Across 73 meta-analyses mortality improvements are uncommon with newer interventions in adult cardiac surgery

We aimed to assess how often randomized controlled trials (RCTs) in adult cardiac surgery found significant mortality benefits for newer interventions vs older ones, whether observed treatment effect estimates changed over time and whether RCTs and nonrandomized observational studies gave similar results.

We searched journals likely to publish systematic reviews on adult cardiac surgery for meta-analyses of mortality outcomes and that included at least 1 RCT, with or without observational studies. Relative treatment effect sizes were evaluated overall, over time, and per study design.

A total of 73 meta-analysis comparisons (824 study outcomes on mortality, 519 from RCTs, 305 from observational studies) were eligible. The median mortality effect size was 1.00, IQR 0.54-1.30 (1.00 among RCTs, 0.91 among observational studies, P = .039). Four RCTs and six observational studies reached P < .005 favoring newer interventions. Two meta-analyses reached P < .005 favoring newer interventions. Effect size for experimental interventions relative to controls did not change over time overall (P = .64) or for RCTs (P = .30), and there was a trend for increase in observational studies (P = .027). In 34 meta-analyses with both RCTs (n = 95) and observational studies (n = 305), the median relative summary effect (summary effect in observational studies divided by summary effect in RCTs) was 0.87 (IQR, 0.55-1.29); meta-analysis of the relative summary effects yielded a summary of 0.93 (95% CI, 0.74-1.18).

The vast majority of newer interventions had no mortality differences over older ones both overall and specifically in RCTs, while benefits for newer interventions were reported more frequently in observational studies.