Objective: Clinical experience suggests that some (but not all) patients with attention-deficit/hyperactivity disorder (ADHD) are highly responsive to the nonstimulant atomoxetine. We conducted a retrospective analysis of randomized controlled trials (RCTs) to identify potential baseline (moderator) and on-treatment (mediator) predictors of responses.
Method: Data from 6 U.S. RCTs among patients aged 6 to 18 years were pooled (N = 1,069; subjects treated with atomoxetine, n = 618). Subjects were categorized as much improved (> or = 40% decrease in ADHD Rating Scale-IV-Parent Version: Investigator Administered and Scored total score), minimally improved (25%-< 40% decline), or nonresponders (< 25% decrease). Logistic regression, analyses of variance, and repeated-measures analyses were used to explore associations between baseline and on-treatment variables, achieving a much improved response at trial endpoint (6-9 weeks).
Results: Forty-seven percent of patients showed a much improved clinical response, and 40% did not respond. Only 13% of the patients had a minimal response. No baseline characteristics predicted achieving a much improved clinical response; the only predictor of achieving this response was being at least minimally improved by treatment week 4 (sensitivity = 81%, specificity = 72%, positive predictive value = 75%, and negative predictive value = 79%).
Conclusions: Clinical response to atomoxetine was bimodal, with most subjects being either responders who were much improved or nonresponders. There were no demographic or clinical predictors of response. However, subjects who ultimately achieved a much improved response were likely to be at least minimal responders by week 4. The recommendation to consider either augmenting or switching treatment in patients who do not achieve at least this level of response to atomoxetine by 4 weeks offers a method for limiting the extended duration of titration to subjects who are most likely to benefit further, while minimizing the duration of exposure in those less likely to achieve an excellent response.