Twenty-three retrospective studies were included with 7,759 participants (range 39 to 1,314) 2,822 of whom were off-pump and 4,937 on-pump. Quality scores ranged from 2 to 11. Ten studies were considered high quality. Matching scores ranged from 1 to 21. Eleven studies were considered to have high matching. Funnel plots did not show evidence of publication bias.
Compared to on-pump bypass, off-pump bypass was associated with lower 30-day mortality (OR 0.64, 95% CI 0.51 to 0.81; Ι²=0%; 22 studies) but no statistically significant difference in mid-term mortality (OR six studies; HR seven studies), long-term mortality (four studies) and myocardial infarction (Ι²=4%; 18 studies). There was no significant statistical heterogeneity. In meta-regression analyses total arterial revascularisation in the on-pump groups was associated with lower 30-day myocardial infarction. Other analyses showed no significant effects.
Compared to on-pump, off-pump bypass was associated with less complete revascularisation (OR 0.23, 95% CI 0.12 to 0.42; four studies with no significant heterogeneity), fewer grafts to circumflex branch (OR 0.06, 95% CI 0.01 to 0.23; five studies with significant heterogeneity) and fewer grafts in the posterior descending artery (OR 0.13, 95% CI 0.04 to 0.38; four studies with significant heterogeneity).
Subgroup analyses for only studies with participants with poor left ventricular function (≤30%), outcomes for short (13 studies; Ι²=0%), mid-term (OR three studies; HR four studies) and long-term (two studies) mortality were similar to the main analyses (no significant heterogeneity). Results for other subgroups were reported.