Complementary and Integrative Health Approaches to Manage Chronic Pain in U.S. Military Populations: Results from a Systematic Review and Meta-Analysis, 1985–2019

Abstract

Objective:

To examine the efficacy of complementary and integrative health (CIH) approaches for reducing pain intensity (primary outcome) and depressive symptoms (secondary outcome) as well as improving physical functioning (secondary outcome) among U.S. military personnel living with chronic pain.

Method:

Studies were retrieved from bibliographic databases, databases of funded research, and reference sections of relevant manuscripts. Studies that (a) evaluated a CIH approach to promote chronic pain management among military personnel, (b) used a randomized controlled trial design, and (c) assessed pain intensity were included. Two coders extracted data from each study and calculated effect sizes. Discrepancies between coders were resolved through discussion.

Results:

Comprehensive searches identified 12 studies (k = 15 interventions) that met inclusion criteria. CIH practices included cognitive-behavioral therapies (k = 5), positive psychology (k = 3), yoga (k = 2), acupuncture (k = 2), mindfulness-based interventions (k = 2), and biofeedback (k = 1). Across these studies, participants who received the intervention reported greater reductions in pain intensity (d+ = 0.44, 95% CI = 0.21–0.67, k = 15) compared to controls. Statistically significant improvements were also observed for physical functioning (d+ = 0.36, 95% CI = 0.11–0.61, k= 11) but not for depressive symptoms (d+ = 0.21, 95% CI = −0.15−0.57, k= 8).

Conclusions:

CIH approaches reduced pain intensity and improved physical functioning. These approaches offer a nonpharmacological, nonsurgical intervention for chronic pain management for military personnel. Future studies should optimize interventions to improve depressive symptoms in military populations experiencing chronic pain.

Impact Statement:

CIH approaches provide an efficacious treatment option for veterans experiencing chronic pain. This meta-analysis provides support for the use of CIH practices to manage chronic pain symptoms with active duty and military veterans experiencing chronic pain.

Chronic pain is most often defined as pain that is present most days or every day for a minimum of 12 weeks, typically beyond the time that would be expected for pain related to an injury to resolve (Dahlhamer et al., 2018). Chronic pain occurs frequently among adults in the U.S., with estimates ranging from 20% to 50%, depending on the frequency and severity in which individuals experience pain (Dahlhamer et al., 2018; Nahin, 2017). U.S. military members experience a higher prevalence of musculoskeletal (e.g., low back, neck, and joint pain) and neurological (e.g., migraine) pain than civilians, with nearly two-thirds reporting chronic pain in the past three months (Fontana & Rosenheck, 2008; Nahin, 2017). Active duty U.S. military members are at a higher risk for physical injuries and mental stressors during their service, despite being healthier (nomothetically) than their civilian counterparts (due to pre-enrollment health screening) (Fontana & Rosenheck, 2008; Nahin, 2017). Furthermore, veterans experience severe pain at a rate 50% greater than non-veterans (9.1% vs. 6.4%), with even higher disparities among younger veterans (i.e., 18–39 years of age) (Fontana & Rosenheck, 2008; Nahin, 2017).

The increased risk for chronic pain among military personnel is related to various experiences and exposures occurring throughout their term of service. First, military members experience high levels of musculoskeletal injury, head and brain trauma (e.g., traumatic brain injury [TBI]), and physical amputations due to combat experiences (Hoot et al., 2018). Second, service members are at heightened risk for mood disorders, trauma-related disorders due to combat and military sexual trauma, suicidality, alcohol, tobacco, or drug use disorders (Hoge et al., 2004; Hoot et al., 2018; Pizarro, Silver, & Prause, 2006; Suris & Lind, 2008). For example, the prevalence of TBI, post-traumatic stress disorder (PTSD), and chronic pain among veterans returning from deployment was 9.6%, 29.4%, and 40.2% respectively (Cifu et al., 2013). Military personnel are also more likely to have higher levels of exposure to hazardous chemicals, radiation, air pollutants, warfare agents and harmful noise or vibrations (US Department of Veteran Affairs, 2015). Toxic exposure among veterans deployed to the Iraq and Afghanistan conflicts is associated with chronic multi-symptom illness (i.e., fatigue, mood disturbances and cognitive difficulties, chronic musculoskeletal pain) after controlling for key confounders. Compared to age-matched non-veterans, veterans are also more likely to have one or more chronic health conditions such as asthma, diabetes, or heart disease (Kramarow & Pastor, 2012; Lehavot, Hoerster, Nelson, Jakupcak, & Simpson, 2012). Therefore, military members have an increased risk for combat-related and other chronic health conditions compared to their civilian counterparts. Chronic pain treatment, however, often relies on opioids which increases the risk of functional impairment among military members (Jonas & Schoomaker, 2014).

Pain management treatment approaches typically aim for a reduction in pain intensity. Chronic pain treatments involve medication (e.g., prescription opioids), therapeutic exercise (e.g., physical therapy), chiropractic, and surgery (National Institues of Health, 2011). Despite the benefits of conventional medical approaches, these interventions do not provide individuals with the skills to manage the cognitive, emotional, and physical challenges associated with living with a chronic pain condition. Furthermore, prescription opioids, once considered appropriate for the management of chronic pain conditions, is now a growing health concern in the military (Nelson, Bjarnadóttir, Wolcott, & Agarwal, 2018). This is especially salient given that the prevalence of opioid use for the treatment of chronic pain is higher among military members (15%) compared to civilians (4%) (Toblin, Quartana, Riviere, Walper, & Hoge, 2014).

