Purpose: We aimed to analyze the body composition characteristics of gallstone disease (GD) patients with bioelectrical impedance analysis (BIA) and to construct a nomogram to predict GD based on body composition.
Methods: Patients with or without symptomatic cholecystolithiasis or choledocholithiasis diagnosed in Inner Mongolia People’s Hospital from July 2020 to December 2021 were selected as the case group, and healthy subjects during the same period were selected as the control group. The body composition of the two groups was determined by BIA. The risk predictors for GD were extracted to construct a nomogram based on regression analysis. ROC curves were used to evaluate the predictive power of the nomogram, and calibration curves were drawn to evaluate the consistency of the model. The bootstrap method was used to verify the model and evaluate the generalizability of the model.
Results: A total of 1000 individuals were recruited for the study, including 500 GD cases and 500 controls, to evaluate body composition. Multivariate logistic regression analysis showed that sex (OR = 2.292, 95% CI: 1.436– 3.660), BMI (OR = 1.828, 95% CI: 1.738– 1.929), body fat percentage (BFP) (OR = 1.904, 95% CI: 1.811– 2.205) and waist circumference (WC) (OR = 1.934, 95% CI: 1.899– 1.972) were risk predictors of GD. The AUC was 0.770 (95% CI: 0.741– 0.799). The calibration curve showed that the C-index was 0.767. The prediction model was validated internally with 1000 bootstrap resamples. The accurate value was 0.72, and the kappa value was 0.43. All of the indices indicated that the model was well constructed and could be used to predict the incidence of GD.
Conclusion: A nomogram model of gallstone disease based on sex, BMI, BFP and WC was constructed with good discrimination, calibration and generalizability and can be used for the noninvasive and convenient prediction of gallstone disease in the general population.
Keywords: gallstone disease, cholelithiasis, bioelectrical impedance analysis, BIA, body composition, nomogram, prediction model