An alternative to dietary data exclusions

J Am Diet Assoc. 2007 May;107(5):792-9. doi: 10.1016/j.jada.2007.02.003.


Objective: To examine an alternative to exclusion of apparently implausible data when examining the relationship of dietary energy density to total energy intake and of energy intake to body mass index (BMI). The objective is to show the advantages of retaining all available data but stratifying based on level of energy intake.

Subjects/settings: We examined 24-hour dietary recall data obtained from 7,720 adult participants (18 to 64 years old) in the nationally representative National Health and Nutrition Examination Survey (NHANES) 1999-2002.

Main outcome measures: The relationship of energy density to energy intake, and of energy intake to BMI was assessed using linear regression models adjusting for age, sex, smoking status, and exercise. A sensitivity analysis was done to determine whether the relationship differed when generally accepted exclusionary criteria were applied.

Statistical analysis performed: Although the relationship of energy density to energy intake is similar across a large range of energy intakes, it differs at very low levels of energy intake. Energy intake is much less dependent on energy density at low intakes. The relationship of energy intake to BMI is different at both high and low levels of intake. Furthermore, the nature of the relationship between BMI and energy intake differs based on reporting status (whether reported energy intake is consistent with energy expenditure estimation).

Conclusions: Instead of excluding observations based on energy intake, examining all the data but stratifying by level of intake may be more informative of population nutrient intake.

MeSH terms

  • Adolescent
  • Adult
  • Age Distribution
  • Algorithms
  • Body Mass Index*
  • Dietary Fats / administration & dosage*
  • Energy Intake / physiology*
  • Exercise / physiology
  • Female
  • Health Promotion
  • Humans
  • Linear Models
  • Male
  • Mental Recall
  • Middle Aged
  • Nutrition Surveys*
  • Nutritive Value
  • Public Health
  • Self Disclosure*
  • Sex Distribution
  • Smoking
  • United States


  • Dietary Fats