Non-steroidal anti-inflammatory drugs for low back pain

Clinical bottom line:
The benefit of NSAIDs for treating low back pain is unclear. It is unclear from the present review whether these trials have been adequately designed to measure NSAID efficacy compared with placebo and other treatments. There is no convincing evidence to suggest that one particular NSAID is better than another, or that NSAIDs are better than other treatments.

There is no clear consensus regarding optimal management for low back pain, and management varies widely. The rational for NSAID treatment is based on analgesic and anti-inflammatory properties.

Systematic review

Koes BK, Scholten RJPM, Mens JMA, Bouter LM. Efficacy of non-steroidal anti-inflammatory drugs for low back pain: a systematic review of randomised clinical trials. Ann Rheum Dis. 1997; 56:214-223.

Date review completed: 1994

Number of trials included: 26

Control group: active and placebo

Main outcomes: pain intensity, pain relief, clinical improvement

Inclusion criteria were randomised, controlled trials of NSAIDs for low back pain; full journal publication; English language.

Reviewers assessed trials for quality, and based their conclusions on authors' original conclusions together with data pooling where appropriate. Data were pooled where placebo information was also available, separating out acute and chronic low back pain trials. Peto's odds ratios with 95% confidence intervals were calculated with fixed (heterogeneity present) and random effects models (heterogeneity). Odds ratios were calculated for treatment failures, and therefore a ratio of below one indicates significant benefit of NSAID.

Findings

Trials were of poor to high quality. Unfortunately it was not possible to clearly establish from the review which trials were double-blind, but probably only ten of 26 were double-blind. This makes understanding review findings difficult. Clearly this information should carry more weight than data from non- or single-blind trials. A brief overview of results is provided, but further work would be required to establish treatment efficacy more accurately. Reviewers did not comment on baseline pain intensity. Clearly trials that did not ensure adequate baseline pain were unlikely to detect an improvement.

Trials differed according to drugs, doses, length of intervention, diagnosis, rescue medication allowed, size and outcomes. Interventions were mainly of one to two weeks duration.

NSAID versus placebo

Five of ten trials reported significant benefit with NSAID. However, looking just at the five trials of highest quality suggests that NSAIDs may not be that effective. Only one of these found a significant benefit of NSAID, and this was only on one day. This trial did not allow rescue medication. A second trial found significant benefit only when it looked at the subgroup of patients with moderate to severe baseline pain. It is unclear whether these trials were adequately designed to measure the effect.

NSAIDs compared with other pharmacological interventions

Nine trials were included. Only one trial had a high quality rating. This trial was very small, but reported that patients with chronic low back pain preferred diflunisal 500 mg twice daily for four weeks to paracetamol 1000 mg twice daily (10/16 versus 4/12 rated treatment as good or excellent). Only three of the remaining eight trials reported significant benefit of NSAID compared with a range of other treatments. It is unclear whether any of these trials were adequately designed to measure a difference.

Different NSAID comparisons

Only one low quality trial of the 11 included trials reported a significant difference. It is unclear whether any of these trials were adequately designed to measure a difference.

Adverse effects

Reviewers extracted information on adverse effects where possible. These were usually mild to moderate severity and included abdominal pain and diarrhoea, oedema, dry mouth, rash, dizziness, headache, tiredness.