- Research article
- Open Access
- Open Peer Review
Socioeconomic factors and adolescent pregnancy outcomes: distinctions between neonatal and post-neonatal deaths?
© Markovitz et al; licensee BioMed Central Ltd. 2005
- Received: 16 December 2004
- Accepted: 25 July 2005
- Published: 25 July 2005
Young maternal age has long been associated with higher infant mortality rates, but the role of socioeconomic factors in this association has been controversial. We sought to investigate the relationships between infant mortality (distinguishing neonatal from post-neonatal deaths), socioeconomic status and maternal age in a large, retrospective cohort study.
We conducted a population-based cohort study using linked birth-death certificate data for Missouri residents during 1997–1999. Infant mortality rates for all singleton births to adolescent women (12–17 years, n = 10,131; 18–19 years, n = 18,954) were compared to those for older women (20–35 years, n = 28,899). Logistic regression was used to estimate adjusted odds ratios (OR) and 95% confidence intervals (CI) for all potential associations.
The risk of infant (OR 1.95, CI 1.54–2.48), neonatal (1.69, 1.24–2.31) and post-neonatal mortality (2.47, 1.70–3.59) were significantly higher for younger adolescent (12–17 years) than older (20–34 years) mothers. After adjusting for race, marital status, age-appropriate education level, parity, smoking status, prenatal care utilization, and poverty status (indicated by participation in WIC, food stamps or Medicaid), the risk of post-neonatal mortality (1.73, 1.14–2.64) but not neonatal mortality (1.43, 0.98–2.08) remained significant for younger adolescent mothers. There were no differences in neonatal or post-neonatal mortality risks for older adolescent (18–19 years) mothers.
Socioeconomic factors may largely explain the increased neonatal mortality risk among younger adolescent mothers but not the increase in post-neonatal mortality risk.
- Infant Mortality
- Neonatal Mortality
- Young Mother
- Adolescent Mother
- Food Stamp
Adolescent pregnancy has long been considered to increase the likelihood of adverse infant outcomes . The infant mortality rate in the United States (U.S.) for young women has remained persistently high despite widespread study of this phenomenon. In 1999, the infant mortality rate for women under 20 years of age was 10.3 deaths per 1,000 live births compared to 7.0 per 1,000 live births for all ages  Although the U.S. teenage (ages 15–19) birth rate has declined substantially during the last decade, from 62.1 to 48.7 per 1,000 women in 2000 , this country still leads the developed countries by wide margins. Simple calculations would suggest that, with an estimated 470,000 births to teen mothers in the year 2000, over 1,500 excess infant deaths occurred among adolescent mothers.
The etiology explaining the higher mortality risk for adolescent mothers has been debated for as long as the phenomenon has been observed. Miller argued that socioeconomic status has a direct effect on infant mortality, confounding the role of young maternal age, in a study of 2,019 U.S. counties from 1971–1975 . Many studies point towards low birth weight (including infants born preterm and those small-for-gestational age) as the predominant proximate marker of infant mortality [5–10]. Fraser, in a large study spanning 20 years of live births from Utah, found maternal age to be an independent predictor for poor birth outcomes, but this study did not specifically address infant mortality or socioeconomic status directly . The defining influence of the biology of adolescent pregnancy, risky adolescent behavior, nutrition, race, socioeconomics, and adequacy of prenatal care – or the interaction among these variables – remains controversial. Certainly poverty, minority status, suboptimal prenatal care, poor education and unmarried status are more common in teenage compared to older mothers, and are known risk factors for low birth weight, preterm birth and higher infant mortality in the United States . Adolescent mothers are disproportionately poor. For example, in Missouri for 1997–1999, 79% of births to women aged 14–19 were covered by Medicaid compared to only 35.7% of births to women 20 and older . Among adolescents living in poor communities, 69% report having had intercourse, compared to 37% of those living in more affluent neighborhoods . Support can be found to defend or refute all or none of these factors as independent risk factors [6, 7, 11, 14–16].
We examined linked birth-death certificate data for Missouri residents to assess the role of poverty status and other potential risk factors on infant mortality among teenage mothers. We hypothesized that teen motherhood is highly confounded by poverty, and that there is no independent effect of adolescent age on infant mortality. Our analysis distinguished neonatal and post-neonatal mortality among younger and older adolescent mothers.
This was a population-based cohort study using Missouri's linked birth-death certificate files for 1997–1999. This period was chosen since it provided a sufficiently large study population with the most current birth-related data. To test the primary hypothesis, we compared infant mortality rates of adolescent and older mothers while controlling for socioeconomic status (defined by participation in poverty assistance programs) and other known and purported risk factors.
The study population included all singleton live births for mothers who were Missouri residents and less than 20 years of age during 1997–1999. The study population was stratified by maternal age to identify younger (12–17 years) and older (18–19 years) adolescent mothers. A comparison population included a random sample of singleton live births for mothers who were 20–34 years old during 1997–1999, since the infant mortality rates were relatively stable for this age group but higher for older mothers. Women other than non-Hispanic white or African American were excluded, because they comprised less than 3% of all live births in Missouri during the study period.
The following covariates were extracted from the birth certificate: race/ethnicity (non-Hispanic black or non-Hispanic white), maternal education level (recorded as age appropriate based on the number of years of formal education completed by the mother), parity, marital status, tobacco or alcohol use (as dichotomous and categorical covariates based on the number of cigarettes per day or drinks per week), maternal weight gained during pregnancy, adequacy of prenatal care utilization (based on the Kotelchuck index), and socioeconomic status. The poorest women were defined as those receiving food stamps, since this program includes only those with personal incomes less than 130% of the federal poverty level. Poor women were those defined as receiving Medicaid or WIC. In Missouri, pregnant women at or below 185% of the federal poverty level are eligible for Medicaid and WIC. While personal incomes can be used to identify women participating in poverty programs, we cannot assume that all non-participants are above 185% of the federal poverty level. All other women included those who did not participate in any of the three poverty assistance programs. Other covariates were clinical estimate of gestational age (20–28, 29–32, 33–36, ≥37 weeks) and birth weight (<2500, ≥ 2500 grams). Infant mortality included deaths that occurred during the neonatal (0–27 days) and post-neonatal (28–364 days) period. The underlying causes of death were divided into five categories: 1) accidental, 2) respiratory arrest or sudden infant death syndrome, 3) infectious, 4) perinatal or congenital, 5) other deaths.
Chi-square tests and student's unpaired t-tests were used to determine if categorical and continuous covariates, respectively, were significantly different for younger versus older mothers. Overall infant mortality, as well as neonatal mortality and post-neonatal mortality, was assessed as a function of the three age cohorts and stratified by race. Logistic regression was used to assess how each covariate affected the role of maternal age on infant mortality risk. Covariates that altered the odds ratio (OR) for the primary relationship by more than 10% were considered significant and included in the full model. 95% confidence intervals (CI) were computed to estimate the precision of the OR. First-order interaction terms between each covariate and maternal age were tested in a secondary analysis. The significance of adding each interaction was tested with a p-value <0.01. The final model was examined for outliers, influential data and goodness of fit (Omnibus Test and Hosmer and Lemeshow test, p <0.05). The role of each covariate in the full model was also assessed using neonatal and post-neonatal mortality as the primary outcome in separate logistic regression analyses. SPSS (Chicago, IL, version 10.0) was used for all statistical analyses.
This research, reviewed by the Saint Louis University institutional review board, was classified as exempt from the U.S. Department of Health and Human Services regulations for the protection of human subjects. The exemption 45 CFR 46.101(b)(4) permits epidemiologic research that uses existing publicly available data that are maintained in such a manner that subjects cannot be identified directly or through identifiers linked to the subjects.
Characteristics of Missouri women with singleton live births, 1997–1999, by maternal age.
12 to 17 (N = 10,131)
18 to 19 (N = 18,954)
20–35 (N = 28,899)
Clinical Estimate of Gestational Age
Birth Weight Category
Normal (> 2500 gm)
Low (≤ 2500 gm)
Age appropriate education level
Food Stamp Program
Tobacco use During Pregnancy
Alcohol use during pregnancy
Infant, neonatal and post-neonatal mortality risk by age group, stratified by race
Rate (per 1000 births)
cOR (95% CI)
Rate (per 1000 births)
cOR (95% CI)
1.63 (1.02, 2.47)
2.58 (1.56, 4.24)
1.14 (0.79, 1.64)
1.75 (1.11, 2.75)
1.41 (0.84, 2.39)
1.23 (0.74, 2.04)
1.15 (0.59, 2.26)
Logistic regression models: effect of maternal age on infant mortality.
Maternal Age (years)
Age-inappropriate educational Level
Underlying causes of post-neonatal death by maternal age (n = 180).
Cause of death (%)
Maternal age (years)
Accidental n (%)
SIDS/respiratory arrest n (%)
Perinatal/congenital n (%)
Infectious n (%)
All other n (%)
Total n (%)
We conclude that neonatal mortality rates do not differ by maternal age, but post-neonatal rates may be higher for mothers aged 12–17 years after adjusting for socioeconomic and other known risk factors. Our results are most similar to Reichman and Pagnini, who studied all New Jersey births in 1989 and 1990 . They found that controlling for medical and socioeconomic factors accounted for some, but not all the effect of young maternal age on infant mortality. The unadjusted odds ratio of infant mortality among white 15–17 year olds in their study was 2.5; the adjusted value was 1.6. Their study was strengthened by access to linked uniform billing hospital discharge records in addition to state vital statistics. They were therefore able to assess socioeconomic status much more directly in their model, comparing Medicaid recipients and "self-pay" to the privately insured.
In our study, we were limited to inferring poverty status by the dichotomous values of participation in WIC, Medicaid, or food stamps programs as noted on the birth certificate. Recognizing the eligibility criteria for these programs (130% of federal poverty level for food stamps and less than 185% of poverty level for WIC and Medicaid) allowed us to predict confidently income levels of program participants, but we cannot conclude that those who are not participants are of higher income levels. Particularly in the aftermath of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, many individuals who were eligible may not have enrolled in such programs. Nevertheless, 75–80% of the two younger maternal cohorts were enrolled in Medicaid and/or WIC, compared to only 36% of the older mothers. Such dramatic differences are likely to be truly related to socioeconomic distinctions and not bureaucratic or regulatory obstacles unique to older women. This study is also limited by the validity of the original data entered on birth and death certificates. Of particular concern would be the possibility of differential bias in reporting of information between older and younger mothers since there is evidence that reporting infant mortality may suffer from a racial bias and the race distributions are not identical in our age cohorts .
Our finding that participation in WIC and Medicaid appears to be protective against neonatal mortality is consistent with a study by Moss and Carver (1998) . Using data from the 1988 National Maternal and Infant Health Survey, among women whose household income was below 185% of federal poverty level, they found an adjusted odds ratio of infant mortality for WIC participation of 0.64 (CI 0.52, 0.77). In their full model, there was no significant effect of Medicaid, but this compares to an adjusted odds ratio of 1.39 (CI 1.04, 1.86) for "self-pay" individuals versus privately insured as the reference population. Recognizing that WIC and Medicaid, as well as the food stamps program, are not simply markers of low socioeconomic status but are also intervention programs, designed to improve the economic and nutritional condition of the disadvantaged – in particular pregnant women – may explain the apparent protective effect. Our study provides additional evidence that this is a biologically plausible observation. If we accept the premise that low socioeconomic status increases infant mortality by increasing the risk of preterm births, thus resulting in higher neonatal mortality, then adding the infant's gestational age into the logistic regression model should not significantly affect the primary relationship between maternal age and neonatal mortality, and indeed this is the case in our study.
Infant mortality is not a "black box;" the reasons newborns may die are very different from why 11 month olds die. As we can see, distinguishing neonatal from post-neonatal mortality is critical to appreciate fully the relationship between young maternal age and infant mortality. Certainly this increased post-neonatal mortality rate cannot be ascribed to "biologic" factors of younger mothers, where the debate over the mechanism of increased infant mortality with young maternal age has so often raged. The increased risk of post-neonatal mortality among the 12–17 year old mothers remains despite adjustment for covariates in our model suggesting social factors not explained by poverty (at least as measured on our model). It is interesting that the factors protective against neonatal mortality – participation in Medicaid, WIC and/or food stamp programs – are apparent risk factors for post-neonatal mortality (aOR 1.60 for poor and aOR 2.29 for poorest women in the 12–17 year olds). We speculate that whatever protective effect these programs have on neonatal mortality has dissipated after the newborn period, and then they are only markers of lower socioeconomic status. It is intriguing to consider what the causes of death might be in the post-neonatal period and how they differ in younger mothers. Despite our cohort of almost 60,000 infants and an apparent doubling of the accidental and infectious death rates in the youngest mother cohort, these differences did not achieve statistical significance.
In summary, this population-based cohort study confirms the known association of young maternal age and infant mortality, but adds considerably to our understanding by the distinction of younger from older teens and neonatal from post-neonatal mortality. Socioeconomic factors likely account for most of the young mothers' increased risk of neonatal mortality, but despite adjustment, a substantially increased risk of post-neonatal mortality exists for the youngest mothers. Our analysis points towards an increased risk of accidental and infectious deaths in these infants, raising questions of maternal maturity and ability to supervise adequately these developing infants.
