Investigating the significance of the data collection period of household consumption and expenditures surveys for food and nutrition policymaking: Analysis of the 2010 Bangladesh household income and expenditure survey

Efforts to improve the use of Household Income and Expenditure Surveys (HIES) for nutrition policy decisions face numerous methodological issues. Our objective was to explore how nutrition indicators vary by data collection period (number of days of household consumption data available) using the 2010 Bangladesh HIES. Household availability and individual apparent consumption of nutrients, foods, and food groups were estimated for each of 7 consecutive 2-day recall periods, both separately and in combinations representing 2, 4, 6, 8, 10, 12, or 14 days of data. Differences among cumulative data collection periods in mean apparent nutrient consumption and proportion with low values were statistically significant but small (<4 percentage points), suggesting that a data collection period of one week or less would yield the same conclusions as a longer (and more costly) data collection period of 14 days for these outcomes if the recall period were held constant. For all food groups except staple foods, the proportion of households consuming food groups and selected potentially fortified foods increased, and the mean amount consumed among consumers (the 'conditional mean') decreased with longer data collection periods; this observation is likely related to frequency of consumption and suggests that the choice of data collection period for these items will be determined by the intended application. Additional research is necessary to confirm the external validity of these findings.