Validity of self-reported BMI in older adults and an adjustment model

Study objectives To determine reporting bias of self-reportedvs. measured anthropometrics (body weight, height, bodymass index, BMI), and to adjust self-reported BMI regardingthe bias.

Methods We compared self-reported with measured anthropometricsutilizing 659 control persons (age mean 68; range37 to 80 years) from a case-control stroke study. The Bland-Altman approach examined the agreement between selfreportedand measured values. A linear model was appliedto correct the bias dependent on sex, age and self-reportedBMI.

Results Under-reporting of weight and over-reporting ofheight was found. On average, this resulted in lower selfreportedBMIs by 1.0 kg/m2 in men, 1.2 kg/m2 in women(p<0.001). Bias correction of self-reported BMIs was derivedfrom self-reported BMI (p<0.001), age (p<0.001), age-BMIinteraction (p<0.001) and sex (p<0.05). Under-estimation ofcorrect BMI resulted in the under-estimation of an overweightprevalence, with relatively low sensitivity regarding selfreportedvalues (88 %). Our estimates should be recalibrated,if applied to other studies.

Conclusion Self-reported anthropometric measures are systematicallybiased despite high correlations with measuredvalues. A correction removes the average bias and improvesaccuracy.