Key Findings and Strength of Evidence

The key findings of this review, including strength of evidence ratings, are summarized for each chronic pain condition in Tables 47–61 (interventions and comparators with no evidence for either function or pain outcomes are not shown); domains used to determine the overall strength of evidence are shown in Appendix G. All outcomes were considered direct. The strength of evidence was low or insufficient for many interventions and was limited by small numbers of trials for specific comparisons and for our specified time frames, particularly for long term. We focused on evaluating the persistence of effects for therapies beyond the course of treatment, using the following definitions for postintervention followup: short term (1 to <6 months), intermediate term (≥6 to <12 months) and long term (≥12 months). Evidence was particularly limited on effects on long-term outcomes.

The majority of trials compared interventions with usual care with very few trials employing pharmacological treatments or exercise as comparators. In general, effect sizes for most interventions were small, based on mean differences. There tended to be more evidence for the effects of interventions on pain than for function, and the effects on function were generally smaller or not clearly present.

No trials directly compared interventions with opioids and few trials reported effects of intervention on opioid use. Our previous reviews found opioids associated with small to moderate effects on pain during treatment (effects would not be expected to persist) with evidence almost exclusively from short-term (≤3 month) trials.^11,16,25,247 Information on adherence to interventions was not well-reported; poor adherence may have impacted some of our findings. Harms were poorly reported across interventions. No serious intervention-related adverse events requiring medical attention were identified; reported adverse events were generally minor (e.g., muscle soreness or increased pain with exercise, bruising with acupuncture) and time-limited (e.g., temporary worsening of pain).

Table 47

Chronic low back pain: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 48

Chronic low back pain: effects of nonpharmacological interventions compared with exercise.

Table 49

Chronic neck pain: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 50

Chronic neck pain: effects of nonpharmacological interventions compared with pharmacological treatments.

Table 51

Chronic neck pain: effects of nonpharmacological interventions compared with exercise.

Table 52

Osteoarthritis of the knee: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 53

Osteoarthritis of the knee: effects of nonpharmacological interventions compared with exercise.

Table 54

Osteoarthritis of the hip: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 55

Osteoarthritis of the hip: effects of nonpharmacological interventions compared with exercise.

Table 56

Osteoarthritis of the hand: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 57

Fibromyalgia: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 58

Fibromyalgia: effects of psychological therapies compared with pharmacological treatments.

Table 59

Fibromyalgia: effects of nonpharmacological interventions compared with exercise.

Table 60

Chronic tension headache: effects of nonpharmacological interventions compared with usual care, placebo, sham, attention control, or waitlist.

Table 61

Chronic tension headache: effects of nonpharmacological interventions compared with pharmacological treatments.

Low Back Pain. For chronic low back pain, compared with usual care, attention control, sham, or placebo, there was moderate evidence of slight improvement in function, at least in the short term, for massage, yoga, psychological therapies (cognitive-behavioral therapy [CBT]) (strength of evidence [SOE]: moderate) and low evidence for exercise, acupuncture, low-level laser therapy, spinal manipulation, multidisciplinary rehabilitation (SOE: low). With the exception of spinal manipulation, these interventions also showed slight improvement (exercise, acupuncture, massage, psychological therapies, multidisciplinary rehabilitation, SOE: low) or moderate improvements (yoga, low-level laser therapy, SOE: low) in pain short term. The slight improvements in function compared with controls were sustained into the intermediate term for yoga, spinal manipulation, psychological therapies, and multidisciplinary rehabilitation, with low strength of evidence for all but the psychological therapies, for which SOE was moderate. No clear improvement in function was seen at intermediate term for exercise, acupuncture, massage or low-level laser therapy (SOE: low for all). Improvements in pain were seen in the intermediate term for exercise (slight effect SOE moderate) and yoga (moderate effect, SOE low) and mindfulness-based stress reduction (MBSR) (slight effect, SOE: low) as well as spinal manipulation, psychological therapies and multidisciplinary rehabilitation (slight effects, SOE: moderate). Long-term evidence was available for four intervention categories: psychological therapies, multidisciplinary rehabilitation, exercise, and acupuncture. The strongest evidence was for psychological therapies (CBT primarily), which were associated with slightly greater effects than usual care or attention control on both function and pain at short, intermediate, and long term (SOE: moderate for all time frames). Neither exercise nor acupuncture was associated with improved function long term, even though both demonstrated continued pain improvement (SOE: low for all). For multidisciplinary rehabilitation, effects on function from earlier time frames were not sustained in the long term versus usual care (SOE: low). High intensity multidisciplinary rehabilitation (≥20 hours/week or >80 hours total) was not clearly better than nonhigh intensity programs. Short-term effects on function and pain were somewhat larger with high intensity multidisciplinary rehabilitation than with nonhigh intensity interventions but the tests for interaction were not statistically significant. At intermediate term, estimates were similar for high intensity and nonhigh intensity programs.

