[Validity of notified anthropometric data for determining the prevalence of obesity]

The aim of this study was to evaluate the validity of the anthropometric data declared by participants in the Survey of Nutrition and Health in the Community of Valencia, Spain in 1994 to estimate the prevalence of obesity using the values obtained by direct measurement in the participants themselves as a reference.

The characteristics of the people who did not declare their weight and/or height were analyzed. Complete information on self declared and measured weight and height was collected in 1,387 subjects (700 males and 687 females). The mean values and proportion of indexes declared and measured were compared and the sensitivity (S), specificity (SP) and predictive values (PV) of a Quetelet Index (QI) QI >or= 30 kg/m2 were estimated to detect obesity in reference to the measured values.

Those who did not declare their weight and/or height demonstrated a higher prevalence of obesity than those who did; 27.9% versus 13.1%, the difference being statistically significant (p < 0.001). The subjects who did declare were found to underestimate their weight, overestimate their height and thus, underestimate their relative weight (RW). This phenomenon was found to be greater in women and in older subjects. The prevalence of undeclared obesity was 10% versus 16.3% in that measured. The S of QI >or= 30 kg/m2 for screening obesity was 66.5%, being 69.3% in women and 63% in men, with a SP of 98.7% and positive PV of 92.4%.

The estimation of the prevalence of obesity from a Quetelet Index >or= 30 kg/m2 based on self-reported data leads to a considerable underestimation of this problem at population level therefore questioning its validity.