Background: Using diagnostic information obtained at two different times, we incorporated error of measurement into structural equation twin models to evaluate the contribution of additive genetic, common environmental, and individual-specific environmental factors to the liability to binge-eating and broadly defined bulimia nervosa (BN). We also evaluated the validity of the equal environment assumption (EEA) with reference to these two phenotypes. Methods: We interviewed 1897 female twins (including both members of 854 twin pairs) from a population-based register about their lifetime history of binge-eating and of broadly defined BN twice, approximately 5 years apart. Results: The reliabilities of a lifetime history of binge-eating (kappa = .34) and of broadly defined BN (kappa = .28) were low. Based on single interviews, the heritability of binge-eating was estimated to be 50% and broad BN 60%, with the remaining variance attributable to individual-specific environment. Common environmental influences had no effect on liability to either trait. By combining information from two interview waves and thereby incorporating error of measurement into a structural equation model, the estimated heritability of the latent vulnerability to binge-eating (82%) and broadly defined BN (83%) increased substantially. Although there were no violations of the EEA detected for binge-eating, cosocialization influenced twin concordance for broadly defined BN. Conclusions: Lifetime histories of binge-eating and broadly defined BN appear to be highly heritable conditions of low reliability. Biol Psychiatry 1998;44:1210-1218. (C) 1998 Society of Biological Psychiatry.
