Su1912 Risk Factors and Outcomes of Upper Gastrointestinal Bleeding in Severely Burned Patients

Abstract

G A A b st ra ct s for in-patient morality but there are no studies looking at 30-day mortality or need-forintervention (blood transfusion, endoscopic or surgical intervention to stoop bleeding). We sought to determine prognostic accuracy AIMS65 in predicting short-term mortality and need-for-intervention. Methods: All adult (18 years or older) patients admitted to medical service for UGIB during October 2005 to September 2011 were identified from International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9CM) discharge database. Four components of AIMS65 score were pooled with help of medical informatics department and altered mental status was extracted by physician researchers; internal medicine residents. Altered mental status was defined as Glasgow Coma Scale score of less than 14 or mention of disoriented, lethargy, stupor, or coma in physician admission note. Main outcomes measured were all cause short-term mortality (30-day and 90day from hospital admission) and need-for-intervention. Need-for-intervention was identified using ICD-9CM procedure coding, 44.43 (endoscopic control of gastric or duodenal bleeding), 44.44 (transcatheter embolization for gastric or duodenal bleeding), 44.49 (other control of hemorrhage of stomach or duodenum) or 99.04(packed cell transfusion). Receiver operating curve statistics was used to determine predictive performance. Results: Over period of 6 years, a total of 802 admissions were identified with mean (SD) age of 69.5 (18.2) years, 59.2% were over 65 years of age, 392 (48.9%) were male. Total of 42 patients (5.2%) expired during index hospitalization or were sent to hospice and 54 (6.7%) expired within 30-day and 80 (10%) within 90-day from admission. 40 (5.0%) patients required endoscopic or surgical intervention to stop bleeding and 274 (34.2%) required blood transfusion. The area under receiver operating curve (AUC) was 0.78 (95% CI; 0.72-0.85) for in-hospital, 0.77 (95% CI, 0.71-0.83) for 30-day, and 0.79 (95%CI, 0.74-0.84) for 90-day mortality. The AUC for need-for-intervention was 0.53 (95%CI, 0.48-0.57) (Figure 1). Conclusion: The AIMS65 score provides accurate estimation of short-term mortality but does not estimate need-for-intervention in acute UGIB related hospitalizations.