Health Insurance Coverage and Access to Care for Community Health Center Patients: Evidence Following the Affordable Care Act

INTRODUCTION

Community Health Centers (CHCs) provide primary care services to 26 million low-income patients annually, offering preventive services, chronic disease management services, and some mental health and substance abuse services to patients without regard for ability to pay. Despite experiencing substantial coverage gains following the Affordable Care Act (ACA), CHC patients continue to have high rates of uninsurance: in 2014, 23% of CHC patients in Medicaid expansion states and 39% of CHC patients in non-expansion states remained without coverage.¹ Lack of insurance coverage among CHC patients may compromise access to important health services that are often not directly provided in CHCs, including specialty care,² prescription medications,³ and behavioral health services. This in turn may lead to adverse health outcomes for these patients.

Little is known about the role of health insurance in access to care for CHC patients, particularly regarding access to non-required services that are not always directly provided by CHCs. While pre-ACA and state-specific evidence suggests insurance is associated with better access to and quality of primary care for CHC patients,^4,5 no recent evidence exists. Thus, our objective was to estimate the association of having health insurance with access to care among CHC patients in 2014–2015.

METHODS

We used a nationally representative sample of 5040 non-elderly adult CHC patients from the 2014 Health Resources and Services Administration Health Center Patient Survey (HCPS),⁶ representing the 13.9 million adult patients served by US CHCs, to examine differences in access to care for CHC patients with and without insurance coverage. The HCPS, which conducted in-person one-on-one interviews from September 2014 to April 2015, is the first and only survey to be administered following implementation of the ACA that is representative of all CHC patients. We assessed 14 patient-reported outcomes related to access and delayed access to medical care, specialty care, behavioral health care, follow-up care after abnormal cancer screenings, and medications. Final sample sizes varied by measure.

For each outcome, we calculated inverse probability of treatment weights (IPTWs) based on propensity scores to estimate average treatment effects. Propensity scores, which minimize selection bias by balancing on observable characteristics for insured versus uninsured patients, included 19 patient-level sociodemographic and clinical covariates (Table 1), 8 state-level covariates, and survey weights. We used logistic regression models with IPTWs to estimate the effect of having health insurance on each outcome. Models directly adjusted for the patient-level covariates included in the propensity score, thus producing doubly robust estimates, and standard errors were clustered at the state level. A two-sided α level < .05 was considered statistically significant.

Table 1.

Study Sample Characteristics by Insurance Coverage Status (2014–2015)

Characteristic %	Insured (n = 3304†)	Uninsured (n = 1736†)	P value‡
Age			< 0.001
18–25	20.2	11.3
26–34	21.2	23.6
35–44	20.4	21.7
45–54	19.6	25.9
55–64	18.6	18.5
Race/ethnicity			< 0.001
White	49.7	44.5
Hispanic	20.6	35.2
Black	21.6	15.9
Asian	3.0	0.6
Other	5.2	4.2
Sex			0.61
Female	66.0	64.7
Male	34.0	35.9
Income as % of federal poverty level (FPL)			0.05
≤ 100% FPL	54.3	63.3
101–199% FPL	28.6	29.3
≥ 200% FPL	16.8	8.6
English is primary language			0.003
Yes	77.1	64.1
No	22.9	36.5
Education			0.37
Less than high school	33.3	38.5
High school	30.7	28.2
More than high school	36.0	33.8
Urban/rural location			0.05
Rural	44.4	56.0
Urban	55.6	44.6
Other patient characteristics
Married	26.8	30.4	0.27
Non-US born	13.2	33.8	< 0.001
Homeless	2.5	2.4	0.82
Not heterosexual or straight	6.7	3.9	0.07
Self-reported health status			0.25
Excellent	5.9	11.4
Very good	13.0	10.5
Good	39.5	36.1
Fair	30.9	31.5
Poor	10.7	11.1
Indication of select medical conditions
Diabetes	19.0	22.1	0.291
Hypertension	42.0	41.7	0.983
Asthma	19.3	10.0	0.001
Depression	15.1	16.3	0.731
Anxiety	31.6	26.0	0.227
Patient type			0.63
Community Health Center	90.5	91.3
Public housing	1.2	1.3
Migrant	3.2	3.9
Homeless	5.1	4.1
Medicaid expansion state as of 2014			< 0.001
Yes	62.7	32.2
No	37.3	68.4

