Emotional Awareness and Expression Therapy vs Cognitive Behavioral Therapy for Chronic Pain in Older Veterans: A Randomized Clinical Trial

Chronic pain affects many older adults and is a risk factor for cognitive decline and premature death. Medical and surgical interventions, particularly opioids, often provide only partial efficacy and pose risks including polypharmacy, drug-drug interactions, and falls. Psychological interventions may be safer, but standard options such as cognitive-behavioral therapy (CBT) provide only small effect size benefits. Veterans have high rates of severe pain and overlapping psychiatric conditions (depression, anxiety, PTSD), which may further limit CBT effectiveness. Adverse childhood experiences, military combat, racism or discrimination, and psychiatric conditions are associated with presence and severity of chronic pain but are not directly addressed by CBT. Emotional awareness and expression therapy (EAET) was developed to target these drivers of chronic pain and has demonstrated efficacy in several trials, including some showing superiority to CBT, though most samples were largely white, female, and younger. Testing in an older, diverse population is needed. This trial evaluated whether group-based EAET is superior to CBT among a racially and ethnically diverse cohort of older veterans on pain severity (primary outcome) and secondary mood and functional outcomes from baseline to posttreatment (week 10) and to 6 months posttreatment, and assessed whether baseline depression, anxiety, and PTSD symptoms moderate pain reduction.

Prior research indicates that CBT yields modest benefits for chronic pain in adults, with smaller effects in complex cases with comorbid depression and anxiety. Multiple factors—including adverse childhood experiences, trauma, and discrimination—are linked to chronic pain severity yet are not directly targeted by CBT. EAET was designed to address trauma- and emotion-related drivers of pain by facilitating emotional processing and expression, with controlled and uncontrolled trials demonstrating efficacy across conditions such as fibromyalgia, irritable bowel syndrome, medically unexplained symptoms, and somatic symptom disorder. Two trials have shown some superiority of EAET to CBT, including a preliminary randomized comparison in older veterans and a large fibromyalgia trial showing EAET advantages on several secondary outcomes. However, comprehensive evaluation in racially and ethnically diverse, older, predominantly male populations has been limited, motivating the current full-scale randomized clinical trial.

