Tuesday, May 23, 2006

How much should we trust that test?

Imagine you've been under the weather lately. You've popped OTC analgesics, antipyretics, and what have you. Half a week has gone by and you just seem to be getting worse, so you go see your physician. He asks for the symptoms and does all those things with his stethoscope.

"I don't want to alarm you but there are indications you may have Krupp-Messerschmitt Disease," he tells you. You give your doctor a blank stare. "Well, it's a rather infrequent type of tumorous growth that can affect various organs. Afflicts 1 in 500 persons." "Are there drugs for this?" you inquire. "Unfortunately, surgery is the only remedy. But the important thing right now is to find out whether you actually have it. So I'm ordering a Tolkien-Rowling Test." Knowing that practically no diagnostic test is perfect you ask how accurate it is. Hoping to allay your fears the good doctor reassures you, "The latest research on this shows that if you actually have KM then 90% of the time the test will correctly come out positive, and should you not have the disease then 85% of the time it will correctly be negative."

Given the above figures how confident can you be that you have KM if the test comes out positive? That you're KM-free if it comes out negative?

Solution and answers.

How close were your estimates?

No comments: