An Analysis of Payment Mix Patterns of Preterm Births in a Post-Affordable Care Act Insurance Market: Implications for the Medicaid Program

  • Rose Meltzer
    Affiliations
    Avalere Health, Washington, DC
    Search for articles by this author
  • Anne Rossier Markus
    Correspondence
    Correspondence to: Anne Rossier Markus, PhD, MHS, JD, Chair and Professor, Department of Health Policy and Management, Milken Institute School of Public Health, The George Washington University, 950 New Hampshire Ave. NW, Washington, DC 20052. Phone: 202-994-4139; fax: 202-994-4040.
    Affiliations
    Department of Health Policy and Management, Milken Institute School of Public Health, The George Washington University, Washington, DC
    Search for articles by this author
Open AccessPublished:June 03, 2020DOI:https://doi.org/10.1016/j.whi.2020.04.003

      Abstract

      Background

      The United States has a relatively high preterm birth rate compared with other developed nations. Before the enactment of the Affordable Care Act in 2010, many women at risk of a preterm birth were not able to access affordable health insurance or a wide array of preventive and maternity care services needed before, during, and after pregnancy. The various health insurance market reforms and coverage expansions contained in the Affordable Care Act sought in part to address these problems. This analysis aims to describe changes in the patterns of payer mix of preterm births in the context of a post-Affordable Care Act insurance market, explore possible factors for the observed changes, and discuss some of the implications for the Medicaid program.

      Methods

      We applied a repeated cross-sectional study design to explore payment mix patterns of all births and preterm births between 2011 and 2016, using publicly available National Vital Statistics Birth Data. We included an equal number of years with payment source available in the dataset before and after January 1, 2014, when the coverage expansions became effective.

      Results

      We found a small relative change in payment mix during the study period. Private health insurance (PHI) paid for a higher percentage of all births and this rate increased steadily between 2011 and 2016. Preterm births paid by PHI increased by 1.4 percentage points between 2011 and 2016 and self-pay/uninsured preterm births decreased by 0.3 percentage points over the same time period. Medicaid had the highest, and a relatively stable, preterm birth coverage percentage (48.9% in 2011, 49.2% in 2014, and 48.9% in 2016). Medicaid was also more likely to pay for preterm births than PHI, but this likelihood decreased by more than one-half after 2014 (8.2% in 2013 vs. 3.8% in 2014).

