Disseminating EMDR

Sandra:

I am have quite mixed feelings about the success of EMDR. Given that it has been shown to be an effective treatment for PTSD, I certainly would prefer that a loved one with PTSD be treated with EMDR than with, say, Thought-Field-Therapy. But this is an outcome. What about the process of dissemination? EMDR was aggressively marketed on claims that it was successful in treating PTSD, it was superior to other treatments, and it has special unique features that require special training. To date, only the first of these three claims is empirically justifiable and even that one was not justifiable until 1995 at the earliest, and more reasonably 1997. Yet by 1995, over 10,000 therapists had shelled out their money to get training.

I've made some bold claims here , so I accept my burden of proof and I lay out the evidence for my claims below. I review the published treatment studies by year of publication and discuss relevance for conclusions about efficacy and mechanism and juxtapose it with estimates of the number of therapists trained in EMDR. Now, one can argue that year of publication is conservative because it can take a long time to get through the publication process and that the results were known through conferences much earlier. While this is true, conference presentations do not permit the kind of careful scrutiny and widea vailability to the general population of clinical scientists for review. Thus, I think publication and professional debate is an important aspect of the validation process. If my reconstruction of history is accurate, then EMDR was disseminated between 1989 and 1997 on the basis of claims that were not justified, and even today is being disseminated on the basis of claims that, while being researched, are not yet resolved. If this is true, why would I be happy about the dissemination of EMDR? Would you be happy if someone re-packaged EMDR, gave a new name and added a few procedureal twists, conducted a study to show it beat waitlist, and then went on to market it by claiming it was better than EMDR and, despite the similarity, worked through different mechanisms? Would you find this a bit underhanded?

Shapiro, 1989
I hope that we can agree that this study has little probative value because of the numerous and significant methodological limitations, including: (1) no formal diagnosis at pretreatment to assure a relatively homogenous population; (2) all evaluations were conducted by the therapist (i.e., not by a blinded evaluator); (3) did not use reliable and validated measures to assess outcome; (4) use of single therapist; (5) no evidence that therapy sessions were ever evaluated by someone other than the therapist to determine whether the treatment was implemented as planned. Despite these limitations, Shapiro made some pretty bold statements about the efficacy of EMDR (e.g., complete desensitization of a single trauma memory in one 60 minute session for 80-100% of people) and about the role of eye movements in EMDR (eye movements were crucial to outcome).

Boudewyns et al., 1993
Dismantling study with veterans comparing two sessions of EMDR with 2 sessions a no-eye movement condition that duplicated all procedures of EMDR except there were no eye movement and a treatment-as-usual condition. SUDs in session reduced for EMDR but not the no-eye movement condition and therapists rated more EMDR subjects as "treatment responders." However, in a posttreatment assessment, there were no differences between any of the three groups on standardized measures. AND, there were no differences between groups on SUDs in response to the very memory targeted in treatment.

So, by 1993 there were two studies that found changes on SUDs obtained in session, but the one study that included standardized measures found no differences on these measures and in the psychophysiological assessment at the end of treatment, there was no evidence that treatment had altered the veteran's response to the target memory. Despite this, Shapiro in a letter published in the Behavior Therapist claimed that 4,000 people had already been trained in EMDR.

Jensen, 1994
Another study of veterans comparing 2 sessions of EMDR with treatment as usual. Once again, there was a decrease on SUDs for the treatment group compared to the control group, but there was no difference between groups on standardized measures. These researchers had several of their therapy tapes rated by an EMDR expert who thought they terminated treatment sessions prematurely.

Renfrey & Spates, 1994
A dismantling study that compared standard EMDR with a group that had all components of EMDR but the eye movements were induced by alternating flashing lights and a third condition that had all components of EMDR except eye movements. The study included several standardized measures, but oddly, only reported on SUDs, VoC, and heart rate. These measures showed equivalent improvement across groups. I contacted Richard Spates back in 1999 who informed me that the same pattern of results was observed for the standardized measures. It is important to note that, because there were no differences between groups in conjunction with the lack of a waitlist control, we cannot rule out the possibility that the observed improvements would have occurred in the absence of treatment.

Vaughan et al., 1994
Four conditions: EMDR, Image Habituation Training (IHT), and relaxation with waitlist (WL). The three treatments were combined and compared to WL and found that treatment was superior. They then conducted several analyses group X time analyses to determine if there were any differential effects of the three treatments. None of the interactions were significant. Despite this, they went and conducted a bunch of post-hoc analyses on the Impact of Event Scale and on the re-experiencing symptoms from their diagnostic measure. They found a hodge-podge of results, with EMDR being superior to one or the other alternative treatment. These analsyes, however were not justified given the lack of a significant interaction. Plus, the analyses on re-experiencing symptoms from the diagnostic measures selected, without any theoretical justification, 2 out of five symptoms and the each symptom is measures with a single item. Single items are notoriously unreliable, further questioning the meaning of the results. Overall, there were no credible results from this study to justify concluding EMDR was superior to any of the control treatments.

