Under-Reporting Nutrient Intakes in Nutrition Research
Have you ever kept a diet record or filled in a dietary questionnaire? Was it accurate? Did you forget to include something that you ate? Did you change your diet to make it seem like you eat healthier than you normally do? These are some of problems that make accurate assessments of food and nutrient intakes difficult. Last month, Archer and colleagues reported results of an analysis of under-reporting in the NHANES nutrition survey. There has been a recent reply by Mitka. Regular readers of TalkingNutrition will be familiar with the NHANES survey, and it is also the basis for the data for our Micronutrient Calculator.
Archer and colleagues used NHANES data from all published cycles (1971 to 2010) and calculated the ratio of reported energy intake from two 24-hour dietary recalls to calculated Basal Metabolic Rate (BMR) according to the Schofield equations. The ratio of reported energy intake (rEI) to BMR gives an indication of the amount of energy used for physical activity compared to energy intake. When the ratio of reported energy intake to BMR was less than 1.35, subjects were considered to have under-reported their energy intakes. A cut-off of 1.35 indicates that individuals are confined to bed. For example, using a Total Energy Expenditure calculator, my BMR is around 1300 kcal. If I spent 8 hours in bed and the rest of my time sitting, indicating that I was completely sedentary, I would expect to have total calorie needs of around 1700, and this corresponds to a ratio of 1.3. I am not confined to bed, however, and when I estimate time spent on my normal activities such as cycling, walking to the shops and doing housework, my energy needs increase to around 2100 and the ratio of intake to BMR increases to 1.6. Estimates of my energy intake from questionnaires should therefore be in the range of 2100 kcal, and if the estimates fall to 1700 or less, it would indicate that I had restricted my energy intake to an unsustainable level (for example, to lose weight), I was confined to bed and had adjusted my energy intakes accordingly, or that I had under-reported my energy intake. While it is likely in population studies that a certain proportion of any population has a reduced energy intake due to weight loss attempts or illness, studies using more reliable measures of energy intakes (such as the doubly-labeled water technique) have found that under-reporting of food intakes is an important problem.
The main results of the study by Archer and colleagus indicate chronic underreporting in the NHANES dataset occurs, with average rEI/BMR of 1.19 for women and 1.31 for men in total. Body weight appears to be a predictor of accurate reporting, with the greatest proportion of “plausible” rEI/BMR ratios reported by people with a body mass index between 18.5 and 25. The rEI/BMR ratio in the overweight and obese was less likely to be plausible for free-living adults. The discrepancies between total energy expenditure and rEI were also calculated, assuming constant physical activities levels (even though the NHANES dataset does contain measures of physical activity for a subset of participants). These results indicate a reporting deficit of 100-200 kcal in most years of the survey for normal weight participants, 300-400 kcal for overweight participants, and 500-600 kcal for obese participants. The authors suggest that calorie-dense foods are underreported because people are influenced by public policies aimed to improve dietary behaviors. In terms of measuring micronutrient intakes, this may mean that nutrients such as vitamin E are also under-reported as this vitamin is obtained. Other foods such as vegetables and protein-rich foods are less likely to be under-reported, therefore estimates of micronutrient intakes found in these foods are more likely to be accurate (for example, vitamin C, vitamin B6, vitamin B12, niacin, vitamin K).
The authors conclude with some strongly-worded criticisms of the way that energy intake is measured at the population level. However, they do not suggest more accurate means of obtaining these measurements. The authors point to the increase in consumption of commercially prepared foods as a contributor to inaccuracies. My experience has been that commercially prepared foods are more likely to have been assessed for their energy and nutrient contents, and are available in defined portion sizes, unlike foods cooked at home for which nutrient compositions vary and must be calculated.
Nutrition researchers are generally aware of the limitations of dietary assessment instruments (under-reporting in NHANES is mentioned as a caveat of measuring nutrient intakes by the National Cancer Institute). Research is ongoing to identify more accurate ways to improve the accuracy of dietary assessments, particularly at the population level. For example, Stumbo recently describe various methods using hand-held and digital technologies to obtain more accurate estimates of food intakes. The current method of dietary assessment in NHANES has been refined over the years and represents the best estimate possible given the size of the study and the resources available. Technological advances in future will hopefully overcome the current difficulties in accurately assessing nutrient intakes in populations.
Archer E, Hand GA, Blair SN. Validity of u.s. Nutritional surveillance: national health and nutrition examination survey caloric energy intake data, 1971-2010. PLoS One. 2013 Oct 9;8(10):e76632. doi: 10.1371/journal.pone.0076632.
Mitka M. Do Flawed Data on Caloric Intake From NHANES Present Problems for Researchers and Policy Makers?. JAMA. 2013;310(20):2137-2138. doi:10.1001/jama.2013.281865.
Stumbo PJ. New technology in dietary assessment: a review of digital methods in improving food record accuracy. Proc Nutr Soc. 2013 Feb;72(1):70-6. doi: 10.1017/S0029665112002911.