Screening for Free: A Value-Based Insurance Design Natural Experiment

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Aligning health insurance benefits with clinical goals, aka value-based insurance design (VBID), holds much promise for reducing financial barriers to recommended care, improving quality, and modulating spending growth. While studies have found that patient cost-sharing can reduce the use of necessary care and increase spending for downstream care such as hospitalizations, there have been few rigorous VBID studies, especially of the effects on overall clinical or spending outcomes. The Patient Protection and Affordable Care Act (ACA) requires that all Medicare beneficiaries have free access to preventive services starting in 2011, thus creating a VBID natural experiment. This elimination of cost-sharing has the potential to mitigate underuse of cancer screening tests recommended by the US Preventive Services Task Force (USPSF). Using information from large Medicare Advantage plans, this project will examine the clinical and economic effects of this ACA- mandated Medicare policy change. Our 2007-2014 data contain detailed, comprehensive information on insurance benefits, individual characteristics, testing, clinical events, and medical spending. The study sample includes beneficiaries with a range of pre-policy cost-sharing amounts with the staggered elimination of cost- sharing (free screening) over time. For example, 46% of the study population had free colorectal cancer screening in 2007, 74% in 2010, and 100% in 2011; all benefit changes occur on January 1st of the year. The three aims examine the cost-sharing changes' direct and overall effects: Aim 1) screening test rates; Aim 2) clinical event rates; and Aim 3) total medical spending. We will evaluate the outcomes in all subjects and illustrative subgroups, e.g., subjects for whom screening may be of high or low value as per the USPSTF recommendations. Our primary analyses will focus on within-person changes (fixed effects estimation), and will adjust for a rich set of demographic, clinical, physician, and system characteristics. We also will compare subjects with cost-sharing versus those without cost-sharing, using as concurrent controls subjects who consistently have free screening throughout the study period. With several million person-years of data, we will have adequate power to detect even small changes in our outcomes. This study provides an opportunity to evaluate the largest VBID natural experiment to date, to examine the early effects of a major ACA provision on access to preventive care, and to inform future Medicare policy decisions.