Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Topical NSAIDs

Systematic review
Results
Acute conditions
Chronic conditions
Topical versus oral NSAID
Comment

It is always interesting when there is a debate about whether an intervention works or not. Often that debate then extends into whether the intervention is worth using or not. One of the best examples is the debate over topical NSAIDs, where despite having a systematic review [1], doubts remain. There are two arguments used to discredit the review. One is that it is 'just the rubbing' that produces pain relief. The other is that there must be masses of negative unpublished information that, if we had it, would change the conclusion.

Systematic review

The review searched several electronic databases and a specialist database of trials with pain as an outcome developed by hand-searching journals, and that was subsequently included in the Cochrane Library. Pharmaceutical companies in the UK were also asked for any unpublished studies. Included trials had to be full journal publications of trials that were both randomised and double blind, and with placebo or active comparators.

Acute conditions were defined as sprains or strains, and for these the outcome was one that equated to at least half pain relief after one week. Chronic conditions were predominantly single joint arthritis or rheumatological disorders with the same outcome equating to at least half relief at two weeks.

Results

Overall 86 randomised trials with 10,160 patients were found. Over 75% of placebo-controlled studies had a quality score of 3 out of 5 on a popular scale, known to be associated with minimisation of bias.

Acute conditions

There were 40 trials in strains and sprains with 1,747 patients treated with topical NSAID and 1,492 with placebo (Figure 1). Topical NSAID was better than placebo, with an NNT at one week of 3.9 (95% confidence interval 3.4 to 4.4); 71% of patients had a successful outcome with topical NSAID and 38% with placebo. Topical preparations of ketoprofen, felbinac and ibuprofen had NNTs of 2.6 to 3.5. There were few local or systemic adverse effects, or withdrawals, and no difference between NSAID and placebo.

Figure 1: Topical NSAID versus placebo in strains and sprains after seven days



Chronic conditions

There were 13 placebo-controlled trials in chronic conditions with 547 patients treated with topical NSAID and 550 with placebo (Figure 2). Topical NSAID was better than placebo, with an NNT at two weeks of 3.1 (2.7 to 3.8); 65% of patients had a successful outcome with topical NSAID and 30% with placebo. There were few local or systemic adverse effects, or withdrawals, and no difference between topical NSAID and placebo.

Figure 2: Topical NSAID versus placebo in chronic conditions after 14 days



Topical versus oral NSAID

Five studies compared topical with oral NSAID, with no benefit of oral over topical treatment (three in acute and two in chronic conditions).

Comment

So is it just the rubbing? A simple look at the two L'Abbé plots (Figures 1 and 2) tells us that cannot be the case. In these trials, most of which we know had matched placebo, the only difference is that one version of the gel or cream had NSAID in it. Topical NSAIDs produced about twice the effect of placebo, despite any rubbing that took place with topical NSAID or placebo.

What about unpublished material? This review uncovered previously unpublished trials, with about 17% of the total unpublished, so it was hardly ignored. And the message from an HTA report is that unpublished and published information do not differ in their conclusions, and we now know funnel plots to be unhelpful in finding publication bias [2].

So we have known efficacy with known safety against the gastrointestinal harm associated with oral NSAIDs [3]. There may still be arguments about whether these topical preparations are worth using, and particularly in which patients with chronic conditions they will work best. But is there room for argument about whether they work?

References:

  1. RA Moore et al. Quantitative systematic review of topically applied non-steroidal anti-inflammatory drugs. BMJ 1998 316: 333-338.
  2. JA Sterne et al. Publication and related bias in meta-analysis. Power of statistical tests and prevalence in the literature. Journal of Clinical Epidemiology 2000 53: 1119-29.
  3. JM Evans et al. Topical non-steroidal anti-inflammatory drugs and admission to hospital for gastrointestinal bleeding and perforation: a record linkage case-control study. BMJ 1995 311: 22-26.
previous or next story in this issue