Filling a dietary data gap? Validation of the adult male equivalent method of estimating individual nutrient intakes from household-level data in Ethiopia and Bangladesh

- Adult male equivalents are used to estimate individual nutrient intake from household data.
- AME-based predictions were reasonably accurate for all but the youngest children.
- AME-based predictions were more accurate in Ethiopia than Bangladesh.
- AME predictions were less accurate for animal protein than energy, protein, & iron.
- Individual dietary data should complement household data to assess diet deficiency.

Many researchers use data from Household Consumption and Expenditure Surveys (HCES) to estimate individual food and nutrient intake when individual dietary data are not available. They assume that food is allocated within households according to members' proportional energy requirements relative to an adult male (called an adult male equivalent, or AME). This study sought to validate AME-based estimates of individual consumption of calories, protein, iron, and animal source protein (ASP) across 10 age-sex categories, using data from Bangladesh and Ethiopia containing both household and individual-level consumption data. The study also assessed the accuracy of adjusting for meal partakers and physical activity levels (PAL), and compared energy-weighted AMEs to nutrient-specific AME predictions.Energy AME-based predictions of nutrient intake were generally accurate within ten percentage points of individually reported intakes, but were less accurate for infants 6-23 months and children in Bangladesh than for other demographic groups. AME predictions were more accurate: (1) in Ethiopia than in Bangladesh, (2) for predicting intake of the three nutrients rather than ASP, (3) for estimating nutrient intake rather than adequacy, (4) using energy-weighted AMEs rather than nutrient specific weights, and (5) using moderate PAL for youth and adults rather than high PAL. Adjusting for meal partakers did not consistently improve the AME-based predictions. Energy based AME estimates from household data can produce a useful proxy of average intake for certain population subgroups, however individually measured dietary assessment remains the best approach to identify groups at risk of nutrient inadequacy.