Achieving the Triple Aim in Oregon: Effects of Coordinated Care Organization implementation on access, quality, and cost

In 2012, Oregon established Coordinated Care Organizations (CCOs) which function as community-based networks of providers and insurers, and which bear financial risk for enrolled Medicaid members.  To assess the effects of potential variation among CCOs, and between the CCO model and traditional models of health service delivery, a multi-disciplinary, mixed-methods research study was conducted. Study goals are to assess what CCOs in Oregon actually do; how they ultimately impact health care access, use, quality, costs, and health outcomes; and how CCO outcomes may be associated with specific organizational structures and operational approaches.

Leveraging the Oregon Health Study (OHS), a longitudinal panel design was employed to follow a cohort of Medicaid members through CCO implementation and evaluate the impact of CCOs on health care access, quality, patient engagement, health behaviors, and health outcomes. A baseline survey (n=9,350) of Medicaid members shortly prior to CCO implementation was completed and followed 18 months after CCO implementation (n=4,360 responses, 47% response rate). The follow-up surveys exited the field in February 2014; data entry and analysis will be completed by May 2014.  We will report outcomes for CCOs as a whole and variation in outcomes across CCO types. 

Claims data from the Medicaid components of Oregon’s APAC database was used to analyze utilization patterns and calculate per-member costs for all members before and after CCO implementation, and expenditure totals and patterns were analyzed to identify CCO-related changes in the probability of use, expenditures per user, and expenditures per person. Difference-in-difference and two-part model designs assessed net percentage changes in use and expenditures in total and for specific service categories.  Methodologies to define service categories (BETOS) and prices (CPT, HCPCS, or DRG) are discussed.  Analysis is ongoing and inlcudes comparison of overall CCO study members’ experience to a control group, as well as comparisons among CCOs. 

CCO's defining characteristics, similarities and differences within four domains (governance, organization, finance, and operations) were assessed through document reviews and in-depth, qualitative interviews (n=75). Preliminary findings indicate substantial variation among CCOs in all four domains. Linear regression analysis will be used to integrate CCO typologies encompassing each of the four domains to the results of survey and administrative data analyses. 

Because CCO designs in Oregon vary substantially, these early results will suggest to other states which designs are most likely to produce the desired effects.  As a result, states can choose model elements that are both transformative and cost-effective, accelerating the positive impacts of reform.