Risk of Preterm Birth in Multiparous Teenagers FREE

STUDIES^{1 - 8} examining outcomes of teenage pregnancies have demonstrated increased risk for a variety of poor birth outcomes. Most studies^{1 - 4 ,7 - 8} attribute this increased risk to socioeconomic characteristics rather than to physiologic factors related to young maternal age. Indeed, teenagers who bear a child are more likely than older mothers to be socioeconomically disadvantaged, to be unmarried, and to underachieve in school.^{1 ,5 ,8 - 10} However, the debate about the contribution of young maternal age to adverse birth outcomes has continued with a recent study⁵ demonstrating that young age remains a risk factor for low birth weight (LBW) and preterm birth in first births to teenagers after controlling for confounding socioeconomic characteristics. The relative contributions of young age and confounding socioeconomic risk factors to poor birth outcomes are even less clear for repeat births to teenagers, partly because much of the research on teenage childbearing has focused on outcomes for first births.

Birth outcome in teenage multiparas is an important public health concern because of the high rate of repeat pregnancies in teenagers. Estimates of repeat pregnancy rates in teenagers within 12 months postpartum range from 17% to 25%, and more than 75% of these pregnancies are likely to be unplanned.^{11 - 12} Studies^{6 - 7 ,9 ,13 - 16} focusing on birth outcomes in multiparas present conflicting findings about the risk for teenagers bearing a subsequent child. Sweeney¹³ found higher mean birth weight in second births compared with first births but did not examine LBW or other adverse pregnancy outcomes. Graham⁷ and Jekel et al¹⁶ found increased LBW in second births to a cohort of teenage mothers. Eisner et al⁶ concluded that maternal age younger than 18 years remains as a risk factor for LBW for multiparas after adjusting for confounding factors. In a retrospective case-control study, Santelli and Jacobson¹⁴ found no increase in LBW for second births compared with first births in a cohort of teenage multiparas. However, this study demonstrated a higher occurrence of LBW to teenage multiparas compared with teenage mothers who postponed a second pregnancy until after adolescence. Finally, Blankson et al¹⁷ found lower occurrence of fetal growth retardation but higher occurrence of preterm delivery in second births compared with first births in a cohort of low-income teenagers.

A possible source of confusion in interpreting findings of past studies is the frequent use of LBW as an outcome measure when 2 distinct causes compose LBW: prematurity (birth at <37 weeks' gestation) and small size for gestational age (inadequate intrauterine growth). Our analysis focused on preterm birth because it accounts for most LBW in the United States (64%)¹⁸ and has a clear clinical relationship to infant morbidity and mortality. The morbidity associated with prematurity poses a considerable challenge to teenage mothers who are more likely to be facing additional sources of socioeconomic stress than older mothers. Teenagers with more than 1 birth face even greater prevalence of socioeconomic risk factors for preterm birth and social stress than first-time teenage mothers, including belonging to a lower socioeconomic group, having a first birth at a very young age (<16 years), and having lower educational achievement.^{10 - 11 ,14 ,16 ,19} We undertook this study to determine whether teenage multiparas are at higher risk for preterm birth than adult multiparas after controlling for health behavior and socioeconomic characteristics shown to explain the risk associated with teenage childbearing in studies of teenage primiparas.^{4 ,6 ,9}

We used the 1990-1996 US Natality Files, compiled from all US birth certificates by the National Center for Health Statistics, to compare the occurrence of preterm birth in 10- to 20-year-old multiparas with that in 25-year-old multiparas. We chose the years 1990 to 1996 as the most recent period that provided adequate numbers of very young teenagers. All second births to white non-Hispanic, black non-Hispanic, and Hispanic multiparas aged 10 to 20 years and 25 years residing in the United States were included. The youngest multiparas in the sample were aged 13 years. Sample sizes after exclusions (exclusions are discussed in the following paragraph) were 899,393 multiparas aged 13 to 20 years and 440,462 multiparas aged 25 years. We limited the analysis to second births for 2 reasons. First, because there are no unique identifiers in the US Natality Files, a multipara could be included more than once in a data set spanning several years unless a specific parity is chosen. Second, teenagers with more than 2 births are a select group more likely to face additional risk factors for poor birth outcome, some of which we are unable to take into account.

