Eighteen studies (n=1,181) met the inclusion criteria.
Thirteen studies achieved the maximum Jadad score of 5. Eighty-nine per cent of studies fulfilled the quality criteria of describing the randomisation method and double-blinding. Seventy-eight per cent accounted for drop-outs, but only 56% reported adequate allocation concealment.
Gabapentin was associated with a 35% reduction in analgesic use in the first 24 hours post-operatively (ratio of means 0.65, 95% CI: 0.59, 0.72, p<0.001). However, there was evidence of significant heterogeneity (I-squared 84.4%), which was not accounted for by surgical procedure, dosage or study quality. Gabapentin delayed time to first analgesic by 7.9 minutes (WMD 7.9, 95% CI: 4.2, 11.6, p<0.001). There was no evidence of statistical heterogeneity.
Gabapentin significantly reduced pain by 27% (95% CI: 6.8 mm, 15.8 mm) at rest 2 hours post-operatively and by 39% (95% CI: 8.5 mm, 20.2 mm) at rest 4 hours post-operatively. This reduction in pain was maintained at 12 and 24 hours post-operatively. However, there was evidence of statistical heterogeneity in measurements at 12 hours (I-squared 73.9%) and 24 hours (I-squared 64.7%). With the exception of 24 hours after surgery, gabapentin use was also associated with a significant reduction in pain with movement, ranging from 18 to 28% after surgery. There was, however, evidence of significant heterogeneity in the 12 hour post-operative measurements (I-squared 71.9%).
Gabapentin was associated with a significantly increased risk of dizziness (RR 1.40, 95% CI: 1.06, 1.84, p<0.02) and a decreased risk of pruritus (RR 0.30, 95% CI: 0.13, 0.70, p<0.005) and vomiting (RR 0.73, 95% CI: 0.56, 0.95, p<0.02). There was no evidence of statistical heterogeneity for these analyses. There was also a borderline increased risk of sedation with gabapentin (RR 1.65, 95% CI: 1.00, 2.74, p=0.05), although there was evidence of significant heterogeneity in this analysis (I-squared 83.3%). The use of gabapentin did not have a significant impact on the occurrence of respiratory depression or nausea.
The results of the review did not change when only high-quality studies or studies with severe post-operative pain were included.
There was no evidence of publication bias.