ACA Medicaid Expansions Increased Preconception Health Counseling, Folic Acid Intake, and Effective Postpartum Contraception

Abstract

The period prior to pregnancy is critically important for the health of a woman and her infant, yet not all women have access to health insurance during this time. We evaluated whether increased access to health insurance under the ACA Medicaid expansions affected ten preconception health indicators, including the prevalence of chronic conditions and health behaviors, birth control use and pregnancy intention, and receipt of preconception health services. By comparing changes in outcomes for low-income women in expansion and non-expansion states, we document greater preconception health counseling, prepregnancy folic acid intake, and postpartum use of effective birth control methods among low-income women in expansion states associated with Medicaid expansion. We do not find evidence of changes on the other preconception health indicators examined. Our findings indicate that expanding Medicaid led to detectable improvements on a subset of preconception health measures.

Maternal and infant mortality are higher in the United States than most developed countries,^1,2 particularly among American women in low-income communities.^3,4 While national policies designed to improve pregnancy and birth outcomes have historically targeted disadvantaged women during pregnancy, there is growing recognition that the period prior to pregnancy also shapes the subsequent health of expectant mothers and their children.^5–8 Maternal risk factors associated with poor pregnancy outcomes often arise years before conception, making intervention during pregnancy “too little, too late.”⁵ In addition, the period following pregnancy provides an important opportunity for continued care related to maternal health and the receipt of family planning services, which may affect the outcomes for subsequent pregnancies.⁹

Expanding health insurance coverage has the potential to increase women’s access to affordable health care during non-pregnancy periods. The Affordable Care Act (ACA) has drastically changed the health policy landscape by introducing new coverage options for low-income women. While all states provided Medicaid coverage during pregnancy prior to the ACA, eligibility criteria for women before and after pregnancy were far more restrictive. Women without children did not qualify for Medicaid in most states and income eligibility thresholds were low for women with children (median state threshold for parental eligibility in 2013 was 64 percent of the federal poverty line (FPL)).¹⁰ This contributed to gaps in coverage and access to health services before and after pregnancy.^11–12

Under the ACA, thirty-six states and the District of Columbia have implemented optional Medicaid expansions for adults with incomes up to 138 percent of the FPL,¹³ providing a new eligibility pathway for women to obtain Medicaid coverage when not pregnant. This change led to an immediate and large increase in insurance coverage among low-income women of reproductive age in expansion states when compared to non-expansion states. In general, women without children experienced the largest changes in insurance coverage, although women with children in states with low parental eligibility thresholds also experienced large increases in insurance coverage.¹⁴ The Medicaid expansions and other ACA coverage expansions improved access to care for reproductive-aged women, including increased access to physicians and fewer cost barriers to receiving care.^14–16 These changes may translate into improved maternal and infant health if they support the prevention and treatment of important medical and behavioral risk factors during non-pregnancy periods.

This paper evaluates the impact of the ACA Medicaid expansions on women’s health before pregnancy using ten key indicators of preconception health. These indicators measure risky behaviors and the prevalence of chronic health conditions among women of reproductive age who may become pregnant, monitor birth control use and pregnancy intention status, and track the use of essential preconception health services among women who become pregnant. Nine of these indicators were identified as prioritized measures for monitoring preconception health status by the National Preconception Health and Health Care (PCHHC) initiative; the last was a core state preconception health indicator identified by the Centers for Disease Control and Prevention (CDC).^17,18

The impact of Medicaid eligibility expansions on preconception health is difficult to predict and could change over time. On the one hand, Medicaid expansions could increase diagnosis of chronic disease by increasing access to care,^19,20 leading to an uptick in measured health risks prior to pregnancy. On the other hand, improved access to tools to identify and manage chronic disease risk factors could influence health behaviors, as well as reduce the true prevalence of health risks in the long run. A recent analysis examined the health of women of reproductive age and found no changes in chronic conditions, smoking cessation, or BMI associated with the ACA Medicaid expansions using data through 2016. The authors did find, however, improved self-reported health and decreased binge drinking associated with expanded Medicaid eligiblity.¹⁶

We extend this line of inquiry by using more recent data (through 2018) and examining changes in a prioritized list of preconception health indicators.^17,18 In addition to measures of chronic conditions and health behaviors among women of reproductive age, we examine changes in the use of essential preconception health services among women who become pregnant. These include preconception health counseling and daily folic acid intake, which is recommended by the United States Preventive Services Task Force (USPSTF) due to the large effect of early pregnancy folic acid consumption on the prevention of neural tube defects among newborns.^21,22 We also examine whether there are changes in pregnancy intention, which may result from increased access to family planning services, and postpartum use of effective methods of contraception. Contraception use following pregnancy may reduce future unintended pregnancies or short interpregnancy intervals, which are associated with increased risk of adverse maternal and infant outcomes.^23,24 For this reason, the postpartum period can represent an important component of the preconception period for subsequent pregnancies.

