Fourteen trials reported in 17 publications (more than 2,617 participants) were included in the review. At most, three of the six quality criteria were met by individual studies.
Fentanyl patches versus buprenorphine patches: Fentanyl patches were associated with more nausea (OR 4.86, 95% CI 2.14 to 11.07) and vomiting (OR 17.32, 95% CI 4.43 to 67.71) and a higher risk of treatment discontinuations due to adverse events (OR 4.37, 95% CI 1.81 to 10.50).
Buprenorphine patches versus morphine: Buprenorphine patches were associated with better control of pain intensity (MD -16.20, 95% CI -28.92 to -3.77); morphine was associated with a greater risk of constipation (OR 5.63, 95% CI 1.73 to 18.29), nausea (OR 4.23, 95% CI 1.83 to 9.79), vomiting (OR 15.85, 95% CI 3.92 to 64.13) and a higher risk of treatment discontinuations due to adverse events (OR 4.26, 95% CI 1.68 to 10.81).
Placebo versus buprenorphine patches: Placebo was associated with a higher risk of treatment discontinuations due to lack of effect (OR 3.31, 95% CI 1.60 to 6.88) or any cause (OR 1.93, 95% CI 1.36 to 2.72).
Differences in other outcomes for each drug comparison, where they could be estimated, were not statistically significant between treatment arms.
Results from the enriched and non-enriched subgroup analyses supported the full analyses.