The indicator for preterm birth within 4 years was equal to 1 if the subsequent birth within 4 years was preterm (less than 37 weeks) and 0 if there was no birth in the next 4 year or if the next birth was not preterm. The indicator for low birth weight (less than 2500 g) was equivalently defined. The dotted line indicates the start of the Medicaid policy change (2012). LARC indicates long-acting reversible contraception.
Difference-in-differences estimates from linear probability models that included an interaction term between an indicator variable for birth in an implementing hospital and a postpolicy indicator for 2012 to 2015, a set of binary variables for each hospital, year, and season of birth. Models controlled for age group (mother aged 12 to 19 years, 20 to 24 years, 25 to 29 years, or 30 to 51 years). Standard errors were adjusted for clustering at the hospital level. The indicator for preterm birth within 4 years was equal to 1 if the subsequent birth within 4 years was preterm (less than 37 weeks) and 0 if there was no birth in the next 4 year or if the next birth was not preterm. The indicator for low birth weight (less than 2500 g) was equivalently defined. Error bars indicate 95% CIs. LARC indicates long-acting reversible contraception.
eTable 1. Parallel trend tests.
eTable 2. Difference-in-differences estimates of association between Medicaid payment policy change and study outcomes stratified by race, Medicaid-paid births South Carolina 2009-2015.
eFigure. Distribution of birth interval in days overall and by maternal race.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Steenland MW, Pace LE, Cohen JL. Association of Medicaid Reimbursement for Immediate Postpartum Long-acting Reversible Contraception With Infant Birth Outcomes. JAMA Pediatr. 2022;176(3):296–303. doi:10.1001/jamapediatrics.2021.5688
Are policies that provide hospital reimbursement for immediate postpartum long-acting reversible contraception (LARC) associated with changes in the timing of subsequent births and birth outcomes?
In this cohort study including 186 953 individuals with a Medicaid-paid childbirth between 2009 and 2015, comparing hospitals that implemented immediate postpartum LARC with those that did not, the policy was associated with a decrease in the probability of preterm birth and low-birth-weight birth. The policy was also associated with a decrease in the probability of a short-interval birth and an increase in days to next birth among non-Hispanic Black individuals.
Policies increasing access to postpartum contraception may improve infant outcomes.
Together, preterm birth and low birth weight are the second-leading cause of infant mortality in the US and occur disproportionately among Medicaid-paid births and among the infants of Black birthing persons. In 2012, South Carolina’s Medicaid program began to reimburse hospitals for immediate postpartum long-acting reversible contraception (LARC) separately from the global maternity payment.
To examine the association between South Carolina’s policy change and infant health.
Design, Setting, and Participants
This population-based cohort study using a difference-in-differences analysis included individuals with a South Carolina Medicaid-paid childbirth between January 2009 and December 2015. Data were analyzed from December 2020 to July 2021.
Medicaid-paid childbirth after March 2012 in South Carolina hospitals that had implemented the policy.
Main Outcomes and Measures
Immediate postpartum LARC uptake, subsequent birth within 4 years, subsequent short-interval birth, days to subsequent birth, subsequent preterm, and low-birth-weight birth within 4 years.
The study sample included 186 953 Medicaid-paid births between January 2009 and December 2015 in South Carolina (81 110 births from 2009 to 2011, 105 843 births from 2012 to 2015, and 46 414 births in exposure hospitals). The policy was associated with an absolute 5.6–percentage point (95% CI, 3.7-7.4) increase in the probability of receiving an immediate postpartum LARC overall, with significantly larger effects for non-Hispanic Black individuals than non-Hispanic White individuals (difference in coefficients 3.54; 95% CI, 1.35-5.73; P = .002). The policy was associated with a 0.4–percentage point (95% CI, −0.7 to −0.1) decrease in the probability of subsequent preterm birth and a 0.3–percentage point (95% CI, −0.7 to 0) decrease in the probability of subsequent low birth weight. No significant difference in the association between the policy and preterm birth or low-birth-weight birth between non-Hispanic Black and non-Hispanic White individuals was found. The policy was associated with a 0.6–percentage point (95% CI, −1.2 to −0.1) decrease in the probability of short-interval birth and a 27-day (95% CI, 11-44) increase in days to next birth among non-Hispanic Black individuals. The policy was associated with a significant decrease in the probability of a subsequent birth overall; however, confidence in this result is attenuated somewhat by nonparallel trends for this outcome before the policy change.
