Background: Previously, we developed a prostate cancer (PC)-specific health state classification system, the Patient Oriented Prostate Utility Scale (PORPUS). In this study, we developed a scoring system to allow indirect calculation of utilities from the PORPUS.
Methods: We interviewed 234 PC outpatients, including those with newly diagnosed and metastatic disease, to obtain rating scale (RS) values on 4 to 6 levels of each of the 10 attributes of the PORPUS, and on 10 corner states (worst level on 1 attribute, best on 9). Patients also completed standard gamble (SG) and RS tasks on 4 multiattribute states (impotence and pain corner states, mild and severe PC symptoms). We used the RS and SG scores for multiattribute states to determine a risk aversion function for mapping values to utilities. We then tested 15 different strategies to estimate the multiattribute utility function (MAUF), using the single attribute disutilities for each level of the 10 PORPUS attributes, and the disutilities for the corner states. The root mean squared error (RMSE) of prediction of the SG on the 4 multiattribute states was used to identify the optimal strategy and scoring system.
Results: The optimal strategy gave an RMSE of 0.06. Comparison of mean MAUF-predicted utilities to directly elicited SG utilities for the 2 multiattribute states from patients in 2 previously published studies (n = 248 and n = 141) supported the validity of the MAUF.
Conclusions: The scoring system together with the PORPUS comprise an indirect utility instrument, the PORPUS-U, which can be used in clinical and research settings.