We also examine changes in insurance status during the prepregnancy and postpartum periods, which are the hypothesized pathways for Medicaid eligibility expansions to improve preconception health. Expanded eligibility for health insurance coverage prior to and following pregnancy may promote more continuous coverage and increase the receipt of recommended health services, including family planning services, preventive care, and treatment for chronic health conditions. Previous studies have documented increased or more stable Medicaid coverage following the ACA Medicaid expansions during both the prepregnancy and postpartum periods.^25–27 We present these estimates for our specific study sample mainly to help with the interpretation of our findings.

STUDY DATA AND METHODS

Data and Sample

Because more than half of births are unintended at the time of conception,²⁸ preconception health indicators measure health not only among women who become pregnant, but also among all nonpregnant women of reproductive age. We used data from two distinct CDC surveillance systems that track each of these populations in our analyses.

First, to measure changes in preconception health among non-pregnant women of reproductive age, we used the 2011–2018 years of the Behavioral Risk Factor Surveillance System (BRFSS), a survey of noninstitutionalized adults in all 50 states and Washington D.C.²⁹ The BRFSS collects detailed information on respondents’ health risk behaviors and chronic health conditions. We selected 2011 as the start of the study period due to significant changes in the sampling scheme in that year.²⁹ Using information on household size and income, we identified women with household incomes at or below 138 percent of the FPL who would be eligible for the Medicaid expansions. Household size was calculated as the sum of the number of children and adults per household. In the BRFSS, income is reported in discrete bins; we used the midpoint of each bin to assign a precise income level to the household.

Second, we examined changes in preconception health among women with a recent live birth using the 2012–2017 years of the Pregnancy Risk Assessment Monitoring System (PRAMS). PRAMS collects information on women’s experiences prior to, during, and following pregnancy using interviews starting 2 to 4 months after delivery. State participation is optional and only states meeting a response rate threshold are included in annual data releases. Our analysis used information for states with data available for the entire study period, yielding a sample of thirteen states (8 expansion and 5 non-expansion states); see Appendix Exhibit C.³⁰ To restrict the sample to women with household incomes at or below 138 percent FPL, we used information on household income and size during the 12 months prior to birth. Income is reported by category and we assigned household income using the midpoint of the category, as in the BRFSS.

Outcomes

Our primary outcomes of interest were ten indicators of preconception health constructed using data available in the BRFSS and PRAMS. The indicators in the BRFSS measured the prevalence of health conditions (ever diagnosed with depression, diabetes, or hypertension) and health behaviors (current smoking, normal weight, and recommended physical activity) among non-pregnant women of reproductive age. Two of these measures, hypertension and recommended physical activity, were only available for a subset of years. Indicators in the PRAMS measured preconception health among women with a recent live birth. These indicators included receipt of preconception health counseling during the 12 months prior to pregnancy; daily folic acid intake; whether the pregnancy was unwanted; and postpartum use of a more effective method of contraception such as sterilization or an implant, intrauterine device, injectable, pill, patch, or vaginal ring. Each of these measures in the PRAMS refers to the preconception or postpartum period for the pregnancy that resulted in a recent live birth.

Secondary outcomes of interest were also constructed to examine changes in insurance coverage during the month prior to pregnancy (pre-pregnancy period) and at the time of the post-pregnancy interview (postpartum period).

Appendix Exhibits A and B include additional details on outcome definitions and availability.³⁰

Study Design

The analyses used a difference-in-differences research design to estimate the effects of the ACA Medicaid eligibility expansions. Under this design, changes in outcomes before and after implementation of ACA Medicaid expansions (first difference) are compared in expansion versus non-expansion states (second difference). This comparison of outcomes in states with vs. without implementation of Medicaid expansions allowed us to control for the role of concurrent national changes affecting both groups of states, such as national expansions in access to private insurance coverage in 2014. The underlying assumption is that outcomes would have evolved similarly in the two groups of states in the absence of the ACA Medicaid expansions.

