Using conjoint analysis to study health policy changes: An example from a cohort of persons who use drugs

Background: Historically, persons who use drugs (PWUDs) in the United States have often lacked health insurance, as Medicaid did not cover low-income adults. The Affordable Care Act of 2010 (ACA) increased access to insurance in states, such as Kentucky, that expanded their Medicaid programs. A cohort study of PWUDs in Kentucky found the prevalence of being insured increased from 34% pre-ACA to 87% post-ACA. However, changes to Medicaid were proposed that intended to restrict access to this program. This manuscript describes the feasibility and utility of conjoint analysis, an innovative method for studying decision-making, to identify potential impacts of health policy changes on PWUDs. Methods: IBM SPSS's “orthogonal design” command was used to construct 12 policy profiles that presented varying combinations of the proposed policy changes to Kentucky's Medicaid plan. Each policy profile presented information on (1) monthly premium costs, (2) penalties for not paying monthly premiums, (3) weekly work requirements, and (4) if their preferred physician accepted the plan for payment. Readability of the policy profiles was analyzed using the Readable.io application. The policy profiles were included in a recent follow-up of a longitudinal cohort of PWUDs in Appalachian Kentucky (n = 355). Participants rated the likelihood of enrolling in each policy profile's Medicaid plan, using a scale ranging from 0=not at all likely to 10=extremely likely. Data were analyzed using SPSS's conjoint analysis commands. Results: Readability results indicated the policy profiles required a 3^rd grade education. Nearly all participants responded to each of the 12 policy profiles. Across the policy profiles, the mean response for willingness to enroll was 3.43 (SD = 3.61), indicating relatively low willingness to enroll. Conjoint analysis revealed the two most influential factors on willingness to enroll were work requirements (importance score, IS = 77.63) and monthly premium costs (IS = 17.76). Penalties for nonpayment (IS = 0.43) and physician acceptance (IS = 4.13) had minimal influence. Conclusions: This research demonstrates the feasibility of using conjoint analysis to study the impacts of potential policy changes on PWUDs.