Objective: To examine the validity of self-reported values for current anthropometric measurements and factors related to misreporting.
Design: E3N, a prospective cohort study of cancer risk factors, conducted in France and part of the European Prospective Investigation on Cancer. E3N comprises 100,000 women, born between 1925 and 1950, followed with self-administered questionnaires sent every 18 to 24 months starting in 1990.
Subjects: 152 women for the validation study of self-reported anthropometric measurements, and 91,815 women selected to evaluate factors affecting misreporting of body silhouette.
Statistical analysis: Paired t tests, Pearson and Spearman correlations were applied to evaluate the validity of self-reported measures, and analysis of variance and logistic regression were used to assess the factors influencing misreporting of silhouette.
Results: The correlation coefficients between self- and external measurements were high. All but sitting height (r = 0.56) were more than 0.80, with weight and bust (nipples) measurement correlation coefficients attaining 0.94. The correlation between body mass index (BMI), measured by the technician and the self-reported silhouette, was 0.78. Small height was always associated with misclassification. Specific factors related to a more favorable perception of body silhouette were: being overweight, small height, younger age, and a lower level of education. These women were also more frequently unmarried, more physically active, and had had a slender body shape during adolescence. Results denoting a less favorable perception of body shape were reversed. CONCLUSION/APPLICATIONS: Self-reported measurements (made with or without help) are valid measures in epidemiological studies. Body silhouettes are simple and useful indicators of body mass index. However they should be interpreted with caution in certain instances, especially for overweight subjects.