Reader Comments

Post a new comment on this article

Selective statistics

Posted by rfisken on 18 Feb 2013 at 09:58 GMT

This article has caused a great deal of negative comment in the U.K., and doubtless elsewhere, about the use of NSAIDs. Yet it does not report any primary data about the safety of anti-inflammatory drugs and the data that it does report is all about relative, and not absolute, risk. For someone who is in severe pain but has no risk factors for CVD a small increase in absolute risk might well be acceptable, as might also be the case for a person of advanced years for whom quality of life may be more important than living a few months longer. Papers like this are of very limited value unless they contain data on absolute risk, preferably in patients of different ages. To say that such data can be obtained from reading the references is, frankly, not good enough, especially if the paper is being used to justify an important claim about public policy such as that a drug should be banned or severely restricted.

No competing interests declared.

RE: Selective statistics

mddah02 replied to rfisken on 19 Feb 2013 at 21:56 GMT

This paper was written to draw attention to the possible public health impacts of the widespread use of high risk NSAIDs (diclofenac and etoricoxib) in LMIC facing a rising epidemic of cardiovascular disease. It is not written to help individual choices made by patients with painful disorders living in high income countries. That is not to say that the latter are unimportant, but it’s hard to accommodate all of this in a single paper. The relative risk values reported in the paper (and in many other publications) can be applied to a range of background risks to derive estimates of the excess risk associated with use of the drug. If a patient has a background risk of (say) 1/100 then this will be increased to around 1.4/100 by either diclofenac or rofecoxib (Vioxx). That difference of 0.4% means that of every 250 treated patients one would suffer an additional heart attack with diclofenac or rofecoxib. However, if the background risk is 1/10 the excess risk is 4% meaning that of every 25 patients in that high risk group who receive treatment one will suffer an additional heart attack. These risks are easily calculated and may be helpful for individual decision making but they are not the focus of this paper.

David Henry FRCP
CEO, Institute for Clinical Evaluative Sciences, Toronto

No competing interests declared.

RE: RE: Selective statistics

PMcGettigan replied to mddah02 on 19 Feb 2013 at 22:59 GMT

Our work on determining risk was published in earlier meta-analyses of data from dozens of observational studies. In this paper, we concentrated on how global usage of NSAIDs reflected information on cardiovascular risk associated with individual agents. Our concern was the public health relevance of the failure to translate evidence to practice.
The question of absolute risk increase for an individual may be illustrated as follows. Overall, diclofenac increases cardiovascular risk by 40%, and doubles it at doses above 100mg/day. The clinical importance of this increase depends on the underlying cardiovascular risk of the individual taking the drug.
For example, a young fit sportsman has a cardiovascular risk that is very low, fractions of a percent. Increase this by 40% - 100% by taking diclofenac; it is still a very low risk. So for him, the choice of NSAID to control the pain of a sports injury is no big issue as long as the NSAID is effective.

At the age of 68 or so, this man’s cardiovascular risk may be very different – let’s say he smokes 12 cigarettes/day, has high blood pressure and diabetes with consequent renal impairment and that his father had a MI aged 59. His risk of a cardiovascular event in the next year is in the range 5%- 10% (QRISK®2-2012 risk calculator: Taking diclofenac for arthritic or other pains increases this by 40% or even 100%, taking it to 7%-14%. So for this man & others like him, the absolute risk increase is 2%-4% (a number needed to treat to harm of 25-50 ) or even up to 20% with diclofenac >100mg/day. This is a risk that in our view is well worth avoiding when there are lower risk options, in particular naproxen.

We know that there are some people with underlying cardiac risk levels like the man described who don’t get relief with one NSAID and need to try others. Some will find that diclofenac is the one & only effective NSAID. If so, the benefit of good pain relief may be well worth the increased absolute risk. This is their decision though, made by considering the risks and benefits with their doctors. These people are generally a minority. For most, one NSAID will work as well as another in the right dose. If diclofenac was being used only in the people for whom no other NSAID relieved pain, we would expect to see it with a consistent but low level of use, not to find it as the most-prescribed NSAID.

Patricia McGettigan BSc(Pharm) MD FRCPI FRACP

No competing interests declared.

RE: Selective statistics

PMcGettigan replied to PMcGettigan on 19 Feb 2013 at 23:32 GMT

I notice that I split my sentence above on risk wrongly when I wrote in the absolute numbers. It should read: Taking diclofenac for arthritic or other pains increases this by 40% or even 100%, taking it to 7%-14% or indeed up to 20%. So for this man & others like him, the absolute risk increase is 2%-4% (a number needed to treat to harm of 25-50 ) or up to 10% with diclofenac >100mg/day (NNT-H of 10).'

Patricia McGettigan.

No competing interests declared.