Expansion states were states that implemented the ACA Medicaid expansion at any point during the study period. The post-expansion period in each state began on the date the expansion was implemented in that state. In total, 27 states expanded Medicaid eligibility in 2014, the first year of the policy, and an additional 5 states expanded Medicaid eligibility over 2015 to 2018. We excluded four states (Delaware, Massachusetts, New York, and Vermont) and Washington, DC due to enactment of similar Medicaid eligibility expansions prior to the ACA. We also excluded Wisconsin, which implemented an alternate Medicaid expansion for adults up to 100 percent of the federal poverty level (FPL) in 2014.¹³

Statistical Analysis

To identify the effects of the ACA Medicaid eligibility expansions, we estimated a linear probability model for each outcome in which the main predictor was an indicator of whether the respondent’s state of residence had expanded Medicaid. In analyses with the BRFSS, expansion status was measured quarterly using available information on year and quarter of the interview. In the PRAMS, information is available on the year of child’s birth only. Therefore, the period of exposure to the Medicaid expansions for all outcomes related to the preconception period was the calendar year prior to birth. For postpartum outcomes, the year of birth was the relevant exposure period. A state was considered to have expanded Medicaid during a given year if implementation occurred prior to July of that year; see Appendix Exhibit C.³⁰

All regressions included state and time indicator variables to control for time-invariant differences across states and secular trends that were unrelated to the ACA Medicaid eligibility expansions. Regressions controlled for respondent age, race/ethnicity, marital status, education level, and the number of children in the household. Regressions also controlled for the annual state unemployment rate, a measure of local economic conditions that could influence poverty and access to employment-based insurance. Analyses of outcomes among women with a recent birth using the PRAMS data additionally controlled for state policies relevant to fertility that changed during our study period, including the maximum welfare benefit for a family of three, an indicator for state mandates for private health insurance coverage of contraceptives, and whether states had Medicaid family planning waivers.

All regressions were weighted using the relevant survey weights. Heteroskedasticity-robust standard errors were clustered by state to account for within-state correlation of the error terms.³¹ In the analyses with PRAMS data, due to the smaller number of states, we used a wild cluster bootstrap procedure to estimate the confidence intervals and p-values.³² In all analyses, a significance level of 0.05 was used to determine statistical significance. Appendix Exhibit D provides additional details on the models and control variables.³⁰

Sensitivity Analyses

We conducted additional analyses to assess the validity of our findings. The assumption underlying the research design is that, in the absence of Medicaid expansion, outcomes would have evolved similarly for the study population in expansion and non-expansion states (i.e. the “parallel trends” assumption). To examine whether there were differential trends in outcomes by expansion status prior to the policy change, we estimated event study models, that is, models that measured the adjusted differences in outcomes between expansion and non-expansion states every year prior to the Medicaid eligibility expansions, and tested for changes in these differences over time. We tested for the joint significance of the pre-expansion coefficients in this model. In an additional check, we tested for differences in adjusted linear trends between expansion and non-expansion states prior to the expansions. See Appendix Exhibit D for further details.³⁰

We also explored whether changes in the preconception health indicators measured in the PRAMS may have reflected changes in the composition of women giving birth as a result of the ACA Medicaid expansions. For example, better access to contraception or family planning services under the Medicaid expansions could affect the timing and frequency of births and, therefore, the characteristics of women observed in the PRAMS. To investigate this, we examined whether there were changes in the observable characteristics (age, race/ethnicity, education, and family composition) of low-income women with a recent live birth associated with the Medicaid eligibility expansions using the difference-in-differences model described above.

Limitations

This study had several limitations. First, although our specification treats Medicaid eligibility expansions as a quasi-experiment, Medicaid expansion was not randomly assigned. Thus, potential confounding from unobserved, concurrent changes that varied by state may have impacted our results. Second, income levels were calculated using the midpoints of discrete bins reported in the BRFSS and PRAMS, leading to some mismeasurement in which respondents were Medicaid expansion eligible, which may attenuate our estimates. Third, while we use the same data sources used by the CDC to track this set of preconception health indicators, these data are subject to self-report bias or recall bias, which could vary in unknown ways across respondents. Fourth, due to limited sample sizes, we did not examine heterogeneity in the impact of Medicaid eligibility expansions on preconception health outcomes by race, ethnicity, parental status, or health status. These would be important directions for future research, given documented differences in perinatal insurance coverage by race, ethnicity, and parental status.^14,33 Fifth, we did not adjust for multiple comparisons in our analyses.

Finally, analyses with the PRAMS used data from participating states only (see Appendix Exhibit C) and may not necessarily represent the experiences of women in other states. In addition, this data source samples women with recent live births and does not allow us to directly test for any changes in fertility in response to the Medicaid expansions. However, in supplemental analyses, we found no significant changes in the observable characteristics of mothers associated with Medicaid expansion during our study period (see Appendix Exhibit E).³⁰ This suggests that changes observed in the following analyses are not driven by changes in the composition of women giving birth associated with the policy change.