Conclusions and Relevance
Findings of this cohort study suggest policies increasing access to immediate postpartum LARC may improve birth outcomes but should be accompanied by other policy efforts to reduce inequity in these outcomes.
Access to comprehensive postpartum contraceptive services at the time of childbirth is recommended to ensure that postpartum individuals achieve their reproductive goals and to help them avoid unintended pregnancy. Unintended pregnancy and short-interval births are associated with low birth weight1,2 and preterm birth,3,4 which together are the second-leading cause of infant mortality in the US.5 Among Medicaid-paid births, 80% of pregnancies that occur within the first 6 months of childbirth are unintended.6 Unintended pregnancy is particularly common for adolescents, among whom 75% of pregnancies are unintended.7
Policy interventions to reduce low birth weight and preterm birth have included expanding eligibility for public insurance coverage before pregnancy8-12 and reimbursement for clinical and social support services during pregnancy.13-16 However, postpartum people face numerous barriers to accessing the full range of contraceptive options. Many postpartum individuals miss their postpartum visit and have gaps in insurance coverage after childbirth, making it difficult to obtain postpartum contraception.17-19 Access to the full range of medically safe contraceptive options before leaving the hospital after childbirth may help postpartum individuals avoid barriers to outpatient postpartum contraceptive access.
Until 10 years ago, long-acting reversible contraception (LARC; ie, intrauterine devices [IUDs] and contraceptive implants) were unavailable for postpartum people in hospitals. LARCs are the most effective type of reversible contraception and are recommended as a safe and effective postpartum option.20 Historically, immediate postpartum LARC (LARC placed while still in the hospital following childbirth) has been reimbursed as part of a global obstetrics payment for all care provided during an admission for labor and delivery.21 As the cost of a LARC device is between $500 and $800,22 use of this global payment has been an important disincentive for health care professionals to offer immediate postpartum LARC.23
In 2012, South Carolina’s Medicaid program became, to our knowledge, the first to begin reimbursing hospitals for immediate postpartum LARC placement separate from the global maternity payment. Since then, similar policies have been adopted by most other state Medicaid programs. With failure rates of less than 1%, adding LARC as a contraceptive option could improve birth outcomes by reducing unintended and/or short-interval pregnancies among women who choose to use these methods. Currently, limited rigorous evidence exists on the infant health impact of policies that increase contraceptive access. In a previous study, we found that South Carolina Medicaid’s policy change increased provision of immediate postpartum LARC and reduced short-interval births among adolescents.24 The objective of the current study was to use hospital discharge and birth certificate data to analyze the association between South Carolina’s policy change and rates of preterm birth and low birth weight among all persons with Medicaid coverage during childbirth.
The data used in this study come from South Carolina’s Revenue and Fiscal Affairs Office. They created a data set containing selected variables from birth certificate records for births to state residents between 2009 and 2019 and identified all hospital discharge records associated with these births from the Inpatient Hospitalization Encounter-Level Data.25 We linked these data sets using an individual identifier for the birthing individual as well as year of birth. Brown University’s Human Subjects Protections Program determined that this study was exempt from human subject’s regulations. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Each observation in our data set was an individual childbirth event (an admission for labor and delivery). We defined an index birth as any birth taking place between January 2009 and December 2015, and subsequent births as any birth from the same person after an index birth for up to 4 years (2019). The inclusion criteria for the study analysis were all births that were paid for by Medicaid between January 2009 and December 2015. Births between January 2016 and December 2019 were dropped from the analytical data set so that we could follow each person’s childbirth outcomes for 4 full years.
