Association of Medicaid vs Marketplace Eligibility on Maternal Coverage and Access With Prenatal and Postpartum Care

Key Points Question How does preconception and postpartum coverage and care use compare among women with low incomes gaining eligibility for Medicaid vs marketplace coverage? Findings In this cohort study of 11 432 women with low incomes, Medicaid eligibility relative to marketplace eligibility was associated with significantly increased Medicaid coverage (20.3 percentage points), decreased private insurance coverage (−10.8 percentage points), and decreased uninsurance (−8.7 percentage points) in the preconception period. It was also associated with increased postpartum Medicaid (17.4 percentage points) and increased adequate prenatal care (4.4 percentage points) but not with significant changes in early prenatal care, postpartum checkups, or postpartum contraception. Meaning The finding of lower rates of preconception uninsurance among Medicaid-eligible women in this study suggests that women with low incomes were facing barriers to marketplace enrollment.

The β1 coefficient estimates changes in outcomes among women gaining Medicaid eligibility relative to Marketplace eligibility after policy implementation. Yit represents the dependent variable. γ`xit is a vector of control variables, including the state unemployment rate for women from the Bureau of Labor Statistics, mother's age, marital status, race, Hispanic ethnicity, and high school educational attainment.
The table below presents an analysis of the difference-in-differences assumption that the trends prior to policy implementation are similar for mothers gaining Medicaid and Marketplace eligibility. This pre-policy trend comparison follows the approach of other health insurance studies that tests pre-trends for difference-in-differences models. Analysis is limited to prior to 2014 to compare parallel trends in the pre-period, and each outcome is modeled as an interaction of a linear time trend and treatment group status to assess significant differences in trends prior to the policy implementation. The β2 coefficient, presented in the table, indicates any significant difference in pre-policy trends between mothers gaining Medicaid and Marketplace eligibility, as specified: yit= β0 + β1LinearTimeTrendt + β2TreatmentStatusi *LinearTimeTrendt + (StateFE)i + γ`xit+ +ɛit  (2013). Data are weighted using PRAMS sample weights. Estimates expressed as percentage points. Estimates are adjusted for mother's age, marital status, race/ethnicity, educational attainment, state level rates of unemployment for women, and state and year fixed effects. Fstatistics represent joint tests of the null hypothesis that all pre-policy interaction terms are equal to zero.

eFigure2. Plots of Placebo Test Coefficients
Estimates expressed as percentage points. Error bars represent 95% CIs. Placebo test coefficients represent the difference in the outcome in Medicaid states relative to Marketplace states in each year compared to the last year of the pre-policy period (2013). Data are weighted using PRAMS sample weights. Estimates are adjusted for mother's age, marital status, race/ethnicity, educational attainment, state level rates of unemployment for women, and state and year fixed effects. Authors' analysis of Pregnancy Risk Assessment Monitoring System data from fourteen states and New York City. Sample includes low-income women 100-138 percent of the federal poverty level.  (-11.4, 30.4) 0.780 0.594 Notes: Authors' analysis of Pregnancy Risk Assessment Monitoring System data from fourteen states and New York City. Sample includes lowincome women 100-138 percent of the federal poverty level. Women 18-26 years old are identified using birth year and year of survey, as maternal age is measured through categories that do not align with excluding women under the age of 26, and may therefore not perfect capture women in this age group. Data are weighted using PRAMS sample weights. Estimates are presented as percentage points. Model is adjusted for marital status, race/ethnicity, educational attainment, state level rates of unemployment for women, and state and year fixed effects. P-values and confidence intervals are derived from wild bootstrap cluster standard errors by state. Boldface indicates p<0.05. †p<0.10, *p<0.05, **p<0.01, ***p<0.001

eTable7. Estimates of the Effects of Medicaid versus Marketplace Eligibility on Maternal Coverage and Access to Prenatal and Postpartum Care, Omitting States with Generous Pre-ACA Medicaid Eligibility
Medicaid  (-5.7, 22.7) 0.481 0.421 Notes: Authors' analysis of Pregnancy Risk Assessment Monitoring System data from fourteen states and New York City. Sample includes low-income women 100-138 percent of the federal poverty level. Data are weighted using PRAMS sample weights. Estimates are presented as percentage points. Model is adjusted for marital status, race/ethnicity, educational attainment, state level rates of unemployment for women, and state and year fixed effects. P-values and confidence intervals are derived from wild bootstrap cluster standard errors by state. Boldface indicates p<0.05. †p<0.10, *p<0.05, **p<0.01, ***p<0.001

eTable9. Estimates of the Effects of Medicaid versus Marketplace Eligibility on Maternal Coverage and Access to Prenatal and Postpartum Care, Excluding Women with PRAMS Follow-Up within 2 Months
Medicaid  (<= 8; 9-11; 12+). The lowest number is assigned for each category (1,9,12). Data are weighted using PRAMS sample weights. Model is adjusted for mother's age, marital status, educational attainment, state level rates of unemployment for women, and state and year fixed effects. P-values and confidence intervals are derived from wild bootstrap cluster standard errors by state. Boldface indicates p<0.05. †p<0.10, *p<0.05, **p<0.01, ***p<0.001 Notes: PRAMS survey responses indicating non-group private insurance changed over the study period. As a result, our ability to capture specific increases in Marketplace coverage are limited by the change in this measure 2011-2018. Examining rates for this non-group private insurance separately from the broader private insurance category resulted in lower reports of non-group private insurance compared to other estimates for this population, such as those presented in Blavin F, Karpman M, Kenney GM, Sommers BD. Medicaid versus marketplace coverage for near-poor adults: Effects on out-of-pocket spending and coverage. Health Aff (Millwood). 2018;37 (2): 299-307. doi:10.1377/hlthaff.2017.1166. For these reasons, we use a measure of private insurance in our analyses that encompasses all PRAMS private coverage variables.