      Conclusions

      After the 2010 reforms, Medicaid remained a constant source of coverage for the most vulnerable women in society when faced with the high cost of a preterm birth. Nationwide, of the 64 million women ages 15 to 44, 4% gained PHI (directly purchased or employer sponsored) and another 4% Medicaid, with a concomitant 8% decrease in uninsured women of reproductive age between 2013 and 2017. More research is needed to conclude with certainty that the reforms worked as intended, but the important role of Medicaid as a financial safety net is undeniable.
      Of the 3.7 million registered births that occurred in the United States in 2018, 379,171 were preterm (babies born before 37 weeks of gestation) (
      • March of Dimes
      2019 March of Dimes Report Card. Washington, DC.
      ,
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). The preterm birth rate in the United States decreased steadily between 2007 and 2015, from 10.4% to 9.6% (
      • Ferré C.
      • Callaghan W.
      • Olson C.
      • Sharma A.
      • Barfield W.
      Effects of maternal age and age-specific preterm birth rates on overall preterm birth rates — United States, 2007 and 2014.
      ,
      • March of Dimes
      2019 March of Dimes Report Card. Washington, DC.
      ,
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      Births in the United States, 2016. U.S. Department of Health and Human Services.
      ). However, it increased for the next 3 years, reaching 10% in 2018 (
      • March of Dimes
      2019 March of Dimes Report Card. Washington, DC.
      ). Although not all babies born preterm are unhealthy, they face an overall higher risk of short- and long-term health problems than babies born at 40 weeks. Respiratory distress, digestive issues, vision and hearing loss, and cerebral palsy are more common among preterm infants (
      • March of Dimes
      Long-term health effects of premature birth. Washington, DC.
      ). Later, children born preterm are more likely to have developmental and learning disabilities (
      • March of Dimes
      Long-term health effects of premature birth. Washington, DC.
      ). Because preterm birth can affect an individual for life, the Centers for Disease Control and Prevention (CDC) calls reducing the preterm birth rate “a national public health priority” (
      • Centers for Disease Control and Prevention (CDC)
      ).
      The United States spends much more on health care than comparable countries. At the same time, its poorer health outcomes contribute to excess spending. In 2007, the then-Institute of Medicine estimated that preterm births cost American society nearly $26.2 billion (in 2005 dollars; the equivalent of $34.2 billion in 2019 dollars) (
      Preterm birth: Causes, consequences, and prevention.
      ,
      U.S. Bureau of Labor Statistics
      CPI inflation calculator. Washington, DC.
      ). Of that amount (in 2005 dollars), $16.9 billion was attributed to direct medical care, $1.9 billion to labor and delivery, $611 million to early intervention services, and $1.1 billion to special education programs. According to the
      • March of Dimes
      Medicaid covers fewer births but more preterm births than private coverage - Study by GWU, March of Dimes shows disproportionate percentage of preterm births are covered by Medicaid. Washington, DC.
      , “a significant portion of the estimated $26 billion in annual costs associated with preterm birth likely accrues to Medicaid programs.” The nation also indirectly lost nearly $5.7 billion in household and labor market productivity (
      Preterm birth: Causes, consequences, and prevention.
      ). In terms of the financial burden on insurers, a more recent study found that preterm births cost employer-sponsored health plans $6 billion annually (
      • Grosse S.D.
      • Waitzman N.J.
      • Yang N.
      • Abe K.
      • Barfield W.D.
      Employer-sponsored plan expenditures for infants born preterm.
      ). Although no national figure of Medicaid cost burden exists, one study estimates that at least $4 billion in annual costs could be averted by the program overall if preterm births were significantly reduced (
      • Markus A.
      • Krohe S.
      • Garro N.
      • Gerstein M.
      • Pellegrini C.
      Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
      ).
      Several factors increase a woman's risk of giving birth before 40 weeks of gestation (
      • Shapiro-Mendoza C.K.
      • Barfield W.D.
      • Henderson Z.
      • James A.
      • Howse J.L.
      • Iskander J.
      • Thorpe P.G.
      CDC Grand Rounds: Public health strategies to prevent preterm birth.
      ). Women with certain health conditions, including hypertension and diabetes, are more likely to deliver preterm than women who do not have these conditions (
      • Goldenberg R.L.
      • Culhane J.F.
      • Iams J.D.
      • Romero R.
      Epidemiology and causes of preterm birth.
      ) Lifestyle factors—smoking, drinking alcohol, stress, being overweight or underweight—can also increase this risk (
      • March of Dimes
      Preterm labor and premature birth: Are you at risk? Washington, DC.
      ). When pregnant women receive prenatal care, they can lower their risk of delivering preterm, as they can monitor their baby's health and be advised to improve their own (
      • Vintzileos A.M.
      • Ananth C.V.
      • Smulian J.C.
      • Scorza W.E.
      • Knuppel R.A.
      The impact of prenatal care in the United States on preterm births in the presence and absence of antenatal high-risk conditions.
      ). Having health insurance coverage mitigates the financial burden of paying for the cost of prenatal care and other maternity-related services, particularly labor and delivery, which on average cost $4,600 in 2014 (
      • McDermott K.W.
      • Elixhauser A.
      • Sun R.
      Trends in hospital inpatient stays in the United States, 2005–2014. HCUP Statistical Brief #225.
      ).
      In 2010, Congress adopted and the president signed the Affordable Care Act (ACA). The ACA contains a number of health insurance market reforms (such as benefit mandates, prohibition of coverage exclusions for preexisting conditions, and premium limits). Several of these reforms are particularly important and relevant for women of reproductive age (defined as the period from menarche to menopause). Since Fall 2010, all new health plans must cover without cost sharing several federally recommended preventive services, including contraception care. A number of these preventive services are federally recommended for pregnant women (e.g., screens for anemia, hepatitis B, chlamydia, gonorrhea, syphilis, bacteriuria, Rh incompatibility, gestational diabetes; alcohol and tobacco counseling and cessation interventions; and folic acid supplements). As of 2013, states can receive a 1-percentage point increase in their Federal Medical Assistance Percentage if they offer preventive services without cost sharing in their traditional (i.e., pre-ACA) Medicaid program, which includes the pregnancy program. Effective January 1, 2014, women can no longer be refused coverage because they are pregnant, which insurers considered a preexisting condition before the reforms (,
      • Claxton G.
      • Pollitz K.
      • Semanskee A.
      • Levitt L.
      Would states eliminate key benefits if AHCA waivers are enacted? Kaiser Family Foundation.
      ). They also can no longer be subject to higher premiums compared with men or because of their health status. Finally, since Fall 2010, young women have been able to stay covered under their parents’ private plans until they turn 26 years old.
      Since Fall 2013 and annually since then, women regardless of state of residence who do not qualify for Medicaid or do not have access to an affordable employer-sponsored health insurance policy have been able to directly purchase coverage through new, federally or state run, health insurance Marketplaces. Coverage begins on January 1 of the following calendar year, starting with January 1, 2014, as the first year when this coverage expansion became effective. Women can also access public subsidies to lower their premiums and cost-sharing contributions if enrolled in a Marketplace plan and if their income is between 100 and 400% or 100 and 250% of the federal poverty level (FPL), respectively. In addition, this Marketplace coverage must meet certain benefit requirements, such as covering maternity and newborn care and preventive services, including contraception. Among the 97.4 million women ages 19–64 in 2017, 9% purchased insurance in the individual market, up from 7% in 2013 (
      Kaiser Family Foundation
      Women’s health insurance coverage. Fact Sheet. Washington, DC.
      ).
      Effective January 1, 2014, women in states that have expanded Medicaid to nonelderly adults, including parents who have dependent children, are eligible to enroll in the program as long as their income is below 133% FPL, or 138% with the 5% income disregard (
      Kaiser Family Foundation
      Medicaid income eligibility limits for adults as a percent of the federal poverty level.
      ). The income disregard in effect increases the coverage reach of the program and therefore allows for a greater number of women to enroll in the program. The ACA also required states to apply the 5% income disregard to traditional populations, in effect raising coverage levels from 133% to 138% FPL for pregnant women, which again allows for an expansion of coverage for women (
      Medicaid and CHIP Payment and Access Commission (MACPAC)
      Maternity Services: Examining Eligibility and Coverage in Medicaid and CHIP.
      ). Women applying for coverage under the adult expansion are not eligible for that coverage if they apply while pregnant. Furthermore, if they indicate that they are pregnant at the time of application, states should automatically enroll them in the traditional pregnant women group and will receive federal funds at the normal matching rate, which is lower than the matching rate for the expansion adult group (
      Medicaid and CHIP Payment and Access Commission (MACPAC)
      Maternity Services: Examining Eligibility and Coverage in Medicaid and CHIP.
      ). Thus, per federal law, if an expansion woman becomes pregnant once covered by Medicaid, she should be covered under the pregnant women category under her state's Medicaid program, where she has access to a wide array of pregnancy-related services (
      • Medicaid and CHIP Payment and Access Commission (MACPAC)
      Issues in pregnancy coverage under Medicaid and exchange plans.
      ). However, a Congressionally mandated report of the MACPAC found that states are not required to track the pregnancy status of women enrolled through the new adult group, so it is likely that some women who become pregnant will stay enrolled as expansion adults (
      Medicaid and CHIP Payment and Access Commission (MACPAC)
      Maternity Services: Examining Eligibility and Coverage in Medicaid and CHIP.
      ). A pregnant woman could still request to switch to the eligibility group for pregnant women if the benefits offered are more generous than those offered under the expansion adult group (
      Medicaid and CHIP Payment and Access Commission (MACPAC)
      Maternity Services: Examining Eligibility and Coverage in Medicaid and CHIP.
      ). A recent study, however, found that coverage of pregnancy-related services seemed to be generally consistent across eligibility categories based on self-reported information by states (
      • Gifford K.
      • Walls J.
      • Ranji U.
      • Gomez I.
      Medicaid coverage of pregnancy and perinatal benefits: Results from a state survey.
      ). After childbirth, a woman may reenroll in the adult expansion part of the program if still income eligible. However, transitions into and out of Medicaid (known as “churn”) owing to pregnancy and postpartum status have been documented as an issue in the past (
      • Daw J.
      • Hatfield L.
      • Swartz K.
      • Sommers B.
      Women in the United States experience high rates of coverage ‘churn’ in months before and after childbirth.
      ).
      In states that did not expand Medicaid under the ACA, women of reproductive age are eligible for the program only if they are pregnant, the parent/caretaker of a dependent child, or disabled and meet the state's income eligibility levels. In these states, many pregnant women often enter prenatal care later than recommended but have access to a full array of pregnancy-related services, including labor and delivery and postpartum care. However, if after 60 days postpartum they no longer meet the eligibility criteria, they find themselves with no coverage at all, a common transition for more than one-half of women who gain Medicaid coverage by the month of delivery (
      • Daw J.
      • Hatfield L.
      • Swartz K.
      • Sommers B.
      Women in the United States experience high rates of coverage ‘churn’ in months before and after childbirth.
      ). This loss of coverage occurs because the criteria for a parent's coverage are typically much lower than for pregnant women or because states do not have a family planning waiver or a state plan amendment to extend coverage for these specific (i.e., family planning) services beyond the 60 days postpartum.
      In contrast, women in expansion states can remain covered after delivery if still income eligible as an expansion adult and thus benefit from continuous coverage and better access to care, including earlier entry into prenatal care, if they were to become pregnant again. A study published in 2017 and using Behavioral Risk Factor Surveillance System data found that Medicaid expansions significantly decreased the uninsured rate among poor (<100% FPL) women of reproductive age (19–44 years) between 2012 and 2013 and 2014 and 2015 by 13.2% (48% were uninsured in 2012–2013 and 39.6% were uninsured in 2014–2015) (
      • Johnston E.M.
      • Strahan A.E.
      • Joski P.
      • Dunlop A.L.
      • Adams E.K.
      Impacts of the Affordable Care Act's Medicaid expansion on women of reproductive age: Differences by parental status and state policies.
      ). In 2014, the most recent year available for national enrollment estimates, 25 million of the 69.3 million individuals enrolled in Medicaid were women older than 19 years of age (representing 36% of total Medicaid enrollment); of these 25 million women, the vast majority, 67%, were of reproductive age (19–49 years old) (
      Kaiser Family Foundation
      Medicaid’s role for women. Fact Sheet. Washington, DC.
      ). In general, Medicaid is an important source of women's health services for adult, nonelderly women (97.4 million women between the ages of 19 and 64 in 2017), regardless of their ability to have children, and it has grown in importance since the 2014 Medicaid expansions (in 2017, Medicaid covered 17% of all nonelderly women in the United States, increasing from 13% in 2013) (
      Kaiser Family Foundation
      Women’s health insurance coverage. Fact Sheet. Washington, DC.
      ).
      Prior research shows that, just before the expansions, Medicaid covered a disproportionate percentage of preterm births relative to PHI (
      • Markus A.
      • Krohe S.
      • Garro N.
      • Gerstein M.
      • Pellegrini C.
      Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
      ). Evidence also indicates that by providing Medicaid coverage to individuals early in life as infants, the program can have longer term impacts on health (
      • Dave D.M.
      • Kaestner R.
      • Wehby G.
      Does Medicaid coverage for pregnant women affect prenatal health behaviors? IZA Discussion Papers, 9712.
      ,
      • Miller S.
      • Wherry L.R.
      The long-term health effects of early life Medicaid coverage.
      ,
      • Sohn H.
      Medicaid's lasting impressions: Population health and insurance at birth.
      ,
      • Wherry L.R.
      • Fabi R.
      • Schickedanz A.
      • Saloner B.
      State and federal coverage for pregnant immigrants: Prenatal care increased, no change detected for infant health.
      ). The Agency for Healthcare Research and Quality found that Medicaid covered a disproportionate percentage of deliveries with any severe maternal morbidity relative to PHI (51.4% vs. 42.1%) just after the expansions in 2015, compared with deliveries with no severe maternal morbidity (Medicaid: 42.9%; PHI: 51.3%) (
      • Fingar K.
      • Hambrick M.
      • Heslin K.
      • Moore J.
      Trends and disparities in delivery hospitalizations involving severe maternal morbidity, 2006-2015. HCUP Statistical Brief #243.
      ). A study published in 2019 on the impact of the Medicaid expansion on low birth weight and preterm births, using the National Center for Health Statistics restricted access Birth Data Files from 2011 to 2016, concluded that the expansion did not lower the rate of either birth outcome (
      • Brown C.
      • Moore J.
      • Felix H.
      • Stewart M.K.
      • Bird T.M.
      • Lowery C.
      • Tilford M.
      Association of State Medicaid Expansion Status with Low Birth Weight and Preterm Birth.
      ). However, the same study found that it reduced disparities between Black and White babies in both instances (e.g., by almost one-half of a percentage point for preterm births), but did not reduce differences between Latinx and White babies (
      • Brown C.
      • Moore J.
      • Felix H.
      • Stewart M.K.
      • Bird T.M.
      • Lowery C.
      • Tilford M.
      Association of State Medicaid Expansion Status with Low Birth Weight and Preterm Birth.
      ). Another study examining the growing effect of the Medicaid expansion on poverty and out-of-pocket financial burden between 2010 and 2016 found that it reduced both; the authors calculate out-of-pocket costs in the range of $1,000 less annually relative to a world without Medicaid (
      • Zewde N.
      • Wimer C.
      Antipoverty impact of Medicaid growing with state expansions over time.
      ). Overall, these studies point to the importance of the Medicaid program as a health insurance safety net for pregnant women and newborns.
      One issue that has yet to be studied is whether the role of Medicaid as a safety net has remained constant or changed (either grown or decreased) since the enactment of the ACA relative to PHI as a main payer of labor and delivery (hereafter referred to as payment mix). At a minimum, one might expect the importance of the program to remain the same, that is, coverage levels of births, including preterm births, are stable during the decade following the ACA, even after the coverage expansions that took place in 2014, given that most of these expansions did not target pregnant women specifically. Alternatively, one might—perhaps—expect some increase, even if modest, in coverage of deliveries by Medicaid in the period after 2014 compared with the period before 2014, owing to two possible factors.
      The first factor is known as the woodwork or welcome mat effect, which “describes the increase in enrollment that can occur after programs are expanded or changed, encouraging eligible participants to ‘come out of the woodwork’ to enroll in them” (
      • Caffrey M.
      After the "woodwork effect," will the newly enrolled stay in Medicaid? AJMC.
      ,
      • Hudson J.
      • Moriya A.
      Medicaid expansion for adults had measurable ‘welcome mat’ effects on their children.
      ,
      • LaPlante M.
      The woodwork effect in Medicaid long-term services and supports.
      ,
      • Sonier J.
      • Boudreaux M.
      • Blewett L.
      Medicaid ‘welcome-mat’ effect of Affordable Care Act implementation could be substantial.
      ). In 2014 alone, it was estimated that 2 million individuals, including many women, who were eligible but not enrolled in Medicaid obtained coverage by becoming aware of the coverage expansions under the ACA, and this increase in coverage occurred in both expansion and non-expansion states (
      • Caffrey M.
      After the "woodwork effect," will the newly enrolled stay in Medicaid? AJMC.
      ). Considering that some women may have been pregnant when they applied or may have become pregnant after enrollment, it could lead to a bump upward in labor and delivery services paid for by Medicaid through the nonelderly adult expansion pathway. Alternatively, because the welcome mat effect is less documented for pregnant women at the time of delivery, some experts point out that the effect is less likely to occur owing to what is commonly referred to as emergency Medicaid, the part of the program that covers labor and delivery services for pregnant women who are presumed to be Medicaid eligible (
      • D’Angelo D.
      • Le B.
      • O’Neil M.E.
      • Williams L.
      • Ahluwalia L.
      • Harrison R.
      • Grigorescu V.
      Patterns of health insurance coverage around the time of pregnancy among women with live-born infants – Pregnancy Risk Assessment Monitoring System, 29 States, 2009.
      ). Both explanations are plausible but ultimately empirical questions.
      The second factor pertains to the pathways to Medicaid coverage for women of reproductive age. Women will either seek coverage when pregnant and thus will enroll in the traditional Medicaid pregnancy program and now a few more are eligible owing to the 5% income disregard. Or, they will apply for coverage as a Medicaid expansion adult. If they are pregnant at the time of application, they will be automatically enrolled in the pregnant women group. If not, they may become pregnant at a later time after enrollment and may or may not switch to the traditional pregnancy program. The two pathways could also have different implications for access to prenatal care. Early entry into prenatal care has been recognized as an important (albeit not only) contributing factor to more appropriately managing health and behavioral risks and chronic conditions associated with maternal morbidity (e.g., controlling hypertension to reduce preeclampsia risk). Enrolling in Medicaid on the basis of pregnancy (one of the pre-ACA traditional eligibility pathways) can delay early entry into care because women are already pregnant and their pregnancy may already be in the second or third trimester. A 2018 MACPAC study found that women with Medicaid were less likely to enter prenatal care in the first trimester compared with privately insured women (
      • Medicaid and CHIP Payment and Access Commission (MACPAC)
      Access in brief: Pregnant women and Medicaid.
      ). By enrolling in Medicaid as an adult or as a parent of a dependent child without the requirement to be pregnant, women are able to access services before they become pregnant. Whenever they find out that they are pregnant after enrollment, they are then able to start prenatal care early, preferably in their first trimester.
      According to CDC data from 2013 to 2015, 50% of women of reproductive age indicated that they expect to have a child in the future (
      • Daugherty J.
      • Martinez G.
      Birth expectations of U.S. women aged 15–44. NCHS Data Brief No. 260.
      ). In addition, the increase in the rate of women's coverage through Medicaid was double the increase for directly purchased private insurance between 2013 and 2017 (
      Kaiser Family Foundation
      Women’s health insurance coverage. Fact Sheet. Washington, DC.
      ). Therefore, it is not unreasonable to make the assumption that an increase in enrollment could potentially lead to an increase, even if small, in the percentage of births, including preterm births, paid for by Medicaid. In contrast, it could lead to fewer Medicaid births if newly enrolled women do not have children after enrollment (a possibility under the adult expansion pathway). It could also lead to more or fewer preterm births, depending on the relative health of new enrollees.
      The study presented in this article sought to explore which of these possible scenarios (i.e., stable, increased, or decreased Medicaid payment relative to PHI payment) could be shown empirically, with as a backdrop all of the changes and factors that have been described that could contribute to the observed changes in payment mix post-ACA market reforms and coverage expansions. The purpose of this article is to describe the payment mix (Medicaid relative to PHI and self-payments) of preterm births through an analysis of publicly available data from the National Center for Health Statistics Birth Data Files covering the 2011 to 2016 period, with an equal number of years before and after January 1, 2014, when coverage expansions became effective. The article concludes with a discussion of the implications of the findings for the Medicaid program.

      Methods

       Variables

      The main outcome for this study was preterm births and the main independent variable of interest was source of payment for deliveries. We used the variable gestational age at birth based on the best obstetric estimate of completed weeks of gestation available on the newborn section of the birth certificate to construct the preterm birth variable and identify births that were considered preterm (i.e., births occurring before week 37). Source of payment for deliveries was categorized as private, Medicaid, self-pay/uninsured, and other. Later stages of the analysis sought to control for demographic and medical factors known to be associated with preterm birth based on prior literature and variables available in the dataset (age, race, ethnicity, education level, marital status, prepregnancy body mass index, cigarette smoking, infertility treatment, interval between pregnancies, prior preterm birth, prenatal care initiation, method of delivery, gestational diabetes, hypertension, and eclampsia) (
      • Markus A.
      • Krohe S.
      • Garro N.
      • Gerstein M.
      • Pellegrini C.
      Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
      ).

       Data Sources

      This study used publicly available birth data from CDC's National Vital Statistics System for 6 years, 2011 through 2016 (
      National Bureau of Economic Research (NBER)
      NCHS' vital statistics natality birth data.
      ). Each of these datasets contains between 3 and 4 million observations, with one observation equaling one live birth. Although the publicly available data do not include state identifiers, we do know from the CDC National Center for Health Statistics background materials that not all states provided source of payment, our key explanatory variable, for all 6 years. However, by 2016, all states did provide this information. In terms of expansion status (information we obtained from another source, the Kaiser Family Foundation, and combined with the state lists from the CDC databooks), in 2014, 24 of the 48 states included in the dataset (50%) had expanded Medicaid. By 2015, 28 of 49 states in the dataset (57%) had expanded Medicaid, and by 2016, that number increased again, to 32 out of 51 states (63%). Data for Washington, DC, were available for each year. For this study, we limited the dataset for each year of data analyzed to all states that reported payment source information for that year.

       Approach

      We adopted a national-level, repeated cross-sectional study design to explore and describe annual, point-in-time distributions of payment source for all U.S. recorded births between 2011 and 2016. For each year between 2011 and 2016 we captured the annual payment mix for preterm birth and estimated the likelihood of a given payment source for a preterm birth, controlling for known covariates. As described elsewhere in this article, we assumed going into the analysis that Medicaid would either remain stable throughout the study period, that is, covering the same percentage of all births and preterm births between 2011 and 2016, regardless of the coverage expansions of 2014, or that the percentage would increase slightly relative to PHI. Regardless of which of the scenarios held true empirically, we assumed that Medicaid would remain more likely to pay for preterm births compared with other sources of payment because of how it was intentionally designed to cover the lowest-income women, who also are the most at risk of having a preterm birth.

