Factors associated with the outcomes of a novel virtual reality therapy for military veterans with PTSD: Theory development using a mixed methods analysis

The study addresses why some military veterans with PTSD do or do not benefit from novel trauma-focused therapy. PTSD is prevalent and disabling, with evidence for trauma-focused CBT and EMDR, yet many veterans do not engage, drop out, or fail to respond. There is limited understanding of person, intervention, and contextual factors that influence treatment trajectories. Recognizing psychological therapies as complex interventions delivered within complex systems, the paper aims to develop and apply a theory-informed model to examine how characteristics of the person, the 3MDR intervention, and the context interrelate and associate with distinct outcome typologies among veterans undergoing 3MDR.

Prior work supports trauma-focused therapies for PTSD but highlights variable responses and high non-engagement/dropout in veterans. EMDR is not routinely recommended for combat-related trauma in some guidelines due to insufficient evidence. Systematic reviews note limited knowledge of facilitators/barriers to progress and call for examining personal, social, and contextual factors. 3MDR, developed for combat-related PTSD, leverages VR immersion, walking toward trauma reminders, multi-sensory cues (images, music), and dual-attention tasks to tax working memory, aiming to promote reconsolidation. Related VR therapies show promise across mental and neurological conditions. Process evaluation frameworks for complex interventions (e.g., MRC guidance) and realist approaches underscore the need to consider complexity, variation, and interconnectedness. Emerging 3MDR evidence suggests effectiveness and acceptability, but mechanisms and critical ingredients remain unclear, motivating granular, theory-driven analysis across person–intervention–context domains.

Design and source trial: Data derive from a phase II randomized controlled trial of 3MDR for male UK military veterans with treatment-resistant PTSD (ISRCTN80028105). Forty-two participants were randomized to immediate or 12-week delayed treatment. Ethical approval was obtained; all participants consented. 3MDR protocol: Following preparation, participants received six weekly sessions plus one concluding session, delivered in a Motek GRAIL VR lab with treadmill and motion capture. A Zephyr BioHarness recorded heart rate and breathing rate. Sessions included participant-selected music (trauma-associated at start; grounding at end) and seven trauma-associated images per session. For each image, participants verbalized thoughts/feelings (displayed on-screen), performed a dual-attention task tracking a moving red ball and naming numbers, then rated SUD (0–10). Measures: Primary outcome was PTSD severity (CAPS-5). Additional instruments included LEC-5 (traumatic exposures), PHQ-9 (depression), GAD-7 (anxiety), MSPSS (perceived social support), WSAS (functional impairment), and EQ-5D-5L (quality of life). Nested qualitative process evaluation: 11 veterans were interviewed post-therapy for experiences and reflections; 10 with therapy exposure were included in this analysis. Mixed methods analysis: A four-step convergent approach was used. Step 1: collect quantitative and qualitative data. Step 2: separate analyses—clinical outcomes summarized at pre-therapy and post-therapy; physiological summaries (breathing rate, heart rate, walking pace); content analysis of 757 in-session words/phrases into emotional categories and valence; full transcription and within-case qualitative summaries in NVivo. Step 3: integrate analyses into a person–intervention–context table for each participant. Step 4: interpret convergence/divergence patterns across participants grouped into three outcome typologies (dramatic, moderate, minimal improvement) based on CAPS-5 change and movement across severity bands. Some proxies were used (e.g., breathing rate as a proxy for cognitive engagement/working memory load), and session number was fixed by protocol.

