You do not currently have permission to view the full text for this document. Please sign in to an account with an active subscription or contact support.
The evidence base for psychodynamic therapy continues to grow, but has tended to be somewhat inconsistent with some meta-analyses showing its usefulness against waiting lists or for specific disorders, and other meta-analyses suggesting that CBT (cognitive behavioural therapy) is better. Most comparison research has been done to show whether one therapy is superior to another and this is a different (and easier) test than whether a therapy is as good as another for complex statistical and practical reasons. In statistical terms, the smaller the difference that is expected, the larger the sample size has to be to find it, because ordinarily-occurring differences unrelated to the action of the therapy have to be evened-out first, and this can only happen if there are lot of people involved. For studies reasonably to be able to test no-difference there needs to be a minimum of 275 people in each of the therapies that are being compared, and this is very uncommon. As a result “equivalence” testing, where no real differences in outcomes are expected, is rare in the therapy world, as it is in the whole field of medicine generally. It is even rarer in meta-analyses where data from different studies are combined to give a sample size that is large enough to find equivalence and be compared.
This recent meta-analysis by Steinert and colleagues is the first to take up the challenge and to test whether psychodynamic therapy is equivalent to therapies that had already been established as efficacious (as opposed to “treatment as usual” or some other such comparator). The researchers put together a rigorous set of criteria to judge whether a study should be included in the analysis. These included a manual, a recognised disorder, reliable and valid outcome measures, adequate sample size, and whether it is tested and controlled for researcher bias (dealing with the problem that research tends to favour the preferred method of the researcher, trainer, and supervisor involved). They