*Percentages are calculated with analytic survey weights that reflect the distribution of patient characteristics for all health center patients in the USA

^†n represents the unweighted number of health center patients surveyed in the study sample, representing population sizes of 9.1 million insured and 4.8 million uninsured patients

^‡P value represents whether there is a statistically significant difference in the characteristic between CHC patients who are insured versus uninsured

RESULTS

In 2014–2015, approximately 34% of the sample was uninsured. Compared to uninsured CHC patients, insured patients were more likely to be younger, non-Hispanic, above 200% of the federal poverty level, US-born, English-speaking, and living in urban areas and in states that expanded Medicaid eligibility (Table 1). After balancing on observable characteristics, having health insurance was associated with better access for 9 of 14 measures (Table 2). For instance, compared to similar CHC patients without insurance, CHC patients with insurance coverage were more likely to have access to necessary medical care (aOR = 2.12), see a recommended specialist (aOR = 2.73), see a mental health professional if advised (aOR = 1.74), receive recommended follow-up care after an abnormal pap (aOR = 3.44), and get necessary prescription medications (aOR = 2.10), particularly for patients with high cholesterol (aOR = 2.25).

Table 2.

Access to Care for Health Center Patients with Versus Without Health Insurance in 2014–2015

		Health center patients (%)
	No.	Insured	Uninsured	Adjusted odds ratio* (95% CI)
Any medical care
Able to access necessary medical care	3556	86.2	74.9	2.12	(1.74–2.58)
No delay in getting care, test, or treatment	3557	82.5	76.9	1.50	(1.23–1.83)
Specialty care
Saw a specialist if advised	1898	75.3	51.4	2.73	(2.15–3.46)
Behavioral health care
Saw mental health professional if advised	1381	74.1	63.7	1.74	(1.31–2.32)
Able to get needed mental health care	1380	85.2	76.5	1.73	(1.23–2.41)
No delay in getting needed mental health care	1380	81.1	75.3	1.29	(0.93–1.79)
Follow-up care
Follow-up after pap test if recommended	378	84.7	69.0	3.44	(1.80–6.54)
Follow-up after mammogram if recommended	276	84.5	78.3	1.42	(0.61–3.29)
Follow-up after colorectal cancer screening if recommended	212	63.8	50.0	2.02	(0.88–4.66)
Medications
Able to get medication if needed	4038	82.4	70.5	2.10	(1.78–2.23)
No delay in getting medication	4038	77.1	68.5	1.60	(1.34–1.91)
Taking BP medication if BP high in last visit	1182	87.0	87.2	0.98	(0.64–1.50)
Taking asthma medication if needed	472	77.0	75.0	1.16	(0.60–2.24)
Taking cholesterol medication if needed	1514	91.8	84.1	2.25	(1.48–3.43)

BP, blood pressure

*Regression models apply inverse probability of treatment weights, which balance on 19 patient-level sociodemographic and clinical covariates (Table 1) and 8 state-level covariates (primary care physicians per capita, physician assistants and nurse practitioners per capita, specialists per capita, percent of counties with medically underserved area, percent of counties with medically underserved population, expansion status in 2014, Medicaid managed care penetration rate, Medicaid physician fee index). Regression models also directly adjust for the same 19 patient-level covariates; odds ratios compare access for insured versus uninsured (the reference category) and represent average treatment effects; an odds ratio > 1.0 indicates that insurance coverage is associated with better access

DISCUSSION

Among CHC patients, those with health insurance reported significantly better access to medical care, specialty care, follow-up care, and medications. Our findings highlight the vital role of insurance in accessing care within the safety-net, particularly for non-primary care services. Potential reversals to Medicaid expansions may erode access to care for CHC patients, as CHCs may be unable to compensate for coverage losses, while expanding Medicaid in current non-expansion states could improve access for millions of uninsured CHC patients. Furthermore, these findings contribute to our larger understanding of access challenges faced by the uninsured in a post-ACA era, where expanding safety-net capacity to provide both primary and non-primary care services for uninsured patients remains critical. Additional policy options include further investments to expand CHCs’ scope of services and capacity to care for the uninsured, including sustained levels of federal grant funding, and increasing funds available to offset uncompensated care for specialists serving uninsured patients.

Compliance with Ethical Standards

Conflict of Interest

Dr. Cole discloses that from 2011 to 2017, she was employed by The Lewin Group—a subsidiary of Optum. All other authors have no conflicts of interest to disclose.