Establishment of Prediction Model for Psoriatic Arthritis Based on Logistic Regression

Abstract

Background: Psoriatic arthritis (PsA) is an inflammatory musculoskeletal disease associated with psoriasis (PsO) that appears approximately 7-8 years after psoriasis and remains undiagnosed in most of patients, which is disabled and difficult to treat. So early risk assessment is particularly important. However, at present, the risk assessment of the status quo is not optimistic. The aim of this study was to establish and validate a prediction model for assessing the risk of PsA in psoriasis patients. Methods: Ambispective cohort study was conducted. Demographic and clinical records were collected and followed up and blindly reviewed. Single and multiple factors analysis were used to analyze the factors influencing PsA, and Logistic regression equation models were developed to verify their predictive value. Results: Among 1141 Pso patients, 221 patients had PsA. Logistic regression equation model showed that Arthralgia (OR= 438.079; 95% CI: 175.154-1095.682), nail dystrophy (OR= 7.453; 95% CI: 4.116-13.498), scalp lesions (OR= 3.047; 95% CI: 1.297-7.159), Inverse (intertriginous) lesions (OR= 2.45; 95% CI: 1.382-4.342) and BMI (OR= 0.909; 95% CI: 0.846-0.977) were identified as potential predictors affecting the risk of transition from PsO to PsA. And the receiver operating characteristic (ROC) curve of PsA was drawn, the area under the curve (AUC) was 0.974 (95% CI 0.963 - 0.985), the prediction sensitivity was 96.40%, the specificity was 92.10%, the calibration curve showed that the predicted results was in good agreement with the observed results. Conclusion: The independent predictors of PsA involve Arthralgia, nail dystrophy, scalp lesions, Inverse (intertriginous) lesions, BMI. The Logistic prediction model based on these predictors has reliable predictive value and can help early clinical assessment and treatment.