Lack of dietary diversity is particularly a considerable problem among poor populations of developing world as their diets are predominantly based on starchy staples [1
]. In the present study, cereal consumption was 100% and almost all other food group consumption was low. Especially, their animal food consumption was poor. Furthermore, carbohydrate was the main energy contributor (68%) to the diet than other macronutrients. This condition may be due to their daily consumption of rice based diet with few other food commodities. Present study mainly focused on validation of dietary diversity indicators as an indicator of the nutrient adequacy of rural elderly people in Sri Lanka. As dietary diversity measures, simple count of individual food items (FVS), simple count of food groups (DDS) and dietary serving score (DSS) were used. In DDS, DDS and DDS-half serving scores were used separately. The average (standard deviation) of the food variety score (FVS), dietary diversity score (DDS), DDS-half and dietary serving score (DSS) was 8.4 (2), 4.4 (0.9), 3.8 (1.0) and 11.4 (2.5), respectively. Any of those scores were not up to their theoretical maximum. It reflects that their dietary diversity was not up to the optimum level. FVS counts all the food items consumed over the previous 24-h of survey date. When computing FVS, both beverages and condiments were not considered. Among the beverages we only excluded tea and coffee without milk i.e. milk and coffee infusion as the frequency of consumption of tea and coffee is high among the Sri Lankans and therefore to avoid overestimation of diversity of the diet. FVS could vary from the maximum value of 15 to minimum value of zero. Indeed, when computing FVS, we count the individual food items eaten without considering the group of the food. Therefore considering only FVS alone can therefore give a falsely favorable impression of the quality of diet. When constructing DDS, both nutritional aspects and local food group culture were considered. In the present study, six food groups for the DDS were adapted, since consumption of animal protein sources was low, animal (meat/fish/dairy) and plant protein sources (legumes/lentils) were taken as two different food groups. One of the limitations of this study is we excluded the counting of food group such as oils and fats and sugar and sweets. In DDS-half serving, other than just counting of food group, consumption of at least a half serving of that specific food was considered. When comparing to DDS, DDS-half serving improved the association with MAR, indicating that the performance of dietary diversity as an indicator of adequate nutrient intake is improved when a minimum intake for each food group was considered. This finding has important implications for field use of the indicator, it highlighted that the importance of considering the quantity consumed. In most field survey, we used to record simply the number of food groups consumed than information on quantities of food consumed. Moreover, it is important to identify which food group of their daily consumption may be the major contributor to the nutrition adequacy. Table shows the mean dietary serving scores of each food group. The lower mean of the diversity score was related to fruit group and the higher one was for the cereals/roots group. For cereals/roots group, DDS was 4 and that was same as the theoretical score of 4 and other all food group scores were below the recommended serving score. It indicates that they could not achieve daily recommended serving sizes of these food groups considered in the present study to fulfill the nutrient adequacy. Sri Lankan Food Based Dietary Guidelines were used to adapt serving scoring system in this study [13
]. According to the serving scoring system, the mean DSS was about 11 out of 20. It showed the poor diversity of the diet among this elderly population.
The mean dietary serving score (DSS) within the food groups in the study population
Validation studies were performed according to the several criteria because validated dietary diversity indicators can be used as a precious and reliable nutrition tool in relation to health promotion and evaluation. As gold standards of nutrient adequacy, both NAR and MAR were used. For NAR values of nutrients, mean values were taken because intake of most of the nutrients was normally distributed. In the present study, mean MAR was about 0.4. The ideal cut-off point should be 1. It is convinced that the all nutrient requirements have been covered. But according to the results, only about 40% of the requirement of total nutrients has been covered. That may be the result of the diets that they were taken, because those diets were poor both in quality and quantity. In the present study, all four dietary diversity measures were significantly correlated with MAR, illustrating the potential of simple scores of dietary diversity for use as indicators of nutrient adequacy of the diet. These findings are similar to those of previous studies, testing the utility of dietary diversity as an indicator of nutrient adequacy in the diet of children, adolescents and adult women [1
In furthermore validation, dietary diversity indicators were validated against nutrient NARs. Vitamin B12, vitamin D and folate were not significantly correlated with FVS, DDS and DDS-half serving. Vitamin B12 is found only in animal source foods, particularly liver, dairy products and eggs. The best sources of vitamin D are dairy products, legumes and green leafy vegetables and folate can be found in both animal source foods and plant based foods. Meat, egg, green leaves and dairy were the least consumed food groups in the study sample and lack of these groups could explain the poor correlation with DDS. Although they consumed legumes and fish commonly, the portion size consumed tended to be small.
An additional methodological contribution to validation of study was the sensitivity-specificity analysis and that was carried out to identify best cut-off points for predicting nutrient adequacy for both diversity indicators. Similar to the study done by Kennedy et al. 2007 [15
], results of this study found the best cut-off points to maximize sensitivity and specificity was FVS of 9 and DDS of 4.5. Determining a fixed cut-off point, where elders can be defined as having greater or less risk of inadequate nutrient intake has potential application in immediate population nutrition assessment.