Introduction

Reviews

False Positive Tests

Timothy B. McCall, M.D.

The more tests that are done, the greater the likelihood of at least one false positive result. Consider routine chemistry tests which are often run in panels consisting of up to 20 individual tests. The cutoff point between normal and abnormal for each of the individual tests is set to allow a 5% false positive rate. If two tests are done, there is therefore a 10% chance of getting a false positive. If six are done, there is a 26% chance. For a battery of 20 tests, two thirds of normal people will have at least one abnormal result.

• The less likely a problem is, the more likely an abnormal test result is a false positive. This is a basic principle of test interpretation, yet many physicians don’t seem to understand it. Let me explain. Say that only one person in 1000 has Disease X and that the test for it yields a false positive result 5% of the time. If we tested 1000 randomly-chosen people, we would expect to find one person with the disease. Fifty people per 1000 or 5%, would get false positive test results. In other words, one person in 51 with an abnormal result would actually have the disease. This is not the impression many patients get when they’re told their test results are abnormal. Doctors may not be much better at interpreting abnormal results.

STUDY: How Well Do Doctors Interpret Abnormal Test Results?

Researchers stood in the hallways of four Harvard Medical School hospitals and quizzed resident physicians, fourth-year medical students and faculty members about a disease with a prevalence of one in 1000, the test for which has a false positive rate of 5%. The doctors were asked to assess the likelihood that someone with a positive result had the disease. Only 20% of the Harvard faculty members gave the correct answer of one in 51. Almost half the participants answered that 95% of the people with a positive result had the disease—an estimate almost 50 times too high. If the answers at Harvard Medical School were this bad, we can only guess how well the average doctor would do.

• Some tests have extremely high rates of false positives. The newly-developed screening test for prostate cancer, the PSA, is notorious for its high rate of false positives. In other words, many people who don’t have prostate cancer, come up with abnormal results. Another example is the blood test for Lyme disease.

Example: Tests for Lyme Disease

The antibody test for the tick-borne infection Lyme disease is inherently inaccurate. Because there are so many false positives, experts recommend that the tests should only be done on people who’ve been in areas known to harbor infected ticks or who have suggestive symptoms. These symptoms include a characteristic skin rash early on and, if the person isn’t treated with antibiotics, arthritis and certain heart and neurological problems months later. Doctors don’t always follow the experts’ advice. The diagnosis of Lyme disease has in fact become trendy, with many people misdiagnosed based on false positive tests. Some of them end up being treated with expensive and potentially dangerous drugs that have no chance of benefiting them.

STUDY: The Overdiagnosis of Lyme Disease

Researchers at the Lyme disease clinic at the New England Medical Center, probably the leading clinic in the world for this problem, studied almost 800 patients referred to them for Lyme disease. They found that in only 23% was Lyme disease causing their symptoms. Another 20% had Lyme disease in the past and still had antibodies to it but their symptoms were caused by other medical conditions. The remaining 57% had never had Lyme disease. Many of these people who were incorrectly diagnosed had false positive test results at other labs and had already been treated without benefit before being referred to the clinic.

• The more abnormal the result, the greater the likelihood of a problem. Say the normal range for a test extends from 10 to 30. A result of 31 or 32 is much more likely to be a false positive than a result of 132. When in doubt and there’s no emergency, repeat the test a few weeks or a few months later. If you are healthy and a test result is slightly abnormal, it will tend to stay about the same or improve slightly. If you are developing a problem, the results will usually worsen over time.
  * The more likely a problem is, the more likely a normal test result is a false negative. False negatives are also a big problem. In the example mentioned earlier, false negative mammograms led doctors to miss the diagnosis of cancer in women who developed lumps in their breast. Doctors and patients alike tend to accept uncritically the results of sophisticated tests., If the odds of a serious problem, like cancer, are great enough, you may need additional tests to be sure the diagnosis is not missed since so much is at stake.

Next: Follow the Money

Return to Examining Your Doctor