We identified six randomised, double-blind studies involving 438 participants with suitably characterised neuropathic pain. In each, tramadol was started at a dose of about 100 mg daily and increased over one to two weeks to a maximum of 400 mg daily or the maximum tolerated dose, and then maintained for the remainder of the study. Participants had experienced moderate or severe neuropathic pain for at least three months due to cancer, cancer treatment, postherpetic neuralgia, peripheral diabetic neuropathy, spinal cord injury, or polyneuropathy. The mean age was 50 to 67 years with approximately equal numbers of men and women. Exclusions were typically people with other significant comorbidity or pain from other causes. Study duration for treatments was four to six weeks, and two studies had a cross-over design.
Not all studies reported all the outcomes of interest, and there were limited data for pain outcomes. At least 50% pain intensity reduction was reported in three studies (265 participants, 110 events). Using a random-effects analysis, 70/132 (53%) had at least 50% pain relief with tramadol, and 40/133 (30%) with placebo; the risk ratio (RR) was 2.2 (95% confidence interval (CI) 1.02 to 4.6). The NNT calculated from these data was 4.4 (95% CI 2.9 to 8.8). We downgraded the evidence for this outcome by two levels to low quality because of the small size of studies and of the pooled data set, because there were only 110 actual events, the analysis included different types of neuropathic pain, the studies all had at least one high risk of potential bias, and because of the limited duration of the studies.
Participants experienced more adverse events with tramadol than placebo. Report of any adverse event was higher with tramadol (58%) than placebo (34%) (4 studies, 266 participants, 123 events; RR 1.6 (95% CI 1.2 to 2.1); NNH 4.2 (95% CI 2.8 to 8.3)). Adverse event withdrawal was higher with tramadol (16%) than placebo (3%) (6 studies, 485 participants, 45 events; RR 4.1 (95% CI 2.0 to 8.4); NNH 8.2 (95% CI 5.8 to 14)). Only four serious adverse events were reported, without obvious attribution to treatment, and no deaths were reported. We downgraded the evidence for this outcome by two or three levels to low or very low quality because of small study size, because there were few actual events, and because of the limited duration of the studies.