D. Wilson et al., 1995
Another dismantling study that compared EMDR with another condition that replaced eye movements with alternating thumb tapping and an eyes fixed condition. The y found SUDs reduction and changes in skin conductance during treatment for EMDR but not the other conditions. However, they didn't bother to assess PTSD prior to or after treatment and they did not use any reliable and valid measures to assess outcome.

S. Wilson et al., 1995
The first study to be both relatively rigorous in its methodology and to produce unequivocal evidence for the efficacy of EMDR. About half of the subjects in this study were reported to have PTSD at the outset and the follow-up study (S. Wilson et al., 1997) reports that 84% of those who had PTSD at the beginning of the study did not have it 15 months after treatment. However, there is a very serious problem with the PTSD incidence data that has never been adequately discussed in the literature. S. Wilson et al. (1995) never reported the outcome of PTSD incidence for acute treatment. In a footnote, the state the following:

"The items on the PTSD-I [PSD Interview, the measure of PTSD used in this study] are worded in a way that makes them unsuitable for measuring change in symptoms. For example, many of the items ask if respondent has 'ever' experienced a PTSD symptom…In an attempt to measure changes in symptoms as a result of treatment, we reworded the items to assess occurrence of symptoms 'within the last 7 days'…This revision of the PTSD-I was used all measurement times after T1 [pretreatment]. The rewording resulted in a significant drop in the percentage of participants who were diagnosed as having PTSD at T2 for participants in the delayed-EMDR condition [i.e., after the WL period was over]."

Just to be clear, they used a measure that assess life time PTSD to determine if people had PTSD at the beginning, and then used a measure of current PTSD at the end of treatment. They found that a "significant", but unspecified percent of people who were diagnosed with lifetime PTSD at pretreatment did not have current PTSD after 4 weeks of WL. There are [at least] three possible explanations for this: (1) changing the wording of the measure invalidated the instrument, which is Wilson et al.'s explanation; (2) lots of people with PTSD no longer had PTSD after a four week WL period, which is not very likely given that subjects did not change much on other reliable measures of psychopathology; or (3) many of the people diagnosed with PTSD at pretreatment had PTSD at an earlier point in their life, but did not have it at the time of the pretreatment assessment, which was revealed when they changed the time frame from "ever" to the "last seven days."

If number 1 is true, then we cannot trust the follow-up data reporting 84% recovery from PTSD because they used an invalid measure. If number 2 is true, this is an unusual population, as other studies find that few people actually improve enough on WL in order to lose the diagnosis of PTSD. If number three is true, less than 50% of their subjects actually had PTSD to begin with AND the 84% "cure rate" is an overestimate because it is attributing to treatment a reduction that is in fact due to changing from a lifetime assessment to a current assessment. By the way, lifetime prevalence rates are ALWAYS lower than current because the fact of the matter is that for every trauma population studied thus far, most people recover without intervention.

So, in 1995 we see our first reasonably well-done study that included at least some people with PTSD (up to 50%, probably "significantly less" but due to the authors not reporting how many subjects in each group lost the diagnosis after acute treatment, we cannot estimate how many) and showed unequivocally significant improvement on several standardized measures of psychological functioning. This is the same year Shapiro published her treatment manual. On the dust jacket of the book, it is reported that more than 10,000 people had already been trained in EMDR.

Pitman et al., 1996
Dismantling study with veterans. Did not include a waitlist condition. There were no differences between standard EMDR and an eyes fixed condition. Results showed some, but quite limited, improvement but the lack of a WL group precludes attributing even those modest changes to EMDR per se. Interestingly, they did find a positive correlation between treatment fidelity and outcome. However, this was equally true for both conditions, so the results provide no evidence for a contribution of eye movements.

Marcus et al., 1997
A study that compared EMDR delivered by two study therapists with "treatment as usual" as delivered by regular care providers in the context of an HMO. EMDR was superior to treatment as usual. However, this study confounded treatment type with treatment provider, as different people provided the different treatments. Moreover, treatment as usual was a hodge-podge of interventions including medication, group therapy, and numerous kinds of individual therapy. Thus, we don't know exactly what EMDR was superior to. In fact, it is logically possible that some subjects in treatment as usual would have received treatment that was more effective than EMDR but, because most people in treatment as usual did not get that particular treatment, there was no evidence for this. Thus, the study was actually somewhat biased towards finding an effect of EMDR by the fact that you were comparing a homogenous treatment with a non-homogenous alternative.

