Thanks Graham,
I thought I should also add my small contribution (and not just give you more homework).
From the supplementary material:
SMC was given by a total of 38 different physicians. 27/38 were liaison psychiatrists, 4-GPs and 7-other (non specified physicians).
The treatment in this group included symptomatic pharmacotherapy (especially for insomnia, pain and mood)-not clearly specified. (but my " educated guess" would be mostly antidepressants).
Harm:
In the CBT group (which is harmless " talk therapy") there were 4 patients with episodes of self harm or worsening of depression/CSF symptoms, which were possibly related to the assigned treatment.
In the other groups there were 2 patients with such symptoms.
This probably did not reach statistical significance because of the small sample size, but may possibly be significant. (there were 160 patients in each group).
Furthermore, there was no comparison to patients receiving no intervention. (possibly in this group there would be 0 episodes?).
There was a total of 49 episodes of SAEs-serious adverse events (defined as life threatening, requiring hospital admission, severe worsening of a known condition etc). Those are clustered together for all the patients, without mentioning of group. Those are also not mentioned in the abstract or text (at least I could not find them) and only shown in one table (which can easily be missed) and in the supplementary material. 9 and 8% were in the more physically active groups (APT and GET) and 4% in the SMC and CBT. Could be a coincidence. There is no mentioning of which SAEs were in which group (eg-if there were more orthopedic surgeries and injuries,as opposed to ectopic pregnancy and pneumonia in the GET it could be meaningful)
They include: orthopedic surgeries, epileptic seizures, asthma attacks, accidental head injury which may or may not be related to the treatment assigned.
Which ever way you look at it, this is a relatively high number of SAEs for a group of fairly healthy young people receiving harmless treatment.
90% of the patients had non-serious adverse events. Those are poorly defined. Again, you are not expected to have adverse events (which means any worsening which does not require hospitalization or causes long-term disability) in such a significant percent of patients with a relatively minor illness receiving harmless treatments. Most had more than one episode (there were a total of about 1000 episodes in each group). So, either the illness is not as mild as portrayed or the interventions are not so harmless.
There was no true control group (patients in which there was no intervention) so it is hard to know what is the natural course of the illness and what is harm as the result of those interventions.
Overall most patients (including those in the SMC group) received some form of behavioral/psychiatric therapy.
So, a possible conclusion is that CBT and GET are better than antidepressants?