Eleven trials (n=2,709) were included: three of duloxetine (n=1,018); six of pregabalin (n=1,466); and two of gabapentin (n=225).
Direct comparisons:
All three drugs showed statistically significant benefits compared with placebo for efficacy outcomes.
Mean differences in 24-hour pain severity compared with placebo were: duloxetine -1.13 (95% CI -1.36 to -0.89, I2=0%), gabapentin -1.44 (95% CI -2.21 to -0.66, I2=34%) and pregabalin -0.90 (95% CI -1.23 to -0.57; I2=46%).
For treatment response, log odds ratios were: duloxetine 0.86 (95% CI 0.63 to 1.09; estimated NNT 5, 95% CI 3 to 7) and gabapentin 0.84 (95% CI 0.52 to 1.16; estimated NNT 5, 95% CI 4 to 8). Between-study variance was low (range zero to 0.154).
For overall health improvement, log odds ratios were: duloxetine -0.76 (95% CI -1.00 to -0.51) and gabapentin -1.29 (95% CI -1.72 to -0.86). Between-study variance was low (range zero to 0.019).
For tolerability, duloxetine significantly reduced discontinuation due to lack of efficacy compared with placebo (log OR -0.96, 95% CI -1.80 to -0.12) and so did pregabalin (log OR -0.71, 95% CI -1.21 to -0.22). There was no difference between gabapentin and placebo. Both duloxetine and pregabalin led to a significant increase in discontinuation due to adverse events. There was no difference between gabapentin and placebo. Other results for individual types of adverse event were reported in the paper.
Indirect comparisons:
For reduction in 24-hour pain severity, duloxetine was not inferior to pregabalin (MD -0.248, 95% CrI -0.667 to 0.162), but was superior to pregabalin for health improvement (log OR 0.54, 95% CrI 0.02 to 1.06). There was no difference for treatment response. Duloxetine resulted in a significantly lower incidence of dizziness than pregabalin. There were no differences for other safety outcomes or between duloxetine and gabapentin for any efficacy or safety outcome.