Improving the accuracy of self-reports on diet and physical exercise: the co-twin control method.

The objective was to examine the association between several obesity-related nongenetic behaviors and body mass index (BMI) and waist circumference (WC) in young adult twins using reports from both twins on their similarities and differences. A total of 713 monozygotic (MZ) and 698 same-sex dizygotic (DZ) twin pairs aged 22-28 years filled in structured questionnaires to compare their eating, physical activity and dieting behavior with their co-twin's behavior, and to report their own eating and exercise habits. In both MZ and DZ pairs, the co-twins for whom both twin pair members concordantly answered that this twin eats more, snacks more, eats more fatty foods and sweet and fatty delicacies, chooses less healthy foods, eats faster and exercises less, had significantly higher BMIs (0.6-2.9 kg/m(2)) and WCs (1.5-7.5 cm). Multivariate regression analysis identified co-twin differences in the amount of food consumed as the strongest independent predictor of intrapair differences in BMI (beta = 0.63 and 1.21, for MZ and DZ, respectively, p < .001) and WC (beta = 1.52 and 3.53, for MZ and DZ, respectively, p < .001). Higher leisure-time physical activity and healthier dietary choices clustered in the same subjects. The measurement of habitual dietary intake and physical activity has previously relied on subjective self-reports that are prone to misreporting. By using comparative measures within twin pairs we found that the amount of food consumed is the major contributor to obesity independent of genetic predisposition.