I have written too much over the years on anxiety disorders. Most of the time I have not discussed this paper, which does not fit neatly into the modern diagnostic classification, but shows that CBT is as effective or more so than antidepressants in neurosis. neurosis still has clinical utility: Generalized Anxiety disorder, Dysthymia and Panic frequently co occur. This is one of the few RCTs that compared CBT with benzodiazepines and antidepressants directly.

But the paper does not nearly fit into the DSM system, and is generally ignored.

You can see from this that benzodiazepines did worse than placebo with both Hamilton scale (HAM-A for anxiety and HAM-D for depression). antidepressants did better. However, how many people required rescue medication?

It appears the placebo group were medicated according to symptoms, and thus did better than the benzodiazepine group. It also appears that CBT is about as effective as medication. This has led to most clinical guidance groups not recommending benzodiazepines for depression of psychosis.

But current review methodology is removing these older papers from our memory: it is as if any work done before the mi 1980s. when the consensus to use the American system was forged — in part because the FDA required it — is discounted. As if the previous generation could not do a clinical trial. And as if mathematical models are more valid — such as network meta analysis — than a five armed trial.

I am a fanboy for CBT. Because of trials such as this. Not because of the meta analyses and systematic reviews, including the ones I am guilty of producing.