Ir Med J. 2008 May;101(5):155.

Sir
A recent meta-analysis concerning new-generation antidepressant medication has generated substantial media controversy¹. In their valuable paper, Kirsch et al re-analysed data from all clinical trials submitted to the US Food and Drug Administration (FDA) for the licensing of four new-generation antidepressant medications: fluoxetine, venlafaxine, nefazadone and paroxetine. They focused in particular on the trials’ reports of depressive symptoms rated on the Hamilton Rating Scale of Depression (HRSD), a commonly-used and well-validated rating scale for depression.

The authors did not, however, focus their conclusions on the comparison of depressive symptoms before and after treatment; i.e. they did not focus on the question that matters most to individuals with depression: Will I get better on this medication? Instead, they focussed on a quite different question: Does the improvement on antidepressant medication exceed that seen on placebo? The distinction between these two questions is particularly important in the context of depression, a condition in which, as Kirsch et al note, the placebo effect is “exceptionally large.”

Kirsch et al duly found “a weighted mean improvement” of “9.60 points on the HRSD in the drug groups and 7.80 in the placebo groups”; i.e. there was substantial improvement in both groups. In addition, the difference between the improvement seen on antidepressants and that seen on placebo “easily attained statistical significance”; i.e. there was significantly greater improvement on antidepressant medication than on placebo.

Noting that the magnitude of improvement was related to initial severity, Kirsch et al went on to comment that the difference in responses between medication and placebo groups achieved “clinical significance” only for individuals with severe depression. This does not, however, mean than individuals with mild or moderate depression did not experience significant improvement on antidepressant medication; it simply means that the magnitude of their improvement (which was considerable) was not significantly greater than the improvement experienced by individuals on placebo (which was also considerable). Based almost exclusively on the latter observation, and with scant acknowledgement of the substantial improvement seen on antidepressant medication, Kirsch et al concluded that “there seems little evidence to support the prescription of antidepressant medication to any but the most severely depressed patients, unless alternative treatments have failed to provide benefit.”

In my view this is not the most direct, pragmatic or clinically informative conclusion to be drawn from these data. The following conclusions appear to reflect the data more directly and are more clinically useful:

1. Antidepressant medication produces improvement in individuals with depression, and the magnitude of improvement increases with increasing initial severity;
2. Treatment with placebo also produces substantial improvement, although it is less than that produced by antidepressant medication.

It would be regrettable if the data re-analysed by Kirsch et al were interpreted in such a way as to deny individuals with depression a trial of antidepressant medication. Notwithstanding their own interpretation of these data, Kirsch et al have performed an important task for mental health service-users and providers everywhere: they have confirmed that antidepressant medication is effective in the treatment of depression.

BD Kelly
Department of Adult Psychiatry, University College Dublin, Mater
Misericordiae University Hospital, 62/63 Eccles Street, Dublin 7.
Tel. + 353 1 803 4474
Fax + 353 1 830 9323
E-mail: [email protected]

Reference

1. Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, Johnson BT. Initial severity and antidepressant benefits: A meta-analysis of data submitted to the Food and Drug Administration. PLoS Medicine 2008; 5: 260-8.