- Three outcome response typologies emerged among 10 analyzed participants: dramatic improvement (n=4: P1, P4, P9, P12), moderate improvement (n=4: P8, P10, P18, P26), minimal improvement (n=2: P2, P25). Typology assignment used CAPS-5 changes and severity-band transitions during active therapy. - Person domain: • Mobility capacity differentiated groups; only dramatic improvers reported no mobility problems at baseline EQ-5D-5L, suggesting capacity to walk continuously (45–60 minutes) may facilitate engagement/exposure. • Commitment/ability to complete therapy: all dramatic improvers completed (one ended early due to transformative benefit), whereas moderate improvers included one crisis dropout and others experiencing exhaustion or contemplating discontinuation. • SUD trajectories: dramatic improvers often showed substantial SUD reductions (two to absent), while moderate/minimal groups had variable or persistently high SUDs. • Cognitive/emotional language: across groups, in-session language was predominantly negative valence; anxiety, anger, and depression commonly expressed (≈80–90% negative; dramatic ~80% negative, moderate ~90% negative, minimal ~70% negative per group summaries). - Intervention domain: • Image selection/use: dramatic improvers sourced highly evocative images and replenished selections to sustain elicitation; moderate improvers’ images sometimes lost impact or left them feeling “stuck”; minimal improvers struggled with ordering or focused on a single key image. • Therapeutic alliance: trust and familiarity with therapists mattered. The crisis dropout (moderate group) returned to a usual therapist, underscoring alliance importance. • Dosing preferences: all three moderate completers wished for additional sessions; one dramatic improver also would have continued. Minimal improvers preferred predefined or tailored dosing rather than more of the same. • Psychophysiology/engagement: all groups walked slowly (low physical demand). Very rapid breathing responses occurred across sessions and participants, potentially reflecting cognitive/emotional engagement to trauma reminders. Heart rate increases were modest and sometimes near resting, weaker than expected given the context. - Context domain: • Clinic environment: dramatic improvers commonly reported strong initial reactions to the technology/setting; some moderate responders did as well. Minimal responders included one familiar with the lab environment and one distressed by the broader hospital setting. • Travel/logistics: mixed modes; some described post-session travel as a “daze.” No typology-specific pattern. • Social support and functioning: MSPSS varied (low/medium/high across groups) without typology-specific patterning. High functional impairment (WSAS) was common; dramatic group showed more secondary improvements in WSAS, PHQ-9, GAD-7, and MSPSS compared to minimal responders, whose secondary outcomes often worsened; moderate responders had mixed secondary outcomes. - Non-patterned factors: No clear typology differences in sociodemographics, trauma types, baseline comorbidity (depression/anxiety common in all groups), or overall quality of life domains beyond mobility. - Sample characteristics: All 10 had severe to moderate PTSD at baseline, extensive trauma exposure, and prior trauma-focused treatment without loss of PTSD diagnosis.

Findings support a theory-informed model where outcomes reflect interactions among person, intervention, and context. The capacity to walk continuously, sustain engagement, and tolerate immersion appears linked to better outcomes, consistent with 3MDR’s theoretical emphasis on walking toward trauma reminders and VR-based presence. Intervention components—particularly the ability to select and sequence evocative images across sessions, the quality of therapeutic alliance, and flexible dosing—were associated with outcome typologies. Contextual responses to the VR clinic environment may modulate engagement; dramatic responders were often impressed by the setting, whereas minimal responders included those either over-familiar or distressed by the broader hospital environment. Secondary outcomes clustered with primary outcomes, bolstering the typology distinctions. Some factors showed no clear patterning (e.g., social support levels, employment/finance), underscoring individual variability and complexity. Psychophysiological data suggested high breathing rates despite slow walking may reflect cognitive/emotional engagement, while heart rate changes were smaller than expected. The study advances a generalizable person–intervention–context model for analyzing complex psychological therapies and provides granular insights specific to 3MDR’s mechanisms and delivery considerations.

This theory-driven mixed methods analysis identifies patterned associations between person, intervention, and context factors and outcome typologies in 3MDR for veterans with treatment-resistant PTSD. Practical implications include assessing mobility, facilitating effective image selection and sequencing, prioritizing therapeutic alliance, and adopting flexible, tailored dosing to avoid feeling “stuck.” The model presented offers a structure for future personalized intervention research and evaluation. Future studies should prospectively select measures aligned with the model, incorporate direct metrics of cognitive engagement and therapeutic alliance, gather expectations pre-therapy, expand psychosocial and physiological profiling, and test adaptive dosing within 3MDR.

The analysis is based on a purposive subsample (n=10) of a larger trial, limiting generalizability and statistical inference. Some secondary outcome data were missing. Several constructs used proxies (e.g., breathing rate as a proxy for cognitive engagement/working memory load). The fixed session number in the parent trial constrained evaluation of individualized dosing. Causal attributions cannot be made; patterns are exploratory and hypothesis-generating.