Figure 2.

A-B, All-cause mortality. A, Meta-analysis of all-cause mortality in eight trials with low risk of bias (RoB) and five some concern/high concern RoB trials. The conventional meta-analysis of low RoB trials showed no statistically significant difference on mortality with lower vs higher fluid volumes (fixed effect model; relative risk, 0.99; 97% CI, 0.89-1.10; P = .69; I² = 0%). B, Trial sequential analysis (TSA) for all-cause mortality in eight low RoB trials. We used a control event proportion of 31.1%, α of 3.3% (two-sided), β of 10% (power 90%), and an a priori relative risk reduction of 15% in the analysis. The TSA-adjusted CI in the fixed effect model was 0.89 to 1.11 with a diversity D² of 0%. The blue cumulative z curve crossed the area of futility. Therefore, the TSA is conclusive, and a relative risk reduction of 15% is unlikely. A total of 83% (n = 3,626) of the required information size of 4,380 patients was accrued. MH = Mantel-Haaenszel; RRI = relative risk increase; RRR = relative risk reduction.