Authors' reply

Dr Halvorsen quite rightly draws attention to the various definitions of clinically significant change, which all have their advantages and disadvantages. We especially agree with the comment that the threshold for clinically significant change should at least coincide with the threshold for reliable change (18.66 in our sample). However, using the threshold of 10 points, as promoted by Schnurr,1 has specific value in our study. First, the 10-point threshold has been shown to be related to changes in quality of life in several samples.2,3 Second, clinically significant change refers to both clinical improvement and deterioration. Most clinicians and researchers would agree that a deterioration of more than 10 points is clinically meaningful and undesirable. Third, treatment duration in our study was relatively short, justifying a relatively low threshold for clinically significant change.

Obviously, if the threshold for clinically meaningful change is increased, the number of participants who benefit from the treatment will decrease. In our study, using the 30% rule (which is based on each participant's score at baseline rather than on the participants' mean score at baseline), the number of participants who experience clinically significant change decreases to six in both conditions. Using the two-standard-deviations rule, the number is reduced even further, to one in the EMDR condition and three in the stabilisation condition.

However, the informative value of this, in our opinion, is limited. The general conclusion of our paper is that six 90-min sessions of EMDR preceded by three 60-min preparatory sessions are of limited efficacy. This conclusion is based not only on calculations of clinically significant change but also on the slopes of clinician-administered and self-reported measures and on the number of PTSD cases (which may also be subjected to different scoring rules4). An alternative conclusion would be that EMDR is not efficacious at all, but that would not be defensible given the significant decrease in self-reported PTSD severity in the EMDR sample and the significant, medium-sized differences between the EMDR condition and the non-randomised wait-list condition.

What the rules for calculating clinically significant change have in common is that they are defined a priori by researchers. In reference to the discussion of our paper, it may be argued that participants themselves should be invited to define at which point they feel that they have meaningfully changed and which criteria define for them where this point lies. By listening to refugee patients' opinions and insights on what is helpful and meaningful to them, we might all be in for a surprise.