Medicine's Secret Stat

Researchers can gather all the hard-nosed evidence they want about the effectiveness of a particular drug or treatment. But there's one figure doctors don't much talk about despite its importance. It's called number needed to treat, or NNT, a new measure developed in the past 20 years that's one of the best-kept statistical secrets in medicine.

The idea of NNT is simple enough. Most clinical trials look at how much better people do on a particular medicine compared with how they would do without it or whether they should be on a different medicine. Take statins, drugs that aim to reduce bad cholesterol. A typical trial might give one group of men a statin for, say, five years and give a second group a placebo, or fake pill.

Generally, you will see fewer heart attacks in the statin group (about 30% fewer in one real-world trial). Reducing the risk by a third sounds like a lot, which is one reason many hundreds of thousands of men with no sign of heart disease take statins. But that number is meaningless unless you take into account the percentage of men in both groups who have heart attacks in the first place. If those people represent only a tiny fraction of the two populations, an improvement of 30% isn't much--maybe one heart attack fewer in a group of thousands.

the rest...


http://www.time.com/time/magazine/article/0,9171,1590464,00.html?iid=sphere-i...


grz-