Variables associated with joint involvement and development of a prediction rule for arthritis in psoriasis patients. An analysis of the Italian PsoReal database.

BACKGROUND: Limited data exist to predict the development of psoriatic arthritis (PsA) in psoriasis (PsO) patients OBJECTIVE: To analyze factors associated with incident PsA in PsO patients, and to develop a predictive algorithm for progression to arthritis using a full set of variables and a restricted one applicable to administrative data.

METHODS: Cohort study within the PsoReal registry in Italy. Multivariable generalized linear models were used to assess factors associated with PsA and to derive a predictive model.

RESULTS: Among 8895 patients, 226 PsA cases were identified (incidence 1.9 per 100 patient-years). Independent predictors in the full model were: female sex, age 40-59 years, BMI≥25, chronic-plaque PsO features, presence of palmoplantar pustulosis, hospitalization for PsO in the last 5 years, and previous use of systemic PsO therapy, area under the curve of Receiver Operating Characteristics curve (AUC-ROC) = 0.74. Female sex, age 40-59 years, hospitalization for PsO, previous use of systemic PsO therapy were independent predictors in the restricted model, AUC-ROC = 0.72.

LIMITATIONS: Lack of other potential predictors for PsA.

CONCLUSION: Our models could be used by clinicians and health authorities when planning intervention and population surveillance. Future studies should confirm our models using larger datasets and additional variables.