Obstacles to a randomised controlled trial of intensive dynamic psychotherapy: an account of the New South Wales Section of Psychotherapy outcomes project

Designs for studying the effectiveness of long-term psychoanalytic treatments: balancing level of evidence and acceptability to patients

Long-Term Psychoanalytic Treatments (LTPT) include both long-term psychoanalytic psychotherapy (LTPP) and psychoanalysis (PsAn). Current opinion seems to be that there is some evidence for the effectiveness of LTPP, but none for that of PsAn. This may be due in part to researchers not balancing the level of evidence of randomized controlled studies (RCTs), cohort studies, and pre-post studies with patients' acceptance of these various research designs used in studying the effectiveness of LTPT. After a review of the merits of eight possible control conditions for LTPT in RCTs and cohort studies, and a consideration of the limitations and merits of pre-post studies, it was found that RCTs pair high levels of evidence with limited degrees of patient acceptance, especially where PsAn is concerned. Cohort studies appear to provide at most a moderate level of evidence. Their acceptability is hardly better than that of RCTs, as it depends on the acceptability of control conditions similar to those of RCTs. The acceptability of pre-post studies is much better, but they can provide, at most, a moderate level of evidence. Apart from randomization, they can meet all methodological criteria for high-quality research (often they do not, but there are ways to correct this). In summary, in the long-term treatment of complex mental disorders with LTPT, RCTs often pair a high level of evidence with limited patient acceptance of the method. Compared to RCTs, cohort studies show a lower level of evidence without much gain in acceptability. Pre-post studies pair the highest level of acceptability with the lowest level of evidence of the three designs. Limited acceptability is not to be confused with no acceptability, nor moderate level of evidence with none.