Despite the relative scarcity of children following the ODD→CD pathway, we found that a diagnosis of ODD at wave 1 was a significant risk factor for a later CD diagnosis independent of CD continuity for boys, but not girls. This finding was highlighted in our analyses of the first four waves of GSMS (Rowe et al., 2002), and was confirmed over the more extended observation period available here. As noted in the introduction, Lahey et al. (2009) recently reported negligible prediction from questionnaire measures of ODD symptoms to later CD symptoms once baseline levels of CD symptoms were controlled. We extended our analyses of ODD→CD in boys to control for sub-threshold CD symptoms at baseline in a parallel way. Adding this covariate reduced associations between ODD and later CD, but still showed that prior ODD was a significant predictor. In part, these discrepant findings seem likely to reflect the more severe difficulties indexed by a diagnostic measure than by questionnaire approaches. In addition, as outlined earlier, the measures used by Lahey et al. may not have provided the most effective separation of oppositional and conduct problem constructs.
