Background:

Upper gastrointestinal hemorrhage (UGIH) is a common reason for emergency department (ED) evaluation and hospital admission. Despite validated risk‐assessment methods, it is not known how risk stratification is used to influence early endoscopy (EGD) and subsequent triage. We evaluated risk‐stratified clinical outcomes, resource utilization, and time of presentation for UGIH patients over 2 years.

Methods:

Analyses included 417 consecutive UGIH patients evaluated in the ED and admitted to 6 academic medical centers that did not have open‐access endoscopy services available to the ED. Outcomes included time to EGD, inpatient death, 30‐day readmission, and in‐hospital complications related to UGIH. Resource utilization was assessed by hospital length of stay (LOS) and costs. Data were obtained from chart abstraction and administrative files.

Results:

Mean age of subjects was 59.3 years; 60% were male, 36% were white, and 41% had Medicare. The most common diagnoses were erosive disease (54%), PUD (48%), and varices (15%). Mean LOS was 4.8 days, with a mean hospital cost of $12,052. Overall, 72% of patients presented to the ED during weekdays, of which 29% presented between 8 am and noon, 43% between midnight and noon, and 52% between 8 am and 4 pm; low‐risk patients were more likely to present after hours (P = .04). Based on a validated rebleeding risk score (Rockall), 13.0% were low‐risk pre‐EGD and 42.7% were low‐risk post‐EGD. EGDs were performed during business hours in 85.6% of patients, yet only 23.5% received EGD within 12 hours, and only 48% had EGD within 24 hours, with no difference by risk score. Hospital LOS was more than 2 days shorter if EGD was performed within 1 day of admission (4.06 vs. 6.74 days; P < .001) but was not related to risk score. For clinical outcomes, 1.2% died in the hospital (0.0% low risk vs. 2.1% high risk; P = .05), 10.3% experienced a complication (5.1% low risk vs. 14.2% moderate/high risk; P < .01), and 5% were readmitted within 30 days (3.4% low risk vs. 6.3% high risk; P = .18). Discharge within 1 day of EGD did not lead to higher readmission rate, regardless of rebleed risk.

Conclusions:

UGIH risk scores did not predict timing of EGD or early discharge post‐EGD, suggesting patients were not appropriately risk‐stratified. In addition, 13% of patients were considered low‐risk pre‐EGD, for which endoscopy is not necessary, yet still received EGD and admission. Our findings suggest that 30%‐50% of UGIH patients present before midday during the week, making same‐day open‐access EGD feasible. Of these, 30%‐40% could potentially be discharged from the ED or after 24‐hour observation. Emergency departments and hospitalists should work with GI services to develop innovative strategies to maximize the availability of endoscopy services for UGIH. Early EGD and appropriate triage are factors critical to reducing resource utilization while maintaining excellent clinical outcomes for UGIH.

Author Disclosure:

P. J. Kaboli, None; C. C. Wyatt, None; A. D. Auerbach, None; J. L. Schnipper, None; T. B. Wetterneck, None; D. O. Meltzer, None.