Can self-reported dieting and dietary restraint identify underreporters of energy intake in dietary surveys?

Nutrition Research Newsletter, Nov, 2006

Underreporting is endemic in most dietary studies and presents a serious challenge when interpreting associations between diet and health to avoid spurious associations. The advent of the doubly labeled water method to measure energy expenditure raised awareness of the extent of underreporting of energy intake, particularly in obese and weight-conscious subjects. Underreporting may be due to a conscious or subconscious misreporting or a true measure of atypical intake, such as during a period of dieting. It is unclear whether individuals who are likely to underreport dietary intake can be reliably identified within dietary surveys. Questions on self-reported dieting and behavior such as dietary restraint may be useful tools in identifying underreporting.

Dietary restraint is defined as any type of attempted or actual self-imposed food restriction for the purpose of weight control. Assessing dietary restraint in studies is difficult. Measures of dietary restraint (Eating Inventory and Dutch Eating Behaviour Questionnaire) were developed in the mid-1980s to measure past and current cognitive and behavioral strategies of dieting and restrained eating. High restraint scores from these measures are commonly observed in the obese and some studies have examined the influence of obesity, dietary restraint, and dieting on underreporting of energy intake. The purpose of this study was to examine whether high dietary restraint scores and self-reported current dieting to lose weight were associated with higher underreporting (assessed using an individualized method) in a random sample of adults participating in the National Diet and Nutrition Survey (NDNS) 2000 and also to assess whether these associations observed were further influenced by BMI status.

NDNS is a population-based dietary survey of adults (18 yrs to 64 yrs) selected using a multistage random probability design. Trained interviewers from the Office of National Statistics undertook the field work for the NDNS survey. Participants were asked to keep a weighed record of all food and drink they consumed over seven consecutive days both in and out of the home. Each participant was given a set of accurately calibrated digital food scales and two recording diaries for foods and drinks consumed at home and a smaller diary for eating and drinking away from home. Participants were instructed by the interviewer on how to use the scales to weigh food, drinks, and leftovers; how to record each item; the brand name of the product; and, where appropriate, the method of preparation. The items consumed and leftovers were weighed. Participants were encouraged by the interviewers to weigh everything they could, and where items could not be weighed, to write down as much information as possible about the food item consumed, including the portion size and an estimate of the amount of any leftovers. Total energy (kilocalories) consumed during the survey week was calculated from a nutrition composition database compiled by the Food Standards Agency and mean reported daily energy intake computed. Physical activity was also recorded during the 7-day dietary record in the dietary diary. The diary asked for details of time spent on a list of specified activities, the duration, and intensity of activity. Information was collected on time spent sleeping, at work, at college, walking, in a range of listed household or similar activities, including housework and gardening, and a range of listed sports and leisure activities. Participants recorded how long they spent doing activities to the nearest 10 minutes. Each subject's physical activity level was calculated from the diary.

Participants were also asked to complete the Dutch Eating Behaviour Questionnaire to assess eating restraint using computer-assisted self-interviewing or paper. Responses to 10 items on restrained eating were used to calculate the restrained eating score. The mean restrained eating scores were categorized into <3 (low-restrained), and >3 (high-restrained) for further analysis with levels of underreporting.

In the NDNS, 668 men and 826 women completed the 7-day dietary record, activity diary, and the Dutch Eating Behaviour Questionnaire. Mean BMI was 27.1 kg/[m.sup.2] in men and 26.5 kg/[m.sup.2] in women. Estimated median underreporting equated to 27% of energy needs in men and 29% in women. The mean restrained eating score was higher in women compared with men. Eighty men (12%) and 199 women (24%) reported dieting to lose weight during the dietary recording. Mean age, estimated energy requirements, estimated underreporting, dieting to lose weight, and restrained eating score were higher in overweight/obese men and women than their lean counterparts. In women, reported energy intake was lower in overweight/obese than lean women, but this was not observed in men.

Significant positive associations between estimated underreporting with restrained eating scores and current dieting were found in both men and women. Underreporting was higher in men and women reporting current dieting than nondieters. Among women, self-reported dieting to lose weight was associated with greater underreporting in both lean and overweight women. However, only lean high-restrained women significantly underreported more than those classified as low-restrained, with similar levels of underreporting in overweight women regardless of restraint score.