       Analysis

      We conducted the study in four steps. First, we generated frequencies and a frequency table for all of the study variables (i.e., payment source, gestational age at birth, and 16 demographic and risk factors associated with preterm birth). Second, we recoded data to make them suitable for this particular study's bivariate and multivariate analyses. Gestation was recoded to a binary variable, with 1 = preterm and 0 = not preterm. Mother's age was collapsed into three categories—19 years old and under, 20–34 years old, and 35 years old and older—because teenagers and women over age 35 face a higher risk of many pregnancy complications compared with women of prime reproductive age (20–34 years old). The Hispanic origin categories were combined to create a binary variable, Hispanic and Not Hispanic. Interval since last pregnancy was condensed to reflect ideal birth spacing (18–59 months), as informed by the March of Dimes (
      • Markus A.
      • Krohe S.
      • Garro N.
      • Gerstein M.
      • Pellegrini C.
      Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
      ). The three types of obesity (I, II, and III) within the prepregnancy body mass index variable were also combined. Missing and unknown observations were removed from all variables. Third, χ2 tests were conducted to determine whether a significant relationship between payment source and gestational age categories existed for each year. Finally, multivariate logistic regression analyses were conducted to estimate the likelihood of preterm birth for each category of payment for each year, controlling for likely confounders. A p value of less than .05 was considered statistically significant for all analyses. All analyses were conducted in RStudio version 1.1.442 with base R version 3.4.4 and several packages downloaded from the CRAN (dplyr, gmodels, PredictABEL, and readr). The George Washington University Institutional Review Board deemed this study exempt from human subjects requirements.

      Results

      Table 1 displays the distribution of maternal characteristics for all births for each year during the study period (i.e., 2011–2016).
      Table 1Distribution of Maternal Characteristics, All Births, 2011 to 2016
      Category201120122013201420152016
      Maternal age, years
       <150.10.10.10.10.10.1
       15–198.37.76.96.25.85.3
       20–2523.423.222.822.121.420.4
       25–2928.528.428.528.729.029.1
       30–3425.025.626.427.127.528.2
       35–3911.712.012.312.813.313.9
       40–442.82.82.82.82.82.9
       45–490.20.20.20.20.20.2
       50–54<0.01<0.01<0.01<0.01<0.01<0.01
      Race
       White76.475.976.075.875.874.7
       Black16.016.016.116.016.116.5
       AI/AN1.21.21.21.11.11.1
       Asian/PI6.46.96.77.17.17.7
      Ethnicity
       Hispanic23.323.123.023.123.423.4
       Not Hispanic76.076.176.376.275.875.7
       Not stated0.70.80.70.80.80.9
      Education level
       ≤8th grade4.54.13.83.63.53.3
       9th–12th grade, no diploma13.512.711.811.310.810.3
       HS graduate24.724.624.724.824.824.8
       Some college20.721.021.121.221.020.5
       Associate's degree7.57.67.98.08.18.2
       Bachelor's degree18.118.418.819.019.419.9
       Master's degree7.78.18.38.58.79.1
       Doctorate/professional2.22.32.42.42.52.6
       Unknown1.21.21.21.21.31.3
      Marital status
       Married59.459.359.459.859.860.2
       Unmarried40.640.740.640.240.239.8
      Prepregnancy BMI
       Underweight3.83.73.63.63.53.4
       Normal45.345.044.444.243.643.1
       Overweight24.324.324.424.625.025.3
       Obesity I12.612.712.913.113.513.9
       Obesity II5.95.96.16.36.56.7
       Obesity III3.94.04.24.44.64.8
       Unknown4.34.34.43.83.22.8
      Cigarette smoking
       Yes8.28.38.18.27.67.1
       No84.086.687.189.891.492.4
       Unknown7.75.14.82.00.90.5
      Infertility treatment
       Yes1.41.51.51.51.61.7
       No98.198.298.298.398.398.2
       Unknown0.40.30.20.20.10.1
      Interval between pregnancies, months
       0–31.21.21.21.21.21.2
       4–112.62.62.62.72.72.8
       12–177.47.37.47.47.47.5
       18–237.57.57.57.67.77.9
       24–3512.012.011.912.012.312.4
       36–477.57.57.57.47.47.3
       48–594.84.94.94.94.84.8
       60–713.13.33.53.53.43.3
       ≥727.88.08.48.79.09.1
       N/A32.932.732.332.031.531.1
       Unknown13.112.912.712.512.512.4
      Prior preterm birth
       Yes2.22.42.62.82.93.1
       No97.397.397.297.197.096.8
       Unknown0.40.30.20.20.10.1
      Prenatal care
       First trimester71.471.571.273.774.374.9
       Second trimester19.419.118.916.716.516.2
       Third trimester4.34.44.54.34.34.5
       None1.41.41.41.51.51.6
       Unknown3.53.74.03.83.42.9
      Delivery method
       Vaginal67.167.167.267.767.968.0
       Cesarean32.732.832.732.232.031.9
       Unknown0.20.20.10.10.10.1
      Gestational diabetes
       Yes4.85.25.25.45.75.9
       No94.794.694.594.494.294.0
       Unknown0.40.30.20.20.10.1
      Prepregnancy hypertension
       Yes1.41.41.51.61.61.7
       No98.298.398.298.298.398.2
       Unknown0.40.30.20.20.10.1
      Gestational hypertension
       Yes4.95.24.95.15.66.0
       No98.298.394.994.794.393.9
       Unknown0.40.30.20.20.10.1
      Eclampsia
       Yes0.20.20.20.30.20.3
       No99.499.599.599.699.799.6
       Unknown0.40.30.20.20.10.1
      Payment source
       Medicaid43.242.842.943.242.642.2
       Private46.046.646.747.348.349.0
       Self-pay4.14.24.34.24.44.3
       Other4.95.04.84.44.13.9
       Unknown1.71.41.30.90.70.6
      Weeks gestation
       <3711.611.411.311.311.311.4
       ≥3788.488.688.788.788.788.6
      Abbreviations: AI/AN, American Indian/Alaska Native; PI, Pacific Islander.
      Values are percent.

       Payment Source and Mix for All Births

      Together, private insurance and Medicaid covered approximately 90% of all births each year (Figure 1). Private insurance covered the greatest percentage of all births for all years and increased by 3% between 2011 and 2016. The percentage of births covered by private insurance grew more slowly between 2011 and 2013 (from 46.0% to 46.6%–46.7%) than between 2014 and 2016, when it jumped from 47.3% in 2014 to 48.3% in 2015, reaching 49.0% in 2016. In contrast, the percentage of births covered by Medicaid fluctuated from 43.2% in 2011, to 42.8% in 2012, to 42.9% in 2013, back to 43.2% in 2014, and back down to 42.6% in 2015, and to its lowest during the study period, 42.2% in 2016, one full percentage point lower than in 2011. The percentage of self-pay births, which are typically uninsured (
      • Centers for Disease Control and Prevention (CDC)
      Births: Final data for 2016: National vital statistics reports; 2017.
      ), also fluctuated between 2011 and 2016, starting at 4.1% in 2011 and reaching 4.3% in 2016. Thus, the payment mix for all births between 2011 and 2016 seems to have changed in favor of private insurance, with slower growth following the young adult (<26 years of age) coverage mandate, which became effective in Fall 2010, and faster growth after 2014 when the coverage expansions through the Marketplaces began. Over the same time frame, Medicaid coverage decreased slightly and uninsured births increased slightly.
      Figure thumbnail gr1
      Figure 1Payment source, all and preterm births, 2011 to 2016.

       Payment Source and Mix for Preterm Births

      Unlike the situation for births overall, Medicaid, not private insurance, covered the greatest percentage of preterm births. Medicaid paid for nearly half of preterm births each year, making it the largest payer of preterm births (Figure 1). Although the percentage of preterm births covered by Medicaid was stable between 2011 and 2016 at 48.9%, the percentage of preterm births covered by private insurance increased by 1.4% (nearly 5,500 births), from 42.1% in 2011 to 43.5% in 2016, and the percentage of preterm births that were uninsured declined very slightly by 0.3% after the 2014 coverage expansion. Thus, the payment mix for preterm births between 2011 and 2016 seems to have changed slightly in favor of private insurance and less uninsurance after 2014, whereas Medicaid coverage as the largest source of payment for preterm births remained at the same level throughout the 6-year period.

       Preterm Birth Rate by Payment Source

      Figure 2 displays the preterm birth rate by payer type over time. For all years sampled, the highest preterm rate occurred among births covered by Medicaid, and the lowest rate occurred among births covered by private insurance. The preterm rate decreased among privately insured births, eventually leveling out at 10% from 2014 onward. The preterm rate also decreased among self-paid/uninsured births. However, among births covered by Medicaid, the preterm rate has fluctuated, and even increased among Medicaid-covered births over the most recent years, rising from 12.7% in 2013 to 13.1% in 2016.
      Figure thumbnail gr2
      Figure 2Preterm Birth rate by payer, 2011 to 2016.

