It is unclear what role platelet transfusion plays in thrombocytopenic patients with gastrointestinal bleeding (GIB). The study explored the effects of platelet transfusion on clinical outcomes in thrombocytopenic GIB patients who were critically ill.

Using the MIMIC-IV database, we conducted a retrospective cohort study in which patients with GIB and a platelet count < 150 × 10 ⁹ /L were identified and stratified into platelet transfusion and non-transfusion groups. Cox regression models, Kaplan–Meier analysis, and propensity score matching (PSM) were applied to study the association between platelet transfusion and mortality over both 28-day (short-term) and 1-year (long-term) periods.

The study included 1275 patients in total. Extended Cox proportional hazards regression analysis demonstrated a significant association between platelet transfusion and increased short-term mortality (HR 1.56; 95 % CI 1.06–2.30), whereas no such association was found for long-term mortality (HR 1.15; 95 % CI 0.89–1.50). PSM resulted in 195 cases being successfully matched in each group. A significant relationship was consistently found between platelet transfusion and short-term mortality (HR 1.74; 95 % CI 1.08–2.79), but not with long-term mortality (HR 1.19; 95 % CI 0.86–1.64). The association was stable in the sensitivity analyses, including residual confounding adjustment, IPTW, and using 14-day and 90-day mortality as surrogates for short-term and long-term mortality, respectively. However, as with any observational study, residual confounding may persist despite rigorous matching and sensitivity analyses.

Despite potential residual confounding, this retrospective cohort study of critically ill patients with thrombocytopenic GIB found that platelet transfusion was associated with higher short-term mortality but did not affect long-term mortality. These findings suggest a cautious approach to platelet transfusion in this population, though they should be validated in larger prospective investigations.