In addition to the physical experience, chronic pain is often accompanied by a multitude of maladaptive cognitive and psychological processes. Persistent negative thinking and catastrophizing about one’s chronic pain leads to a cycle of physical and psychological suffering. For example, patients living with chronic pain often report fear of movement (i.e., kinesophobia), which results in decreased physical activity, muscle loss, and paradoxically exacerbates pain and risk of injury (Crombez, Eccleston, Van Damme, Vlaeyen, & Karoly, 2012). Elevated levels of stress associated with chronic pain also increases cortisol and inflammation (Edwards et al., 2008; Quartana et al., 2010). Therefore, it is critical for pain management approaches to address maladaptive stress and coping responses in the context of treatment. Examples of adaptive stress responses include skills for seeking emotional and social support and safely engaging in physical activity. CIH approaches may be helpful for military members living with chronic pain, as they can provide skills to adjust to, accept, and manage ongoing pain (McAllister, 2013) as well as to cope with life stressors associated with pain (e.g., functional limitations).

CIH practices include a wide range of psychological (e.g., cognitive restructuring, problem-solving skills-training, meditation), physical (e.g., progressive muscle relaxation, acupuncture), or blended psychological and physical (e.g., tai chi, yoga) approaches to manage or reduce stress, improve coping, and promote self-management (Lichtenthal, Simpson, & Cruess, 2005). CIH approaches, such as cognitive behavioral therapy, mindfulness, and yoga, have been shown to improve psychological and physical symptoms across multiple patient populations (e.g., patients living with cardiovascular disease, HIV) (Dunne et al., 2019; Scott-Sheldon, Balletto, et al., 2019; Scott-Sheldon, Gathright, et al., 2020; Scott-Sheldon, Kalichman, Carey, & Fielder, 2008). Additionally, small trials of military samples living with PTSD show that CIH approaches reduce PTSD symptoms (Hollifield, Sinclair-Lian, Warner, & Hammerschlag, 2007; Kearney, McDermott, Malte, Martinez, & Simpson, 2012, 2013; Rosenthal et al., 2011; Staples, Hamilton, & Uddo, 2013). Therefore, nonpharmacological, nonsurgical conventional (e.g., cognitive-behavioral therapy) or nonconventional (e.g., mindfulness, yoga) CIH approaches can help military personnel, active duty, reserves, or veterans, manage their pain symptoms as well as the distress associated with living with chronic pain (Madsen, Vaughan, & Koehlmoos, 2017). Additionally, many CIH approaches provide patients with skills to engage in self-management after the active phases of clinical interventions. Self-management is a valuable component given the chronic nature of these pain conditions and high cost of medical care associated with chronic pain.

A scoping review of CIH practices among military personnel showed that meditation and acupuncture were the most commonly evaluated practices (Elwy, Johnston, Bormann, Hull, & Taylor, 2014). While this scoping review provided a comprehensive overview of CIH approaches used to improve health in the military, the benefits of these approaches for military members remain understudied. Therefore, the purpose of this meta-analysis was to examine the efficacy of CIH practices to promote chronic pain management among military personnel. We hypothesized that CIH interventions would lead to greater reductions in pain intensity (primary outcome) as well as improving physical functioning and reducing depressive symptoms (secondary outcomes) among U.S. military personnel living with chronic pain, relative to controls. Furthermore, we also explored whether sample and intervention characteristics moderated the findings. Specifically, we expected that changes in pain intensity would be moderated by (a) sex (i.e., women would be more likely to use and benefit from alternative approaches such as yoga to manage pain symptoms) (Evans et al., 2018), (b) length of time living with chronic pain (i.e., longer durations of chronic pain would be associated with diminished physical performance and disability) (Jones, Rutledge, Jones, Matallana, & Rooks, 2008), (c) psychological distress (i.e., depressive symptoms are prevalent among patients reporting pain-related symptoms in primary care settings (L. S. Williams, Jones, Shen, Robinson, & Kroenke, 2004), and therefore may benefit more from a nonconventional approach to pain management), (d) intervention duration (i.e., improved outcomes in interventions of shorter vs. longer durations) (Van Daele, Hermans, Van Audenhove, & Van den Bergh, 2012), (e) delivery mode (i.e., individual vs. group-based interventions would be more beneficial given that individuals approaches allow for personally-tailored chronic pain management) (Hawkins, Kreuter, Resnicow, Fishbein, & Dijkstra, 2008), and (f) type of CIH approaches (i.e., cognitive-behavioral therapies would be more effective than mind-body approaches) (Skelly et al., 2018).

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed in the reporting of this meta-analysis (Moher, Liberati, Tetzlaff, & Altman, 2009). The PRISMA Checklist can be found in the Online Supplemental Material (Table S1).

