This study aimed to develop mapping algorithms from the Expanded Prostate Cancer Index Composite (EPIC) and the Short-Form (SF) Health Surveys to the Patient-Oriented Prostate Utility Scale (PORPUS), an econometric instrument specifically developed for patients with prostate cancer.
Data were drawn from 2 cohorts concurrently administering PORPUS, EPIC-50, and SF-36v2. The development cohort included patients who had received a diagnosis of localized or locally advanced prostate cancer from 2017 to 2019. The validation cohort included men who had received a diagnosis of localized prostate cancer from 2014 to 2016. Linear regression models were constructed with ln(1 − PORPUS utility) as the dependent variable and scores from the original and brief versions of the EPIC and SF as independent variables. The predictive capacity of mapping models constructed with all possible combinations of these 2 instruments was assessed through the proportion of variance explained (R ) and the agreement between predicted and observed values. Validation was based on the comparison between estimated and observed utility values in the validation cohort.
Models constructed with EPIC-50 with and without SF yielded the highest predictive capacity (R = 0.884, 0.871, and 0.842) in comparison with models constructed with EPIC-26 (R = 0.844, 0.827, and 0.776). The intraclass correlation coefficient was excellent in the 4 models (>0.9) with EPIC and SF. In the validation cohort, predicted PORPUS utilities were slightly higher than those observed, but differences were not statistically significant.
Mapping algorithms from both the original and the abbreviated versions of the EPIC and the SF Health Surveys allow estimating PORPUS utilities for economic evaluations with cost-utility analyses in patients with prostate cancer.