Author: Don Jewett
Position: Prof. Emeritus
Institution: Univ. of Calif., San Francisco
E-mail: don.jewett@ucsf.edu
Submitted Date: February 27, 2008
Published Date: February 28, 2008
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

There appears to be an "author-bias" in this article that is intent on showing a lack-of-effectiveness from the data, despite that fact that effectiveness is shown.

They say, at the end of the article:
"Thus, the increased benefit for extremely depressed patients seems attributable to a decrease in responsiveness to placebo, rather than an increase in responsiveness to medication."
When a patient takes a medication, the EFFECT can be considered to be a combination of drug effect (if any) and placebo effect (if any). Now, since, in the severe cases, there is a statistical difference between placebo and drug groups, one must conclude that taking the drug had SOME effect (even if it is mediated through a placebo-mechanism). The fact that the score in the severe placebo group is less than that in the less-severe placebo group (the curvilinear data fit) does not imply that the drug is ineffective. Indeed, the effectiveness is shown.
What is NOT shown is the MECHANISM by which it is effective. THAT needs further research.
The further indication of "author-bias" is seen in accepting AS A REQUIREMENT FOR CLINICAL USAGE, the NICE criteria. What justification is there for these criteria to be ABSOLUTE? So placing the statement about the drugs not meeting the NICE criteria has NOT been shown to be relevant to CLINICAL PRACTICE. If the NICE criteria are more than arbitrary end-points, then the article should quote the references that show the efficacy of these end-points.
I am not a defender of drug companies-- I think they often "push" too much. But the arguments against specific drugs should be scientifically strong. The data in this paper is excellent in combining unpublished data with published data. It is correct in its conclusions about mild depression, but reaches the wrong conclusions about severe depression.