Sixty-one patients were referred to the study. Of these, 33 were ineligible, 6 declined to participate, 1 was administratively removed from the SCDMH day program, and one was unreachable after the initial contact. One patient was expelled from her day-treatment program for potential criminal behavior. Thus, we recruited 20 patients who met inclusion/exclusion criteria for the study and form our intent-to-treat sample. Based on chart review, every participant had a diagnosis of either schizophrenia or schizoaffective disorder. On structured clinical psychiatric interviews (i.e., CAPS and MINI) all of our sample met criteria for PTSD and a psychotic disorder, 70% met criteria for current depression, 20% for bipolar disorder, 65% for panic disorder, 30% for agoraphobia, 25% for social phobia, 15% for obsessive compulsive disorder, and 20% for generalized anxiety disorder. Lifetime trauma exposure from the TAA revealed that 45% of the sample reported a serious accident, 60% reported child sexual abuse (CSA) before the age of 13, 55% reported CSA before the age of 18, 50% reported a sexual assault in adulthood, 45% reported a physical assault with a weapon, and 70% reported a physical assault without a weapon. Most patients had experienced multiple traumatic events across categories.
Of the 20 participants who started treatment, 13 completed treatment (i.e., at least 70% of sessions) for a completion rate of 65%. With one exception, there were no statistically significant differences between completers and non-completers (see ). The one exception was gender: completers were more likely to be female than male, 92.3% versus 7.7%, χ2 = 5.93, p = .03. However, this gender difference, may also represent a site effect in that all males in the study were recruited at one clinical program. A site effect may be due to the fact that those participating in a more intensive day-hospital program, with a regular schedule and coordinated transportation, are more likely to complete treatment than those in a more traditional outpatient program. With the exception of NAI scores, there were no statistically significant differences between completers and non-completers on clinical outcomes, treatment satisfaction, or treatment credibility scores. Treatment completers had higher (i.e., more severe) anger scores at pre-treatment than those who dropped out.
Using all patients enrolled for the trial (n = 20), there were no statistically significant differences in average session attendance or average homework compliance by initial CAPS total scores, initial treatment expectancy total scores, race, or gender. There were also no statistically significant differences between any of the individual treatment expectancy items by gender or race. Average number of sessions attended among those who dropped out of treatment was 7.86 (sd = 5.52; range 1-15 sessions).
Using all patients enrolled in the trial, there were no statistically significant gender or race differences on the treatment expectancy total score or any of the individual items. Total treatment expectancy scores ranged from 4 to 40 (m = 30.29, sd = 9.15) and the individual items ranged from 1 to 10 [mean (sd) for item 1 = 8.00 (2.52), mean (sd) for item 2 = 6.83 (2.88), mean (sd) for item 3 = 8.44 (2.43), and mean for item 4 = 7.28 (2.72)]. Twenty percent (n = 4) stated that they would “definitely” recommend the program to a friend or family member and 80% (n = 16) stated they would “probably” recommend the program to a friend or family member, with no statistically significant differences between completers and non-completers on this item, 3.85 (sd = .38) versus 3.71 (sd = .18). t = .68, p = .51.
Of those patients who completed the treatment (n = 13), none (0%) had a job change from pre-assessment to any of the post-assessments, none (0%) had a change in his/her marital status, 4 (30.8%) were hospitalized on a psychiatric unit, 4 (30.8%) were hospitalized on a medical unit, 0 (0%) were arrested for legal difficulties, and 5 (38.5%) changed their place of residence.
3.1. Efficacy Analyses
In the first analysis set (i.e., n = 13; those who completed the treatment), CAPS total, CAPS cluster D, PCL, NAI, and CPOSS satisfaction change scores were significantly improved from pre- to post-treatment (). From pre- to 3-months, CAPS total, CAPS B, C, and D cluster, PCL, NAI, SF-36 mental health, and CPOSS satisfaction change scores were significantly improved. From post- to 3-months, CAPS B cluster and SF-36 mental health change scores were significantly improved. Average patient self-ratings of treatment improvement in this group were 4.23 (sd = .60) (“a little better”) at post-treatment and 3.85 (sd = .90) (between “no change” and “a little better”) at 3-months. There was a statistically significant difference from pre to post-treatment in patients’ self report of the quality of their social relationships, with improved scores at post-treatment. Patients also endorsed fewer primary care visits from the pre-assessment to 3 months. Results from Analysis sets 2 and 3 were quite similar to those for Analysis set 1.
Estimated Mean Clinical and Process Outcomes Change at Pre-Treatment, Post-Treatment and Three-Month Follow-Up (Treatment Com pleters Only, n=13)
Despite the illness burden of this small sample, preliminary findings from this project indicate that the treatment is efficacious. At 3-month follow-up 10 of 13 patients no longer met criteria for PTSD, and 10 of 13 were considered treatment responders, with at least a 15-point decrease in CAPS scores. These two methods for interpreting progress are important to consider together since even incremental decreases in CAPS scores can at times result in a lost diagnosis due to the symptom cluster scoring criteria. When taken together, however, it is worth noting that only one patient failed to either lose his/her diagnosis and/or not meet criteria for a treatment responder. In combination, these data suggest that individuals with SMI and PTSD are able to tolerate exposure based interventions, and most importantly, can benefit from them.
Although the sample size for minority patients (n = 6 completers) was typically too small to yield statistically significant changes, there were significant CAPS total (change score = 27.00, std error = 9.89, t = 2.73, p = .04), CAPS cluster C (change score = 10.33, std error = 3.32, t = 3.11, p = .03), PCL (change score = 15.83, std error = 3.76, t = 4.21, p = .01), NAI (change score = 21.83, std error = 3.44, t = 6.35, p = .00), and SF-36 mental health (change score = -11.68, std error = 2.02, t = -5.78, p = .01) change scores from pre- to 3-months. There were also statistically significant NAI change scores in this group from pre- to post (change score = 21.67, std error = 4.96, t = 4.36, p = .01) and SF-36 mental health change scores from post to 3-months (change score = -9.43, std error = 1.56, t = -6.05, p = .01). The majority of the other outcomes demonstrated a strong trend in the anticipated direction.
With regard to PCL scores collected at every session, we conducted paired t-tests from session 1 to session 14 (beginning of treatment to end of the group component), from Session 1 to session 22 (beginning of treatment to end of treatment), and from session 15 to session 22 (beginning of exposure component to end of treatment). Paired t-tests revealed significant PCL symptom improvement from session 1 to session 22, t = 3.32(12), p = .006, from session 15 to session 22, t = 2.28(12), p = .042, but not from session 1 to session 14, t = 1.71(12), p = .114. Consistent with the downward trend of PCL means over time as shown in , the longitudinal trajectory of individual PCL change scores across 22 sessions of the intervention indicates a significant improvement over time as suggested by a statistically significant negative slope (time coefficient = -0.50, p = 0.0400, SAS Proc Mixed). Taken together, these data suggest that the most significant patient gains were made at the onset of the treatment (during the education and relaxation components; sessions 1 through 4) and from the latter part of the treatment program (during the latter stages of the exposure component).
Posttraumatic Stress Checklist (PCL) Observed Scores for Completers (n = 13) by Session