Statistics Training And The Odds Of Illness Does a positive mammogram mean a woman has breast cancer? Does a positive HIV test mean someone is infected with the virus? As ordinary people confront the laws of probability, the odds of misinterpretation and false alarms rise.
Read the full story

September 17, 2001
By Howard LeWine, M.D.
Harvard Medical School
How does this article relate to you?
Understanding the odds about what can happen to you is essential to making good decisions about medical care. This article tested a traditional way and a more natural way to present medical chance. The more natural way is easier for most people to grasp, and you are more likely to retain that information.
The example used in this article helps to support an important concept. Mammograms do not always show breast cancer even when it is there. The percentage-based rule would state: For women with breast cancer, a mammogram will be positive 80 percent of the time. The natural-frequency rule would state: If a woman has breast cancer, it will show up on the mammogram four out of five times. That means it won't show up one out of five times.
I think it is important to state it both ways: how often it is there and how often it shows up. Everyone hears these numbers with a different slant.
What changes do you need to make?
I don't think that you need to make any change. But health care professionals should give the information in this study careful consideration. When presenting information to patients, I plan to adopt the natural-frequency rules. Given how long researchers and physicians have been using percentage probability, with 100 as the standard, I suspect that the switch will take some time.
What can you expect in the future?
Hopefully, adopting the natural-frequency rules will be a first step toward understanding the chances of different outcomes when you choose to undergo a test or start a new therapy. When the odds are based on good clinical studies, the decisions that you make with your doctor's advice can be guided by evidenced-based medicine.