The Affordable Care Act and Access to Care for Reproductive-Aged and Pregnant Women in the United States, 2010–2016

Abstract

Objectives. To estimate the association between the Affordable Care Act (ACA), health insurance coverage, and access to care among reproductive-aged and pregnant women.

Methods. We performed an observational study comparing current insurance type, cost-related barriers to medical care, and no usual source of care among reproductive-aged (n = 128 352) and pregnant (n = 2179) female respondents to the National Health Interview Survey in the United States, before (2010–2013) and after (2015–2016) the ACA coverage expansions.

Results. Among reproductive-aged women, the ACA was associated with a 7.4 percentage-point decrease in the probability of uninsurance (95% confidence interval [CI] = −8.6, −6.2), a 3.6 percentage-point increase in Medicaid (95% CI = 2.5, 4.7), and a 3.1 percentage-point increase in nongroup private coverage (95% CI = 2.1, 4.1). The ACA was also associated with a 1.5 percentage-point decline in cost-related barriers to medical care (95% CI = −2.6, −0.5) and a 2.4 percentage-point reduction in lacking a usual source of care (95% CI = −4.5, −0.3). We did not find significant changes in insurance or cost-related barriers to care for pregnant women.

Conclusions. The ACA was associated with expanded insurance coverage and improvements in access to care for women of reproductive age, particularly for those with lower incomes.

From 1980 to 2009, rates of uninsurance among reproductive-aged women in the United States steadily increased.^1,2 By 2009, the year before the Affordable Care Act (ACA) was signed into law, 1 in 4 women of reproductive age and 1 in 8 pregnant women reported being currently uninsured.² Rates of uninsurance have historically been higher among reproductive-aged women compared with the general US population because women are less likely to be insured through employment, and younger adults are less likely to have coverage than older adults.³ Before the ACA, public coverage options for nonelderly women were limited. Although Medicaid, the public insurance program for some categories of low-income Americans, has been available to low-income pregnant women since the late 1980s, this coverage is temporary and only insures women from conception to 60 days after delivery. As a result, before the ACA, the majority of low-income women who gained pregnancy-related Medicaid were uninsured 12 months before delivery and 6 months following birth.⁴

Uninsurance among reproductive-aged women is a concern given the large body of evidence that supports the importance of health insurance in ensuring access to health care, promoting adherence to treatment, protecting against the financial burden of care, and supporting self-reported health and well-being.⁵ In addition to supporting women’s health in general, insurance is also important for supporting access to care in the period surrounding pregnancy. Coverage before and after pregnancy can facilitate access to interventions to minimize behavioral risk factors, manage chronic disease, and support general preconception health, which is associated with maternal and infant health outcomes.⁶ Evidence has also shown that health insurance increases access to contraception,⁷ which reduces unplanned pregnancies and improves pregnancy spacing. Finally, after delivery, insurance may facilitate women’s access to diagnosis and treatment of pregnancy-related physical and mental health conditions, including postpartum depression and anxiety.

The ACA’s coverage expansions in 2014 represent the largest coverage expansion in the United States since the establishment of Medicare and Medicaid in 1965.⁸ The ACA expanded Medicaid eligibility to low-income adults (those earning 138% or less of the federal poverty level according to the US Department of Health and Human Services 2017 Poverty Guidelines (https://aspe.hhs.gov/2017-poverty-guidelines); US $22 411 for a family of 2 in 2017) in the states that have thus far chosen to participate and established private health insurance exchanges in all 50 states, which feature subsidized premiums for those with incomes between 100% and 400% of the poverty level (between US $16 240 and US $64 960 for a family of 2 in 2017). These expansions have the potential to dramatically increase access to insurance and reduce cost-related barriers to care for reproductive-aged women. Early evidence has been promising. A previous study based on a small online survey comparing women in 2012 and 2015 found reductions in uninsurance among reproductive-aged women, and an analysis of national survey data from 2012 to 2015 found reductions in insurance and cost barriers to care among low-income women living in states that expanded Medicaid.^9,10

