EMDR FORUM ARCHIVE, 2000
Re: differential mechanism/dose reponse
Cahill · 11/07/01 at 12:09 ET
Hello Bob: Thanks for your contribution. I agree that resolution of these issues will have great practical and theoretical implications. Your main point, if I understand you correctly, is that the critical evidence would be to show that the dose response curves differ in slope between exposure and EMDR. You then give a striking example of this, which, in the absence of a specific citation, I am assuming is a hypothetical example. I agree with you: If such a result were reported, I would have to conclude that something very different was going on here. However, to the best of my knowledge, no study has systematically varied exposure dose for both treatments. The closest thing we have to this is Susan Rogers' study that looked at a variety of measures after a single session of each therapy. Her study reported that EMDR involved an average of 20 minutes of exposure time, while exposure therapy involved an average of 35 minutes. This difference was not significant. Further research is needed to extend these findings to determine whether it is truly the case that EMDR achieves better results with less exposure. Until then, I have to conclude that there is no evidence for differential mechanisms. Turning now to your comment about differential drop out. To the best of my knowledge, there is not a single study to directly compare EMDR with exposure therapy that found greater attrition in the exposure therapy condition. Indeed, there is no evidence of differential attrition in favor of EMDR for any active treatment against which it has been compared. Below I consider each of the published comparative treatment outcome studies (EMDR vs. some other active treatment). Devilley & Spence, 1999 - three of 15 subjects dropped out from the exposure therapy plus stress inoculation training plus cognitive therapy, one after each of sessions 1-3. Six of 17 subjects dropped out from EMDR, 5 occurring after session 1. If we only consider the dropouts that occurred after session 1 (before active treatment began), we are left with 2/14 in cognitive behavioral therapy and 1/12 in EMDR. This difference is not significant. Again no support here for your hypothesis. Lee et al. (in press, available at ww.childtrauma.com/artchris.html) - "three participants dropped out of treatment: one from the SITPE condition, one from EMDR, and one went to prison." (p. 5). Vahghan et al. (1994) - EMDR vs. applied muscle relaxation vs. image habituation training. As best I can tell, there were no dropouts from any group. Carlson et al. (1998) - EMDR vs. WL vs. Biofeedback assisted relaxation. Twelve (of 47) subjects dropped out prior to group assignment. "After group assignment and prior to posttreatment assessment, no participants dropped out of the control or eye movement groups and one participant dropped out of the biofeedback-assisted relaxation group." (p. 6). Marcus et al. (1997) - EMDR vs. standard care. "...10 participants declined to participate and one dropped out." Scheck et al. (1998) - EMDR vs. Active Listening. As best I can tell, there were no dropouts in either group. Edmond et al. (1999) - EMDR vs. standard care. "No attrition occurred during the pre- and posttest phase of the study..." (p. 110). Thus, across a range of studies and a range of alternative treatments, there is no evidence that there is less dropout from EMDR than other treatments, even when they include exposure. At least not yet. There are at least three additional comparative studies pitting EMDR with some form of cognitive behavior therapy, the results of which are yet to be published (these are studies by Ironson et al (in press), Taylor et al. (presented at the World Congress), and Rothbaum et al. (to be presented at this year's AABT). As the results of these studies become more widely circulated and their procedures subjected to scrutiny, the field may be in a better position to answer these questions. So, why do people advance this belief? The "evidence" that has been used to support the claim that there is less dropout for EMDR relative to exposure therapy involves making comparisons across studies. Compared to the above (and other) studies of EMDR, studies of exposure therapy by Foa et al. (1991, 1999) and Marks et al. (1998) have reported higher levels of dropouts from their active treatments. Two points need to be made about these studies. First, the dropouts in exposure therapy in these studies is not larger than dropouts from other active treatments WITHIN THE CONTEXT OF THE SAME STUDY. For example, in the Foa et al. (1991) study, dropouts rates for exposure, stress inoculation, and supportive counseling were 3/17, 4/14, and 3/14, respectively and not significantly different from one another. In the Foa et al. (1999) study, dropout for exposure therapy alone was 2/28, compared to 7/26 for stress inoculation training, and 8/30 for the combination of exposure therapy and stress inoculation training. The comparison between exposure therapy alone with the combined treatment was significant, with exposure alone having fewer dropouts. In the Marks et al. study, dropouts were 3/23, 1/19, 5/24, and 1/21 for exposure, cognitive restructuring, combined exposure plus cognitive restructuring, and relaxation. In the Tarrier et al. (1999) study, dropouts from active treatment were 6/35 and 4/37 for exposure therapy and cognitive therapy, respectively (not significantly different). Second, valid inferences about differential dropouts for different treatments cannot be made by comparing treatments across studies, as this confounds treatment with study. We can only infer that Treatment A causes more dropouts than Treatment B when we know that everything else between the two conditions are the same. We cannot assume this to be the case when comparing treatments across studies. In addition to differences in the treatments, there are differences in the settings in which the treatments take place (e.g., people utilizing their own HMO resources as in the Marcus et al. study vs. going to psychiatric hospital for treatment in the Foa et al. studies); there are likely to be differences in samples (simply requiring a PTSD diagnosis does not equalize the samples); differences in general therapist skills (beyond the specifics of the target treatments); differences in the planned duration of treatment (3-4 sessions in some EMDR studies vs. 9+ in many CBT studies) which permit fewer or greater opportunities for dropout to occur, etc. Indeed, this is the fundamental limitation of the Van Etten and Taylor meta-analysis. By computing within-group effect sizes, rather than between group effect sized, they "invented" comparisons that don't exist in the primary literature (e.g., EMDR vs. medication). However, the validity of such a comparison is predicated on the assumption that the studies are otherwise comparable. But there is no reason to think this assumption has been met.
Replies:
| Behavior OnLine Home Page | Disclaimer |
Copyright © 1996-2004 Behavior OnLine, Inc. All rights reserved.
