Prevalence of Pre-existing Conditions and Adjusted Comparisons, Among a Cohort of Established CHC Patients Pre- and Post-Implementation of the Affordable Care Act (ACA) Insurance Expansions in 2014, by Insurance Type and Race/Ethnicity
| N | Pre-ACA, n (%) | Post-ACA, n (%) | Adjusted Absolute Prevalence Difference, Post vs Pre-ACA (95% CI) | |
|---|---|---|---|---|
| Insurance type | ||||
| Gained Medicaid | 50,839 | 31,454 (61.9) | 40,008 (78.7) | +16.8 (16.5, 17.2) |
| Gained Private | 27,220 | 14,080 (51.7) | 18,987 (69.8) | +18.0 (17.6, 18.5) |
| Race/ethnicity | ||||
| Non-Hispanic White | 29,764 | 20,491 (68.8) | 25,179 (84.6) | +15.8 (15.4, 16.2) |
| Hispanic | 30,296 | 15,108 (49.9) | 20,626 (68.1) | +18.3 (17.8, 18.7) |
| Non-Hispanic Black | 13,090 | 7413 (56.6) | 9827 (75.1) | +18.9 (18.2, 19.6) |
CHC, community health center; CI, confidence interval.
Data from 386 community health centers in 19 states (AK, CA, FL, HI, KS, MD, MN, MO, MT, NC, NM, NV, OH, OR, RI, TX, WA, WI). The retrospective cohort of established patients included those who were uninsured at their last visit pre-ACA (2012 to 2013) and gained insurance coverage post-ACA (2014 to 2015). Conditions identified by ICD 9/10 codes in patients' problem lists or encounter diagnoses based on the Henry J Kaiser Family Foundation list of declinable conditions. We computed within-race/ethnicity and insurance type-group prevalence differences of having at least one pre-existing condition post- vs pre-ACA. Adjusted estimates obtained from generalized estimating equation models specifying a Gaussian distribution, identity link function, an independent working correlation matrix with robust standard errors and adjusted for sex, age, federal poverty level, number of visits, race-ethnicity (insurance type models), post-ACA insurance type (racial/ethnic models), state Medicaid expansion status, and health system. Standard errors were clustered by patient nested within their primary clinic to account for temporal correlation of observations within patients over the ACA periods and intracluster correlation of patients within clinics. Pre-post prevalence differences were significant (P < .001) for all comparisons.