PURPOSE: Opioids are the primary choice for managing chronic cancer pain. However, many nonopioid therapies are currently prescribed for chronic cancer pain with little published evidence comparing their efficacy.
METHODS: Electronic databases were searched for randomized controlled trials (RCTs) comparing any systemic pharmaceutical intervention and/or combination thereof in treating chronic cancer pain. The primary outcome was global efficacy reported as an odds ratio (OR). The secondary outcome was change in pain intensity reported as a standardized mean difference (SMD).
RESULTS: We included 81 RCTs consisting of 10,003 patients investigating 11 medication classes. Most RCTs (80%) displayed low risk of bias. The top-ranking classes for global efficacy were nonopioid analgesics (network OR, 0.30; 95% credibility interval [CrI], 0.13 to 0.67), nonsteroidal anti-inflammatory drugs (network OR, 0.44; 95% CrI, 0.22 to 0.90), and opioids (network OR, 0.49; 95% CrI, 0.27 to 0.86), whereas the top-ranked interventions were lidocaine (network OR, 0.04; 95% CrI, 0.01 to 0.18; surface under the cumulative ranking curve analysis [SUCRA] score, 98.1), codeine plus aspirin (network OR, 0.22; 95% CrI, 0.08 to 0.63; SUCRA score, 81.1), and pregabalin (network OR, 0.29; 95% CrI, 0.08 to 0.92; SUCRA score, 73.8). In terms of reducing pain intensity, we found that no class was superior to placebo, whereas the following top-ranked interventions were superior to placebo: ziconotide (network SMD, -24.98; 95% CrI, -32.62 to -17.35; SUCRA score, 99.8), dezocine (network SMD, -13.56; 95% CrI, -23.37 to -3.69; SUCRA score, 93.5), and diclofenac (network SMD, -11.22; 95% CrI, -15.91 to -5.80; SUCRA score, 92.9).
CONCLUSION: There are significant differences in efficacy among current regimens for chronic cancer pain. Our evidence suggests that certain nonopioid analgesics and nonsteroidal anti-inflammatory drugs can serve as effectively as opioids in managing chronic cancer pain.
This study adds value to the therapy choice of cancer patients. In addition, the Bayesian meta-analysis technique is interesting.
The challenge with using the conclusions from this paper in clinical practice is that the paper's methodology is fairly opaque. Most of us are not very familiar with network meta-analysis techniques, which makes it difficult to say how much confidence to put in the result.
I struggled to rank this effectively as my understanding of the statistical methods is limited (which I suspect may be true for many others in my field) but my main concern is that I fail to understand how one can view as a homogeneous group (as implied in the discussion) a group containing oral acetaminophen, ??iv/po/sc ketamine, po nimodipine and iv lidocaine....these have radically different mechanisms of action were undoubtedly used in very different situations to conclude 'nonopioid analgesics were the best'. This seems very suspect. Equally so, the choices in 'group M' are again very divergent and I struggle to see how the article can claim class effects/lack of effect. As such, I do not feel I can rank the article highly though, I agree, it raises intriguing points about the importance of looking beyond opioids.
As explained in the editorial that accompanies the article, attention to the triad of common sense clinical practice—efficacy, feasibility, and risk avoidance—will protect us from misinterpreting high-level scientific evidence that could lead us down the wrong path.