In people with chronic low back pain, there were no clear differences in short-term function for comparisons of qigong, yoga, or spinal manipulation with exercise even though small improvements in pain were seen for yoga (SOE: low for all). Multidisciplinary rehabilitation was associated with small effects on function short term as well as pain (SOE: moderate). For Qigong, results for intermediate-term function and short-term pain slightly favored exercise (SOE: low for all). Again, multidisciplinary rehabilitation was associated with slight improvements in function and pain at intermediate term (SOE: moderate), but this was not sustained in the long term (SOE: low). Long-term data were only available for multidisciplinary rehabilitation.

Neck Pain. For chronic neck pain, in the short term, moderate effects on function and pain were seen for low-level laser therapy (SOE: moderate). In the short term and intermediate term, acupuncture and Alexander Technique were associated with slightly greater effect on function compared with usual care (both interventions), sham acupuncture or sham laser (SOE: low). The effect of acupuncture was not sustained long term (SOE: low) compared with sham acupuncture, sham laser, or usual care, and no improvement in pain was seen at any time frame (SOE: low). There were no clear improvements in function or pain across types of exercise (short term) or for psychological therapies or massage compared with usual care, sham procedures, or attention controls (SOE: low for all).

Knee Osteoarthritis. For knee osteoarthritis (OA), exercise, microwave diathermy and ultrasound were associated with functional improvement in the short term compared with usual care, attention control, or sham procedure; the effect size was small for exercise and ultrasound, and larger for diathermy (SOE: moderate for exercise, low for ultrasound, diathermy). While the small effects of exercise on function persisted into the intermediate and long term (SOE: low for both), there were no clear benefits to ultrasound at intermediate term (SOE: low). Similarly, small short-term effects of ultrasound on pain did not persist to intermediate term (SOE: low) in contrast to moderate improvement in pain for exercise (SOE: low). Long term, the small improvement in function seen with exercise was sustained, but there was no clear effect on pain (SOE: low). There were no clear differences in function or pain associated with electromagnetic fields (short-term SOE: low), with psychological therapies for any time frame (SOE: low), or with acupuncture at short (SOE: moderate) or intermediate term (SOE: low) versus usual care, attention control, or sham procedure. There was no difference in function or pain between pain coping skills training and exercise at short term or intermediate term in one trial (SOE: low).

Hip and Hand Osteoarthritis. Evidence was sparse on interventions for hip and hand OA. Exercise was associated with slightly greater function than usual care at short and intermediate-term (SOE: low), but data were in sufficient to determine long-term effects. For pain, a small effect was seen only at short term; no differences were seen at the other time points (SOE: low for short term and intermediate term, insufficient for long term). Compared with exercise, a small effect on function was seen with manual therapy in the short and intermediate term, and small improvement in pain short term (SOE: low for all). For hand OA, no clear differences were seen for low-level laser therapy versus sham or for multidisciplinary rehabilitation versus waitlist control at short term for either function or pain (SOE: low).

