we used the relative size of the baby at birth as reported by
mothers. Several evaluation studies have shown that the
reported size of babies at birth is a reliable proxy for birth
weight [25–27]. Nevertheless, when considering the findings
of this study we need to keep in mind the problems
associated with using the reported size as a proxy for birth
weight and the fact that information on mothers’ BMI and
anaemia status was collected at survey date.
We made a distinction between two sets of covariates:
those that directly affect birth weight (proximate determinants)
and those that affect low birth weight only through
the proximate determinants (underlying socio-economic
factors). And we focused on the relationship between 7
proximate variables and the likelihood of low birth weight.
Of special consideration was the link between mothers’
nutritional status and low birth weight.