Developing a preference-based utility scoring algorithm for the Psoriasis Area Severity Index (PASI)

Introduction: It is challenging to identify health state utilities associated with psoriasis because generic preference-based measures may not capture the impact of dermatological symptoms. The Psoriasis Area Severity Index (PASI) is one of the most commonly used psoriasis rating scales in clinical trials. The purpose of this study was to develop a utility scoring algorithm for the PASI. Methods: Forty health states were developed based on PASI scores of 40 clinical trial patients. Health states were valued in time trade-off interviews with UK general population participants. Regression models were conducted to crosswalk from PASI scores to utilities (e.g. OLS linear, random effects, mean, robust, spline, quadratic). Results: A total of 245 participants completed utility interviews (51.4% female; mean age = 45.3 years). Models predicting utility based on the four PASI location scores (head, upper limbs, trunk, lower limbs) had better fit/accuracy (e.g. R², mean absolute error [MAE]) than models using the PASI total score. Head/upper limb scores were more strongly associated with utility than trunk/lower limb. The recommended model is the OLS linear model based on the four PASI location scores (R² = 0.13; MAE = 0.03). An alternative is recommended for situations when it is necessary to estimate utility based on the PASI total score. Conclusions: The derived scoring algorithm may be used to estimate utilities based on PASI scores of any treatment group with psoriasis. Because the PASI is commonly used in psoriasis clinical trials, this scoring algorithm greatly expands options for quantifying treatment outcomes in cost-effectiveness analyses of psoriasis therapies. Results indicate that psoriasis of the head/upper limbs could be more important than trunk/lower limbs, suggesting reconsideration of the standard PASI scoring approach.