Adolescent Interpregnancy Interval in Colorado Among Women With a Last Live Birth Between 2004 and 2013

Abstract

Purpose:

The purpose of the article was to determine risk factors associated with interpregnancy interval (IPI) and how IPI is associated with subsequent pregnancy outcomes.

Methods:

We performed bivariate and multivariable analyses of the Colorado Birth Certificate Registry data from women with a last live birth from 2004 to 2013, among Coloradan women aged 19 years and younger.

Results:

Our multivariate analysis found that older adolescents aged 17–19 years had a reduced likelihood of an IPI >18 months (odds ratio [OR] .8; confidence interval [CI] .7–.9). Self-identifying as Hispanic increased the likelihood of an IPI <18 months, whereas delivering during or after 2009 or having Medicaid insurance significantly increased the odds that an adolescent would have an IPI >18 months (OR 1.5 [CI: 1.4–1.6]; OR 1.2 [CI: 1.1–1.2]; OR 1.5 [CI: 1.4–1.6]). We found that an IPI >18 months was associated with the following dependent variables in unique logistic regressions adjusted for age, ethnicity, year of last live birth, and insurance status: reduced risk of neonatal complications (OR .9 [CI: .8–.9]), reduced risk of overweight/obesity (OR .9 [CI: .8–.9]), and an increased likelihood of normal birthweight (OR 1.2 [CI: 1.1–1.4]), term birth (OR 1.3 [CI: 1.1–1.4]), and attending 14 or more prenatal visits (OR 1.3 [CI: 1.2–1.4].

Conclusions:

Having an IPI >18 months in adolescent Coloradan women is associated with an increased likelihood of prenatal care attendance, term birth, and normal birthweight and with a reduced likelihood of neonatal complications.

Women who become pregnant as adolescents have a high likelihood of repeat adolescent pregnancy [1]. Repeat pregnancy among teenagers often occurs at less than the recommended interpregnancy interval (IPI) of 18 months [2]. Regardless of whether adolescent pregnancies are desired or undesired, March of Dimes recommends that for best pregnancy outcomes, all pregnancies be properly spaced [3]. Adolescent pregnancy is associated with adverse outcomes such as preterm birth, anemia, hypertensive disorders, and low birthweight; repeat adolescent pregnancy compounds these adverse outcomes [4–7]. Reducing adolescent pregnancy is a national and global public health priority.

Adolescent pregnancy in Colorado is no less important in terms of its public health importance. Colorado has targeted this issue for a number of years, one example of which is the implementation of the Colorado Family Planning Initiative (CFPI) in 2009, a concerted effort to reduce adolescent pregnancy in the state [8]. In 2008, a private donor gave the Colorado Department of Public Health and Environment funds to support the CFPI [9]. This program was focused on providing placement and counseling training to providers, operational support, and low- or no-cost long-acting reversible contraceptives (LARC) to low-income women across the state [9]. The designation of LARC includes intrauterine devices and implants [10,11].

The CFPI is credited with having been associated with the following results: reducing the adolescent birth and abortion rates almost by 50%, reducing the number of births to women without a high school education by 38%, reducing the second and higher order births to adolescents by almost 66%, increasing the average age of first pregnancy in Coloradan women by over a year, and avoiding an estimated almost 70 million dollars in cost to the state from enti-tlement programs [9,12]. In addition, researchers found that improved access to family planning from the CFPI and increased use of LARC were associated with a reduced risk of preterm birth [13].

Despite the published effectiveness of the association of the CFPI with reduced rates of adolescent pregnancy, there is a gap in knowledge about the impact of the program on the adolescent IPI in Colorado. With this analysis, we want to understand the picture of IPI in Coloradan adolescents and how IPI impacts subsequent pregnancy outcomes. The objective of this study was to determine risk factors associated with a short IPI among Coloradan adolescent women and determine how IPI impacts pregnancy outcomes. We wanted to see if the year of delivery of last live birth, which we used as a proxy measure of the CFPI, had any impact on IPI, as well.

Methods

Study overview

To conduct this retrospective cohort analysis, we used the Colorado Birth Certificate Registry data of all live births that occurred between 2005 and 2016.

Only women with a history of a prior live birth were included in the analysis. We determined a woman to have had a prior live birth if there was informative data in the “Date of last live birth” field. Data in this field were taken as evidence that they had experienced a prior pregnancy. They also had to be aged <20 years at that last delivery. In addition, their index pregnancy had to be a live birth, as opposed to a miscarriage or abortion, to be included in the birth certificate registry database.