RESULTS

Preconception Health Indicators

Exhibit 1 compares the characteristics of low-income women of reproductive age across expansion and non-expansion states prior to Medicaid expansions. A smaller proportion of women in the sample were Hispanic in non-expansion states than expansion states. In addition, women with a recent live birth were younger in non-expansion states than expansion states.

EXHIBIT 1.

Baseline Characteristics of Low-Income Women by Medicaid Expansion Status

	Expansion states	Non-expansion states
Women of Reproductive Age, 20–44
Number of observations	27,118	23,094
Age in years, mean** (SD)	31.4 (6.9)	31.7 (7.3)
Race/ethnicity, %****
White, non-Hispanic	40.6	39.8
Black, non-Hispanic	15.2	25.0
Other, non-Hispanic	8.4	5.0
Hispanic	35.9	30.1
Education, %***
Less than High School	29.1	26.9
High School	30.3	32.5
Some College	32.2	32.7
College or more	8.4	8.0
Family composition
Married, %***	31.8	34.1
Number of children, mean*** (SD)	1.9 (1.4)	2.0 (1.5)
Women with a Recent Live Birth
Number of observations	9,135	5,924
Age, %****
20–24	34.3	43.0
25–29	32.3	32.7
30–34	21.1	16.4
35–39	9.9	6.2
40+	2.4	1.8
Race/ethnicity, %****
White, non-Hispanic	40.8	64.1
Black, non-Hispanic	20.5	12.4
Other, non-Hispanic	9.4	10.3
Hispanic	29.3	13.2
Years of education, %****
<12 years	23.0	21.8
12 years	37.3	37.1
13–15 years	31.4	32.6
>=16 years	8.3	7.9
Family composition
Married, %****	37.8	43.8
Number of previous live births, mean (SD)	1.4 (1.0)	1.3 (13)

SOURCE: Authors’ analysis of PRAMS and BRFSS data.

NOTES: Statistically significant difference in means between the two groups denoted by an asterisk.

^**p < 0.05,

^***p < 0.01,

^****p < 0.001.

Exhibit 2 reports estimated changes in preconception health indicators among low-income women associated with the ACA Medicaid expansions. Among nonpregnant women of reproductive age, there were no significant changes in any of the health indicators (depression, diabetes, hypertension, smoking, weight, and physical activity).

EXHIBIT 2.

Changes in Preconception Health Indicators for Low-Income Women Associated with ACA Medicaid Expansions

	Baseline mean in expansion states (%)	Difference-in-differences estimate
		Percentage Point Change	(95% CI)
a) Women of reproductive age (BRFSS)
Depression	27.5	1.1	(−1.5 to 3.6)
Diabetes	5.1	−0.1	(−0.8 to 0.7)
Hypertension	13.8	0.5	(−1.0 to 2.1)
Current smoking	27.6	−0.7	(−2.3 to 0.9)
Normal weight	35.1	0.8	(−0.2 to 1.8)
Recommended physical activity	46.4	1.6	(−1.7 to 4.8)
b) Women with a recent live birth (PRAMS)
Preconception health counseling	18.1	4.0**	(0.8 to 7.0)
Folic acid intake	19.6	1.9**	(0.2 to 4.8)
Unwanted pregnancy	11.9	−2.1	(−8.5 to 4.5)
Postpartum use of effective contraception	53.4	3.8**	(0.3 to 11.0)

SOURCE: Authors’ analysis of PRAMS and BRFSS data.

NOTES: Weighted regression models adjusted for the covariates noted in the text.

^**p < 0.05,

^***p < 0.01,

^****p < 0.001.

Among low-income women with a recent live birth, there were significant improvements in three preconception health indicators that were associated with Medicaid expansion. There was a significant increase in the share of women reporting a preconception health conversation with a health care provider prior to pregnancy of 4 percentage points, or a 22 percent increase over the baseline mean. In addition, the share of women reporting daily folic acid intake (via a multivitamin, prenatal vitamin, or folic acid supplement) in the month prior to pregnancy increased by 1.9 percentage points, or a 9.7 percent increase over the baseline rate. Finally, the use of effective methods of contraception during the postpartum period increased by 3.8 percentage points, or a 7.1 percent increase over baseline.

