Processing Speed Mediates the Longitudinal Association between ADHD Symptoms and Preadolescent Peer Problems

The Strengths and Difficulties Questionnaire (SDQ) is a brief behavioural screening questionnaire that asks about children's and teenagers' symptoms and positive attributes; the extended version also includes an impact supplement that asks if the respondent thinks the young person has a problem, and if so, enquires further about chronicity, distress, social impairment, and burden for others. Closely similar versions are completed by parents, teachers, and young people aged 11 or more. The validation study involved two groups of 5-15-year-olds: a community sample (N = 467) and a psychiatric clinic sample (N = 232). The two groups had markedly different distributions on the measures of perceived difficulties, impact (distress plus social impairment), and burden. Impact scores were better than symptom scores at discriminating between the community and clinic samples; discrimination based on the single "Is there a problem?" item was almost as good. The SDQ burden rating correlated well (r = .74) with a standardised interview rating of burden. For clinicians and researchers with an interest in psychiatric caseness and the determinants of service use, the impact supplement of the extended SDQ appears to provide useful additional information without taking up much more of respondents' time.

To determine any long-term effects, 6 and 8 years after childhood enrollment, of the randomly assigned 14-month treatments in the NIMH Collaborative Multisite Multimodal Treatment Study of Children With Attention-Deficit/Hyperactivity Disorder (MTA; N = 436); to test whether attention-deficit/hyperactivity disorder (ADHD) symptom trajectory through 3 years predicts outcome in subsequent years; and to examine functioning level of the MTA adolescents relative to their non-ADHD peers (local normative comparison group; N = 261). Mixed-effects regression models with planned contrasts at 6 and 8 years tested a wide range of symptom and impairment variables assessed by parent, teacher, and youth report. In nearly every analysis, the originally randomized treatment groups did not differ significantly on repeated measures or newly analyzed variables (e.g., grades earned in school, arrests, psychiatric hospitalizations, other clinically relevant outcomes). Medication use decreased by 62% after the 14-month controlled trial, but adjusting for this did not change the results. ADHD symptom trajectory in the first 3 years predicted 55% of the outcomes. The MTA participants fared worse than the local normative comparison group on 91% of the variables tested. Type or intensity of 14 months of treatment for ADHD in childhood (at age 7.0-9.9 years) does not predict functioning 6 to 8 years later. Rather, early ADHD symptom trajectory regardless of treatment type is prognostic. This finding implies that children with behavioral and sociodemographic advantage, with the best response to any treatment, will have the best long-term prognosis. As a group, however, despite initial symptom improvement during treatment that is largely maintained after treatment, children with combined-type ADHD exhibit significant impairment in adolescence. Innovative treatment approaches targeting specific areas of adolescent impairment are needed.

The purpose of model selection algorithms such as All Subsets, Forward Selection and Backward Elimination is to choose a linear model on the basis of the same set of data to which the model will be applied. Typically we have available a large collection of possible covariates from which we hope to select a parsimonious set for the efficient prediction of a response variable. Least Angle Regression (LARS), a new model selection algorithm, is a useful and less greedy version of traditional forward selection methods. Three main properties are derived: (1) A simple modification of the LARS algorithm implements the Lasso, an attractive version of ordinary least squares that constrains the sum of the absolute regression coefficients; the LARS modification calculates all possible Lasso estimates for a given problem, using an order of magnitude less computer time than previous methods. (2) A different LARS modification efficiently implements Forward Stagewise linear regression, another promising new model selection method;