Novel and Simple Criteria for Predicting Mortality of Peptic Ulcer Disease

Objective Conventional risk scores of peptic ulcer disease (PUD) are based on many parameters, and their application in clinical practice is therefore limited. The aim of this study was to establish simple and reliable criteria for predicting PUD-associated mortality. Methods A total of 499 patients with PUD were divided into 2 groups: the training cohort (n=333) and the validation cohort (n=166). To minimize selection bias due to missing values, we used imputed datasets generated by the multiple imputation method (training-cohort dataset, n=33,300; validation-cohort dataset, n=16,600). Results In the training-cohort dataset, the heart rate-to-systolic blood pressure ratio (HR/SBP) and serum albumin (s-Alb) level were significant independent predictive factors for mortality according to the multivariate analysis (HR/SBP, odds ratio [OR]: 1.72; 95% confidence interval [CI], 1.06-2.80, p=0.028; s-Alb, OR: 0.23, 95% CI, 0.11-0.51, p<0.001). The model comprising HR/SBP and s-Alb was able to detect mortality due to PUD with an area under the curve (AUC) of 0.855. In the validation-cohort dataset, this model also showed good efficacy with an AUC of 0.835. The novel criteria combining HR/SBP and s-Alb developed by a decision tree analysis showed 73.3% sensitivity and 87.6% specificity for predicting mortality in the total-cohort dataset. Our criteria were superior to the Glasgow Blatchford and Rockall scores and similar to the AIMS65 and Progetto Nazionale Emorragia Digestiva scores for predicting mortality. Conclusion The combination of the HR/SBP ratio and s-Alb level is a good predictor of mortality in patients with PUD.