Effects of Pantoprazole Based on Dialysis Status: A Protocol and Statistical Analysis Plan for a Secondary Analysis of the REVISE Trial

Background A recent international randomized trial found that the proton pump inhibitor (PPI) pantoprazole significantly reduced the risk of clinically important upper gastrointestinal (GI) bleeding among invasively ventilated patients in the intensive care unit (ICU), with no effect on mortality. Pantoprazole may be of particularly high utility among patients at high-risk of bleeding, such as those receiving renal replacement therapy (RRT) prior to ICU admission. However, given the complex mechanisms by which renal insufficiency may increase bleeding risk, as well as the possible association between PPI use and adverse kidney outcomes, investigating whether kidney failure treated with RRT at baseline modifies the treatment effect of PPIs is clinically relevant. Methods We have planned a post hoc secondary analysis of the Re-evaluating the Inhibition of Stress Erosions (REVISE) Trial ( NCT03374800 ). The primary objective if this study is to evaluate whether any RRT on study day 1 (e.g., maintenance RRT in the outpatient setting, or RRT initiated in the hospital [intermittent hemodialysis, continuous renal replacement therapy, sustained low efficiency dialysis, or peritoneal dialysis]) modifies the treatment effect of pantoprazole on clinically important upper GI bleeding and 90-day mortality. The secondary objective is to evaluate whether maintenance RRT (intermittent hemodialysis or peritoneal dialysis) in an outpatient setting before the index hospital admission modifies the effect of pantoprazole on clinically important upper GI bleeding and 90-day mortality. Secondary outcomes for both analyses are patient important upper GI bleeding, ICU and hospital mortality. We will perform Cox proportional-hazard analyses, including the RRT exposure of interest, randomized group, and their interaction for each outcome, reporting results as hazards ratios with 95% confidence intervals, and p-values for the interaction term. Conclusion Results of this study will inform critical care practice, particularly for the care of patients at risk of stress-related GI ulceration in the ICU who have received RRT.