To investigate the relationship between adolescent pregnancy and neonatal mortality in a nutritionally deprived population in rural Nepal, and to determine mechanisms through which low maternal age may affect neonatal mortality.
Nested cohort study using data from a population-based, cluster-randomized, placebo-controlled trial of newborn skin and umbilical cord cleansing with chlorhexidine.
Sarlahi District of Nepal.
Live-born singleton infants of mothers younger than 25 years who were either parity 0 or 1 (n = 10 745).
Maternal age at birth of offspring.
Crude and adjusted odds ratios of neonatal mortality by maternal age category.
Infants born to mothers aged 12 to 15 years were at a higher risk of neonatal mortality than those born to women aged 20 to 24 years (odds ratio, 2.24; 95% confidence interval, 1.40-3.59). After adjustment for confounders, there was a 53% excess risk of neonatal mortality among infants born to mothers in the youngest vs oldest age category (1.53; 0.90-2.60). This association was attenuated on further adjustment for low birth weight, preterm birth, or small-for-gestational-age births.
The higher risk of neonatal mortality among younger mothers in this setting is partially explained by differences in socioeconomic factors in younger vs older mothers; risk is mediated primarily through preterm delivery, low birth weight, newborns being small for gestational age, and/or some interaction of these variables.
clinicaltrials.gov Identifier: NCT00109616
Pregnancy during adolescence is a significant problem globally, with the highest incidence rates occurring in developing nations.1 It is estimated that more than 14 million females between 15 and 19 years of age give birth each year, and more than 90% of these births occur in developing countries.2 Although early childbearing has often been regarded as a social issue, there is mounting evidence that young maternal age may be linked to adverse infant outcomes including low birth weight (LBW), preterm birth, and intrauterine growth restriction resulting in newborns small for gestational age (SGA), as well as neonatal mortality.3- 8 Attempts to elucidate the etiology of these poorer pregnancy outcomes among adolescent women have produced conflicting data, and considerable debate remains as to whether the excess risks are due to biologic immaturity or are the consequence of deleterious social and environmental factors.3,4,9- 11
Since younger mothers are more likely to be poor and less educated and to have inadequate prenatal care and fewer social supports than older mothers, socioeconomic and lifestyle factors often have been cited as the main explanatory variables for disparities in reproductive outcomes.12 However, a number of studies have shown strong associations between maternal age and adverse infant outcomes even after controlling for these factors.4,5 Thus, investigations in both industrialized and developing nations lend support to an intrinsic biologic risk associated with young maternal age.3,7,8
The adolescent period is a time of significant growth; 45% of adult weight and 15% of adult height are attained during this stage.13 Continued growth during pregnancy could result in competition between the mother and fetus for important nutrients and may be associated with increased risk of adverse pregnancy outcomes.14,15 In countries where chronic malnutrition is prevalent, the consequences of this competition may be even more detrimental to the mother and infant. Moreover, in adolescents, chronic malnourishment is associated with delayed age at menarche and prolonged puberty and thus may contribute to poor reproductive outcomes.16,17 In addition to pubertal growth, postmenarcheal growth may also affect pregnancy outcomes in such populations. Few studies, however, have investigated the risks associated with adolescent pregnancy in areas of high malnutrition.
In Nepal, chronic malnutrition is common and early marriage is customary; more than two-thirds of rural females are married by the age of 20.18 Rates of pregnancy are reported to be as high as 89 births per 1000 females aged 15 to 19, but births among younger females are also frequent.1 Early childbearing, however, is considered a successful outcome in Nepal, and the extended family network often helps to care for the infant, thereby eliminating some of the social and environmental problems associated with adolescent pregnancy.
The primary aim of this study was to investigate the relationship between young maternal age and neonatal mortality in a nutritionally deprived population in rural Nepal. We also sought to determine whether preterm birth, LBW, and being SGA are mechanisms through which low maternal age affects the risk of neonatal mortality in this setting.
This study used data from the Nepal Newborn Washing Study, a cluster-randomized, placebo-controlled, community-based trial of newborn skin and umbilical cord cleansing with chlorhexidine conducted in the Sarlahi District of Nepal between 2002 and 2005. The details of the methods and the main results of this trial have been reported elsewhere.19,20 Briefly, 413 sectors were randomized for newborn infants to receive 1 of 2 skin-cleansing regimens by a local female ward distributor (WD) immediately after delivery. The regimens consisted of full-body skin cleansing of the infant excluding the eyes and ears with either Pampers baby wipes (Proctor and Gamble Co, Cincinnati, Ohio) containing 0.25% free chlorhexidine or with baby wipes that lacked chlorhexidine (placebo).
Women were recruited for participation in the study at approximately 6 months' gestation by WDs who visited women in their area on a weekly basis. All women received weekly vitamin A supplements, iron–folic acid supplements, albendazole as well as tetanus immunization if deficient, and a clean delivery kit. At recruitment, project workers provided education regarding proper nutrition, hygienic delivery, and neonatal care, and collected data on educational level, literacy, and maternal health. Information regarding socioeconomic status including ethnic group, caste, latrine and cattle ownership, the presence of electricity in the home, and maternal occupation was also obtained at this time.
In this population, most women deliver at home with the assistance of family members or untrained traditional birth attendants. Ward distributors, who were alerted by relatives once labor began, visited the woman's home during or soon after delivery (mean, 6.8 hours after delivery) to provide the skin-cleansing intervention to the newborn. Subsequently, a birth assessment team arrived at the home to collect information regarding the delivery process and condition of the newborn and to measure axillary temperature and infant birth weight using a digital infant scale (Seca 727; Seca, Hamburg, Germany). Infant vital status was assessed on days 2, 3, 4, 6, 8, 10, 12, 14, 21, and 28 after birth. Infants with specific signs and symptoms were referred for medical care.
Only live-born singleton infants of women younger than 25 years who were either parity 0 or 1 were included in this analysis. In this population, most adolescents are parity 0 or 1 and almost all women aged 25 years or older are of parity greater than 1. Hence, confining the analysis to women of parity 0 or 1 and younger than 25 years would reduce confounding of the association between maternal age and survival by parity. Older parity 0 or 1 women also were excluded because their low parity may have been associated with reproductive or other health problems. Maternal age was calculated as the age of the mother at time of delivery based on information solicited from the mother at the enrollment interview. Gestational age was estimated from 2 maternal reports of time since last menstrual period; these estimates were provided at enrollment and at first assessment after delivery. Delivery before 37 weeks' gestation was defined as preterm. For birth weight data, only those weights measured within 72 hours of delivery were considered in this analysis, and infants were classified as SGA if their weight fell below the 10th percentile for gestational age and sex as defined by the US reference for fetal growth.21
Verbal informed consent was obtained from all participants. The study received ethical approval from the Nepal Health Research Council and the Committee on Human Research of the Johns Hopkins Bloomberg School of Public Health.
Statistical analysis was performed using Stata 9.0 (StataCorp, College Station, Texas). Four maternal age categories (12-15 years, 16-17 years, 18-19 years, and 20-24 years) were defined a priori based on evidence from similar research studies.3,5The distributions of various maternal, socioeconomic, and infant characteristics were compared across maternal age groups using χ2 tests for categorical data or analysis of variance for continuous measurements. To investigate the relationship between maternal age and death during the neonatal period, neonatal mortality rates (NMRs), calculated as the number of deaths within the first 28 days divided by the number of live births, were assessed by maternal age and stratified by parity. Because a number of deaths occurred prior to receiving the assigned intervention, a 2-level “treatment received” variable was created by categorizing infants into those who received chlorhexidine vs those who received either placebo or no treatment. This variable was assessed as a potential confounder and/or effect modifier.
Logistic regression models were constructed to compute crude and adjusted odds ratios (ORs) for neonatal mortality by maternal age category. Potential confounders significantly associated (P < .05) with both maternal age and neonatal mortality were included in the multivariable models. Interactions between maternal age and parity or treatment received were assessed through testing of product terms. The Hosmer-Lemeshow goodness-of-fit statistic was used to assess adequacy of model fit. Odds ratios and their 95% confidence intervals (CIs) were corrected using generalized estimating equations with an independent correlation structure to account for clustering within a subset of participants who contributed more than 1 pregnancy to the study.22
Finally, ORs for the risks of other adverse outcomes by maternal age category, including LBW, preterm delivery, and SGA, were obtained using logistic regression. Since these were thought to be in the causal pathway of infant survival, we assessed whether they were mediators of the relationship between maternal age and neonatal mortality by creating models that further adjusted for LBW, preterm delivery, or SGA, or both preterm delivery and SGA.