Observations missing information for birth weight, gestational age, or maternal education and observations for which the gestational age was implausible were excluded: 42,478 multiparas aged 13 to 20 years (4.1%) and 17,717 multiparas aged 25 years (3.4%). The problem of implausibly high or low gestational ages given the recorded birth weight has been noted in previous studies^{20 - 21} using vital records, especially at low gestational ages. A study²² of misdating in adolescent pregnancies also raised concern that gestational age based on last menstrual period (LMP) might be less accurate for adolescents than for adults. To eliminate observations with implausible gestational ages, we first analyzed gestational age based on LMP. If this age was implausible for the recorded birth weight based on published birth weight–for–gestational age curves,²¹ we then examined the clinical estimate of gestational age if available. Observations with implausible values for LMP-based gestational age and clinical estimates of gestational age (or only implausible LMP if no clinical estimate was available) were excluded according to the algorithm²¹ (1.3% of observations for 13- to 14-year-olds, 0.8% for 15- to 17-year-olds, 0.6% for 18- and 19-year-olds, 0.6% for 20-year-olds, and 0.4% for 25-year-olds). For the small number of observations with an implausible LMP but a plausible clinical estimate for gestational age, the clinical estimate was used and the observation was included in the data set.

The study population was stratified into 3 race/ethnicity categories—white non-Hispanic, black non-Hispanic, and Hispanic—because past research²³ has shown that the impact of socioeconomic risk factors for poor birth outcome differs by race. Age groups in the study population were defined as 13 and 14 years, 15 to 17 years, 18 and 19 years, and 20 years. Although most births to teenagers occur between the ages of 15 and 19 years, we included 13 and 14 years to capture the spectrum of risk during adolescence. Among teenagers 15 to 19 years old, 15- to 17-year-olds differ significantly from 18- and 19-year-olds with respect to sexual experience, pregnancy rates,^{18 ,24} and intentional pregnancies.^{5 ,25} Therefore, they are analyzed as separate groups. We included births to mothers aged 20 years because most of these births were conceived while the mothers were still aged 19 years. Mothers aged 25 years served as the comparison group for each race/ethnicity and age group strata because they are, in general, at relatively low risk for adverse birth outcomes. Furthermore, we wanted to choose a relatively young comparison group to minimize unmeasured socioeconomic characteristics between teenage and older mothers.

To include a wider range of socioeconomic and obstetric characteristics, it was necessary to construct 3 separate models. Model 1 includes the full sample, and models 2 and 3 include partial samples based on availability of data on selected characteristics. Sociodemographic characteristics available from the US Natality Files included marital status and education. Marital status was defined as married or unmarried. For 13- to 18-year-olds, adequate education was defined as completing the minimum number of expected grade levels for age.¹⁵ For those older than 18 years, completion of grade 12 was defined as adequate education. Obstetric characteristics available in the US Natality Files include history of previous LBW births, prenatal care, interpregnancy spacing, and maternal smoking. Prenatal care was categorized as (1) care initiated in the first or second trimester or (2) care initiated in the third trimester or no prenatal care. Previous work²⁶ has shown that outcomes in those missing prenatal care information on the birth certificate are most similar to those with no prenatal care. Thus, this group was included with those receiving late or no prenatal care. The interpregnancy interval reported by the National Center for Health Statistics is the interval between the current live birth and the previous live birth. To adjust for varying lengths of gestation, we subtracted the gestational age of the current live birth from the reported interval to calculate the time in months between delivery of the previous live birth and the conception of the current pregnancy. Using this calculated measure, short interpregnancy interval was defined as 6 months or less.^{6 ,27 - 28} Because data on interpregnancy interval is available for all states from 1990 to 1993 only, a separate analysis of the association between interpregnancy interval and preterm birth was performed using data from this period (model 2). Use of tobacco during pregnancy was reported for each state between 1990 and 1996 except for California, Indiana, New York, and South Dakota. A separate analysis was done eliminating observations from the nonreporting states (model 3). This omission decreased the sample of white and black mothers by approximately one third and Hispanic mothers by approximately one half.

