The interesting bits, for me, are the imaging scans which are expensive and slow and you need more than one for comparison per subject. Secondly the 'covert tr

The interesting bits, for me, are the imaging scans which are expensive and slow and you need more than one for comparison per subject. Secondly the 'covert treatment' aspect, and as far as I know only Hjorth et al. do in any reasonable numbers is the giving someone the treatment and usual and tell them it's a placebo and they "shouldn't expect to get better". It's ethically tricky to do that with an anxiety disorder but I imagine a doctor telling someone with depression that will be painfully close to "you're not getting better", "we're not going to help you" and other quite rejecting disheartening ideas. But plain placebo controlled SSRI trials I think show about a 0.2 absolute benefit in SSRIs in depression. i.e. if you compare SSRI with placebo you need to treat 5 people with SSRI to see one more person respond. If you compare SSRI with no treatment, you only need to treat 3 people with depression, to have 1 more respond. Those 3 and 5 are called the 'number needed to treat' and can be compared to the 'number needed to harm'.