An in-depth analysis of mortality of hospitalized patients with suspected non-variceal upper gastrointestinal bleeding – data from a prospective registry

Objectives: Non-variceal upper gastrointestinal bleeding (nvUGIB) mortality has improved but remains high. Guidelines on nvUGIB often focusing on pre- and endoscopic management. However, more data on causes of death are needed, because non-bleeding specific mortality is unlikely to be improved by interventional advances.
Aims & Methods: The aims of this retrospective analysis of a single tertiary center registry of patients with nvUGIB were to assess causes of death in patients with nvUGIB and characterize this cohort. All hospitalized patients evaluated for suspected non-variceal UGIB (i.e., patients undergoing endoscopy) at our endoscopy unit between January 2018 and October 2020, either due to presentation for nvUGIB or with a suspected bleeding episode during hospitalization for any reason, were included in a prospective registry. Data on pre-endoscopic signs/symptoms of UGIB, epidemiological data, laboratory values, and medications were extracted from the electronic health record. Endoscopy reports were individually searched and relevant data including the most likely culprit lesion in the opinion of the endoscopist as well as respective hemostatic treatment extracted.
Results: Among 1135 patients – 41.7% women, mean age 70.4 (±15.7) years, median length of stay 8 (IQR 4;16) days – 92 (8.1%) died during hospitalization and were further analyzed. Mean age was 73.2 (±12.9) years, 44.5% were women, 36% of patients were on anticoagulation, 27% on antiplatelet therapy, and 9% on both. The median Charlson comorbidity index was 5 (IQR 3;6.8), the CHA2DS2-VASc-Score 3 (IQR 2;5). However, the median AIMS65 score was only 2 (IQR 1;2). Median time from admission to death was 17 (IQR 6.5;30) and from endoscopy to death 11 (IQR 4;21.5) days. The most common sources of bleeding as determined by endoscopy reports were: uncertain (24%), peptic ulcer (21%), esophagitis (14%), and non-peptic ulcers (9%). Adherence to guideline-based endoscopic peptic ulcer management recommendations was 70%. Radiological intervention for bleeding was performed in 1%, surgical intervention in 8.7%. Patients who died were older (73.2 vs. 70 years, p=0.0336), had lower hemoglobin nadirs (7.1 vs. 7.8 g/dL, p<0.0001), high c-reactive protein levels (p<0.0001) and creatinine (p<0.01), had longer lengths of stay (LOS) (median LOS 17 vs. 8 days, p<0.0001), and longer times from admission to endoscopy (median 3 days vs. 1 day, p<0.0001). However, sex was not associated with mortality (p=0.58). Looking at bleeding in association with causes of death, mostly (n=55, 59.8%), bleeding was categorized as a bystander. Only 32.6% of causes of death were judged as directly due to or likely associated with the bleeding episode, and these patients had a higher AIMS65 (p=0.0301). Only 12% died due to an unmanageable bleeding (or change to best supportive care). Overall, the following major groups of causes of death were identified: malignancy (18.5%), multi-organ failure (17.4%), infection/sepsis (15.2%), bleeding-associated (15.2%), cardiovascular (13.0%), and others (23.9%).
Conclusion: UGIB mortality remains relatively high despite management advances. However, only a minority of patients die due to UGIB and even among those most deaths are related to sequelae of the bleeding instead of unmanageable bleeding. Patients who die are older, likely sicker, and suffer from many comorbidities. Based on these data, it is uncertain whether focusing on endoscopic management improvements will significantly change UGIB-related mortality. Possibly, the focus should be shifted to improving general/internal medicine or geriatric care.