Inclusion Criteria, Information Sources, and Search Strategy

Studies were included if they: (1) evaluated a CIH approach to promote chronic pain management among military personnel, including active duty and veterans; (2) used a randomized controlled trial (RCT) design; and (3) assessed pain intensity. CIH approaches included nonpharmacological, nonsurgical conventional (e.g., cognitive-behavioral therapy) or nonconventional (e.g., mindfulness, yoga) approaches to pain management. The studies sampled patients experiencing any type of chronic pain resulting from an injury or accident (e.g., musculoskeletal, traumatic brain injuries) or painful conditions (e.g., rheumatoid arthritis). Three strategies were used to identify studies: First, we searched electronic bibliographic databases (i.e., PubMed, PsycINFO, CINAHL, Embase, ProQuest Dissertations and Theses Full Text, ERIC, Cochrane Library, SocIndex, and Web of Science: Science Citation Index) using a Boolean search strategy and database-specific limiting and expanding fields and terms. For example, the PubMed search string was: [((“complementary and alternative medicine”) OR (autogenic training[MeSH]) OR (biofeedback [MeSH]) OR (“cognitive restructuring”) OR (“cognitive behavioral stress management”) OR (“deep breathing”) OR (“emotional freedom technique”) OR (guided imagery [MeSH]) OR (“mindfulness-based stress reduction”) OR (mindfulness [MeSH]) OR (meditation [MeSH]) OR (“problem-solving training”) OR (“progressive muscle relaxation”) OR (“relaxation techniques”) OR (self-disclosure) OR (self-hypnosis) OR (tai chi [MeSH]) OR (“transcendental meditation”) OR (yoga[MeSH]) OR (“stress management”)) AND ((military [MeSH]) OR (veterans [MeSH]))]. These databases were searched on three separate occasions (February 2017, April 2018, and August 2019) to ensure the retrieval of all studies available through August 2019. Second, we reviewed the reference lists of relevant reviews and other related manuscripts retrieved from our database searches. Finally, we reviewed databases of funded research (NIH RePORTER) and clinical trials (ClinicalTrials.gov).

Study Selection

All unique records (after duplicate removal) retrieved from the electronic bibliographic database searches were screened for inclusion based on title and abstract. Full-text manuscripts of any potentially relevant records were retrieved and reviewed for inclusion by two authors (MLD, JD), and verified by the principal investigator (LAJSS). Records that reported the same study and/or sample across multiple manuscripts were linked in the database and represented as a single unit. The manuscript reporting the most complete data was selected as the primary manuscript; additional papers were considered supplemental but used in the data collection process. One study author was contacted for additional data and provided the requested information (Groessl, Liu, Goodman, et al., 2017).

Data Collection Process, Data Items, and Reliability

Two of three trained independent coders (MLD, JD, BLB) extracted study information (e.g., publication year), sample characteristics (e.g., age, gender), study design (e.g., randomized controlled trial), intervention details (e.g., number of sessions), and intervention technique (e.g., cognitive-behavioral therapy, yoga) using a coding manual and form developed by the principal investigator (LAJSS) and co-investigators, then pilot tested by the research team. The methodological quality (MQ) of each study was determined using 17 items adapted from validated measures (Downs & Black, 1998; Fowkes & Fulton, 1991; Jadad et al., 1996; Miller et al., 1995). The highest possible MQ score was 25, with higher scores reflecting higher MQ. Inter-rater reliability was assessed for study, sample, design, methodological, and intervention characteristics. For categorical variables, coders agreed on 91% of the judgments (mean Cohen’s κ = 0.84). For continuous variables, reliability yielded an average intra-class correlation coefficient (ρ) of 0.85 across categories (median = 0.99). Coders discussed and resolved discrepancies; the principal investigator (LAJSS) resolved any remaining unresolved discrepancies.

Study Outcomes

Our primary outcome was pain intensity; physical functioning and depressive symptoms were secondary outcomes. Pain intensity was assessed using self-report measures that asked individuals to (a) rate their pain on a visual analog scale (VAS) or a numerical rating scale (daily diary) or (b) complete a multi-item pain assessment (e.g., Brief Pain Inventory (Cleeland & Ryan, 1994); McGill Pain Questionnaire (Melzack & Torgerson, 1971)). For the secondary outcomes, all studies used self-report measures to assess physical functioning (e.g., Brief Pain Inventory (Cleeland & Ryan, 1994); Roland-Morris Disability Questionnaire (Roland & Fairbank, 2000)) and depressive symptoms (e.g., Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961)). As is typical in many areas of research, a wide range of measures (or subscales) were used to assess pain. Two authors (ED, LAJSS) reviewed each pain measure in depth to determine the content. For the present meta-analysis, “pain intensity” and “pain severity” represented the pain intensity factor. Similarly, “functional limitations,” “disability,” and “pain interference” represented the physical functioning factor. Due to the limited number of studies measuring additional outcomes of interest, we were unable to assess stress processes (e.g., problem-focused coping), health behaviors (e.g., sleep quality), physiological indicators (e.g., heart rate variability), and other outcomes (e.g., cortisol) in the current meta-analysis.

