The Journal : Current Issue
Bruce V Stadel, Eric Colman, Todd Sahlroot
Risk ratios are widely misused in ways that exaggerate both the benefits and harms of drugs. This is especially true when a risk ratio is called "relative risk".
Relative risk does not measure "risk" at all, because risk has dimensions, such as observed deaths per 100 or 1000 people. However, a risk ratio has no dimensions because they cancel in calculating the ratio. Thus, if a drug changes risk from two deaths per 100 people to one death per 100 people, the risk ratio (0·5) is the same as if the drug changes risk from two deaths per 1000 people to one death per 1000 people. It is wrong to call these changes a "50% decreased risk"--the change from two deaths per 100 people to one death per 100 people is a 1% decreased risk, and the change from two deaths per 1000 people to one death per 1000 people is a 0·1% decreased risk.
No comments:
Post a Comment