Tuesday, April 17, 2012

Relative Risk Reduction vs Absolute Risk Reduction Part 2

So what did yesterday's drivel have to do w/today's topic?  Well, a year ago this month, I noted a study suggesting an increase risk of thromboembolism (blood clots) in those women who used oral contraceptives containing drospirenone.  Of course, all the evidence did not flow against this particular progestin (very different from bio-identical progesterone).  Still, given the large number of options available, I've been recommending other options since.  The Food & Drug Administration (FDA) did not have enough information as of last September but as luck would have it, more evidence built up against drospirenone in October & November last year.

Finally, the FDA acknowledged last week that drospirenone is associated w/threefold increase in risk of blood clots based upon observational data and will change labeling to show this.  While a 3x increase in relative risk is impressive (and statistically significant), I also think that it's important to consider absolute risk (the graphic on the left is worth a thousand words).  Compared to a non-pregnant female not taking any combination oral contraceptive, use of drospirenone will double or triple the risk of blood clots from 1-5 to 3-9 in 10,000 women-years.  However, to get a grasp of the big picture, it should be noted that this is still less than the 5-20 blood clots in pregnant women per 10,000 women-years and the 40-65 blood clots in the 12 weeks immediately after delivery.  In other words, while hormonal contraception increases relative risk, the overall absolute risk is still small compared to what will happen if they get pregnant.

I point this out not to increase sales of birth control pills nor because I receive remuneration (in fact, I do not) but rather to demonstrate the importance of understanding relative risk vs absolute risk, in order to determine whether or not your benefit will exceed your potential for side effects.  It's rather unfortunate that this isn't something that we can explain adequately in a typical 15 minute office visit.  What's really scary is how little teaching in statistics we receive in medical school, yet our ability to clearly explain research outcomes depends upon it.



Health
Top Blogs

No comments:

Post a Comment