Comparison of the Risk Stratification Tools Including the Blatchford Score and the Clinical Rockall for Nonvariceal Upper Gastrointestinal Bleeding in the Emergency Department

Lee, Han You; Jeon, Woochan; Kim, Kyung Hwan; Park, Joon Min; Kim, Hyunjong; Park, Junseok; Shin, Dong Wun; Kim, Hoon

Abstract

PURPOSE
Upper gastrointestinal bleeding (UGIB) is a common medical emergency condition in the emergency department (ED). Patients with UGIB show a wide range of clinical severity, from mild bleeding to death. The objective of this study was to evaluate methods for risk stratification of active UGIB in the ED.
METHODS
We retrospectively reviewed patients with UGIB who were admitted to the ED of a tertiary care, university-affiliated hospital center from January 2011 to December 2011. Our study subjects were patients over 20 years old who complained of symptoms and signs of gastrointestinal tract bleeding and underwent endoscopic gastroduodenoscopy (EGD) evaluation. However, patients diagnosed with variceal gastrointestinal bleeding, disseminated malignancy, coagulopathy, and lower gastrointestinal bleeding and patients who did not undergo EGD within 6 hours were excluded. The Blatchford score and the clinical Rockall score were calculated for the enrolled patients. In cases where the value of each score was greater than 0, the scores were considered high risk. Active UGIB was defined as a symptom of patients who underwent emergency endoscopic intervention such as ligation or sclerotherapy. We compared the proportions of patients identified as high risk using chi tests. The areas under the receiver operating characteristic (AUROC) curve for detection of patients requiring emergency endoscopic intervention were calculated for both the Blatchford score and the clinical Rockall score.
RESULTS
The numbers of patients with high risk according to the Blatchford score and the clinical Rockall were 220 (93.6%) and 192 (81.7%) of 235 patients, respectively. The number of patients with active UGIB was 96 (40.9%) of 235 patients. The sensitivity and specificity of risk stratification based on the Blatchford score was 100% (96/96) and 10.8% (15/139) (p=0.001), respectively, while those based on the clinical Rockall score were 80.2% (77/96) and 17.3% (24/139) (p>0.05). The AUROC curves of the Blatchford score and the clinical Rockall score were 0.617 (95% CI; 0.546-0.688) and 0.495 (95% CI; 0.420-0.571), respectively.
CONCLUSION
The Blatchford score could be more useful as a risk stratification tool than the clinical Rockall score for active UGIB patients in the ED. The Blatchford score would be preferable as a clinical tool that can discriminate patients who need emergency endoscopic intervention for control of UGIB.