing a Bayesian analysis.

Results: Four RCTs including a total of 4324 participants were analyzed. Neither the fixed-effect nor random-effects models demonstrated a significant reduction in mortality, with risk ratios (RR) of 1.16 (95% CI 0.95-1.40) for the fixed-effect model and 1.13 (95% CI 0.67-1.91) for the random-effects model (GRADE: low certainty of evidence). Sensitivity analyses, including the exclusion of two high-risk-of-bias studies and a Bayesian analysis, were consistent with the primary analysis. For the composite outcome death or MI both fixed-effect and random-effects models showed a statistically significant RR of 1.18 (95% CI 1.01-1.37) with negligible heterogeneity (I² = 0%, p = 0.46), indicating results unfavorable to restrictive transfusion (GRADE: very low certainty of evidence). However, this result was primarily driven by a single study. For cardiac mortality, the fixed-effects model indicated a significant RR of 1.42 (95% CI 1.07-1.88), whereas the random-effects model showed non-significant RR of 1.05 (95% CI 0.36-3.80). Analyses of other secondary endpoints did not show statistically significant results.

Conclusions: Our analysis did not demonstrate a significant benefit in early mortality with a liberal transfusion strategy compared to a restrictive strategy for AMI, low certainty of evidence. Liberal transfusion may reduce the risk of the composite outcome death or MI, with very low certainty of evidence. These findings should be interpreted with caution in critically ill patients.