We used standard methodological procedures expected by Cochrane.
MAIN RESULTS: Five multicentre, randomised, placebo-controlled, double-blind trials with a total of 862 participants were eligible for inclusion in this updated review. Three trials involved participants with painful diabetic
peripheral neuropathy (
DPN) (n = 634), one included people with
neuropathic pain due to
radiculopathy (n = 145), and one, which was newly identified at this update, involved participants with peripheral
neuropathic pain of mixed origin (
polyneuropathy,
peripheral nerve injury or
postherpetic neuralgia) (n = 83). Some studies did not report all outcomes of interest. For painful
DPN, compared to the baseline, the proportion of participants who reported at least a 50% or 30% reduction of
pain scores after 16 weeks of treatment in the
oxcarbazepine group versus the placebo group were: at least 50% reduction: 34.8% with
oxcarbazepine versus 18.2% with placebo (risk ratio (RR) 1.91, 95% confidence interval (CI) 1.08 to 3.39, number of people needed to treat for an additional beneficial outcome (NNTB) 6, 95% CI 3 to 41); and at least 30% reduction: 44.9% with
oxcarbazepine versus 28.6% with placebo (RR 1.57, 95% CI 1.01 to 2.44; NNTB 6, 95% CI 3 to 114; n = 146). Both results were based on data from a single trial, since two trials that found little or no benefit did not provide data that could be included in a meta-analysis. Although these trials were well designed, incomplete outcome data and possible unblinding of participants due to obvious adverse effects placed the results at a high risk of bias. There was also serious imprecision and a high risk of publication bias. The
radiculopathy trial reported no benefit for the outcome 'at least 50%
pain relief' from
oxcarbazepine. In mixed neuropathies, 19.3% of people receiving
oxcarbazepine versus 4.8% receiving placebo had at least 50%
pain relief. These small trials had low event rates and provided, at best, low-quality evidence for any outcome. The proportion of people with 'improved' or 'very much improved'
pain was 45.9% with
oxcarbazepine versus 30.1% with placebo in
DPN (RR 1.46, 95% CI 1.13 to 1.88; n = 493; 2 trials; very-low-quality evidence) and 23.9% with
oxcarbazepine versus 14.9% with placebo in
radiculopathy (RR 1.61, 95% CI 0.81 to 3.20; n = 145).We found no trials in other types of
neuropathic pain such as
trigeminal neuralgia.Trial reports stated that most adverse effects were mild to moderate in severity. Based on moderate-quality evidence from the three
DPN trials, serious adverse effects occurred in 8.3% with
oxcarbazepine and 2.5% with placebo (RR 3.65, 95% CI 1.45 to 9.20; n = 634; moderate-quality evidence). The number needed to treat for an additional harmful (serious adverse effect) outcome (NNTH) was 17 (95% CI 11 to 42). The RR for serious adverse effects in the
radiculopathy trial was 3.13 (95% CI 0.65 to 14.98, n = 145). The fifth trial did not provide data.More people withdrew because of adverse effects with
oxcarbazepine than with placebo (
DPN: 25.6% with
oxcarbazepine versus 6.8% with placebo; RR 3.83, 95% CI 2.29 to 6.40;
radiculopathy: 42.3% with
oxcarbazepine versus 14.9% with placebo; RR 2.84, 95% CI 1.55 to 5.23; mixed
neuropathic pain: 13.5% with
oxcarbazepine versus 1.2% with placebo; RR 11.51, 95% CI 1.54 to 86.15).
AUTHORS' CONCLUSIONS: This review found little evidence to support the effectiveness of
oxcarbazepine in
painful diabetic neuropathy,
neuropathic pain from
radiculopathy and a mixture of neuropathies. Some very-low-quality evidence suggests efficacy but small trials, low event rates, heterogeneity in some measures and a high risk of publication bias means that we have very low confidence in the measures of effect. Adverse effects, serious adverse effects and adverse effects leading to discontinuation are probably more common with
oxcarbazepine than placebo; however, the numbers of participants and event rates are low. More well-designed, multicentre RCTs investigating
oxcarbazepine for various types of
neuropathic pain are needed, and selective publication of studies or data should be avoided.