The IPI, which was the primary outcome (dependent variable) for objective 1 and the independent variable for objective 2, was calculated by determining the time between the delivery date of the last live birth and the time of conception of the index delivery, described by the birth certificate. The index delivery is the birth that was documented in the birth certificate registry between 2005 and 2016. IPI only refers to the time between a live birth and subsequent conception; as such, women with abortion or miscarriage were not included in the analysis. To calculate this, first, the difference in time in days between the delivery date published on the birth certificate record and the field, “Date of last live birth” was calculated. Then the gestational age of the infant at birth, as listed in the gestational age at birth field, minus 14 days (for date of conception) was subtracted from that difference. This gave us the time from last live birth to the date of conception, in days, which was then converted to months.

We compared women with an IPI ≤18 months to >18 months. However, we wanted to set a time to censor women who had not experienced a repeat live birth. We set this censor point at 36 months. This resulted in all women with a last live birth between April 15, 2004, and April 15, 2013, being included in the dataset. Women with a last live birth after 2013 would not have the full 36 months to conceive, and as such, were excluded. Women who delivered before April 15, 2004, would only be captured if they had an IPI longer than 36 months and therefore were excluded as well so as not to include outliers in the dataset. If women with an IPI >36 months were included in the analysis, it would skew the data to make it seem like women who delivered during earlier years had a longer IPI, and this would impact our data with respect to the CFPI. It should be noted that the definition of a long IPI is >59 months; therefore, adolescents with a long IPI were excluded from this analysis [14].

Setting

The Colorado Birth Certificate is a document registered with the Office of the State Registrar of Vital Records at the Colorado Department of Public Health and Environment as evidence of the occurrence of a live birth [15]. It serves principallyas documentation of identity and place of birth while collecting a host of medical items concerning characteristics of pregnancy and delivery, maternal and infant risk factors, and infant characteristics and outcomes [15].

Population

Our analysis included women who delivered a live birth in the state of Colorado who had a documented previous live birth on index birth certificate record and were aged 19 years or under at the time of that last live birth. Records were excluded if there was no report of a previous live birth or if the date of last live birth was outside the range of April 15, 2004, to April 15, 2013.

Ethics approval

This analysis of deidentified data was approved by the Colorado Multiple Institutional Review Board #18–0862.

Primary outcomes

Objective 1 is to determine if there is an association between the implementation of the CFPI (proxied by year of last live birth) and IPI, controlling for age, race, ethnicity, and insurance type. The IPI was separated into ≤18 months and >18 months to reflect the March of Dimes recommendation that the IPI or time from delivery to subsequent conception be at least 18 months [3]. Women were considered to have a short IPI if they conceived ≤18 months after their last live birth. They were considered to have had a longer, or optimal, IPI if they conceived between 18 and 36 months after their last live birth.

Dichotomous covariates in objective 1 included age (≤16 or 17–19 years), race (white or non-white), ethnicity (Hispanic or non-Hispanic), insurance status (Medicaid or non-Medicaid), and year of delivery (2004–2008 or 2009–2013). Year 2009 was chosen at the inflection point for the year of comparison as this was when the CFPI was implemented, which was aimed at increasing the use of LARC in the state [9]. Therefore, we used the year of last live birth as a proxy measure for exposure to the CFPI as we could not directly measure involvement in this program with our dataset. Non-Medicaid insurance includes private insurance, self-pay, unknown, Tricare, other government, and Indian health service.

Secondary outcomes

Our secondary objective was to determine how, in turn, IPI as the independent variable was associated with pregnancy outcomes. Body mass index (BMI; divided into <25 and ≥25 kg/m², which was reported on the birth certificate at delivery so likely represents delivery BMI), adherence to prenatal care (<14 visits or ≥14 visits), method of delivery (vaginal or cesarean delivery), gestational age at delivery (<37 or ≥37 weeks), infant birthweight (<2,500 or ≥2,500 g), and maternal and neonatal complications of pregnancy as composite variables (yes or no) were included in the analysis. Of note, the American Congress of Obstetricians and Gynecologists recommends 14 prenatal visits [16]. We chose to use BMI categories (generally reserved for adults) as this was a pregnant population and BMI is the measure of body mass measurement endorsed for monitoring during prenatal care, regardless of age [17].

