Physician Networks and Ambulatory Care-sensitive Admissions.
Lawrence P Casalino, Michael F Pesko, Andrew M Ryan, David J Nyweide, Theodore J Iwashyna, Xuming Sun, Jayme Mendelsohn, James Moody
Author Information
Lawrence P Casalino: *Division of Health Policy and Economics, Department of Healthcare Policy and Research †Department of Healthcare Policy and Research, Weill Cornell Medical College, New York, NY ‡Department of Health Management and Policy, School of Public Health, University of Michigan, Ann Arbor, MI §Centers for Medicare & Medicaid Services, Center for Medicare and Medicaid Innovation, Baltimore, MD ∥Department of Pulmonary and Critical Care, University of Michigan, Ann Arbor, MI ¶HIV Epidemiology and Field Services Program, Department of Health and Mental Hygiene, Bureau of HIV/AIDS Prevention and Control, New York, NY #Department of Sociology, Duke University **King Abdulaziz University, Durham, NC.
BACKGROUND: Research on the quality and cost of care traditionally focuses on individual physicians or medical groups. Social network theory suggests that the care a patient receives also depends on the network of physicians with whom a patient's physician is connected. OBJECTIVES: The objectives of the study are: (1) identify physician networks; (2) determine whether the rate of ambulatory care-sensitive hospital admissions (ACSAs) varies across networks--even different networks at the same hospital; and (3) determine the relationship between ACSA rates and network characteristics. RESEARCH DESIGN: We identified networks by applying network detection algorithms to Medicare 2008 claims for 987,000 beneficiaries in 5 states. We estimated a fixed-effects model to determine the relationship between networks and ACSAs and a multivariable model to determine the relationship between network characteristics and ACSAs. RESULTS: We identified 417 networks. Mean size: 129 physicians; range, 26-963. In the fixed-effects model, ACSA rates varied significantly across networks: there was a 46% difference in rates between networks at the 25th and 75th performance percentiles. At 95% of hospitals with admissions from 2 networks, the networks had significantly different ACSA rates; the mean difference was 36% of the mean ACSA rate. Networks with a higher percentage of primary-care physicians and networks in which patients received care from a larger number of physicians had higher ACSA rates. CONCLUSIONS: Physician networks have a relationship with ACSAs that is independent of the physicians in the network. Physician networks could be an important focus for understanding variations in medical care and for intervening to improve care.
