Development of a Predictive Model for Screening Patients with Psoriasis at Increased Risk of Psoriatic Arthritis

INTRODUCTION: This study aimed to develop a predictive model based on ultrasound variables which can be used to screen patients with psoriasis who are prone to progress to psoriatic arthritis (PsA) in clinical practice. METHODS: This is a cross-sectional study conducted in a single center from October 2018 to November 2020. All subjects (non-PsA group, PsA group, and control group) underwent an ultrasound examination and their ultrasound abnormalities were recorded. On the basis of statistical analysis and clinical experts’ advice, several variables were selected for modelling. We used logistic regression to establish the prediction model. To assess the discrimination and accuracy of this model, internal validation and external validation were performed. RESULTS: A total of 852 patients with psoriasis but without PsA, 261 patients with PsA, and 86 healthy volunteers were included. Ultimately, the predictive model consisted of six variables, namely hand joint power Doppler (PD) signals (grade 0: OR 2.94, 95% CI 1.94–4.47; grade ≥ 1: OR 109.30, 95% CI 14.35–832.27; P < 0.001), wrist joint synovial thickening (grade 1: OR 1.29, 95% CI 0.69–2.43; grade 2: OR 4.30, 95% CI 1.92–9.65; grade 3: OR 11.05, 95% CI 1.01–120.64; P = 0.001), knee joint PD signals (grade 0: OR 1.01, 95% CI 0.56–1.80; grade ≥ 1: OR 14.77, 95% CI 3.99–54.69; P < 0.001), toe joint PD signals (grade 0: OR 1.18, 95% CI 0.78–1.79; grade ≥ 1: OR 5.74, 95% CI 2.84–11.63; P < 0.001), quadriceps tendon and patellar tendon enthesitis (OR 1.95, 95% CI 1.36–2.78, P < 0.001), Achilles tendon and plantar aponeurosis enthesitis (OR 1.63, 95% CI 1.14–2.32, P = 0.007). C-index for the predictive model was 0.80 (95% CI 0.76–0.83). After bootstrapping validation (1000 times), it was confirmed to be 0.79. The external validation showed the accuracy of the predictive model is 0.87 (95% CI 0.69–0.95). CONCLUSION: This study succeeded in developing a predictive model with a high degree of accuracy to predict the risk of PsA in patients with psoriasis.