Seventeen randomised controlled trials (RCTs) were included (n=1620 patients). Twelve RCTs were double blind and the remaining were open-label or no details of blinding were provided. Five trials reported an adequate method of allocation concealment, but for most of the remaining trials it was unclear.
Chest-tube drainage: Both aprotinin (12 RCTs, n=992 patients) and lysine analogues (three RCTs, n=259 patients) were more effective than placebo at reducing chest-tube drainage; this was statistically significant for both interventions. For the overall pooling, the mean difference between antifibrinolytic and placebo was 374 millilitres (mL) (95% confidence interval (CI): 275 to 473). Statistical heterogeneity was high for the overall pooling and the two subgroups. There was no statistically significant difference in chest-tube drainage between aprotinin and lysine analogues based on three trials (mean difference 24 mL, 95% CI: -12 to 60; n=502 patients), and there was no statistical heterogeneity.
Transfusion: Patients receiving an antifibrinolytic were significantly less likely to receive any blood products than those receiving placebo (odds ratio 0.37, 95% CI: 0.27 to 0.49; n=935 patients) based on 10 trials of aprotinin and one of a lysine analogue. There was also a statistically significant benefit for patients receiving an antifibrinolytic agent compared to placebo for mean number of packed red blood cells transfused per patient, the mean number of units transfused for all blood products and proportion of patients receiving at least one unit of packed red blood cells. There was evidence of statistical heterogeneity for the analysis of mean number of units of packed red blood cells transfused.
Other outcomes: There was no statistically significant difference between patients receiving an antifibrinolytic agent compared to placebo for the following outcomes: need for surgical re-exploration (nine trials reporting 17 events for 461 patients); mortality (seven trials reporting five deaths in 646 patients); myocardial infarction (seven trials reporting 14 events in 361 patients); stroke (five trials reporting six events in 241 patients); and thrombotic complications (11 trials reporting 28 events in 786 patients). None of the trials reported on renal failure.
Sensitivity analyses: Sensitivity analyses investigated the impact of excluding individual trials on the pooled estimates of effect; excluding the three trials with greatest impact on chest-tube drainage, or excluding some trials that were not of on-pump coronary artery bypass graft, did not alter the results. Analysis investigating effect of aprotinin dose, whether last exposure to aspirin was more than 48 hours prior to surgery, use of a transfusion algorithm at time of recording test-tube drainage, and trial quality, also did not alter the results. Based on the funnel plots, the authors reported no evidence of publication bias.