       Likelihood of a Preterm Birth by Payment Source

      The regression analysis results, showing the odds ratios for preterm birth for each year by payment source after adjusting for confounders, are displayed in Table 2.
      Table 2Odds of Preterm Birth by Payment Source and Other Characteristics, 2011 to 2016
      Variable and Reference CategoryOther Categories in the Variable201120122013201420152016
      Insurance: PrivateMedicaid1.0717 (1.0599–1.0836)
      p < .001.
      1.0771 (1.0654–1.0889)
      p < .001.
      1.0821 (1.0705–1.0938)
      p < .001.
      1.0384 (1.0281–1.0489)
      p < .001.
      1.0336 (1.0235–1.0438)
      p < .001.
      1.0468 (1.0367–1.0569)
      p < .001.
      Self-pay1.0272 (1.0045–1.0503)
      p < .001.
      1.0227 (1.0007–1.0452)1.0227 (1.0010–1.0448)0.9061 (0.8876–0.9250)
      p < .001.
      0.8785 (0.8605–0.8968)
      p < .001.
      0.8673 (0.8498–0.8850)
      p < .001.
      Other1.0500 (1.0294–1.0710)
      p < .001.
      1.0567 (1.0369–1.0769)
      p < .001.
      1.0564 (1.0361–1.0770)
      p < .001.
      1.0529 (1.0329–1.0733)
      p < .001.
      1.0260 (1.0065–1.0459)1.0327 (1.0129–1.0528)
      Age, years: 20–34≤191.2098 (1.1952–1.2246)
      p < .001.
      1.1150 (1.0980–1.1324)
      p < .001.
      1.1194 (1.1019–1.1373)
      p < .001.
      1.0902 (1.0736–1.1070)
      p < .001.
      1.0890 (1.0721–1.1062)
      p < .001.
      1.1296 (1.1118–1.1477)
      p < .001.
      ≥351.1342 (1.1171–1.1516)
      p < .001.
      1.2064 (1.1922–1.2207)
      p < .001.
      1.2067 (1.1927–1.2208)
      p < .001.
      1.2111 (1.1980–1.2243)
      p < .001.
      1.2207 (1.2079–1.2336)
      p < .001.
      1.2194 (1.2070–1.2319)
      p < .001.
      Race: WhiteBlack1.4833 (1.4658–1.5010)
      p < .001.
      1.4848 (1.4679–1.5020)
      p < .001.
      1.4938 (1.4771–1.5107)
      p < .001.
      1.4490 (1.4338–1.4643)
      p < .001.
      1.4442 (1.4293–1.4592)
      p < .001.
      1.4351 (1.4209–1.4495)
      p < .001.
      American Indian/Alaska Native1.2043 (1.1602–1.2502)
      p < .001.
      1.1850 (1.1412–1.2306)
      p < .001.
      1.1523 (1.1110–1.1952)
      p < .001.
      1.1539 (1.1164–1.1926)
      p < .001.
      1.1622 (1.1246–1.2010)
      p < .001.
      1.1556 (1.1185–1.1939)
      p < .001.
      Asian/Pacific Islander1.0337 (1.0150–1.0527)
      p < .001.
      1.0353 (1.0174–1.0535)
      p < .001.
      1.0364 (1.0184–1.0547)
      p < .001.
      0.9893 (0.9731–1.0058)1.0276 (1.0115–1.0439)
      p < .001.
      1.0307 (1.0153–1.0462)
      p < .001.
      Ethnicity: Not HispanicHispanic1.0521 (1.0408–1.0635)
      p < .001.
      1.0698 (1.0585–1.0812)
      p < .001.
      1.0402 (1.0292–1.0514)
      p < .001.
      1.0217 (1.0115–1.0320)
      p < .001.
      1.0280 (1.0179–1.0381)
      p < .001.
      1.0136 (1.0039–1.0233)
      Education: HS diplomaLess than HS1.1105 (1.0875–1.1339)
      p < .001.
      1.1089 (1.0856–1.1327)
      p < .001.
      1.0980 (1.0743–1.1222)
      p < .001.
      1.0993 (1.0766–1.1225)
      p < .001.
      1.0908 (1.0682–1.1138)
      p < .001.
      1.0822 (1.0598–1.1051)
      p < .001.
      Grade 9–121.1005 (1.0859–1.1153)
      p < .001.
      1.1016 (1.0871–1.1164)
      p < .001.
      1.1146 (1.0996–1.1297)
      p < .001.
      1.1074 (1.0934–1.1216)
      p < .001.
      1.1066 (1.0926–1.1208)
      p < .001.
      1.0998 (1.0859–1.1139)
      p < .001.
      Some college0.9376 (0.9263–0.9490)
      p < .001.
      0.9425 (0.9314–0.9537)
      p < .001.
      0.9388 (0.9278–0.9499)
      p < .001.
      0.9407 (0.9305–0.9511)
      p < .001.
      0.9406 (0.9305–0.9508)
      p < .001.
      0.9357 (0.9257–0.9458)
      p < .001.
      Associate's degree0.9058 (0.8900–0.9219)
      p < .001.
      0.9219 (0.9062–0.9379)
      p < .001.
      0.9100 (0.8948–0.9255)
      p < .001.
      0.9347 (0.9202–0.9494)
      p < .001.
      0.9288 (0.9147–0.9432)
      p < .001.
      0.9104 (0.8968–0.9243)
      p < .001.
      Bachelor's degree0.7990 (0.7872–0.8110)
      p < .001.
      0.8066 (0.7950–0.8184)
      p < .001.
      0.8036 (0.7921–0.8152)
      p < .001.
      0.8182 (0.8073–0.8293)
      p < .001.
      0.8122 (0.8016–0.8229)
      p < .001.
      0.8114 (0.8011–0.8219)
      p < .001.
      Master's degree0.7602 (0.7455–0.7753)
      p < .001.
      0.7916 (0.7768–0.8067)
      p < .001.
      0.7856 (0.7711–0.8004)
      p < .001.
      0.8066 (0.7928–0.8207)
      p < .001.
      0.7916 (0.7784–0.8052)
      p < .001.
      0.7878 (0.7750–0.8009)
      p < .001.
      Doctorate/professional degree0.7620 (0.7383–0.7865)
      p < .001.
      0.7755 (0.7523–0.7994)
      p < .001.
      0.8043 (0.7808–0.8285)
      p < .001.
      0.8160 (0.7938–0.8389)
      p < .001.
      0.7973 (0.7760–0.8192)
      p < .001.
      0.7857 (0.7652–0.8067)
      p < .001.
      Marital status: MarriedUnmarried1.1448 (1.1332–1.1566)
      p < .001.
      1.1529 (1.1414–1.1645)
      p < .001.
      1.1454 (1.1340–1.1568)
      p < .001.
      1.1409 (1.1304–1.1516)
      p < .001.
      1.1384 (1.1280–1.1489)
      p < .001.
      1.1459 (1.1356–1.1562)
      p < .001.
      BMI: NormalUnderweight1.2385 (1.2140–1.2635)
      p < .001.
      1.2611 (1.2365–1.2862)
      p < .001.
      1.2377 (1.2132–1.2626)
      p < .001.
      1.2398 (1.2168–1.2633)
      p < .001.
      1.2523 (1.2287–1.2763)
      p < .001.
      1.2361 (1.2128–1.2599)
      p < .001.
      Overweight0.9579 (0.9482–0.9678)
      p < .001.
      0.9555 (0.9460–0.9652)
      p < .001.
      0.9554 (0.9459–0.9650)
      p < .001.
      0.9474 (0.9386–0.9564)
      p < .001.
      0.9619 (0.9530–0.9709)
      p < .001.
      0.9655 (0.9567–0.9745)
      p < .001.
      Obese0.9401 (0.9302–0.9502)
      p < .001.
      0.9508 (0.9409–0.9607)
      p < .001.
      0.9434 (0.9337–0.9531)
      p < .001.
      0.9536 (0.9445–0.9628)
      p < .001.
      0.9745 (0.9654–0.9837)
      p < .001.
      0.9742 (0.9652–0.9832)
      p < .001.
      Cigarette smoking: NoYes1.1636 (1.1469–1.1807)
      p < .001.
      1.1849 (1.1681–1.2020)
      p < .001.
      1.2193 (1.2020–1.2368)
      p < .001.
      1.1856 (1.1698–1.2015)
      p < .001.
      1.2030 (1.1868–1.2195)
      p < .001.
      1.2168 (1.2002–1.2337)
      p < .001.
      Infertility treatment: NoYes2.6295 (2.5613–2.6995)
      p < .001.
      2.6674 (2.6008–2.7356)
      p < .001.
      2.5051 (2.4430–2.5687)
      p < .001.
      2.5053 (2.4471–2.5650)
      p < .001.
      2.4338 (2.3785–2.4903)
      p < .001.
      2.3481 (2.2970–2.4003)
      p < .001.
      Interval between pregnancies, months: 18–590–310.4744 (10.2330–10.7216)
      p < .001.
      10.2075 (9.9736–10.4469)
      p < .001.
      10.5514 (10.3134–10.7949)
      p < .001.
      10.6019 (10.3787–10.8300)
      p < .001.
      10.9215 (10.6922–11.1557)
      p < .001.
      10.5619 (10.3409–10.7876)
      p < .001.
      4–171.3787 (1.3606–1.3969)
      p < .001.
      1.3499 (1.3325–1.3675)
      p < .001.
      1.3666 (1.3491–1.3843)
      p < .001.
      1.3403 (1.3243–1.3566)
      p < .001.
      1.3584 (1.3424–1.3746)
      p < .001.
      1.3662 (1.3504–1.3821)
      p < .001.
      ≥601.1229 (1.1079–1.1380)
      p < .001.
      1.1036 (1.0894–1.1181)
      p < .001.
      1.1146 (1.1004–1.1289)
      p < .001.
      1.1343 (1.1210–1.1478)
      p < .001.
      1.1261 (1.1131–1.1393)
      p < .001.
      1.1366 (1.1237–1.1497)
      p < .001.
      N/A (first birth)1.1357 (1.1240–1.1476)
      p < .001.
      1.1238 (1.1124–1.1354)
      p < .001.
      1.1620 (1.1502–1.1739)
      p < .001.
      1.1670 (1.1559–1.1782)
      p < .001.
      1.1881 (1.1769–1.1994)
      p < .001.
      1.1858 (1.1748–1.1970)
      p < .001.
      Prior preterm birth: NoYes3.2203 (3.1545–3.2875)
      p < .001.
      3.2036 (3.1414–3.2670)
      p < .001.
      3.2669 (3.2066–3.3282)
      p < .001.
      3.1743 (3.1211–3.2285)
      p < .001.
      3.1575 (3.1064–3.2093)
      p < .001.
      3.1077 (3.0589–3.1573)
      p < .001.
      Prenatal care: First trimesterSecond trimester0.8525 (0.8435–0.8617)
      p < .001.
      0.8448 (0.8359–0.8537)
      p < .001.
      0.8516 (0.8427–0.8605)
      p < .001.
      1.2627 (1.2506–1.2749)
      p < .001.
      1.2432 (1.2313–1.2551)
      p < .001.
      1.2652 (1.2533–1.2773)
      p < .001.
      Third trimester0.6689 (0.6540–0.6841)
      p < .001.
      0.6625 (0.6481–0.6773)
      p < .001.
      0.6675 (0.6533–0.6821)
      p < .001.
      1.0610 (1.0416–1.0808)
      p < .001.
      1.0395 (1.0207–1.0587)
      p < .001.
      1.0321 (1.0139–1.0507)
      p < .001.
      No prenatal care2.1962 (2.1356–2.2586)
      p < .001.
      2.2872 (2.2250–2.3511)
      p < .001.
      2.3266 (2.2646–2.3904)
      p < .001.
      2.5559 (2.4936–26198)
      p < .001.
      2.5349 (2.4739–2.5975)
      p < .001.
      2.5741 (2.5137–2.6360)
      p < .001.
      Delivery: VaginalCesarean1.6664 (1.6523–1.6806)
      p < .001.
      1.6878 (1.6738–1.7019)
      p < .001.
      1.7205 (1.7064–1.7348)
      p < .001.
      1.7339 (1.7205–1.7474)
      p < .001.
      1.7567 (1.7433–1.7702)
      p < .001.
      1.8095 (1.7959–1.8233)
      p < .001.
      Gestational diabetes: YesNo1.2876 (1.2654–1.3101)
      p < .001.
      1.2366 (1.2161–1.2575)
      p < .001.
      1.2664 (1.2457–1.2874)
      p < .001.
      1.2307 (1.2121–1.2496)
      p < .001.
      1.2468 (1.2286–1.2653)
      p < .001.
      1.2233 (1.2059–1.2410)
      p < .001.
      Hypertension: YesNo1.9055 (1.8534–1.9590)
      p < .001.
      1.8717 (1.8219–1.9229)
      p < .001.
      1.9074 (1.8589–1.9573)
      p < .001.
      1.8651 (1.8212–1.9101)
      p < .001.
      1.9599 (1.9158–2.0051)
      p < .001.
      1.9982 (1.9552–2.0422)
      p < .001.
      Eclampsia: YesNo4.4310 (4.1726–4.7054)
      p < .001.
      4.1871 (3.9530–4.4350)
      p < .001.
      4.3236 (4.0893–4.5712)
      p < .001.
      4.2282 (4.0261–4.4405)
      p < .001.
      4.2020 (4.000–4.4141)
      p < .001.
      4.1950 (4.0023–4.3970)
      p < .001.
      Abbreviation: BMI, body mass index.
      Values are odds ratio (95% confidence interval).
      p < .001.
      The odds of a birth covered by Medicaid being preterm decreased most notably in 2014, the first year of Marketplace coverage and the year that states began to expand Medicaid. In 2013, births paid for by Medicaid were 8.2% more likely to be preterm than privately insured births were. In 2014, that value dropped by more than one-half, to 3.8%.
      Of note is the change in the likelihood of having a preterm birth based on entry into prenatal care, controlling for payment source. Women who received prenatal care in the second or third trimester of pregnancy compared with entering prenatal care in the first trimester were less likely to have a preterm birth before 2014, but more likely to have one after 2014, although women who did not receive any prenatal care at all remained 2.2 to 2.6 times more likely to have a preterm birth before and after 2014.