Even after reimbursement, important barriers to providing immediate postpartum LARC remain, such as the need to train health care professionals, stock the devices in hospital pharmacies, and alter payment software to include LARC billing.26,27 We showed in previous work that adoption of the policy across hospitals in South Carolina was not uniform, with approximately half of hospitals offering patients immediate postpartum LARC placement after the start of Medicaid reimbursement.28 To focus in this study on births in which immediate postpartum LARC may have been offered, we defined the exposure as having given birth after the policy (2012 to 2015) in an implementing hospital. We considered a hospital to have implemented the policy if they provided at least 1% of postpartum people an immediate postpartum LARC in any postpolicy year (2012 to 2015). All childbirths throughout the study period were designated as exposed based on the timing and location of the index birth, whereas any subsequent birth following the index birth was used to define the study outcomes.
Using the linked data set, we created variables for birth interval in days, birth weight, and gestational age for subsequent births from the same person. Specifically, for each index birth, we created an indicator variable that was equal to 1 if the patient’s subsequent birth was preterm (less than 37 weeks’ gestation) within the next 4 years and that was equal to 0 if the subsequent birth was not preterm or if the patient did not have another birth within 4 years. In a similar manner, we also created an indicator variable for low birth weight (less than 2500 g)29 defined within a 4-year window.
As our previous studies did not examine the outcomes of the policy separately by race and ethnicity and used a different data source (Medicaid claims), we also included immediate postpartum LARC placement and birth intervals as outcomes in this study. We created a variable for receipt of immediate postpartum LARC if the childbirth discharge record contained an insertion and/or device code for a contraceptive implant or IUD (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] diagnosis code of V2511, V2513, V255, Z30430, Z30433, Z30014, or Z30017; ICD-10 procedure code of 697, 0UH97HZ, 0UH98HZ, 0UHC7HZ, 0UHC8HZ, or 0UH90HZ). The policy may have affected infant outcomes through several different pathways, including by changing the share of women who go on to have another birth within 4 years (ie, a decrease in births overall), through a decrease in short-interval births or a decrease in unintended births. To examine 2 of these potential pathways, we also created an indicator variable equal to 1 if the patient had a subsequent birth within the next 4 years and an indicator variable for a short-interval birth, defined as another childbirth within 15 months.30 A final variable measured the total number of days between index and subsequent births.
We estimated the association between implementation of the policy and adverse subsequent birth outcomes using a quasiexperimental difference-in-differences approach31 comparing the change in the study outcomes in implementing hospitals with the concurrent changes in nonimplementing hospitals. By comparing the change in implementing and nonimplementing hospitals, this analysis controls for any potential secular changes in the study outcomes over time that are unrelated to the intervention. We fitted linear probability models for each outcome that included an interaction term between an indicator variable for birth in an implementing hospital and a postpolicy indicator for 2012 to 2015, a set of binary variables for each hospital, year, and birth quarter. Regression models also controlled for age group (mother aged 12 to 19 years, 20 to 24 years, 25 to 29 years, or 30 to 51 years), self-reported race and ethnicity (non-Hispanic Black, non-Hispanic White, Hispanic, and other or unknown race [including American Indian or Alaska Native, Asian or Pacific Islander, Chinese, Japanese, Hawaiian, and Filipino]), and education level (less than high school, high school or GED, and more than high school or GED). Standard errors were adjusted for clustering at the hospital level. The coefficient on the interaction term in these regression models is interpreted as the percentage-point difference in the change in the study outcomes comparing index childbirths that took place in implementing and nonimplementing hospitals.
