This review concluded that a 66% increase in published trials since 2005 showed only a limited improvement in neuropathic pain. A large proportion of patients with neuropathic pain had insufficient pain relief. Limitations of the statistical methods and heterogeneity between studies suggest the results should be viewed with caution when comparing the effectiveness of types of interventions.

To investigate the efficacy and safety of treatments for neuropathic pain conditions by updating a previous systematic review.

MEDLINE and EMBASE were searched from April 2005 to April 2010 for studies in English. Search terms were not reported and the authors reported that a free text search was used. Further papers were identified from published reviews and references of retrieved studies. Clinical Study Results Database was searched to May 2009. Retrieved studies were added to the studies from the previous review that included publications up to April 2005 (see Other Publications of Related Interest).

Randomised controlled trials (RCTs) that were double-blinded and compared pharmacological interventions for the treatment of neuropathic conditions to a placebo control were eligible for inclusion in the review. Trials needed to include at least 10 patients and report pain as the primary outcome. No minimum follow-up was required. Trials could compared first-line drugs with each other (tricyclic antidepressants, serotonin noradrenaline reuptake inhibitors), topical lidocaine, gabapentin and pregabalin for neuropathic pain. Various named neuropathic pain conditions were eligible for inclusion. Enriched-enrolment and pre-emptive studies were excluded. Studies on radiculopathy, complex regional pain syndrome and cancer neuropathic pain were excluded.

The included studies investigated painful poly-neuropathy (mainly due to diabetes), postherpetic neuralgia, peripheral nerve injury, HIV neuropathy, central pain, trigeminal neuralgia and mixed neuropathic pain. The drug classes investigated were antidepressants, anticonvulsants, opioids and other. The primary outcome of interest was the number needed to treat for 50% reduction in pain (alternatively 30% reduction in pain or at least good pain relief).

The authors did not state how many reviewers performed study selection.

Study quality was assessed using the Jadad scale of randomisation, blinding, withdrawals and drop-outs. A quality score of at least 2 out of 5 (randomised and double-blind) was required for inclusion of the study.

The authors did not report how many reviewers performed quality assessment.

Data were extracted to calculate the number needed to treat (NNT) for 50% pain intensity reduction (or 30% pain reduction or at least good pain relief). Number needed to harm (NNH) was calculated based on how many patients needed to be treated for one patient to drop out due to adverse effects. Where the percentage of patients with 50% pain relief was reported and actual numbers of patients were not, these were calculated under the assumption that the percentage was of an intention-to-treat population. Ninety-five per cent confidence intervals (95% CIs) of NNT and NNH were calculated (using the normal distribution) as the reciprocal value of the 95% CI for the absolute risk difference. Authors were contacted for data not reported in the paper.

The authors did not report how many reviewers extracted data.

For the placebo-controlled trials NNT and NNH, with 95% CIs, were obtained by pooling raw data by drug type stratified by pain condition. Heterogeneity was assessed through examination of L'Abbe plots; statistical heterogeneity was not assessed. For the trials that compared first-line drugs, some studies were pooled (no further details provided).

One hundred and seventy-four placebo-controlled RCTs were included in the review: 105 from the earlier review and 69 from the update searches. Eighty trials had a cross-over design and 94 had a parallel design. The average Jadad score was 4.1 (standard deviation 0.9). Seven comparative trials of first-line drugs for neuropathic pain conditions were included.

NNT ranged from 2.1 (95% CI 1.9 to 2.6) to 6.8 (3.9 to 27) for different types of antidepressants, 2.1 (1.4 to 4.2) to 6.4 (4.3 to 12) for different types of anticonvulsants, 2.1 (1.5 to 3.3) to 5.1 (2.7 to 36) for opioids and from 2.3 (1.5 to 4.7) to 11 (5.5 to 316) for other drugs.

NNH ranged from 13.1 (9.6 to 21) to 15.9 (11 to 26) for antidepressants, 6.3 (5.1 to 8.1) to 32.5 (18 to 222) for anticonvulsants and 13.3 (8.8 to 27) to 17.1 (9.9 to 66) for opioids. NNH was not available for most of the other drug types.

The authors stated that differences in study designs and patient populations meant that a direct comparison of NNT across the drug classes was not appropriate.

Pooling of six trials that compared a tricyclic antidepressant to gabapentin or pregabalin showed no significant difference in the proportion of patients with 50% pain relief (49% for tricyclic antidepressant and 43% for gabapentin and pregabalin, p=0.19). Withdrawal due to side-effects was similar.

Despite there being 66% more published trials than were available for the previous review (in 2005), only a limited improvement in neuropathic pain was obtained. A large proportion of neuropathic pain patients were left with insufficient pain relief and this called for other treatment options to target chronic neuropathic pain.

The research question was supported by clear inclusion criteria. Only a single potential source of unpublished studies was searched and only studies in English were included, so publication and language biases could not be ruled out. The authors did not report whether they took steps to minimise reviewer error and bias (such as involvement of two reviewers). Study quality was investigated using the Jadad scale to assess some potential sources of bias. Details of the evidence synthesis methods were fairly limited and this made it difficult to judge their appropriateness. The authors acknowledged limitations in using NNT to compare the efficacy of different drug classes. The included studies were heterogeneous in terms of drug dose, placebo response and study design; only limited exploration of this was undertaken,

As noted by the authors, it was not appropriate to use data from this review to compare NNT values across drug classes. Their conclusion focused on the general issue of limited improvement in neuropathic pain treatment despite the volume of new trials since the earlier review.

Practice: The authors did not state any implications for practice.

Research: The authors stated a need for large-scale drug trials to identify possible subgroups of patients who were likely to respond to specific drugs.

Velux Foundation, Denmark; Lundbeck Foundation, Denmark; Innovative Medicines Initiative.

Finnerup NB, Otto M, McQuay HJ, Jensen TS, Sindrup SH. Algorithm for neuropathic pain treatment: an evidence based proposal. Pain 2005; 118(3): 289-305.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.