A Comparative Analysis of Food Consumption Data From 24-h Dietary Recall and Household Consumption and Expenditure Surveys in Tanzania
Fanny Sandalinas, Rie Goto, Lilia Bliznashka, Fusta Azupogo, Mohammed Osman, Joyce Kinabo, Deanna K. Olney, Sonja Y. Hess, Evangelista Malindisa, Kidola Jeremiah, Edward J.M. JoyObjective
Household Consumption and Expenditure Surveys (HCES) are increasingly used to assess diets in low- and middle-income countries, but their validity compared to individual-level dietary data remains uncertain. We assessed the strengths and limitations of HCES data for informing strategies to improve diets and nutrition in Tanzania.
Design
Exploratory analysis of food group consumption from HCES (individualised using the adult female equivalent approach) and 24-h dietary recall (24hR). We examined concordance and trends by socioeconomic characteristics between methods for 10 food groups and fortifiable food vehicles.
Setting
Rural Arusha and Kilimanjaro regions, and national data from the Tanzania National Panel Survey Wave 5.
Participants
The analysis included 2599 adult women who completed a 24hR and lived in 2604 households contributing to HCES data in Arusha and Kilimanjaro. Nationally, 4469 households were included, with a regional subsample of 370 households from Arusha and Kilimanjaro.
Results
Dietary patterns were similar using HCES and 24hR data, including low consumption of nutrient-dense foods, while HCES were effective at capturing usual intake of food items eaten episodically. However, compared to 24hR data, energy intakes were substantially lower using HCES data, particularly in large households (42% difference), while there was poor concordance between methods for fruit and meat consumption and for wealth-related trends in cereal and vegetable intake.
Conclusion
HCES data can provide valuable insights for nutrition policy and planning; however, careful communication and interpretation of evidence is required, given limitations such as assumptions on within-household allocation of foods. Methods development could reduce measurement error.