Jacques Barber et al. have recently published the results of a randomized, controlled study conducted between 2001 and 2007, comparing antidepressant therapy, short-term dynamic psychotherapy, and placebo in a 16-week course of treatment for 156 depressed adults. Here is a link to the abstract: http://www.ncbi.nlm.nih.gov/pubmed/22152401
The bottom line in the study was that there was no significant difference between antidepressants, psychotherapy, or placebo. Response rates were 31% for medication, 28% for psychotherapy, and 24% for placebo -- which has a low probability of being statistically different. Remission rates were 26% for medication, 22% for psychotherapy, and 20% for placebo.
Critics trying to explain these findings might attempt to argue that the psychotherapy or the medication regime was not sufficient, etc. -- but I do not see this to be true. The medications (venlafaxine or sertraline) were given at quite sufficient doses for good lengths of time. The psychotherapy was not CBT (which has a larger research evidence base) but there is little reason, in my opinion, to believe that the therapy style was inferior.
The authors attempt to do some secondary analyses looking for explanations, but their conclusions seem quite weak to me (e.g. regarding race or gender). The fact that they spin these conclusions into a prominently framed set of "clinical points" seems quite inappropriate to me -- this is a negative study, there are no "clinical points" to be found here, unless they recommend placebos and cessation of other therapies!
There are a number of issues from this study that I do find very important to discuss:
1) despite a massive amount of data showing that various therapies (e.g. antidepressants or psychotherapy) are effective for various problems, there are examples of carefully-conducted negative studies, such as this one. These results cannot simply be explained away as statistical aberration: there must be a reason why one group of people responds to a treatment, while another does not. Many of these reasons are poorly understood. It may be that the diagnostic category of "major depressive disorder" is inadequate, in that it correlates poorly on its own with treatment responsiveness.
2) the subjects in this study had a high degree of comorbidity (e.g. substance abuse problems, anxiety disorders, and axis II problems). While the severity and chronicity of depression was not found to actually correlate with treatment responsiveness, I suspect that the comorbidities would substantially affect response to a relatively short-term course of therapy.
3) the subjects in this study were socioeconomically disadvantaged; while the effect of SES was also not found to "influence the initial findings," I believe that low SES is not necessarily a direct negative influence upon mental health; rather it is an indirect factor which for many people increases the likelihood of some profound mental health negatives (e.g. unemployment, lack of meaningful or satisfying employment, lack of healthy or safe community, lack of availability to do healthy or meaningful leisure activities, not enough money to eat healthily, etc.). I believe that the environmental adversities need to be looked at very closely in a study of this type.
This leads to what I believe is an obvious explanation for the findings here: there is no therapy for depression that is likely to help unless ALL contributing factors (including obvious environmental contributing factors) are addressed. By way of analogy, I believe it is pointless to treat insomnia using a powerful sedative if a person is sleeping in a room which is continuously noisy, cold, and prone to break-ins by violent intruders. The environmental issues need to be addressed first! Another analogy I have often used is of trying to repair a water supply system for a city: it is a waste of effort to pipe in more water from rivers, or to dig a deeper reservoir, if the walls of the reservoir and the pipes are leaking or bursting because of structural defects. In order for a therapeutic strategy to work, the "leaks" have to be repaired first. For a person with anemia, it is not an appropriate strategy to simply give a blood transfusion: while a transfusion may be necessary, it will not be sufficient--and could even make matters worse-- if the underlying cause of blood loss is not addressed and treated.
In the case of medications or psychotherapy, I believe these can be very helpful, but only if environmental adversity is also remedied. In some instances, of course, relief of a psychiatric symptom could help a person to improve the environmental circumstances. But in most other cases, I think the issue is broader, and could be considered a political or social policy matter.
Another related issue is that I do not believe "depression" can be treated on its own without addressing all psychiatric and medical comorbidities at the same time. Ongoing substance abuse, in my opinion, is often a powerful enough factor--psychologically as well as neurophysiologically--to completely dominate and dissolve the positive influences of psychotherapy or effective medication. In this study, 30-40% of the cohort reported substance use problems.
As a final thought, I think the "5 axis" model of diagnosis in the DSM system deserves some affirmation; many times, however, we only pay attention to Axis I (diagnoses such as depression or schizophrenia, etc.) or Axis II (personality disorder). I think that studies such as this one highlight the necessity to look closely at Axes III (medical illnesses) and IV (social, community, financial, and relational problems). It is likely that issues on these latter two axes can prevent any resolution of problems on the first two.