A Diagnostic Model for Gerd Phenotypes Based on Epidemiological and Clinical Data of the Russian Registry

Abstract

The aim of the study was to identify the epidemiological and clinical factors predisposing to non-erosive reflux disease (NERD), erosive esophagitis (EE) and Barrett’s esophagus in Russian population of patients, and to develop a diagnostic model for GERD phenotypes.
 Materials and methods. 1433 patients with GERD from Moscow, Moscow region, St. Petersburg and Leningrad region were studied. During the study complaints and anamnestic data were gathered, anthropometry, tests for H. pylori infection and upper endoscopy were performed. Spearman rank correlation was used to identify the correlation of clinical characteristics with NERD, EE and Barrett’s esophagus. 880 patients with GERD (407 with NERD, 443 with EE, 27 with Barrett’s esophagus, 3 with esophageal stricture) were selected for correlation analysis due to the presence of all necessary parameters. To develop multivariable model for GERD phenotype diagnosis (EE/NERD), multivariate analyses was used. The prognostic value of multivariable model was determined based on ROC analysis.
 Results. The prevalence of two main GERD phenotypes - EE and NERD - was approximately equal in the patients of Russian megapolises. Male gender was significantly negatively correlated with NERD development (p=0.002011). An inverse correlation was found between NERD and older age groups (p=0.002459), high waist circumference (WC) (p=0.003733), long duration of GERD symptoms (p=0.012984), the results of the GerdQ≥8 points (p=0.040030). In the group of patients with EE grade A-B, a significant inverse correlation was revealed with age over 40 years (p=0.002708), high body mass index (BMI) (p=0.020957), and high WC (p=0.001293). The male gender (p=0.000108) and the duration of GERD symptoms more than 3 years (p=0.028292) were significantly associated with severe grades of EE development. The male gender (p=0.007505) and not using of PPIs over the past 6 months (p=0.015930) were positively associated with Barrett's esophagus. Based on correlation analyses and predictive value 6 variables associated with EE/NERD were chosen: gender, age, GERDQ results, GERD symptoms duration, WC, BMI. Variables were divided into classes with similar risk with corresponding scores. The cut-off point (total score) with balance of sensitivity and specificity (Se≈Sp) was 549. Area under the curve (AUC) was 0,73, which indicates good quality of the diagnostic model. 
 Conclusion. Female gender, young age, normal BMI and WC, short duration of symptoms and GERD-Q less than 7 points were positively associated with NERD compared with EE among Russian patients. Male patients with long-term GERD symptoms, with overweight and obesity, who do not take PPIs are at high risk of Barrett's esophagus.