Before the Affordable Care Act (ACA), an estimated 25% of reproductive-aged women in the United States reported no insurance in the past year.1 Although low-income pregnant women were eligible for Medicaid in all states before the ACA, research suggests that low-income women experienced high rates of coverage changes, known as “churning,” in the period surrounding pregnancy.2 From 2005 to 2013, more than half of those who held pregnancy-related Medicaid in the month of delivery experienced a change in coverage or at least 1 month of uninsurance in the 12 months before delivery.2 Because preconception care is associated with earlier initiation and more adequate prenatal care, uninsurance may limit a woman’s access to these important services.3–6
Before the ACA, the income eligibility for Medicaid for nonpregnant women varied widely by state. The ACA included federal support for the expansion of state Medicaid programs to all nonelderly adults with incomes at or below 138% the federal poverty level for those states that elected to participate.7 Several studies of these expansions found that they resulted in increased Medicaid enrollment and preventive service utilization such as primary care visits in the nonelderly adult population.8 The objective of this study was to determine the effects of the ACA Medicaid expansions on insurance coverage among low-income women 1 month before pregnancy. We hypothesized that preconception Medicaid enrollment would increase among women at or below 138% the federal poverty level in states that expanded Medicaid relative to those states that did not.
MATERIALS AND METHODS
We used a quasiexperimental, difference-in-difference design to compare changes in preconception insurance coverage among women with incomes at or below 138% the federal poverty level in expansion compared with nonexpansion states before and after the Medicaid expansions. The difference-in-difference design uses longitudinal data to compare the effect of an intervention between an exposed and unexposed group. The differences in the outcomes preintervention and postintervention are calculated for both the exposed and unexposed groups and then compared; differences between the exposed and unexposed groups are attributed to the effect of the intervention.
We used individual-level data from the Pregnancy Risk Assessment Monitoring System, which is administered by the Centers for Disease Control and Prevention and overseen by state health departments.9 The Pregnancy Risk Assessment Monitoring System data set comprises survey responses from a sample of women who had a livebirth; survey data are paired with birth certificate data. States that expanded their Medicaid program after January 1, 2014, or had similar Medicaid coverage eligibility before 2014 were excluded to isolate the effect of the policy implementation. Year 2014 was considered the policy transition period and excluded in the main analyses, because the index time was childbirth in our study, and as a result of the duration of pregnancy, it may take months for the Medicaid expansion to have a potential effect on preconception coverage. Patients from states with data available in the prepolicy analysis period (2009–2013) and the postpolicy analysis period (2015) were included. The primary outcome was self-reported coverage in the month before conception, categorized as “uninsured,” “Medicaid,” or “non-Medicaid.” We compared patient demographics (maternal age, race, Hispanic ethnicity, education, marital status) before the policy between expansion and nonexpansion states using χ2 tests to understand differences in the baseline characteristics of women in both groups.
We used multivariate linear regression models to compare the changes in preconception coverage between the nonexpansion and expansion states among women whose reported household incomes were at or below 138% the federal poverty level. The regression coefficients estimated for the interaction terms between expansion status and the postpolicy period represent the mean difference in the outcome between expansion and nonexpansion states before and after the policy (ie, the difference-in-difference estimate). Adjusted models controlled for patient demographics (maternal age, race, Hispanic ethnicity, education, marital status), state and year fixed effects, and annual state unemployment rates among reproductive-aged women.10 Because insurance coverage is strongly correlated with employment, state unemployment rates were included to adjust for extrinsic economic factors that may be influencing uninsurance rates during the same time period. We applied survey weights provided by the Centers for Disease Control and Prevention to account for the Pregnancy Risk Assessment Monitoring System sampling design and estimated Huber-White robust standard errors clustered at the state level.
The main analysis group was women with incomes at or less than 138% the federal poverty level, because women in this group were newly eligible for Medicaid in expansion states. We expected any association between the outcome and the expansion to be concentrated among women in this income range. The outcome was also tested in two other subgroups of women: 1) all women, regardless of income and 2) women with prenatal Medicaid coverage. The outcome was tested in all women to compare the effects of Medicaid expansion in the larger population, because overall population uninsurance rates are commonly tracked and reported. Second, we selected the subgroup of women with Medicaid prenatal coverage to determine whether the Medicaid expansion potentially resulted in more women experiencing continuous Medicaid coverage between the preconception and prenatal periods. We conducted additional analyses to test the assumptions of our model and assess the sensitivity of the findings, including testing for differential prepolicy trends and reestimating the models without state unemployment rates and with an alternate study period definition. A full description of the data source, models, and sensitivity and subgroup analyses can be found in Appendix 2, available online at http://links.lww.com/AOG/B194.
