Poulos C, Ng X, Mansfield C, Rosenberg S, Peay H, Lee J, Irony T, Ho M. Logical soundness in a discrete choice experiment study of benefit risk preferences. Presented at the ISPOR 2022 Conference; May 15, 2022. Washington, DC. [abstract] Value Health. 2022 Jun; 25(6 S1).

OBJECTIVES: A recent FDA-sponsored PPI study of knee osteoarthritis (KOA) treatments incorporated assessments of comprehension and response consistency to evaluate the logical soundness of the data.

METHODS: US adults with KOA completed an online discrete-choice experiment survey. In a series of treatment-choice questions, patients chose their preferred KOA treatment. Each treatment was characterized by four attributes with varying levels: improvement in pain, improvement in function, duration of treatment effect, and risk of tissue overgrowth within the knee joint. Five questions assessed comprehension of attribute descriptions, choice tasks, and the risk communication; two dominated choice tasks (one weakly dominated and one strongly dominated); and scope tests. Random parameters logit (RPL) analysis was conducted to compare subgroups that correctly answered all comprehension questions vs. those who answered one or more incorrectly.

RESULTS: Of the 250 survey respondents, the majority (ranging from 58.4% to 95.0%) answered each of the 5 comprehension questions correctly. The two risk comprehension questions were most likely to be answered correctly (94.1% and 95.0%, respectively). Overall, 41.9% of the respondents answered all five comprehension questions correctly; 38.4% answered only one question incorrectly; and 14.7% answered two or more incorrectly. RPL results suggest preferences did not vary systematically among those who answered all correctly vs. those who did not. More respondents selected the superior treatment in the strongly dominated choice task (96.3%) than the weakly dominated (86.9%). As expected, preference weights indicated that better clinical outcomes were preferred to worse outcomes. Results for the scope tests were mixed and warrant further investigation.

CONCLUSIONS: The multiple assessments of logical soundness in this study compare favorably with other PPI studies and suggest that KOA patients understood the survey and were attentive to the choice tasks. The study demonstrated how logical soundness tests can be used and interpreted in a PPI study.

Share on: