study outcome, and neither study included overall PTSD outcome measures (Boudewyns and Hyer, 1990; Chemtob et al., 199714). Falsetti et al. (2003) was excluded because it is an additional analysis of Falsetti et al. (2001) that does not include PTSD outcome data (although it includes other data for the complete sample, unlike the 2001 publication, which was preliminary). See Tables 4-1, 4-2, and 4-3 for a summary of included studies.
The committee identified a diverse literature of 10 randomized trials of EMDR compared with various other therapies and wait list or alone compared with wait-list control. The mean age in these studies was in the 30s to the 40s (with a wider range for civilian studies, typically including participants from age 18 to the 70s, and a narrower range for studies in veterans, generally of the Vietnam War). The sex of participants varied in a pattern similar to that described in Chapter 3—in four studies where the trauma was combat, most or all participants were male; participants in the two studies with sexual assault/abuse victims were all female, and participants with a variety of trauma types included a mix of men and women. Approximately half of the studies provided race/ethnicity data, with the range of white participants from 54 to 68 percent. Most studies reported duration of PTSD diagnosis or exposure to index trauma with a range from approximately 1 year in a study of occupational witnessing man-under-train accidents to two decades in the case of veterans. Treatment length ranged from 2 sessions to 10 weekly sessions, and duration of sessions was generally 90 minutes. Most studies provided information about therapists administering the treatment, and they typically were reported as being licensed, trained at master’s level or above, and having received EMDR training (some had level II training). Most therapists also were supervised. Some studies did not conduct follow-up after the completion of treatment, while others conducted follow-up at 3, 6, 12, or 15 months.
Six trials had major limitations such as lack of assessor blinding or independence, high dropout rates, or weak (or no) treatment of missing values (Boudewyns et al., 1993; Jensen, 1994; Marcus et al., 1997; Power et al., 2002; Rothbaum, 1997; Silver et al., 1995). Four studies had few or no major limitations, and of those, two showed statistically significant improvement in CAPS score or a significant difference in loss of diagnosis in the treated group (Carlson et al., 1998; Hogberg et al., 2007; Rothbaum et al., 2005; van der Kolk et al., 2007). The study by Carlson and colleagues was a small trial in male veterans, and it showed no effect post-treatment. The study by van der Kolk and colleagues was an RCT comparing EMDR,