Differential effects of socio-demographic factors on maternal haemoglobin concentration in three sub-Saharan African Countries

Objective To investigate the effects of socio-demographic factors on maternal haemoglobin (Hb) at different points of the conditional distribution of Hb concentration. Methods We analysed the Demographic and Health Surveys data from Ghana, Democratic Republic of the Congo (DRC) and Mozambique, using Hb concentration of mothers aged 15-49 years as an outcome of interest. We utilise quantile regression to estimate the effects of the socio-demographic factors across specific points of the maternal Hb concentration. Results The results showed crucial differences in the effects of socio-demographic factors along the conditional distribution of Hb concentration. In Ghana, maternal education had a positive effect on Hb concentration in the 5 th and 10 th quantiles. The positive effect of education on maternal Hb concentration occurred across all quantiles in Mozambique, with the largest effect at the lowest quantile (5 th ) and the smallest effect at the highest quantile (90 th ). In contrast, maternal education had a negative effect on the Hb concentration of mothers in the 50 th , 75 th and 90 th quantiles in DRC. Maternal body mass index (BMI) had a positive effect on Hb concentration of mothers in the 5 th , 10 th , 50 th and 90 th , and 5 th to 50 th quantiles in Ghana and Mozambique, respectively. Breastfeeding had a significant positive effect on Hb concentration across all countries, with the largest effect on Hb concentration of mothers in the lower quantiles. All the household wealth indices had positive effects on maternal Hb concentration across quantiles in Mozambique, with the largest effect among mothers in the upper quantiles. However, in Ghana, living in a poor wealth index was inversely related with Hb concentration of mothers in the 5 th and 10 th quantiles. Conclusions Our results showed that the effects of socio-demographic factors on maternal Hb concentration vary along its distribution. Interventions to address maternal anaemia should take these variations into account to identify the most vulnerable groups. What this study adds Quantile regression can be used effectively to analyse anaemia data Socio-demographic factors have differential effects on Hb at different points of its distribution Interpreting results based on the mean effect (as in OLS) only provides a partial picture Breastfeeding has positive effect on maternal Hb concentration The use of multicountry data revealed differences and commonalities between countries