Tai Chi and Wellness for Gulf War Illness: Combined Analysis of In-Person and Remote Randomized Controlled Trials

Abstract

Background:

Veterans with Gulf War Illness (GWI) experience a wide range of physical, neurological, cognitive, and mood complications that can lead to distressing psychological symptoms. Despite its debilitating impact on physical and psychological functioning, effective treatment options for GWI remain limited. Tai Chi, an ancient Chinese mind-body practice, may be efficacious in reducing GWI symptoms and associated mental health challenges. The present study combined data from two randomized controlled trials (RCTs) examining a Tai Chi group intervention relative to a Wellness comparison group intervention. Primary outcomes were GWI symptoms (health symptoms, pain interference, fatigue, and depressed mood), with secondary outcomes including anxiety, stress, sleep disturbances, mindfulness, health-related quality of life, and PTSD symptoms. We also evaluated whether treatment delivery modality (in-person versus remote) influenced outcomes.

Methods:

Participants were 114 veterans with GWI and chronic pain. They completed self-report measures at baseline, posttreatment, 3-month follow-up, and 9-month follow-up.

Results:

Preliminary multilevel modeling indicated that treatment modality (in-person versus remote) did not significantly influence outcomes. Both Tai Chi and Wellness interventions were associated with improvements in health symptoms, pain interference, depression, anxiety, mindfulness, mental and physical health-related quality of life, and PTSD symptoms.

Conclusions:

Tai Chi and Wellness group interventions may demonstrate similar efficacy in improving GWI symptoms and mental health outcomes among veterans with GWI and chronic pain. These interventions can be delivered in-person and remotely with comparable therapeutic benefits, offering flexible and accessible treatment options for this underserved population.

Introduction

Since the 1990-1991 Persian Gulf War, many of the deployed U.S. military personnel have reported persistent and debilitating symptoms of Gulf War Illness (GWI), including musculoskeletal pain, fatigue, and mood-cognition problems (Fukuda et al., 1998; Steele, 2000). Longitudinal studies indicate that these symptoms often endure and amplify over time (Krengel et al., 2022; Yee et al., 2020), contributing to poor health-related quality of life (Gifford et al., 2022). Additionally, veterans with GWI experience a broad range of symptoms that frequently lead to psychological comorbidities (Engdahl et al., 2018; Zundel et al., 2019). Gulf War veterans, especially those with severe GWI, demonstrate high rates of depression, anxiety, and Posttraumatic Stress Disorder (PTSD) (Black et al., 2004; Gifford et al., 2022), along with elevated levels of daily psychological stress (Botz-Zapp et al., 2025). These factors likely exacerbate sleep disturbances, which are also common in this population (Gifford et al., 2022).

Despite the significant functional impairments related to GWI, treatment options remain limited (Kaimal & Dieterich-Hartwell, 2020). As noted in a systematic review, the strength of evidence is low to moderate for existing GWI interventions, such as cognitive behavioral therapy, exercise, and mindfulness-based stress reduction (Nugent et al., 2021). Because GWI symptoms tend to remain stable or worsen over time, additional research to strengthen the evidence supporting efficacious interventions for GWI symptoms and associated mental health problems is necessary (Chester et al., 2019).

Mindfulness, defined as purposeful attention and awareness of present-moment experiences, is negatively related to psychological stress with moderate strength and may be particularly relevant for veterans with GWI (Brown & Ryan, 2003; Kearney et al., 2016). Evidence indicates that mindfulness is inversely associated with core GWI symptoms, including pain, fatigue, and dysphoric mood (Ikeuchi et al., 2020; Nigol & Di Benedetto, 2020; Westphal et al., 2021). Despite scant evidence, there is an increasing body of literature supporting mindfulness-enhancing interventions for Veterans with GWI, demonstrating reductions in pain, fatigue, cognitive problems, and depressive symptoms (Breneman et al., 2023; Kearney et al., 2016; Nakamura et al., 2017). While these interventions hold promise, more rigorous studies are needed to substantiate their efficacy and to investigate novel mind-body approaches for treating GWI.

