Saturday, June 11, 2005

The Other Side of Probablity

Every week or so a story comes out regarding the increased risk of this or that malady associated with a particular medication. A for-instance is this article detailing increased risk of heart attack in response to the uses of NSAID pain relievers. Included in that group is the over-the-counter analgesic ibuprofin (Motrin). The article includes the standard verbiage:
The worst risks were with diclofenac, a prescription painkiller, which increased the risk by 55 per cent.

The increased risk of a heart attack for those taking rofecoxib was 32 per cent. For ibuprofen it was 24 per cent and for celecoxib 21 per cent.

The big problem with these articles is that they assume that the reader already knows what the original risk was and can then calculate the new risk. Not knowing those values, and relying on the "double the risk" statements" frequently lead to hysteria and legislation, two things that rarely end well.

So, I would like to thank Celia Hall, the author, for including other aspects of the statistics.
Julia Hippisley-Cox, professor of clinical epidemiology and general practice at Nottingham, said this meant for those over 65 taking diclofenac that one extra patient for every 521 would have a heart attack linked to the drug.

For those who had taken rofecoxib the figure was one patient at risk for every 695 taking the drug and for ibuprofen the figure was one for every 1,005 patients.

This way, customers can choose whether they want to take the drug and risk being the extra one-in-a-thousand.

Link via Roger L. Simon

No comments: