Gastrointestinal bleeding (GIB) in the setting of thrombocytopenia raises concerns about endoscopic procedure risk. We aimed to assess the safety and outcomes of endoscopy for overt GIB in the setting of severe thrombocytopenia in liver cirrhosis (LC) and non-liver cirrhosis (NLC).

This is a retrospective study on inpatients who underwent endoscopy within 24 hours of presentation for overt GIB with a platelet count (PC) of 20 to <50 × 10³/mL. Outcomes included diagnostic and therapeutic yields, procedural adverse events, packed red blood cell (pRBC) and platelet transfusions, recurrent bleeding rate, and all-cause and GIB-related mortality.

One hundred forty-four patients were identified. The median PC was 41 × 10³/mL and 61% had LC. The diagnostic yield was 68% (LC = 61%, NLC = 79%, P = .04). Therapeutic yield was 60% (59% vs 60%, P = 1.00). The initial hemostasis rate was 94% with one adverse event. The median number of pRBC and platelet transfusions decreased after intervention in the entire cohort. Recurrent bleeding rates were 22% at 1 month and 30% at 1 year, with no differences between groups. An increased international normalized ratio (INR) >2 was a predictor of recurrent bleeding. All-cause mortality was 19% at 1 month and 37% at 1 year, whereas GIB-associated mortality in our cohort was only 3% at 1 month and 4% at 1 year, with no significant difference between LC and NLC. Predictors of mortality were INR >2, activated partial thromboplastin time >38 seconds, hypotension, intensive care unit admission, and pulmonary comorbidities.

In this study cohort, we observed that endoscopy for overt GIB in the setting of severe thrombocytopenia in patients with LC and NLC appears safe, has moderate diagnostic and therapeutic yields with high initial hemostasis rate, and is associated with a significant decrease in pRBC and platelet transfusions. Recurrent bleeding and all-cause mortality rates remain high.