Psychosocial Rehabilitation After Moral Injury and Loss With Adaptive Disclosure

The investigators conducted a multi-site randomized trial comparing Adaptive Disclosure for Moral Injury and Loss (AD-MIL; an expanded version of Adaptive Disclosure) for war-related PTSD stemming from Moral Injury (MI) and traumatic loss (TL) to Present Centered Therapy (PCT; Frost et al., 2014). The primary endpoint was psychosocial functioning. The investigators had six hypotheses:

A: Functional and behavioral changes:

A.1. Immediately post-treatment, and 3- and 6-months post-treatment, Veterans with PTSD randomized to AD-MIL will have greater reductions in social, educational, and occupational disability.

A.2. COVID-19 aim. In the extended time period for the trial, at the post-treatment, or post-treatment and 3-months post-treatment intervals (depending on when the participant was randomized), Veterans with PTSD randomized to AD-MIL will have greater reductions in social, educational and occupational disability (the investigators also explored whether there would be greater PTSD and depression symptom reduction in this interval).

B: Mental health changes:

B.3. AD-MIL will lead to greater reductions in PTSD symptom severity and a smaller percentage of PTSD cases.

B.4. AD-MIL will lead to greater reductions in depressive symptoms. B.5. AD-MIL will lead to greater reductions in shame and guilt B.6. AD-MIL will lead to greater reductions in psychological distress.

Exploratory hypotheses: AD-MIL will lead to greater reductions in aggressive behaviors, suicidal ideation, and alcohol use, and greater increases in compassion for the self and others and social connectedness.

BACKGROUND PTSD is a functionally disabling condition among war Veterans. Approximately 20% of Veterans of post-9-11 operations have clinically significant PTSD and suffer from a variety of co-morbid mental and physical health conditions. They also have extensive functional impairments, aggressive and risky behaviors, and reduced quality of life. Although considerable gains have been made in the VA's dissemination of PTSD treatments that are highly effective with civilian trauma, these therapies have been shown to work considerably less well for war trauma. The investigators have argued that this is partly due to a lack of attention to the military culture and warrior ethos and the unique harms of war trauma, namely, moral injury (MI) and traumatic loss (TL). In addition, PTSD treatments have failed to demonstrate an impact on functioning and quality of life, problems that are no less impacted by the warzone trauma. Instead, symptom change is typically the sole metric of success, and functional deficits are rarely considered. The investigators argue that PTSD symptoms should be conceptualized and targeted as part of the fabric of the whole Veteran and his or her context. The investigators aimed to fill a care-gap in the VA by creating an evidence-based treatment for war-related PTSD stemming from MI and TL, focusing on improving psychosocial functioning. The investigators extended Adaptive Disclosure to treat MI and TL (AD-MIL) by: (1) facilitating emotional-processing of TL and MIs and a healing and action plan by letter writing tasks to the lost person or, depending on the nature of the MI, the person or persons harmed by transgressive acts (MI stemming from personal actions or the failure to act) or the person or persons who were the transgressors (MI stemming from being the victim of others' transgressions or bearing witness to grave transgressive acts); (2) skills training and behavioral contracting to improve functioning and targeting MI- and TL-related psychological and behavioral obstacles to habilitative engagements in occupational, relationship, and family roles; and (3) teaching self- and other-compassion and mindfulness. If found to be effective, AD-MIL will fill a care-gap in the VA, reduce PTSD patients' suffering, and help Veterans reclaim or establish positive relationships, work roles, and self-care routines.

METHOD Overview: The VA sites were Boston, Minneapolis, San Francisco, San Diego, and Central Texas. The coordinating/lead site was VA Boston led by the study PI, Brett Litz, PhD. The trial followed the consensus recommendations for clinical trials in the VA (VA ORD, 2008): (1) clearly defined target symptoms: functional and clinical outcomes were operationalized; (2) reliable and valid measures: assessment tools were selected for their content relevance and psychometric properties; (3) use of blind evaluators of outcome: the evaluator was independent and blind to treatment condition. This assessor reminded participants to help maintain their blind; (4) assessor training: the independent evaluator was carefully trained to criteria and monitored on an ongoing basis; (5) manualized, replicable, specific treatment programs; (6) there was unbiased assignment to treatment arms and (7) treatment adherence: sessions were be recorded, and a random percentage was used to assess treatment integrity. Adherence to the therapy manuals was monitored by senior supervisors. The investigators followed the CONSORT guidelines for randomization and participant tracking.

Participants: The trial required 148 Veterans (including women and members of diverse ethnic and racial groups) with PTSD as a result of military trauma.