P values <.05 were considered statistically significant. All analyses were performed in StataSE 14.1. The Partners Healthcare Human Research Committee exempted this study from review because this study used an existing, deidentified, publicly available data set.
There were 76,587 women from eight expansion states (Hawaii, Illinois, Maryland, Michigan, New Jersey, Oregon, Washington, West Virginia) and 61,910 from seven nonexpansion states (Arkansas, Maine, Missouri, Nebraska, Oklahoma, Utah, Wyoming) in the data set. The primary analysis was conducted in the subset of these women who were low income: 30,495 in expansion states and 26,561 in nonexpansion states. The states and sample sizes for each year are listed in Appendix 3, available online at http://links.lww.com/AOG/B194. Table 1 compares the patient characteristics in the prepolicy period between the nonexpansion and expansion states among women with incomes 138% the federal poverty level or less. The most notable difference was the distribution of education levels between the two groups: women in the expansion states had attained higher levels of education compared with those in nonexpansion states.
Figure 1 shows the unadjusted trends of preconception insurance coverage in low-income women. Before the expansion, uninsurance rates, Medicaid enrollment, and private insurance coverage followed the same trends in both the expansion and nonexpansion groups. The unadjusted trends for the subgroups are shown in Appendix 4, available online at http://links.lww.com/AOG/B194.
Table 2 reports the unadjusted and adjusted results from the difference-in-difference models. The percent of women with preconception Medicaid coverage was 30.8% prepolicy and 35.6% postpolicy in nonexpansion states and 43.2% prepolicy and 56.8% postpolicy in expansion states. There was a significantly greater increase in Medicaid coverage in expansion states after the policy implementation (unadjusted difference-in-difference estimate +8.5% points, 95% CI 1.2–15.9; adjusted difference-in-difference estimate +8.6, 95% CI 1.1–16.0). Rates of preconception uninsurance among low-income women were 44.2% prepolicy and 34.3% postpolicy in nonexpansion states and 37.4% prepolicy and 23.5% postpolicy in expansion states. There was no significant difference in the changes in uninsurance between the two groups in the postpolicy period (unadjusted difference-in-difference estimate −3.9% points, 95% CI −10.6 to 2.8; adjusted difference-in-difference estimate −4.1, 95% CI −11.1 to 2.9). In this population of low-income women, non-Medicaid insurance coverage was 25.3% prepolicy and 30.5% postpolicy in nonexpansion states and 19.4% prepolicy and 19.7% postpolicy in expansion states. Relative to nonexpansion states, there was a significant decrease in non-Medicaid coverage in the expansion states in the postpolicy period (unadjusted difference-in-difference estimate −4.8, 95% CI −8.5 to −1.2; adjusted difference-in-difference estimate −4.7, 95% CI −8.3 to −1.1).
Table 3 presents the difference-in-difference DID estimates for the two subgroup populations: all women in the data set, regardless of income, and women who had Medicaid coverage for their prenatal care. In all women, regardless of income, enrollment in Medicaid was 16.6% prepolicy and 18.5% postpolicy in nonexpansion and 21.5% prepolicy and 28.1% postpolicy in expansion states. Medicaid enrollment increased more in the expansion states than the nonexpansion states in the postpolicy period (unadjusted difference-in-difference estimate +4.8% points, 95% CI 1.0–8.5; adjusted difference-in-difference estimate +4.7, 95% CI 0.0004–9.4). Rates of uninsurance were 26.1% prepolicy and 19.2% postpolicy in nonexpansion and 21.8% prepolicy and 14.2% postpolicy in expansion states. There was no significant differences in the rates of uninsurance in the expansion compared with nonexpansion states postpolicy (unadjusted difference-in-difference estimate +0.2, 95% CI −3.6 to 3.9; adjusted difference-in-difference estimate +0.5, 95% CI −4.1 to 5.2). Non-Medicaid coverage was 16.6% prepolicy and 18.5% postpolicy in nonexpansion and was 21.5% prepolicy and 28.1% postpolicy in expansion states. Relative to nonexpansion states, the non-Medicaid coverage decreased in the expansion states among all women (unadjusted difference-in-difference estimate −5.0, 95% CI −7.2 to −2.7; adjusted difference-in-difference estimate −5.4, 95% CI −10.1 to −0.6).