The outcome variable, preterm birth, was defined as gestational age at birth of less than 37 weeks. Further analysis of preterm birth focused on distinguishing moderate preterm birth (33-36 weeks' gestation) from very preterm birth (<33 weeks' gestation). Crude odds ratios (ORs) and 95% confidence intervals (CIs) for preterm birth, moderate preterm birth, and very preterm birth were calculated for each age group compared with 25-year-olds. These unadjusted results showed that teenagers generally faced greater odds of having a very preterm than a moderate preterm birth, an outcome associated with much greater health risk for the infant. Therefore, we limited the remainder of the analysis to very preterm births. Adjusted ORs (AORs) and 95% CIs for very preterm birth were estimated using a logistic regression model (SAS version 6.09; SAS Institute Inc, Cary, NC). Dichotomous covariates in model 1 included marital status, education, prenatal care, and previous preterm birth. As mentioned in the previous paragraph, interpregnancy interval and smoking status were included in separate models (models 2 and 3) with subsets of the full data set.

From January 1, 1990, to December 31, 1996, 899,393 second births to adolescents aged 13 to 20 years and 440,462 second births to 25-year-old white non-Hispanic, black non-Hispanic, and Hispanic women met the study criteria. Of singleton first and second births to women aged 10 to 20 years during the study period, 22.7% were second births. Table 1 shows characteristics of mothers aged 13 to 20 years and the comparison group (25-year-old mothers) stratified by race/ethnicity. Compared with multiparas aged 25 years, those aged 13 to 20 years were more likely to be unmarried, to have inadequate education for age, and to obtain late or no prenatal care. Black and Hispanic multiparas were more likely than white multiparas to be unmarried, to receive inadequate prenatal care, and to belong to younger age groups. White and Hispanic multiparas were more likely than black multiparas to have inadequate education.

Table 2 shows percentages of preterm births (<37 weeks' gestational age), moderate preterm births (33-36 weeks' gestational age), and very preterm births (<33 weeks' gestational age) by race/ethnicity group. Percentages of moderate preterm and very preterm births decline with increasing maternal age. However, wide racial disparities exist. Black non-Hispanic multiparas have nearly twice the percentage of preterm births as white and Hispanic mothers. Black mothers aged 25 years have a percentage of preterm births similar to 15- to 17-year-old white non-Hispanic and Hispanic mothers.

Consistent with the pattern for preterm births, crude ORs of preterm birth decline with increasing maternal age (Table 3). Multiparas aged 13 and 14 years have 2.7 to 4.1 higher odds of preterm birth compared with 25-year-olds, depending on race/ethnicity. By age 20 years, the odds decline to 1.3 to 1.5. Table 3 also shows ORs for preterm births separated into 2 outcome categories: moderate preterm births (33-36 weeks' gestational age) and very preterm births (<33 weeks' gestational age). For very preterm births, the crude ORs are generally greater than those for all preterm births in teenage mothers. Although moderate preterm births make up most preterm deliveries, and thus have ORs similar to those for all preterm births, infants born very prematurely face the greatest risk of morbidity and mortality. Because younger mothers face substantially higher odds of this adverse birth outcome, the results in the remainder of this article focus on the risk of very preterm birth.

To account for confounding factors related to maternal age, AORs for very preterm birth were calculated using logistic regression (Table 4, model 1). As expected, the AORs are lower than the crude ORs (Table 3), but the same relationship remains: teenage multiparas remain at higher risk for very preterm birth after adjusting for differences in education, marital status, prenatal care, and previous preterm births. Adjusted ORs of very preterm birth for white non-Hispanic and Hispanic multiparas are of similar magnitude, and only as mothers reach age 20 years does their risk for preterm birth approach levels for 25-year-old mothers. The pattern among black non-Hispanic mothers differs. By age 20 years, black multiparas are not at greater odds of having a very preterm birth compared with 25-year-old mothers. However, black multiparas aged 13 to 19 years remain at higher risk for very preterm birth.

