The Diffusion of the Patient-centered medical home

The patient-centered medical home (PCMH) has received remarkable attention, due in no small part to its central role in the 2010 Affordable Care Act.  There are numerous and evolving definitions of PCMH practices and general agreement about the large heterogeneity even among recognized PCMH practices. However, several core elements are required of all practices recognized as PCMHs.  These elements include greater involvement of patients and families in medical decisions, comprehensive and coordinated care which generally integrates medical and mental health care, greater accessibility to patients, and greater use of evidence-based approaches to improve quality and safety. Recent literature on primary care practices that have adopted elements consistent with PCMH standards have found generally lower costs, fewer patients using emergency department services and inpatient admissions, greater access to outpatient care and greater use of preventative services. While the PCMH model of primary care may hold enormous potential for improving medical practice by increasing efficiency in the consumption of medical care and improving patient-centered outcomes, little is known about the barriers and facilitators of PCMH transformation and its diffusion among primary care practices nationally. We examine whether factors such as market characteristics, patient populations, Medicaid generosity, and other state level factors affect the diffusion of PCMH recognition in a national population of PCMH-recognized practices. We use data on the National Committee for Quality Assurance (NCQA) recognized primary care practices and providers from 2008-2014 merged with a variety of state-level characteristics reflecting potential barriers and facilitators of innovative medical care. Non-linear generalized estimating equation models with polynomial functions of time, log links, negative binomial distributions to account for the count nature of the dependent variable, and exchangeable correlations were used. These models allow the diffusion patterns to deviate from the standard S-shaped diffusion curves, due in part to the time-limited recognition periods (3 years) per practice and evolving recognition standards. We find that beginning in April of 2008, when the first PCMH was recognized, there was substantial diffusion of PCMH recognized practices, with average states seeing growth of two additional practices per month. More populous states and those with greater capitated managed care plans saw greater growth in the diffusion of PCMHs, with every 100,000 state inhabitants associated with 0.5 more PCMH practices (p<0.01) and every 10,000 Medicaid enrollees in a capitated program associated with 0.3 more PCMH practices (p<0.01) while states with greater enrollment in Medicaid programs and those with greater federal matching rates for Medicaid expenditures, a marker of poverty, had lower levels of PCMH recognition (every 10,000 Medicaid enrollees are associated with 0.2 fewer PCMH practices and every 1% point increase in the FMAP associated with 0.1 fewer PCMH practices). These facilitators and barriers of PCMH diffusion were very similar for PCMHs recognized at the highest level (level 3). These results have important implications for policy makers, as the benefits of PCMH practices are disproportionately accruing in states that may already have made other investments in their health care system, such as transforming the Medicaid program through capitation.