Background: As the use of self-reported data to classify obesity continues, the temporal change in the accuracy of self-report measurement when compared to clinical measurement remains unclear. The objective of this study was to examine temporal trends in misclassification patterns, as well as sensitivity and specificity, of clinically measured versus self-report based body mass index (BMI) from three national lifestyle surveys over a 10-year period.
Methods: The Surveys of Lifestyle Attitudes and Nutrition (SLAN) were interview based cross-sectional survey/measurements involving nationally representative samples in 1998, 2002 and 2007. Data from a subsample of both self-reported and measured height and weight were available from 66 men and 142 women in 1998, 147 men and 184 women in 2002 and 909 men and 1128 women in 2007. Respondents were classified into the BMI categories normal (< 25 kg m(-2)), overweight (25- < 30 kg m(-2)) and obese (>= 30 kg m(-2)).
Results: Underreporting of BMI increased across the three surveys (14%-> 21%-> 24%; p = 0.002). Sensitivity scores for the normal category exceeded 94% in all three surveys but decreased for the overweight (75%-> 68%-> 66%) and obese categories (80%-> 64%-> 53%). Simultaneously, specificity levels remained high.
Conclusions: BMI values based on self-reported determinations of height and weight in population samples are underestimating the true prevalence of the obesity epidemic and this underestimation is increasing with time. The decreased sensitivity and consistently high specificity scores in the obese category across time, highlights the limitation of self-report based BMI classifications and the need for simple, readily comprehensible indicators of obesity.