Construction and validation of predictive model for depression risk in patients with rheumatoid arthritis

Objectiv e: To analyse the factors influencing depression in rheumatoid arthritis patients and construct and validate a depression risk prediction model. Design : This study conducted a cross-sectional study Methods: The convenience sampling method was used to select rheumatoid arthritis patients admitted to the inpatient department of three tertiary hospitals in Jiangsu from January 2021 to January 2023 as the survey subjects. General information, disease-related data, and laboratory test indicators of patients were collected. The Hamilton Depression Scale and Health Assessment Questionnaire were applied to assess the patients' depression and functional limitation status. Univariate and logistic regression analyseswere conductedto identifythe factors that influencedepression. The R software was employed to developa nomogram prediction model and evaluate its predictive performance. Results : A total of 504 rheumatoid arthritis patients were included in this study, with the incidence of depression at 55.56%. According to the Logistic regression analysis results, educational level, disease activity, pain level, anti-CCP antibody, and Health Assessment Questionnaire score were the influencing factors for depression in rheumatoid arthritis patients. For this model, Deviance χ2 = 286.072, P > 0.999, Pearson χ2 = 412.053, P = 0.665, indicating the logistic regression model fits well with the data in this study. The area under the receiver operating characteristic curve is 0.944, with a 95% confidence interval (CI) of 0.925–0.963, P < 0.001. The results of ten-fold cross-validation show that the model's accuracy is 86.33%, sensitivity is 89.54%, and specificity is 81.75%, indicating the model has substantial predictive capability. Conclusion: The depression risk prediction model for rheumatoid arthritis patients developed in this study demonstrates high accuracy and discrimination, which supports clinical medical staff in screening high-risk patients for depression and offers a basis for early intervention. Clinical relevance : This study created a validated model to predict depression in RA patients. By integrating factors like pain, disease activity, and physical function, the model achieves 86.33% accuracy. Clinicians can use its nomogram for rapid risk assessment, enabling early support and improved quality of life through targeted interventions.