UNCERTAINTY, GAINS, LOSSES, PROBABILITY… WHAT IS MORE IMPORTANT IN TREATMENT PREFERENCES?

Purpose: Two treatment preference measures were designed with behavioral economic principles in mind; with a goal of studying whether variations in the symptoms alleviated (gains), side effects (losses), overall chances of these (probability of joint receipt of gains and losses), or uncertainty was the most influential on a patient’s responses.

Method: Participants were 55 male veterans with prostate cancer receiving care from the urinary and cancer clinics at Edwards Hines Jr. VAMC. Two preference measures featuring hypothetical health treatments were given to all participants. Each measure varied the gains, losses, probability, and uncertainty in alleviations and chances. The two preference measures were the Preference Indicator for Cancer Treatments with Uncertain and Risky Elements (PICTURE), and the Standard Gamble (SG). Conjoint analysis was used to extract the part-worth utility and “importance” of each experimentally controlled feature of six treatment decisions presented to each veteran. Choices from the PICTURE were converted into ranks, and used a nonmetric model. Preferences from the SG were analyzed with a metric model. 

Result: Analyses were conducted within each uncertainty type and preference measure, and at group- and individual-levels. Higher importances indicated which treatment features seemed to induce the most variation in responses. There was variability in importances of the treatment features; between preference measures and over all participants. The majority of veterans followed the same patterns in individual importances. Part-worth utility values indicated details of treatment feature preferences.

Conclusion: Gains, losses, probability, and uncertainty were emphasized differently across the preference measures, and across different types of uncertainty. With the PICTURE preference measure, the probability of jointly receiving alleviated symptoms and side effects, then symptom alleviation, appear to direct the treatment decisions. When using the Standard Gamble preference measure, the probability is more influential than side effects, on decisions between certain treatments, and between treatments with uncertain improvements. In treatments with uncertain chances, surprisingly, probabilities are not influential and preferences vary only with side effects and then improvements. The PICTURE is more stable in preference elicitation when different types of uncertainty are introduced in the treatment description. SG responses change greatly with uncertain information. Information presentation in treatment options can affect individual- and group-level treatment decisions.