Influence of socio-demographic factors on the breastfeeding period of women in Bangladesh: a polytomous logistic regression model

Background
In Bangladesh, terrible degradation in the breastfeeding period has occurred with rapid urbanization in recent years that is causing a shortage of child nourishment. Identifying the risk factors of breastfeeding duration is important for planning nutritional programs and strategies.

Objectives
This study tries to identify influential demographic and socio-economic factors that affect the breastfeeding period for reducing child nutrition deficiency.

Material and methods
The study attempts to proceed with data collected from an observational study entitled the Bangladesh Demographic and Health Survey (BDHS) 2014. The breastfeeding period (Ordinal exogenous variable) is classified into three groups: 0–5-months, 6–23 months and at least 24 months. Gamma, chi-square and linear-by-linear statistics are used to identify the associated factors that have an impact on the breastfeeding period. A test of parallelism is conducted to evaluate the proportional odds. The polytomous logistic regression (PLR) model and the proportional odds (PO) model are used to find the marginal effect of demographic and socio-economic predictors that affect the breastfeeding period.

Results
Parental educational attainment, wealth index, division, religion, mother’s BMI, drinking water source, household members, amenorrhea and abstaining, respectively, are the most significant factors that influence the breastfeeding period. The PLR model is also more precise than the PO model for indicating the marginal effect among those vital factors for the breastfeeding period.

Conclusions
PLR is an appropriate model to recognize the effect of predictors of breastfeeding duration instead of the PO model and other measures.