Rothbaum, 1997
This study met very rigorous design standards. The only two limitations were the small sample size and the fact that a single therapist administered the treatment. Thus, the excellent results could be attributed to Rothbaum's general clinical skills, rather than to EMDR. Treatment was clinically and statistically superior to waitlist. Importantly, all subjects had PTSD at the outset. This study, therefore, replicated and extended the results of S. Wilson et al.

In my humble opinion, 1997 is when we can say with some confidence that EMDR works better than WL for PTSD. There are four studies (S. Wilson, Vaughan et al., Marcus et al., and Rothbaum) that are reasonably well done, not perfect, but no single study will ever be perfect. It is enough of a pattern to convince me that there is something there, but no good evidence that eye movements are an effective ingredient. Given that EMDR has some exposure, then it is most parsimonious to assume (i.e., not proven) that the effective ingredients are (1) the common factors associated with most forms of psychotherapy and with the placebo effect (e.g., talking to a knowledgeable and kind professional, the belief that the therapist has something useful to offer, etc.) and (2) repeated exposure to trauma-related thoughts and memories.

So, EMDR was marketed quite effectively to more than 10,000 therapists before there was adequate research that was available for public review demonstrating its efficacy. Moreover, it was marketed as being better than other available treatments even though there was no credible evidence to support this claim, and it was marketed with the idea there was something special about eye movements or perhaps laterally alternating stimuli or dual attention again without any credible evidence.

If you go to the EMDR website, you will see the following quote from the Washington post in 1995: "EMDR (Eye Movement Desensitization and Reprocessing) therapy has emerged as a procedure to be reckoned with in psychology....Almost a million people have been treated ....Also, further research appears to support the remarkable claims made for EMDR therapy."

Now, I ask you, given the state of the research in 1995, I ask you what evidence was there at that time the EMDR was "a procedure to be reckoned with"? Moreover, what were the "remarkable claims" that were being supported by research at that point? Surely, the findings that EMDR was superior to WL, about the same as IHT or relaxation, and no evidence for a role of eye movements should not be called "remarkable." However, claims that EMDR was superior to other forms of cognitive behavior therapy (CBT) or that it operated through some special mechanisms would indeed be remarkable, but what research available in 1995 supported such claims? We still have, at the beginning of 2003, no credible evidence that eye movements contribute to outcome of EMDR for PTSD and we are only beginning to see the publication of studies comparing EMDR with other validated treatments, and the studies have produced conflicting results:

Devilly & Spence 1999 - CBT was superior to EMDR, but there are problems with the randomization in this study making conclusions weak.

Lee et al., 2002 - EMDR was superior to exposure therapy plus cognitive restructuring. However, the authors put far more work into insuring and assessing fidelity of EMDR than CBT.

Ironson et al., 2002 - EMDR plus in vivo exposure instructions was equivalent to exposure therapy. There are claims that there was evidence of superiority for EMDR in terms of greater efficiency. Specifically, the report that more people in the EMDR group had 70% reduction in PTSD severity (measured by the PSS-SR) than in exposure therapy. However, I suggest that it is a statistical artifact.

The amount of change after three prep sessions and three treatment sessions was nearly identical for the two treatments. However, the percent change in EMDR was greater. Mathematically, the only way for this to be true is if the denominator of the percent ratio (the pretreatment score in this case) for EMDR was smaller than for exposure therapy. Specifically, the average pretreatment PSS-SR score for exposure was 34.56, the average posttreatment score was 15.78, resulting in an 18.78 point change, which translates to an average 54.3 percent reduction. In EMDR, the pretreatment score was 26.58 and the posttreatment score was 9.1, for a mean reduction of 17.48 points, which is numerically SMALLER than the reduction for exposure therapy, but it translates to a larger average percent change (65.8%). Therefore, I don't believe their conclusions about efficiency are warranted and that both treatments were roughly equivalent.

Powers et al., 2002 - Compared EMDR with the combination of exposure therapy and cognitive restructuring. There were no significant differences between groups, but there were numerical differences in favor of EMDR.

There are two more studies that have been completed (Taylor et al. and Rothbaum et al.) that have been reported at conferences and are being prepared for publication. Because the results are not yet widely available and that details of results can change in the process of review and publication, I will not comment on these studies at this time.

Bottom line is that the data are mixed and there are two more studies whose results need to be scrutinized before we can reasonably advance any data-based conclusions about the relative efficacy of EMDR and exposure therapy. Also, EMDR has not been compared with cognitive therapy by itself nor with stress inoculation training, so we don't know how EMDR compares with these alternatives. And despite the "conclusion" from the Van Etten and Taylor meta-analysis that EMDR is as good as SRI medication, there is not a single published study comparing the efficacy of medication of any sort with psychotherapy of any sort for PTSD.

Replies:

There are no replies to this message.

| Behavior OnLine Home Page | Disclaimer |

Copyright © 1996-2004 Behavior OnLine, Inc. All rights reserved.