Tai Chi is an ancient Chinese mind-body practice that integrates sequences of flowing movements, stretching, balance training, diaphragmatic breathing, and mindful relaxation. These components work synergistically to improve health (Niles et al., 2022; Wang et al., 2004). A growing body of evidence indicates that Tai Chi confers benefits for individuals with chronic health conditions, improving both physical and psychological symptom burden and quality of life (Reid et al., 2019; Wang et al., 2004). Preliminary studies also demonstrate that Tai Chi may reduce GWI symptoms as well as stress, anxiety, and PTSD symptoms (Abbott & Lavretsky, 2013; Niles et al., 2022; 2024). Therefore, studies investigating the salutary effect of Tai Chi on GWI symptoms and mental health outcomes are warranted.

To address this gap, we conducted a randomized controlled trial (RCT) with a parallel-group superiority design to compare the efficacy of Tai Chi and Wellness group interventions in veterans with GWI. The onset of the COVID-19 pandemic necessitated a transition from in-person delivery to synchronous remote (videoconference) delivery. Initially conceptualized as two related but distinct trials, we previously reported findings from the in-person cohort (Niles et al., 2024). Findings indicated that (1) Tai Chi was associated with greater reductions in depressed mood compared to Wellness, (2) both interventions were related to similar reductions in pain interference, and (3) there were no significant changes in fatigue (Niles et al., 2024).

Because many of the same psychometric instruments were administered across both delivery modes, we elected to combine these data. The present study aims to replicate and extend our prior findings by evaluating the effects of Tai Chi versus Wellness on GWI symptoms (health symptoms, pain interference, fatigue, and depressed mood) and associated mental health outcomes (anxiety, stress, sleep disturbances, mindfulness, health-related quality of life, and PTSD symptoms). We hypothesized that Tai Chi would be associated with greater improvements in these outcomes relative to the Wellness comparison condition. We further explored potential differences between in-person and remote delivery; we considered these analyses exploratory and expected no mode-specific effects.

Method

Participants

The sample consisted of 114 veterans deployed to the Persian Gulf region in 1990-1991 for Operation Desert Shield and/or Operation Desert Storm and reported symptoms consistent with GWI. Participant demographic characteristics are presented by treatment in Supplementary Table 1 and by modality in Supplementary Table 2. Based on established cutoff scores (reported in Measures below), 59.6% (n=68) of the sample reported clinically elevated PTSD symptoms and 60.5% (n=69) endorsed clinical insomnia.

Inclusion criteria were: (1) veteran of the 1990-1991 Gulf War; (2) met GWI criteria of at least one symptom in two or more of the following domains: (a) musculoskeletal pain, (b) fatigue, and (c) mood-cognition difficulties; (3) one GWI symptom must be musculoskeletal or joint pain or stiffness pain for at least 6 months; (4) spoke and understood English; and (5) able to attend scheduled group intervention sessions. For the remote trial, access to a computer or tablet for telehealth appointments was added as an inclusion criterion. Exclusion criteria were: (1) lacked capacity to provide informed consent; (2) severe medical, psychiatric, or neurological condition or brain injury that could interfere with safe engagement in study procedures; (3) psychotropic or pain medication change in past month; (4) currently engaged in practicing any mind-body intervention, such as Tai Chi, Yoga, Qigong, or meditation, for at least 3 hours per week for longer than 3 months; (5) trouble standing for 60 minutes; and (6) current participation in another treatment study that could confound findings. For the remote participants, inappropriate behavior toward staff or other participants and unresponsive to boundary-setting was added as an exclusion criterion.

Procedure

The present study was conducted at the VA Boston Healthcare System in Boston, MA, USA, and the facility’s Institutional Review Board (IRB) approved all study procedures. The original RCT was registered and updated on ClinicalTrials.gov (Identifier: NCT02661997; Registration Date: January 25, 2016). Detailed descriptions of the in-person and remote RCT protocols, including recruitment, randomization, and safety procedures, are available elsewhere (Niles et al., 2023; 2024). Figure 1 displays the CONSORT diagram summarizing the combined participant flow across both trials. Recruitment strategies for the in-person RCT included pamphlets, flyers, newsletters, mailings, clinician referrals, and a voluntary recruitment database. For the remote RCT, study materials such as flyers, brochures, and web links were disseminated through veteran social media platforms.

Figure 1.

CONSORT Chart. This figure presents the combined participant flow that includes both the in-person and remote randomized controlled trials.

Study staff conducted telephone screenings to provide an overview of the trial and assess preliminary eligibility. Electronic medical records were then reviewed to confirm additional criteria, after which eligible veterans were scheduled for baseline assessments. A total of 12 cohorts (6 in person and 6 remote) were recruited. At the baseline visit, informed consent was obtained either face-to-face for in-person participants or via Webex for remote participants, who subsequently returned signed consent documents by mail. Following confirmation of eligibility and completion of baseline assessments, participants were randomized in a 1:1 ratio to receive either the Tai Chi (n=58) or Wellness (n=56) intervention using an online randomization tool (Random.org). Both interventions were delivered via groups twice weekly over 12 weeks (24 sessions total). In total, 53 veterans participated in person and 61 participated remotely. Participants completed psychometric assessments at baseline, 12 weeks (post-intervention), 24 weeks (approximately 3 months post-intervention) and 48 weeks (approximately 9 months postintervention). For in-person participants, outcome assessors who administered the self-report measures were blinded. Remote participants completed the surveys independently and returned them by mail. Data analysts were also blinded. Other study staff were unblinded to track attendance, dropout, and safety issues.

Measures

Health Symptom Checklist (HSC)

The HSC is a 34-item self-report inventory that measures health symptoms related to GWI (Yee et al., 2020). Originally adapted from Bartone and colleagues (Bartone et al., 1989), the version administered in the present study has been used in longitudinal studies tracking GWI symptoms over time (Krengel et al., 2022; Yee et al., 2020). Consistent with previous Gulf War research, a total score was calculated by summing all the items that were endorsed as occurring at least weekly. Internal consistencies ranged from α=.92 (baseline) to α=.94 (9-month follow-up) in the present study.

Brief Pain Inventory-Short Form (BPI)

The BPI is a 9-item self-report measure that assesses pain and its impact on daily functioning in the past 24 hours (Cleeland & Ryan, 1994). We used the pain interference subscale (BPI-Pain Interference) that gauges the extent to which pain disrupts various domains of life. Research supports the validity and reliability of the BPI (Cleeland & Ryan, 1994; Ferreira et al., 2023). Internal consistencies ranged from α=.93 (baseline) to α=.96 (post-intervention and 3-month follow-up) in the present study.

Multidimensional Fatigue Inventory (MFI)

The MFI is a 20-item self-report survey that indexes various dimensions of fatigue, including general fatigue, physical fatigue, reduced motivation, reduced activity, and mental fatigue (Smets et al., 1995). As a global index of fatigue, we used the MFI’s general fatigue subscale (MFI-General Fatigue) (Smets et al., 1995). The MFI possesses good psychometric properties (Lin et al., 2009; Smets et al., 1995). Internal consistencies ranged from α=.70 (baseline) to α=.82 (3-month follow-up) in the present study.

Depression Anxiety Stress Scales (DASS)

The DASS is a 21-item self-report questionnaire that measures state levels of depression, anxiety, and stress (Lovibond & Lovibond, 1995). We used the three DASS subscales of depression (DASS-Depression), anxiety (DASS-Anxiety), and stress (DASS-Stress). The DASS possesses excellent validity, reliability, and sensitivity to change (Lovibond & Lovibond, 1995; Ronk et al., 2013). Internal consistencies ranged from α=.84 (DASS-Anxiety at baseline) to α=.95 (DASS-Depression at 9-month follow-up) in the present study.

Insomnia Severity Index (ISI)

The ISI is a 7-item self-report measure that assesses perceived sleep difficulties (Bastien et al., 2001). The ISI exhibits good validity, reliability, and sensitivity to change (Bastien et al., 2001; Manzar et al., 2021). A cutoff score of 15 was used to determine clinical insomnia (Morin & Espie, 2003). Internal consistencies ranged from α=.90 (3-month follow-up) to α=.93 (9-month follow-up) in the present study.

Mindful Attention Awareness Scale (MAAS)

The 15-item MAAS is a self-report survey that measures mindfulness with respect to self-regulation and well-being (Brown & Ryan, 2003). The MAAS possesses strong psychometric properties (Brown & Ryan, 2003; Osman et al., 2016). Internal consistencies ranged from α=.92 (baseline) to α=.94 (post-intervention and 3-month follow-up) in the present study.