      Discussion

      This study showed that the mix of sources of payment for all births and preterm births changed during the 2011 to 2016 period, with a slight change in payment mix favoring PHI. Private insurance payment of all births increased by 6.5% during the study period, representing more than 118,000 births (
      • Centers for Disease Control and Prevention (CDC)
      Births: Final data for 2016: National vital statistics reports; 2017.
      ). This increase in private insurance coverage is consistent with other research on ACA insurance reforms affecting young women and nonelderly women overall.
      • Antwi Y.A.
      • Ma J.
      • Simon K.
      • Carroll A.
      Dependent Coverage under the ACA and Medicaid Coverage for Childbirth.
      , who analyzed vital statistics data for the 2009 to 2013 period applying a difference-in-difference approach, found that private insurance payment of childbirth before and after 2010, when the young adult (under age 26) expansion was implemented, increased by 2.5%. Employer-sponsored and individually purchased insurance grew by a total of 4% (2% each) between 2013 and 2016 for women ages 19 to 64 (
      Kaiser Family Foundation
      Women’s health insurance coverage. Fact Sheet. Washington, DC.
      ).
      Although the total number of births declined between 2011 and 2013, it increased in 2014 (by 55,895 births) (
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). Since 2014, the total number of births has been declining again (
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). In 2018, this decline was the fourth year in a row (
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). The most recent CDC data show that, as a percentage of births, PHI covered 49.4% in 2016 (1.94 million births), 49.1% in 2017 (1.89 million births), and 49.6% in 2018 (1.88 million births) (
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). Medicaid covered 42.5% of all births in 2016 (1.68 million births), 43% in 2017 (1.65 million births), and 42.3% in 2018 (1.60 million births) (
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). Medicaid-covered births decreased by 2.5% during the study period (2011–2016). Finally, self-pay/uninsured births represented 4.1% of all births in 2016 (161,781 births) and 2017 (158,076 births) and 4.2% in 2018 (159,252 births) (
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). Self-pay/uninsured births decreased by 0.2% during the study period (2011–2016).
      According to our calculations, the total number of preterm births decreased between 2011 and 2013 (from 387,452 to 377,489 preterm births), increased in 2014 (to 382,855 preterm births), decreased in 2015 (to 381,936 preterm births), increased in 2016 (to 386,696 preterm births), and has been decreasing since 2016 to 381,695 preterm births in 2017 and 379,171 preterm births in 2018 (
      • March of Dimes
      2019 March of Dimes Report Card. Washington, DC.
      ,
      • Martin J.A.
      • Hamilton B.E.
      • Osterman M.J.K.
      • Driscoll A.K.
      Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
      ). We found that Medicaid covered around 184,400 to 189,400 preterm births during the study period (2011, 189,464; 2012, 187,878; 2013, 184,492; 2014, 188,365; 2015, 186,003; and 2016, 189,241 preterm births), and followed a pattern similar to the numbers of preterm births overall (i.e., a decrease between 2011 and 2013 followed by an increase in 2014, a decrease in 2015, and an increase in 2016). As a percentage of all preterm births, Medicaid covered a disproportionate number of preterm births relative to PHI and self-pay/uninsurance, with 48.9% in 2011 (189,464) and 48.9% in 2016 (189,241), representing a change of 223 preterm births, or a 0.1% decrease during the study period.
      Our results confirm the significant size and relative stability of Medicaid as an insurance source for the most vulnerable women in American society, fulfilling its role as a financial safety net that provides access to necessary maternity care. This finding is also consistent with prior research. Although Medicaid payment of all births decreased by 2.5% during the study period, which is comparable to a decrease of over 2% observed in another study (
      • Antwi Y.A.
      • Ma J.
      • Simon K.
      • Carroll A.
      Dependent Coverage under the ACA and Medicaid Coverage for Childbirth.
      ), Medicaid remained the second-largest source of payment for childbirth for women ages 15 to 49 at around 42.2%–43.2%, which is comparable to the 43.4–45% range documented in the prior literature (
      • Markus A.
      • Andres H.
      • West K.
      • Garro N.
      • Pellegrini C.
      Medicaid-Covered Births, 2008-2010, in the Context of the Implementation of Health Reform.
      ,
      • Markus A.
      • Krohe S.
      • Garro N.
      • Gerstein M.
      • Pellegrini C.
      Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
      ). According to previous research, Medicaid was the largest source of payment for births to younger women, ages 19 to 25, between 2009 and 2013, at over 60% (
      • Antwi Y.A.
      • Ma J.
      • Simon K.
      • Carroll A.
      Dependent Coverage under the ACA and Medicaid Coverage for Childbirth.
      ).
      Our study also showed that the preterm birth rate was highest for Medicaid, at 13.1% in 2016, and lowest for private insurance, at 10%, and this rate for Medicaid increased by 0.4 percentage points between 2013 and 2016 and decreased by approximately 0.2 percentage points for private insurance. Although Medicaid continues to pay for a disproportionate percentage of preterm births, remaining stable at 48.9%, and to be more likely to pay for preterm births than private insurance, which is consistent with prior research (
      • Markus A.
      • Krohe S.
      • Garro N.
      • Gerstein M.
      • Pellegrini C.
      Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
      ), the likelihood of this in relation to private insurance has diminished noticeably by roughly one-half since 2014, the year when most Medicaid expansions began and the Marketplaces were launched.
      Medicaid still provides coverage for the lowest-income women, but the eligibility criteria for pregnancy coverage under Medicaid did not change with the ACA (other than the 5% income disregard, which has the practical effect of expanding coverage for women in the traditional pregnant women category), and although eligibility was expanded in 2014, it was for nonpregnant, nonelderly adult women who may or may not become pregnant after enrollment (MACPAC, 2013). At the same time, private insurance has undergone a distinct shift since 2010 in terms of coverage requirements regarding women and maternity benefits, improved affordability through Marketplace subsidies, contraception coverage mandates, and other factors beyond changes in coverage and benefits, which could explain our results. Several factors could be contributing to the patterns observed, but it is beyond the scope of this article to determine which ones and to what extent.
      Between 2011 and 2016, the percent of self-paid/uninsured preterm births decreased by 0.3% and the odds of self-paid births being preterm fell to less than 1.0 in 2014 and the subsequent two years. In other words, self-paid preterm births were less likely than privately paid preterm births from 2014 to 2016. The fact that self-payments and Medicaid payment of all births seemed to have decreased (by 0.2% and 2.5%, respectively) and private insurance payment increased (6.5%) during the study period may reflect a shift in financing of childbirth, although it is beyond the scope of this study to establish this fact empirically.