Summary Measures and Synthesis of Results

Summary effect sizes (d) were calculated based on the standardized mean-change, controlling for baseline (Becker, 1988). Positive effect sizes indicated that participants who received the intervention reported lower pain intensity, improved physical functioning, or fewer depressive symptoms relative to controls. Effect sizes were weighted by sample size to correct for sample size bias (Hedges, 1981). Two independent coders calculated effect sizes. Discrepancies between coders were discussed and corrected. Weighted mean ES (d+) and 95% confidence intervals were calculated using random-effects assumptions using a maximum likelihood approach (Lipsey & Wilson, 2001). Box plots for each outcome identified outliers (Emerson & Strenio, 1983). Heterogeneity was assessed using the Q statistic; the I² index and 95% confidence intervals were calculated to assess the proportion of heterogeneity across the studies (Higgins, Thompson, Deeks, & Altman, 2003; Huedo-Medina, Sanchez-Meca, Marin-Martinez, & Botella, 2006). Moderator analyses were performed with meta-regression and an analog to the ANOVA using a maximum likelihood random-effects model (Hedges, 1994; Lipsey & Wilson, 2001). The analyses were performed using Stata/SE 15.1 using published macros (StataCorp, 2017). Comprehensive Meta-Analysis was used to generate the forest plot for pain intensity (Borenstein, Hedges, Higgins, & Rothstein, 2013).

Risk of Bias Across Studies

Two methods (visual inspection and statistical tests of funnel plots) were used to evaluate the evidence for the presence of small-study effects (Begg & Mazumdar, 1994; Egger, Davey Smith, Schneider, & Minder, 1997; Sterne & Egger, 2001). These tests were conducted only for dependent variables with ≥10 studies (Lau, Ioannidis, Terrin, Schmid, & Olkin, 2006).

RESULTS

Study Selection

Our searches identified 2,909 unique records of which 2,108 records were excluded based on title and abstract review (559 records did not meet any inclusion criteria; 760 records were reviews, commentaries, or editorials; 108 records were protocols; 94 records were qualitative). An additional 587 records were excluded because they did not sample military personnel with chronic pain. Upon review of the full text of the remaining 801 records, an additional 774 records were excluded (see Figure 1 for details). Thus, the final sample included 12 studies and 15 supplemental manuscripts providing additional study information. An overview of the study, sample, and intervention details for the 12 included studies (k = 15 interventions) can be found in Table 1.

Figure 1.

PRISMA Flow Diagram Screening and Selection Process

Table 1.

Study, Sample, and Intervention Characteristics of the 10 Studies (k = 12) included in the Meta-Analysis

