2039.  Incremental Value of Endoscopy in Identifying Low-Risk Persons with Acute Upper GI Hemorrhage (UGIH)
IM Gralnek, VA Greater Los Angeles Healthcare System, GS Dulai, VA Greater Los Angeles Healthcare System, TT Oei, Keck-USC School of Medicine, J Gornbein, UCLA School of Medicine

Objectives: Investigators have developed and validated an easy to use score (Blatchford Score, Lancet 2000; 356) for risk-stratifying individuals with acute UGIH. Like the clinical Rockall Score, the Blatchford Score uses only clinical and lab data.  Proponents of Blatchford Score suggest its use at the point of hospital admission to identify persons at low-risk for adverse outcomes (<5% rebleeding and <1% mortality) who could be managed as outpatients. From a healthcare policy perspective, this could increase efficiency of care and decrease resource utilization.  The complete Rockall Score, is a valid and well-accepted risk stratification tool that uses both clinical and endoscopic data in risk-stratifying individuals with acute UGIH. We hypothesized that as compared to the Blatchford Score, the complete Rockall Score would safely and correctly identify a greater percentage of low-risk individuals presenting with acute UGIH.

Methods: Using ICD-9 codes for discharge diagnosis,  we identified, at a university tertiary care hospital, a cohort of patients (n=175) with acute non-variceal UGIH. Medical record data were abstracted by two data abstractors blinded to study intent. Blatchford and Rockall scores were generated for each case.  Low-risk cases were defined as Blatchford Score=0, clinical Rockall Score=0, and complete Rockall Score <=2.

Results: We found that the Blatchford Score, stratified only 14/175 (8%) of the cohort with acute, non-variceal UGIH as "low-risk".  In comparison, we found that the clinical Rockall Score (pre-endoscopy) stratified 21/175 (12%) as low-risk and the complete Rockall Score (post-endoscopy) stratified 53/175 (30%) as low-risk for adverse outcomes. No patient stratifed as low-risk using the Blatchford Score rebled or died (100% negative predictive value). 2/53 (3.8%) low-risk patients by the complete Rockall Score had rebleeding, none died (96.2% NPV).

Conclusions: In this retrospective cohort study, the complete Rockall Score accurately and safely identified an additional n=39 (22%) low-risk patients with acute, non-variceal UGIH. These data suggest that endoscopy is an integral component in identifying low-risk individuals presenting with acute UGIH and has incremental clinical value. Moreover, the use of urgent endoscopy may improve efficiency of care and decrease healthcare resource utilization. We recommend the prospective comparison and evaluation of these risk stratification tools to confirm these findings.