Anna Herforth, Harvard T.H. Chan School of Public Health
@AnnaWHerforth @HarvardChanSPH Development of a low-burden diet quality questionnaire (DQ-Q) for measuring dietary diversity and other indicators of diet quality across countries Presentation | Slides
Kate Schneider, Tufts University
@GlFoodPolNut @TuftsNutrition Household diet quality and the cost of nutritious diets in Malawi~ Presentation | Slides
Maria Garza, Royal Veterinary College
@RoyalVetCollege Framework for food safety assessment of livestock derived food retailers, in selected low and middle-income areas of Nairobi Presentation | Slides
Sabri Bromage, Harvard T.H. Chan School of Public Health
@HarvardChanSPH Development of the Global Diet Quality Score Presentation | Slides
Development of a low-burden diet quality questionnaire (DQ-Q) for measuring dietary diversity and other indicators of diet quality across countries
1Harvard T.H. Chan School of Public Health, USA
2University of São Paulo, Brazil
3Teachers College, Columbia University, USA
There is a need for low-burden survey tools to collect information on diet quality. Low-burden methods are fast to administer, do not require subject expertise or probing by enumerators, and require low cognitive burden on respondents. The aim of low-burden survey tools for diet quality is to facilitate and democratize the collection of valid dietary data, for example in large-scale multi-topic surveys, and in nutrition-sensitive projects that are not managed by nutrition experts (such as agriculture project planning, monitoring and evaluation). Few low-burden diet survey methods exist, and they need to be rigorously tested to collect consistent and valid information.
A common method for collecting dietary diversity data consists of open-ended food group questions, e.g. "Yesterday, did you eat any porridge, bread, rice, pasta or other foods made from grains?" (FAO and FHI360, 2016). We used this as a starting point for developing a 28-item diet quality questionnaire (DQ-Q), where each yes/no question asks about consumption of a distinct food group in the previous day or night. We conducted 82 cognitive interviews in five languages São Paulo and New York City, in which we tested different ways of introducing and asking the questions, compared responses to closed-ended questions to open-ended questions, and interviewed respondents about their thought processes while answering the questions. Simultaneously, we analyzed nationally representative dietary intake data in two countries (Brazil and the United States) to determine the extent to which a short list of foods (<7 items) for each food group could capture the majority of people consuming each food group. A national pilot test of 1,000 respondents was implemented in Brazil within the Gallup World Poll, a multi-topic survey carried out by non-nutritionist enumerators.
We found that (1) respondents in both countries sometimes miscategorized foods when asked open-ended questions, and open-ended questions presented an additional cognitive burden; (2) respondents varied in their ability to think of other foods that belonged to a specific named food group; (3) cognitive burden on respondents was generally low, and was reduced by asking closed-ended questions, by questions with relatively short lists of food examples, and by specific design aspects of the questionnaire including a comprehensive introduction and intentionally-placed question stems; and (4) the majority of consumption of each food group could be represented by a few foods in each country setting (sentinel foods). On average, 1-7 sentinel foods captured 96-97% of people who consumed each food group (range 85-100%). These findings provide evidence supporting the approach of asking closed-ended questions using sentinel foods. The DQ-Q requires adaptation to each country context: the food groups are the same across countries, but need to be populated with sentinel food examples that capture consumption of the food group specific to each country. The finalized DQ-Q took 3-5 minutes to administer, had very low training requirements for non-specialist enumerators, and was easy for them to administer.
Collecting valid proxy indicators of diet quality with low-burden methods, such as dietary diversity scores or other indicators of diet quality, requires careful attention to the tool design for valid and comparable results. These results have important implications for large multi-topic surveys that aim to measure diet quality, and the DQ-Q can be used in whole or in part for the measurement of diet quality in the Gallup World Poll, the measurement of MDD-W in the DHS, the measurement of diet quality by other national surveys, and by agriculture projects that aim to collect diet quality information.
FAO and FHI 360. (2016). Minimum dietary diversity for women: A guide to measurement. FAO: Rome.