Study	Sample	Recruitment/Location	Condition	Current Treatment	Control	Intervention Details	Sessions	Dosea	Outcomes
Appelbaum (1988); Linked: Appelbaum (1986)	N = 18 (39%); 11% F; Mage = 62; Myears DX = 15; 100% veterans	Outpatients referred through rheumatology or rehabilitation medicine services at the Albany VAMC; Albany, NY	RA (Stage II or III)	Pharma	WL	Cognitive Behavioral Pain Management. Theory-based cognitive behavioral pain management intervention delivered over 6 weeks that included PMR, thermal biofeedback, and cognitive pain management strategies (i.e., stress and coping, problem-solving skills). Problem-solving homework was required; HP of relaxation and biofeedback exercises was encouraged.	10	630	PI; PF; Dep
Berry (2014)	N = 14 (100%); 7% F; Mage = 45; 100% veterans	Patients at the Wm. Jennings Bryan Dorn VAMC; Columbia, SC	Chronic pain, injury-related	NR	SC	Biofeedback. Heart rate variability coherence biofeedback delivered over 4 weeks that included controlled breathing and self-regulation training. No HP was reported.	4	160b	PI;PF
Carmody (2013)	N = 101 (67%); 3% F; Mage = 68; Myears dx = 18; 100% veterans	Physician referrals, flyers, and letters sent to veterans ≥55 years of age at the San Francisco VAMC or affiliated VA community-based outpatient clinics; San Francisco, CA	Chronic pain, multiple	Pharma	Phone-delivered pain education , TM	Cognitive-Behavioral Therapy (telephone-delivered). Manualized form of cognitive-behavioral therapy for pain management delivered over 20 weeks, adapted for low literacy rural populations with a primary goal to facilitate adjustment to chronic pain by teaching methods to manage negative emotions/thoughts, and improve social functioning and coping. Handouts were provided to guide home assignments.	12	540	PI; Dep
Groessl (2017); Linked: Groessl (2015); Groessl, Liu, and Schmalzl (2017); Groessl et al. (2016)	N = 152 (72%); 26% F; Mage = 53; Myears DX = 15; 100% veterans	VA San Diego Healthcare System; San Diego, CA	Chronic LBP	Pharma	WL	Yoga. Hatha yoga consisting of physical yoga postures, movement sequences, regulated breathing, and brief meditation delivered over 12 weeks. A HP manual was provided; participants were encouraged to practice 15–20 minutes on days without an instructor-led yoga session.	24	1440	PI; PF; Dep
Hausmann (2018)c; Linked: Hausmann, Ibrahim, et al. (2018)	N = 360 (86%); 24% F; Mage = 64; 100% veterans	VA Pittsburg Healthcare System, Pittsburg, PA; VA Medical Center, Philadelphia, PA	KneeOA	Pharma and Nonpharma	Neutral activities, TM	Positive Psychology. Home-based intervention that included a series of activities to build positive psychology skills (e.g., reflection, gratitude, kindness, mindfulness) introduced via a brief, 5 to 15-minute telephone call once a week for 6 weeks. A workbook was provided; participants were encouraged to practice the weekly skill each day.	6	60	PI;PF
Hausmann (2017)	N = 42 (100%); 17% F; Mage = 68; 100% veterans	VA Pittsburg Healthcare System, Pittsburg, PA	Knee or HipOA	Pharma and Nonpharma	Neutral activities, TM	Positive Psychology. Home-based intervention that included a series of activities to build positive psychology skills (e.g., reflection, gratitude, kindness, mindfulness) introduced via a brief, 10 to 15-minute telephone call once a week for 6 weeks. A workbook was provided; participants were encouraged to practice the weekly skill each day.	6	75	PI;PF
Highland (2018); Linked (2017)	N = 68 (91%); 63% F; Mage = 44; 28% veterans, 59% active duty	Patients at Walter Reed National Medical Unit; Bethesda, MD	Chronic LBP	Pharma	SC	Yoga. Weekly individual therapeutic yoga sessions that focused on postural alignment, breath work, and centering delivered over 8 weeks. This program allowed for modification in postures. An audio CD was provided for optional HP.	12	720	PI;PF
Kearney (2016); Linked: Martinez (2016); Stephenson, Simpson, Martinez, and Kearney (2016)	N = 55 (82%); 15% F; Mage = 50; 100% veterans (100% PG); 78% on service connection disability	Flyers posted in clinics and letters sent to GW veterans in the VA Puget Sound Healthcare System; Seattle, WA	Chronic pain, GWI	Pharma and Psych	SC	Mindfulness-based Stress Reduction. Mindfulness meditation, body scan meditation, breathing practices, gentle yoga, walking meditation and loving kindness meditation delivered over 8 weeks. Participants were provided an audio CD and assigned 30–45 minutes of HP per day, 6 days per week. Informal practices (e.g., walking or eating mindfully) were also encouraged. Participants received additional readings and a workbook for tracking progress.	9	1620	PI; Dep
Lathia (2009)	N=31 (90%); 10% F; Mage = 62 Myears DX = 4; 100% veterans	Physical therapy unit of Philadelphia VAMC; Philadelphia, PA	Chronic shoulder pain	Pharma	Placebo	Acupuncture (individualized). Individualized treatment points based on patients’ symptoms delivered over 6 weeks. Participants were taught a 10-minute home exercise program (consisting of range-of-motion movements) to be completed 3x/day.	12	360	PI;PF
						Acupuncture (standard). Used fixed, standard protocol treatment points delivered over 6 weeks. Participants were taught a 10-minute home exercise program (consisting of range-of-motion movements) to be completed 3x/day.
Moore and Chaney (1985); Linked: Moore (1983)	N=43 (79%); 2% F; Mage = 49; Myears dx = 17; 81% receiving disability compensati on; 100% veterans	Patients in the VA Puget Sound Healthcare System; Seattle, Washington	Chronic pain, multiple	Pharma	WL	Cognitive-Behavioral Therapy (individual). Group-based lecture/discussion on gate control concepts of pain; goal-setting exercises; and relaxation techniques (i.e., imagery, autogenic training, PMR, breathing exercises) delivered over 4 weeks. Participants were given audiotaped relaxation exercises and asked to practice relaxation 2x/day for 15 minutes.	8	960	PI; PF; Dep
						Cognitive-Behavioral Therapy (couples). Group-based lecture/discussion on gate control concepts of pain; goal-setting exercises; and relaxation techniques (i.e., imagery, autogenic training, PMR, breathing exercises) delivered over 4 weeks. Participants were given audiotaped relaxation exercises and asked to practice relaxation 2x/day for 15 minutes.
Nassif (2016); Linked: Nassif (2013); Nassif, Norris, Soltes, Blackman, etal. (2014); Nassif, Norris, Soltes, Sandbrink, etal. (2014)	(77%); 0% F; Mage = 47; 100% veterans (100% OIF/OEF), 78% army	War Related Illness and Injury Study Center, Washington DC VAMC; Washington, DC	Chronic musculosk eletal pain	NR	WL	Mindfulness Meditation. Included guided imagery, PMR, and body sensing techniques delivered over 8 weeks. Participants were provided with audio recordings for HP and a workbook to log self-practice.	16	960	PI; Dep
Otis (2017)	N = 23 (74%); 9% F; Mage = 50; 100% veterans	VA Boston and Connecticut Healthcare Systems; West Haven, CT; Boston, MA	Comorbid chronic pain and PTSD	NR	WL	Intensive Treatment. Outpatient therapy sessions for pain and PTSD, including cognitive-behavioral therapy and cognitive processing therapy components, delivered over 3 weeks. Both cognitive-behavioral therapy and cognitive processing therapy require HP.	6	540	PI

Note: N (%), number of participants who consented to participate in the study (retention); F, female; Dx, diagnosis; OIF, Operation Iraqi Freedom; OEF, Operation Enduring Freedom; PG, Persian Gulf war; GW, gulf war; GWI, gulf war illness; VA, Veteran’s Administration; VAMC, Veteran’s Administration Medical Center; RA, rheumatoid arthritis; OA, osteoarthritis; LBP, lower back pain; PTSD, posttraumatic stress disorder; Pharma, pharmacological; Psych, psychological; AO, assessment only; WL, waitlist control; SC, standard of care; TM, time-matched; PMR, progressive muscle relaxation; HP, home practice; NR, not/none reported; PF, physical functioning; Dep, depressive symptoms; PI, pain intensity.

^aEstimated total dose (sessions × minutes) of the intervention.

^bDose was estimated from Thurber et al. (2010).

^cData were only reported by race (White/African American); the effect size computed for each group was treated as a separate intervention in the analyses because each effect size represented an independent subgroup.