Furthermore, although the ACA did not specifically target pregnant women, coverage could also be improved for this group through (1) increased private insurance among women not eligible for pregnancy-related Medicaid and (2) improved timing and uptake of Medicaid coverage among women eligible for pregnancy-related Medicaid. However, no study has examined overall national changes in insurance and access to care before and after the ACA’s implementation for both reproductive-aged and pregnant women. The objective of this study was to use nationally representative data to measure the association between the ACA’s major coverage expansions and insurance coverage, cost-related barriers to care, and access to a usual source of care, for reproductive-aged and pregnant women.

METHODS

In this observational study, we used national survey data from 2010 to 2016 to compare changes in insurance status and affordability of care among reproductive-aged and pregnant women before and after the major coverage expansions of the ACA, implemented on January 1, 2014. Similar to other assessments of coverage changes for the ACA as a whole,^8,11 our study design was an interrupted time series without a control group, as coverage gains under the ACA have occurred in all 50 states. We used multivariate adjustment to control for prepolicy trends in the outcomes and potential observable confounders. We considered the prepolicy period to be 2010 to 2013 and the postpolicy period to be 2015 to 2016. Similar to other studies of the ACA, we considered 2014 to be a policy transition period because it may take several months for the coverage changes to affect access outcomes, some of which are based on a 12-month recall period.¹² Over the study period, the number of states (including Washington, DC) participating in the ACA Medicaid expansions grew from 26 as of January 1, 2014, to 32 as of December 31, 2016.

Data Sources and Study Sample

Our primary data source was the 2010 to 2016 panels of the National Health Interview Survey (NHIS), a nationally representative survey of US households conducted by the US Centers for Disease Control and Prevention (CDC). The NHIS collects detailed information on sociodemographics, health insurance status, health care access, and health care utilization via an in-person interview. The sample included all female respondents of reproductive age (18–44 years). We also identified a subset of women who were pregnant at the time of survey by using the NHIS “Sample Adult” file, which includes additional data from a randomly selected adult from each surveyed family. Because of the small sample of pregnant women in the NHIS, we conducted a sensitivity analysis with data from a large national telephone survey, the Behavioral Risk Factor Surveillance System (BRFSS), which samples a larger sample of pregnant women but has a higher nonresponse rate, does not include detailed information on insurance type, and has more limited data on family income.

Study Outcomes

The study outcomes were insurance status and measures of access to care. We coded current health insurance status in 5 mutually exclusive categories: private (group), private (nongroup), Medicaid, other insurance, or uninsured, according to a hierarchy in which Medicaid took precedence over private insurance or other coverage (for those reporting more than 1 type of insurance). Private (group) coverage includes health insurance available through the workplace or school, whereas private (nongroup) includes insurance purchased on the individual market, including the ACA Marketplaces. To capture cost-related barriers to care, we used an indicator of whether a woman reported delaying or not receiving medical care because of cost in the past 12 months. To measure access to primary care, we used an indicator of whether a woman reported currently having a usual source of health care.

Statistical Analysis

We used a logistic regression model to estimate the association between the ACA and the binary outcomes described previously. Based on the model, we calculated the difference in the predicted probability of the outcomes in the first transition year of the ACA (2014) and the second and third years of the policy (2015–2016) compared with the prepolicy baseline (2010–2013). We calculated predicted probabilities by using the Stata SE version 15 (StataCorp LP, College Station, TX) MARGINS command, which represent average marginal effects (i.e., the discrete change in the probability of the outcome after the policy averaged across the observed values of the covariates). The model adjusted for age, race, Hispanic ethnicity, marital status, US citizenship, region, family income as a percentage of the federal poverty level, and a linear yearly prepolicy time trend. Because insurance rates vary with the economy and employment rates, we also adjusted for both individual-level self-reported employment status and age-group specific quarterly national female unemployment rates obtained from the US Bureau of Labor Statistics.

