Social, biological, and programmatic factors linking adolescent pregnancy and early childhood undernutrition: a path analysis of India's 2016 National Family and Health Survey

Abstract

Background: Adolescent pregnancy and child undernutrition are major social and public health concerns. We aimed to examine associations between adolescent pregnancy and child undernutrition in India, where one in five adolescents live, and one in three of the world's stunted children.

Methods: Data were from India's fourth National Family Health Survey, 2015-16. Primiparous women aged 15-49 years who gave birth between 2010 and 2016 were classified on the basis of age at first birth: 10-19 years (adolescence), 20-24 years (young adulthood), and 25 years or older (adulthood). Primary outcomes were anthropometric measures of offspring undernutrition. Multivariable regression and structural equation models were used to understand the extent to which these measures were linked to adolescent pregnancy and the potential social, biological, and programmatic pathways.

Findings: Of the 60 096 women in the sample, 14 107 (25%) first gave birth during adolescence. Children born to adolescent mothers had lower Z scores for length or height-for-age (mean difference -0·53 SD), weight-for-age (-0·40 SD), and weight-for-length or height (-0·16 SD) than children born to adult mothers. Compared with adult mothers, adolescent mothers were shorter (-1·21 cm, 95% CI -1·78 to -0·65), more likely to be underweight (18 percentage points, 15-21) and anaemic (8 percentage points, 6-11), less likely to access health services (-4 to -15 percentage points), and had poorer complementary feeding practices (-3 to -9 percentage points). Adolescent mothers also had less education (-3·30 years, 95% CI -3·68 to -2·91), less bargaining power (-7 to -15 percentage points), and lived in poorer households (-0·66 SD, 96% CI -0·82 to -0·50) with poorer sanitation (-28 percentage points, -32 to -24). In the path analysis, these intermediate factors predicted child anthropometry, with the strongest links being mother's education (18%), socioeconomic status (13%), and weight (15%).

Interpretation: Children born to adolescent mothers are at risk of being undernourished. Adolescent pregnancy is related to child undernutrition through poor maternal nutritional status, lower education, less health service access, poor complementary feeding practices, and poor living conditions. Policies and programmes to delay pregnancy and promote women's rights could help break the intergenerational cycle of undernutrition through many routes.

Funding: Bill & Melinda Gates Foundation through Partnerships and Opportunities to Strengthen and Harmonize Actions for Nutrition in India, led by the International Food Policy Research Institute.

Figure 1

Conceptual framework for linking adolescent pregnancy and early childhood undernutrition

Figure 2

HAZ score and prevalence of stunting for first-born children by child's age and mother's age at first birth in India, 2016 (A) HAZ score. Curves are smooth local polynomials with 95% CIs. (B) Stunting. Adjusted coefficient (95% CI) for models are shown below each panel in the figure. OLS regression models were adjusted for child age, sex, maternal religion, and caste fixed effects and controlled for the cluster sampling design and sampling weights used in the survey. Data are from India's fourth National Family Health Survey (NFHS-4, 2016). HAZ=Length or height-for-age Z score. OLS=ordinary least squares.

Figure 3

Pathways from adolescent pregnancy to child undernutrition through maternal nutrition, health services, infant and young child feeding practices, living conditions, women's education, and bargaining power Bold font and shading for the mediating variables indicates that the variable was important in mediating the association between adolescent pregnancy and child height-for-age Z score (p<0·1 for both steps). Numbers in the paths are regression coefficients. Figure presents paths from the three path models that were run. Model 1 included variables for maternal nutritional status, access to health and nutrition services, living conditions, and women's education. Model 2 included all variables in model 1 plus the child feeding variables shown, in the subsample of mothers with children 6–24 months of age (n=25 494). Model 3 included all variables in model 1 plus the bargaining power variables shown, in the subsample with data available (n=10 340). All models adjusted for maternal caste and religion and child age and sex and controlled for the cluster sampling design and sampling weights used in the survey. Coefficients shown in the figure are from Model 1, except for those for child feeding (from Model 2) and bargaining power (from Model 3). HAZ=Length or height-for-age Z score. *Constructed from principal components analysis of asset ownership.