Objective: The primary objective of this study is to test the validity of 24-hour dietary
recall methods compared to the gold standard of measured portions and observed food
intake in the study population for improvement of this assessment in similar lowresource
settings.
Study Population: Located in Western Kenya in Nyanza Province, Migori County is the
setting for this research. Children were recruited from two out of four primary schools
taking part in a larger study. Children ranged in age from eight to sixteen, an age range
within which accuracy of diet recall has been found to increase as children age.
Methods: Thirty-eight children were observed eating a measured meal at their school.
Thus, for our study sample we know exactly what children consumed at this meal. The
next day, the selected children were invited back to complete an interviewer-assisted 24-
hour dietary recall using methods adapted for use in the study population. Single
sample t-tests were conducted in order to test the validity of the dietary recall methods
by comparing recalled food amounts with the gold standard of measured portions and
observed food intake. Finally, two-sample t-tests were conducted in order to compare
differences across age, sex, standard and school.
Results: The dietary recall methods used to obtain recalled food amounts do not appear
to be a valid method for obtaining detailed information about children’s food intake in
v
our study population. For the large majority of food volume and macro- and micronutrient
categories, the mean deviation for recalled values and those measured and
observed was statistically different from 0 (p<.05). However, results appear to have been
largely influenced by serving methodology, a potential confounder in this study,
suggesting that the dietary recall method may in fact be valid after accounting for
serving methodology.
Conclusions and Implications: Although the small sample size limits firm conclusions,
the results provide evidence-based direction for future food intake validation studies
and improvement of this assessment in similar low-resource settings. Future studies
should increase the sample size in order to reach more robust conclusions and carefully
consider approaches to improve the validity of this assessment in low-resource
populations