Patient-Reported Outcomes Measurement Information System Scale v1.2-Global Health (PROMIS-GH)

The 10-item PROMIS-GH is a self-report questionnaire that indexes health-related quality of life across the following domains: general health, overall quality of life, physical health, mental health, satisfaction with social activities, physical function, pain intensity, fatigue, ability to carry out social activities, and emotional problems (Hays et al., 2009). We used the Global Physical Health (PROMIS-GPH) and Global Mental Health (PROMIS-GMH) subscales. The PROMIS-GH and its subscales demonstrate good validity and reliability (Chan et al., 2025; Hays et al., 2009). Internal consistencies ranged from α=.76 (PROMIS-GPH at baseline) to α=.90 (PROMIS-GMH at 9-month follow-up) in the present study.

PTSD Checklist for DSM-5 (PCL-5)

The 20-item PCL-5 is a gold-standard self-report measure of PTSD symptoms as specified in the DSM-5 (Weathers et al., 2013). Substantial research supports the PCL-5’s validity and reliability in veterans (Forkus et al., 2023). A cutoff score of 33 was used to determine clinically elevated PTSD symptoms (Bovin et al., 2016). Internal consistencies ranged from α=.96 (baseline, post-intervention, and 3-month follow-up) to α=.97 (9-month follow-up) in the present study.

Interventions

Full descriptions of the interventions under study have been provided elsewhere (Niles et al., 2023; 2024). In brief, the Tai Chi group intervention was based on other Yang-style Tai Chi programs examined in prior RCTs (Wang et al., 2016; 2018). Tai Chi sessions included: (1) warm up and review of Tai Chi principles, (2) Tai Chi movements with meditation, (3) breathing practice, and (4) relaxation. The Wellness comparison group intervention aligned with the Veterans Affairs’ (VA) Whole Health approach that addresses the following domains: physical activity, mind-body connection, spirituality, interpersonal relationships, sleep and relaxation, healthy eating and substance use, personal development, and surroundings (U.S. Department of Veterans Affairs, 2025). Each Wellness intervention session focused on one of these domains by (1) reviewing relevant written materials and video clips, (2) setting, refining, and reviewing Specific, Measurable, Attainable, Realistic, and Timely (SMART) goals, and (3) engaging in a mindfulness exercise. Participants in both interventions were encouraged to continue practicing their skills using treatment manuals, videos (Tai Chi), and goal-tracking practices (Wellness) even after the interventions concluded. There were no treatment-related adverse events.

Data Analyses

A two-sided power analysis determined a sample size of 120 to be of adequate power (85%) to detect a Cohen’s d effect size of 0.62 given an alpha level of .05 and estimated 20% withdrawal rate by the primary endpoint of 12 weeks (see Niles et al., 2024, for details). Due to disruptions associated with the COVID-19 pandemic, 114 participants were ultimately included in the analyses.

We conducted analyses in R version 4.5 using an intent-to-treat approach with all available data at each timepoint (R Core Team, 2023). We calculated mean scores for each outcome variable, except for the HSC, which used a total score. Participant scores were considered missing if they did not complete at least 90% of a scale. We addressed missing data using listwise deletion. We computed the raw means and standard deviations for outcomes with significant changes across timepoints (see Table 1). For the primary analyses, we conducted random effects mixed (i.e., multilevel) models estimated using restricted maximum likelihood with the lme4 package (Bates et al., 2015). These models included random intercepts for participants. We specified time as number of months since baseline (coded: 0=Baseline, 3=Post-Intervention, 6=3-Month Follow-Up, 12=9-Month Follow-Up). Models included linear and quadratic terms (i.e., main effects) for time. They also included a main effect for intervention (coded: 0=Tai Chi, 1=Wellness) and two-way interactions between intervention and each term for time.

Table 1.

Raw Means and Standard Deviations of Outcome Variables that Significantly Changed across Timepoints by Treatment

