Evaluation of dietary diversity scores for assessment of micronutrient intake and food security in developing countries

Micronutrient malnutrition and food insecurity are widespread global public health problems. Micronutrient deficiencies affect one-third of the global population. Household food insecurity, often results in monotonous diets, a contributing factor to malnutrition. Consuming a varied diet comprised of diversity across food groups is a recommended approach to achieving nutritional requirements. Many organizations promote the collection of information on dietary diversity to inform food security and nutritional assessments, but there is not an agreed upon set of indicators used for this purpose. Simple, standardized diet-based indicators are needed to assess, monitor and evaluate individual micronutrient intake and household food security. This thesis has a two-fold aim i) to examine relationships between dietary diversity and adequate intake of micronutrients and ii) to consider outstanding methodological questions. These include, determining an appropriate cut-point for (in)adequate intake of micronutrients from the diet and the effect of length of recall period on characterizing dietary patterns. The association between probability of adequate micronutrient intake and individual dietary scores was determined by secondary analysis of data sets of non-breastfed Filipino children (n=2805, 2-5.9y), South African children (n=2200, 1-8y), and urban Malian women (n=102, 15-49y). Dietary diversity scores were positively and significantly correlated with intake of micronutrients, with correlation coefficients of 0.36, 0.63, and 0.33 in the Philippines, South Africa, and Mali respectively. Using a minimum quantity for a food group to count in the score improved the correlations to 0.44 in the Philippines and 0.48 in Mali. The best score cut-offs for a dichotomous indicator of inadequate intake of micronutrients were four and six food groups in South Africa and six in the Philippines. In Mali, five or more food groups was the best cut-off point for increased mean probability of micronutrient adequacy. In Mali dietary diversity scores using six and nine food groups had slightly higher correlations and indicator performance than scores based on thirteen or twenty-one food groups. Differences in household-level dietary patterns over one and seven day recall periods were tested through secondary analysis in Somalia (n=430 hh), Burkina Faso (n=3640 hh), Lao PDR (n=3913 hh) and Northern Uganda (n=1956 ). In Somalia, the median dietary diversity score was four for both a one day and seven day recall period. The main food groups consumed by fifty percent or more of households in the lowest dietary diversity tertile were cereals, sugar and oil for both recall periods. The dietary patterns based on food group consumption using score tertiles in Burkina Faso, Lao PDR and Northern Uganda were similar for one or seven days. This thesis shows that dietary diversity scores are acceptable indicators of micronutrient intake from the diet. Requiring a minimum quantity of consumption for a food group to count in the score is better, but jeopardizes simplicity. The best cut-off for a dichotomous indicator differed across studies, but fell within the range of four to six food groups for predicting poor intake in children and five or more food groups to predict higher probability of adequate intake in women of reproductive age. Cost implications of misclassification and social unacceptability of false results are low for this indicator. For the purposes of characterizing dietary patterns of households at population level a one day recall period is sufficient. The simplicity of data collection and analysis of indicators of dietary diversity should enable more widespread and repeated collection of information on dietary intake in resource constrained, food insecure environments. Areas for further research include, the added value of assigning weights to food groups and studies in populations with higher average micronutrient intakes to better define an appropriate cut-point for a dichotomous indicator. It is recommended that dietary diversity scores be incorporated as monitoring and evaluation tools in food security and nutrition assessments at national level and below.