Thirty-six trials (5,576 patients; sample size range 20 to 1,440) were included. Overall, the authors stated that there was moderate risk of bias across trials. All except one trial were single centre, and the processes of randomisation and blinding were not reported in most cases.
There were no statistically significant differences (using fixed-effect meta analyses) between blood and crystalloid cardioplegia for risk of mortality (RR 0.95, 95% CI 0.6 to 1.51; 24 trials; Ι²=0%), perioperative myocardial infarction (RR 0.80, 95% CI 0.55 to 1.12; 25 trials; Ι²=0%) or perioperative low cardiac output syndrome (RR 0.77, 95% CI 0.58 to 1.14; nine trials; Ι²=18.6%). Random-effects meta-analyses showed similar results.
There was no evidence of publication bias, and subgroup/meta-regression analyses did not alter the main findings.