	Baseline		Post-Intervention		3-Month Follow-Up		9-Month Follow-Up
Outcome	Wellness	Tai Chi	Wellness	Tai Chi	Wellness	Tai Chi	Wellness	Tai Chi
HSC	17.49 (8.20)	19.24 (7.51)	15.27 (8.38)	17.66 (8.71)	14.66 (7.78)	16.67 (8.66)	14.94 (9.26)	17.24 (9.30)
BPI-Pain Interference	4.69 (2.76)	4.50 (2.58)	3.66 (2.62)	3.51 (2.93)	3.83 (2.83)	3.77 (2.88)	4.11 (2.79)	4.01 (2.68)
DASS-Depression	1.01 (.86)	1.18 (.85)	.82 (.74)	.96 (.97)	.85 (.77)	.99 (.93)	.94 (.89)	.94 (.86)
DASS-Anxiety	.83 (.68)	.96 (.69)	.69 (.62)	.85 (.75)	.73 (.62)	.86 (.75)	.81 (.78)	.84 (.78)
MAAS	3.82 (1.10)	3.80 (1.14)	4.18 (1.13)	3.99 (1.24)	4.01 (1.07)	3.87 (1.24)	4.24 (1.10)	4.08 (1.07)
PROMIS-GPH	2.94 (.73)	2.79 (.68)	3.02 (.74)	2.99 (.80)	3.06 (.81)	3.06 (.74)	3.01 (.85)	2.96 (.75)
PROMIS-GMH	2.71 (.89)	2.43 (.93)	2.83 (.82)	2.73 (1.11)	2.88 (.85)	2.70 (1.12)	2.75 (.94)	2.70 (1.06)
PCL-5	1.85 (1.00)	1.91 (.98)	1.60 (.96)	1.71 (1.00)	1.49 (0.92)	1.62 (1.00)	1.59 (1.04)	1.59 (1.09)

Note. Standard deviations are presented in parentheses. Wellness: n = 58. Tai Chi: n = 56.

Abbreviations: BPI = Brief Pain Inventory; DASS = Depression Anxiety Stress Scale; HSC = Health Symptom Checklist; MAAS = Mindfulness Attention Awareness Scale; PCL-5 = PTSD Checklist for DSM-5; PROMIS-GMH = Patient-Reported Outcomes Measurement Information System Scale v1.2- Global Mental Health; PROMIS-GPH = Patient-Reported Outcomes Measurement Information System Scale v1.2- Global Physical Health

Because the effect of treatment might depend on modality (coded: 0=In Person, 1=Remote), we first fitted models with the modality × time × intervention three-way interaction and its component two-way interactions and main effects. After finding that interactions involving modality were non-significant, we conducted final models that accounted for modality only as a covariate (coded: −1=In Person, 1=Remote). Models with the three-way interaction did not include a quadratic term for time or its interaction with intervention to balance model complexity with statistical power. To visualize and describe change over time, we plotted models with significant effects using model-implied trajectories and 95% confidence intervals. To make plots and tables, we used the marginaleffects package and the easystats package collection (Arel-Bundock et al., 2024; Lüdecke et al., 2022).

Results

Health Symptoms

For the model with the HSC as the outcome (see Table 2), there was a significant main effect of linear time, b=−11.42, SE=5.59, p=.042, and all other effects were non-significant (p-values≥.066). The plot of the HSC model demonstrated that both Tai Chi and Wellness were associated with similar reductions in health symptoms over time, with 9-month follow-up scores lower than baseline for both groups. However, between the 3-month and 9-month follow-up period, health symptoms remained constant in Tai Chi but increased in Wellness (see Figure 2).

Table 2.

Results from Multilevel Models with Significant Effects over Time

	Outcome Variable
	HSC	BPI-Pain Interference	DASS- Depression	DASS- Anxiety	MAAS	PROMIS- GPH	PROMIS- GMH	PCL-5
Effect: b(SE)
Intercept	17.76(1.03)***	3.97(.33)***	1.01(.11) ***	.87(.09)***	3.93(.14)***	2.90(.09)***	2.61(.12)***	1.75(12)***
Linear Time	−11.42(5.59)*	−1.31(2.23)	−1.44(.58)*	−1.13(.50)*	2.22(.99)*	.80(.54)	1.23(.60)*	−1.92(.66)**
Quadratic Time	9.80(5.31)	5.32(2.13)*	1.16(55)*	.26(.47)	.47(.94)	−1.17(51)*	−1.62(.57)**	1.03(.63)
Treatment (Wellness)	−2.02(1.46)	.12(.47)	−.10(.15)	−.12(.12)	.09(.20)	.12(.13)	.21(.17)	−11(17)
Modality	1.26(0.73)	.39(.23)	.08(.08)	.04(.06)	−.08(.10)	−.09(.07)	−.12(.09)	−.001(.09)
Linear Time × Treatment (Wellness)	−1.62(7.75)	−.81(3.06)	1.47(.80)	1.11(.69)	−.16(1.36)	−.30(.74)	−1.02(.83)	.28(.91)
Quadratic Time × Treatment (Wellness)	6.69(7.40)	1.98(2.94)	.45(.77)	.94(66)	−1.37(1.30)	.34(.71)	−.22(.79)	.90(.07)

^*p ≤ .05.