Maternal complications were composite outcomes that included blood transfusion, uterine rupture, intensive care, placental abruption, and maternal death; the binary composite outcome was considered to have occurred if any record reported any one or more of these complications.

Similarly, neonatal complications were composite outcomes that included if the infant had a 5-minute Apgar score of ≤5, required assisted ventilation or intensive care admission, received antibiotics, experienced neonatal seizures, hyaline membrane disease, or meconium aspiration, or underwent transfer to another facility within 24 hours of life. Neonatal death is not captured on the birth certificate.

Analysis plan

For our primary objective and outcome, we tested bivariate associations between the aforementioned mother’s characteristics and the IPI, as the dependent variable. These analyses were tested with Fisher’s exact test (and the chi-square for BMI). We then performed a logistic regression with IPI as a binary outcome comparing ≤18 to >18 months.

For our secondary objective and outcomes, we performed a series of multiple logistic regression analyses to see if IPI was associated with the aforementioned adverse pregnancy outcomes while controlling for all other covariates that were significant to less than p < .05 in the bivariate analyses. BMI was considered a dichotomous variable for this analysis.

Results

Cohort description

Figure 1 illustrates the population included in this study. Of the 809,676 birth certificate records in the database, 272,986 were excluded because they had no prior births, 244,699 were excluded based on IPI criteria, and 39,454 by age criteria. This left 19,130 adolescents, 10,799 (56%) of whom had an IPI ≤18 months, and the remaining 8,351 (44%) had an IPI >18 months.

Figure 1.

CONSORT Diagram of Coloradan adolescents included in this interpregnancy interval analysis with a last live birth between 2004 and 2013.

Population characteristics

Table 1 describes the mother’s characteristics that we felt might impact the IPI and were used to address objective one (age, race, ethnicity, and year of last live birth). There are a few notable findings in the table. Sixty-seven percent of adolescent women who experienced repeat pregnancy were covered by Medicaid. More than 80% were racially white non-Hispanic; this is compared with 84% of adolescents without a repeat pregnancy (data not shown). Sixty percent of the cohort were ethnically Hispanic, which was nearly the same proportion as adolescents without a repeat pregnancy (data not shown).

Table 1.

Mother’s demographic characteristics, pregnancy characteristics, and outcomes of adolescents with interpregnancy interval ≤36 mo, Colorado, 2004–2013

	Frequency
Mother’s characteristic
Mean age (y) last live birth (standard deviation)
≤16 (n = 5%)	15.4 (.02)
17–19 (n = 95%)	18.2 (.01)
White race (%)	15,394 (83%)
Non-white race (%)	3,196 (17%)
Hispanic ethnicity (%)	11,465 (60%)
Year of last live birth (%)
2002–2008	13,070 (68%)
2009–2013	6,060 (32%)
Medicaid insurance (%)	12,802 (67%)
Pregnancy characteristic
Prenatal visits (%)
<14	16,626 (87%)
≥14	2,504 (13%)
BMI (%)
Underweight	1,099 (6%)
Normal weight	8,023 (42%)
Overweight	3,999 (21%)
Obese	5,846 (31%)
Method of delivery (%)
Vaginal (assisted/operative)	12,161 (83%)
Cesarean	2,421 (17%)
Gestational age (wk) (%)
<24 to 27+6	104 (1%)
28 to 33+6	356 (2%)
34 to 36+6	1,347 (7%)
≥37	17,323 (90%)
Birthweight (g) (n, %)
<l,500	222 (1%)
1,500–2,499	1,511 (8%)
2,500–3,499	13,142 (69%)
3,500–4,499	4,197 (22%)
≥4,500	58 (<1%)
Neonatal complications of pregnancy (%)	1,388 (7%)
Maternal complications of pregnancy (%)	268 (1%)

BMI = body mass index.

Pregnancy characteristics

Table 1 also describes the pregnancy and neonatal outcome measures that were present during the pregnancy following the “last live birth,” which will be referred to as the repeat pregnancy, which were used for analysis of our second objective. These include prenatal visits, BMI, method of delivery, gestational age, birthweight, and neonatal/maternal complications. As a cohort, few adolescents had the American College of Obstetricians and Gynecologists recommended number of 14 antenatal visits (13%). Just more than half of adolescents were overweight or obese in their repeat pregnancy (52%). Cesarean birth occurred in 17% of the population, preterm birth occurred in 10% of mothers, and birthweights of <2,500 grams occurred in 9% (sum) of the study population. Neonatal complications of pregnancy were more prevalent occurring in almost 7% of all deliveries. In terms of maternal outcomes, the composite outcome was rare, being present in 1% of the maternal adolescent population.

Objective 1

Tables 2 and 3 consider IPI as the dependent or outcome variable. Table 2 illustrates the bivariate associations (unadjusted), and Table 3 illustrates the multivariate (adjusted) associations of mother’s characteristics with the IPI in Coloradan women under age 19 years. Table 2 shows that younger adolescents with a prior live birth (≤16 years) were more likely to have an IPI >18 months (p = .02). Hispanic ethnicity and white race were associated with a short IPI (p < .001 and p =.02, respectively). Medicaid insurance was associated with an IPI >18 months (73% vs. 64%, p < .001). Adolescents who delivered their last live birth during or after 2009 were more likely to have an IPI >18 months (42% vs. 47%, p < .001).

Table 2.

Characteristics associated with IPI among adolescent women with a last live birth, Colorado, 2004–2013

	Interpregnancy interval
	≤18 mo	>18 mo	p value
	(n = 10,799)	(n = 8,351)
Age at last live birth ≤16 (column %)	426 (4%)	385(5%)	.02
Age at last live birth > 16	9,522 (96%)	7,206 (95%)
White race (column %)	9,038 (84%)	6,896 (83%)	.02
Non-white race	1,741 (16%)	1,455 (17%)
Hispanic ethnicity (column %)	6,892 (64%)	4,573 (55%)	<.001
Non-Hispanic ethnicity	3,887 (36%)	3,778 (45%)
Last live birth (row %)			<.001
2004–2008 (13,070)	7,593 (58%)	5,477 (42%)
2009–2013 (6,060)	3,186(53%)	2,847 (47%)
Medicaid insurance (column %)	6,741 (63%)	6,061 (73%)	<.001
Non-Medicaid insurance	4,038 (37%)	2,290 (27%)

Table 3.

Mother’s characteristics and their association with interpregnancy interval among Coloradan adolescent women with a last live birth between 2004 and 2013

Variable	Adjusted odds ratio of having an IPI < 18 mo	Confidence interval
Hispanic ethnicity	1.5*	1.4–1.6
Age 17–19 y	1.3*	1.1–1.5
White race	1.0	.9–1.1
Year of last live birth in or after 2009 (CFPI)	.9*	.8–.9
Medicaid insurance	.7*	.6–.7

CFPI = Colorado Family Planning Initiative.

^*p < .05.

Table 3 presents the results of multivariate logistic regression of the association of maternal characteristics and IPI. Hispanic ethnicity and age 17–19 years were associated with an increased likelihood of an IPI <18 months (odds ratio [OR] 1.5; confidence interval [CI; 1.4–1.6] and OR 1.3 [CI: 1.1–1.5], respectively). White race was not associated with IPI (OR 1.0 [CI: .9–1.1]). Delivering during or after 2009 was independently associated with a reduced odds an adolescent would have an IPI <18 months (OR .9 [CI: .8–.9]), as was having Medicaid insurance coverage (OR .7 [CI: .6–.7]).

Objective 2

We found that an IPI >18 months was associated with the following dependent variables in independent multivariate logistic regressions adjusted for age, ethnicity, year of last live birth, and insurance coverage, which are shown in Table 4: reduced risk of neonatal complications (OR .9 [CI: .8–.9]) and being overweight or obese in the repeat pregnancy (OR .9 [CI .8–.9]), and an increased likelihood of term birth (OR 1.3 [CI: 1.1–1.4]), normal birthweight (OR 1.2 [CI: 1.1–1.4]), and 14 prenatal visits (OR 1.3, [CI: 1.2–1.4]).

Table 4.

