The objective of this study was to evaluate the efficacy and tolerability of atomoxetine for the treatment of ADHD in 5- and 6-year-old children. This age group was selected because at the time of the study design, atomoxetine was relatively new to the market and few data were available with young children (under 7 years of age). The use of atomoxetine in children under 6 years of age is off-label.
The hypothesis was that atomoxetine would be well tolerated and more efficacious than placebo for reducing core symptoms of ADHD, as measured by the investigator-administered ADHD-RS. An 8-week, double-blind placebo-controlled randomized trial was conducted at 3 academic research sites and was approved by each site's institutional review board, and oversight was provided by an independent data safety-monitoring committee. Recruitment began in October 2005 and ended in June 2008. The last patient visit was September 2008.
Inclusion criteria were written informed consent by a legal guardian and verbal assent from the child; age 5 to 6 years at the time of consent; criteria met for any subtype of ADHD on the Diagnostic Interview Schedule for Children, clinical interview, and on review by a clinical consensus conference of all 3 sites; ADHD is the primary disorder with symptoms present for ≥9 months; a T score of ≥65 on the ADHD-RS; a Children's Global Assessment Scale score of ≤55; a CGI-S score of ≥4 (at least moderate severity); a Peabody Picture Vocabulary Test-IIIA SS score of ≥70; attending day care, preschool, kindergarten, or elementary school for ≥2 half-days per week with a peer group of 8 or more; living with the same parent or guardian for ≥6 months; and having a teacher who is able to provide assessments. Exclusion criteria included concurrent use of psychotropic or other medications with significant central nervous system effects; current effective treatment with atomoxetine; medical contraindication to atomoxetine; current diagnosis of adjustment disorder, autism, psychosis, bipolar disorder, or significant suicidality; history of abuse that may confound symptoms of ADHD; and failure to respond to an adequate previous trial of atomoxetine.
The Diagnostic Interview Schedule for Children–IV12
and a clinical diagnostic interview were completed with the parent or guardian by a child and adolescent psychiatrist, advanced-practice registered nurse, or licensed clinical psychologist. The Peabody Picture Vocabulary Test-IIIA,13
an assessment of receptive language skills, was completed as a proxy for general cognitive ability. The Childhood Autism Rating Scale14
was used to screen for autism spectrum disorders.
Subjects were randomly assigned 1:1 to atomoxetine or placebo. The study drug was initiated at 0.5 mg/kg per day. Four weekly and 1 biweekly visit allowed for flexible titration to 0.8, 1.2, 1.4, and a maximum of 1.8 mg/kg per day on the basis of patient response, tolerability, and clinical judgment of the pharmacotherapist. Atomoxetine was administered in a single daily dose; however, divided doses were permitted at the investigator's discretion for effectiveness and tolerability. Each study visit lasted 30 to 40 minutes; half of that time was spent providing psychoeducation about ADHD and behavioral management strategies using handouts adapted from McMahon and Forehand's Helping the Noncompliant Child: Family-Based Treatment for Oppositional Behavior
Although parents were encouraged to implement the strategies, and pharmacotherapists inquired about their success in doing so, no skills training was provided. Parent satisfaction and the perceived efficacy of this approach is described elsewhere.16
The primary efficacy measure for the study was the ADHD-RS17
total score, completed by investigator interview with the parent present at each visit. The CGI-S and the Children's Global Assessment Scale also were completed at each visit and the ADHD-specific Clinical Global Impression–Improvement (CGI-I) at each visit after baseline. The primary efficacy raters/pharmacotherapists were child and adolescent psychiatrists or advanced-practice registered nurses. These raters completed standardized training on the ADHD-RS, CGI-S, and CGI-I using video-taped, ADHD symptom–specific interviews. A scoring guide with specific anchor points was used by raters to facilitate consistent completion of these measures. Scores on the training tapes were individually reviewed by the principal investigator with each rater to enhance consistency of ratings. Weekly teleconferences were used to discuss scoring of the ADHD-RS as well as reporting of adverse events. The teacher version of the ADHD-RS was completed before randomization and again at weeks 3 and 8.
Adverse events and concomitant medications were assessed at each visit by the pharmacotherapist via open-ended discussion with the parent or guardian. Baseline height, weight, pulse, and blood pressure were obtained at study entry. Weight, blood pressure, and pulse were assessed at each subsequent visit, and height was measured again at the final study visit. Laboratory tests (complete blood count, liver function tests, electrolytes, serum urea nitrogen, creatinine, and lead level), an electrocardiogram, and physical examination were performed at the screening visit. Hematology, chemistries, electrocardiogram, and physical examination were repeated at the final study visit.
A target sample size of 76 subjects (38 per group) provided 80% power to detect a true difference of 8 points in the average change in the ADHD-RS over the 8-week treatment period between atomoxetine and placebo (12-point versus 4-point reduction), assuming an SD of 12 and a 2-sided α level of .05.18
Given that the sample size was increased to 96 subjects (48 per group) to account for a 20% drop-out rate, the trial was powered to identify a moderate standardized effect size of d
Analyses were performed by using SAS.19
Distributions of baseline characteristics for each group were compared using a 2-sample t
test and a χ2
or Fisher's exact test. Generalized estimating equations were used to fit linear or logistic regression models to compare the mean changes in ADHD-RS measures from baseline or log odds for the CGI-I and CGI-S categorical measures.20
Changes in continuous outcomes were calculated as the follow-up minus the baseline value. Primary analysis focused on 8-week changes. Regression models were adjusted for randomization stratification factors (age and study center) and baseline outcome scores (except for CGI-I, for which a baseline score was not available). A similar modeling approach was used for the analysis of blood pressure, weight, and pulse. A 2-sided α level of .05 was used, unless otherwise specified.
As a secondary analysis, weekly changes relative to baseline were compared between treatment groups. Covariates in the regression models included time, treatment, and their interaction, as well as age, center, and baseline outcome. An autoregressive working correlation structure was used. If the time-by-treatment interaction was statistically significant, time-specific comparisons between treatment groups were made. A Pocock group-sequential 2-sided α level of .014 was used when making time-specific comparisons between treatment groups at weeks 1 through 8, and a level of .03 was used when making comparisons at weeks 3 and 8.21,22
Data from all subjects who were dispensed medication and had at least 1 postrandomization visit were analyzed. A multiple imputation approach, based on predictive mean matching, was used to impute the missing measures. Results presented are based on imputed data, which were similar to the results using only observed data (not presented).
Adverse events were defined, and a patient-level adverse-event indicator was coded for each type of event if the event occurred any time during the treatment period and was either not present at baseline or progressed in severity from baseline. A χ2 test, or Fisher's exact test, was used to compare the distribution of adverse events between treatment groups.