In her March 2002 editorial, Dr Moncrieff raises doubts about the efficacy of antide-pressant drugs, and argues that side-effects of the active drug may explain some of the differences owing to the associated increased expectancy of a positive effect. Our 24-week study (Reference Malt, Robak and MadsbruMalt et al, 1999) comparing the efficacy of empathic primary-care counselling and support combined with placebo, a selective serotonin reuptake inhibitor (sertraline) or an α₂/5-HT_2/3 antagonist (mianserin) in 372 subjects with depressed mood does not support her arguments.

In our study the general practitioners were required to systematically explore possible side-effects. This method yields a greater prevalence of side-effects than when only spontaneously reported side-effects are considered. The mean numbers of baseline-corrected UKU-elicited side-effects (Reference Lingjaerde, Ahlfors and BechLingjaerde et al, 1987) during the study were 7.11, 6.51 and 6.45 after 8 weeks of treatment with sertraline, mianserin and placebo, respectively, and 3.16, 3.09 and 3.02 after 24 weeks of treatment (NS). This means that prevalence of side-effects is unlikely to explain the difference in response.

Another observation arguing against the hypothesis that non-specific side-effects may explain differences between active drug and placebo is the fact that we obtained differences in response over time among the three treatment arms. As would be expected by the pharmacodynamic profiles of the drugs, mianserin induced a faster initial response, while sertraline demonstrated an advantage in the long run explained by better efficacy among subjects with high neuroticism. At the end of the study, the physicians were not able to identify reliably the treatment given to each of their patients.

Furthermore, differences in effect size between the treatments (see Table 1) clearly demonstrated the advantage of anti-depressant drugs on core symptoms of depression. These differences are well beyond the estimated mean effect size of 0.27 reported for active placebo.

Table 1 Differences in effect size on item levels measured on the Montgomery—Åsberg Depression Rating Scale between three forms of treatment in a 24-week randomised treatment trial of patients with depression (n=372) in primary care. Intention-to-treat data obtained from Malt et al. (Reference Malt, Robak and Madsbru1999)

	Sertraline v. placebo	Mianserin v. placebo
Observed sadness	0.04	0.19
Reported sadness	0.47	0.39
Anxiety	0.34	0.22
Insomnia	0.59	0.36
Appetite	-0.09	0.05
Concentration problems	0.08	-0.15
Lassitude	0.36	0.24
Inability to feel	0.10	0.22
Pessimistic thoughts1	0.54	0.45
Suicidal ideation1	0.13	0.04

Instead of questioning the efficacy of antidepressant drugs in depression, attention should be directed at the critical question regarding the characteristics of those patients who will benefit from receiving antidepressant drugs in addition to psychological intervention.