We appreciate Dr Carroll's interest in our article. Dr Carroll states that (1) study LYAQ included atomoxetine with and without fluoxetine in patients with attention-deficit/hyperactivity disorder, (2) less than half of the patients in study LYAQ had major depressive diagnoses, and (3) pretreatment depression scores in study LYAQ matched posttreatment scores in other youth studies, concluding that inclusion of LYAQ “invalidates” our aggregate analysis. Dr Carroll is mistaken. We based our inclusion criteria on all fluoxetine monotherapy vs placebo randomized controlled trials. The first 2 active treatment visits did not include atomoxetine and our analysis of the data from this study was restricted to this period before atomoxetine treatment (days 1-42 postbaseline depending on the subject). Furthermore, although patients in this study had comorbidities including attention-deficit/hyperactivity disorder, the majority (81%) of patients had a depressive disorder. Our statistical approach directly incorporates study-to-study variability in baseline severity and background incidence rate of suicidal events, as well as variability in rates of change over time. As such, our inclusion of this study was scientifically appropriate and in no possible way biased or invalidated our overall conclusion and avoided cherry-picking studies to be included in the analyses.