In a recent article, cardiologist, Dr. Eric Topol's briefly and to the point explains:
Evidence-based medicine is a fancy buzz term. You've got people who draw out guidelines from shaky data and tell the world this is the way it's got to be. A great example is statins. The medicine community promotes them like crazy. But for people without previous heart disease only one in a 100 people will benefit in terms of preventing heart disease. And then you have the FDA issuing a warning that taking statins can increase the risk of diabetes, which is at least 1 in 200 for the more potent statins. For people who have heart disease, statins are great. But if all you've had is high cholesterol, what you're doing is taking this 1/100 chance of getting a benefit and offsetting it with 1/200 chance of getting diabetes. What kind of trade off is that? We've got to get more intelligent about the individual's true benefit to risk story.
Source: Destroying Medicine to Rebuild It
Here's another good site that explains ... The NNT. NNT stands for Number Needed to Treat.