Cognitive Testing of Dietary Assessment and Receipt of Nutrition Services for Use in Population-Based Surveys: Results from a Demographic and Health Surveys Pilot in Uganda.

BACKGROUND

Assessing dietary intake and receipt of nutrition services is essential for monitoring health and development, yet accurately measuring these indicators via population-based surveys remains challenging.

OBJECTIVES

We investigated how respondents understood survey instructions and questions used to measure dietary intake and receipt of nutrition services. We also explored the possible influence of social desirability on responses.

METHODS

In a Demographic and Health Surveys pilot conducted in Uganda, women were randomly assigned to hear either a short or long introduction to dietary intake questions regarding themselves ( = 1116) and/or their children ( = 289). Dietary indicators were compared between groups using a Pearson chi-square test. Cognitive interviews about questions on both dietary intake and receipt of nutrition services were conducted with a subset of respondents with children under age 2 ( = 19). Transcripts were analyzed using deductive and inductive approaches.

RESULTS

Minimum dietary diversity did not differ significantly between respondents who received the short compared with long introduction to dietary questions (42.5% compared with 37.0%, respectively, for women; 39.7% compared with 46.6%, respectively, for children). The most common issues with dietary questions were inclusion or omission of items consumed in small quantities and from mixed dishes, terminology for milk feeding and yogurt consumption, and misclassification of foods by food group. Questions on receipt of nutrition services were generally well understood, although there was some confusion with terminology related to rooming-in after birth and breastfeeding observation by a health worker. Social desirability bias likely influenced responses to questions on exclusive breastfeeding in the first 2 d after birth and soda consumption and may have influenced responses to questions on meat consumption.

CONCLUSIONS

Small modifications to survey questions may improve questions on dietary intake and receipt of nutrition services. Other identified issues may be best addressed through interviewer training or data interpretation. Further survey implementation research may be needed.