Because the interpregnancy interval is available from the US Natality Files for 1990 to 1993 only, results from a separate analysis are shown in Table 4 (model 2) and Table 5. Young multiparas are more likely to have short interpregnancy intervals (≤6 months) than older multiparas (Table 5). This higher prevalence is in part due to the definition of a teenage multipara: to give birth at least twice during adolescence, a relatively short interpregnancy interval is necessary. Most first births to teenagers occur after age 15 years,¹⁸ and those who become multiparas must give birth again before reaching age 20 years. In general, multiparas with a short interpregnancy interval face a higher occurrence of very preterm birth than those with intervals greater than 6 months (Table 5). When interpregnancy interval was added to the logistic regression model, the AOR of preterm birth associated with young age generally decreased (Table 4). The magnitude of the association between a short interpregnancy interval and preterm birth, however, was similar between age groups. These results suggest that although teenage multiparas are more likely to have a short interpregnancy interval, having a short interval does not pose a greater risk for very preterm birth for a teenage multipara than for an older multipara.

Another possible confounder not included in the full model is maternal smoking. Table 4 (model 3) and Table 6 show results for analysis of effects of smoking using data for 1990 to 1996, excluding the states not reporting tobacco use—California, Indiana, New York, and South Dakota. Table 6 shows that white non-Hispanic mothers have the highest prevalence of smoking and that the prevalence of maternal smoking for this group peaks between ages 15 and 19 years. Black non-Hispanic mothers have an increasing prevalence of smoking as maternal age rises. Hispanic multiparas have a relatively low prevalence of maternal smoking for all age groups. Except for the youngest white and black non-Hispanic multiparas, very preterm births tend to be more frequent in smokers than nonsmokers. When smoking was added to the basic logistic regression model (Table 4), the AOR of young age increased among 13- to 17-year-old black and white multiparas, reflecting the pattern of greater smoking in older black mothers and the apparently greater impact of smoking on preterm births in older black and white mothers. In contrast, the youngest Hispanic mothers face a lower AOR after adding smoking to the model. Overall, adding maternal smoking to the analysis resulted in relatively minor changes in the magnitude and no change in the pattern of AORs of preterm birth associated with young maternal age.

Results of this analysis suggest that repeat childbearing in teenagers is associated with an increased risk of very preterm birth. Compared with 25-year-old multiparas, risk is highest for the youngest teenagers but remains present through 15 to 19 years, the age group in which multiparous births to teenagers are most common. Although previous studies^{6 - 7 ,9 ,14} have demonstrated increased risk of LBW or small size for gestational age in teenage multiparas associated with young maternal age, only 2 studies to our knowledge specifically explored the risk of preterm birth in teenage multiparas. Jekel et al¹⁶ and Blankson et al¹⁷ demonstrated a higher occurrence of preterm birth among subsequent births to a cohort of teenagers compared with an index birth despite the fact that the mothers were older for the second birth. The present study provides population-based evidence that subsequent pregnancy in teenagers carries a risk of preterm birth that decreases progressively as maternal age approaches 25 years.

Our findings also show that socioeconomic and health factors explain in part the relationship of young age and higher risk of preterm birth, as seen in previous studies.^{3 ,6 ,8 - 9 ,13 - 14 ,16 ,29} Multiparous teenagers, especially minorities, are more likely to have inadequate education and to be unmarried than older mothers. Nevertheless, young age remains an independent risk factor for very preterm birth in our analysis after adjustment for the socioeconomic and risk factors available in national vital statistics data sets.

