Validating the New Primary Care Measure in the Medical Expenditure Panel Survey

• Published in: Medical care Vol. 58; no. 1; p. 52
• Main Authors: Olaisen, R Henry, Flocke, Susan A, Smyth, Kathleen A, Schluchter, Mark D, Koroukian, Siran M, Stange, Kurt C
• Format: Journal Article
• Language: English
• Published: United States 01.01.2020
• Subjects: Adult; Aged; Female; Health Care Surveys - methods; Health Care Surveys - statistics & numerical data; Health Expenditures - statistics & numerical data; Health Services Research - methods; Humans; Logistic Models; Male; Middle Aged; Primary Health Care - methods; Primary Health Care - statistics & numerical data; Process Assessment, Health Care - methods; Process Assessment, Health Care - statistics & numerical data; Reproducibility of Results; Retrospective Studies; United States; United States
• ISSN: 1537-1948
• DOI: 10.1097/MLR.0000000000001220

The advancement of primary care research requires reliable and validated measures that capture primary care processes embedded within nationally representative datasets. The objective of this study was to assess the validity of a newly developed measure of primary care processes [Medical Expenditure Panel Survey (MEPS)-PC] with preliminary evidence of moderate to excellent reliability. A retrospective cohort study of community-dwelling adults with history of office-based provider visit/s using the MEPS (2013-2014). The 3 MEPS-PC subscales (Relationship, Comprehensiveness, and Health Promotion) were tested for construct validity against known measures of primary care: Usual Source of Care, Known Provider, and Family-Usual Source of Care. Concurrent and predictive logistic regression analyses were calculated and compared with a priori hypotheses for direction and strength of association. For concurrent validity, all odds ratio estimates conformed with hypotheses, with 91% displaying statistical significance. For predictive validity, all estimates were in the direction of hypotheses, with 92% displaying statistically significant results. Although Relationship and Health Promotion subscales conformed uniformly with hypotheses, the Comprehensiveness subscale yielded significant results in 60% of bivariate odds ratio estimates (P<0.05). The MEPS-PC composite measures display modest to strong preliminary evidence of concurrent and predictive validity relative to known indicators of primary care. The MEPS-PC composite measures display preliminary evidence of concurrent and predictive construct validity, and it may be useful to researchers investigating primary care processes and complexities in the health care environment.