^**p ≤ .01.

^***p ≤ .001.

Abbreviations: BPI = Brief Pain Inventory; DASS = Depression Anxiety Stress Scale; HSC = Health Symptom Checklist; MAAS = Mindfulness Attention Awareness Scale; PCL-5 = PTSD Checklist for DSM-5; PROMIS-GMH = Patient-Reported Outcomes Measurement Information System Scale v1.2- Global Mental Health; PROMIS-GPH = Patient-Reported Outcomes Measurement Information System Scale v1.2- Global Physical Health

Figure 2.

Model-Implied Trajectories for Changes in GWI Symptoms. This figure shows the changes in health symptoms, pain interference, and depression over the study period across both treatments. Vertical lines indicate 95% confidence intervals. Abbreviations: BPI = Brief Pain Inventory; DASS = Depression Anxiety Stress Scale; GWI = Gulf War Illness; HSC = Health Symptoms Checklist; time_month = number of months since baseline

Pain Interference

The model with BPI-Pain Interference as the outcome (see Table 2) revealed a significant main effect of quadratic time, b=5.32, SE=2.13, p=.013, but all other effects were non-significant (p-values≥.093). The graph of the BPI-Pain Interference model indicated similar reductions in pain interference over time for both interventions. In Tai Chi, pain interference steadily decreased from baseline to 9-month follow-up. Pain interference declined from baseline to 3-month follow-up but rose from 3-month to 9-month follow-up in Wellness (see Figure 2).

General Fatigue

The model with MFI-General Fatigue as the outcome exhibited all non-significant effects (p-values≥.066).

Depression, Anxiety, and Stress

For the model with DASS-Depression as the outcome (see Table 2), the main effects of linear time, b=−1.44, SE=0.58, p=.014, and quadratic time, b=1.16, SE=0.55, p=.036, were significant. All other effects were non-significant in this model (p-values≥.066). The graph of the DASS-Depression model indicated that both treatments were related to similar reductions in depression from baseline to 3-month follow-up. However, from 3-month to 9-month follow-up, depression increased in both treatments. In Wellness, depression increased to the extent that levels were similar at 9-month follow-up and baseline, whereas depression at 9-month follow-up was lower than at baseline in Tai Chi (see Figure 2).

For the model with DASS-Anxiety as the outcome (see Table 2), there was a significant main effect of linear time, b=−1.13, SE=0.50, p=.024, and all other effects were non-significant (p-values≥.105). The plot of the DASS-Anxiety model revealed similar reductions in anxiety over time for both treatments. Anxiety declined persistently in Tai Chi from baseline to 9-month follow-up. In Wellness, anxiety decreased from baseline to 3-month follow-up, but then increased from 3-month to 9-month follow-up such that levels were similar at 9-month follow-up and baseline (see Figure 3).

Figure 3.

Model-Implied Trajectories for Changes in Mental Health Outcomes. This figure shows the changes in anxiety, mindfulness, physical and mental health-related quality of life, and PTSD symptoms over the study period across both treatments. Vertical lines indicate 95% confidence intervals. Abbreviations: DASS = Depression Anxiety Stress Scale; MAAS = Mindfulness Attention Awareness Scale; PCL-5 = PTSD Checklist for DSM-5; PROMIS = Patient-Reported Outcomes Measurement Information System Scale v1.2; PTSD = Posttraumatic Stress Disorder; time month = number of months since baseline

For the model with DASS-Stress as the outcome, all effects were non-significant (p-values≥.103).

Sleep Disturbances

The model with the ISI as the outcome demonstrated all non-significant effects (p-values≥.102).

Mindfulness

The model with MAAS as the outcome (see Table 2) indicated a significant main effect of linear time, b=2.22, SE=0.99, p=.025, and all other effects were non-significant (p-values≥.296). The graph of the MAAS model demonstrated that both treatments were associated with similar improvements in mindfulness from baseline to 9-month follow-up (see Figure 3).

