Objective: To evaluate the relative validity of recalled intake among schoolchildren and mothers in rural Kenya.
Design: Cross-sectional study. Mothers' recall of both the schoolchild's intake and her own intake on the previous day were compared with intakes that were weighed by an interviewer in the home.
Setting: Karurumo location of Embu District in Kenya.
Subjects: A total of forty-two sets of grade 1 students and their mothers.
Results: Between 08.00 and 17.00 hours, when foods were both weighed and recalled, approximately 70 % of weighed food items were recalled the next day. Under-reporting of food amounts was seen across most food categories for the recall, with added sugars, sweets and fats being most affected. The recall underestimated energy intakes by approximately 6-9 % during this period. Correlation coefficients between nutrient intakes ranged from 0.43 to 0.65, while weighted kappa values ranged from 0.22 to 0.50. Higher levels of agreement were noted for nutrient densities, with correlation coefficients between 0.46 and 0.82 and weighted kappa values between 0.30 and 0.73.
Conclusions: Although the recall method provides an acceptable alternative to the more labour-intensive and expensive food-weighing method, there is need to further improve its performance in this population through more accurate recall of single foods, especially fruits, as well as added sugars, fats, dairy products and meats, which are often added in small amounts to mixed dishes.