County-level clustering and characteristics of repeat versus first teen births in the United States, 2015-2017

Abstract

Purpose:

Approximately 16% of U.S. births to women ages 15–19 are repeat (2^nd or higher order) births. Repeat teen mothers are at elevated risk for poor perinatal outcomes. Geographic clustering and correlates of repeat teen birth are unknown.

Methods:

Data from birth certificates on N=629,939 teen births in N=3,108 U.S. counties in 2015–2017 were merged with data on county-level demographic, socioeconomic, and health provider characteristics. We identified contiguous clusters of counties with significantly elevated rates of first teen births only, repeat teen births, both, or neither between 2015 and 2017 and compared demographic, socioeconomic, and medical provider characteristics of counties from between 2010 and 2016 in each cluster type.

Results:

193 counties (6.21%) had high rates of repeat births only; 504 (16.22%) had high rates of first teen birth only; 991 (31.89%) had high rates of both repeat and first teen births and 1,420 (45.69%) had neither. Counties with high repeat (versus first only) birth rates had higher rates of poverty and unemployment, higher levels of income inequality, lower high school graduation rates, a higher share of racial and ethnic minority residents, fewer publicly-funded family planning clinics per capita, and more women receiving contraceptive services at publicly-funded clinics.

Conclusions:

First and repeat teen births cluster in differentially resourced geographic areas. Counties with high repeat teen birth rates have lower socioeconomic conditions than counties with high rates of first teen births only. These counties are more reliant on publicly-funded family planning clinics but have fewer of them per capita.

Research on teen childbearing in the United States has informed the design of interventions that have substantially reduced the birth rate among teenagers ages 15–19 from 59.9 births per 1,000 teens in 1990 to 18.8 births per 1,000 in 2017.¹ However, most existing research overlooks the distinction between first-time and repeat teen births (births to teens who already have one or more children). In 2017, 16.3% of the 194,377 births to U.S. women ages 15–19 were repeat births.² Like first-time teen mothers, most repeat teen mothers report their most recent birth was unintended.³ Limited research has compared these two groups, yet it suggests there are significant demographic and social differences between first-time and repeat teen mothers. Repeat teen births are more common among racial/ethnic minority adolescents.¹ Repeat (versus first-time) teen mothers are more likely to be from low-income families, drop out of high school, and be unemployed.^4–6 They are more likely to delay or forgo prenatal care and to experience low birth weight, preterm birth, and infant mortality compared to first-time teen and older mothers.^7–11 However, most evidence on repeat teenage childbearing derives from international studies^7,9, non-national and dated U.S. studies^5,8,10, or national U.S. data from the 1990s.⁴ The current study employs analyses of recent, national U.S. data to examine county-level characteristics associated with repeat versus first-time teenage births.

Stratifying teen births by parity is important for several reasons. Differences in demographic and socioeconomic characteristics of first-time and repeat teen mothers are likely associated with differential perinatal outcomes and contraceptive use. Further, teen mothers generally receive medical care in different settings than teens who have not yet become pregnant (obstetrics and gynecology (OBGYN) versus pediatric settings). Pregnant and parenting teens are candidates for postpartum contraception and psychosocial interventions such as home visiting, which are not applicable to preventing first-time unintended pregnancy. Such interventions are specifically only available after pregnancy (in the case of home visiting)¹² or birth (in the case of postpartum contraception).

Teen birth rates vary significantly across U.S. geographic regions. Understanding where repeat births disproportionately occur can inform both clinical practice and health services resource allocation on county and state levels. State-level differences in teen birth rates are well documented,¹³ but only a few studies have examined intra-state variability of teen birth rates.^14,15 Two studies examining geographic clustering of counties with elevated teen birth rates found significant county-level clustering, particularly in the Southern and Southeastern U.S. and Northern areas containing Native American reservations.^16,17 However, these studies did not stratify births by parity nor examine characteristics of counties where elevated rates occurred.