In the group of women who had prenatal Medicaid coverage, the percent with preconception Medicaid coverage was 33.6% prepolicy and 39.1% postpolicy in nonexpansion states and 45.7% prepolicy and 60.8% postpolicy in expansion states. There was a significant increase in preconception Medicaid coverage in expansion states compared with nonexpansion states after the policy was implemented (unadjusted difference-in-difference estimate +9.4% points, 95% CI 2.1–16.7; adjusted difference-in-difference estimate +9.8, 95% CI 1.1–18.6). Rates of uninsurance were 45.5% prepolicy and 34.3% postpolicy in nonexpansion states and 38.7% prepolicy and 24.6% postpolicy in expansion states. There were no differences in the rates of uninsurance between the nonexpansion and expansion states after the policy (unadjusted difference-in-difference estimate −2.8, 95% CI −9.9 to 4.4; adjusted difference-in-difference estimate −3.2, 95% CI −11.4 to 5.1). The percent of women with Non-Medicaid coverage was 21.2% prepolicy and 27.0% postpolicy in nonexpansion states and 15.6% prepolicy and 14.6% postpolicy in expansion states. Relative to nonexpansion states, non-Medicaid coverage decreased postpolicy in expansion states in women with prenatal Medicaid coverage (unadjusted difference-in-difference estimate −6.8, 95% CI −8.3 to −5.2; adjusted difference-in-difference estimate −6.8, 95% CI −9.8 to −3.8).
In the multiple sensitivity analyses, we found that the main results were robust to alternate model specifications and study period definitions. The results from these analyses are found in Appendices 5–11, available online at http://links.lww.com/AOG/B194.
In this study, we examined the association between the ACA Medicaid expansions and insurance in the month before conception. We found an 8.6% point increase in preconception Medicaid coverage among low-income women who had a livebirth residing in states that expanded their Medicaid programs compared with those in nonexpansion states, representing a 20% increase relative to the prepolicy baseline. Uninsurance rates decreased in both expansion and nonexpansion states in the postpolicy period; however, we did not detect a significant difference between the rates of change between the two groups. Relative to nonexpansion states, non-Medicaid coverage decreased by 4.1 percentage points in expansion states. In the subgroup analysis of women with prenatal Medicaid coverage, we found a 9.8% point increase in preconception Medicaid coverage among women in expansion states relative to nonexpansion states. This increase represents a 21% increase from baseline and suggests that Medicaid expansion was associated with greater continuity of Medicaid coverage from the preconception to pregnancy period for these women.
Prior studies have demonstrated the effects of health insurance on increasing preventive care, medication adherence, and self-reported health in other fields; it is likely that these findings are generalizable to reproductive-aged women and preconception care.11 Improved access to preconception care can help women to appropriately plan their pregnancy and optimize their health before conceiving. Interventions in the preconception period such as disease screening, disease management, and exposure avoidance counseling reduce pregnancy risks and are associated with improved maternal and fetal outcomes.3–5 Furthermore, preconception care is associated with earlier initiation and more adequate prenatal care.6,12
We found the increase in preconception Medicaid coverage was partially accounted for by a decrease in private insurance. Such private insurance “crowd out,” in which privately insured individuals switch to Medicaid once they become newly eligible, has been demonstrated in some studies of the Medicaid expansion in other populations, although results have been mixed.8,13 Crowd out could be beneficial in this population because fewer low-income women may experience coverage changes from non-Medicaid insurance preconception to Medicaid during their pregnancy. Switching between insurance types is associated with access barriers and care delays.14 Furthermore, compared with private coverage, Medicaid coverage has lower cost-sharing and provides additional benefits such as social services, nutrition, and transportation in some states, which may reduce cost-related and other access barriers for low-income women.
