Abstract
Under nutrition is one of the severe problems around the globe and finds its place in the
global agenda. Sustainable Development Goals (SDGs) highlight the need for special
attention to eradicate malnutrition. Bangladesh having high prevalence of malnutrition is
committed to fulfill the targets of SDGs. Though Bangladesh achieved remarkable success in
improving nutritional status of under-five children at national level, there have been
regional variations. Government is planning to target need based resource allocation to
small administrative levels. To do that real time small area level estimates of nutrition will
be required. Nutritional status often linked with food security condition of the households
hence, these two should be addressed together. For effective policy planning alongside
accurate estimates, underlying factors need to be identified to make the interventions more
focused and sustainable for the long run. This research tried to address these issues using
both secondary (e.g., Census 2011 and BDHS 2011) and primary data. Statistical Package for
Social Science (SPSS), STATA and R software have been used to analyse the data. The study
found that about 42% children under five in rural Bangladesh were stunted, 39% were
underweight and 16% were wasted. Two-level binary logistic regression analyses identified
division, child age in month, twin childs, preceding birth interval, religion, mother's
education, father's education, wealth index, household size, age of household head,
mother's age at first child birth, sources of drinking water, child birth size, sex of child and
child suffered from fever and diarrhea as significant determinants of malnutrition among
the under five children in rural Bangladesh. An attempt was made to investigate the linkage
between malnutrition and food security alongside identifying the determinants of
nutritional status. A primary sample of size 300 was collected from Dharampasha upazila of
Sunamgonj district of Sylhet division for in-depth analysis. Though in all the three models
(stunting, wasting and underweight) we didn't find any direct association between foop
security and child nutritional status, there was an indirect and significant effect of food
security mediated by father's occupation. M-Quantile estimation method was used to
identify small area estimates at upazila level of Sylhet division. The upazilas exhibiting
poorest nutritional status were identified in maps for comparison. Special care should be
given to help these upazilas to come out of the cycle of malnutrition in addition to the
common national programmes. Preceding birth interval, education and source of drinking
water are the major determinants of stunting, wasting and underweight which may be taken
into consideration for policy implications in the current nutrition programmes. Community
effects were found significant in all the three models. Specific programme addressing special
needs of the communities should be ensured to overcome the problem. As there is a link
between food security and malnutrition in different subsections of the population,
interventions addressing food security and nutrition should have synergies. M-quantile
method has been used in Bangladesh for the first time for nutrition estimates. The results are
efficient and may be adopted in future, especially where we have doubt in the distributional
assumption of the data