Documenting characteristics of counties where repeat teen births most often occur, including demographic composition, socioeconomic conditions, density of health care providers, and availability of family planning services, can inform targeted clinical care and prevention efforts. Limited existing research demonstrates teen births (not disaggregated by parity) are more common where incomes are low and economic opportunities are limited.^14,18,19 More residents living below the federal poverty line, lower educational attainment rates among adults, and lower per capita income relate to higher teen birth rates.^14,18 Relatedly, poor teens in states with high levels of income inequality are more likely to become parents than poor teens in states with lower income inequality.²⁰ Finally, availability and use of publicly funded family planning services varies widely between and within states and is linked to rates of unintended births.²¹

Understanding demographic, socioeconomic, and healthcare services characteristics of counties where repeat teen mothers disproportionately live is an crucial first step in better serving this often-overlooked population. This study examined four research questions: 1. Where do elevated repeat teen birth rates cluster geographically in the U.S.? 2. Which clusters have the highest repeat birth rates in both large and small population areas? 3. To what extent do geographic clusters of first and repeat teen births overlap? 4. What are the demographic, socioeconomic and medical provider characteristics of counties with high repeat teen births? We examined county-level indicators of population demographics, socioeconomic conditions (poverty rates, economic inequality, educational attainment, unemployment rates, urbanicity) and density of medical providers in specialties related to women’s health (OBGYN, primary care, and advanced practice midwifery). We hypothesized both repeat and first-time teen births would cluster in the Southern U.S., clusters would somewhat but not completely overlap, and counties with high repeat teen birth rates would have worse socioeconomic conditions and medical provider availability than counties with high first-time teen birth rates only.

METHODS

We examined county-level geographic clustering of births to U.S. women ages 15–19 in years 2015–2017. The study was declared exempt by the sponsoring university IRB. Birth data were from U.S. Vital Statistics Natality files, which reflect virtually all births each year.² Births in the 48 contiguous U.S. states were included (N=629,939 births in N=3,108 counties). Alaska and Hawaii were not included due to the geographic clustering nature of the analyses and their non-adjacency with other states. Births were categorized as either first or repeat (2^nd and higher order) using the indicator of number of previous live births on the birth certificate. Births missing the indicator of previous live births (n=2,090) were excluded. Births to mothers residing outside the U.S., or those whose county of residence was unknown, were also excluded (n=1,265). First-time and repeat birth rates were calculated as the number of each type of birth per 1,000 15–19-year-old females in each county using population data from the US Census Bureau, Population Division, Intercensal Estimates and Annual County Population Estimates. Of 9,324 total county years covered in the current study (3,108 counties for 3 years each), 4 had no CDC-recorded births (0.04%), 324 (3.47%) recorded only births to women ages 20 and older, and 8,996 (96.48%) recorded teen births as well as adult births. County years in which no teen births were recorded were assigned birth rates of zero.

County socioeconomic characteristics and medical provider information were obtained from the Area Health Resource Files (AHRF) County Level Data (US Department of Health and Human Services, County Area Health Resources File, 2016–2017 Release).²² AHRF contains aggregated county-level data from government agencies and medical associations, such as the Bureau of Labor Statistics, Bureau of Economic Analysis, Census Bureau, American Medical Association, and American Hospital Association. Data on availability and use of publicly-funded contraceptive clinics in each county were compiled by the Guttmacher Institute for 2015.²⁰ Data on county income inequality were from the American Community Survey 2011–2015 county-level estimates, downloaded from IPUMS NHGIS database.²³

Measures

Race and ethnicity variables represent the number of females ages 15–19 in each county, of each race and of Hispanic/Latino ethnicity, in 2010. Unemployment rate indicates the percentage of county residents ages 16+ who were unemployed in 2016. Percent urban population is the percentage of county residents living in either urbanized areas (50,000 or more people), or urban clusters (densely developed territory with between 2,500 and 50,000 people), as of 2010. Health insurance coverage indicates the percentage of persons under age 19 without health insurance in 2015. Poverty indicates the percentage of all county residents living below the federal poverty threshold in 2015. High school diploma signifies the percentage of persons 25 years+ with a high school diploma or above from 2011–2015. Income inequality in each county is indicated by its Gini coefficient measured between 2011 and 2015. A Gini coefficient of 0 indicates perfect income equality, while a coefficient of 1 indicates perfect inequality.²⁴

Primary care physicians indicates the number of primary-care medical professionals in 2015, in either an office-based or hospital-based practice dealing with patients, excluding federal employees, hospital residents, and those over age 75. Obstetrician-gynecologists are the number of physicians in the obstetrician-gynecologist specialty, excluding federal employees, in 2015. Advanced practice midwives includes both certified nurse-midwives and certified midwives in 2016. The raw number of health care providers of each type was converted to a measure of provider density by computing number of providers per 1,000 total county residents.