Study and Sample Characteristics

The studies were published between 1985 and 2018 (M = 2011, SD = 12). Most studies were published in journals (92%); the remaining study included a clinical trial (Otis, 2017). Military personnel were nearly always recruited from Veterans Affairs Medical Centers (VAMC; 11 out of 12 studies); one study recruited veterans and active duty military members from a National Military Medical Center (Highland et al., 2018). Patients were recruited directly (referrals from clinic staff; 67%) or indirectly (self-referrals after seeing posted flyers or brochures, receiving a mailed letter, or telephone calls following the receipt of a mailed invitation letter; 33%). A total of 920 military members (96% veterans) living with chronic pain consented to participate in the studies; mean retention rate was 78% (SD = 0.15; range = 39–100%). Samples included 16% women (SD = 0.17) with a mean age of 55 years (SD = 9; range = 44–68 years). Participants were predominately White (M = 59%, SD = 0.23; n_studies = 9) and married (M = 56%, SD = 0.21; n_studies = 7). Six of the twelve studies sampled patients with multiple chronic pain conditions; two studies sampled patients with chronic lower back pain (Groessl, Liu, Chang, et al., 2017; Highland et al., 2018), two studies sampled patients with osteoarthritis (Hausmann, Youk, et al., 2018; Hausmann et al., 2017) one study involved patients with chronic shoulder pain (Lathia, Jung, & Chen, 2009), and one study sampled patients with Stage II/III rheumatoid arthritis (Appelbaum, Blanchard, Hickling, & Alfonso, 1988). One study enrolled patients with comorbid PTSD (Otis, 2017). Five studies reported length of time since pain diagnosis (median = 17, range = 4–21 years).

Intervention Characteristics

All CIH approaches were designed to promote chronic pain management among military personnel. One approach also targeted veterans with comorbid chronic pain and PTSD (Otis, 2017). The CIH practices included cognitive-behavioral therapies (Appelbaum et al., 1988; Carmody et al., 2013; Moore & Chaney, 1985; Otis, 2017), positive psychology program that included mindfulness (Hausmann, Youk, et al., 2018; Hausmann et al., 2017), yoga (Groessl, Liu, Goodman, et al., 2017; Highland et al., 2018), mindfulness-based interventions (Kearney et al., 2016; Nassif, 2013), acupuncture (Lathia et al., 2009), and biofeedback (Berry et al., 2014). Interventions included a median of 10 sessions (M = 10, SD = 5, range = 4–24) lasting a median of 60 minutes each (M = 66, SD = 48, range = 10–180) delivered over a median of 6 weeks (M = 7, SD = 4, range = 3–20). Home assignments and/or practice were encouraged in many interventions (93%); this included practicing new skills (e.g., coping strategies), cognitive restructuring, or mind-body exercises such as yoga or mindfulness (see Table 1 for details).

Design Characteristics

All studies used a RCT design. The control conditions reported were treatment as usual (50%), wait-list or assessment-only (17% of studies), or active comparison (i.e., sham non-penetrating acupuncture, time-matched pain education or neutral activities; 33%). The four studies that used an active comparison condition included a median of 9 sessions (M = 9, SD = 3, range = 6–12) lasting 21 minutes each (M = 24, SD = 16, range = 10–45). The median number of post-intervention follow-up assessments was two (range = 1 to 3). Assessments were conducted at immediate post-intervention through 26 weeks post-intervention. Because all studies measured the primary outcome (pain intensity) within the month following the intervention, our analyses focused on the first post-intervention assessment (Mdn = 0 weeks; range = 0 to 4.33; k = 15).¹ Exploratory analyses examined the overall efficacy of CIH approaches on the primary (pain intensity) and secondary (physical functioning, depressive symptoms) outcomes observed at the last assessment for the six studies that included a delayed assessment (Mdn = 26 weeks; k = 7).² The overall proportion of MQ criteria met across studies was 70% (SD = 10%; range = 56–88%).

Synthesis of Results

The effect sizes for the primary (pain intensity) and secondary (physical functioning, depressive symptoms) outcomes were examined for outliers prior to analyses (Emerson & Strenio, 1983). A single outlier was detected for physical functioning (Berry et al., 2014) and excluded from subsequent analyses.³ Intervention participants reported greater reductions in pain intensity (d+ = 0.44, 95% CI = 0.21–0.67, k = 15) compared to controls (see Table 2). The hypothesis of homogeneity was not supported for pain intensity (Q [14] = 30.23, p = .007); the proportion of variation due to study heterogeneity exceeded the 50% threshold (I²=54; 95% CI = 17, 74). Mixed findings were observed for the secondary outcomes, that is, intervention participants reported greater improvements in physical functioning (d+ = 0.36, 95% CI = 0.11–0.61, k= 11) but not depressive symptoms (d+ = 0.21, 95% CI = −0.15–0.57, k = 7) relative to controls. The hypothesis of homogeneity was not supported for physical functioning (Q [7] = 19.69, p = .032) and the proportion of variance due to heterogeneity exceeded the 50% threshold (I²=49; 95% CI = 0, 75). The hypothesis of homogeneity was supported for depressive symptoms (Q [6] = 5.94, p = .430; I²=0). Exploratory analyses revealed a significant improvement in physical functioning (d+ = 0.33, 95% CI = 0.03–0.63, k = 5) among participants in the intervention vs. controls at a delayed follow-up but the hypothesis of homogeneity was not supported (Q [4] = 16.10, p = .003; I²=75; 95% CI = 39, 90). There was no significant difference between interventions and controls on pain intensity (d+ = 0.24, 95% CI = −0.04–0.51, k = 7) or depressive symptoms (d+ = 0.29, 95% CI = −0.13–0.70, k = 4) at the delayed assessment. The proportion of MQ criteria met was unrelated to the overall weighted mean effect sizes for pain intensity, physical functioning, or depressive symptoms at the first (ps = .368−.492) or delayed (ps = .403–.593) assessments.

