Dietary Diversification

Dietary diversification interventions are interventions that change food consumption at the household level, such as increasing the consumption of animal-source foods (Gibson and Anderson 2009; Gibson, Perlas, and Hotz 2006). In most resource-poor settings, starch-based diets with limited access to meats, dairy, fruits, or vegetables, are the dominant diets. The objective in changing household diet is to increase the variety and quantity of micronutrient-rich foods, to decrease micronutrient deficiencies, including animal-source foods (Nair, Augustine, and Konapur 2016; Gibson 2014). This objective is generally achieved through social and behavior change activities, but can also include increased production of nutrient-rich foods and improved access to diverse foods.

Measurement and data sources

While measuring energy and nutrient intake would ideally rely on observed and weighed food records, such data collection is expensive, may be impractical, and may change the behavior of people being observed. Instead, most information reflects recall of diet intake (at the individual level) or recall of purchasing patterns (at the household level). Diet data collection methods vary and include—

• Repeat 24-hour diet recall, in which the food groups that a household or an individual has consumed are calculated for the preceding 24 hours. This method requires at least two 24-hour recalls from each person, on nonconsecutive days, to report individual-level data.
• List-based diet recall, in which the respondent recalls foods consumed in the past 24 hours, based on a list provided by the interviewer (Kennedy et al. 2011).
• Food frequency questionnaire, in which a respondent is asked questions about foods and beverages consumed over a specified period of time (longer than 24 hours).
• Emerging tools, such as Optifood (FANTA 2016) collects data on food consumption patterns.
• Ethnographic studies explore nutrition practices in various contexts and may include details on dietary intake (Tumilowicz, Neufeld, and Pelto 2015; Pelto et al. 2015).
• Questionnaires, such as those used in Household Consumption and Expenditure Surveys, provide information at the household level on consumption of food groups for approximately 125 predefined food items (Fiedler et al. 2012).

To help aggregate diet data collected through one of the methods above, many population-based surveys will report the types of food consumed, average numbers of predefined food groups included in the diet, and diet diversity scores. At the household level, diet diversity scores represent the economic ability of a household to access a variety of foods, while individual diet diversity scores aim to reflect nutrient adequacy. If inadequate food is consumed, it is unlikely that adequate quality can be met. These scores are most commonly reported for women and children:

• Minimum dietary diversity for women is the percentage of women 15–49 years of age who consumed at least 5 of 10 identified food groups in the last 24 hours (FAO and FHI 360 2016).
• Minimum dietary diversity for children is the percentage of children who consumed at least 4 of 10 identified food groups in the last 24 hours (WHO 2010). When paired with data regarding meal frequency, diet diversity can be used to calculate the minimum acceptable diet for children 6–23 months of age³.

Surveys that collect information related to diet include—

• Demographic and Health Surveys
• household consumption and expenditure surveys
• Multiple Indicator Cluster Surveys
• National Micronutrient Surveys
• Knowledge, Practice, and Coverage Surveys
• other research or evaluation activities.

Methodological issues

• Repeat 24-hour dietary recall assessments are difficult to conduct and data may be unreliable unless managed by trained specialists who are familiar with local dishes and implements, as well as interviewing techniques that can assess the portions consumed by the individual versus the rest of the family.
• Depending on the survey tool, country context, population group (i.e., women or children), and indicators, food groups used in data collection and analysis can vary.
• Population-based surveys that use list-based food questionnaires generally rely on a 24-hour recall period, which is not a good measure of habitual diet; but, the longer recall periods used in some surveys, such as 7 or 14 days, can increase recall bias.
• If consumption patterns are measured during atypical consumption days, for example during harvest seasons, diet indicators may not accurately represent the population studied.