Network influences on chronic illness care in large physician organizations: A study of the California managed care market in 2001 and 2006

Objectives: Care management processes (CMPs) are evidenced-based methods of physician practice redesign consistent with principles of the Chronic Care Model. CMP examples include the use of disease registries, physician practice guidelines, physician feedback on patient quality indicators, and case management. National surveys have found minimal use of these processes in large physician organizations. Physician groups enter into interorganizational relationships in order to achieve economic success, to deliver a broad range of professional services, and to maintain professional autonomy. Little is known regarding the impact of these relationships on CMP use in the practice setting. This study combines social network analysis and statistical methods to examine whether network attributes influence 1) CMP prevalence in large multi-specialty physician organizations; and 2) homogeneity in CMP use among network members in the treatment of patients with asthma, congestive heart failure, diabetes and depression.

Methods: Network analysis methods examined the exchange relationships of physician organizations in the managed care network in California in 2001 and 2006. Physician group ties with other physicians, hospitals, and health maintenance organizations were analyzed using affiliation matrices for each year. Core-periphery analysis identified categories of physician organizations based on their location within the network space. These data were merged with California respondents to the National Study of Physician Organizations; a research project that measured the extent of CMP use in practices of 20 or more physicians. Multivariate regression analyses examined associations of network affiliations and position on CMP prevalence in physician organizations (n=347) and on homogeneity in their use for four chronic conditions among physician group pairs (n=26,857).

Findings: Network effects were found to influence CMP prevalence and homogeneity in their use. CMP use was positively associated with ties to other physician groups and with hospitals. Pairwise homogeneity for specific chronic conditions was greater for organizations with shared network affiliations. CMP homogeneity increased among cohort pairs for each additional shared hospital system affiliation. Core network groups were more homogeneous than other pairs and homogeneity declined among groups in the periphery. CMP use appears to be concentrated within the network core and results of the cohort analysis failed to find evidence of diffusion of these processes toward the periphery.

Conclusion: Implications of these findings suggest that current policy discussions surrounding the development of Accountable Care Organizations (ACOs) should be expanded in order to establish more linkages between core and periphery physician organizations.