To the Editor: We would like to comment on a recent article by Chan et al on clinical practice guidelines for depression in young people.1 We disagree with their proposal to amend National Health and Medical Research Council (NHMRC) guidelines2 to include a statement that "SSRIs [selective serotonin reuptake inhibitors], particularly fluoxetine and paroxetine, should also be considered as a first-line treatment" for major depression in young people. We believe that there is insufficient evidence to assign a grade of "E1"2 to this statement. Chan et al1 quote three randomised controlled trials (RCTs) and one systematic review in support of their argument, but as yet the results of only two of the three RCTs have been published.3,4 Unfortunately, Chan et al do not include a critical appraisal of the significant methodological and analytical problems with each of the studies. Nor is any comment made about risk–benefit ratios, or the fact that even if the results were sound the clinical relevance of such small differences between active drug and placebo is questionable.5 The following brief commentary on the two studies highlights the dangers of carrying out sophisticated procedures such as meta-analysis without sufficient attention to the quality of the trials included in the analysis.

The very high dropout rates (46% of 48 for placebo; 29% of 48 for fluoxetine) in the study by Emslie et al3 raise questions about the reliability of the results. Other interpretations of the findings are plausible. For example, in their study, significant advantage to fluoxetine over placebo on the Clinical Global Impressions improvement rating (a primary outcome measure) was lost when only patients completing the trial were counted (P = 0.2).

More worrying is that Chan and colleagues do not seem to have noticed the dangerously distorted reporting in the study by Keller et al.4 On neither of the two designated primary outcome measures (change from baseline in Hamilton Rating Scale for Depression [HAM-D], and response, set as "fall in HAM-D to ≤ 8 or by ≥ 50%") did paroxetine differ significantly from placebo. But Keller and colleagues never report this negative finding. Instead, the criteria for response are covertly altered (to "fall in HAM-D to ≤ 8", which does achieve significance). The authors then erroneously claim significance on this (altered) primary outcome measure, ignoring the lack of significant change. Thus, a study that showed no significant improvement on either of two primary outcome measures is reported as demonstrating unqualified efficacy.

Similar problems can be found in a more recent article by Emslie and colleagues6 (published after the review by Chan et al1), in which the authors openly acknowledge that the difference between fluoxetine and placebo on their prospectively defined primary outcome measure did not reach statistical significance, yet claim to have demonstrated the drug's efficacy.

Another worry is that Chan and colleagues, in their list of proposed changes to NHMRC recommendations,1 suggest that the availability of SSRIs obviates the need for more expert and thoughtful assessment and management of depression. We are uncomfortable that the prescribing and management of psychotropic medication is portrayed as requiring relatively few skills and resources, to be carried out by those general practitioners who lack training in mental health and/or access to expert mental health services.

We urge the NHMRC to maintain a conservative approach to the use of psychotropic drugs in children with depression unless more convincing evidence is forthcoming.