       Limitations

      This study has several limitations. First, the National Vital Statistics System birth data did not include all 50 states each year. Four of the states that were not included in 2013–2014 were among the five lowest-ranked states in terms of health in 2013 (
      United Health Foundation
      America's health rankings, 2013 edition: A call to action for individuals and their communities. Saint Paul, MN.
      ). Had these states been included, the difference between 2013 and 2014 described may have been more pronounced. However, the 2016 dataset includes all states and Washington, DC, and the estimates are remarkably similar to the previous years of data.
      Second, the National Vital Statistics System does not differentiate between employer-sponsored and individually purchased (or directly purchased) private coverage. As a result, this study could not examine the extent to which the changes observed in private coverage were for Marketplace insurance relative to employer-sponsored insurance.
      Third, this study is limited to a cross-sectional analysis of a limited number of individual years of data, so that it does not allow for any conclusions as to whether changes over time were statistically significant and which of the coverage expansions through the Marketplaces or Medicaid contributed to the changes observed. Furthermore, the analysis does not allow us to conclude whether any of the coverage expansions contributed to patterns observed over the time period of the study. Other possible factors include an economy that has recovered since 2008 and been close to full employment for a while, including for many women, who may now have access to different types of insurance; the contraceptive benefit mandate, which enables women to better control the timing of their pregnancies; and the intrinsic characteristics of women giving birth that cannot be controlled in the dataset but could contribute to more or less healthy pregnancies. Nevertheless, it is consistent with prior research and provides new, policy-relevant data specifically focused on preterm births, which were not available before.

       Implications for Policy and/or Practice

      Republican members of Congress have been trying to repeal the ACA since 2010. In 2017, they made several concrete attempts to repeal the law in whole or in part, but none of the proposed bills passed. Following these unsuccessful repeal attempts, the current administration discontinued funding for the ACA's cost-sharing subsidies and continued to make regulatory changes that affect Marketplace coverage (such as shortening the open enrollment period and allowing short-term, limited insurance policies to be offered). Premiums increased in price across all coverage levels, including an average of 32% for the lowest cost “silver” plan (i.e., the lowest cost plan in the coverage tier with 70% actuarial value) (
      • Semanskee A.
      • Claxton G.
      • Levitt L.
      How premiums are changing in 2018.
      ). If it is true that the ACA's insurance market reforms improved maternity coverage for previously uninsured women by making coverage more affordable, policy actions that undo those reforms and increase the cost of coverage, including litigation that is currently underway in federal courts, are likely to reverse that progress.
      In addition, Medicaid continues to be an important floor under which no one should be uninsured. It remained a constant payment source for millions of births, including thousands of preterm births. While attempting to repeal the ACA in 2017, Congress also considered major changes to the expansion and traditional Medicaid program, which includes coverage for pregnant women, most prominently by giving states the option to block grant the program. Although these efforts have failed to pass, threats to the integrity of the Medicaid program continue at both the federal and state level. Most recently, in late 2019, Tennessee asked the federal government to authorize the state to implement a form of block grant for TennCare, the state's Medicaid program, and in early 2020, the Centers for Medicare and Medicaid Services issued new guidance for states on how to proceed with applications seeking to block grant Medicaid under a Section 1115 waiver. If such changes were to be implemented, they would likely result in restricted access to coverage for low-income and poor women of reproductive age, who are also the most at risk of having a preterm birth.

      Conclusions

      Preventing preterm births remains an important public health issue, and private and public payers play an important role in covering and paying for medically necessary services before, during, and after pregnancy, including prenatal care, labor and delivery, and prevention of recurring spontaneous preterm births. Although payers are key in addressing this problem, they cannot address it alone. The 2014 coverage expansions, which included a benefit mandate to cover maternity care and preventive services (including contraception and pregnancy-related tests, items, and services), have been accompanied by a relative, albeit small, change in birth payment mix. After the coverage expansions, private insurance pays the second-highest percentage of preterm births and self-pay/uninsured women pay for a relatively much smaller percentage of preterm births, whereas Medicaid remains steady as the largest source of payment of preterm births. Future research should continue to monitor the role of Medicaid in this area of public health concern and use evaluation designs and sufficient years of data with state identifiers to help tease out the actual impact of Marketplace and Medicaid coverage expansions on birth and preterm birth payment mix.