Availability and use of publicly-funded contraceptive services were indicated by the number of publicly-funded family planning clinics and the number of female contraceptive clients served in each county in 2015. Publicly-funded family planning clinics were defined as those offering contraceptive services to the general public with public funds such as Title X, Medicaid, or the federally qualified health center program to provide free or reduced-fee services to at least some clients. Female contraceptive clients refer to the number of distinct women of all ages who made at least one initial or subsequent visit for contraceptive services in the reported year. The raw number of family planning clinics was converted to the rate per 1,000 total county residents. The raw number of clients served was converted to the rate per 1,000 female county residents.

Analysis

To address research question 1, we detected and mapped spatial clusters of counties with unusually high rates of first or repeat teen birth using Kulldorff’s spatial scan approach implemented in the SaTScan software (Kulldorff & Information Management Services, Boston, MA, 2009). The approach is designed to statistically identify a set of likely high-risk geographic clusters based on county-level first and repeat teen births occurring in 2015–2017. This is a widely used approach in surveillance studies where the goal is to detect excess events in geographic areas while adjusting for uneven population distribution, and to test whether such excess can reasonably have occurred by chance (Kulldorff 1997). Our application considers each US county as a geographic center of a potential high-risk cluster. SaTScan scans for events occurring in varying-sized moving circular windows centered at each county and identifies clusters as non-overlapping circular windows where the observed rates are statistically higher than expected (e.g., p < 0.05) under the null hypothesis that the cluster is not at elevated risk²⁵. This procedure identifies a set of elevated-risk spatial clusters rank ordered by their likelihood of being high risk, where each cluster spans one or more contiguous counties. This is a statistical approach to dynamically discover hot spots of high risk that is preferable to approaches relying solely on county-level rates and static risk thresholds to determine elevated spatial risk.

To address research question 2, we categorized each cluster according to whether the total population of all counties in that cluster exceeded 100,000 residents (large populations), or whether the total population was less than 100,000 (small populations), in order to describe where the highest birth rate clusters occurred for large and small populations, following previous work.¹⁶

To address research questions 3 and 4, we categorized each county according to whether it appeared in a cluster of high first births, high repeat births, both, or did not belong to any detected cluster. Subsequent analyses were conducted at the county rather than cluster level in order to illustrate the proximal local contexts related to elevated teen birth rates. Because our primary interest was in comparing counties with high repeat birth rates to those with high first but not repeat birth rates, we combined the “high repeat births” and “both” categories for further analyses. Data categorizing each county were merged with data on county characteristics, described above, using the two-digit state and three-digit county FIPS codes provided by each underlying data source. We then employed ANOVA with post-hoc comparisons using Stata version MP 15.1 (StataCorp LLC, College Station, Texas). Three categories of counties were compared based on their demographic and socioeconomic characteristics and medical provider densities: 1) Repeat Births (combining the “high repeat births” and “both” categories), 2) First Births Only, and 3) Neither First nor Repeat Births. County variables included population by race/ethnicity, share of urban population, level of income equality, as well as rates of poverty, uninsured children, high school graduation, and unemployment. Medical provider densities included the number of primary care physicians, OB/GYNs, advanced practice nurse midwives, publicly funded clinics per 1,000 residents, and number of female contraceptive clients at publicly funded clinics per 1,000 females residents. Power simulations in SaTScan revealed statistical power > 0.99 at α levels of 0.05, 0.01, and 0.005 to detect a cluster relative risk of 1.01 for all identified clusters. Achieved power for ANOVA analyses was calculated using G*Power version 3.1.9.2 (Edgar Erdfelder et al., Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany).²⁶ Power to detect small, medium, and large effects all exceeded 0.99.

