Temporal changes in bias of body mass index scores based on self-reported height and weight

To investigate temporal changes in the bias associated with self-reported (as opposed to measured) body mass index (BMI) and explore the relationship of such bias to changing social attitudes towards obesity.

Using data from the National Health and Nutrition Examination Survey covering two time periods, 1988-1994 and 2005-2008, discrepancy scores between self-reported vs measured BMI were generated. Changes in the sensitivity of BMI categories based on self-reports were examined for six weight groups, both for the US adult population as a whole and major demographic groups. Linear regression models were used to examine temporal changes in average bias, as well as attitudes about weight within each weight category and by demographic group.

Between 2005-2008 and 1988-1994, the bias towards underestimation of a person's BMI based on interview responses has declined among obese individuals, a trend evident in virtually all demographic subgroups explored. Conversely, most demographic groups showed little change in the extent of bias among underweight and normal-weight individuals. Although the 2005-2008 survey respondents underestimated their measured BMI more than the 1988-1994 respondents, this shift can be entirely explained by the increased prevalence of obesity in more recent years. In fact, obese individuals in 2005-2008 were less likely to overreport their height and underreport their weight than their counterparts in the 1988-1994. Evidence from responses to questions about ideal weight and desire to lose weight point in the direction of a shift in social attitudes, which may make it easier to 'admit' to greater weight in surveys.

Over the past 20 years, the bias in self-reported height and weight has declined leading to more accurate BMI categorizations based on self-report. This change is likely to affect efforts to find correction factors to adjust BMI scores based on self-reported height and weight.