Key Opinion Leaders and Paediatric Antidepressant Overprescribing

Abstract

Perhaps the most influential efficacy study of antidepressants in children has been the TADS study [11–13] . The main methodological criticism of the TADS study is that it included an unblinded comparison between cognitive behavioural therapy (CBT) alone and fluoxetine plus CBT. This was carried out in parallel to a doubleblind comparison between fluoxetine and placebo. The lack of patient blinding for medication status and placebo control for one half of the study is likely to have exaggerated the benefit seen in the fluoxetine plus CBT group who knew that they were not receiving placebo. The decision not to include a placebo in the CBT group was based on the belief that it was ‘both too expensive and too artificial to have clinical relevance’ [14] . This, however, is not a plausible explanation for why the CBT-only group could not have been converted into a CBT plus placebo group. In any case, the lack of placebo control for half the patients means that comparing results between parts of the trial, with and without a placebo control, is not valid. Yet The lingering controversy about the usefulness and safety of antidepressants for children and adolescents is likely to leave clinicians in a state of confusion. Some papers report that there is little evidence for efficacy and raise concerns about harm [1–3] ; others state that antidepressants are not only safe and effective, but also prevent suicides [4–7] . This journal has previously noted the influence on decision-making of irrational habits and misplaced trust in data submitted by the pharmaceutical industry [8, 9] . In this editorial we address the manipulation of outcomes that result from academics’ alliance with industry. We explain how industry and key opinion leaders have distorted the clinician’s perception of the safety and usefulness of antidepressants for the treatment of depression in children and adolescents through publication bias, poor methodology, and selective reporting.