RESULTS

Geographic clustering of teen births

Our analysis identified 89 clusters of elevated levels of first births, and 61 clusters of elevated levels of repeat births. Figure 1 depicts counties that appeared in clusters of high first-time teen birth rates, high repeat teen birth rates, both, or neither. Clusters of high first and repeat birth rates overlapped some, but not completely: 991 counties (31.8%) had high rates of both repeat and first births; 1,420 counties (45.7%) had neither. 504 counties (16.2%) had elevated rates of first but not repeat births, indicating a disproportionately low number of their first-time teen mothers became repeat teen mothers. 193 counties (6.2%) had elevated rates of repeat births but not first births, indicating a disproportionately high number of their first-time teen mothers became repeat teen mothers.

Figure 1.

Clusters of counties with elevated first and repeat teen births. Each non-adjacent cluster appearing in the figure represents a statistically separate cluster identified in the analysis.

Largest clusters of teen births

First teen births.

The observed national teen birth rate from 2015–2017 was 20.33; the national first teen birth rate was 16.92. Table 1 depicts the ten clusters with the highest rates of first teen births in clusters with large and small populations. The large population cluster with the highest rate of first births was composed of 18 counties surrounding El Paso, Texas. This cluster had an average first teen birth rate of 35.98. The relative risk of the cluster was 2.13, indicating the observed first birth rate in that cluster was 113% higher than what would be expected under the null hypothesis of no elevated risk in that cluster. All top ten large population clusters of first teen births were in the Southern United States. The small population cluster with the highest rate of first births was composed of one county in Kansas, with a first teen birth rate of 57.70. The relative risk of the cluster was 3.39. Of the ten small population clusters with the highest first birth rates, 3 clusters, located in North Dakota and Nebraska, contained a Native American reservation. The remaining clusters were in Virginia (4), Minnesota (1), and Nebraska (1).

Table 1:

Highest First-Time Teen Birth Rate Clusters in the United States 2015 to 2017

Large City Within Cluster	Number of Counties in Cluster	Population in Cluster, 2017	Average First-Time Teen Birth Rate per 1,000†	Cluster Relative Risk‡
Population Greater than 100,000
El Paso, TX	18	1,383,057	35.98	2.13
McAllen, TX	32	2,637,245	35.35	2.10
Rapid City, SD	34	269,201	33.63	1.98
Midland, TX	41	848,238	32.42	1.91
Lubbock, TX	47	979,217	31.62	1.86
Casper, WY	13	272,202	29.85	1.75
Las Cruces, NM	10	474,827	29.40	1.73
Aztec, NM	5	406,905	29.27	1.72
Watertown, NY	1	114,187	28.70	1.69
Memphis, TN	83	3,449,737	28.35	1.68
Population 100,000 or under
Junction City, KS	1	33,855	57.70	3.39
Petersburg, VA	3	72,201	42.61	2.50
Waynesboro, VA	1	22,327	41.62	2.45
Roanoke, VA	1	99,837	38.05	2.24
Rolla, ND*	5	39,001	37.64	2.21
Worthington, MN	1	21,944	35.89	2.11
Williston, ND*	9	90,652	35.44	2.08
Dakota City, NE*	2	27,409	34.58	2.03
Manassas, VA	1	41,501	33.49	1.97
Lexington, NE	1	23,709	32.87	1.93

Notes:

^*Cluster under 100,000 contains a Native American Reservation or Tribal Entity.

^†The national first-time teen birth rate from 2015–2017 was 16.92 per 1,000 girls ages 15–19.

^‡Cluster relative risk indicates the risk of first-time teen birth versus what woul be expected under the null hypothesis of no elevated risk in the cluster.

Sources: Centers for Disease Control and Prevention. U.S. Vital Statistics Natality files, 2015–2017. https://www.cdc.gov/nchs/data_access/vitalstatsonline.htm

U.S. Dept. of Homeland Security/ Homeland Infrastructure Foundation-Level Data. https://hifld-geoplatform.opendata.arcgis.com/datasets/indian-lands-and-native-entities

Repeat teen births.

