Evaluation of dietary diversity scores to assess nutrient adequacy among rural Kenyan women

Evaluation of dietary diversity scores to assess nutrient adequacy among rural Kenya women S. A. Ngala Abstract Background:The major cause of micronutrient deficiencies are low intake due to monotonous diets, especially among women of child bearing age. Dietary diversity score has been found to be a good proxy indicator for micronutrient adequacy. However, there are still outstanding methodological questions related to seasonal effects, food intake methods, selection of foods and the cut-off for estimating the prevalence of acceptable nutrient adequacy. This thesis evaluated the performance of a simple dietary diversity score for assessing nutrient adequacy in the diets of rural women in Kenya. Methods: The study was conducted in Mbooni Division, Makueni District, Kenya among non-pregnant, non-lactating women of reproductive age having a child between 2-5 years. Food consumption data was collected by 3 non-consecutive 24hour-recalls and a qualitative 24hour-recall in pre-harvest (period 1, October 2007, n=73) and post-harvest (period 2, April 2008, n=203) seasons. Dietary diversity scores (DDS) were derived based on 10 and 13 food groups with minimum intake threshold per food group of 0 and 15 g respectively. Mean probability of adequacy (MPA) was calculated based on intake of 11 micronutrients. Results: The dietary diversity score (DDS) and mean probability of adequacy (MPA) were significantly but moderately associated in both seasons (r=0.40 and r=0.38 period 1 and 2) and the association was independent of season (p=0.45). The DDS from a qualitative 24 hour recall (DDSql) showed little agreement with quantitative 24 hour recall (DDSqn) with a mean difference (DDSqn-DDSql) of -0.51±1.46 (Period 1) and -0.58±1.43 (period 2), with lower correlation between MPA and DDS for DDSql (r=0.14 and 0.19 in period 1 and 2, p>0.05) compared to DDSqn (r=0.40 and 0.54 in period 1 and 2, p<0.01). The Informative food-based scores and the food group-based scores were moderately associated with mean probability of adequacy (r=0.54-0.59 in period 1; r=0.37- 0.45 in period 2) with higher values for informative food based scores. The Minimum Dietary Diversity of Women (MDD-W) and mean probability of adequacy were significantly but moderately associated in both seasons (r=0.43-0.58 in period 1; r=0.24-0.50 in period 2) with but the use of a cutoff of consuming 5 or more food groups as indication of nutrient adequacy resulted in high total misclassification in both periods. Conclusion: A dietary diversity score can be used as a simple proxy for micronutrient adequacy, independent of season. The dietary diversity score derived from qualitative free-listing 24-hour recall formed a poor indicator, needing further refinement to improve its performance. The informative food-based score performs moderately better in predicting nutrient adequacy, but its advantages do not outway those of the food group-based scores, and the latter is therefore preferred. The Minimum Dietary Diversity score for Women, formed a good indicator to predict nutrient adequacy, but using the cutoff of 5 or more food groups resulted in an overestimation of prevalence of adequate intake in our resource poor population.