We thank Garland Land, Joseph Stockbauer, and Janice Bakewell from the Missouri Department of Health and Senior Services for their helpful comments and support while conducting this study.
- Stickle GM, Ma P: Pregnancy in adolescents: scope of the problem. Contemporary OB/GYN. 1975, 5: 85-95.Google Scholar
- National Center for Health Statistics: Infant Mortality Statistics from the 1999 Period Linked Birth/Infant Death Data Set. National Vital Statistics Report. 2002, 50 (4): 2001-1120. (PHS)Google Scholar
- National Center for Health Statistics: Births to Teenagers in the United States, 1940–2000. National Vital Statistics Report. 2001, 49 (10): 1-24. (PHS) 2001-1120Google Scholar
- Miller MK, Stokes CS: Teenage fertility, socioeconomic statue and infant mortality. Journal of Biosocial Science. 1985, 17 (2): 147-155.View ArticlePubMedGoogle Scholar
- Dollfus C, Patetta M, Siegel E, Cross AW: Infant mortality: a practical approach to the analysis of the leading causes of death and risk factors. Pediatrics. 1990, 86 (2): 176-183.PubMedGoogle Scholar
- Geronimus AT: The effects of race, residence, and prenatal care on the relationship of maternal age to neonatal mortality. American Journal of Public Health. 1986, 76 (12): 1416-1421.View ArticlePubMedPubMed CentralGoogle Scholar
- Geronimus AT, Korenman S: Maternal youth or family background? On the health disadvantages of infants with teenage Mothers. American Journal of Epidemiology. 1993, 137 (2): 213-225.PubMedGoogle Scholar
- Zuckerman B, Alpert JJ, Dooling E, et al: Neonatal outcomes: is adolescent pregnancy a risk factor. Pediatrics. 1983, 71 (4): 489-493.PubMedGoogle Scholar
- Hutchins FL, Norman K, Rubino J: Experience with teenage pregnancy. Journal of the American College of Obstetricians and Gynecologists. 1979, 54 (1): 1-5.Google Scholar
- Lee K, Corpuz M: Teenage pregnancy: trend and impact on rates of low birth weight and fetal, maternal, and neonatal mortality in the United States. Clinics in Perinatology. 1988, 15 (4): 929-942.PubMedGoogle Scholar
- Fraser AM, Brockert JE, Ward RH: Association of young maternal age with adverse reproductive outcomes. The New England Journal of Medicine. 1995, 332 (17): 1113-1117. 10.1056/NEJM199504273321701.View ArticlePubMedGoogle Scholar
- Missouri Department of Health and Senior Services. Accessed April 25, 2005, [http://www.health.state.mo.us/BirthMICA/index.html]
- Vital and Health Statistics: Trends in pregnancies and pregnancy rates by outcome: Estimates for the United States, 1976–1996. National Vital Statistics System. 2000, 21 (56):Google Scholar
- Cowden AJ, Funkhouser E: Adolescent pregnancy, infant mortality, and source of payment for birth: Alabama residential live births, 1991–1994. Journal of Adolescent Health. 2001, 29: 37-45. 10.1016/S1054-139X(01)00217-8.View ArticlePubMedGoogle Scholar
- DuPlessis HM, Bell R, Richards T: Adolescent pregnancy: understanding the impact of age and race on outcomes. Journal of Adolescent Health. 1997, 20: 187-197. 10.1016/S1054-139X(96)00174-7.View ArticlePubMedGoogle Scholar
- Lawlor DA, Shaw M: Too much too young? Teenage pregnancy is not a public health problem. International Journal of Epidemiology. 2002, 31: 552-554. 10.1093/ije/31.3.552.View ArticlePubMedGoogle Scholar
- Kotelchuck M: The adequacy of prenatal care utilization index: Its U.S. distribution and association with low birthweight. American Journal of Public Health. 1994, 84: 1486-1489.View ArticlePubMedPubMed CentralGoogle Scholar
- Reichman NE, Pagnini DL: Maternal birth outcomes: data from New Jersey. Family Planning Perspectives. 1997, 29: 268-272.View ArticlePubMedGoogle Scholar
- Farley DO, Richards T, Bell RM: Effects of reporting methods on infant mortality rate estimates for racial and ethnic subgroups. Journal of Health Care for the Poor and Undeserved. 1995, 6 (1): 60-75.View ArticleGoogle Scholar
- Moss NE, Carver K: The effect of WIC and Medicaid on infant mortality in the United States. American Journal of Public Health. 1998, 88: 1354-1361.View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2458/5/79/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.