Of 23 296 live-born singleton infants, 10 745 mother-infant pairs met the selection criteria (Figure). Of these, 9077 infants were weighed within 72 hours of birth. The disproportionately higher mortality in infants without birth weight data is due to the majority of these deaths occurring soon after delivery, before arrival of the birth assessment team to weigh the infant. Similarly, of the 163 infants who did not receive the assigned intervention (chlorhexidine vs placebo), 106 died before the arrival of the WD. In total, the study population consisted of 10 745 infants born to 9733 unique mothers, with the contribution by 1012 women of 2 pregnancies each to the study.
Flowchart depicting selection of mothers and infants included in study.
Selected characteristics of the 10 745 mother-infant pairs are displayed in Table 1. Approximately 3.2% of infants in this study were born to mothers aged 12 to 15 years, 17.1% to mothers aged 16 to 17 years, 31.4% to mothers aged 18 to 19 years, and 48.3% to mothers aged 20 to 24 years. The distribution of characteristics such as ethnic group, caste, literacy, and various socioeconomic attributes differed by age category, with the youngest women more likely to be from the Madeshi ethnic group (people originating from the plains region of Nepal), illiterate, and without a latrine or electricity at their home.
Rates of adverse infant outcomes including LBW, preterm birth, and SGA were also significantly higher in the younger age groups (Table 1). More than 51% of infants born to women aged 12 to 15 years were LBW, 24.0% were preterm, and 73.5% were classified as SGA compared with 27.8%, 16.4%, and 56.3% of infants, respectively, in the 20 to 24 years of age category.
There were 371 infant deaths during the neonatal period, corresponding to an overall NMR of 34.5 per 1000 live births. Neonatal mortality rates were greatest in the youngest women and decreased with increasing maternal age (Table 2). The NMR for infants born to mothers younger than 16 years was more than double that of infants born to mothers older than 20 years (61.8 vs 28.5 per 1000 live births). Trends were similar for both parity 0 and parity 1 mothers, but there was a paucity of parity 1 mothers in the youngest age categories.
Crude and adjusted ORs for neonatal mortality presented in Table 3 show declining risk of neonatal mortality with increasing maternal age. Infants of mothers 12 to 15 years of age were at a more than 2-fold greater risk of mortality than those of mothers aged 20 to 24 years (OR, 2.24; 95% CI, 1.40-3.59). After treatment received, maternal literacy, ethnic group, caste, latrine and cattle ownership, electricity in the home, maternal occupation, parity, and gestational night blindness were adjusted for, infants were found to have a 53% excess risk of neonatal mortality if born to mothers in this youngest age group vs the oldest (1.53; 0.90-2.60), although this association was no longer statistically significant. The adjusted ORs for neonatal mortality associated with mothers aged 16 to 17 years and 18 to 19 years compared with mothers aged 20 to 24 years were 1.17 (0.84-1.64) and 1.00 (0.76-1.32), respectively. Treatment actually received, whether that treatment was the assigned one or not, was a confounder of the maternal age and neonatal mortality relationship. There was no evidence of interaction between maternal age and parity or treatment received. The reversal of the association between caste and neonatal mortality in the multivariable model was due to strong association between caste and ethnicity. Restricting the analysis to early neonatal deaths that occurred during the first week of life yielded a similar but weaker maternal age effect (data not shown). The relationship between maternal age and neonatal mortality also held when the analysis was restricted to infants without birth weight data (those infants who had died before arrival of the WD or those infants who were not weighed within 72 hours after birth) (data not shown).
Ethnicity was one of the strongest predictors of neonatal mortality. Even after adjustment, Madeshi infants were almost twice as likely to die in the first 28 days of life as were Pahadi infants (OR, 2.01; 95% CI, 1.43-2.84). A maternal history of night blindness during pregnancy was also strongly linked to neonatal death (2.03; 1.34-3.07).
Low birth weight, preterm delivery, and SGA were strongly associated with both neonatal mortality (OR, 4.27; 95% CI, 3.06-5.96; 3.61; 2.90-4.49; and 1.77; 1.24-2.53, respectively) and young maternal age (Table 3 and Table 4). To investigate whether the increased risk of neonatal death observed among younger women after adjustment for socioeconomic and other confounders was linked to the higher rates of LBW, preterm birth, and SGA in younger age groups, 5 different models of neonatal mortality that further adjusted for these factors were considered (Table 4). The association between maternal age and neonatal mortality diminished on adjustment for LBW, preterm delivery, or SGA. The addition of both SGA and preterm birth to the model further dampened the effect of maternal age on neonatal mortality. After controlling for these 2 factors, the excess risk of neonatal mortality in infants born to women aged 12 to 15 years compared with women aged 20 to 24 years dropped from 53% to 14% (1.14; 0.50-2.61). Thus, the increased risk of neonatal mortality associated with adolescent pregnancy was primarily mediated through elevated rates of preterm births and SGA infants in these younger women.
This study found a strong association between young maternal age and neonatal mortality that was significantly attenuated after socioeconomic factors and other confounders were controlled for. The effect of maternal age on infant survival in the first week of life was no stronger than for the entire neonatal period, thus confirming the greater influence of socioeconomic rather than biologic factors on the survival of infants born to young mothers. Our results are comparable to a population-based cohort study in Missouri that found a 1.69 times higher risk of neonatal mortality in mothers aged 12 to 17 years vs mothers aged 20 to 34 years.9 In that study, adjustment for socioeconomic status, race, educational level, parity, smoking, and prenatal status accounted for most of the increased risk of neonatal death. Similarly, a hospital-based study found no association between young maternal age and neonatal mortality once maternal race, prenatal care, and other factors had been controlled for.23 Despite similarities with these results, an alternate explanation for our findings could be that the biologic risk associated with young maternal age was expressed through increased rates of miscarriages and stillbirths. In that event, fetuses surviving to be live-born would be at a lower biologic risk than environmental risk once born.
A number of investigations in industrialized and developing countries provide evidence supporting a role for biologic factors in poorer pregnancy outcomes of young mothers. A cross-sectional study in Latin America reported a 50% excess risk of early neonatal mortality among mothers 16 years of age or younger vs mothers 20 to 24 years of age after adjustment for 15 different socioeconomic factors and other confounders.24 A retrospective cohort study of 3.8 million primiparous pregnant women younger than 25 years yielded increased risks of neonatal mortality, LBW, and preterm delivery in adolescents even after taking into account deleterious social and environmental factors.3 A Swedish study found that the increased risk of neonatal mortality among adolescents was explained largely by higher rates of preterm birth, which the authors attributed to biologic causes.5
Preterm birth is one of the leading direct causes of neonatal death.25 Low birth weight, which may indirectly account for 60% to 80% of neonatal deaths, arises through preterm birth, intrauterine growth restriction, or both.25,26 Consistent with other investigations, our study found that LBW, preterm birth, and SGA were strongly associated with both neonatal mortality and low maternal age.3- 8 The excess risk of neonatal mortality in the youngest mothers, although diminished after socioeconomic confounders were controlled for, was further attenuated with adjustment for LBW, preterm birth, SGA, or both preterm birth and SGA. This suggests that although disparities in socioeconomic factors in younger women compared with older women partially explain the increased risk of neonatal mortality among adolescent women, the biologic mechanism of this excess mortality is mediated primarily through preterm birth, SGA, LBW, or some interaction of these variables.
This study also demonstrated a striking relationship between ethnicity and neonatal mortality. Madeshi infants were at much higher survival risk in the neonatal period than Pahadi infants even after confounders were controlled for. This excess risk may be due to differing behavioral practices related to maternal and infant nutrition as well as antenatal and newborn care. For example, breastfeeding was more likely to be initiated early in Pahadi than in Madeshi infants, and the early initiation was shown to be associated with survival.27 Given this association, there are likely other behavioral factors that explain the ethnic difference that were not collected.
Our findings help to illuminate the complex links between maternal age and adverse reproductive outcomes in a nutritionally deficient, low-resource setting where cultural norms favor adolescent pregnancy. In Nepal, early marriage and childbearing are considered socially acceptable and are, in fact, viewed as successful outcomes.28,29 There is significant cultural and family pressure to give birth early and, in particular, to produce a male heir.28,29 That the median age at first birth in Nepal has remained consistent at 19.9 years of age over the 2001 and 2006 Nepal Demographic and Health Surveys, and is similar across all age cohorts, provides evidence that social and cultural attitudes regarding early childbearing continue to be prevalent.30,31 Conversely, in many industrialized nations, adolescent pregnancy often occurs outside the sphere of marriage and is linked to social marginalization, low socioeconomic status, and inadequate prenatal care.12 These factors often are cited as the primary causes of poorer birth outcomes among younger women. In our study population the youngest women still appeared to be the most socioeconomically disadvantaged regardless of social and cultural acceptance of adolescent pregnancy. While poor socioeconomic status may be a more critical influence on adverse reproductive outcomes than social marginalization, preterm birth and SGA are overriding factors in higher mortality of neonates born to adolescent women.
