As in the population cohort, attrition in the clinical cohort was not completely random. Those who missed one or more follow-up waves (34% of the original cohort) were more likely to be come from low-socioeconomic position families (33% vs 21% χ12 = 10.3, P = .002). Furthermore, they had more parent-reported externalizing problems (0.50 ± 0.29 vs 0.44 ± 0.27, t (df = 532) = 2.62, P = .01) and fewer self-reported internalizing problems (0.35 ± 0.24 vs 0.43 ± 0.24, t (df = 532) = −3.49, P < .001) at baseline.16
