Table 3.

Effectiveness and Harms of Nonpharmacologic and Nonopioid Pharmacologic Treatments^a

Source	Topic or Intervention	Participants or Population	Primary Outcomes	Key Findings	Study Quality
Busch et al,48 2007	Exercise training vs untreated control or nonexercise intervention	Systematic review of 33 RCTs with fibromyalgia patients	Global well-being, selected signs and symptoms, and physical function	Exercise training improves global well-being and physicalfunction. Supervised aerobic exercise training has beneficial effects on physical capacity and fibromyalgia symptoms.	Four studies were classified as high quality, 15 as moderate quality, and 14 as low quality
Chaparro et al,49 2014	Noninjectable opioids vs placebo or other treatments	Systematic review of 15 RCTs with patients with chronic low back pain	Pain	One trialfound tramadolsimilar to celecoxib for pain relief. Two trials did not find a difference between opioids and antidepressants for pain or function.	Low- to moderate-quality evidence
Collins et al,50 2000	Antidepressants vs placebo; anticonvulsants vs placebo	Systematic review of 19 RCTs for diabetic neuropathy or postherpetic neuralgia	Pain	For diabetic neuropathy, the NNT for ≥50% pain relief was3.4 for antidepressants (12 trials, 10 evaluated TCAs and 3 SSRIs) and 2.7 for anticonvulsants (3 trials). For postherpetic neuralgia, the NNT was 2.1 for antidepressants (3 studies evaluating TCAs) and 3.2 for anticonvulsants (1 study evaluating gabapentin).	The mean and median quality score for included studies was 4ona scale of 1–5
Fransen et al,51 2015	Exercise vs nonexercise group (active or no treatment)	Systematic review of 54 RCTs or quasi-randomized trials for knee osteoarthritis	Reduced joint pain or improved physical function and quality of life	Exercise reduced pain, improved function, and improved quality of life immediatelyaftertreatment; in studies providing posttreatment follow-up data, improved pain and function were sustained for 2–6 mo.	High-quality evidence for reduced pain and improved quality of life and moderate-quality evidence for improved function
Fransen et al,522014	Exercise vs nonexercise group (active or no treatment)	Systematic review of 10 RCTs or quasi-randomized trials for hip osteoarthritis	Reduced joint pain and improved physical function and quality of life	Exercise reduced pain and improved function immediatelyafter treatment; in studies providing posttreatment follow-up data, improved pain and function were sustained for at least 3–6 mo.	High-quality evidence for reduced pain and improved function
Häuser et al,53 2013	Duloxetine vs placebo; milnacipran vs placebo	Systematic review of 10 RCTs for fibromyalgia patients	Benefits and harms	Duloxetine and milnacipran reduced pain bya smallamount compared with placebo.	Risk of bias in included studies was low
Hayden et al,54 2005	Exercise therapy vs no treatment, other conservative treatments	Systematic review consisting of 61 RCTs for low back pain	Pain, function	Exercise therapy reduces pain and improves function with small magnitudes of effect. Effectiveness of exercise therapy appears to be greater in populations visiting a health care provider compared with the generalpopulation.	Only a small number of studies rated as high quality; potential publication bias
Lee et al,55 2014	CIM therapies vs single self-care CIM, non-self-care CIM, usualcare/no treatment, other multimodalprogram, or other control	Systematic review of 26 RCTs for management of chronic pain	Pain symptoms	Integrative multimodal therapies resulted in positive, but sometimes mixed, effects on pain symptoms compared with active controls or single self-care modalities. More studies are needed to make strong conclusions about effectiveness.	Large majority of poor quality, including weaknesses in randomization and allocation concealment
Lunn etal,56 2014	Duloxetine vs placebo or other controls	Systematic review of 18 RCTs for neuropathic pain, chronic pain conditions without identified cause, or fibromyalgia	Benefits and harms of duloxetine	Duloxetine at 60 mg and 120 mg daily, but not lower dosages, were effective in reducing pain in diabetic peripheralneuropathy pain and in fibromyalgia.	Moderate-quality evidence for diabetic neuropathy; lower-quality evidence for fibromyalgia; some risk of bias
Moore et al,57 2009	Pregabalin vs placebo or any active control	Systematic review of 25 double-blind RCTs for postherpetic neuralgia, painful diabetic neuropathy, central neuropathic pain, or fibromyalgia	Analgesic efficacy and associated adverse events	Pregabalin was effective in patients with postherpetic neuralgia, diabetic neuropathy, central neuropathic pain, and fibromyalgia at doses of 300 mg, 450 mg, and 600 mg (but not at 150 mg) daily. NNTs were generally ≤6 for moderate benefit in postherpetic neuralgia and diabetic neuropathy but ≥7 for fibromyalgia.	Studies all had Oxford quality scores based on randomization, blinding, and reporting of dropout ≥3 (out of maximum of 5)
