Fig. 6

Trial sequential analysis for the overt GI bleeding. Trial sequential analysis using random-effects model with a relative risk reduction of 20%, an adjusted type I error of 3.3%, power of 80%. (panel a) In trials did not receive EN, control event proportion of 21.9%, D2 of 52%, the cumulative Z-curve cross cross the trial sequential monitoring boundary for benefit, the required information size of 6029 patients are not reached. The TSA-adjusted 95% CI for an RR of 0.37 is 0.22 to 0.63. (panel b) In trials with low risk of bias, control event proportion of 8.6%, D2 of 25% (the actual measured D2 was 0%), the cumulative Z-curve cross no boundaries, the required information size of 11,384 patients are not reached. The TSA-adjusted 95% CI for an RR of 0.62 is 0.37 to 1.04. (panel c) In trials with use of EN, control event proportion of 9.2%, D2 of 71%, the cumulative Z-curve cross no boundaries, the required information size of 27,681 patients are not reached. The TSA-adjusted 95% CI for an RR of 0.64 is 0.12 to 3.35. SUP stress ulcer prophylaxis; GI gastrointestinal; TSA trial sequential analysis; EN enteral nutrition