We read with great interest the report by Bukkapatnam et al, in which multivariate analysis indicated that women are at higher risk for operative mortality than men (adjusted odds ratio [OR] 1.61, 95% confidence interval [CI] 1.40 to 1.84) after coronary artery bypass grafting (CABG). In 2003, Nalysnyk et al suggested that female gender is associated with increased risk for death after CABG (pooled unadjusted OR 1.92, 95% CI 1.48 to 2.48) in their meta-analysis of 8 studies published through 2001. No further meta-analyses have been conducted to date. We report here a contemporary meta-analysis of gender differences in mortality after CABG, combining not unadjusted but adjusted ORs for death.
To identify all contemporary studies providing adjusted ORs for perioperative death in women compared to men who underwent CABG, public-domain databases, including Medline and Embase, were searched from January 2000 to June 2010 using Web-based search engines (PubMed and Ovid), with exploding keywords including “sex;” “gender;” “women;” “female;” and “coronary artery bypass.” Studies considered for inclusion met the following criteria: the study population was patients who underwent on-pump and off-pump CABG; and main outcomes included adjusted ORs (or; if unavailable; risk ratios or hazard ratios) for 30-day or in-hospital death using appropriate statistical methods (i.e.; multivariate logistic regression; Cox regression; and propensity score analysis) in women compared to men. We excluded studies reporting only unadjusted mortality or ORs for death. All analyses were conducted with RevMan version 5.0 (Nordic Cochrane Centre; Copenhagen; Denmark) and Comprehensive Meta-Analysis version 2 (Biostat; Englewood; New Jersey).
Our search identified 27 studies providing 25 adjusted OR, 2 risk ratios, and 2 hazard ratios for death after CABG. Pooled analysis (representing 1,715,360 patients) demonstrated a statistically significant increase in mortality in women compared to men in a random-effects model (pooled adjusted OR 1.38, 95% CI 1.29 to 1.49, p <0.00001; Figure 1 ). There was significant between-study heterogeneity (p <0.00001) and little difference in the pooled results of the fixed-effects model (pooled adjusted OR 1.42, 95% CI 1.39 to 1.45, p <0.00001). The exclusion of any single study from the analysis did not substantively alter the overall results of our analysis. There was no evidence of significant publication bias (p = 0.59 by adjusted rank correlation test).