Interpregnancy interval and pregnancy outcomes among Coloradan adolescent women with a last live birth between 2004 and 2013

Pregnancy outcomes (dependent variables of individual logistic regressions)
	Overweight/obese AOR (CI)	Maternal complications AOR (CI)	Neonatal complications AOR (CI)	Cesarean section AOR (CI)	Birthweight >2,500 g AOR (CI)	Gestational age >37 wk AOR (CI)	Attending ≥14 PNVs AOR (CI)
IPI >18 mo	.9 (.8–.9)	.9 (.6–1.0)	.9 (.8–.9)	1.0 (.9–1.1)	1.2 (1.1–1.4)	1.3 (1.1–1.4)	1.3 (1.2–1.4)
Age 17–19y	1.3 (1.1–1.5)	.7 (.4–1.1)	.9 (.7–1.1)	1.2(1.0–1.5)	1.4 (1.2–1.4)	1.4 (1.1–1.7)	1.1 (.9–1.4)
Hispanic	1.8 (1.7–2.0)	.6 (.5–.8)	.7 (.6–.8)	.8 (.7–.9)	1.2 (1.1–1.4)	1.4 (1.1–1.3)	1.1 (1.0–1.2)
Medicaid insurance	.6 (.5–.6)	1.4 (1.1–1.9)	1.6 (1.4–1.8)	1.0 (.9–1.1)	1.0 (.9–1.1)	1.0 (.9–1.1)	1.0 (.9–1.1)
Year of last live birth in or after 2009 (CFPI)	.9 (.8–.9)	.6 (.4–.8)	.4 (.4–.5)	1.0 (.8–1.1)	1.0 (.9–1.1)	1.0 (.9–1.1)	1.3 (1.2–1.5)

Numbers in bold are statistically significant to p < .05.

AOR = adjusted odds ratio; CFPI = Colorado Family Planning Initiative; CI = confidence interval.

Discussion

Our analysis found that adolescent women in Colorado who experienced a repeat birth at ages 17–19 years or self-identified as Hispanic were more likely to have a short IPI. Adolescents who delivered their last baby during or after 2009 and had Medicaid insurance were more likely to have an IPI >18 months. Having an IPI >18 months was associated with better adherence to prenatal care, a reduced likelihood of neonatal complications, a higher likelihood of term birth and normal birthweight, and a reduced risk of being overweight or obese in the repeat pregnancy when controlling for age, ethnicity, year of last live birth, and insurance status.

Our hypothesis was that a shorter IPI would lead to adverse outcomes among adolescents, and we did find that an IPI >18 months was associated with reduced adverse neonatal outcomes, which is consistent with the literature [18]. We also hypothesized that the CFPI, a program aimed at increasing up-take of LARC, would be associated with a longer IPI after it was implemented in 2009. We found that year of last live birth during or after 2009 was associated with an IPI >18 months, which is shown in Table 2. This was consistent with our hypothesis that delivery in or after 2009 would coincide with an IPI >18 months, suggesting that potentially the CFPI was associated with an optimal IPI.

Our analysis was limited in the variables we could test and include in our model by the fields on the birth certificate. Also, the birth certificate was not designed to respond to our hypothesis; therefore, the fields are not precisely calibrated to answer our research question. This also means that there may be unmeasured confounders that impact our ability to draw causal inferences from our work, such as type and use of postpartum contraception after the last live birth. The unknown impact of unmeasured confounders might be particularly pronounced in the case of adolescents on Medicaid; for example, in Table 3, we noted that Medicaid-covered adolescents had a reduced risk of a short IPI but were more likely to have maternal and neonatal complications with a longer IPI, which we cannot easily explain. In addition, we did not have any reason to believe that the predictor variables independently associated with IPI would interact; therefore, we did not consider including interaction terms in the model. Finally, the data on BMI were reported on the birth certificate, which suggests that it was maternal weight and height at delivery, not before pregnancy. The strength of our analysis is that it has a large sample size and offers data on a very specific cohort of the population in which we are interested.

In terms of finding targets for intervention to improve the IPI among adolescents or outcomes related to the IPI, our primary outcome and analysis only found that older adolescents and those that are Hispanic are at risk of a short IPI. Our analysis found that an IPI >18 months is associated with term birth, normal birthweight, and adherence to prenatal care, which has been previously noted in the literature. Therefore, we assert that the implications of this analysis for the health of adolescent women in Colorado state are that older teenagers could benefit from resource investment aimed at increasing their IPI. A potential method of doing this would be an intensive focus on methods of effective postpartum contraception in this population, which could improve pregnancy spacing and adherence to prenatal care in subsequent births [19–22].

IMPLICATIONS AND CONTRIBUTION.

The analysis has important public health implications because it shows that a program aimed specifically at increasing access to long-acting reversible contraceptives in high-risk populations may be associated with improved pregnancy spacing (reducing the number of adolescents with an interpregnancy interval <18 months) and improved pregnancy outcomes.