Fibromyalgia. Short term, in patients with fibromyalgia, there was low-quality evidence that slight improvements on function were associated with exercise, CBT, and mind-body practices of tai chi and qigong (SOE: low for all) compared with wait list and attention control, and moderate-quality evidence for slight functional improvement acupuncture compared with sham acupuncture (SOE: moderate). Improvements in short-term pain were seen with exercise (SOE: moderate) and mind body practices (SOE: low), but not with acupuncture. No clear differences in function or pain outcomes were seen for MBSR (SOE: moderate) or multidisciplinary rehabilitation (SOE: low). Slightly greater effects on function continued into the intermediate term for acupuncture and CBT and massage (SOE: low), and were seen for myofascial release massage and multidisciplinary rehabilitation; there was no clear effect of magnetic mattress pads versus sham pad (SOE: low for all). Slight improvement in pain intermediate-term were seen for massage and multidisciplinary rehabilitation (SOE: low), but not for exercise (SOE: moderate), acupuncture, or magnetic mattress pads (SOE: low). Long term, small improvements in function continued for multidisciplinary rehabilitation but not for exercise or massage (SOE: low for all), and there was no clear impact on pain for exercise (SOE: moderate) or multidisciplinary rehabilitation (SOE: low). No clear differences were seen between multidisciplinary rehabilitation and exercise for the long term on function or pain (SOE: low). CBT was associated with a small benefit for function but not for pain compared with pregabalin at intermediate term (SOE: low).

Chronic Tension Headache. Only nine trials of nonpharmacological treatments for chronic tension headache met the inclusion criteria and all but one was considered poor quality, resulting in a rating of insufficient evidence for comparisons of psychological therapies with waitlist or attention control, electrical stimulation versus sham, and acupuncture versus sham. One fair-quality trial of laser acupuncture versus sham suggested moderate improvement in pain short term (SOE: low), and another fair-quality trial of spinal manipulation versus usual care suggested a small effect on short-term function based on the Headache Impact Test (SOE: low). Approximately 25 percent of the patients in the trial had comorbid migraine headache.

Usual Care/Waitlist and Nonactive Comparators. For comparisons involving usual care/waitlist or nonactive comparators (placebo, sham, attention control), there were some differences depending on the specific comparator evaluated. For some interventions results different by control type. For example, in some analyses, acupuncture was associated with greater effects on pain in patients with chronic low back pain or OA when compared with usual care than when compared with sham acupuncture, suggesting that much of the benefit may be due to placebo or other nonspecific effect.

Harms. Harms were poorly reported across interventions. No serious intervention-related adverse events requiring medical attention were identified; reported adverse events were generally minor (e.g., muscle soreness with exercise, bruising with acupuncture) and time-limited (e.g., temporary worsening of pain).

Medication Use. Few trials compared opioid use pre- and post-intervention, and medication use in general was not well reported across trials.

Subgroups. One fair-quality trial in people with knee OA formally examined factors that might modify the effect of exercise on disability; the effect of exercise on activities of daily living disability did not appear to be modified by age, sex, baseline disability, knee pain score, body mass index, or race.⁵⁰ The few trials that reported subgroup analyses either did not provide sufficient data to assess modification by demographic or other factors or did not formally test for modification; trials were generally too small to effectively evaluate outcomes in subgroups.

Findings in Relationship to What Is Already Known

Many reviews have addressed the effects of interventions for chronic pain management during or immediately following treatments. We focused on evaluating the sustainability of effects for at least 1 month postintervention.

This review updates our previous review on low back pain²⁵ by incorporating new evidence on nonpharmacological treatments for chronic low back pain. Consistent with the prior review, we found exercise, yoga, various psychological therapies, acupuncture, spinal manipulation, and low-level laser therapy with small to moderate effects on function and/or pain. It differs from the prior review in focusing on durability of treatment effects 1 month or longer after completion of a course of treatment and basing estimates on meta-analyses when poolable data were available, and conducting stratified and sensitivity analyses to evaluate sources of heterogeneity and robustness of findings. For example, subanalyses of specific interventions within a given category of intervention (e.g., aerobic exercise within the general category of exercise suggests that despite the inherent heterogeneity within some of the categories, effect estimates results for specific interventions may be similar). Although we found some evidence that beneficial effects of some nonpharmacological therapies persist for up to 12 months following the end of a course of a treatment, data on longer-term (>1 year) outcomes were very sparse.

A recent Institute for Clinical and Economic Review (ICER) review²⁴⁸ on chronic low back pain and neck pain used relevant portions of our previous review for chronic low back pain and updated it with new publications so the findings are generally consistent with our review for this condition. For chronic neck pain, this report and the ICER report both suggest a small benefit for acupuncture. The ICER report focuses on evaluating comparative value for interventions and suggests that cognitive and mind-body therapies for treatment of chronic low back pain and chronic neck pain would be cost-effective, would meet value-based price benchmarks, and may result in only a small increase ($0.75) per member per month for a hypothetical payer plan covering 1 million members, compared with approximately $4.46 per member per month for pain medication.

Our findings indicate that a number of nonpharmacological treatments improve pain and/or function for specific chronic pain conditions included in this review. This is consistent with other reviews, including recent reviews on exercise²⁴⁹ acupuncture,²⁵⁰ and complementary health approaches²⁵¹ for chronic pain management across various conditions, an Agency for Healthcare Research and Quality (AHRQ) report on knee OA treatment,²⁵² and a review of chronic pain treatment guidelines on the use of manual and physical therapies.²⁵³

The protocol for a systematic review and network meta-analysis of interventions for fibromyalgia was identified;²⁵⁴ no publication timeline for this review is currently available.

Applicability

The applicability of our findings may be impacted by a number of factors. Included trials provided limited information on symptom duration, clinical characteristics, comorbid conditions, and concomitant treatments, thus it is not clear to what extent these trials reflect the populations seen in clinical practice or how these impact our results. In addition, with the exception of fibromyalgia, information regarding diagnostic criteria for the pain condition of interest was limited. Information on the presence of overlapping chronic pain conditions or psychosocial factors was generally not provided in included trials, and the extent to which these characteristics were present in trial populations and their impact on our results is not clear. Across conditions, a majority of trial participants were female. The age of included populations generally reflected the ages impacted by the conditions. Evidence to how effectiveness varies by ages was limited. There was also heterogeneity in populations enrolled in the trials with regard to duration of chronic pain, severity of pain (most trials enrolled patients with at least moderate pain at baseline), as well as other factors (e.g., use of medications, medical and psychological comorbidities). Our findings are generally most applicable to people without such comorbidities who have moderate or severe intensity pain that has persisted for more than 12 months. The heterogeneity in populations across included trials likely is consistent with the heterogeneity seen in clinical practice, so our findings may be applicable to most primary care clinical settings.

Variability in interventions, comparators and cointerventions may impact our findings. For interventions, there was variability in the numbers of sessions, length of sessions, duration of treatment, methods of delivering the intervention and the experience and training of those providing the intervention. To address heterogeneity within intervention categories we abstracted details of techniques or methods used, (e.g. specific type of psychological intervention or Yoga) and attempted to stratify by them, however in most cases, data were insufficient to do so. In general, there were no clear differences in effects based on intervention factors or comparators; however analyses were limited by small numbers of trials. In clinical practice, most chronic pain patients likely use a combination of therapies and may continue to receive some types of therapies if benefit is perceived. It is unclear to what extent our findings represent the conditions under which the various interventions are currently delivered. Evidence to identify optimal techniques and delivery of interventions is needed.

To facilitate interpretation of results across trials and interventions, we categorized the magnitude of effects for function and pain outcomes using the system described in our previous review.²⁵ Using this system, beneficial effects identified were generally in the small or moderate range. We recognize that effects that we classified as small (e.g., 5 to 10 points on a 0 to 100 scale for pain or function) may be below some proposed thresholds for minimum clinically important differences for some measures. However, our classification provides some consistent and objective benchmarks to assess magnitude of smaller effects across trials and interventions. Interpretation of clinically important differences in mean change for continuous variables is challenging. If data were provided we also evaluated the proportion of patients who experienced a clinically important improvement in pain or function. This provides valuable insight regarding clinically important improvement. For example, one trial⁸⁹ of MBSR versus usual care in low back pain reported a small improvement in function on a modified Roland Morris Disability questionnaire (1.87, 95% CI −3.14 to −0.60 on 0–23 scale); however, absolute difference between MBSR and usual care on the percentage of participants (20%) achieving a minimally clinically meaningful (≥30%) improvement from baseline (68.8% to 48.6%, risk ratio 1.56, 95% CI 1.14 to 2.14) suggests that the benefits may be more substantial.

Limitations of the Evidence Base

Evidence was sparse for most interventions. Data on long-term outcomes was particularly limited. There were also limited data on outcomes other than pain and function and on harms. Few trials directly compared an included intervention versus pharmacological therapy or the specified active comparator (exercise or biofeedback). Only 5 percent of included trials across conditions were considered to be of good quality; the majority were considered fair (59%). No trial of treatment for chronic tension headache was considered to be of good quality. For some interventions, it is possible to effectively blind participants and providers (e.g. CBT, multidisciplinary rehabilitations, exercise); thus, observed effects may be due in part to placebo, attention, or other nonspecific effects and results may have been susceptible to performance and other biases. Many included trials were small (< 70 participants) and only few or single trials were available for some interventions (e.g., some physical modalities). The combination of these factors led to a determination that evidence was insufficient. There was no or little includable evidence for a number of interventions, including electromuscular simulation, traction, superficial heat or cold, bracing, use of magnets, interferential therapy, transcutaneous electrical nerve stimulation, and manual therapies (other than for low back pain). For most conditions, evidence was also sparse for mindfulness and mind-body practices. Evidence on interventions for hip and hand OA and chronic tension headache was very limited.

Heterogeneity in clinical diagnosis and presentation was present for most of the conditions, with the exception of fibromyalgia. It is likely that included patients may have additional conditions and/or psychological comorbidities that were not described in the trials. Details provided by trials were insufficient to conduct meaningful subanalyses.

Some of the limitations described for the review process reflect limitations of the evidence base, including those related to heterogeneity within and across interventions and heterogeneity within a given condition. Details of concurrent interventions and components of usual care were generally not reported or poorly reported. Additionally, it is assumed that most patients with chronic pain likely continued medications and other therapies or practices during the trials. These factors may have resulted in substantial mixing of effects of the intervention and cointerventions. These factors possibly attenuated observed effects.

Data on potential harms is sparse, although serious harms are not generally expected with the interventions included in this review. Serious treatment-related adverse events were not reported in any of the trials.

Implications for Clinical and Policy Decisionmaking

Our review provides some evidence that an array of nonpharmacological treatments provide small to moderate benefits function and pain that are durable for more than 1 month for the five common chronic pain conditions addressed in this review. Musculoskeletal pain, particularly of back and joint pain, is the most common single type of chronic pain. Age-adjusted rates of adults reporting pain in the last three months were highest for low back pain (28%), neck pain (15%), knee pain (19.5%) and severe headache or migraine (16%).^3,14

The evidence synthesized in this review may help inform guidelines and health care policy (including reimbursement policy) related to use of noninvasive nonpharmacological treatments, and inform policy decisions regarding funding priorities for future research.

