Comparison food and nutrient intake estimates modeled using 14‐day diaries to estimates derived from two 24‐hour recalls

Abstract

The US FDA relies on data from the National Health and Nutrition Examination Survey (NHANES) to estimate food and nutrient intake for the US population. NHANES collect consumption data via two 24‐hour dietary recalls. Short‐term recall data from NHANES do not accurately represent chronic intake of food, and surveys of longer duration are needed to more accurately estimate long‐term intake. We developed a model that combines data from NHANES 24‐hr recall and 14‐day consumption diaries collected from a nationally representative sample of the US population by The NPD Group as part of the National Eating Trends survey. We compared intake estimates derived using our model to 2‐day average estimates based on the dietary recall data. Percent consumers, mean and 90th percentile intake estimates were generally comparable for relatively frequently consumed foods, e.g., milk intake (2‐day: 93%, 285 & 630 g/d; 14‐day: 91%, 280 & 585 g/d). For less frequently consumed foods, larger differences were observed, e.g., egg intake (2‐day: 34%, 67 & 126 g/d; 14‐day: 69%, 18 & 38 g/d). The long‐term estimates were also generally comparable to estimates derived using statistical methods (e.g., NCI method). These results confirm that data from short‐term dietary surveys underestimate the number of consumers and over estimate intakes at the upper percentiles for actual consumers. Supported by the US Food and Drug Administration.