Recruitment: Veterans were recruited and treated at the Minneapolis, San Diego, San Francisco, and Waco VAs. Co-Investigators (Co-Is): (a) provided materials describing the nature of the study and the target populations sought, distributing said materials via formal (e.g., staff meetings) and informal (e.g., bulletin boards) channels; (b) attended clinical staff meetings; (c) gave talks to describe various treatments in staff grand rounds and other contexts (e.g., to trainees); and (d) provided feedback to staff about referred patients.

Assessor Training and Adherence: A co-investigator, Dr. Matt Gray (University of Wyoming) trained the assessors prior to beginning enrollment. Training included reading and viewing training materials, observation of CAPS-5 administration, and supervised administration of at least three CAPS-5s. Dr. Gray has expertise in CAPS-5 training and fidelity monitoring. The assessor was considered trained on CAPS-5 when they matched Dr. Gray on three interviews. To establish matching, Dr. Gray co-rated interviews conducted by the assessor. A match occurred when the assessor and Dr. Gray agreed on the diagnosis and were within 2 severity points on all of the symptom clusters (PTSD criteria B, C, D, and E).

All assessments were audiotaped to ensure that a standardized approach was used across patients (provided that the participant consents). Dr. Gray reviewed a random 10% of the assessments. All recordings were stored on a restricted-access encrypted drive. Selected recordings were transported to Dr. Gray via Federal Express that allows tracking.

Random Assignment: Veterans were assigned to PCT or AD-MIL based on a randomized permuted block scheme, blocked by gender and minority status. Block size for gender and minority status was based on the distribution of these variables at each site. The Boston site used constrained randomization (i.e., biased coin design) if unexpected imbalance arose in gender and minority distribution across treatment groups.

Treatment Fidelity Monitoring: Half-time therapists with Ph.Ds. in clinical or counseling psychology and VA internship experiences treating Veterans with PTSD were trained to deliver AD-MIL and PCT. Training involved a review of the respective manuals and supporting materials, intensive one-on-one training and supervision, and weekly and ad hoc supervision (Dr. Litz for AD-MIL; Dr. Harris for PCT). All sessions were audiotaped. Two random session recordings from a random 20% of the cases were rated to ensure fidelity to each treatment approach.

ANALYSES Data Analysis: The longitudinal and clustered nature of the design produced a multilevel or nested data structure (Raudenbush \& Bryk, 2002), with a likelihood of between-subject variability in treatment trajectory over time (e.g., random slopes). In this study, Veterans and therapists were nested (clustered) within performance sites. The lower level (level-1) data consisted of the repeated measures for each individual at each assessment. Level-1 data is nested within upper level (level-2) person-level variables (e.g., study site).

Using SAS 9.4 and R/R Studio, the investigators explored the change trajectory of endpoints over the course of pre-treatment, during-treatment, and post-treatment time intervals. Specifically, the investigators used the linear mixed model regression framework to assess the differential treatment effects of endpoint symptom burden over time. The investigators examined observed measurement scores and specified linear mixed models that best fit the time-change trends (e.g., linear, piecewise). The investigators tested for the presence of random variation, such as between-site cluster effects and between-subject random slopes of treatment effect. The investigators determined the best fit linear mixed effects model as a combination of most appropriate representation of time and empirically-evidenced random and fixed effects.

The linear mixed effects models incorporated clinically relevant time-invariant and time-varying predictors and used Maximum Likelihood Estimation to produce estimates that were unbiased in the presence of random missingness and dropout. These provided an Intent-To-Treat (ITT) analysis that incorporated all available data. Per-protocol and "completer" subset analyses were performed under the same paradigm. Additionally, a subset of subjects who began therapy during the COVID-19 pandemic were identified and analyzed separately; they comprised subjects which began therapy following December 31st, 2020. The tenability of the "At-Random" missingness assumption was assessed and a sensitivity analysis of the results was conducted to determine robustness of the inferences.

Clinical significance: The investigators used a variation of the two-stage Jacobson \& Truax (J\&T; 1991) method to establish a reasonable cutoff between dysfunctional and functional scores. The investigators first established the cutoff A, defined as the point 2 SDs beyond the range of the pre-therapy mean (cutoff A = M(baseline) - 2 SD(baseline)). Next, the investigators generated a reliable change index (RCI) for each participant to ensure that changes were not due to artifact of measurement error ([x2 - x1]/Sdiff where x1 represents the participant's pre-treatment total score, x2 represents the participant's post-treatment or follow-up total score, and Sdiff is the standard error of difference between the two test scores. Sdiff was calculated from the test-retest reliability of measures. An RCI larger than 1.96 reflects real change. Individuals were classified as recovered (passed both cutoff A and RC criteria), improved (passed RC criterion but not cutoff A), unchanged (passed neither criterion), or deteriorated (passed RC criterion but symptom scores increased) for each follow-up interval. Chi-square analyses were used to compare proportions per arm at each follow-up.