Recent guidelines¹³ from the Centers for Disease Control and Prevention (CDC) in the United States and the Canadian Guideline for Opioid Use in Chronic Non-Cancer Pain²⁵³ recommend nonopioid treatment as the preferred treatment for chronic pain. Further, guidelines from the American College of Physicians recommend nonpharmacological therapies over medications for chronic back pain.¹⁶ Our findings support the feasibility of implementing these guideline recommendations by showing that there are some nonpharmacological treatments for chronic pain that have evidence of sustained effectiveness after the completion of therapy. Importantly, some interventions, such as exercise, CBT, multidisciplinary rehabilitation, mind-body interventions, and some complementary and integrative medicine therapies, such as acupuncture and spinal manipulation, also were associated with some sustained effects on function, although evidence beyond 12 months is sparse. There was no evidence suggesting serious harms from these interventions, although harms data were limited.

Our report reviewed evidence that may also help inform decisions regarding prioritization of nonpharmacological therapies by clinicians selecting therapy. Consistent with a biopsychosocial understanding of chronic pain,^3,7 evidence was somewhat more robust for “active” interventions that engage patients in movement and address psychological contributors to pain, particularly at longer-term followup, versus more “passive” treatments focused on symptom relief such as massage. Active interventions include exercise, multidisciplinary rehabilitation, psychological therapies (particularly CBT), and mind-body interventions. This provides some support for clinical strategies that focus on “active” interventions as primary therapies, with “passive” interventions used in a more adjunctive or supplementary role. Research is needed to compare “active” vs. “passive” strategies.

Our review also has policy implications related to access to treatment and reimbursement. Given heterogeneity in chronic pain, variability in patient preferences for treatments,^3,7 and differential responses to specific therapies in patients with a given chronic pain condition, policies that broaden access to a wider array of effective nonpharmacological treatments may have greater impact than those that focus on one or a few therapies. Several considerations could inform policy decisions regarding access to and coverage of nonpharmacological therapies. Policymakers could prioritize access to interventions with evidence of persistent effectiveness across different pain conditions, such as exercise, multidisciplinary rehabilitation, mind-body interventions, and acupuncture. Because the level of supporting evidence varies from condition to condition, policymakers may need to consider the degree to which evidence may be reasonably extrapolated across conditions (e.g., effectiveness of psychological therapies for chronic low back pain may not necessarily be extrapolated to OA). Although the Affordable Care Act has improved access to complementary and integrative therapies, variability in reimbursement and authorization procedures remain a potential barrier. Although evidence supports the use of multidisciplinary rehabilitation over exercise therapy or usual care, primarily for low back pain, cost and availability remain important barriers particularly in rural areas. Our report suggests that less-intensive multidisciplinary rehabilitation may be similarly effective to high-intensity multidisciplinary rehabilitation, which could inform decisions about more efficient methods for delivering this intervention. Not all patients may require multidisciplinary rehabilitation.²⁵⁵ Policy efforts that focus on use of multidisciplinary rehabilitation in individuals more likely to benefit (e.g., severe functional deficits, failure to improve on standard nonmultidisciplinary therapies, significant psychosocial contributors to pain) could also inform efforts to deliver this modality efficiently.

Limitations of the Systematic Review Process

There were limitations in the systematic review process. Our analysis was restricted to trials that reported outcomes after at least 1 month following the end of therapy (except when therapy lasted at least 6 months; in these cases, we included assessments made immediately post-treatment). We did not include trials of patients with chronic pain conditions other than those specified in the methods and excluded trials of patients with diffuse or mixed pain conditions. Some noninvasive nonpharamcological interventions (e.g., self-management education) were excluded, and we did not address invasive therapies. Trials that evaluated active comparators other than biofeedback (for headache) or exercise (all other conditions) or interventions as adjunctive treatment were excluded. Some meta-analyses were based on two or three trials; findings based on such meta-analyses must be interpreted with caution.

The interventions were grouped a priori to provide an organizational framework for the report. There is some overlap between categories and there a many other methods of grouping interventions. We performed separate or stratified analyses to the extent possible to evaluate specific techniques/methods within broader categories (e.g., we looked at different types of psychological therapies and mind-body practices). We also performed stratified analyses by comparator type where data were available. Sparse literature for many of the interventions precluded extensive examination specific types of intervention within a given category.