      References

        • Antwi Y.A.
        • Ma J.
        • Simon K.
        • Carroll A.
        Dependent Coverage under the ACA and Medicaid Coverage for Childbirth.
        New England Journal of Medicine. 2016; 374: 194-196
      1. Preterm birth: Causes, consequences, and prevention.
        in: Behrman R.E. Butler A.S. 2007 (Available:) (Accessed Fall 2017)
        • Brown C.
        • Moore J.
        • Felix H.
        • Stewart M.K.
        • Bird T.M.
        • Lowery C.
        • Tilford M.
        Association of State Medicaid Expansion Status with Low Birth Weight and Preterm Birth.
        Journal of the American Medical Association. 2019; 321: 1598-1609
        • Caffrey M.
        After the "woodwork effect," will the newly enrolled stay in Medicaid? AJMC.
        (Available:)
        • Centers for Disease Control and Prevention (CDC)
        Births: Final data for 2016: National vital statistics reports; 2017.
        (ASI 4146-5.452; national vital statistics rpt. vol. 66, no. 1; DHHS pub. no. 2018-1120;CS287854. U.S. Department of Health and Human Services. Available:) (Accessed Spring 2018)
        • Centers for Disease Control and Prevention (CDC)
        Preterm birth.
        (Available:) (Accessed Spring 2018)
        • Claxton G.
        • Pollitz K.
        • Semanskee A.
        • Levitt L.
        Would states eliminate key benefits if AHCA waivers are enacted? Kaiser Family Foundation.
        (Available:) (Accessed Spring 2018)
        • D’Angelo D.
        • Le B.
        • O’Neil M.E.
        • Williams L.
        • Ahluwalia L.
        • Harrison R.
        • Grigorescu V.
        Patterns of health insurance coverage around the time of pregnancy among women with live-born infants – Pregnancy Risk Assessment Monitoring System, 29 States, 2009.
        MMWR Surveillance Summaries. 2015; 64
        • Daugherty J.
        • Martinez G.
        Birth expectations of U.S. women aged 15–44. NCHS Data Brief No. 260.
        CDC, Atlanta, GA2016 (Available:) (Accessed May 2019)
        • Dave D.M.
        • Kaestner R.
        • Wehby G.
        Does Medicaid coverage for pregnant women affect prenatal health behaviors? IZA Discussion Papers, 9712.
        (Available:) (Accessed Spring 2018)
        • Daw J.
        • Hatfield L.
        • Swartz K.
        • Sommers B.
        Women in the United States experience high rates of coverage ‘churn’ in months before and after childbirth.
        Health Affairs. 2017; 36: 598-606
        • Ferré C.
        • Callaghan W.
        • Olson C.
        • Sharma A.
        • Barfield W.
        Effects of maternal age and age-specific preterm birth rates on overall preterm birth rates — United States, 2007 and 2014.
        MMWR. Morbidity and Mortality Weekly Report. 2016; 65: 1181-1184
        • Fingar K.
        • Hambrick M.
        • Heslin K.
        • Moore J.
        Trends and disparities in delivery hospitalizations involving severe maternal morbidity, 2006-2015. HCUP Statistical Brief #243.
        AHRQ, Rockville, MD2018 (Available:)
        • Gifford K.
        • Walls J.
        • Ranji U.
        • Gomez I.
        Medicaid coverage of pregnancy and perinatal benefits: Results from a state survey.
        The Henry J. Kaiser Family Foundation, 2017 (Available:)
        • Goldenberg R.L.
        • Culhane J.F.
        • Iams J.D.
        • Romero R.
        Epidemiology and causes of preterm birth.
        Lancet. 2008; 371: 75-84
        • Grosse S.D.
        • Waitzman N.J.
        • Yang N.
        • Abe K.
        • Barfield W.D.
        Employer-sponsored plan expenditures for infants born preterm.
        Pediatrics. 2017; 140 (Available:) (Accessed Spring 2018)
        • Hudson J.
        • Moriya A.
        Medicaid expansion for adults had measurable ‘welcome mat’ effects on their children.
        Health Affairs. 2017; 36: 1643-1651
        • Johnston E.M.
        • Strahan A.E.
        • Joski P.
        • Dunlop A.L.
        • Adams E.K.
        Impacts of the Affordable Care Act's Medicaid expansion on women of reproductive age: Differences by parental status and state policies.
        Women’s Health Issues. 2018; 28: 122-129
        • Kaiser Family Foundation
        Medicaid’s role for women. Fact Sheet. Washington, DC.
        (Available:)
        • Kaiser Family Foundation
        Women’s health insurance coverage. Fact Sheet. Washington, DC.
        (Available:)
        • Kaiser Family Foundation
        Medicaid income eligibility limits for adults as a percent of the federal poverty level.
        (Available:) (Accessed Spring 2018)
        • Kaiser Family Foundation
        Pre-ACA state maternity coverage mandates: Individual and small group markets 2010.
        (Available:) (Accessed Spring 2018)
        • LaPlante M.
        The woodwork effect in Medicaid long-term services and supports.
        Journal of Aging & Social Policy. 2013; 25: 161-180
        • March of Dimes
        Long-term health effects of premature birth. Washington, DC.
        (Available:) (Accessed Spring 2018)
        • March of Dimes
        Medicaid covers fewer births but more preterm births than private coverage - Study by GWU, March of Dimes shows disproportionate percentage of preterm births are covered by Medicaid. Washington, DC.
        (Available:)
        • March of Dimes
        Preterm labor and premature birth: Are you at risk? Washington, DC.
        (Available:) (Accessed Spring 2018)
        • March of Dimes
        2019 March of Dimes Report Card. Washington, DC.
        (Available:)
        • Markus A.
        • Andres H.
        • West K.
        • Garro N.
        • Pellegrini C.
        Medicaid-Covered Births, 2008-2010, in the Context of the Implementation of Health Reform.
        Women’s Health Issues. 2013; 23: e273-e280
        • Markus A.
        • Krohe S.
        • Garro N.
        • Gerstein M.
        • Pellegrini C.
        Examining the association between Medicaid coverage and preterm births using 2010-2013 National Vital Statistics birth data.
        Journal of Children and Poverty. 2017; 23: 79-84
        • Martin J.A.
        • Hamilton B.E.
        • Osterman M.J.K.
        Births in the United States, 2016. U.S. Department of Health and Human Services.
        (Available:) (Accessed Spring 2018)
        • Martin J.A.
        • Hamilton B.E.
        • Osterman M.J.K.
        • Driscoll A.K.
        Births: Final data for 2018. National Vital Statistics Reports; Vol. 68, No, 13.
        National Center for Health Statistics, Hyattsville, MD2019 (Available:)
        • McDermott K.W.
        • Elixhauser A.
        • Sun R.
        Trends in hospital inpatient stays in the United States, 2005–2014. HCUP Statistical Brief #225.
        Agency for Healthcare Research and Quality, Rockville, MD2017 (Available:)
        • Medicaid and CHIP Payment and Access Commission (MACPAC)
        Maternity Services: Examining Eligibility and Coverage in Medicaid and CHIP.
        In June 2013 Report to the Congress on Medicaid and CHIP, Washington, DC2013 (Accessed: February 10, 2020)
        • Medicaid and CHIP Payment and Access Commission (MACPAC)
        Issues in pregnancy coverage under Medicaid and exchange plans.
        MACPAC, Washington, DC2014 (Available:) (Accessed Spring 2018)
        • Medicaid and CHIP Payment and Access Commission (MACPAC)
        Access in brief: Pregnant women and Medicaid.
        (Available:)
        • Miller S.
        • Wherry L.R.
        The long-term health effects of early life Medicaid coverage.
        Journal of Human Resources. 2019; 54: 785-824
        • National Bureau of Economic Research (NBER)
        NCHS' vital statistics natality birth data.
        (Available:) (Accessed Spring 2018)
        • Semanskee A.
        • Claxton G.
        • Levitt L.
        How premiums are changing in 2018.
        Kaiser Family Foundation, Washington, DC2017 (Available:) (Accessed Spring 2018)
        • Shapiro-Mendoza C.K.
        • Barfield W.D.
        • Henderson Z.
        • James A.
        • Howse J.L.
        • Iskander J.
        • Thorpe P.G.
        CDC Grand Rounds: Public health strategies to prevent preterm birth.
        Morbidity and Mortality Weekly Report. 2016; 65: 826-830
        • Sohn H.
        Medicaid's lasting impressions: Population health and insurance at birth.
        Social Sciences & Medicine. 2017; 177: 205-212
        • Sonier J.
        • Boudreaux M.
        • Blewett L.
        Medicaid ‘welcome-mat’ effect of Affordable Care Act implementation could be substantial.
        Health Affairs. 2013; 32: 1319-1325
        • United Health Foundation
        America's health rankings, 2013 edition: A call to action for individuals and their communities. Saint Paul, MN.
        (Available:) (Accessed Spring 2018)
        • U.S. Bureau of Labor Statistics
        CPI inflation calculator. Washington, DC.
        (Available:)
        • Vintzileos A.M.
        • Ananth C.V.
        • Smulian J.C.
        • Scorza W.E.
        • Knuppel R.A.
        The impact of prenatal care in the United States on preterm births in the presence and absence of antenatal high-risk conditions.
        American Journal of Obstetrics & Gynecology. 2002; 187: 1254-1257
        • Wherry L.R.
        • Fabi R.
        • Schickedanz A.
        • Saloner B.
        State and federal coverage for pregnant immigrants: Prenatal care increased, no change detected for infant health.
        Health Affairs. 2017; 36: 607-615
        • Zewde N.
        • Wimer C.
        Antipoverty impact of Medicaid growing with state expansions over time.
        Health Affairs. 2019; 38: 132-138

      Biography

      Rose Meltzer, MPH, was a graduate student at the Milken Institute School of Public Health, George Washington University. She currently is an Associate at Avalere Health. Her primary interests are federal and state policies around the Affordable Care Act and Medicaid.

      Biography

      Anne Rossier Markus, PhD, MHS, JD, is Chair and Professor, Department of Health Policy and Management, the Milken Institute School of Public Health, George Washington University. Her research focuses on the financing and organization of health care and access to quality care.