Table 2.

Summary Effect Sizes for Pain Intensity between Intervention and Controls at Postintervention.

Note. The overall weighted mean effect sizes were calculated using random-effects models with methods of moments (MM) and full information maximum likelihood (ML) methods to estimate the between-study variance. Weighted mean effect sizes (d₊) are positive for differences that favor the intervention relative to control groups. d₊, weighted mean effect size; SE, standard error; CI, confidence interval; CBT, cognitive-behavioral therapy

Predictors of Pain Intensity

Moderator analyses were used to determine if the sample (proportion women, age, mean years diagnosed with chronic pain condition, depressive symptoms) or intervention characteristics such as duration (number of intervention sessions/minutes, number of home practice sessions/minutes), delivery mode (individual or group), type of CIH approach (cognitive-behavioral therapies or mind-body practices), and type of control (active vs. passive) explained the variability in the overall weighted mean effect sizes for the primary outcome (pain intensity). None of the hypothesized sample or intervention characteristics moderated the effect sizes for pain intensity or physical functioning (data not shown).

Risk of Bias Across Studies

The risk of bias was assessed only for pain intensity (k = 15) and physical functioning (k =11) due to the limited number of interventions measuring depressive symptoms (k = 7). The funnel plots for pain intensity and physical functioning appear in the Online Supplemental Materials (Figures S1 and S2). A visual inspection of the funnel plot revealed asymmetries that might be interpreted as small-study effects for pain intensity. Begg (Begg & Mazumdar, 1994) and Egger’s (Egger et al., 1997) statistical tests also indicated evidence for the presence of small study effects (Begg: Δx–y = 43, z = 2.08, p = .038; Egger: bias coefficient = 1.87, SE = 0.67, p = .016). Trim and fill procedures (Duval & Tweedie, 2000) estimated that five studies measuring pain intensity could be missing which may have changed the interpretation of our findings (i.e., estimated effect size with the imputed missing studies: d+ = 0.22, 95% CI = −0.05, 0.48) but the results of this test should be interpreted with caution given that simulation studies have consistently shown that this method performs poorly in the presence of substantial between-study heterogeneity (Peters, Sutton, Jones, Abrams, & Rushton, 2007; Terrin, Schmid, Lau, & Olkin, 2003). The visual inspection of the funnel plot and statistical tests revealed no asymmetries that might be interpreted as small-study effects for physical functioning (Begg: Δx–y = 9, z = 0.62, p = .533; Egger: bias coefficient = 1.39, SE = 0.82, p = .126).

Discussion

Chronic pain causes both physical and emotional suffering and interferes with daily functioning. Furthermore, the psychological stress associated with managing chronic pain symptoms, pain medications and their side effects, and medical follow-up care place a significant burden on patients. Previous research suggests interventions that provide cognitive and behavioral coping strategies to manage stress may be particularly valuable to patients living with chronic pain (Sturgeon, 2014). The current meta-analytic study advances the literature supporting CIH approaches for chronic health conditions (Dunne et al., 2019; Scott-Sheldon, Balletto, et al., 2019; Scott-Sheldon, Gathright, et al., 2020; Scott-Sheldon et al., 2008) by focusing on military populations experiencing chronic pain. Our meta-analysis showed that CIH practices reduced chronic pain intensity and improved physical functioning for active duty military and veterans with chronic pain.

The magnitude and direction of our findings are consistent with the Agency for Healthcare Research and Quality (AHRQ) report on nonpharmacological treatment for chronic pain conditions across multiple patient populations (Skelly et al., 2018).⁴ The AHRQ report found that psychological (e.g., cognitive behavioral therapy) and mind-body (e.g., yoga) interventions lessened pain severity and improved physical functioning among patients with chronic back pain, osteoarthritis, and fibromyalgia, with the overall effect sizes ranging from low to moderate in the short-term. Consistent with the AHRQ report, CIH approaches included in our meta-analysis resulted in an overall effect size of moderate strength for mind-body interventions (pain severity: d = 0.49, 95% CI = 0.21–0.77, k = 10; physical functioning: d = 0.34, 95% CI = 0.07–0.61, k = 8) but, in contrast with the AHRQ report, the effects were not significant for cognitive-behavioral therapies (pain severity: d = 0.34, 95% CI = −0.08–0.77, k = 5; physical functioning: d = 0.49, 95% CI = −0.14–1.13, k = 3). Because few studies included in this meta-analysis evaluated cognitive-behavioral therapies, we may have been underpowered to detect a significant overall between-group difference. Finally, our meta-analysis showed that mind-body interventions resulted in low to moderate improvements in physical functioning among military members at a delayed follow-up (i.e., 26 weeks following completion of the intervention). None of the studies evaluated cognitive-behavioral therapies at a delayed follow-up.