We excluded non-English-language articles; however, we did not identify large numbers of non-English-language articles in our review of bibliographies. We searched ClinicalTrials.gov and identified some potentially relevant studies, but none had results available. We did not search conference proceedings or other sources. We were unable to assess for publication bias using graphical or statistical methods to evaluate any potential impact of small samples, methodological limitations in trials, or heterogeneity in interventions, populations or outcomes. Based on hand searches of reference lists, searches of ClinicalTrials.gov, and suggestions from technical experts, we did not find evidence indicating the presence of unpublished literature sufficient to impact conclusions.

Research Recommendations

The gaps in the available evidence are many across the common conditions we included (Table 62). Four primary issues relate to (1) the need to understand the longer-term sustainability of intervention effects; (2) the need for standardization of interventions for future trials; (3) the standardization of research protocols for collection of and reporting of outcomes including harms; (4) the need for comparisons of interventions with pharmacological interventions. For many of these areas, future research would benefit from considering recommendations from organizations such as the Initiative on Methods, Measurements, and Pain Assessment in Clinical Trials (IMMPACT)^256–261 and the Analgesic, Anesthetic, and Addiction Clinical Trials Translations, Innovations, Opportunities, and Networks (ACTTION)^262,263 and the research priorities outlined in the recent Federal Pain Research Strategy.²⁶⁴

To understand the sustainability of effects, methodologically rigorous traditional (explanatory) trials with longer followup are needed to better understand whether benefits are sustained over time under ideal conditions. In addition, well-designed pragmatic trial designs with long-term followup could facilitate understanding of how interventions are delivered and continued in real-world settings as well as effect sustainability. Methods for enhancing recruitment, adherence and retention need to be incorporated for all trials. Education of researchers examining nonpharmacological approaches to pain management on clinical trial design, execution, and analysis may also assist with improving the quality of the evidence base for many of the interventions.

Research to identify optimal techniques and their delivery would help define more standardized interventions to evaluate in future trials is needed. In addition, there is a need to understand what combinations of interventions may be most logical for a given condition and standardization of methods to study adjunct therapies. Pragmatic trials may help provide insight into these questions.

Standardization of research protocols for reporting and outcomes measures and use of a standard set of measures would facilitate comparison of results across trials. Outcome measures such as the Visual Analog Scale or Numeric Rating Scale may not fully capture the impact of pain or allow for accurate classification or evaluation of changes in chronic pain. Inclusion of recommendations for pain assessment²⁶⁵ assessment that incorporate understanding of pathophysiological mechanisms and address multiple domains of pain, including temporal dimensions, sensory and affective qualities of pain and the location and bodily distribution of pain in trial planning and execution may facilitate more accurate classification and longitudinal tracking of response to interventions. Reporting the proportions of patients achieving a clinically meaningful improvement in pain, function, or quality of life as measures of “success” may provide additional clinical information to complement data on average changes in continuous measures of pain, function, and quality of life for which there is difficulty describing clinically important effects. Routine collection of common or known harms associated with interventions is needed in future trials.

There is heterogeneity with regard to research design, execution, and outcomes reporting in trials of interventions included in this review compared with well-funded trials of devices or pharmacological agents. Lack of funding to design methodologically sound studies with reasonable sample size of nonpharmacological interventions may have contributed to the general low quality of evidence.

Table 62

Summary of evidence gaps and research recommendations.

Conclusions

Exercise, multidisciplinary rehabilitation, acupuncture, CBT, and mind-body practices were most consistently associated with durable slight to moderate improvements in function and pain for specific chronic pain conditions. Our findings provide some support for clinical strategies that focus on use of nonpharmacological therapies for specific chronic pain conditions. Additional comparative research on sustainability of effects beyond the immediate post-treatment period is needed, particularly for conditions other than low back pain.