For all outcome variables except the number of days to subsequent birth, analysis was conducted in the full sample of Medicaid-paid births between 2009 and 2015. Days to subsequent childbirth was examined only among those births followed by another childbirth from the same person within 4 years. Owing to large racial disparities in adverse birth outcomes, we also conducted stratified analysis to compare the outcomes of the policy between non-Hispanic Black and non-Hispanic White individuals. In addition, we tested whether there was a statistically significant difference in the trends in the study outcomes before the policy change in implementing and nonimplementing hospitals. The purpose of these tests was to rule out the possibility that a statistically significant difference-in-differences result was due to diverging trends before the start of the policy change. P values were obtained using z tests for difference in proportions and regression analysis. Significance was set at P < .05, and all P values were 2-sided. The analysis was conducted using Stata version 17 (StataCorp).
The study sample included 186 953 births paid for by Medicaid between January 2009 and December 2015 in South Carolina. A total of 46 414 (24.8%) took place in hospitals that began providing immediate postpartum LARC after the policy change (implementing hospitals) (Table 1). The age distribution was similar in implementing and nonimplementing hospitals. A total of 35.6% (95% CI, 35.0-36.2) in implementing hospitals had less than a high school education compared with 28.4% (95% CI, 28.1-28.7) in nonimplementing hospitals (Table 1). A greater share of patients identified as Hispanic (12.5%; 95% CI, 12.1-12.9) in implementing hospitals than in nonimplementing hospitals (3.3%; 95% CI, 3.2-3.5) (Table 1). Implementing hospitals had a higher volume of annual births paid for by Medicaid (1105 vs 511) and were less likely to be rural (33.3% [95% CI, 4.3-77.7] vs 46.8% [95% CI, 32.1-61.9]). Finally, the share of births that were preterm birth in implementing hospitals (15.5%; 95% CI, 7.1-23.9]) was higher than that in nonimplementing hospitals (9.5%; 95% CI, 7.4-11.6) (Table 1).
Changes in the study outcomes following the policy change are presented in Figure 1, while regression estimates are presented in Table 2. Before the start of the policy change, few patients in South Carolina received an immediate postpartum LARC (Figure 1A). Relative to the change in nonimplementing hospitals, the share of patients who received an immediate postpartum LARC in implementing hospitals increased by 5.6 percentage points (95% CI, 3.7-7.4) after the policy change (Table 2).
Before the start of the policy, rates of preterm and low-birth-weight births within 4 years were higher in implementing facilities compared with nonimplementing facilities (Figure 1B and C). The trends in these outcomes were decreasing before the policy change in both implementing and nonimplementing hospitals, and the differences in prepolicy temporal trends in these outcomes were very small and statistically insignificant (eTable 1 in the Supplement). Before the policy change, the probability of having a preterm birth in the next 4 years was 4.4% (95% CI, 4.1-4.7) and 3.5% (95% CI, 3.4-3.6) among patients treated at implementing and nonimplementing hospitals, respectively (Figure 1B; Table 2). The policy was associated with a 0.4–percentage point (95% CI, −0.7 to −0.1) decrease in the probability of having a subsequent preterm birth among patients at implementing hospitals (Table 2). Before the policy change, the probability of having a low-birth-weight birth within 4 years was 3.6% (95% CI, 3.3-3.8) and 2.9% (95% CI, 2.8-3.0) for patients treated at implementing and nonimplementing hospitals, respectively. The policy was associated with a 0.3–percentage point (95% CI, −0.7 to 0) decrease in the probability of having a subsequent low-birth-weight birth among patients at implementing hospitals.