Design: Two-arm randomized clinical trial conducted at the US Department of Veterans Affairs Greater Los Angeles Healthcare System (VA GLA). Enrollment: May 16, 2019–February 1, 2023; final follow-up: September 14, 2023. The study followed CONSORT guidelines and was IRB-approved and preregistered (NCT03918642). A COVID-19 administrative hold (March 20, 2020–February 4, 2021) paused the trial; a small number of consented patients not yet randomized were transitioned to a separate single-arm EAET telehealth pilot. Participants: Veterans aged 60–95 years with at least 3 months of musculoskeletal pain (including back, leg, pelvic/groin, neck/whiplash, TMJ disorder, fibromyalgia, tension headache). Exclusion: pain secondary to primary disorders (eg, cancer, autoimmune disease, sickle cell disease, neuralgia, burn, infection, cauda equina, gout, migraine/cluster headache, radiography-confirmed hip/knee osteoarthritis, radiculopathy, EMG-confirmed carpal/tarsal tunnel), uncontrolled psychotic or severe mood disorder, severe alcohol/substance use disorder, MMSE ≤24, current pain-related litigation, currently receiving EAET or CBT clinically, CBT participation in past 3 months, not fluent in English, or plan to move within 6 months. Race/ethnicity was self-reported per NIH categories. Randomization and Blinding: Independent statistician generated 1:1 computer randomization in blocks of 16. Investigators/coordinators were blinded to allocation until each cohort of 16 was ready for randomization. Participants were randomized after baseline assessment and informed of assignment at first session; participants were blinded to study hypotheses throughout. Interventions: Both conditions included one 90-minute individual session plus eight 90-minute group sessions (mean group size 6.3). Ten EAET groups and ten CBT groups ran concurrently; all sessions were in person at VA GLA. Fidelity: Both interventions presented as effective, used written manuals with comparable structure and homework. Therapists were nested within condition and supervised by experts; sessions were video recorded and 25% rated by independent fidelity monitor, with high fidelity. EAET: Based on manuals adapted for this population. Conceptual model: unprocessed trauma- and stress-related emotions alter brain pathways for pain and emotion, exacerbating pain. Goals: recognize the influence of avoided emotions (grief, fear, rage, guilt) on pain, and face, process, express, and resolve such emotions to reduce/eliminate pain. Individual session: introduce model and engage in emotional processing (experience emotions in body, verbal expression, adaptive release). Group sessions: psychoeducation on pain/emotions, emotional processing, social disclosure/validation. Homework: identify stress-symptom connections, expressive writing, practice adaptive emotional communication. CBT: VA CBT for Chronic Pain manual with minor timing adaptations to match session structure. Model: chronic pain leads to negative thoughts, feelings, and behaviors that worsen pain/function. Goals: modify maladaptive cognitions/behaviors and learn adaptive coping skills. Individual session: pain/health history, introduction to CBT skills. Group sessions: psychoeducation on coping skills (eg, progressive muscle relaxation, guided imagery), experiential practice, and homework with worksheets. Outcomes: Self-report measures at baseline, posttreatment (week 10), and 6-month follow-up. Primary outcome: Brief Pain Inventory (BPI) pain severity (mean of 4 items, 0–10). Pain response levels: ≥30%, ≥50%, and >70% reduction from baseline. Secondary/exploratory: PROMIS Anxiety, Depression, Fatigue, General Life Satisfaction (NIH Toolbox), Pain Interference, Sleep Disturbance Short Forms; PGIC (post and 6-month); Satisfaction with Therapy and Therapist Scale-Revised (post only). PCL-5 added at start of group 3 per Data Monitoring Committee. Sample Size: Empirical power simulation assuming mean difference −0.70, SD 1.6, baseline-post correlation 0.6; mixed-effects linear regression, two-sided α=.05; target n=120 (60 per arm) for 80% power. Planned consent n=160 (25% attrition); actual consent n=175 due to additional discontinuations during COVID hold. Statistical Analysis: Intention-to-treat including all randomized participants. Mixed-effects linear regression for continuous outcomes with random intercepts for group number and patient, nested time random slope, and clustered robust SEs for treatment group. Fixed effects: treatment, time (categorical, baseline referent), and time×treatment interaction. Satisfaction outcomes: mixed-effects linear regressions with group random intercept and clustered robust SEs; fixed effect: treatment. Prespecified secondary analyses: mixed-effects logistic regression for pain response levels with group random intercept and nested patient random intercept; fixed effects: treatment, time, and interaction; marginal estimates for within-time, between-condition comparisons. Moderation: mixed-effects models with three-way interaction (treatment×time×baseline depression or anxiety), using binary (above/below median) and continuous moderators; post hoc moderation by baseline PTSD in groups 3–10. Analyses in Stata MP 18.0; two-sided P<.05 considered significant.

