Quantifying Pill Disutility Associated With Starting Versus Continuing Cardioprotective Medication: A Randomized Experiment.

Abstract

Background: Quantifying patient-reported pill disutility is important for understanding the risk-benefit tradeoffs of taking medications. The objective of this study was to quantify and compare the pill disutility associated with starting a new medication and continuing an existing medication for cardiometabolic disease prevention in a sample of older adults in the United States.

Methods: We enrolled adults aged ≥60 years from an online panel. Respondents completed a survey that included a 2-armed experiment that randomized them to either a starting or a continuing scenario in which they were instructed that their doctor recommended they start or continue, respectively, a daily medication that prevents heart attacks and strokes. Pill disutility was calculated using a time-tradeoff method with time willing to trade obtained via alternating dichotomous choice contingent valuation design. Pill disutility was described within each scenario overall and by subgroups and then compared across scenarios using the Kruskal-Wallis test and multivariable fractional logistic regression.

Results: A total of 621 respondents with a mean age of 69 years were included in the final analysis. A majority were taking medications (n=84.5%, n=525) and had at least 1 chronic cardiometabolic disease (78.7%, n=489). Pill disutility associated with starting a new medication was 0.0662 (SD, 0.13), while pill disutility associated with continuing an existing medication was 0.0378 (SD, 0.10; P<0.001). Participants randomized to the starting scenario had higher odds of higher pill disutility versus participants randomized to the continuing scenario in both multivariable testing (odds ratio, 1.66 [95% CI, 1.15-2.40]) and across subgroups.

Conclusions: Pill disutility for a daily cardioprotective medication, when obtained from a sample of older adults utilizing rigorous ascertainment methods, is higher than previously reported, especially with regard to starting the medication. These represent the first estimates that can be used in cost-effectiveness modeling involving both prescribing and deprescribing.