How false positives can kill patients

I’ve written in the past that more medicine and tests do not necessarily reflect better care.

There is no test that is 100% specific or sensitive.  That means tests may be positive, when, in fact, there is no disease (“false positive”), or tests may be negative in the presence of disease (“false negative”).

It’s the latter that often gets the most media attention, often trumpeted as missed diagnoses, but false positives can be just as dangerous.

Consider this frightening case report from the Archives of Internal Medicine:

A 52-year-old woman presented to a community hospital with atypical chest pain. Her low-density lipoprotein cholesterol and high-sensitivity C-reactive protein levels were not elevated. She underwent cardiac computed tomography angiography, which showed both calcified and noncalcified coronary plaques in several locations. Her physicians subsequently performed coronary angiography, which was complicated by dissection of the left main coronary artery, requiring emergency coronary artery bypass graft surgery. Her subsequent clinical course was complicated, but eventually she required orthotropic heart transplantation for refractory heart failure. This case illustrates the hazards of the inappropriate use of cardiac computed tomography angiography in low-risk patients and emphasizes the need for restraint in applying this new technology to the evaluation of patients with atypical chest pain.

With the proliferation of CT scans that can detect coronary artery disease, there have been some who advocate more aggressive use of the tests. In the emergency room, for instance, a scan that can accurately detect heart disease would be a huge diagnostic leap forward.

But we’re not there yet.

As this case shows, there are real consequences to false positives. Often times, they lead to more invasive tests, like biopsies, or in this case, a cardiac catheterization that went horrifically awry.

So, if you or your doctor believe in getting a scan “just to be safe,” consider this case where such a mentality led to a heart transplant.

False positives can potentially kill.

 is an internal medicine physician and on the Board of Contributors at USA Today.  He is founder and editor of, also on FacebookTwitterGoogle+, and LinkedIn.

Comments are moderated before they are published. Please read the comment policy.

  • Doug Duffee

    This is an important insight that patients do not understand unless they sit down and talk with their doctor before heading out for the “lifeline screening tests”. Carotid ultrasound is a similar issue. The questionable quality life line screening carotid ultrasound comes back showing “moderate” plaque. This leads to a full diagnostic ultrasound and then most likely an invasive CT angio of the carotids to determine that only medical management is necessary. We potentially place the fortuitious and incidental discovery of surgical plaque up against radiation exposure and contrast nephropathy…

  • Clifton K. Meador, MD

    I have spent a lifetime following the literature of false positives. I wrote a satire in 1965 in the NEJM called the “Art and Science of Nondisease.” it was pure tongue and cheek, classifying the false diagnoses then in existence. In the following 45 years I can find only one article trying to find the rate of false positives at the DIAGNOSIS level at the population level,i.e. how many people in a population have a false diagnosis of a disease. I can find only one. In 1967, Bergman and Stamm examined over 100,000 school chilren in Seattle, calling in for close study those who carried a diagnosis of heart disease. They found that only 20% actually had heart disease but that 75% of all had severe disability, physically or psychologically. The non cardiac disease was generating more disability than the true heart disease. I remain puzzled why the overdiagnosis of non existent disease remains so little studied. I am still hopeul this error will get more attention. False positive lab and other testings get full attention but not at the level of a diagnosis. See Bergman and Stamm, NEJM 276:1008-13. 1967. The Morbidity of Cardiac Nondiseasein School Children.
    If there are other populaton studies of over diagnosis, I hope your readers will share the references.

    Clifton Meador, MD

  • Marc Gorayeb, MD

    I tried to retrieve the article, but don’t have a subscription…
    Would she have undergone the cardiac cath if the CT scan had not been available? Very possibly.
    Could she have had actual coronary pathology? After all, a dissection during cath may indicate that the artery was diseased.
    Could this simply be an unfortunate and rare complication of a routine but invasive procedure? Such an occurrence can be random, unpredictable, and unavoidable from a statistical perspective.
    I agree with the general proposition that the risks associated with a false positive result may exceed the risks of doing nothing in a case with a low pre-test probability. However, this case is a sensationalized exception that does little to advance the argument. And the title of this piece further sensationalizes the case — it’s not helpful and invites demagoguery.
    Remember that hard cases make bad law.

  • Ken Grauer, MD

    Brings to mind 2 sayings:
    1) A clinician deserves the results of the laboratory tests that he or she happens to order.
    2) The “Ulysses Syndrome” – in which a “screening” test yields a result outside of the realm of “normal” – which then sets off a series of additional tests (over time) – at the end of which the patient is finally “restored” to their original state of “good health”. Sometimes we may be better off not knowing …

  • David Winchester

    Using invasive and noninvasive testing to accurately diagnose (without under or over diagnosing illness) is very challenging. However, the authors in Archives have not established that this patient’s test was inappropriately ordered.

    Based on Diamond and Kaul’s table for establishing the pre-test likelihood of CAD, a 52 year old woman with atypical angina has an intermediate likelihood of CAD (the authors refer to her as “low-risk”, but it is not clear what risk they were referring to). Intermediate risk patients with symptoms suggestive of CAD are considered appropriate candidates for noninvasive testing based on guidelines for stress echo, SPECT, and CT angio. Therefore based on the information provided, the test was ordered appropriately.

    Complications are an unfortunate risk of invasive tests, however, consider some alternative situations:
    -A 52 year old woman with atypical chest pain is reassured and undergoes no testing. She has an MI 2 days later.
    -A 52 year old woman with atypical chest pain undergoes treadmill ECG testing and falls, breaking her hip and suffers massive DVT/PE post-op from hip repair
    -A 52 year old woman with atypical chest pain undergoes SPECT which reveals an inferior defect. Ischemia cannot be distinguished from diaphragmatic artifact and she undergoes uneventful coronary catheterization.

    Anecdotes are perhaps the weakest form of data and decisions about the clinical utility of a test cannot be made based on one patient’s outcome.

  • Lezlee Maupin White

    another good article

Most Popular