Topical NSAIDs have unknown effectiveness

Dieppe claimed in an editorial about osteoarthritis that topical non-steroidal anti-inflammatory drugs (NSAIDs) are as effective as oral NSAIDs.¹ He refers to a report from the National Institute for Health and Clinical Excellence (NICE), which describes two meta-analyses. He notes that the authors assessed the trials for quality, but the 424 page report says nothing about that quality. The authors of one the meta-analyses, however, reported that the trials were of poor quality and that the effect decreased with increasing sample size of the trials.² There are other problems with the trials, and the results are difficult to interpret.

Dieppe also quotes a systematic review of topical NSAIDs in people with musculoskeletal pain. The authors reported that small trials exaggerated the effects. When some of the same authors updated this review in people with acute musculoskeletal pain, the effect again decreased significantly and substantially with increasing sample size of the trials.³ However, they paid no attention to this problem as their results build on all the trials, although I pointed out that this was a major compulsory revision they needed to do when I peer reviewed their manuscript.⁴ Their trials were also subject to enormous statistical heterogeneity, but again, the authors refused to pay any attention to this fundamental problem. Their reply to the peer review comments⁴ is interesting as it shows that their review is of very poor quality and that its authors did not take account of the most basic methodological evidence that exists for performing systematic reviews, even though this was pointed out to them.

I have described these issues in my chapter on NSAIDs in Clinical Evidence,⁵ and I have concluded, in accordance with the evidence categories, that topical NSAIDs have unknown effectiveness.