Limitations of this analysis include potential recall bias associated with estimating gestational age based on the self-reported date of the last menstrual period. Although we adjusted for a large number of socioeconomic variables, several well-established maternal risk factors such as body mass index and mid-upper arm circumference, smoking and alcohol use, and weight gain during pregnancy were unavailable. Smoking and alcohol use, however, were low among younger women in a similar population.32
The mothers' age at menarche was also unavailable; hence, it was not possible to distinguish between risks associated with pubertal vs postmenarcheal growth. Low gynecologic age has been associated with increased rates of adverse reproductive outcomes in younger mothers and may be more closely associated with biologic outcomes than chronologic age.33 Hediger et al7 reported that mothers who were young but whose gynecologic age exceeded 2 years were at no greater risk for preterm delivery than older mothers. In our study population, where chronic malnutrition is prevalent, puberty and age at menarche may have been delayed in a substantial proportion of the younger mothers. However, we were unable to ascertain which of the mothers had low gynecologic age and whether those mothers in particular were at a higher risk for neonatal mortality.
All mothers in this study received micronutrient supplementation and educational visits by WDs, which may have affected mortality during the neonatal period. However, the overall NMR in this study was 34.5 deaths per 1000 live births, which is comparable to the most recent national NMR estimate of 33 deaths per 1000 live births reported in the 2006 Nepal Demographic and Health Survey.30 We do not believe the provision of vitamin A and iron–folic acid affected the relationship between maternal age and neonatal mortality since mothers in all age categories benefited from these interventions. Since albendazole was given to everyone and not randomized, we cannot know whether this affected all age groups similarly. While mothers in this study may have been better off with respect to micronutrient status than those not participating in the study, calorie and protein deficiency remained significant problems among our study population and probably were primary contributors to LBW. We therefore believe our findings are generalizable to populations with similar nutritional deficiencies.
In summary, compared with older women, infants born to adolescents were at an increased risk of neonatal mortality that was ascribed largely to social and environmental factors and was mediated through increased rates of SGA, LBW, and preterm delivery among the younger women. Further research is needed to elucidate the complex relationship between adolescent pregnancy and adverse reproductive outcomes, particularly in resource-poor settings where delayed age at menarche due to chronic malnourishment may be an important influence. Delaying the age at first pregnancy may be a valuable strategy to promote and improve infant health and survival.
Correspondence: Joanne Katz, ScD, Department of International Health, Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe St, Room W5009, Baltimore, MD 21205-2103 (firstname.lastname@example.org).
Accepted for Publication: January 24, 2008.
Author Contributions: Drs Katz, Mullany, Darmstadt, and Tielsch had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Sharma, Katz, Mullany, Darmstadt, and Tielsch. Acquisition of data: Mullany, Khatry, LeClerq, Shrestha, and Tielsch. Analysis and interpretation of data: Sharma, Katz, Darmstadt, and Tielsch. Drafting of the manuscript: Sharma and Katz. Critical revision of the manuscript for important intellectual content: Sharma, Katz, Mullany, Khatry, LeClerq, Shrestha, Darmstadt, and Tielsch. Statistical analysis: Sharma, Katz, and Mullany. Obtained funding: Darmstadt and Tielsch. Administrative, technical, or material support: Khatry, LeClerq, Shrestha, Darmstadt, and Tielsch. Study supervision: Katz, Mullany, Khatry, LeClerq, Shrestha, and Tielsch.
Financial Disclosure: None reported.
Funding/Support: This study was conducted by the Department of International Health, Bloomberg School of Public Health, The Johns Hopkins University, under grants from the National Institutes of Health (HD 44004, HD 38753, R03 HD 49406), the Bill and Melinda Gates Foundation (810-2054), and Cooperative Agreements between The Johns Hopkins University and the Office of Health and Nutrition, US Agency for International Development (HRN-A-00-97-00015-00, GHS-A-00-03-000019-00). Commodity support was provided by Procter and Gamble Co.
Sharma V, Katz J, Mullany LC, Khatry SK, LeClerq SC, Shrestha SR, Darmstadt GL, Tielsch JM. Young Maternal Age and the Risk of Neonatal Mortality in Rural Nepal. Arch Pediatr Adolesc Med. 2008;162(9):828-835. doi:10.1001/archpedi.162.9.828