LAUSR

Question In patients with chronic noncancer pain, what are the benefits and harms of opioids? Review scope Included studies compared oral or transdermal opioids with nonopioids in patients with chronic noncancer pain. Studies of interventions that are rarely used for chronic noncancer pain in North America were excluded. Outcomes included pain; physical, emotional, role, and social functioning; sleep quality; and vomiting. PROSPERO CRD42012003023. Review methods MEDLINE, EMBASE/Excerpta Medica, Cochrane CENTRAL, AMED, CINAHL, PsycINFO (all to Apr 2018), and reference lists were searched for randomized controlled trials (RCTs) with follow-up 4 weeks. 96 RCTs (n =26169; mean/median age 40 to 78 y when reported; 61% women in 91 trials) met the eligibility criteria. Pain conditions were central sensitization (33 RCTs), nociceptive pain (32 RCTs), neuropathic pain (25 RCTs), and mixed types of chronic pain (6 RCTs). The median of the average morphine-equivalent doses was 45.0 mg/d where reported; median follow-up was 60 days. {80 trials}* compared opioids with placebo. 51 RCTs adequately generated randomization sequences, 48 had adequate allocation concealment, 84 blinded patients, and 82 blinded outcome assessors; 73 trials had 20% missing outcome data. Quality of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation tool. Main results The main efficacy results for comparisons of opioids with placebo are in the Table. Opioids increased risk for vomiting compared with placebo in trials that excluded patients with problematic adverse events and/or poor response to open-label treatment before randomization (5.9% vs 2.3%, relative risk [RR] 2.50, 95% CI 1.89 to 3.30) and those that did not (9.4% vs 2.3%, RR 4.12, CI 3.34 to 5.07). Meta-analyses of opioids vs active comparators were of low to moderate quality of evidence and are not reported here. Conclusion Opioids improve chronic noncancer pain more than placebo, but important pain reduction over the long term is unlikely. Opioids vs placebo in patients with chronic noncancer pain Outcomes Assessment period (mo) Number of trials (n) Weighted between-group mean difference (95% CI) Modeled risk difference for achieving the MCID (CI) Pain (10 cm VAS) (cm) 3 to 6 42 (16617) 0.69 (0.82 to 0.56) 11.9% (9.7 to 14.1) Physical functioning (SF-36 physical component) (points) 1 to 6 51 (15574) 2.04 (1.41 to 2.68) 8.5% (5.9 to 11.2) Emotional functioning (SF-36 mental component) (points) 1 to 4 23 (8962) 0.44 (1.09 to 0.20) 1.7% (4.2 to 0.8) Role functioning (SF-36 subscale) (points)** 1 to 4 16 (5329) 0.87 (0.54 to 2.28) 1.0% (0.7 to 2.6) Social functioning (SF-36 subscale) (points) 1 to 4 29 (7623) 1.58 (0.45 to 2.70) 2.6% (0.7 to 4.5) Sleep quality (SF-36 sleep quality 100 mm VAS) (mm) 3 to 6 15 (6585) 3.42 (1.58 to 5.26) 5.9% (2.8 to 9.1) MCID = minimal clinically important difference; SF-36 =36-item Short Form Survey; VAS = visual analogue scale; CI defined in Glossary. All evidence was high quality (Grading of Recommendations Assessment, Development and Evaluation criteria). The difference between the proportions of patients who achieved the MCID in each group; positive numbers favor the opioid group. For pain at <3 mo, between-group mean difference is 0.97 (CI 1.16 to 0.78); subgroup analysis indicates the benefit decreased over time (P for interaction =0.04). Score range 0 to 10; higher scores = greater pain; MCID =1 cm. Difference did not reach the cutpoint for the MCID. Score range 0 to 100; higher scores = better functioning; MCID =5 points. **SF-36 subscale for role limitations due to physical problems. Score range 0 to 100; higher scores = better functioning; MCID =10 points. SF-36 subscale for social functioning. Score range 0 to 100; higher scores = better functioning; MCID =10 points. For sleep quality at <3 mo, between-group mean difference is 7.51 (CI 4.89 to 10.12); subgroup analysis indicates the benefit decreased over time (P for interaction =0.03). Score range 0 to 100 mm; higher scores = better sleep quality; MCID =10 mm. Commentary The systematic review by Busse and colleagues found that opioids reduced pain and improved physical functioning and sleep quality compared with placebo, although the average effects were small and did not meet prespecified MCID thresholds. Opioids were also associated with increased risk for vomiting, nausea, constipation, drowsiness, and other adverse events. The review had some important limitations. Because opioids cause tolerance and physical dependence, short-term trials are inadequate to understand long-term effects, but no included trial was longer than 6 months. Most trials excluded patients with substance use disorders or other psychiatric comorbidities. Moreover, >33% of included trials used enriched-enrollment randomized withdrawal designs, which resulted in inclusion of highly selected patients who tolerated and responded well to opioids. The review set the pain MCID threshold at 1 cm on a 0- to 10-cm scale, which is lower than other published estimates (1) and may further diminish the clinical meaningfulness of effects. Finally, the review assumed a normal distribution of outcomes to model the proportion of patients meeting MCID thresholds, but responses to pharmacological pain treatments may be bimodal or otherwise skewed (2). The systematic review compared opioids with active nonopioid comparators as well as placebo, but trials with nonopioids were few and most estimates were imprecise. A recent trial not included in the review found no difference between stepped therapy with opioids versus stepped therapy initiated with nonopioid medications in pain or function at 1 year, underscoring the importance of long-term head-to-head trials (3).