Developing a New Measure of Primary Care Using the Medical Expenditure Panel Survey

Objective: To develop and assess the reliability of a measure of primary care using items from the Medical Expenditure Panel Survey (MEPS), a US representative survey of community-dwelling persons. Methods: On the basis of the domains from the literature on primary care characteristics, we identified relevant items within the 2013-2014 MEPS family of surveys. In a sample of primary survey respondents with at least 1 office-based physician visit in the last 12 months, we conducted exploratory factor analysis, retaining items with a factor loading of 0.30 and factors ≥3 items. Using a hold-out sample, internal consistency, reproducibility, and confirmatory factor analyses were performed. Results: On the basis of 16 care domains, we found 32 candidate items in the MEPS. Factor analyses of data from 4549 persons meeting inclusion criteria (27.6% of the total sample), yielded 3 unique factors involving 24 items. We named these subscales Relationship, Comprehensiveness, and Health Promotion, displaying internal consistency reliability of 0.86, 0.78, and 0.69, respectively. Confirmatory factor analysis corroborated the stability of the exploratory findings in the hold out sample. Sensitivity analyses showed robustness to differences in underlying correlation structure, alternative approach to missing data, and extension to indirect survey respondents. Conclusions: The MEPS Primary Care measure with 3 subscales is reliable and may be useful in conducting primary care health services and outcomes research in the rich MEPS dataset. Further validation is needed, and is described in a companion paper.