Receipt of primary care is associated with better health.1,2 Despite the benefits of primary care, Americans’ receipt of primary care, changes in receipt of primary care over time, and differences in receipt of primary care according to sociodemographic and clinical characteristics are not well known.
We analyzed data from the nationally representative Medical Expenditure Panel Survey (21 915-26 509 individuals yearly) from 2002 through 2015. We used a patient-centered definition of primary care that included the 4 C’s of primary care: first contact, comprehensive, coordinated, and continuous. To estimate factors associated with receiving primary care over time, we used multivariable logistic regression, adjusting for each year from 2002 through 2015;sociodemographic and clinical variables, presented in the Table; and the complex survey design. We examined having primary care over time by decade of age and stratified by comorbidity. We performed all analyses with SAS statistical software (version 9.4, SAS Institute). The Harvard Medical School Institutional Review Board approved this study and determined it not to be human subject research, so patient informed consent was waived.
The proportion of adult Americans with an identified source of primary care decreased from 77% (95% CI, 76%-78%) in 2002 to 75% (95% CI, 74%-76%) in 2015 (odds ratio, 0.90 [95% CI, 0.82-0.98]). During this period, receipt of primary care decreased for every decade of age except for Americans in their 80s, with statistically significant reductions for those in their 30s, 40s, and 50s (Figure, A). For example, 71% of Americans in their 30s had primary care in 2002 compared with 64% in 2015 (P < .001).
Among Americans with no comorbidities (60% [95% CI, 59%-61%] in 2002 and 51% [95% CI, 50%-52%] in 2015), receipt of primary care decreased for every decade of age (Figure, B). For example, for Americans in their 60s with no comorbidities, having primary care fell from 82% in 2002 to 73% in 2015 (P = .003). Having primary care for Americans with at least 3 comorbidities was generally stable (Figure, C).
In multivariable modeling, factors associated with a decreased likelihood of having primary care included calendar year (adjusted odds ratio [aOR], 0.97 for each year from 2002 through 2015 [95% CI, 0.97-0.98]), male sex (aOR vs female sex, 0.59 [95% CI, 0.57-0.60]), Latino race/ethnicity (aOR vs white, 0.80 [95% CI, 0.77-0.84]), black race/ethnicity (aOR vs white, 0.88 [95% CI, 0.84-0.93]), Asian race/ethnicity (aOR vs white, 0.67 [95% CI, 0.62-0.74]), not having insurance (aOR vs private insurance, 0.29 [95% CI, 0.27-0.30]), and Southern US Census Bureau region (aOR vs Northeast, 0.53 [95% CI, 0.48-0.58]) (Table).
From 2002 through 2015, a decreasing proportion of Americans had an identified source of primary care, especially Americans who were younger, less medically complex, of minority background, or living in the South. To improve Americans’ health in an efficient and cost-effective manner, policy makers should prioritize increasing the proportion of Americans with primary care.
The decrease in receipt of primary care, particularly among younger patients or patients with no chronic medical conditions, may be related to their choosing nonlongitudinal interactions over continuity, perhaps related to the convenience revolution3 and a perception that primary care has failed to adopt new modes of delivering treatment that might be more accessible to patients.4,5 Financial barriers, especially among uninsured Americans, may prevent some people from accessing primary care. Shortages in the availability of primary care may pose access barriers even to insured people, with the result that fewer younger and healthier patients have a regular source of care.6
Corresponding Author: David M. Levine, MD, MPH, MA, Division of General Internal Medicine and Primary Care, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St, 3rd Floor, Boston, MA 02120 (dmlevine@bwh.harvard.edu).
Published Online: December 16, 2019. doi:10.1001/jamainternmed.2019.6282
Author Contributions: Dr Levine had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Linder and Landon contributed equally to this article.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Levine, Linder.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Levine, Linder.
Administrative, technical, or material support: Levine, Linder.
Supervision: Linder, Landon.
Conflict of Interest Disclosures: Dr Levine reports funding through Biofourmis to perform a machine learning study of home-based acute care monitoring. No other disclosures were reported.