“…As Pickles and Angold (2003) state, in order to understand the relationships between psychopathology, its aetiology, outcome and other factors, "we need to adopt a truly empirical approach that is unblinkered by either categorical or dimensional prejudices" (p548). Retrospective studies of early predictors of psychopathology are limited by issues of recall bias while prospective cohorts of subjects at 'high risk' (Besag, 2006;Cassell et al, 2007;Gamliel, Yirmiya, & Sigman, 2007;Iverson & Wozniak, 2007;Landa, Garrett-Mayer, Landa, & Garrett-Mayer, 2006;Loh et al, 2007;Merin, Young, Ozonoff, & Rogers, 2007;Toth, Dawson, Meltzoff, Greenson, & Fein, 2007;Yirmiya et al, 2006;Zwaigenbaum et al, 2005b;Yirmiya et al, 2007;Zwaigenbaum et al, 2005a;Zwaigenbaum et al, 2005a) may reveal early indicators of neurodevelopmental disorders, such as autism, but are subject to a variety of forms of selection and ascertainment bias. For example, prospective studies of children exposed to smoking in pregnancy and of children with very low birth weights have shown increased rates of ADHD and of ADHD/autism symptoms (Hultman et al, 2007;Indredavik, Brubakk, Romundstad, & Vik, 2007), but to what extent such factors alone can contribute to these difficulties can best be determined in the context of general population birth cohort studies.…”