The Prevalence, Correlates, and Impact of Logically Inconsistent Preferences in Utility Assessments for Joint Health States in Prostate Cancer

Abstract

Variations in health state utilities can impact cost-effectiveness analyses. One potential source of error is when joint health state (JS) utilities are rated higher than the embedded single state (SS) utilities. Knowing when and in whom this occurs can improve cost-effectiveness analyses. Men (n = 323) were surveyed at the time of prostate biopsy. Time tradeoff SS and JS utilities for prevalent prostate cancer (PCa) health states were collected. JS utilities assessed included those most prevalent for PCa. "Inconsistency" was defined in the following 3 ways: (1) any size rank order violation; (2) total number of violations; and (3) differences greater than 1 standard deviation (SD). Regression analysis assessed independent patient characteristics associated with inconsistent responses. Aggregate JS utilities were consistent. At the individual level, 36% to 41% of responses violated rank order and 12% to 14% were larger than 1 SD. In all, 69% of respondents had at least 1 JS inconsistency, and 24% had >1 SD inconsistencies. Being married and feeling anxious were independently correlated with giving all types of inconsistent ratings, and lower education correlated with making >SD errors. SS utilities, and not JS utilities, were significantly lower for the inconsistent group. "Correcting" JS inconsistencies decreased aggregate utilities 1 to 9 units. Inconsistent JS utilities for PCa are prevalent in men at biopsy. Being married, more anxious, and having less education are correlated with inconsistencies. It is the SS utilities, rather than the JS utilities, that differ between consistent and inconsistent raters. Better understanding of the source of these inconsistencies is needed.