Considering only pre-post effect sizes as evidence of treatment effect is problematic in depression research as depression has natural tendency to wax and wane. In RCTs, generally the remission or recovery rates in placebo controls are compared against antidepressant treated samples. You may be able to get baseline average clinical scores reported in a group-wise fashion in most RCTs, and if the primary studies also report these scores in addition to dichotomous outcomes, you may be able to calculate pre-post changes in the active treatment groups. Hope this helps!

Thanks for your response, Lena! I have values from a benchmark (Minami et al., 2007) showing that mean pre- post-treatment ES for natural history of depression (no-treatment and waiting list control groups basically) are around 0.371with the BDI-II (they are larger with HRSD). Therefore I was wondering whether I could find any study, preferably meta-analysis, doing what exactly you suggested (gathering pre and post scores from primary studies) in a systematical fashion. Do you know any?

Hi Rodrigo. For the effect sizes of various antidepressants for the treatment of depression, you might want to review the Turner meta analysis that caused such a stir a few years ago. It's linked below. This should provide you a good starting point.

http://www.nejm.org/doi/full/10.1056/NEJMsa065779

Thanks Afton! It is a great starting point, indeed. Unfortunately, it still doesn't help me in my question because it doesn't provide enough information for calculating the mean effect size expressed in Cohen's d (n for drug and controls would be the most important).

As psychoanalyst I cannot calculate as you do, and don't like that (sorry, it is not meant as critic) In my practice there were many analysands and more therapy patients who came with heavy medication without help. They left their medication quite soon and began to work for finding a way out from their problems.So they did. I understand that diagnosis depression is a large variety of structural guestions which have to be studied before you choose the therapy form both in case of therapy, analysis or medication.

Marira

Thanks, Jan and Marita, for your input.

You can have a sense of effect sizes (and can use imperfect estimation methods to derive effect sizes), even in the absence of pre-post mean ES for the purpose of interpretation of available data. Effect sizes for any treatment approach in depression are modest. However, the heterogeneity of the current construct of depression is a major obstacle. The "true" relative effectiveness of medication vs psychotherapy vs placebo will be different for discrete latent classes of depression.

Dear Rodrigo,

i was was also also wondering about this and found this article Article Antidepressants versus placebo in major depression: An overview

You may also be interested to see this review https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(17)32802-7/fulltext?code=lancet-site