Insurance Coverage

Exhibit 3 presents findings from analyses of secondary outcomes among low-income women with a recent live birth. Medicaid coverage increased during the preconception and postpartum periods by 11.1 percentage points and 8.5 percentage points, respectively, after Medicaid expansion. These increases did not, however, translate into statistically significant (p<0.05) decreases in the overall uninsurance rate during these periods.

EXHIBIT 3.

Changes in Prepregnancy and Postpartum Coverage and Utilization for Low-Income Women With a Recent Live Birth Associated with Medicaid Expansion

	Baseline mean in expansion states (%)	Difference-in-differences estimate
		Percentage point change	(95% CI)
Prepregnancy health insurance
No insurance coverage	34.6	−8.2	(−17.7 to 3.9)
Medicaid coverage	38.3	11.1**	(0.7 to 18.1)
Postpartum health insurance
No insurance coverage	23.2	−5.7	(−20.8 to 0.5)
Medicaid coverage	70.5	8.5***	(3.2 to 24.5)

SOURCE: Authors’ analysis of PRAMS data.

NOTES: Weighted regression models adjusted for the covariates noted in the text.

^**p < 0.05,

^***p < 0.01,

^****p < 0.001.

Sensitivity Analyses

Tests for differential pre-period trends suggested the validity of the research design and causal interpretation of these findings. Examining our outcomes of interest prior to the onset of Medicaid eligibility expansions, we found no evidence of significant differences in trends between the two groups of states in the pre-period for the outcomes studied; see Appendix Exhibits F, G, and H.³⁰ These tests, however, do not definitively rule out the presence of differences in pre-trends, which may bias the estimated effects of Medicaid expansion.

DISCUSSION

We provide new evidence that expanded Medicaid coverage led to improvements in addressing certain preconception health risks among low-income women. First, we found that state Medicaid expansions were associated with increased receipt of preconception health counseling - i.e., a conversation targeted to addressing key risks such as smoking, depression, or uncontrolled diabetes and hypertension prior to conception. Our point estimates imply that gaining Medicaid coverage increased the likelihood that a low-income woman received preconception health counseling by 36 percent. This is important because an estimated half of pregnancies in the United States are affected by chronic health and behavioral risk factors that increase the risk for adverse maternal and infant outcomes,^34,35 and for which effective preconception medical interventions exist.^{21,22,36–38}

Second, we document an increase in the proportion of low-income women who reported daily folic acid intake during the month prior to conception associated with Medicaid expansion. Recommended for all women capable of having children by the USPSTF, daily folic acid intake reduces the likelihood of neural tube defects among newborns.^21,22 Our point estimates imply that new Medicaid coverage increased daily folic acid intake by 17 percent among its recipients. And, third, we document that the ACA Medicaid expansions led to increased use of effective birth control methods following childbirth among low-income women, which may decrease the likelihood of future unplanned pregnancies or short inter-pregnancy intervals. We estimate that the gain in postpartum Medicaid coverage increased effective contraception use by 45 percent among new enrollees.

We did not find any evidence, however, that the ACA Medicaid expansions were associated with changes in preconception health indicators related to chronic disease or other health behaviors among low-income women. This is consistent with prior research that found no impact of Medicaid eligibility expansions on BMI or on diagnoses of diabetes or hypertension for women of reproductive age.¹⁶ Our data spans 5 years after the Medicaid expansions; this window may be too short to expect detectable changes in the prevalence of chronic conditions or many health behaviors. While these changes were not statistically significant, the confidence intervals on our estimates included values that could be considered clinically meaningful changes.

Even if coverage does not change the prevalence of risky health conditions in the short term, it can improve uptake of proper treatment and management of these conditions,¹⁹ which may also have important implications for maternal and infant health. Supporting this possibility, prior research suggests that treatment of diabetes increased under the ACA Medicaid expansions,^39–41 as did treatment of diabetes and hypertension among women of reproductive age even as diagnosis rates for these conditions did not change.¹⁶

These findings have direct relevance to ongoing policy discussions. We provide the first evidence that the ACA Medicaid expansions led to increased preconception health counseling, folic acid intake, and effective postpartum contraception use. In addition, our results for contraception use, which asked respondents about their utilization at the time of interview (at least 2–4 months after delivery), indicate that expanded Medicaid coverage also affects postpartum health care use. Expanded postpartum coverage, such as the national extensions in postpartum Medicaid coverage that have been recently proposed as part of a bipartisan bill in Congress,⁴² may help to increase continuity of insurance coverage during this critical window and promote the receipt of recommended postpartum care.²⁶

In sum, ACA Medicaid expansions - a public policy still under active debate in many states - can help address some of the risk factors that shape maternal and child health in the United States.