We weighted all estimates by using the survey sample weights provided by the CDC. For each outcome model, we excluded observations with missing outcome data. We used multiple imputation files provided by the CDC to address nonresponse in family income in the NHIS. For other covariates, we included observations with missing covariate data by using indicator variables for missing values in a particular covariate. The eMethods of the Appendix provides additional detail on the model specification and multiple imputation procedure. We conducted all analyses with Stata SE version 15.

Subgroup Analyses

We modeled outcomes separately for the subgroup of pregnant women. We also conducted subgroup analyses for reproductive-aged women with family incomes above and below 185% of the federal poverty level, the median state income threshold to qualify for pregnancy-related Medicaid coverage when the ACA was implemented. The ACA did not change the eligibility criteria for pregnancy-related Medicaid coverage; however, it is relevant to measure coverage changes after the ACA among women who are likely to enroll in public insurance if they become pregnant. We did not stratify analyses for pregnant women because of small sample size.

RESULTS

We identified 123 352 reproductive-aged women aged 18 to 44 years (74 175 before the ACA and 54 177 after the ACA) and 2179 pregnant women (1263 before the ACA and 916 after the ACA). Table 1 shows the sociodemographic characteristics of the sample.

TABLE 1—

Sample Characteristics, Unweighted Sample Size and Weighted Percentage: National Health Interview Survey, United States, 2010–2016

	Reproductive-Aged Women, No. (%)	Pregnant Women, No. (%)
Sample size
Total	128 352	2 179
Pre-ACA (2010–2013)	74 175	1 263
Post-ACA (2014–2015)	54 177	916
Age, y
18–21	17 706 (15)	234 (13)
22–29	37 684 (30)	1 037 (47)
30–39	48 511 (36)	847 (37)
40–44	24 451 (18)	61 (3)
Race
White	94 424 (76)	1 593 (77)
Black	20 736 (15)	389 (15)
Other	13 192 (8)	197 (8)
Hispanic ethnicity	32 996 (19)	571 (21)
Marriage status
Married	58 912 (45)	1 318 (65)
Missing	360 (0.3)	2 (0.04)
Education
< High school	17 412 (11)	366 (15)
High school	58 638 (46)	907 (41)
Any postsecondary	50 849 (42)	900 (44)
Missing	1 453 (1)	6 (0.2)
Citizenship status
US citizen	109 725 (89)	1 798 (84)
Missing	637 (0.5)	1 (0.02)
Region of residence
Northeast	20 008 (17)	281 (15)
Midwest	24 898 (22)	470 (24)
South	46 197 (37)	805 (38)
West	37 249 (24)	623 (23)
Currently employed	92 934 (73)	1 528 (71)
Family income ≤ 185% FPL	51 109 (36)	998 (40)

Note. ACA = Affordable Care Act; FPL = federal poverty level according to the US Department of Health and Human Services 2017 Poverty Guidelines (https://aspe.hhs.gov/2017-poverty-guidelines).

Figure 1 shows the unadjusted trends in current insurance status for reproductive-aged women from 2010 to 2016. Table 2 shows the unadjusted percentage of women reporting each outcome and the adjusted changes in the predicted probabilities of reporting each outcome after the ACA. Table A shows the odds ratios from the logistic regression models used to calculate these predicted probabilities. Among all reproductive-aged women, the first year of the ACA was associated with an adjusted 3.9 percentage-point (95% confidence interval [CI] = −4.8, 2.9) decrease in the predicted probability of uninsurance (Table 2). In the second and third year of implementation, the probability of uninsurance decreased by 7.4 percentage points (95% CI = −8.6, −6.2), a 33% decrease from the pre-ACA baseline. The reduction in uninsurance was accounted for by increases in Medicaid coverage (3.6 percentage points in the postpolicy years, 25% increase from baseline) and nongroup private insurance (3.1 percentage points, 58% increase from baseline). The ACA was not associated with significant changes in group private insurance among reproductive-aged women.

FIGURE 1—

Unadjusted Trends in Current Insurance Coverage for (a) Reproductive-Aged Women and (b) Pregnant Women: National Health Interview Survey, United States, 2010–2016

Note. Reproductive-aged women, n = 128 352; pregnant women, n = 2179. Error bars represent 95% confidence intervals. “Other insurance” category not shown.

TABLE 2—

Changes in Insurance Coverage and Access to Care After the Affordable Care Act (ACA) Among Reproductive-Aged and Pregnant Women: National Health Interview Survey, United States, 2010–2016

	Unadjusted Percentage of Women				Adjusted Post-ACA Changes
	Pre-ACA Baseline	2014	2015	2016	Year 1 (Transition), Change (95% CI)	Years 2 and 3 (Postpolicy), Change (95% CI)
Reproductive-aged women 18–44 y
Current insurance status
Medicaid	14.5	17.2	17.9	19.7	2.0 (1.1, 3.0)	3.6 (2.5, 4.7)
Private (group)	55.3	55.0	57.5	57.2	−0.7 (−1.9, 0.5)	0.3 (−1.1, 1.7)
Private (nongroup)	5.3	8.2	8.3	8.0	3.0 (2.3, 3.7)	3.1 (2.1, 4.1)
Other	2.7	2.6	2.9	2.8	−0.3 (−0.8, 0.1)	−0.2 (−0.7, 0.4)
Uninsured	22.0	16.9	13.4	12.3	−3.9 (−4.8, −2.9)	−7.4 (−8.6, −6.2)
Delayed or did not receive medical care because of cost	14.0	11.4	10.2	9.5	−0.9 (−1.8, 0.01)	−1.5 (−2.6, −0.5)
No usual source of care	17.8	14.5	15.9	14.7	−3.6 (−5.3, −1.9)	−2.4 (−4.5, −0.3)
Pregnant women
Current insurance status
Medicaid	39.3	31.2	33.4	33.9	−7.6 (−16.2, 1.1)	−1.7 (−11.8, 8.4)
Private (group)	45.2	50.4	51.1	48.2	1.3 (−8.3, 10.9)	−5.7 (−15.4, 4.0)
Private (nongroup)	3.0	5.0	6.6	6.4	2.6 (−1.3, 6.6)	3.3 (−2.2, 8.8)
Other	3.5	4.0	3.3	4.5	2.8 (−0.7, 6.4)	3.9 (−0.8, 8.6)
Uninsured	9.0	9.4	5.7	7.0	0.8 (−3.5, 5.1)	−1.0 (−6.4, 4.4)
Delayed or did not receive medical care because of cost	10.2	5.5	7.6	6.2	−0.9 (−6.5, 4.7)	3.7 (−3.5, 11.0)
No usual source of care	11.6	6.6	7.3	11.3	−4.8 (−12.1, 2.6)	−0.7 (−8.5, 7.1)

Note. CI = confidence interval. Adjusted changes represent the change in the predicted probability of the outcome in the post-ACA year(s) relative to the predicted probability in the pre-ACA baseline period (2010–2013). Adjusted logistic regression models control for race, Hispanic ethnicity, marital status, family income as a percentage of federal poverty, employment status, citizenship, region, yearly time trend, and national age-group-specific female unemployment rates.

Table 2 and Figure 2 also show changes in access to care for this population. The probability of reproductive-aged women reporting delaying or not receiving medical care because of cost in the past 12 months decreased by 1.5 percentage points (95% CI = −2.6, −0.5; 11% decrease from baseline) in the second and third year of the ACA. In addition, the predicted probability of not having a usual source of care declined by 2.4 percentage points (95% CI = −4.5, −0.3), a 14% decrease from baseline.

FIGURE 2—

Unadjusted Trends in Reproductive-Aged Women Who (a) Were Uninsured, (b) Delayed/Did Not Receive Medical Care Because of Cost, and (c) Had No Usual Source of Care: National Health Interview Survey, United States, 2010–2016

Note. Reproductive-aged women, n = 128 352; pregnant women, n = 2179. Error bars represent 95% confidence intervals.

Changes in insurance status and access to care were larger among reproductive-aged women with family incomes less than or equal to 185% of the federal poverty level compared with those earning above this threshold (Table B; Figures A and B, available as supplements to the online version of this article at http://www.ajph.org). In 2015 to 2016, the predicted probability of uninsurance decreased by 13.3 percentage points (95% CI = −15.8, −10.9) compared with 2010 to 2013 among lower-income reproductive-aged women compared with 4.2 percentage points (95% CI = −5.4, −2.9) among higher-income women—though in both cases this represented a roughly one third relative decrease from the pre-ACA mean. Lower-income women also experienced significant reductions in delaying or not receiving medical care because of cost and improvements in having a usual source of care after the ACA. We observed no significant changes in access outcomes among higher-income women. Figures A and B show the unadjusted trends for each outcome by family income.

Figure 1b shows the unadjusted trends in current insurance status for pregnant women from 2010 to 2016. We observed no statistically significant changes in insurance status or access to care among pregnant women (Table 2). Our results were consistent in the sensitivity analyses that replicated these models by using the larger sample (n = 18 382) from the BRFSS data (Tables C and D, Figure C, available as supplements to the online version of this article at http://www.ajph.org).

DISCUSSION

In this national study of the first 3 years of the ACA’s major coverage expansions in the United States, we found significant reductions in uninsurance and increases in nongroup private insurance and Medicaid among reproductive-aged women. We also found significant reductions in women reporting not having a usual source of care and cost-related barriers to medical care. In subgroup analyses, these changes were larger among lower-income women. We did not find significant changes in these outcomes for pregnant women.

Our results suggest that the decades-long trend of increasing uninsurance among reproductive-aged women reversed after the major ACA coverage expansions in 2014, resulting in a significant reduction in uninsurance that grew from a 18% decrease from baseline in the first year of the policy to a 33% decline in the second and third years of implementation. The corresponding reductions in cost-related barriers to medical care and improvements in having a usual source of care after the ACA are consistent with previous evidence on the benefits of insurance coverage,⁵ but our findings extend these results to women of reproductive age.

We found that reductions in uninsurance and improvements in access to a usual source of care were concentrated among reproductive-aged women earning less than 185% of federal poverty, the median state eligibility threshold for pregnancy-related Medicaid at the time of ACA implementation. Women in this income range are more likely to experience inadequate prenatal care and severe maternal morbidity.^13–15 Our results suggest that the ACA was associated with increased coverage and access to care before conception for low-income women, many of whom will subsequently obtain Medicaid for pregnancy and childbirth. This may improve outcomes among Medicaid-covered births by increasing access to family planning services and interventions to reduce risk factors for adverse pregnancy outcomes before conception. Having a regular provider before pregnancy and receiving preconception care are also associated with increased odds of timely and adequate prenatal care.^16,17

Increases in Medicaid coverage rates among reproductive-aged women may also improve postpartum health outcomes for low-income women, as more women will be able to maintain Medicaid long term rather than losing it when pregnancy-related Medicaid coverage expires 60 days after delivery. Access to diagnosis and treatment of physical and mental health conditions in the postpartum period is critical to reducing maternal morbidity and mortality. A recent CDC report estimated that nearly 20% of all deaths occur from 43 days to 1 year after delivery, 32% and 16% of which are associated with cardiovascular and mental health conditions, respectively.¹⁸

The ACA did not change Medicaid eligibility for pregnant women, and we did not find evidence that the ACA was associated with increased coverage rates among pregnant women. Although the findings were not statistically significant, the point estimates from the adjusted models suggest a decrease in uninsurance and Medicaid coverage and an increase in nongroup private insurance among pregnant women (which includes plans purchased through the ACA Marketplaces). Additional research, ideally with a larger sample and more years of data, is needed to assess whether the ACA was associated with a shift from Medicaid to private Marketplace coverage in the prenatal period and, if so, whether this had any impact on access to care and out-of-pocket costs for pregnant women. As private insurance is a more stable source of coverage compared with time-limited pregnancy-related Medicaid, such a shift could result in improvements in continuity and coverage of care from before to after pregnancy.⁴ However, unlike Marketplace plans, pregnancy-related Medicaid coverage has little to no cost-sharing requirements and often includes additional services such as dental care and transportation, which could be important benefits for low-income women.

Despite the significant gains in coverage we observed, we found that approximately 1 in 8 reproductive-aged women and 1 in 15 pregnant women were still uninsured in 2016. Several groups of women are at risk for lacking insurance after the ACA. One such population is low-income women living in the 19 states that did not implement the ACA Medicaid expansion. Research on the impact of the ACA on low-income adults has found that the uninsurance rate declined significantly in expansion states relative to nonexpansion states¹²; however, 3 to 4 million low-income adults remain in the so-called “Medicaid gap,” ineligible for any publicly subsidized health insurance because of their states’ decisions not to expand coverage.¹⁹ Another population at risk for remaining without coverage are undocumented immigrants. Women in this group account for nearly 7% of births in the United States and are 4 times more likely than US citizens to be uninsured, reflecting more limited access to employer-sponsored coverage and restrictions on eligibility for public programs such as Medicaid.^20,21

Finally, preliminary evidence indicates that after several years’ decline under the ACA, uninsurance rates have begun to rise under the Trump Administration, which may in part reflect changes the administration has taken to reduce enrollment in the Marketplaces and reduce enforcement of the individual mandate for coverage, and ongoing debate over potentially repealing the law.²²

Limitations

This study has limitations that should be considered when one is interpreting the results. First, like all observational studies, our results could be subject to unobserved confounding. While adjusted models controlled for potential individual-level confounders and unemployment rates, other unmeasured changes in the economy or demographics of reproductive-aged women over time could also have contributed to the observed changes. However, the magnitude of the estimates and the abrupt and persistent changes detected are consistent with the ACA playing a major role. Second, the small sample size for the subgroup of pregnant women resulted in imprecise estimates, even in the larger BRFSS data set. Although we did not find significant changes among this group, the wide CIs associated with these estimates suggest they should be interpreted with caution. Third, our data are all self-reported, which may be subject to recall, reporting bias, or both, and previous research indicates that some survey respondents may confuse Medicaid and private coverage; however, measures of uninsurance are generally less subject to this concern.²³ Fourth, the results may not generalize to all subgroups of reproductive-aged and pregnant women. While we conducted prespecified subgroup analyses by income, the association between the ACA and the outcomes may vary by race, ethnicity, health status, or other factors.

Public Health Implications

Lack of health insurance among reproductive-aged women could be one factor driving the United States’ poor performance on maternal and infant health outcomes relative to other high-income nations.²⁴ Although we did not find significant short-run changes for pregnant women, the improvements observed in access and coverage for reproductive-aged women have the potential to result in long-run improvements in US maternal health outcomes by increasing access to preconception and postpartum care, as well as improving the timing and continuity of prenatal care. Early evidence is suggestive of possible health benefits associated with the ACA. For example, the ACA’s dependent coverage provision, which extended the age to which young adults could stay on their parents’ private insurance from 18 to 25 years, was associated with an increase in private insurance payment for birth, improvements in prenatal care use, and modest reductions in preterm birth and cesarean deliveries among unmarried women.²⁵ Recent expansions of prenatal Medicaid coverage to undocumented and new immigrant women in some states have also been associated with improvements in prenatal care use, though evidence on birth outcomes is mixed.^26,27 Finally, a study of the relationship between ACA Medicaid expansion and infant mortality found greater declines in expansion states compared with nonexpansion states.²⁸ However, more research is needed to examine the long-run impacts of the ACA on maternal and infant health and to determine the mechanisms through which expanded coverage and access to care may lead to improvements in maternal and infant outcomes.