ICU Mortality Associated With Clinically Important Upper Gastrointestinal Bleeding: A Multistate Analysis.
Wang Yang Y, Heels-Ansdell Diane D, Parpia Sameer S, Cook Deborah D et al.
Clinically important upper gastrointestinal bleeding may increase mortality and time to discharge from the ICU. Conventional analyses ignore timing and treat bleeding as a binary event. We evaluated the association between time-dependent clinically important upper gastrointestinal bleeding and ICU mortality. A multicenter cohort study using data from an international stress ulcer prophylaxis trial. A total of 4821 invasively ventilated adult participants. None. A multistate model was specified with two transient ICU states (bleed-free and post-bleed) and four absorbing states (ICU discharge or death, each with or without bleeding). The Aalen-Johansen method was used to estimate transition and state occupancy probabilities and to describe risks of different clinical courses. Follow-up was censored at 90 days. We fit two Cox models for time to death in the ICU regressing on time-dependent bleeding history while adjusting for age, sex, baseline Acute Physiology and Chronic Health Evaluation II score, admission category (medical/surgical/trauma), vasopressor/inotrope use and renal replacement therapy on study day 1 and randomized allocation of pantoprazole or placebo; a second model also adjusted for time-dependent covariates of advanced life support (invasive mechanical ventilation, vasopressor/inotrope use, and renal replacement therapy). Clinically important upper gastrointestinal bleeding developed in 104 patients (2.2%) in the ICU. Median ICU length of stay for those who experienced bleeding was 19 days (interquartile range [IQR], 11-30 d) vs. 9 days (IQR, 5-15 d) for those who did not. Accounting for immortal time bias, the time-dependent indicator of clinically important upper gastrointestinal bleeding was associated with increased ICU mortality (hazard ratio, 1.56; 95% CI, 1.13-2.15). At any time, patients who experienced bleeding had approximately 1.6 times the hazard of ICU mortality compared with those without a history of bleeding. Clinically important upper gastrointestinal bleeding as a time-dependent event is associated with an increased risk of ICU mortality.