Tuesday, September 22, 2009

Latest HRT scare - more epidemiological rubbish

The original journal article is here. The article below glides over the fact that the effect of HRT on incidence of cancer was NON-SIGNIFICANT statistically. And given the large numbers involved, the effect had to be pretty tiny in absolute terms for that to be so. The marginally significant difference in mortality among HRT users who DO get cancer is more a puzzle than a threat. In absolute terms, women on HRT are most UNLIKELY to die of lung cancer -- unless they smoke. Smoking is the real risk factor

The constant reliance on relative risk ratios is very annoying and uninformative. It may be good for getting journal articles published but it is useless to the average person trying to work out a reasonable policy for themselves. What they need to know is the ABSOLUTE risk -- and in all the reports about the evils of HRT it is negligible

If taking HRT raises your risk of some illness from 1 in 4000 to 1 in 3000, that gives a relative risk ratio big enough to generate a scary academic journal article but in absolute terms the risk is still negligible.

So unfortunately the crucial distinction between relative and absolute risks is glossed over or ignored in the pursuit of eye-catching headlines.

Posted via email from Garth's posterous

No comments: