Impact of individual, household and community characteristics on children's nutritional indicators.
Bottom Line: The results show that variation in nutritional status of under-five children in Botswana is a function of characteristics of the households and communities within which they live.As much as 17% of variation is due to differences in the communities and households.Economic status of households holds an important key in predicting the nutritional status of children.
This study analyzed WHO-standardized nutritional indicators of children from selected households within communities that were sampled from all districts of Botswana. Data from the 2007 Botswana Family Health Survey were fitted into multilevel models that seek to account for variability due to the macro- and micro-units that have been hierarchically selected. This allowed for estimation of different levels of intraclass correlations while simultaneously assessing the model-fit by accounting for the influence on the nutritional indicators due to the fixed variables attributable to these macro- and micro-units. The results show that variation in nutritional status of under-five children in Botswana is a function of characteristics of the households and communities within which they live. As much as 17% of variation is due to differences in the communities and households. Economic status of households holds an important key in predicting the nutritional status of children.
Mentions: The intra-class correlation at Level 3, expressing the likeness of height-for-age z-scores for children in the same community, is 2% while that for children in different households within the same community is and is significantly different from zero. This shows that, while ignoring other factors, 17% of variation in height-for-age is attributed to differences in households within a community, (Figure 2). Considering the Level-2 model, we see that the intra-class correlation, which expresses likeness in height-for-age for children in different households within a community, is estimated at 0.99. This suggests that households within a community contribute more variability in the height-for-age z-scores than when community groupings are considered alone (Table 2).