Nomogram for predicting reflux esophagitis with routine metabolic parameters: a retrospective study.

Abstract

The prevalence of reflux esophagitis (RE) is relatively high around the world. We investigated routine metabolic parameters for associations with RE prevalence and severity, creating a user-friendly RE prediction nomogram. We included 10,881 individuals who had upper gastrointestinal endoscopy at a hospital. We employed univariate and multivariate logistic regression for independent risk factors related to RE prevalence, and conducted ordinal logistic regression for independent prognostic factors of RE severity. Subsequently, a nomogram was constructed using multivariate logistic regression analysis, and its performance was assessed through the utilization of receiver operating characteristic (ROC) curves, calibration curves, decision curve analysis (DCA), and clinical impact curve (CIC) analysis. In this study, 43.8% (4769 individuals) had confirmed RE. Multivariate analysis identified BMI, age, alcohol use, diabetes, Helicobacter pylori, systolic blood pressure (SBP), diastolic blood pressure (DBP), glucose, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), total cholesterol (TC), albumin, uric acid (UA), fT3, and fT4 as independent RE risk factors (p < 0.05). The personalized nomogram used 17 factors to predict RE, with an AUC of 0.921 (95% CI: 0.916-0.926), specificity 84.02%, sensitivity 84.86%, and accuracy 84.39%, reflecting excellent discrimination. Calibration, decision, and CIC analyses affirmed the model's high predictive accuracy and clinical utility. Additionally, ordinal logistic regression linked hypertension, diabetes, HDL-C, LDL-C, TG, and TC to RE severity. Our study highlights the association between the routine metabolic parameters and RE prevalence and severity. The nomogram may be of great value for the prediction of RE prevalence.