OBJECTIVES: The objective of this research was to predict the Psoriasis Area and Severity Index (PASI) for patients in the real-world dermatology setting using only physician global assessment (PGA) and body surface area (BSA).

METHODS: Patients from 6 specialty dermatology networks within the OMNY Health real-world data platform with a diagnosis code for psoriasis and with PASI, PGA, and BSA scores recorded on the same day from 2017 to 2023 were included. BSA, PGA, and/or their product (PGA x BSA) were considered as candidate predictor variables. The distribution of PASI scores in relation to the candidate predictor variables was examined, and multiple linear regression models were employed where the logarithm of the PASI score (with 0.5 continuity correction) was modeled as the outcome. Each combination of candidate predictor variables was attempted while employing leave-one-out cross validation. Model performance was assessed by root mean squared error (RMSE) and adjusted R squared (adjR2) values.

RESULTS: From the 298,326 psoriasis patients, 339 patients with 660 observations were included. The best performing model included BSA, PGA (categorical variable), and PGA x BSA (continuous variable) as predictor variables and yielded an RMSE of 0.59 and adjR2 of 0.78. For comparison, the worst performing model included only PGA x BSA (continuous variable) with an RMSE of 0.94 and adjR2 of 0.45. In the best performing model, BSA and PGA each had positive associations with the PASI outcome while PGA x BSA had a slightly negative association.

CONCLUSIONS: Administering the PASI tool in the real-world setting is time consuming, and development of an equation to predict PASI from BSA and PGA may offer an opportunity to measure psoriasis disease activity more efficiently in routine clinical practice. Additional analyses accounting for patient characteristics may be beneficial to develop a higher performing algorithm.