Participants: 126 randomized (EAET 66; CBT 60); mean age 71.9 years; 92% male. Posttreatment completion 88%; 6-month follow-up 82%. Primary outcome (BPI pain severity): EAET achieved greater reduction than CBT at posttreatment (difference estimate −1.59; 95% CI −2.35 to −0.83; P<.001) and at 6 months (difference estimate −1.01; 95% CI −1.78 to −0.24; P=.01). Within-condition changes: EAET −2.18 (95% CI −3.00 to −1.37) at post; −1.26 (95% CI −1.83 to −0.68) at 6 months. CBT −0.60 (95% CI −1.13 to −0.06) at post; −0.25 (95% CI −0.72 to 0.23) at 6 months. Responder analyses: At posttreatment, ≥30% pain reduction: 63% EAET vs 17% CBT (OR 21.54; 95% CI 4.66–99.56; P<.001). ≥50% pain reduction: 35% EAET vs 7% CBT (OR 11.77; 95% CI 2.38–58.25; P=.002). At 6 months, ≥30% reduction: 41% EAET vs 14% CBT (OR 7.24; 95% CI 1.74–30.06; P=.006). ≥50% reduction at 6 months: 17% EAET vs 4% CBT (OR 6.58; 95% CI 0.99–43.67; P=.05). ≥70% reduction at post: 13% EAET vs 2% CBT (OR 13.93; 95% CI 0.86–225.44; P=.06). Secondary/exploratory outcomes (baseline to posttreatment difference favoring EAET): Anxiety (estimate −2.49; 95% CI −4.30 to −0.68; P=.006); Depression (−3.06; 95% CI −5.88 to −0.25; P=.03); General Life Satisfaction (+1.23; 95% CI 0.36–2.10; P=.005); PTSD symptoms (−4.39; 95% CI −8.44 to −0.34; P=.03); PGIC (+1.46; 95% CI 0.77–2.15; P<.001); Global satisfaction (+0.28; 95% CI 0.12–0.45; P<.001). At 6 months, significant differences favored EAET for Depression (−2.39; 95% CI −4.33 to −0.45; P=.02) and PGIC (+1.24; 95% CI 0.62–1.86; P<.001). Moderation: Higher baseline depression, anxiety, and PTSD symptoms were associated with greater pain severity reduction after EAET but not CBT. Interaction (difference-in-DID) estimates: Depression −1.55 (95% CI −2.73 to −0.37; P=.01); Anxiety −1.53 (95% CI −2.19 to −0.88; P<.001); PTSD −1.69 (95% CI −2.96 to −0.42; P=.009). Using continuous depression moderator, a 1-SD increase in baseline depression yielded greater EAET vs CBT pain reduction at posttreatment (estimate −0.83; 95% CI −1.32 to −0.35; P<.001).

EAET demonstrated superiority over CBT in reducing pain severity among older veterans with chronic musculoskeletal pain at both posttreatment and 6-month follow-up. EAET also yielded greater improvements across multiple secondary domains at posttreatment—anxiety, depression, life satisfaction, PTSD symptoms, patient global impression of change, and global satisfaction—and sustained advantages for depression and PGIC at 6 months. Moderation analyses indicated that patients with higher baseline depression, anxiety, or PTSD symptoms experienced particularly robust pain reductions after EAET, addressing a clinical gap where CBT often underperforms in the presence of such comorbidities. These findings support conceptual models positing that changing beliefs about pain and engaging in targeted processing of trauma-related emotions can substantially reduce chronic pain. The superiority of EAET suggests that integrating its principles—emotional processing, expression, and resolution—into mainstream pain management may improve outcomes, especially for medically and psychiatrically complex patients.

In a racially and ethnically diverse cohort of older veterans, time-limited, group-based EAET was superior to CBT for reducing pain severity and improving multiple psychosocial outcomes at posttreatment, with sustained benefits for pain severity (vs baseline), depression, and PGIC at 6 months. EAET was particularly effective for patients with higher baseline depression, anxiety, and PTSD symptoms, indicating its suitability for complex clinical populations. Incorporating EAET principles into clinical pain medicine may help reduce the societal burden of chronic pain. Future research should evaluate generalizability to younger and nonveteran populations, telehealth delivery, and effectiveness in patients with cognitive impairment.

Generalizability may be limited due to focus on older veterans; replication in younger and nonveteran populations is needed. The sample comprised mostly men and had a high prevalence of PTSD, which may have influenced responsiveness to EAET; the trial was underpowered to formally test these subgroup effects. Therapist backgrounds were not exactly matched between conditions. All sessions were delivered in person; effectiveness via telehealth was not evaluated within this trial. Patients with cognitive impairment were excluded, limiting applicability to that group.