Background: Unity in pursuit of the Triple Aim: better health, better care, and lower per capita cost, can be achieved through a well-designed health care delivery system. The accountable care organizations (ACOs) model is considered a key component of health care delivery system improvement because the model fosters better coordination of care through clinical integration and financial accountability. Within the six Centers for Medicaid & Medicare Services (CMS) ACO programs, the Medicare Shared Savings Program (MSSP) ACO has the largest size with a total of 432 ACOs formed; the service subjects of the MSSP ACO are the fee-for-service beneficiaries. Recently, academicians and researchers have been attracted to exploring ACOs' formation and performance. However, most of the early ACO research types are either descriptive or case study. Also, early researchers had limited access to ACO data sets, so they could utilize only regional and demographic factors to identify the predictors of ACO formation. Purpose: An integrative theoretical framework, Rogers' diffusion of innovation theory and Duncan's POET model, was used to examine ACO formation and performance. The first purpose of this study was to determine the relative influences of contextual variables and ACO characteristic variables on how early an ACO model was adopted. The second purpose was to examine how executives' perceptions of ACO performance and the ACO first-year performance are influenced by the contextual variables, ACO characteristic variables, and timing of the adoption of an ACO model. Methods: A cross-sectional design was formulated to gather data from a survey supplemented by secondary data with the analysis unit at the organization level. Study participants in the ACO survey included 2012, 2013, 2014, and 2015 ACO cohorts. Logistic regression was performed to examine the effects of POET and Rogers' five core characteristics in the early adoption of an ACO model (dichotomous). Additionally, multiple linear regression analysis was used to examine the effects of POET and the timing of adoption of an ACO model in the perceptions of ACO performance. ACO first-year performance dataset consisted only of ACO cohorts from 2012 through 2014. Finally, confirmatory factor analysis and structural equation modeling were conducted to examine the measurement model of the ACO first-year performance and a full latent variable model, respectively. Major Findings: A survey of ACO executives/managers between October 2015 and February 2016 was conducted. The 447 MSSP ACOs in my mailing list yielded a response rate of 13.65 % (n=61). Of the 61 MSSP ACOs, 42 (52.5%) were late adopters whose contractual agreement with CMS started in 2014 or 2015, and 36 (59.0%) were with hospital-based composition. Among ACOs that participated in my survey, their current degree of IT adoption in functionalities (62.27 vs. 52.50 points), usage levels (65.19 vs. 49.49 points), and integration levels (62.24 vs. 53.37 points) were better than their initial years. The multiple logistic regression presented that MSSP ACOs were more likely to be early adopters of a CMS if their service areas had high unemployment rates (OR=2.23; 95% CI: 1.13 - 4.39). In the multiple linear regression analysis, the executives in the early ACOs perceived their organizations as more effective than the late adopters, with 12.65 points higher in an aggregate of eight ACO quality domains (p = .005). Three hundred and seventeen MSSP ACOs, with contractual agreements with CMS before 2015, had retained their year-one performance records (the actual ACO performance with eight quality domains). The variability in the actual ACO performance was explained by the predictor variables of the study with an R-square of 15%. The actual ACO performance was likely to be improved if ACOs had more Medicare assigned beneficiaries or had the hospital-based composition. On the other hand, if ACOs' service areas were located in areas of high poverty concentration, a high unemployment rate, or a lower competitive index, their ACO performance was relatively lower than their counterparts. Implications: The findings suggest that managers should consider strategies to increase economies of scale in size and to have hospital involvement in their ACOs in order to increase effective management. Inadequate capital for information technology improvements is the biggest barrier inhibiting healthcare providers' willingness to join an ACO. Regardless of rural or urban areas, financial support is still important for those potential ACO participants who are planning to invest in necessary infrastructure. ACOs that involved hospitals also showed better performance than those ACOs without hospital involvement. This information may help health policy makers to define core principles of the best ACO model in the future. Conclusions: This study makes a unique contribution using a theoretically integrative framework with Rogers' diffusion of innovation theory coupled with Duncan's POET model to examine ACO formation and ACO performance. In the early ACO adopters, three-fifths of the ACOs had hospital involvement; and the levels of their current IT degree in functionalities, usage levels, and integration levels are higher than the late ACO adopters. This study demonstrates that contextual variables, such as unemployment rates at ACO service areas, relatively influence how early an ACO model was adopted. Executives in the early ACOs had higher perceptions of overall organizational effectiveness as compared with the late adopters. The first-year performance of 2012, 2013, and 2014 ACO cohorts is positively influenced by the size of assigned Medicare beneficiaries and hospital-based ACO and is negatively influenced by the poverty rate, unemployment rate, and market competition scores (Herfindah-Hirschman Index).
Doctor of Philosophy (Ph.D.)
College of Health and Public Affairs
Public Affairs; Health Services Management and Research
Length of Campus-only Access
Doctoral Dissertation (Open Access)
Lin, Yi-Ling, "The Diffusion and Performance of the Accountable Care Organization Model" (2016). Electronic Theses and Dissertations. 5185.