Objective: To develop and validate a prediction model for psychosocial problems in preadolescence using data on early developmental factors from routine Preventive Child Healthcare (PCH).
Study design: The data come from the 1692 participants who take part in the TRacking Adolescents' Individual Lives Survey, a longitudinal study. Information on early developmental factors (ages 0 to 4 years) was collected from the PCH file. Parents complete the Child Behavior Checklist when their child is age 11. To examine the predictive value of PCH-registered developmental factors on preadolescent problems, several multiple logistic regression analysis were performed, in a derivation sample (n = 1058). The predictive performance of the models was then assessed with area under the curve (AUC) in a validation sample (n = 643) to evaluate the validity of these models.
Results: PCH-registered behavioral problems, attention/hyperactivity problems, enuresis, education level of the father, and being male were found to significantly predict externalizing problems (odds ratios [OR] between 1.4 and 3.7). Internalizing problems were predicted by maternal smoking during pregnancy, sleep problems, and being male (ORs between 1.7 and 3.0). The model for externalizing problems had a modest discriminatory power (AUC 0.66, 95% confidence interval 0.59-0.72). However, for internalizing problems the AUC was 0.54 (95% confidence interval 0.47-0.60), indicating poor discriminatory power.
Conclusions: Findings on early development as registered by PCH are modestly predictive for externalizing problems in preadolescents, but only slightly for internalizing problems.
Copyright (c) 2010 Mosby, Inc. All rights reserved.