The observed national repeat teen birth rate from 2015–2017 was3.35. Table 2 depicts the ten clusters with highest rates of repeat teen births in large and small populations. The large population cluster with the highest rate of repeat births was composed of 32 counties surrounding McAllen, Texas, with an average repeat teen birth rate of 10.85 and a relative risk of 3.29. Except for a cluster surrounding Fort Wayne, Indiana, all large population clusters of elevated repeat teen birth rates were located in Southern states. The small population cluster with the highest rate of repeat teen births was Benson County, North Dakota, with an average repeat teen birth rate of 29.79 and a relative risk of 8.84. Of the ten small population clusters with the highest repeat teen birth rates, six contained Native American reservations: 3 clusters in Montana and one each in South Dakota, North Dakota, and Wyoming. The remaining three clusters were located in Virginia (2 clusters), Florida (1 cluster), and South Dakota (1 cluster).

Table 2:

Highest Repeat Teen Birth Rate Clusters in the United States 2015 to 2017

Large City Within Cluster	Number of Counties in Cluster	Population in Cluster, 2017	Average Repeat Teen Birth Rate per 1,000†	Cluster Relative Risk‡
Population Greater than 100,000
McAllen, TX	32	2,637,245	10.85	3.29
Amarillo, TX	33	436,747	9.69	2.88
Lubbock, TX	38	1,197,093	9.07	2.71
Wilkesboro, NC	4	101,897	8.68	2.58
Decatur, IL	1	105,801	8.33	2.47
Peoria, IL	1	183,011	8.05	2.39
St. Louis, MO	1	308,626	7.08	2.10
El Paso, TX	7	1,189,095	7.02	2.09
Garden City, KS	14	108,705	6.82	2.02
Twin Falls, ID	5	160,744	6.73	2.00
Population 100,000 or under
Minnewaukan, ND*	1	6,936	29.79	8.84
Wolf Point, MT*	2	14,567	20.93	6.21
Mission, SD*	7	21,775	19.09	5.67
Huron, SD	1	18,157	14.19	4.21
Hardin, MT*	4	35,008	12.30	3.65
Petersburg, VA	1	31,750	11.40	3.38
Lander, WY*	1	39,803	11.30	3.36
Okeechobee, FL	1	41,605	10.69	3.17
Havre, MT*	6	49,461	9.58	2.84
Roanoke, VA	1	99,837	7.97	2.37

Note:

^*Cluster under 100,000 contains a Native American Reservation or Tribal Entity

^†The national repeat teen birth rate from 2015–2017 was 3.35 per 1,000 girls ages 15–19.

^‡Cluster relative risk indicates the risk of repeat teen birth versus what would be expected under the null hypothesis of no elevated risk in the cluster.

Sources: Centers for Disease Control and Prevention. U.S. Vital Statistics Natality files, 2015–2017. https://www.cdc.gov/nchs/data_access/vitalstatsonline.htm

U.S. Dept. of Homeland Security; Homeland Infrastructure Foundation-Level Data, https://hifld-geoplatform.opendata.arcgis.com/datasets/indian-lands-and-native-entities

County characteristics

Table 3 describes county-level demographic and socioeconomic characteristics and medical provider densities of counties that appeared in clusters of elevated 1) Repeat Birth (including counties with high repeat birth rates only and counties with both high first and repeat birth rates), 2) First Birth Only, or 3) Neither First nor Repeat Birth. Neither First nor Repeat counties differed significantly from Repeat Birth and First Birth Only counties on all county-level characteristics.

Table 3:

US County Demographic and Socioeconomic Characteristics and Medical Provider Density by Teen Birth Cluster Type

