CHILDHOOD PREGNANCY (10-14 YEARS OLD) AND RISK OF STILLBIRTH IN SINGLETONS AND TWINS

HAMISU M. SALIHU, MD, PHD, PUZA P. SHARMA, MD, MPH, OLANIYI J. EKUNDAYO, MD, MPH, SIBYLLE KRISTENSEN, MPH, ABIODUN P. BADEWA, MD, PHD, RUSSELL S. KIRBY, PHD, AND GREG R. ALEXANDER, SCD

Objective To clarify the association between childhood pregnancy and risk of stillbirth. Study design We analyzed singleton and twin pregnancies that occurred in children (10-14 years old) in the United States
from 1989 to 2000. We estimated the absolute and relative risks of stillbirth by using 15- to19-year-old and 20- to 24-year-old mothers as comparison groups. Results The analysis involved 17.8 million singletons and 337,904 individual twins. The rate of stillbirth was highest in pediatric mothers for both singletons (12.8/1000) and twins (56/1000) compared with adolescent (6.8/1000 in singletons and 29/1000 in twins) and mature (5.5/1000 in singletons and 20/1000 in twins) mothers. After adjusting for confounding characteristics, pediatric mothers continued to exhibit significantly elevated risk for stillbirth in both singletons (odds ratio, 1.57; 95%CI, 1.49-1.66) and twins (odds ratio, 1.97; 95%CI, 1.42-2.73). Preterm birth rather than small size for gestational age was revealed by means of sequential modeling to account for the excess risk of stillbirth observed in pediatric gravidas. Conclusion Pregnancy in childhood is a risk factor for stillbirth; shortened gestation rather than reduction in fetal growth is the mediating pathway. (J Pediatr 2006;148:522-6)

espite a progressive decrease in rates of birth to teenagers since the early 1990s, teen pregnancy remains a significant public health concern in the United States. The birth rate among this is high (41.7 per 1000 in 2003, thrice the national average of 14.1 per 1000), with much higher rates in minority subgroups.1 Teen pregnancies are associated with higher rates of adverse birth outcomes.2-12 Compared with infants of older mothers, those of teenage mothers are more likely to be preterm,2,5 very preterm,3,9,11 small for gestational age,11 or of low and very low birth weight.5,6,10,11 Neonatal,5,9,11 postneonatal,7,9 and infant mortality rates also are higher in infants of teen mothers than in those born to older mothers.3,6,10 Compared with those of 15- to 19-year-old mothers, pregnancy outcomes of the youngest adolescents or pediatric mothers (Ͻ15 years) are less well studied. The latest statistics indicate that the birth rate is as high as 0.6 per 1000 in this age group.1 Earlier studies on adverse fetal outcomes in pediatric mothers have only rarely examined the occurrence of stillbirth in this age group.13,14 This is unfortunate, because stillbirth could lead to subsequent poor obstetric outcomes and disturb future reproductive See related article, p 527 functions and cause psychological trauma with long-lasting scars as these mothers grow older.13,15 We therefore carried out this study to estimate the risk of stillbirth in singleton From the Department of Maternal and Child Health, Department of Epidemiology, and twin deliveries to pediatric mothers. The inclusion of multiples (twins) is necessitated University of Alabama at Birmingham, Birby the current increase in multiple births in the United States across all demographic mingham, Alabama; Department of Family 16,17 subgroups. Medicine, Baptist Health Systems, Mont-

D

METHODS
We used the “natality data files” and the “fetal death files” assembled by the National Center for Health Statistics for 1989 to 2000. The natality files contain individual records of all live births, whereas the fetal death files include all fetal deaths that occurred in the United States during the stated period. The procedures for quality control of the data are explained in detail elsewhere.18,19 The data source forms the basis for official US birth and death statistics. The study group for analysis was pediatric mothers, defined here as gravidas 10 to
LMP Last menstrual period SGA Small for gestational age

gomery, Alabama. Supported by a Young Clinical Scientist Award to H.M.S. by the Flight Attendant Medical Research Institute (FAMRI). Submitted for publication Apr 22, 2005; last revision received Sep 20, 2005; accepted Nov 3, 2005. Reprint requests: Hamisu Salihu, MD, PhD, Department of Maternal and Child Health, University of Alabama at Birmingham, 1665 University Blvd, Room 320, Birmingham, AL 35294. E-mail: hsalihu@uab.edu 0022-3476/$ - see front matter Copyright © 2006 Elsevier Inc. All rights reserved. 10.1016/j.jpeds.2005.11.018

522

14 years of age. We used the words “pediatric” and “childhood” pregnancy interchangeably in this study to refer to pregnancy in 10- to 14-year-old girls. We constructed 2 comparison groups consisting of women aged 15 to 19 years old (also designated as adolescent mothers) and 20 to 24 years old (also designated as mature mothers). We selected live births and fetal deaths of singleton and multiple pregnancies between 20 and 44 gestational weeks inclusive. We restricted our analysis of multiples to twins, because higher-order multiples were observed rarely among pediatric mothers. We compared these sociodemographic characteristics between pediatric mothers and the older maternal cohorts: maternal race, marital status, reported use of tobacco during pregnancy, and adequacy of prenatal care. Adequacy of prenatal care was assessed by using the revised graduated index algorithm,20,21 which has been found to be more accurate than several others, especially in describing the level of prenatal care use in groups of high risk.21 This index assesses the adequacy of care on the basis of the trimester in which prenatal care began, the number of visits, and the gestational age of the infant at birth. We also performed crude frequency comparisons for the presence of common obstetric complications, namely diabetes mellitus, pregnancy-associated hypertension, pre-eclampsia, eclampsia, chronic hypertension, abruption of the placenta, and placenta previa. We constructed a composite variable indicating the presence of at least 1 of these conditions. The main outcome of interest was stillbirth, which we defined as intrauterine fetal death at Ն20 weeks of gestation. The interval between the first day of the last menstrual period (LMP) and the date of birth was used as a means of computing gestational age in completed weeks. Records missing the date of the LMP were imputed by the National Center for Health Statistics when there was a valid month and year. Clinical estimate of gestation was used in the computation of gestational age in cases in which the date of the LMP was not reported or in which the LMP date was inconsistent with the birth weight.18 Approximately 4% to 5% of the gestational ages during the period were based on clinical estimate of gestation.

ourselves and other investigators to be accurate.22,23 Using this algorithm, we obtained a perfect 1:1 matching in 314,246 of the 337,904 individual records (93.0%), yielding 157,123 unique sibling pairs. This level of accuracy is similar to that previously reported for matching twin siblings for data sets from the same source.22,23 Crude frequency comparisons across maternal age groups were conducted on both matched and unmatched twins combined. To obtain adjusted estimates in singletons, multivariable logistic regression modeling was conducted. For twins, adjusted analysis was performed on matched pairs, and the generalized estimating equations 24 framework was applied to control for intra-cluster correlations by using the PROC GENMOD in SAS software version 9.1 (SAS, Cary, NC). All tests of hypothesis were 2-tailed with a type-1 error rate fixed at 5%. This study was approved by the institutional review board for human subjects at the University of Alabama at Birmingham.

RESULTS
We analyzed approximately 17.8 million singleton and 337,904 individual twin deliveries. The incidence of twin births increased with advancing age: 1.14%, 1.48%, and 2.05% in pediatric, adolescent, and mature gravidas, respectively. Among singleton deliveries, 130,620 were to 10- to14year-old mothers, 5,841,847 to mothers aged 15 to19 years, and 11,870,000 were to 20- to 24-year-old mothers. Of the total individual twins, 1500 were delivered to pediatric mothers, 88,269 to adolescent mothers, and 248,135 to mature mothers. The distribution of selected maternal sociodemographic characteristics by plurality status across maternal age categories is given in Table I. In both singleton and twin births, pediatric mothers were more likely to be single (Ͼ90%) compared with adolescent and mature mothers. Almost half all pediatric mothers were black, compared with one-fourth and one-fifth of adolescent and mature mothers, respectively. The level of prenatal smoking among pediatric mothers was about 50% less than that of the other 2 groups. Among singletons, adequacy of prenatal care was lowest for pediatric gravidas; in contrast, prenatal care was highest among twin gestations of pediatric mothers. The risk for pregnancy-associated maternal complications also was elevated in pediatric mothers compared with their older counterparts, regardless of plurality. Regardless of race/ethnicity, absolute risk for stillbirth was greatest in mothers aged 10 to 14 years in both singletons and twins. In singletons (Figure 1), Hispanic mothers had the lowest stillbirth rates, whereas black mothers had the highest stillbirth rates across all maternal age categories. However, in twin births (Figure 2) there were no striking differences among the 3 racial groups, except in 10- to 14-year-old mothers, in whom the crude frequency of stillbirth was greatest in white mothers. We present adjusted estimates for both singletons and twins by maternal age category in Table II. In model I, the estimates were generated after taking into account the effects
523

Statistical Analysis We computed stillbirth rates by dividing the number of stillbirths by the sum of live births and stillbirths, multiplied by 1000. For the analysis in twins, matched sibling twin pairs were not readily identifiable. Because it was imperative to match twin siblings to establish and capture the correlations that existed within twin sets, we applied an algorithm that consisted of maternal and infant identifier-characteristics to find matched twin pairs. Infants were considered to be matched twin pairs when the data for these variables matched: day of the week of birth, month of birth, state and county of birth, maternal age, maternal race/ethnicity, level of maternal education, place of birth of the mother, number of prenatal visits, month prenatal care visits began, and gestational age. The algorithm has been previously validated and found by

Childhood Pregnancy (10 –14 Years Old) And Risk Of Stillbirth In Singletons And Twins

Table I. Sociodemographic characteristics of mothers with singleton and twin pregnancies by maternal age categories
Singletons 10–14 (N ‫؍‬ 130,620; %) Marital status Married Single Missing Race White Black Hispanics Unknown Smoking Yes No Missing Adequacy of prenatal care Yes No Complications* Yes No 6.4 93.3 0.3 22.1 49.1 27.4 1.4 6.0 75.4 18.6 15–19 (N ‫؍‬ 5,841,847; %) 26.4 73.4 0.2 46.0 27.0 25.6 1.5 14.8 65.3 19.9 20–24 (N ‫؍‬ 11,870,000; %) 56.7 43.1 0.2 55.1 19.1 24.3 1.6 14.8 64 21.2 10–14 (N ‫ ؍‬1500; %) 6.5 91.8 1.7 26.3 47.2 24.5 2.0 6.1 76.6 17.3 Twins 15–19 (N ‫ ؍‬88,269; %) 24.7 74.4 0.9 42.4 34.7 21.5 1.5 12.2 68.3 19.5 20–24 (N ‫ ؍‬248,135; %) 53.1 46.2 0.7 52.5 26.7 19.3 1.6 14.0 66.2 19.8

P value Ͻ.001

P value Ͻ.001

Ͻ.001

Ͻ.001

Ͻ.001

Ͻ.001

20.3 79.7 6.9 93.1

29.9 70.1 5.9 94.1

37.1 62.9 6.1 93.9

Ͻ.001 Ͻ.001

64.2 35.8 13.4 86.6

50.4 49.6 12 88

40.7 59.3 11.2 88.8

Ͻ.001 Ͻ.001

N, Number. *Complications: The variable codes for the presence of at least 1 of the following: diabetes mellitus, pregnancy-associated hypertension, pre-eclampsia, eclampsia, chronic hypertension, abruption placenta, and placenta previa.

Figure 1. Race/ethnicity specific stillbirth rates by maternal age categories in singletons.

Figure 2. Race/ethnicity specific stillbirth rates by maternal age categories in twins.

of confounding characteristics. Using subsequent models (models I-VI), we sought to discern the pathway through which the elevated risk for stillbirth associated with young maternal age could have been mediated, testing for maternal complications (model II), congenital anomalies (model III), both maternal complications and congenital anomalies (model IV), feto-maternal competition for nutrients leading to fetal growth reduction or small for gestational age (model V), and shortened gestation or preterm birth (model VI). Only modest changes in the odds ratios were observed in both
524 Salihu et al

singletons and twins in models that tested for maternal complications as mediator (model II), and no change was detected when the congenital anomalies variable was assessed as a mediator (model III). The loading of pregnancy-associated complications and congenital anomalies onto the model simultaneously further confirmed the lack of effect of the congenital anomaly variable (model IV). The inclusion of small for gestational age (SGA) in the model after accounting for the effects of congenital anomalies and obstetric complicaThe Journal of Pediatrics • April 2006

Table II. Adjusted odds ratios for the association between-maternal age and stillbirth in singletons and twins
10–14 yr Model I Singletons Twins Model II (model I ϩ complications) Singletons Twins Model III (model I ϩ anomaly) Singletons Twins Model IV (model I ϩ complications and anomaly) Singletons Twins Model V (model IV ϩ SGA) Singletons Twins Model VI (model IV ϩ preterm birth) Singletons Twins 1.57(1.49–1.66) 1.97(1.42–2.73) 15–19 yr 1.05(1.04–1.07) 1.22(1.13–1.31)

1.67(1.58–1.77) 2.04(1.46–2.85) 1.57(1.49–1.66) 1.97(1.42–2.73)

1.07(1.06–1.09) 1.18(1.09–1.28) 1.05(1.04–1.07) 1.22(1.13–1.31)

1.67(1.58–1.77) 2.04(1.46–2.85) 1.65(1.55–1.75) 1.91(1.33–2.74)

1.07(1.06–1.09) 1.18(1.09–1.28) 1.04(1.02–1.05) 1.18(1.09–1.28)

0.90(0.85–0.96) 1.47(1.02–2.13)

0.87(0.86–0.88) 1.07(0.99–1.16)

P for trend Ͻ 0.001 Adjusted estimates were obtained after controlling for the confounding effects of race, marital status, parity, prenatal smoking, year of birth, adequacy of prenatal care received, and sex of the infant. The estimates for singletons were based on a logistic regression model. For twins, we went further and adjusted for intra-cluster correlations by applying the generalized estimating equation. In model II, we adjusted for pregnancy complications by loading the coded variable onto model I. The steps for model construction for subsequent models are as described in parentheses in the table.

tions (model V) did not alter the estimates further. The most important finding was the influence of preterm birth which caused a remarkable reduction in stillbirth risk for both singletons and twins of younger mothers. In singletons, accounting for the effects of preterm birth, young motherhood changed from being a risk to a protective factor for stillbirth. In twins, a substantial reduction in the risk estimate also was observed, although not as pronounced as in singletons.

DISCUSSION
We found very young maternal age to be a risk factor for stillbirth in both singletons and twins. The likelihood of stillbirth was Ͼ50% greater in singletons and almost doubled in twins of pediatric mothers compared with infants of mature mothers (20-24 years old). Previous studies have reported heightened risk for unfavorable fetal outcomes in pediatric mothers,14,25–32 but information on stillbirth in these mothers is limited.13,28 In a crude analysis involving 13 cases of stillbirth in 1622 mothers aged 11 to 15 years, the authors observed similar rates for fetal mortality in 11- to 15-year-old (0.8%), 16- to 19-year-old (0.9%), and Ն20-year-old (0.9%) mothers.13 However, that study could not make adjustments for observed differences in

baseline sociodemographic characteristics because of insufficient sample size. A similar shortcoming was observed in another study in which fetal death was examined in 1120 adolescent black mothers, of whom 203 were Ͻ15 years old.28 Several mechanisms have been postulated to explain the elevated risk for adverse birth outcomes in pediatric mothers. Proponents of the biologic immaturity hypothesis speculate that in the pregnant state, a direct feto-maternal competition for available nutrients occurs, undermining the ability of the growing fetus to achieve its optimal potential growth,26,33 leading to a higher-than-expected level of low birth weight. By examining the contribution of SGA and preterm birth separately, we found that shortened duration of pregnancy rather than reduced fetal growth was the mediating pathway in singletons and in twins, with more pronounced results in singletons. After accounting for preterm births, our model showed that pediatric mothers were at lower risk for stillbirth compared with mature mothers in single births, whereas the risk thresholds for stillbirth were similar among twin births. This suggests that to prevent excess stillbirth in pediatric mothers it will be more effective to address factors associated with preterm delivery than those causing fetal structural growth reduction. Because information on timing of stillbirth in the data set is limited, these findings must be interpreted with caution. The social environment theory posits that the adverse birth outcomes observed in very young mothers are attributable to their deprived social environment, in which poverty, limited education, and belonging to a racially/ethnically disadvantaged group exert a cumulative effect.26,29,31,32,34 –36 We adjusted for race as an indicator for minority status in all the models examined. Although race/ethnicity per se might have accounted to some degree for the increased risk, the relative risk still remained high after controlling for this variable, indicating that race/ethnicity did not sufficiently explain our findings. However, we could not adjust for the role of income inequality because this information was lacking in our data set, nor could we model the contribution of cultural practices and health beliefs that probably impact the seeking of modern obstetrical care during pregnancy. The third theory relates the use of preventive care (specifically, prenatal care) by young mothers to pregnancy outcomes. Horon et al hypothesized that younger mothers were more likely to obtain prenatal care late or not at all compared with mature mothers 37 and that this disparity may account for the unfavorable birth outcome disadvantage in younger mothers. Our data showed that although the level of prenatal care was lowest in pediatric mothers with singletons, pediatric mothers with twins had the highest level of prenatal care. This contrasting picture may result from a heightened awareness on the part of general health care providers when they encounter a child with twin pregnancy. Because our final models controlled for adequacy of prenatal care, it is unlikely that receipt of suboptimal prenatal care is the explanation for the increased risk of stillbirth among pediatric gravidas in this study.
525

Childhood Pregnancy (10 –14 Years Old) And Risk Of Stillbirth In Singletons And Twins

A limitation in this study is its inability to provide information on the causes of stillbirth. The study, however, includes the most comprehensive data to date on stillbirth occurrence. The results have important health policy implications, namely, the need for more resource infusion to prevent pregnancy in childhood, which, as this study reveals, occurs not infrequently.

REFERENCES
1. Martin JA, Kochanek KD, Strobino DM, Guyer B, MacDorman MF. Annual summary of vital statistics—2003. Pediatrics 2005;115:619-34. 2. Daltveit AK, Vollset SE, Otterblad-Olausson P, Irgens LM. Infant mortality in Norway and Sweden 1975-88: a cause-specific analysis of an increasing difference. Paediatr Perinat Epidemiol 1997;11:214-27. 3. Smith GCS, Pell JP. Teenage pregnancy and risk of adverse perinatal outcomes associated with first and second births: population based retrospective cohort study. BMJ 2001;323:1-5. 4. Meng-Chih L. Family and adolescent childbearing. J Adolesc Health 2001;28:307-12. 5. Blake DM, Lee MI. Twin pregnancy in adolescents. Obstet Gynecol 1990;75:172-4. 6. Woolbright LA. Postneonatal mortality in Alabama: why no progress in the 90s. Ann Epidemiol 2001;11:208-12. 7. Orvos H, Nyirati I, Hajdu J, Pal A, Nyari T, Kovacs L. Is adolescent pregnancy associated with adverse perinatal outcomes. J Perinat Med 1999;27:199-203. 8. Olausson PO, Cnattingius S, Haglund B. Teenage pregnancies and risk of late fetal death and infant mortality. Br J Obstet Gynecol 1999;106:116-21. 9. Botting B, Rosato M, Wood R. Teenage mothers and the health of their children. Popul Trends 1998;93:19-28. 10. Leland NL, Petersen DJ, Braddock M, Alexander G. Variations in pregnancy outcomes by race among 10-14-year-old mothers in the United States. Public Health Rep 1995;110:53-8. 11. Pillai VK, Bandyopadhyay S. Age effects on infant mortality controlling for race: a meta-analytical study. Health Care Women Int 1997;18:115-26. 12. Stevens-Simon C, Beach RK, Klerman LV. To be rather than not to be—that is the problem with the questions we ask adolescents about their childbearing intentions. Arch Pediatr Adolesc Med 2001;155:1298-300. 13. Satin AJ, Leveno KJ, Sherman ML, Reedy NJ, Lowe TW, McIntire DD. Maternal youth and pregnancy outcomes: middle school versus high school age groups compared with women beyond the teen years. Am J Obstet Gynecol 1994;177:184-7. 14. Scholl TO, Decker E, Karp RJ, Greene G, De Sales M. Early adolescent pregnancy: a comparative study of pregnancy outcome in young adolescents and mature women. J Adolesc Health Care 1984;5:167-71. 15. Fielding JE. Adolescent pregnancy revisited. N Engl J Med 1978;299:893-5. 16. Russell RB, Petrini JR, Damus K, Mattison DR, Schwarz RH. The changing epidemiology of multiple births in the United States. Obstet Gynecol 2003;101:129-35. 17. Salihu HM, Aliyu MH, Rouse DJ, Kirby RS, Alexander GR. Potentially preventable excess mortality among higher-order multiples. Obstet and Gynecol 2003;102:679-84.

18. Martin J, Curtin S, Saulnier M, Mousavi J. Development of the matched multiple birth file. In: 1995-1998 matched multiple birth dataset. NCHS CD-ROM series 21, no.13a. Hyattsville, Md: National Center for Health Statistics; 2003. 19. National Center for Health Statistics. 1995-1998 linked birth/infant death data set. Vital statistics of the United States: quality control procedures. Hyattsville, Md: US Department of Health and Human Services, Centers for Disease Control and Prevention; 2000. 20. Alexander GR, Cornely DA. Prenatal care utilization: its measurement and relationship to pregnancy outcome. Am J Prevent Med 1987;3:243-53. 21. Kogan MD, Martin JA, Alexander GR, Kotelchuck M, Ventura SJ, Frigoletto FD. The changing pattern of prenatal care utilization in the United States, 1981-1995, using different prenatal care indices. JAMA 1998;279:1623-8. 22. Salihu HM, Alexander MR, Shumpert NM, Pierre-Louis BJ, Alexander GR. Infant mortality among twins born to teenagers in the United States: black-white disparity. J Reprod Med 2003;48:257-67. 23. Pollack H, Lantz PM, Frohna JG. Maternal smoking and adverse birth outcomes among singletons and twins. Am J Public Health 2000;90:395-400. 24. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42:121-30. 25. Friede A, Baldwin W, Rhodes PH, Buehler JW, Strauss LT, Smith JC, et al. Young maternal age and infant mortality: the role of low birth weight. Public Health Rep 1987;102:192-9. 26. Naeye RL. Teenaged and pre-teenaged pregnancies: consequences of the fetal-maternal competition for nutrients. Pediatrics 1981;67:146-50. 27. Cooper LG, Leland NL, Alexander G. Effect of maternal age on birth outcomes among young adolescents. Soc Biol 1995;42:22-35. 28. Chang SC, O’Brien KO, Nathanson MS, Mancini J, Witter FR. Characteristics and risk factors for adverse birth outcomes in pregnant black adolescents. J Pediatr 2003;143:250-7. 29. Reichman NE, Pagnini DL. Maternal age and birth outcomes: data from New Jersey. Fam Plann Perspect 1997;29:268-81. 30. Fraser AM, Brockert JE, Ward RH. Association of young maternal age with adverse reproductive outcomes. N Engl J Med 1995;332:1113-7. 31. DuPlessis HM, Bell R, Richards T. Adolescent pregnancy: understanding the impact of age and race on outcomes. J Adolesc Health 1997;20:187-97. 32. Elster AB. The effect of maternal age, parity, and prenatal care on perinatal outcome in adolescent mothers. Am J Obstet Gynecol 1984;149:845-7. 33. Scholl TO, Hediger ML, Ances IG, Cronk CE. Growth during early teenage pregnancies. Lancet 1988;2:701-2. 34. Macleod S, Kiely JL. The effects of maternal age and parity on birth weight: a population based study in New York City. Int J Gynaecol Obstet 1988;26:11-9. 35. Berenson AB, Wiemann CM, McCombs SL. Adverse perinatal outcomes in young adolescents. J Reprod Med 1997;42:559-64. 36. Ekwo EE, Moawad A. Maternal age and preterm births in a black population. Pediatr Perinat Epidemiol 2000;14:145-51. 37. Horon IL, Strobino DM, MacDonald HM. Birth weights among infants born to adolescent and young mothers. Am J Obstet Gynecol 1983;146:444-9.

526

Salihu et al

The Journal of Pediatrics • April 2006