Maternal smoking and short interpregnancy interval are other possible explanations for the relationship between age and birth outcome. Adjusting for maternal smoking actually increased the risk of very preterm birth associated with young maternal age, as found in an earlier study.⁸ In contrast, the odds of very preterm birth in teenage mothers decreased slightly after adjustment for interpregancy interval. Many studies have found no association between short interpregnancy interval and poor birth outcomes.^{9 ,13 - 14 ,30} Studies^{6 ,27 - 28 ,31} that demonstrated a risk of preterm birth associated with short intervals focused on multiparas with very short intervals (<3-6 months). Our results (Table 5) suggest a modestly increased risk of preterm birth with an interval of 0 to 6 months compared with longer intervals but that short intervals do not have greater impact on the outcome of a teenage pregnancy compared with that of an older mother. Instead, young multiparas are disproportionately affected by the adverse risk of short interpregnancy interval because teenagers have a higher prevalence of short intervals. The higher prevalence is because of selection: to become a teenage multipara, one must give birth at least twice during adolescence. There are also more complex issues involving motivation of teenage mothers to use contraception and to space births that might play a role in this observed pattern.

The racial disparity in percentage of preterm births is striking. As is well known, black infants have substantially higher rates of prematurity than white or Hispanic infants.^{18 ,23 ,32} Although the absolute percentage diminishes with increasing maternal age, the relative difference between black and white infants does not decrease. Comparing black teenage multiparas with black 25-year-old multiparas masks the racial disparity because the percentage of preterm births in 25-year-olds is similar to that experienced by 15- to 17-year-old white and Hispanic multiparas. Thus, although AORs for young black multiparas are lower than for white and Hispanic teenage multiparas, the overall risk is far greater. Although analyzing this gap is beyond the scope of this study, addressing the racial disparity in preterm birth rates remains a vital part of any intervention to reduce adverse outcomes in the most vulnerable teenage mothers and in adult mothers.

The main limitation of this study is the limited data available in the US Natality Files for socioeconomic status and family background. We were unable to control for factors such as income and characteristics of place of residence. The socioeconomic status of adolescents relies more heavily on family background than for older women. In general, education has been shown to be a good indicator of socioeconomic status and a strong predictor of reproductive outcomes.^{15 ,33 - 34} However, the significance of this indicator for adolescents who are more likely to be in the process of completing their education is less clear, mainly because there is limited variation in this group by level of schooling.³⁴ In this study, we use a measure that better distinguishes educational achievement in those still younger than the age of high school graduation. Nonetheless, it is difficult to define socioeconomic status as clearly for adolescents as for adults. Another limitation is underreporting of maternal smoking on birth certificates. This might affect the reported prevalence of smoking and, to some degree, preterm rates by smoking status, as shown in Table 6. However, unless there are substantial differences in underreporting by maternal age, the relationship shown in model 3 (Table 3) should not be markedly affected.

Whether due to young age, unmeasured factors associated with young age, or a combination, teenage multiparas are at increased risk for preterm birth compared with older mothers after controlling for established socioeconomic and obstetric risk factors. Although teenage pregnancy rates are declining, the population of teenage mothers remains large and vulnerable, and efforts to prevent repeat pregnancies continue to deserve high priority. Teenage mothers are a unique group in that they are at high risk for a poor health outcome but they can be easily identified, unlike teenagers at risk for a first-time pregnancy. In contrast to other girls in their peer group, teenage mothers have contact with medical providers for prenatal care, delivery, and well-child visits for their infants. Health practitioners who interact with teenage mothers, even if exclusively for the care of the infant, are in a position to counsel them about contraception and family planning. Opportunities are also present in the public support systems associated with health care. For example, social workers who enroll teenage mothers and their infants in Medicaid and other programs can also be instrumental in referring these mothers to family planning services. Examination of the interaction between teenage mothers and the health care system could optimize the role of health care professionals in preventing repeat pregnancies and poor birth outcomes in teenage mothers.

Although our study is cross sectional, the results imply that the longer an adolescent can postpone a second pregnancy, the lower her risk of a very preterm birth. Until we are better able to identify and ameliorate the risks associated with young maternal age and multiparity, the clearest approach to reducing adverse outcomes for teenage mothers is postponing subsequent births until after adolescence.

Accepted for publication June 12, 2000.

Reprints: Lara J. Akinbami, MD, Infant and Child Health Studies Branch, National Center for Health Statistics, 6525 Belcrest Rd, Room 790, Hyattsville, MD 20782.