This study has several limitations. First, difference-in-difference analyses rely on the assumption that the outcome trends among expansion and nonexpansion states would have evolved similarly in the postpolicy period if not for Medicaid expansion. Although this assumption is not directly testable, our validity checks suggest that the potential bias resulting from differential trends is small relative to the difference-in-difference effects observed. However, we are unable to fully exclude the possibility that differential changes in the outcomes over time (eg, as a result of differentially changing demographics or other policies) biased our results. Second, this analysis relies on self-reported survey data that were collected after delivery, introducing the possibility of recall bias or inaccurate responses on preconception insurance coverage. A validation study of the Pregnancy Risk Assessment Monitoring System data noted that women accurately reported their delivery insurance, although preconception insurance was not specifically studied.15 Third, states’ annual response rates for the Pregnancy Risk Assessment Monitoring System survey are not published, although a minimum threshold set by the Centers for Disease Control and Prevention must be met to be published; the results could be biased if the characteristics of responders changed differentially between the nonexpansion and expansion groups over time. Fourth, income data were missing in 5.8% and 9.6% of patients in the nonexpansion and expansion groups, respectively, which, in combination with their reported number of dependents, was used to determine the relation to the federal poverty level. The sensitivity analyses regarding income demonstrate that the missing data are unlikely to bias our findings. In addition, the retrospective design of the Pregnancy Risk Assessment Monitoring System survey limited the postpolicy analysis to 1 year, because the preconception period for most women who delivered in 2014 was in the prepolicy period. Finally, our ability to meaningfully assess the effects of preconception coverage changes on health care and birth outcomes was limited by the sample size of the data set.
In summary, this study demonstrates the association between the ACA Medicaid expansion on preconception coverage among women with livebirths. Among states that expanded their Medicaid eligibility to women at or below 138% of the federal poverty level, there was a significant increase in low-income women enrolled in Medicaid before pregnancy. Subsequent studies should determine the significance of increased preconception Medicaid coverage among low-income women on the use of preconception care services, preconception health status, access to and quality of prenatal care, and, ultimately, birth outcomes.
1. Kozhimannil KB, Abraham JM, Virnig BA. National trends in health insurance coverage of pregnant and reproductive-age women, 2000 to 2009. Womens Health Issues 2012;22:e135–41.
2. Daw JR, Hatfield LA, Swartz K, Sommers BD. Women in the United States experience high rates of coverage ‘churn’ in months before and after childbirth. Health Aff (Millwood) 2017;36:598–606.
3. Atrash HK, Johnson K, Adams M, Cordero JF, Howse J. Preconception care for improving perinatal outcomes: the time to act. Matern Child Health J 2006;10(suppl):S3–11.
4. Stephenson J, Heslehurst N, Hall J, Schoenaker DAJM, Hutchinson J, Cade JE, et al. Before the beginning: nutrition and lifestyle in the preconception period and its importance for future health. Lancet 2018;391:1830–41.
5. Recommendations to improve preconception health and health care—United States: a report of the CDC/ATSDR preconception care work group and the select panel on preconception care. Available at: https://www.cdc.gov/mmwr/preview/mmwrhtml/rr5506a1.htm
. Retrieved May 11, 2018.
6. Wally MK, Huber LRB, Issel LM, Thompson ME. The association between preconception care receipt and the timeliness and adequacy of prenatal care: an examination of multistate data from Pregnancy Risk Assessment Monitoring System (PRAMS) 2009–2011. Matern Child Health J 2018;22:41–50.
7. U.S. Congress. Patient protection and affordable care act. 42:2010.
8. Miller S, Wherry LR. Health and access to care during the first 2 years of the ACA Medicaid expansions. N Engl J Med 2017;376:947–56.
9. Centers for Disease Control and Prevention (CDC). Pregnancy risk assessment monitoring system. 2017. Available at: https://www.cdc.gov/prams
. Retrieved September 27, 2017.
10. U.S. Department of Labor, Bureau of Labor Statistics. Databases, tables & calculators by subject. Available at: https://www.bls.gov/data/
. Retrieved February 23, 2018.
11. Sommers BD, Gawande AA, Baicker K. Health insurance coverage and health—what the recent evidence tells us. N Engl J Med 2017;377:586–93.
12. Braveman P, Marchi K, Egerter S, Pearl M, Neuhaus J. Barriers to timely prenatal care among women with insurance: the importance of prepregnancy factors. Obstet Gynecol 2000;95:874–80.
13. Antonisse L, Garfield R, Rudowitz R, Artiga S. The effects of Medicaid expansion under the ACA: updated findings from a literature review. Henry J. Kaiser Family Foundation; 2018. Available at: https://www.kff.org/medicaid/issue-brief/the-effects-of-medicaid-expansion-under-the-aca-updated-findings-from-a-literature-review-march-2018/
. Retrieved May 22, 2018.
14. Lavarreda SA, Gatchell M, Ponce N, Brown ER, Chia YJ. Switching health insurance and its effects on access to physician services. Med Care 2008;46:1055–63.
15. Dietz P, Bombard J, Mulready-Ward C, Gauthier J, Sackoff J, Brozicevic P, et al. Validation of self-reported maternal and infant health indicators in the Pregnancy Risk Assessment Monitoring System. Matern Child Health J 2014;18:2489–98.