It seems odd that WHO would recommend a dose of only 20 mg/day. According to the product labeling agreed upon between the FDA and the sponsor:

===========

Initial Treatment

Celexa (citalopram HBr) should be administered at an initial dose of 20 mg once daily, generally with an increase to a dose of 40 mg/day. Dose increases should usually occur in increments of 20 mg at intervals of no less than one week. Although certain patients may require a dose of 60 mg/day, the only study pertinent to dose response for effectiveness did not demonstrate an advantage for the 60 mg/day dose over the 40 mg/day dose; doses above 40 mg are therefore not ordinarily recommended.

============

This labeling is freely available at http://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?id=9114#nlm34068-7

Because our reading of the above that 20 mg is not an FDA-approved dose--only a starting dose--we excluded data from 20 mg/day dose groups in our study of selective publication of antidepressant trials in the 2008 NEJM.

Therefore it does not come as a big surprise to me that 20 mg/day is not particularly effective.

• competing interests: no
• Invited by the author to make a review on this article? :
• Experience and credentials in the specific area of science:

  psychiatrist who has done research on the efficacy of antidepressants

• Publications in the same or a related area of science: Yes
  
• References: Turner et al. Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med (2008) vol. 358 (3) pp. 252-60

I have written a blog post commenting on this study:

http://pyjamasinbananas.blogspot.com/2011/01/is-low-dose-citalopram-as-effective-as.html

In particular I was unable to replicate the low p-values of the authors' correlation between citalopram dose and effect size and also it was unclear how the authors were able to use change scores based on the data that was made available from the Ministry of Health which appears to be only final scores with the corresponding standard deviations.

• competing interests: None
• Invited by the author to make a review on this article? :
• Experience and credentials in the specific area of science:

  Have performed meta-analyses before

• Publications in the same or a related area of science: Yes
  
• References: None