CIH approaches can provide military populations with the skills to manage their chronic pain. For instance, cognitive-behavioral therapy for chronic pain has been tailored to military veterans and includes sessions focused on restructuring maladaptive pain-related thinking, relaxation strategies, and physical activity pacing (Murphy et al., 2014). Yoga is effective at reducing pain intensity (Büssing, Ostermann, Lüdtke, & Michalsen, 2012; K. A. Williams et al., 2005) and improving mood-related outcomes (e.g., depressive symptoms, anxiety) for patients with chronic pain conditions (Lavey et al., 2005). Similarly, tai chi, considered as a gentle “moving meditation,” has been found to reduce pain (Hall, Maher, Latimer, & Ferreira, 2009) and positively influence physiological biomarkers of stress (e.g., heart rate, cortisol level) (Sandlund & Norlander, 2000). Lastly, acupuncture is frequently offered for military veterans experiencing tinnitus and symptoms of post-traumatic stress. Veterans with chronic pain may benefit from engaging in one or more of these CIH practices.

Contrary to study hypotheses, our meta-analysis did not find a significant difference between intervention participants and controls on depression, nor did depression serve as a moderator for treatment effects. These findings contrast with the prior literature (Gilpin, Keyes, Stahl, Greig, & McCracken, 2017), and may have been a result of the limited number of studies assessing depressive symptoms. In fact, depressive symptoms were reduced in six out of the seven interventions measuring this outcome, but these reductions were typically not significantly different with controls (one exception: Kearney et al. (2016)). Additionally, there may have been a restricted range of baseline depressive symptoms, as none of the included studies reported recruiting patients experiencing high levels of psychological distress. CIH approaches are likely to be more beneficial for individuals who are experiencing distress, yet few studies adequately assess samples for baseline levels of distress. Future studies should evaluate the efficacy of CIH practices among military populations with different levels of psychological distress. In addition to depressive symptoms, the present study examined several hypothesized moderators (e.g., participant sex, type of CIH approach, intervention dose) and did not find significant effects. This was surprising, given the substantial heterogeneity observed but may be most parsimoniously explained as the result of the variability in reporting of demographic and secondary factors across studies, which limited the power of the moderator tests.

Over the past decade, the U.S. military has made substantial efforts to promote chronic pain management, to reduce opioid prescribing, and to increase access to nonpharmacological and integrative medicine approaches. The Army appointed a Pain Management Task Force to coordinate treatment efforts within the Department of Defense and the Veterans Health Administration in 2009 to offer “holistic, multidisciplinary, and multimodal” treatment for veterans with chronic pain (Pain Management Task Force, 2010). The military also deployed the Sole Provider Program and the Controlled Drug Management Analysis and Reporting Tool to help identify and monitor the risk and misuse of opioid use (Sharpe Potter, Bebarta, Marino, Ramos, & Turner, 2014). These changes in chronic pain monitoring and treatment was necessitated by the significant opioid epidemic in the U.S. and the realization by the scientific and medical communities that opioid medications do not provide a long-term solution for chronic pain. Pharmacological interventions are insufficient for addressing the functional and emotional symptoms of living with chronic pain. In fact, a critical component of effective chronic pain management is motivating patients to engage in physical activity. Often patients with chronic pain develop a fear of movement, or kinesiophobia, which leads to a decline in activity and muscle deconditioning. VAMCs are increasingly using psychological, physical, and mind-body approaches to chronic pain management (Elwy et al., 2014). For example, “Whole Health” initiatives within the VA system are transforming management of chronic health conditions by utilizing “physical, mental, emotional, spiritual, and environmental elements that work together to provide the best quality of life for each Veteran” (U.S. Department of Veterans Affairs, 2017).

Limitations

The current meta-analysis is limited by several factors. First, only a small number of studies met the inclusion criteria limiting our ability to examine additional psychological outcomes often associated with chronic pain (e.g., anxiety, pain catastrophizing). In addition, the current meta-analysis is also limited by the heterogeneous group of chronic pain types given the small number of studies published using CIH approaches for pain among military populations. Future meta-analyses targeting specific chronic pain conditions among U.S. military members should be conducted as additional studies become available. Second, moderator tests were restricted by the data reported in the included studies. Potential predictors of pain severity that could not be tested included status (active-duty vs. military veterans), “theater” (i.e., where veterans served in combat), type of pain medications, and level of disability related to military service. Future studies should report these factors as they may be informative to clinical decision-making. Third, we could not detect long-term benefits because none of the studies included an extended follow-up (i.e., >6 months). Because many of the studies included in this meta-analysis included small samples, with relatively short-term improvements, we recommend that future studies recruit larger samples and follow participants longer (e.g., 12 to 18 months). Finally, study samples were primarily limited to White male veterans which limits the generalizability of the findings to all military members. Future studies should include a representative sample of military members including the active-duty force.

Conclusions

Increasingly, patients and providers are looking to nonpharmacological, nonsurgical treatment options for chronic pain management. Our results indicated that CIH approaches provide an efficacious treatment option for veterans experiencing chronic pain. This research provides support for the use of CIH practices with active duty and military veterans experiencing chronic pain. Additional randomized studies are needed with longer follow-up periods to determine the long-term benefits of CIH practices in chronic pain management. Future research should continue to evaluate CIH practices with military populations, as rates of chronic pain continue to be a major concern for returning veterans and their families.