Family and genetic specificity of ADHD phenotypes in North American and Australian general population twin samples

Two recent approaches for detecting ADHD liability genes use DSM-IV and latent class criteria. The family/genetic specificity of these approaches was examined in 4,036 North American twins, 2,802 Australian twins and 1,013 Australian non-twin siblings. A multinomial logit model was used to predict each twins ADHD phenotype from the phenotype of their co-twin (and non-twin sibling). All estimated sibling coefficients predicting the same and different twin phenotype were significant, indicating both classification systems are familial, but non-specific. However, when the MZ:DZ odds ratio was examined, differences in the classification systems became apparent. Prediction of different DSM-IV subtype combinations resulted in evidence for genetic non-specificity in the North American sample only; given a DSM-IV Combined co-twin, the MZ:DZ ratio for DSM-IV Inattention was 12.4 (2.4-63.4), χ² = 9.1, P = 0.01. Similarly, given a DSM-IV Inattentive cotwin, the MZ:DZ ratio for DSM-IV Combined was 6.1 (1.4-26.5), χ² = 5.9, P = 0.06. Predicting different latent class combinations, co-twin Inattentive-Impulsive had a significant genetic influence on Mild Inattention class membership for Australians (MZ:DZ = 5.2 (1.8-15.2), χ² = 9.2, P = 0.03) and a significant genetic influence on Talkative-Impulsive class membership among North American females (MZ:DZ = 17.0 (2.1-23.4). χ² = 9.9, P = 0.1 ). Overall, clinically relevant ADHD latent classes showed more genetic specificity than DSM-IV subtypes across samples.