NPC Archive Item: All non-analgesic drug classes are similarly effective in the treatment of neuropathic pain

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo

26th January 2009

A Canadian Health Technology Assessment concludes that, in patients with neuropathic pain, there is no statistically significant difference in clinical response rates between tricyclic antidepressants, anticonvulsants and serotonin-norepinephrine reuptake inhibitors (SNRIs).

No significant difference was seen between the three drug classes in terms of 30% reduction in pain, 50% reduction in pain, or dropouts due to adverse effects. As with all medicines, where there is no evidence to distinguish treatments regarding safety or effectiveness, choice should be based on both individual patient preference and cost. As they are likely to be the least costly, the meta-analysis suggests that tricyclic antidepressants should be the first-line treatment option in patients whose neuropathic pain is not controlled using simple analgesia.

What does this study claim?
This meta-analysis (28 RCTs) assessed the clinical response rates in adults diagnosed with neuropathic pain who were taking tricyclic antidepressants (e.g. amitriptyline), SNRIs (duloxetine▼, venlafaxine) or anticonvulsants (gabapentin, pregabalin▼) compared with placebo.

A partial response (30% reduction in pain on a visual analogue scale) was seen in 66.0% (95% CI 61.7 to 70.4) of patients taking an SNRI, 54.5% (95% CI 47.3 to 61.7) of patients taking an anticonvulsant, and 49.2% (95% CI 36.3 to 62.2) of patients taking a tricyclic antidepressant. For full response (50% reduction in pain on a visual analogue scale) the rate was 45.9% (95% CI 41.3 to 50.4) for SNRIs and 36.3% (95% CI 30.1 to 42.5) for anticonvulsants. No rates were available for tricyclic antidepressants.

Note that as the 95% confidence intervals for all of these overlap, there is no statistically significant difference between them.

The placebo response rates varied between the studies included in the meta-analysis. This may have been due to chance or because patients recruited into some studies had more or less severe disease than those in other studies. Because of this, response rates were adjusted for placebo effect. When this was done the response rate for tricyclic antidepressants became highest for both outcomes. However, statistical analyses could not identify any significant difference between the response rates in the three drug classes.

The dropout rates due to adverse drug reactions were similar between the three drug classes: 12.3% for anticonvulsants, 12.0% for SNRIs and 11.7% for tricyclic antidepressants.

So what?
The meta-analysis has several limitations, mainly relating to the quality of the included RCTs. Compared with studies looking at SNRIs and anticonvulsants, trials of tricyclic antidepressants were smaller, older and of a lower quality. In addition, the average pain score was lower at baseline than those from the analyses of the other two drug classes, and the average trial length was shorter. However, even the more recent higher quality studies of the newer agents have their limitations. For example, trial populations exclude patients with co-morbid conditions and concomitant medications that might interfere with interpretation of efficacy or safety data, and there are few head-to-head comparisons of relative effectiveness of different drugs and drug classes. In addition, mean duration of follow-up in the studies included in the meta-analysis is less than 10 weeks so we do not know how effective these drugs will in the long-term for what is frequently a chronic condition.

Uncertainty remains about the optimal treatment for individual patients. This meta-analysis shows that there is little to choose between the pain response rates and adverse effect dropout rates of SNRIs, anticonvulsants and tricyclic antidepressants. Because the evidence to support the use of treatments for neuropathic pain is generally limited, treatment should be tailored to the individual’s circumstances, taking into account any contra-indications, co-morbidities etc. It is difficult to predict how a person will respond and trials of several drugs may be necessary to obtain optimal pain relief and minimise adverse effects. However, evidence supports tricyclic antidepressants as a first-line treatment of neuropathic pain and they are likely to be more cost-effective than SNRIs or anticonvulsants.

A NICE clinical guideline on the pharmacological management of neuropathic pain is currently under development but is not expected to be published until 2010. However, in their clinical guideline on the management of type 2 diabetes, NICE recommend that a tricyclic antidepressant should be used to treat neuropathic discomfort if standard analgesia is ineffective. Treatment should be initiated at a low dose and titrated upwards as tolerated. If a trial of tricyclic antidepressant does not provide effective pain relief, the patient should be offered a trial of duloxetine▼, gabapentin or pregabalin▼.

Guidance on the management of several types of neuropathic pain is available on the NHS Clinical Knowledge Summaries website e.g. neuropathic pain – adjunct drugs, post-herpetic neuralgia, shingles and trigeminal neuralgia. The neuropathic pain section of NPC is currently under development.

Study details

Canadian Agency for Drugs and Technologies in Health. Anticonvulsants, Serotonin-Norepinephrine Reuptake Inhibitors, and Tricyclic Antidepressants in Management of Neuropathic Pain: A Meta-Analysis and Economic Evaluation. Technology Report 116. December 2008

Design: Meta-analysis of 28 RCTs that compared the safety and efficacy of various non-analgesic drug treatments with placebo in adults with neuropathic pain.

Patients, intervention and comparison: Thirteen studies (n=1,257) evaluated anticonvulsants (gabapentin, pregabalin▼), 5 (n=781) studied SNRIs (duloxetine▼, venlafaxine) and 10 (n=249) focused on tricyclic antidepressants (e.g. amitriptyline). One study evaluated both tricyclic antidepressants and SNRIs.

Outcomes: Two main outcomes were considered: proportion of patients who achieved a 30% decrease in pain on a visual analogue scale and the proportion achieving a 50% decrease.

Results: The mean duration of follow-up was 9.1 weeks for anticonvulsants, 9.9 weeks for SNRIs and 6.6 weeks for tricyclic antidepressants. The average baseline pain score on the visual analogue scale was 6.6 for anticonvulsants, 6.2 for SNRIs and 5.2 for tricyclic antidepressants.

Unadjusted rate (standard error [SE]) (95% CI)

Adjusted rate ‘from placebo’† (95% CI)

Adjusted rate ‘through placebo’† (95% CI)

30% response


66.0% (2.2%)
(61.7% to 70.4%)

(43.4% to 56.0%)

(36.3% to 44.9%)


54.5% (3.7%)
(47.3% to 61.7%)

(49.9% to 59.0%)

(46.3% to 60.7%)

Tricyclic antidepressant

49.2% (6.6%)
(36.3% to 62.2%)

(42.4% to 76.5%)

(54.4% to 80.3%)

50% response


45.9% (2.3%)
(41.3% to 50.4%)

(33.1% to 43.5%)

(29.0% to 38.1%)


36.3% (3.2%)
(30.1% to 42.5%)

(38.3% to 46.3%)

(45.6% to 58.0%)

Tricyclic antidepressant

No data

(32.8% to 59.2%)

(47.4% to 73.4%)

* Because there were no studies that reported 50% response rate for tricyclic antidepressants this adjusted rate was pro-rated from 30% response rate data.

† ‘From placebo’ and ‘through placebo’ are two different approaches used to adjust for the placebo effect.

Meta-analytic dropout rates were 12.3% (SE 1.8%) for anticonvulsants, 12.0% (SE 2.3%) for SNRIs and 11.7% (SE 4.4%) for tricyclic antidepressants.

Sponsorship: Health Canada and various Canadian provincial and territorial governments

Please comment on this blog in the NPC discussion rooms, or using our feedback form