Utility of RHEMITT Score in a retrospective cohort with overt small bowel bleeding

Introduction The RHEMITT score (RS) has showed an excellent accuracy to predict rebleeding after video capsule endoscopy (VCE), in patients with small bowel bleeding (SBB). However, RS was developed and subsequently validated in a heterogeneous population of patients with overt and occult SBB. As the bleeding presentation is a recognised independent predictor for rebleeding, the aim of this study was to assess the RS utility in a homogeneous cohort of patients admitted with only overt SBB. Methods Retrospective, single centre study, including patients admitted with overt SBB. RS was calculated and the rebleeding rates compared. The follow-up was 12 months. Univariate analysis used simple logistic regression, chi-squared test or Mann-Whitney as needed. To test the performance of the score, ROC analysis was performed selecting the optimal cut-off point maximized by the Youden index. A rebleeding-free survival was also assessed. Results 79 patients were included, 62% showed relevant findings on VCE. 27.8% rebled, of which 63.3% rebled within the first 6 months. The rate of rebleeding in low-risk RS category was 10.5%, in the intermediate-risk 30.2% and in the high-risk 57.1%. Urea (p=0.008), relevant findings on VCE (p=0.021) and high-risk RS category (p=0.048) were predictive factors for rebleeding. The area under the curve for RS was 0.665 (95% CI 0.773 - 0.526). Kaplan-Meier plots weren't statistically different according to the RS categories (log-rank p=0.113). Conclusion High-risk RS performed well in predicting rebleeding in this cohort of patients with overt SBB. However, in low and intermediate RS categories further research is warranted.