Article type
Year
Abstract
Background: Opioids are often prescribed for chronic noncancer pain (CNCP). Thus far, systematic reviews have pooled different opioids assuming a common effect; however, the comparative effectiveness of individual opioids, and of long-acting (LA) versus short-acting (SA) opioids, on pain relief is uncertain.
Objectives: We conducted a network meta-analysis (NMA) to inform whether pooling opioids is a source of heterogeneity or not.
Methods: We searched EMBASE, MEDLINE, CINAHL, AMED, PsycINFO and the Cochrane Central Registry of Controlled Trials, from inception to March 2019, to identify trials that randomized CNCP patients to an oral or transdermal opioid vs. placebo, or other opioids, and followed patients for ≥4 weeks. We performed a frequentist NMA exploring effects on a 10 cm visual analogue scale for pain (1 cm is the minimally important difference) and assessed the quality of evidence using the GRADE approach (dichotomized as “moderate-to-high” or “low-to-very-low” quality of evidence). We estimated ranking probabilities using the surface under the cumulative ranking curve (SUCRA), which assigns the
probability for each opioid to be the best. We categorized opioids first based on their effectiveness vs. placebo and then vs. other competing opioids, and finally according to GRADE quality of evidence
ratings.
Results: We included 76 studies with 21,752 patients that evaluated 15 individual opioids, of which 4 were SA and 11 were LA. The SUCRA values suggested codeine-extended release (ER) (94.2%) and
oxymorphone-ER (88.1%) as the best opioids for pain relief. The certainty of evidence for both these drugs relative to placebo was, however, low. All comparisons supported by moderate-to-high quality evidence demonstrated that opioids were more effective than placebo, but that none were superior to others. Low certainty evidence suggested a statistically significant, but clinically unimportant, advantage of SA vs. LA opioids for pain relief (weighted mean difference 0.18cm, 95% CI: 0.06, 0.29).
Conclusions: Our findings suggest that apparent differences in effectiveness between opioids, when ranked according to SUCRA values, result from the failure to consider the certainty of evidence.
Patient or healthcare consumer involvement: Using the minimally contextualized approach for interpreting results of the NMA highlights its advantages relative to relying solely or largely on SUCRA values.
Objectives: We conducted a network meta-analysis (NMA) to inform whether pooling opioids is a source of heterogeneity or not.
Methods: We searched EMBASE, MEDLINE, CINAHL, AMED, PsycINFO and the Cochrane Central Registry of Controlled Trials, from inception to March 2019, to identify trials that randomized CNCP patients to an oral or transdermal opioid vs. placebo, or other opioids, and followed patients for ≥4 weeks. We performed a frequentist NMA exploring effects on a 10 cm visual analogue scale for pain (1 cm is the minimally important difference) and assessed the quality of evidence using the GRADE approach (dichotomized as “moderate-to-high” or “low-to-very-low” quality of evidence). We estimated ranking probabilities using the surface under the cumulative ranking curve (SUCRA), which assigns the
probability for each opioid to be the best. We categorized opioids first based on their effectiveness vs. placebo and then vs. other competing opioids, and finally according to GRADE quality of evidence
ratings.
Results: We included 76 studies with 21,752 patients that evaluated 15 individual opioids, of which 4 were SA and 11 were LA. The SUCRA values suggested codeine-extended release (ER) (94.2%) and
oxymorphone-ER (88.1%) as the best opioids for pain relief. The certainty of evidence for both these drugs relative to placebo was, however, low. All comparisons supported by moderate-to-high quality evidence demonstrated that opioids were more effective than placebo, but that none were superior to others. Low certainty evidence suggested a statistically significant, but clinically unimportant, advantage of SA vs. LA opioids for pain relief (weighted mean difference 0.18cm, 95% CI: 0.06, 0.29).
Conclusions: Our findings suggest that apparent differences in effectiveness between opioids, when ranked according to SUCRA values, result from the failure to consider the certainty of evidence.
Patient or healthcare consumer involvement: Using the minimally contextualized approach for interpreting results of the NMA highlights its advantages relative to relying solely or largely on SUCRA values.