Figure 2 and eTable 2 in the Supplement present the results from the analysis stratified by patient race and ethnicity. Before the policy change, the probability of having a subsequent preterm birth was higher among non-Hispanic Black individuals than among non-Hispanic White individuals in implementing hospitals (5.5% [95% CI, 5.0-6.0] vs 4.1% [95% CI, 3.7-4.5]; P < .001) and nonimplementing hospitals (4.2% [95% CI, 4.0-4.4] vs 2.9% [95% CI, 2.8-3.1]; P < .001) (eTable 2 in the Supplement). The policy was associated with a decrease in the probability of preterm birth among non-Hispanic Black individuals (0.5 percentage points; 95% CI, −1.1 to 0) and non-Hispanic White individuals (0.5 percentage points; 95% CI, −0.8 to −0.1) (Figure 2; eTable 2 in the Supplement). The association between the policy implementation and subsequent preterm birth did not differ between groups. Before the policy change, the probability of having a subsequent low-birth-weight birth was higher among non-Hispanic Black individuals than among non-Hispanic White individuals in implementing hospitals (5.0% [95% CI, 4.6-5.5] vs 2.8% [95% CI, 2.4-3.2]; P < .001) and nonimplementing hospitals (3.9% [95% CI, 3.7-4.2] vs 2.1% [95% CI, 2.0-2.3]; P < .001) (eTable 2 in the Supplement). In stratified analysis, the policy was not associated with changes in the probability of having a subsequent low-birth-weight birth in either group, and there was no significant difference in the difference-in-differences coefficients between groups (Figure 2; eTable 2 in the Supplement).
The temporal trends in prepolicy outcomes were not statistically different in implementing and nonimplementing hospitals for short birth interval (less than 15 months) and birth interval in days (eTable 1 in the Supplement). However, there was a statistically significant difference in the trend before the policy change for having another birth within 4 years (eTable 1 in the Supplement). The policy was not significantly associated with the probability of having a short-interval birth in the full study population or separately among non-Hispanic White individuals but was associated with 0.6–percentage point (95% CI, −1.2 to −0.1) decrease in the probability of having a short interval birth among non-Hispanic Black individuals (Table 3). The policy was associated with a 1.6–percentage point (95% CI, −2.8 to −0.3) decrease in the probability of having another birth within 4 years overall but not among non-Hispanic Black and non-Hispanic White individuals separately (Table 3). Among those who had another birth within 4 years, the distribution of birth interval in days shifted to the right in the full population, and in particular among non-Hispanic Black individuals (eFigure in the Supplement). Consistent with this visual evidence, the policy was associated with an increase of 27 days (95% CI, 11-44) between births among non-Hispanic Black individuals but was not significantly associated with the number of days between births in the full population or separately among non-Hispanic White individuals (Table 3).
This study found that South Carolina’s policy to begin Medicaid reimbursement for immediate postpartum LARC was associated with a decrease in the probability of having a subsequent preterm birth and of having a low-birth-weight birth. The policy was also associated with a decrease in the probability of having a short-interval birth and an increase in days between births among non-Hispanic Black individuals. This study also found an association between this policy and the probability of having another birth within 4 years in the full population, but significant differences in trends for this outcome before the policy change should be considered when inferring policy impact for this outcome.
Very little evidence exists documenting the association between policies or programs that increase access with postpartum contraception and infant health outcomes.32 Medical recommendations on birth spacing rely on literature documenting the association between short birth intervals of less than 6 months and adverse infant outcomes.4,32,33 Other literature has examined the association between the duration of postpartum contraceptive use and preterm birth.34 This study adds to this literature by demonstrating that increasing immediate postpartum contraceptive options is associated with reductions in low-birth-weight and preterm birth. For both of the study outcomes, the policy change was associated with approximately a 9% decrease in the outcome rate. The magnitude of the policy’s association with the study outcomes is limited by the share of people who chose an immediate postpartum LARC, which ranged between 1% and 13% in implementing hospitals after the policy change. Additionally, the policy may have shifted use away from other effective methods, such as postpartum sterilization, which would also limit the strength of the association between the policy and the study outcomes. To strengthen the evidence on postpartum contraception and infant health and to establish a stronger causal evidence base, future studies could examine the policy rollout in other states and in places where there is more widespread policy adoption.
The policy’s impact on preterm birth and low birth weight could occur through several potential mechanisms. First, declines could result from a decrease in overall births within the 4-year follow-up period in our study. However, as the trends in this outcome were not parallel before the policy change, this study cannot identify the degree to which a decline in births contributed to improved birth outcomes. Second, birth outcomes may have improved because of decreases in short-interval births. While no significant associations with short-interval births were found in the full population, among non-Hispanic Black individuals, who had higher rates of immediate postpartum LARC uptake than non-Hispanic White individuals, the policy was associated with a significant decrease in short-interval births. In addition, although our data cannot speak to this, the policy may have affected the intendedness of subsequent pregnancies, leading to healthier behaviors before and during pregnancy, such as early initiation of prenatal care.1
Medicaid reimbursement for immediate postpartum LARC is part of a broader set of policy changes implemented over the last 10 years intended to increase access to LARC.35 More recently, experts have raised concerns that the enthusiasm for LARC over the last decade may have resulted in the overemphasis of LARC in contraceptive counseling. Biased counseling practices could undermine trust in family planning health care professionals and programs, particularly among racial and ethnic minority groups, perpetuating a long-standing history in the US of reproductive coercion and oppression.36,37 Although we find encouraging downstream associations of immediate postpartum LARC payment with infant health, it is essential that immediate postpartum LARC policy implementation be grounded in principles of patient-centeredness and antiracism to ensure efforts meant to increase the availability of immediate postpartum LARC do not reduce contraceptive autonomy. Further, large racial disparities in infant outcomes persist in South Carolina and nationally, and policy efforts that directly address disparities in birth outcomes are needed.
This study has several limitations. First, our data set did not include information on pregnancy intention or abortion, so we could not test whether the policy decreased unintended pregnancy. In addition, we lack patient-reported outcomes, such as information about receipt of patient-centered contraceptive counseling, so we could not evaluate whether the policy resulted in biased counseling.37,38 We also lacked data on infant mortality from death records and could not include this outcome in our analysis. Further, our findings could be confounded if the policy affected the risk profile of patients delivering in implementing hospitals. However, as individual-level covariates are included in the regression model, the analysis likely controls for this potential bias. Another analytical limitation is that we used a conservative threshold (1%) to categorize hospitals into the implementing group, which limited the magnitude of the effect size that we could identify in our analysis. Finally, this analysis provides evidence at the population level and does not provide evidence to guide individual decision-making about how long to wait before becoming pregnant after childbirth. Future studies examining the effect of waiting 24 months compared with 12 months, for example, could inform individual decision-making about how to balance potential infant health effects with other considerations (eg, employment, age, health, and relationship status) when deciding when to become pregnant again.
In this study of all Medicaid-paid births in South Carolina between 2009 and 2015, giving birth in a hospital that had begun to offer immediate postpartum LARC was associated with a decrease in preterm birth and low birth weight. While our findings support the role of increased access to highly effective forms of postpartum contraception in improving birth outcomes, it is imperative that efforts to expand access focus on assuring comprehensive access to all forms of contraception without coercion. Additional policy solutions are needed to improve infant health, including those that directly address structural and interpersonal racism,39,40 to reduce racial disparities in infant health.
Accepted for Publication: October 18, 2021.
Published Online: January 10, 2022. doi:10.1001/jamapediatrics.2021.5688
Corresponding Author: Maria W. Steenland, SD, Population Studies and Training Center, Brown University, 68 Waterman St, Providence, RI 02912 (email@example.com).
Author Contributions: Dr Steenland had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Steenland, Pace.
Critical revision of the manuscript for important intellectual content: Pace, Cohen.
Statistical analysis: Steenland, Cohen.
Obtained funding: Steenland.
Study supervision: Pace, Cohen.
Conflict of Interest Disclosures: None reported.
Funding/Support: Research reported in this article was supported by the National Institute for Child Health and Human Development under award R03HD099428. Dr Steenland was also supported by other National Institutes of Health support (grant P2C HD041020).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the South Carolina agencies and programs where the data originated or the National Institutes of Health.