Health-Related Quality of Life

The model with PROMIS-GPH as the outcome (see Table 2) revealed a significant main effect of quadratic time, b=−1.17, SE=0.51, p=.023, but all other effects were non-significant (p-values≥.137). The graph of the PROMIS-GPH model indicated a similar trajectory of change for both treatments: physical health-related quality life increased from baseline to 3-month follow-up, then declined from 3-month follow-up to 9-month follow-up, though levels remained greater than at baseline (see Figure 3).

For the model with PROMIS-GMH as the outcome (see Table 2), there were significant main effects for both linear time, b=1.23, SE=0.60, p= 041, and quadratic time, b=−1.62, SE=0.57, p=.005. All other effects in this model were non-significant (p-values≥.180). The plot of the PROMIS-GMH revealed mental health-related quality of life similarly improved from baseline to 3-month follow-up for both interventions. Although scores declined somewhat from 3-month follow-up to 9-month follow-up, they did not return to baseline levels and, especially for Tai Chi, improvements were maintained (see Figure 3).

PTSD Symptoms

The model with the PCL-5 as the outcome (see Table 2) demonstrated a significant main effect of linear time, b=−1.92, SE=0.66, p=.004, and all other effects were non-significant (p-values≥.103). The graph of the PCL-5 model revealed that both treatments were related to similar reductions in PTSD symptoms over time. PTSD symptoms steadily decreased from baseline to 9-month follow-up in Tai Chi. In Wellness, there was a steep decline in PTSD symptoms from baseline to 3-month follow-up, then increased from 3-month to 9-month follow-up, though did not return to baseline levels (see Figure 3).

Discussion

This study utilized data from both in-person and remote RCTs to evaluate the comparative efficacy of Tai Chi and Wellness group interventions for GWI symptoms and related mental health outcomes in veterans with GWI and chronic pain. Both interventions were related to significant benefits across multiple outcomes, including health symptoms, pain interference, depression, anxiety, mindfulness, health-related quality of life, and PTSD symptoms. However, Tai Chi did not demonstrate greater improvements relative to Wellness. Importantly, treatment effects did not differ by delivery modality, as evidenced by the absence of significant three-way interactions, thereby supporting the combining of in-person and remote trial data.

Both Tai Chi and Wellness were efficacious in reducing health symptoms, pain interference, and depressive symptoms. Improvements in health symptoms are particularly noteworthy given that prior longitudinal studies have reported stability or worsening of HSC scores over time (Krengel et al., 2022; Yee et al., 2020). The observed reductions in pain interference replicate our earlier findings from the in-person trial (Niles et al., 2024). In contrast to previous results indicating that Tai Chi was associated with greater reduction in depression than Wellness (Niles et al., 2024), the present analysis found comparable effects across both interventions. This discrepancy highlights the need for further understanding the mechanisms underlying symptom improvement. Given that both interventions emphasized mindfulness, they may reduce GWI symptoms through similar pathways, such as promoting mind-body awareness and parasympathetic activation (Reid et al., 2019). Future mechanistic studies are needed to clarify these potential shared and distinct pathways.

Beyond GWI symptoms, both interventions yielded benefits in mental health outcomes, including anxiety, mindfulness, health-related quality of life, and PTSD symptoms. These findings underscore the interplay between psychological distress and GWI (Engdahl et al., 2018; Zundel et al., 2019), and imply that mind-body and health-focused approaches may enhance functioning in this population. Further trials should examine how changes in specific mental health processes, such as mindfulness or pain interference, temporally influence other symptom clusters. For example, mindfulness gains may precede reductions in pain (de França Moreira et al., 2024), whereas improvements in pain interference may facilitate reductions in PTSD symptoms (Lee et al., 2019). Identifying such temporal relations could inform the optimization of integrative interventions for veterans with GWI, a population with persistently low health-related quality of life (Gifford et al., 2022).

Encouragingly, improvements in health symptoms, pain interference, mindfulness, quality of life, and PTSD symptoms were sustained at 9-month follow-up, indicating enduring treatment effects. These findings raise the possibility that holistic interventions may slow or counter the typical progression of GWI symptomatology (Krengel et al., 2022; Yee et al., 2020). However, improvements in depression and anxiety were not maintained, with symptoms returning to baseline by 9 months. This pattern is consistent with prior research on complementary and integrative interventions (Ley & Putz, 2024), and highlights the need to develop strategies to maintain treatment benefits over time, such as booster sessions or ongoing engagement in health-promoting behaviors.

