Risk factors for patient-important upper gastrointestinal bleeding: A protocol and statistical analysis plan for a secondary analysis of the REVISE (Re-Evaluating the Inhibition of Stress Erosions) trial

Background: While studies have reported risk factors for clinically-important upper gastrointestinal bleeding among critically ill patients, the risk factors for patient-important upper gastrointestinal bleeding, as defined by ICU survivors and family members, are unknown. Further, although trials show that stress ulcer prophylaxis with pantoprazole decreases the risk of upper gastrointestinal bleeding, some research suggests that mortality may be increased among patients with high illness severity who are exposed to pantoprazole. The risk of death during critical illness is greater than the risk of upper gastrointestinal bleeding.Objective: To identify risk factors for patient-important upper gastrointestinal bleeding among invasively ventilated critically ill adults, taking into account illness severity and the competing risk of death.Materials and Methods:  This pre-planned secondary analysis of the REVISE trial database will be guided by this protocol. We will use Cox proportional hazards regression analysis to assess the effect of candidate risk factors on the hazard rate of the primary outcome of time to patient-important upper gastrointestinal bleeding, considering the competing risk of death. We propose 2 sensitivity analyses.  Secondary analyses will evaluate risk factors for clinically-important upper gastrointestinal bleeding, and another model will evaluate whether enteral nutrition attenuates the effect of pantoprazole on bleeding prevention compared to placebo by including an interaction term between pantoprazole vs. placebo and amount of enteral nutrition received.Results: This study will identify conditions of critical illness which confer an increased risk of patient-important upper gastrointestinal bleeding in critically ill patients, and factors that decrease the risk.Conclusions: The findings may inform bedside care, practice guidelines, and the design of future studies.