To what extent can we predict the future?


Möricke et al. (this issue) are to be commended for their study on behavioral and developmental profiles of 14–15 months old children. The sample encompassing 6.330 children was derived from a birth cohort study which included children born between August 2000 and August 2001 in the province of Utrecht/The Netherlands. Parents filled in the 74-item Utrecht Screening Questionnaire, which is based on selected items of the Childhood Behavior Checklist, the Vineland Social-Emotional Early Childhood Scales and the Early Screening of Autistic Traits Questionnaire. All items were deemed as being suitable for children younger than 18 months and had to be specific for externalizing, internalizing or social-communicative problems. The investigators used exploratory factors and latent class analyses to identify subgroups of children with homogeneous behavioral and developmental profiles; the respective subgroups were compared with existing classifications on a descriptive level. For the exploratory factor analysis, a solution with nine factors fitted the data best, which were termed as deviant communication, negative emotionality, deviant reactive behavior, deviant play behavior, demanding behavior, social anxiety/inhibition, advanced social interaction problems, basic social interaction, measures of fit and accuracy, and sleep problems. Measures of fit and accuracy indicated that a solution with five classes best fitted the data of the latent class analyses. These classes encompassed between 10.8 and 35.9 % of the children and partially differed with respect to age, gender distribution, familial socio-economic status, and the proportions of children of Dutch and non-European origin. Upon combination of both the exploratory and the latent class analyses, three main groups were identified: (1) normal children, (2) children with problems in communication and/or social interaction and (3) children with negative and demanding behavior. The authors discuss that the broad domain of internalizing and externalizing behaviors of the CBCL 1.5–5 are discernible; however, no separate factors for mood problems, anxiety problems, pervasive developmental disorders, attention problems or aggressive behavior were detected. According to the conclusions of the authors, the identified groups may represent predictors of disorders described in existing classification systems (DSM-IV TR, ICD-10, DC 0-3). Obviously, the usefulness of such behavioral and developmental profiles rests upon their predictive value of future behavior. If prediction is poor, we can merely use profiles to characterize current behavior. If, however, prediction of future behavior is good in clinical terms, we can potentially base decisions regarding diagnostic issues, treatment, and prevention on such categorisations. It seems trivial to point out that as a general rule short-term is better than mediumand long-term prediction. In addition, age at assessment constitutes an important variable. Thus, the prediction of future behavior is better if the child is older at baseline. Fortunately, the authors will separately publish follow-up findings at the age of 3 years. Weight adjusted for height (e.g. body mass index; BMI/ kg/m) represents a phenotype that has been studied extensively with respect to tracking over time. According to a US study published in 1994 [5], BMI at age 1 was only weakly correlated with BMI at age 35 (r = 0.15) and r values increased up to 0.5 and 0.7 for 10 and 18 year olds at baseline [5]. The prediction can be improved if parental J. Hebebrand (&) Department of Child and Adolescent Psychiatry, Universitätsklinikum Essen, University of Duisburg-Essen, Wickenburgstr. 21, 45147 Essen, Germany e-mail:


0 Figures and Tables

    Download Full PDF Version (Non-Commercial Use)