	Mean±Standard Deviation
	High Repeat Teen Birth Rate Only & Both High Repeat & First‐time Birth Rates*   n = 1,184	High First‐time Teen Birth Rate  n = 504	Neither n = 1,420
Race: Females 15 to 19 ‐ 2010
Percent(%) White	73.95±21.76†,§	78.59±20.76† ,‡	83.55±14.76‡ ,§
Percent (%) Black	16.05±21.32†,§	9.65±18.02†,‡	6.79±11.79‡,§
Percent (%) Native American	2.31±8.68†,§	3.84±12.30†,‡	1.55±4.42‡,§
Percent (%) Asian	0.65±0.94†	0.72±1.15†	1.69±2.79‡,§
Percent (%) Two or more races	2.60±2.42†,§	3.10±1.98‡	3.03±1.85‡
Percent (%) Other race	4.44±6.48†	4.10±6.13†	3.38±4.69‡,§
Ethnicity: Females 15 to 19 ‐ 2010
Percent (%) Hispanic/Latino	12.19±19.14†	10.72±15.10†	8.86±12.05‡,§
Percent (%) Non‐Hispanic/Latino	87.81±19.14†	89.28±15.10†	91.14±12.05‡,§
Percent (%) Urban Population ‐ 2010	38.50±28.58†	37.81±31.40†	45.22±33.26‡,§
Percent (%) Poverty Rate ‐ 2015	19.86±6.73†,§	16.56±6.01†,‡	13.25±4.58‡,§
Percent (%) Under 19 Uninsured ‐ 2015	6.95±3.68†	7.19±3.54†	5.87±3.03‡,§
Percent (%) with High School Diploma, ages 25+ ‐ 2011‐2015	81.29±6.47†,§	85.28±5.93†,‡	88.83±4.82‡,§
Percent (%) Unemployed Rate, ages 16+ ‐ 2016	5.93±1.99†,§	5.28±1.72†,‡	4.62±1.44‡,§
GINI Coefficient ‐ 2011‐2015║	0.453±0.035†,§	0.440±0.035† ,‡	0.436±0.032‡,§
Medical Providers¶
Primary Care Physicians ‐ 2015	0.436±0.285†	0.468±0.320†	0.614±0.395‡ ,§
Obstetrician‐Gynecologists ‐ 2015	0.044±0.063†	0.042±0.066†	0.061±0.080‡ ,§
Adv. Practice Nurse Midwives ‐ 2016	0.008±0.027†	0.011±0.033†	0.016±0.031‡ ,§
Publicly Funded Clinics ‐ 2015	0.094±0.119†,§	0.120±0.171† ,‡	0.067±0.130‡ ,§
Female contraceptive clients at clinics ‐ 2015#	42.83±57.00†,§	36.99±59.91†,‡	30.28±41.78‡,§
ANOVA results on all measures were significant at the 1% level.

^*This category includes clusters with high rates of repeat teen births only and clusters with high rates of both repeat teen births and first‐time teen births.

^†Significantly different than group with neither high repeat nor first‐time teen birth rate, α = 0.05

^‡Significantly different than group with high repeat teen birth rate only or both high repeat and first‐time teen birth rate, α = 0.

^§Significantly different than group with high first‐time teen birth rate, α = 0.05

^║A Gini coefficient of 0 indicates perfect income equality, while a coefficient of 1 indicates perfect inequality.

^¶Density per 1,000 total county residents

^#Density per 1,000 female county residents

C. Gini.”On the Measure of Concentration with Special Reference to Income and Statistics.” 1936. Colorado College Publication, General Series No. 208, 73–79.

Centers for Disease Control and Prevention. U.S. Vital Statistics Natality files, 2015‐2017. https://www.cdc.gov/nchs/data_access/vitalstatsonline.htm

JJ Frost, et al. Publicly Funded Contraceptive Services at U.S. Clinics, 2015. New York: Guttmacher Institute. 2017. https://www.guttmacher.org/report/publicly‐funded‐contraceptive‐services‐us‐clinics‐2015

Steven Manson, et al. IPUMS National Historical Geographic Information System: Version 13.0 [Database]. Minneapolis: University of Minnesota. 2018. http://doi.org/10.18128/D050.V13.0

U.S. Department of Health and Human Services. County Area Health Resources File, 2016‐2017 Release. https://datawarehouse.hrsa.gov/topics/ahrf.aspx.

The comparison of interest, Repeat Birth versus First Birth Only, also revealed a number of differences in county characteristics. In Repeat Birth counties, the adolescent female population was more likely to be Black and less likely to be White, Native American, and multiracial. There were also higher rates of county-level poverty, fewer residents with a high school diploma, a higher unemployment rate, and higher income inequality. There were no significant differences in medical provider density for any provider type. However, Repeat Birth counties had significantly fewer publicly-funded family planning clinics, and clinics in those counties served significantly more female contraceptive clients per capita.