Moore et al,58 2014	Gabapentin vs placebo	Systematic review of 37 RCTs for neuropathic pain or fibromyalgia	Analgesic efficacy and adverse effects	Gabapentin was significantly more effective than placebo in reducing pain in diabetic neuropathyand postherpeticneuralgia. Evidence was insufficient for other conditions.	“Second-tier” evidence (some risk of bias, but adequate numbers in the trials)
Roelofs et al,59 2008	NSAIDs and COX-2 inhibitors vs control	Systematic review of 65 RCTs for nonspecific low back pain	Acute low back pain	NSAIDs are more effective than placebo for acute and chronic low back pain without sciatica, but have more adverse effects. NSAIDs are not more effective than acetaminophen but had more adverse effects. No type of NSAIDs, including COX-2 inhibitors, was found to be more effective than other NSAIDs.	Mixed high- and low-quality studies
Saarto et al,60 2010	Antidepressants vs placebo or other controls	Systematic review of 61 RCTs for neuropathic pain	Pain	TCAs and venlafaxine have low NNTs (3.6 and 3.1, respectively) for at least moderate pain relief.	Study quality limited byinsufficient reporting detail
Salerno et al,61 2002	Antidepressants vs placebo	Systematic review of 9 RCTs for chronic back pain	Back pain	Antidepressants were associated with small but significant improvement in pain severity; improvements in function were not significant. Most (6) studies evaluated TCAs.	Moderate-quality studies
Staiger et al,62 2003	Antidepressants vs placebo	Systematic review of 7 RCTs in patients with chronic low back pain	Back pain	Four of 5 studies evaluating TCA and tetracyclic antidepressants found significant improvement in chronic low back pain. Other antidepressants studied (2 studies evaluating SSRIs and 1 evaluating trazodone) did not show significant pain improvement.	Mixed quality (quality scores ranged from 11–19 out of 22)
Trelle et al,63 2011	NSAIDs vs other NSAIDs or placebo	Meta-analysis of 31 RCTs comparing any NSAID with other NSAID or placebo for any medical condition	Myocardial infarction, stroke, cardiovascular death, death from any cause	Compared with placebo, NSAIDs were associated with increased risk of myocardial infarction, stroke, and cardiovascular death.	Generally high
Welsch et al,64 2015	Opioids (including tramadol) vs nonopioids (including acetaminophen, NSAIDs/COX-2 inhibitors, mexiletine, anticonvulsants, antidepressants, and muscle relaxants)	Systematic review of 10 RCTs in patients with neuropathic pain, low back pain, or osteoarthritis	Efficacy (including various pain measures), tolerability, and safety	There was no significant difference between opioids and nonopioid analgesics in pain reduction; nonopioids were superior to opioids in improving physical function and were better tolerated. When patients from tramadol trials (n randomized = 2788) were removed from results of the review, results for pain and function for patients receiving opioids (morphine) compared with alternative drugs (n randomized = 223) had wide, overlapping confidence intervals. Improved tolerability foralternative drugs vs morphine remained significant.	One study had a high, 2 studies a moderate, and 7 studies a low study quality
Wiffen et al,65 2014	Carbamazepine vs placebo or other active control	Systematic review consisting of 10 RCTs in adults with chronic neuropathic pain or fibromyalgia	Pain relief	Carbamazepine provided better pain relief than placebo for trigeminal neuralgia, diabetic neuropathy, and poststroke pain for ≤4 weeks. Dizziness and drowsiness were commonly reported with carbamazepine. In 4 studies, 65% of patients receiving carbamazepine vs 27% receiving placebo experienced ≥1 adverse event. In 8 studies, 3% of patients receiving carbamazepine withdrew because of adverse events (vs 0% taking placebo).	Third-tier evidence (trials involving small numbers of participants; considered likely to be biased, with outcomes of limited clinical utility, or both)
Williams et al,66 2012	Cognitive behavioral therapy or behavioral therapy	Systematic review of 42 RCTs for patients with nonmalignant chronic pain except headache	Pain, disability, mood, and catastrophic thinking	Cognitive behavioral therapy was found to have small to moderate effects on pain, disability, mood, and catastrophic thinking immediatelyafter treatment when compared with usual treatment or deferred cognitive behavioral therapy, but only effects on mood persisted at follow-up. Behavioral therapy had a positive effect on mood immediately after treatment.	Mean quality of study design, 15.8 out of 26 (SD 4.3; range, 9–24 out of 26)

Abbreviations: CIM, complementary and integrative multimodal; COX-2, cyclooxygenase 2; NNT, number needed to treat; NSAID, nonsteroidal anti-inflammatory drug; RCTs, randomized clinical trials; SSRIs, selective serotonin reuptake inhibitors; TCA, tricyclic antidepressants.

^aAll the studies in this table were included in the contextual evidence review.