The transition from in-person to remote delivery during the COVID-19 pandemic offered a unique opportunity to examine treatment modality effects. Consistent with our hypothesis, outcomes did not differ between modalities. This finding implies that synchronous telehealth delivery of Tai Chi and Wellness is a viable option for veterans with GWI, a geographically dispersed and aging population (Dursa et al., 2021). Remote interventions may reduce isolation, foster social connection, and enhance accessibility, particularly in areas with limited availability of specialized care (Madigan et al., 2021; Teo et al., 2025).

This study has several strengths, including the comparison of two theoretically grounded interventions, the use of validated GWI common data elements (e.g., HSC, BPI, PCL-5) (Cohen et al., 2022), and a longitudinal analytic approach. Nevertheless, there are some limitations. The sample size was modest and composed primarily of older White male veterans, reflecting the demographic composition of the Gulf War cohort but limiting generalizability. The study may also have been underpowered to detect the modality × time × intervention interactive effects. All outcomes were based on self-report, which may be subject to response bias. Though expected, dropout rates increased from post-intervention to the 9-month follow-up, raising concerns about potential attrition bias. Additionally, the definition of GWI used in this study was intentionally broad and inclusive to maximize external validity and the applicability of the findings. However, this approach may reduce construct validity compared to studies using a narrower definition.

Demographic analyses revealed that a higher proportion of White participants and a lower proportion of Black participants were randomized to Tai Chi compared to Wellness (see Supplementary Table 1), likely reflecting a failure of the randomization process to produce equal groups. Future studies investigating Tai Chi would benefit from strategies aimed at bolstering the recruitment of non-White participants to facilitate evaluation of differences among racial groups in treatment engagement and efficacy. Demographic variables such as age, race, and sex were not modeled as covariates given the preliminary nature of the study. Future trials should consider evaluating their potential moderating effect. Finally, at the follow-up assessments, we did not collect data on continued practice of skills learned in treatment. Future research would benefit from collecting this information to better understand treatment effects.

In conclusion, this trial provides preliminary support for both Tai Chi and Wellness interventions in improving core GWI symptoms and associated mental health outcomes. Benefits were observed across both in-person and remote modalities, underscoring the feasibility of offering these approaches in flexible formats to broaden access. Future studies should investigate mechanisms of change, identify moderators of treatment response, and evaluate strategies to sustain improvements over time. Given the paucity of effective treatments for GWI, these findings represent an important step toward expanding therapeutic options for this underserved veteran population.

Highlights.

1. Tai Chi may reduce symptoms and mental health challenges of Gulf War Illness (GWI)

2. Tai Chi and Wellness reduce health symptoms, pain interference, and depression

3. Anxiety, mindfulness, quality of life, and PTSD symptoms improve in the treatments

4. No differences in outcomes emerged between in-person and remote modalities

5. Tai Chi and Wellness can benefit veterans with GWI, regardless of delivery method

Glossary:

α

Cronbach’s alpha

b

Beta weight

BPI

Brief Pain Inventory-Short Form

COVID-19

Coronavirus Disease 2019

d

Difference

DASS

Depression Anxiety Stress Scales

DSM-5

Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition

GWI

Gulf War Illness

HSC

Health Symptom Checklist

IRB

Institutional Review Board

ISI

Insomnia Severity Index

MA

Massachusetts

MAAS

Mindful Attention Awareness Scale

MFI

Multidimensional Fatigue Inventory

n

Sample size

N

Population size

p

Probability value

PCL-5

PTSD Checklist for DSM-5

PROMIS-GH

Patient-Reported Outcomes Measurement Information System Scale v1.2-Global Health

PROMIS-GMH

PROMIS-Global Mental Health

PROMIS-GPH

PROMIS-Global Physical Health

PTSD

Posttraumatic Stress Disorder

RCT

Randomized Controlled Trial

SD

Standard deviation

SE

Standard error

SMART

Specific, Measurable, Attainable, Realistic, and Timely

time_month

Number of months since baseline

U.S.

United States

USA

United States of America

VA

Veterans Affairs