DISCUSSION

Consistent with previous research mapping teen birth rates, we found counties with high rates of both first-time and repeat teen births in large population areas were concentrated in the Southern United States.^15,16 However, in small population areas, counties with the highest first and repeat teen birth rates were located primarily in Northern states and in states with large Native American populations, as noted in previous work.^3,16,27

Our results extend previous studies by revealing the extent to which geographic clustering of teen births varies by parity. We found that while most counties experiencing elevated first teen birth rates also experience elevated repeat birth rates, first and repeat births do not completely overlap. Counties with elevated repeat teen birth rates had worse socioeconomic conditions, including higher poverty rates, lower high school graduation rates, higher unemployment rates, and higher income inequality compared to counties with high first birth rates only. Previous work has shown higher rates of community-level poverty and economic inequality to be associated with higher rates of teen pregnancy.^14,18,28 Economic inequality likely relates particularly to elevated repeat teen birth rates via two mechanisms: 1) fewer perceived economic opportunities among low-SES teen mothers in high inequality areas and 2) unequal access to pregnancy prevention services. Previous research demonstrates lower levels of perceived community opportunity predict higher teen pregnancy rates, whereas higher educational expectations (signaling higher perceived opportunity to obtain education) are related to lower teen pregnancy rates.^14,28,29 Our results indicate this may be particularly true for repeat teen mothers.

Compounding the perceived lack of opportunity are inequalities in access to services that prevent teen pregnancies, including sex education and access to family planning services.^30,31 Counties with high rates of repeat teen births had fewer publicly-funded clinics and more female contraceptive visits to those clinics compared to counties with high rates of first teen births only. Greater use of fewer clinics alongside higher rates of repeat teen birth likely indicates a need for additional, accessible contraceptive services in those counties. The Guttmacher Institute estimated that publicly-funded family planning clinics met only 47% of the need for family planning services among low-income women in 2010.²¹ Further, the presence of obstetrician-gynecologists and family planning clinics in a community does not guarantee the availability of highly-effective contraception methods, nor abortion services, for teens.^31–33 Thus, our findings point to a need for additional providers in counties with high repeat teen birth rates to work to ensure teen mothers’ contraceptive needs are adequately addressed.

These results reinforce the need to consider parity in the research and monitoring of teen births. They further demonstrate a need to allocate resources and tailor teen pregnancy prevention programs to the types of teen births that are most common in each respective area. Counties with elevated repeat teen birth rates may wish to increase access to immediate postpartum long-acting reversible contraceptives or home visiting programs, which are known to delay or reduce repeat teen pregnancies but are not relevant strategies for preventing first births.^34,35

Strengths, Limitations, and Future Research

Strengths of the current study include its inclusion of virtually all births occurring to U.S. teen mothers during the study period, linkage to both county-level demographic and socioeconomic characteristics and provider counts, and attention to parity in teen childbearing, an oft-neglected factor. The primary study limitation relates to the data source, birth certificates, which afford the ability to research parity but not gravidity. We were not able to capture subsequent pregnancies to teenage mothers that did not result in births (i.e., those ending in miscarriage or abortion). Our data therefore do not reflect all teens who experienced a pregnancy, only those who gave birth. Further, our analyses are cross-sectional, and measures of county characteristics were not available in every year; therefore, results should be interpreted as associations but not causal. Future research on teen pregnancy prevention and optimizing pre/perinatal care for adolescents should stratify teen births by parity to consider specific needs and characteristics of these two populations. Future research should examine both parity and gravidity in order to understand the extent to which county-level characteristics relate to teenage pregnancies as well as births. Longitudinal and/or quasi-experimental studies examining changes in county socioeconomic and medical provider characteristics and their effects on teen birth rates would also illuminate the extent to which contextual changes, rather than more common individual-level interventions, can reduce rates of unplanned teen childbearing.

Conclusions

In sum, our results indicate that parity is a significant consideration in the epidemiology of teen childbearing. Repeat teen mothers experience disparate risk factors and perinatal outcomes compared to first-time teen mothers and reside in differentially-resourced geographic areas. Adolescent health care providers in regions with elevated teen birth rates should be aware of the unique characteristics and environments of repeat teen mothers in order to offer tailored contraceptive care strategies and improve perinatal outcomes among this vulnerable population.

IMPLICATIONS AND CONTRIBUTION.

Teen pregnancy prevention strategies and pre/perinatal care for adolescents should be implemented with consideration of the specific contexts that give rise to higher rates of repeat birth and tailored accordingly in order to prevent both first and repeat births.