Skip navigation
Acute Pain | Chronic Pain | General

Anticonvulsants for diabetic neuropathy and postherpetic neuralgia

Clinical bottom line

Anticonvulsants are effective treatments for diabetic neuropathy (NNT 2.7, 95% confidence interval 2.2 to 3.8) and postherpetic neuralgia (NNT 2.9, 95% confidence interval 2.4 to 3.7) compared with placebo, based on four trials. The NNH for any patient to have a minor adverse effect with anticonvulsants was 2.7 (2.2 to 3.4).

For over thirty years the management of neuropathic pain has involved the use of both antidepressants and anticonvulsants, but which drug class should be first line choice remains unclear. The dogma that the character of the pain was predictive of the response, burning pain responding to antidepressants and shooting pain to anticonvulsants, was shown to be incorrect in diabetic neuropathy, where patients experiencing both burning and shooting pain responded to tricyclic antidepressants. Most studies on neuropathic pain have involved diabetic neuropathy and postherpetic neuralgia, because these conditions represent the majority of patients with neuropathic pain.


SL Collins et al. Antidepressants and anticonvulsants for diabetic neuropathy and postherpetic neuralgia: a quantitative systematic review. Journal of Pain and Symptom Management 2000 20: 449-458.

Searching involved numerous electronic databases including MEDLINE, EMBASE, CINAHL, Sigle, PubMed and the Cochrane Library, plus in-house databases of randomised controlled trials in pain. The inclusion criteria used were: full journal publication, adult patients, double-blind design, random allocation to treatment groups which included placebo and either an antidepressant or an anticonvulsant for the treatment of chronic pain due to diabetic neuropathy or postherpetic neuralgia. An adequate description of the original authors’ method of clinical diagnosis was required to ensure an accurate diagnosis of the two conditions. For postherpetic neuralgia the pain must have been present for more than three months after zoster eruption to limit the chance of spontaneous cessation of symptoms.

A clinically relevant outcome was defined as a measure equivalent to at least 50% pain relief after the longest reported duration of treatment. This was extracted as dichotomous information from the following hierarchy of outcome measures:

The majority of studies used a cross-over design and results are presented in terms of patient episodes rather than actual numbers of patients. One patient episode represents the result for one patient completing one part of the cross-over. So for a trial where the patient was crossed-over from placebo to active this would generate two patient episodes.

Adverse effects were classified as minor if reported by a patient who then continued to take the medication and completed the trial. A major adverse effect was one causing the patient to withdraw from the study. Withdrawal due to lack of efficacy was not counted as an adverse effect.


The studies of anticonvulsants in diabetic neuropathy and postherpetic neuralgia were of medium size, ranging from 76 to 225 patient episodes in total. Consequently the results of individual trials varied greatly (Figure 1).

Figure 1: Trials of anticonvulsants versus placebo in diabetic neuropathy (red) and postherpetic neuralgia (white)

Pooled results are shown in Table 1. Over both conditions, 54% of patients had an outcome equivalent to more than 50% pain relief with anticonvulsant, compared with 20% with placebo. The NNT for at least 50% pain relief compared with placebo was 2.9 (2.4 to 3.7). For every patient who received benefit, one also had a minor adverse effect that did not lead to discontinuation. Nearly 1 patient in 8 discontinued treatment because adverse effects were intolerable, though this was not significantly different from placebo.

Table 1: Summary results of efficacy and harm for anticonvulsants in diabetic neuropathy and postherpetic neuralgia

Number of patients improved or harmed
number/total (%)
Trials Patient episodes Anticonvulsant Placebo Relative benefit (95% CI) NNT
(95% CI)
Diabetic neuropathy 3 321 100/161 (62%) 41/160 (26%) 2.4 (1.8 - 3.2) 2.7 (2.2 - 3.8)
Postherpetic neuralgia 1 225 47/109 (43%) 14/116 (12%) 3.6 (2.1 - 6.1) 3.2 (2.4 - 5.0)
Both conditions 4 546 147/270 (54%) 55/276 (20%) 2.7 (2.1 - 3.5) 2.9 (2.4 - 3.7)
Harm NNH
(95% CI
Minor adverse effect 4 492 153/24 (62%) 60/245 (24%) 2.5 (2 - 3.1) 2.7 (2.2-3.4)
Major adverse effect 5 548 32/276 (12%) 19/272 (7%) 1.6 (0.97 - 2.7) N/A


This review updated previous reviews and confirmed that anticonvulsants are effective in patients with diabetic neuropathy and postherpetic neuralgia. The degree of benefit, an outcome equivalent to at least 50% pain relief, was valuable, and with an NNT of 3, efficacy was equivalent to that seen in effective analgesics in acute pain (10 mg intramuscular morphine, for instance). Over half of patients given anticonvulsants benefited. For every patient who benefited, one had a minor adverse effect, but continued with the treatment. There was no overall difference between results for anticonvulsants and antidepressants.