The impact of young maternal age at birth on neonatal mortality: Evidence from 45 low and middle income countries

Objectives This study explores the impact of early motherhood on neonatal mortality, and how this differs between countries and regions. It assesses whether the risk of neonatal mortality is greater for younger adolescent mothers compared with mothers in later adolescence, and explores if differences reflect confounding socio-economic and health care utilisation factors. It also examines how the risks differ for first or subsequent pregnancies. Methods The analysis uses 64 Demographic and Health Surveys collected between 2005 and 2015 from 45 countries to explore the relationship between adolescent motherhood (disaggregated as <16 years, 16/17 years and 18/19 years) and neonatal mortality. Both unadjusted bivariate association and logistic regression are used. Regional level multivariate models that adjust for a range of socio-economic, demographic and health service utilisation variables are estimated. Further stratified models are created to examine the excess risk for first and subsequent births separately. Findings The risk of neonatal mortality in all regions was markedly greater for infants with mothers under 16 years old, although there was marked heterogeneity in patterns between regions. Adjusting for socio-economic, demographic and health service utilisation variables did not markedly change the odds ratios associated with age. The increased risks associated with adolescent motherhood are lowest for first births. Conclusion Our findings particularly highlight the importance of reducing adolescent births among the youngest age group as a strategy for addressing the problem of neonatal mortality, as well ensuring pregnant adolescents have access to quality maternal health services to protect the health of both themselves and their infants. The regional differences in increased risk are a novel finding which requires more exploration. The study is based on analysis of 64 Demographic and Household Surveys (DHS) from 45 countries carried out between 2005 and 2015 in low and middle income countries (see S1 Table for list of surveys). The DHS are large, nationally representative surveys that collect data on a range of demographic and health variables that are comparable across time and countries [24]. Thirty one countries were in sub-Saharan Africa, eight in South and South-East Asia and the remaining six were in Latin America and the Caribbean. In a number of countries more than one survey was available since 2005, and when this occurred, surveys were pooled to increase sample size. A cut-off of 2005 was included to ensure the datasets were reasonably contemporaneous while still providing a large sample of countries reflecting a wide geographic spread. Women aged 15 to 49 in sampled households provide a full birth history, with further questions about births that occurred within a defined period before the survey (usually of five years). This included the survival status of the child, with the age at death for those that died. The outcome of analysis is neonatal mortality, defined as death within the first 28 days of life. Initial bivariate binary logistic regression was used to estimate unadjusted odds ratios (ORs) for neonatal mortality by age of the mother at the time of the birth. Age at birth was grouped as <16 years, 16/17 years and 18/19 years with a reference category of 20–29 years. Individual models were run for each country, as well as for data pooled at the regional and aggregate regional level in order to increase the sample size. The dependant variable for analysis of individual countries and small regions was neonatal mortality, but separate models for early and late mortality were developed for aggregated regions. Early neonatal mortality refers to a death of a live-born baby within the first seven days of life, while late neonatal mortality covers the time after 7 days until before 28 days. Further models were run for neonatal mortality with maternal age at birth in individual years (e.g. <15, 15, 16, 17, 18 and 19 years) for the aggregate regional level to examine patterns in more detail. In addition, ORs were also calculated for models based on the aggregate regional level adjusted for a range of socio-economic, demographic and health service utilisation factors that are known to be associated with both adolescent motherhood and the risk of neonatal death. These include urban / rural residence, maternal education (grouped into no education, primary and secondary/tertiary), birth order (grouped into first birth, 2/3rd birth and 4th or higher birth), antenatal care utilisation (ANC; no ANC, 1–3 visits or 4 or more visits) and place of birth (home or facility). DHS surveys measure wealth through asset indices, which are a composite measure of a household's cumulative living standard. The index is calculated using data on a household’s ownership of selected assets, housing construction and access to water and sanitation facilities. However, these indices were not included in the analysis as they are not comparable across countries, although analysis using asset indices was included as part of the sensitivity analysis (not presented). Further separate stratified models were created for first births and subsequent births. In order to account for the differential chances of households being selected into the surveys, sample weights were applied in all analyses. This was conducted within each country and survey. No weighting was conducted to reflect the population sizes of each country in the regional level estimates. Sensitivity testing was carried out by re-running the models with no weighting, and also using only one survey per country (as well as with asset index included) to check this was not introducing bias.