Below, a transcript of a column by George Lundberg, M.D., Editor-at-Large of MedPage Today that does a very neat job of summing up why healthy people should not be tested for disease unless they are experiencing symptoms or have good reason to believe that they are “at risk.” Put simply, the math is against you.
Dr. Lundberg serves as president of the Lundberg Institute, and is a consulting professor of pathology and health research policy at Stanford University School of Medicine. (Full disclosure: I’m on the Advisory Board of the Lundberg Institute.)
Lundberg co-authored this piece with Dr. Clifton Meador. (If you’ve seen the film version of Money-Driven Medicine, you will remember Meador as the doctor who takes you on a tour of Nashville.) They promise a sequel on “false diagnosis” next Monday.
You will find the audio version of the column here.
Hello and Welcome. I'm Dr. George Lundberg speaking for myself and lead author Dr. Clifton Meador of Nashville, Tenn., in this At Large at MedPage Today.
Over the past several decades, there has been a shift in the kinds of patients seeking medical care.
The progression has been from sick to early sick to well to worried well to worried sick.
The reasons are beyond the scope of this article. There is a subtle and hidden, but potentially very damaging, factor operating in the diagnostic process when large numbers of well people seek medical attention in a system designed to care for the sick. This factor is the prevalence of disease.
The accuracy of the diagnostic process is ruled by strict arithmetic.
Specificity and sensitivity of the test itself and the prevalence (or pretest probability) of the disease in the population being tested control the accuracy of the process.
Pretest probability means the percent of people being tested who have the disease in question.
A pretest probability of 2% means that 20 people out of 1,000 have the disease. Obviously, 980 people do not have the disease.
At 2% prevalence, using a test with 95% specificity and 95% sensitivity, the rate of false positives will be 72%.
So, 5% of 980 well people will generate 49 false positive tests while 95% of 20 people with the disease will yield 19 true positives.
1,000 people; 49 "wild goose chases," and only 19 geese caught.
Trouble, especially for the 49.
When the prevalence of a disease decreases, the rate of false positives increases.
Even with a very specific and sensitive test, if the pretest probability in the population is low, there will be more false positives than true positives.
False positives lead to one of two things — one, they must be tracked down with additional tests, which are costly and will generate more false positives. Often this leads to unnecessary biopsies and anxiety. This is part of the sad story of mammograms and PSA testing. Or, two, the false positive test result becomes a label of a nonexistent disease and the person carries the burden of a disease he or she does not have. The amount of false diagnoses of any disease, other than heart disease in children, is unknown and generally unstudied.
It is pure arithmetic.
The increase of well people seeking medical care lowers the prevalence of all diseases and increases the rates of false diagnoses.
Beware of this current active Great American Medical Tragedy.
That's our opinion. We are lead author Dr. Clifton Meador of the Meharry Vanderbilt Alliance and Dr. George Lundberg, At Large for MedPage Today