Postpartum contraception method type and risk of a short inter-pregnancy interval in a state Medicaid population

Abstract

Objective:

To evaluate the likelihood of a short inter-pregnancy interval (IPI) resulting in a birth among women covered by Medicaid, as a function of postpartum contraceptive method type.

Study Design:

We used Medicaid claims and eligibility data to identify women (aged 15–44) who had a Medicaid-financed birth in Delaware in the years 2012–2014 (n=10,328). Claims were analyzed to determine postpartum contraceptive type within 60 days of the index birth, and linked birth certificates were used to determine the incidence and timing of a subsequent birth through 2018 (regardless of payer). We used logistic regression to analyze the likelihood of having a short IPI following the index birth as a function of postpartum contraceptive type, controlling for pre-term births, parity, having a postpartum checkup, and maternal characteristics including age, race, education, and marital status.

Results:

Compared to patients receiving postpartum long-acting reversible contraceptive methods (LARC), patients with no contraceptive claims had nearly five times higher odds (OR=4.98, CI=3.05–8.13) and those with claims for moderately effective methods (injectable, pill, patch, or ring) had 3.5 times higher odds (OR=3.51, CI=2.13–5.77) of a subsequent birth following a short IPI.

Conclusions:

In a state population of Medicaid-enrolled women, women with claims for postpartum LARC had substantially lower risk of a short IPI resulting in a birth.

1. Introduction

Almost one-third of pregnancies in the United States are classified as having a short inter-pregnancy interval (IPI), defined in the Healthy People 2030 family planning objectives as conception within 18 months of a previous live birth [1]. The American College of Obstetricians and Gynecologists recommend that IPIs of less than 6 months be avoided and that women should be aware of modestly increased risks associated with IPIs of less than 18 months [2]. The evidence on the causal impact of short interval pregnancies and births on maternal and infant health points to IPIs of less than 12 months being more likely to be associated with adverse outcomes [3,4,5,6,7]. These include gestational diabetes, premature membrane rupture, uteroplacental bleeding disorders, preterm birth and low birth weight babies [8,9,10,11,12]. Therefore, reducing the proportion of short IPIs from the 2015–2017 baseline level of 33.8% of all IPIs is among the U.S. Department of Health and Human Services’ “Healthy People 2030” 10-year objectives for improving national health [1].

Unintended pregnancy is a related public health problem and more than half of short IPIs are unintended [13]. Increasing access to postpartum long-acting reversible contraception (LARC) has been promoted to address both unintended pregnancy and short IPIs [14]. Reducing barriers to LARC may be particularly important for addressing adverse health outcomes among low-income women and women of color, who are at higher risk for unintended pregnancy [15,16]. This may also increase women’s control over their reproductive health, including enabling more autonomy in decision-making related to birth spacing [17]. Low-income women are, by program design, more likely to have their births covered by Medicaid. Associations of LARC use with reduced risk of short IPI have been found to be stronger among Medicaid-insured women [18], and having a birth covered by Medicaid is associated with higher risk of a short IPI [13]. Additionally, being Medicaid insured is associated with adverse birth outcomes including higher risks of preterm births [19], and higher incidence and longer duration of hospital stays for newborn complications for Medicaid covered births [20], suggesting benefits to reducing incidence of short IPI as an additional risk factor for Medicaid-covered births. Many state Medicaid programs have recently sought to improve access to postpartum LARC by decoupling device and insertion payments from global labor and delivery fees [21,22,23].

The goal of the current study is to investigate the association of postpartum LARC versus other postpartum contraception with risks of short IPI for Medicaid-covered births in the state of Delaware. We examine the association of contraceptive type obtained within 60 days after delivery with occurrence and timing of a subsequent pregnancy that resulted in a live birth. Delaware provides an important context, having the highest rate of unintended pregnancies in the US in 2010 and, along with Florida, the highest rate again in 2014 (48%) [23]. Our study considers postpartum contraception in the time period prior to the state carving out immediate postpartum LARC payment and undertaking other access initiates as part of the Delaware Contraceptive Access Now (DelCAN) initiative, providing baseline data from the period when postpartum contraceptives could only be obtained in ambulatory settings, after the delivery discharge [24].

2. Materials and Methods

We conducted a retrospective cohort study of women (age 15–44 years old) enrolled in Delaware’s Medicaid program who had a live birth in any year from 2012 to 2014. Women were identified as having had a birth using diagnosis and procedure claims from administrative Medicaid data, from 2012 to 2014. Code sets were obtained from the Office of Population Affairs (OPA) postpartum contraceptive performance measures [25]. In partnership with the Delaware Division of Public Health (DPH), women were matched to birth certificates by using the mother’s name and birthday, as well as the infant’s name and birthday, where the mother could be matched to an infant using the household identifier in the Medicaid data. Using social security numbers associated with these index birth certificates, DPH was able to identify subsequent live births to women with an initial Medicaid-covered birth, regardless of the payer on the subsequent birth certificate; birth certificate data was tracked through 2018.

Our study cohort was restricted to women who were enrolled in any of Medicaid’s full-benefits aid categories for at least two months postpartum, and who had not had a tubal ligation. The code sets used to identify contraceptive method types were based on the contraceptive performance measures developed by the Office of Population Affairs; these measures include detailed lists of diagnostic, procedural and prescription codes associated with contraceptive types [25]. Several previous studies have used the Office of Population Affairs’ contraceptive performance measures to identify utilization of LARCs and other contraceptive methods among similar populations of Medicaid women [26,27]. This study was approved by the University of Delaware IRB (#930265).

2.1. Outcomes

The primary outcome of interest was the length of IPI following a Medicaid-covered index birth. We used the gestational age of the infant, recorded on the birth certificate, to calculate the IPI. Our outcome was a binary indicator of IPI less than 12 months. We chose 12 months given that that interval is strongly associated with adverse health outcomes and are more likely to result from unintended pregnancies [28]

2.2. Postpartum Contraceptive Method Use

Women were identified as having received postpartum contraceptives if they had at least one paid procedure, diagnosis or prescription claim for a contraceptive method within 60 days of the index birth. We excluded women who received a tubal ligation from the sample due to its very high effectivenesswhich was confirmed in our data by an unreportable low number of subsequent births to women who had claims for postpartum tubal ligation (the Centers for Medicaid & Medical service (CMS) suppression rules restrict the reporting of outcomes with less than 11 observations). The postpartum period of 60 days is used because Medicaid-enrolled women should have insurance coverage for at least two months postpartum, including coverage of the full range of contraceptive methods, which is also why postpartum contraception within 60 days is one of the Office of Population Affairs established contraceptive care measures.

In addition to LARC (IUD or contraceptive implants) we also measured use of shorter-acting moderately effective prescription methods: pill, patch, ring, and injection. We could not observe use of less effective methods (e.g. barrier methods) and therefore we interpret the absence of a claim for a prescription method as use of a less effective method or no method.

2.3. Covariates

We included a set of covariates obtained from the birth certificate: marital status, maternal education, race/ethnicity, age, parity, and whether the birth was pre-term (<37 weeks gestation). The categories we used were marital status (married, unmarried), education (less than college, any college), race/ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, non-Hispanic Other), age (15–19, 20–24, 25–29, 30–34, and 35 and older), and parity (first, second-fourth, and fifth and over). We also included whether the woman had a claim for a postpartum checkup within 60 days of the birth.

2.4. Analyses

We used a multivariable logistic regression to model the odds of having a short IPI (<12 months). We also generated a Kaplan-Meier survival curve to graphically demonstrate the relative risk of short IPI by contraceptive method type. SAS v.9.4 was used for all analysis.

3. Results

The cohort was comprised of 10,328 women who met the inclusion criteria (Table 1). For many of the women (43.4%), the index birth was their first child. Just over thirteen percent of the women were teenagers at the time of the index birth. The majority were unmarried (78.6%) and had a high school degree or less (62.1%). The cohort was diverse in terms of race and ethnicity. Just over 10.3% of the births were pre-term, occurring before 37 weeks gestation, according to the obstetrician’s estimate of gestational age.

Table 1.

Postpartum contraceptive uptake, subsequent births and pregnancies, and demographic characteristics of women enrolled in Delaware Medicaid with an Index Birth^a between 2012–2014

Characteristics	N=10,328
Contraceptives (within 60 days of birth)	(%)
No prescribed contraceptivesb	63.4
Moderately effective methods	30.5
LARC	6.1
Postpartum Checkup Claim
Yes	59.5
No	40.5
Inter-pregnancy Interval
Less than 12 months (11 or less)	10.5
12 to 17 months	6.1
18 months or more	25.3
No subsequent birth	58.0
Age at index birth
15–19 years	13.2
20–24 years	36.2
25–29 years	28.3
30–34 years	15.5
35 years and older	6.8
Marital status at index birth
Not married	78.6
Married	21.4
Race/Ethnicity
White, non-Hispanic	44.8
Black, non-Hispanic	41.5
Hispanic	10.9
Other	2.8
Mother’s education at index birth
High school or less	37.9
College or above	62.1
Parity (index birth order)
First	43.4
Second – Fourth	51.5
Fifth or higher	5.1
Pre-term index birth (<37 weeks)
Yes	10.3
No	89.7

Notes:

^aAn index birth is the first birth that we have a record for, but it may not be the woman’s firstborn child as she may have had a birth prior to 2012.

^bWomen had no prescribed method but may have been using a less-effective method.

^cNo subsequent birth during the follow up period (through December 2018)

Just under 60% of the women had claims for a postpartum checkup within 60 days of the birth (Table 1). We performed Chi-squared tests and determined that there were no statistically significant differences in the race/ethnicity distribution of women with or without a postpartum checkup (ChiSq=1.02, p=0.7975), but age was statistically significant, with younger mothers (15–24) having a slightly lower likelihood of postpartum checkup claims, at 57% vs. 61% for older mothers (ChiSq=20.19, p=0.0005). With respect to contraceptive method claims within 60 days after the index birth, 6.1% of the women had claims for LARC and 30.5% had claims for moderately effective methods. Follow-up on birth certificates demonstrated that through the end of 2018, 42% of the women had a subsequent birth; 25% of those subsequent births followed a short IPI (<12 months). In the overall sample of patients including those with no subsequent birth, 10.5% had an IPI of less than 12 months.

3.1. Survival Analysis

The Kaplan-Meier survival curve graphically presents the proportion of women with a subsequent birth by contraceptive claim status within 60 days of the index birth (Figure 1). Survival curves were generated for women who had claims for a LARC method, a claim for a moderately effective method, or no contraceptive method claim during that postpartum time frame. Women with a LARC method stand out from the other two groups, with only 2.7% having an IPI shorter than 12 months. In contrast, 8.6% of women with a postpartum claim for a moderately effective method and 12.2% of women with no contraceptive method claim had a subsequent birth following a short IPI.

Figure 1.

Survival curve for time to subsequent pregnancy, by postpartum contraceptive method claim type, Delaware Medicaid women with a 2012–2014 index birth (N=10,328)

3.2. Logistic Regression

The results of logistic regression are presented in Table 2. Controlling for all other variables in our model, LARC methods had a strong protective effect against the risk of a short IPI. Women not using any prescription method had 4.98 times greater odds of an IPI shorter than 12 months compared to women using LARC (95% CI 3.05–8.13). Moderately effective contraceptives had a smaller protective effect compared to LARC methods (AOR 3.51, 95% CI 2.13–5.77). Having a postpartum checkup within 60 days of the birth was not statistically significant; nor was the mother’s race/ethnicity, marital status or education. Age was important; the youngest mothers were over 3 times more likely to have a short IPI than women 35 or older. All age categories had greater odds of a subsequent birth following a short IPI compared to women 35 or older. Parity of the index birth was also significant; mothers with the most children were significantly more likely have a subsequent birth following short IPI after the index birth.

Table 2.

Adjusted Odds Ratio of a short (<12 months) inter-pregnancy interval among women enrolled in Delaware Medicaid with a 2012–2014 Index Birth (N=10,328)

Variable	OR	95% CI
Postpartum Checkup claim within 60 days	0.89	0.78	-	1.02
Contraceptive Type (ref=LARC)
Moderately effective method claim	3.51	2.13	-	5.77
No contraceptive claim	4.98	3.05	-	8.13
Race (ref=White, Non-Hispanic)
Hispanic	0.99	0.79	-	1.23
Non-Hispanic Black	1.17	1.02	-	1.34
Other, non-Hispanic	1.17	0.79	-	1.73
Age (ref=age 35–44)
15–19 years old	3.29	2.24	-	4.84
20–24 years old	2.84	2.01	-	4.01
25–29 years old	1.89	1.34	-	2.66
30–34 years old	1.77	1.24	-	2.52
Education – High School or less (ref=more than high school)	1.15	1.00	-	1.33
Married (ref=unmarried)	1.13	0.96	-	1.34
Index birth parity (ref=2–4)
First birth	0.85	0.73	-	0.99
Fifth birth or more	2.09	1.61	-	2.72
Preterm (<37 weeks) index birth	1.11	0.91	-	1.36

Notes: Estimated using a binary logistic regression model. OR, odds ratio, CI, confidence interval

4. Discussion

We found that LARC provision within 60 days postpartum had a strong protective effect against a short IPI, controlling for socio-demographic characteristics. Use of LARC in the U.S. increased substantially from 2008 to 2014 [29], especially among postpartum women, at one in six births in 2012–2015 [30]. In this study of Delaware women with a Medicaid-covered birth from 2012–2014, just under one in 12 women had a Medicaid claim for LARC within two months of delivery, suggesting potentially unmet need for increased LARC access among this population, immediately prior to a Medicaid reform in the state to decouple LARC device and insertion payments from global labor and delivery fees [24].

Only a few studies, to our knowledge, have investigated the association of postpartum contraceptive method type and risk of short birth intervals among women enrolled in Medicaid or equivalent public insurance programs, respectively conducted in California and South Carolina [31,32]. Those studies also demonstrated that increased uptake of postpartum LARC is associated with a lower risk of short IPI. Significant variation in postpartum contraceptive use across has been found across states, particularly in relation to LARC use, which has been attributed to a range of potential factors including differences in state policies and funding for family planning [33]. Therefore, it is significant that the direction and strength of our results are comparable to findings from both California and South Carolina. Our study adds to those previous studies, moreover, by including in the analysis all women who gave birth, and not only those who were also observed to have a subsequent birth. Women who would, without access to postpartum LARC, be at increased risk of a short IPI, may also include those who do not desire an additional birth. This is especially relevant given evidence that as many as half of women who are current LARC users express that they do not want any further births [34].

Our research has several limitations. Our analysis did not require that women remain continuously eligible in Medicaid beyond 60 days postpartum. We made this decision in order to be able to observe follow-up births that may not have been covered by Medicaid, however, while 95% of women did remain enrolled at least part of the time when a birth following a short IPI would be observed (11 to 20 months after the index birth), only 60% of the women had continuous eligibility in full benefits up to 20 months following the birth. Some women may have moved out of state, in which case their subsequent births would not have be registered on a Delaware birth certificate. Nonetheless, in sensitivity analysis we came to nearly identical results when including only women continuously enrolled in Medicaid up to 20 months postpartum (results not shown).

Another limitation is that we did not investigate contraceptive initiation or discontinuation after the 60-day postpartum period. During the years we analyzed (2012–2014) the recommendation from the American College of Obstetricians and Gynecologists (ACOG) was that a postpartum checkup should occur within 60 days, and the percentage of women we observed with a visit during that time frame was very similar to national estimates. However, ACOG has subsequently extended the recommended window of time for a postpartum visit to a longer 84 days, and in the years we analyzed some women may have had a later visit. We did not look at method discontinuation, and therefore do not know what share of subsequent births to LARC users (or users of moderately effective methods) followed intentional discontinuation of a method. Short IPI, while potentially problematic, may also be intended and a lack of knowledge of pregnancy intention of the second birth is a limitation that cannot be analyzed using administrative data.

There are also inherent limitations of administrative data. While race/ethnicity was not statistically significant in our analysis, we acknowledge that the interpretation of our results is limited to the finding that LARC methods provided strong protection against a subsequent birth following a short IPI among a cohort of Medicaid-enrolled women and cannot fully assess important differences that women of color may experience. Given the evidence suggesting that low-income women and women of color experience coercion in postpartum contraceptive counseling, more research is needed to better understand the relationships between LARC access, utilization, and choice [35]. We also did not attempt to identify pregnancy outcomes that did not result in a live birth (such as elective abortion, miscarriage, or ectopic pregnancy), so our results are not necessarily indicative of the degree to which various contraceptive options protect against short IPI if those pregnancies are not carried to term.

Despite these limitations, our study of pregnancies that result in subsequent births has important policy and practice implications given the potential of postpartum LARC to influence birth outcomes in Medicaid-covered populations [18,19].

Implications.

Women who received LARC within 60 days postpartum are less likely to experience a short inter-pregnancy interval resulting in a birth. The evidence suggests that recent state policy changes that make postpartum LARC more accessible to those that desire it will be an effective strategy in helping patients obtain desired birth intervals.