Study design

We created and analyzed a novel, nationally representative dataset that links survey data on patient perceptions of care integration to medical group data. To collect patient data, we refined, extended and administered an existing instrument, the Patient Perceptions of Integrated Care (PPIC) survey, which includes 36 integrated care items. It was theoretically derived and refined through pilot-testing, cognitive testing, and input from an advisory panel of survey measurement and care integration experts, patient representatives, and patients, and tested for reliability and validity. The survey measures five psychometric dimensions, plus an index describing integration following hospitalization. Medical group data were acquired from the National Study of Physician Organizations (NSPO3). We estimated patient-level multiple regression models assessing the relationship between the survey-based measures of patient-perceived integration and dichotomized variables measuring organizational integration, including medical group size, hospital ownership, specialty mix, information technology sophistication, and care management process capabilities (CMPs). We controlled for primary provider specialty and patient characteristics, including demographics, health status, and life orientation (optimism), clustering standard errors by practice.

We administered the PPIC survey across a stratified random sample of 150 medical groups including 12,364 Medicare beneficiaries with multiple chronic conditions, oversampling people who were hospitalized in the past six months. Our final sample included 3,226 respondents (27%). The majority of respondents were high need; 93% had more than two chronic conditions. We linked 84% of respondents with a specific medical group.