Increasingly sensitive tests redefine what is pathologic

There is a story reported in the New York Times (NYT) today which picked up on a paper published out of Boston University in the British Medical Journal (BMJ) entitled: "When a test is too good: How CT pulmonary angiograms find pulmonary emboli that don't need to be found."

The story sounds familiar:
1. New technique or technology heralded for early discovery of disease
2. New technology deployed and used widely
3. Explosion of silent disease diagnosed
4. Claims of lives saved used to justify widespread use
5. Overall death rates remain unchanged

In this case, the technology is CT angiogram to diagnose pulmonary embolism and with the widespread use of this technology, many more emboli are found. However, the question arises whether treatment of what are generally small and localized pulmonary emboli (PE) makes any difference in outcomes. PE's are generally treated with a course of anticoagulants, agents which carry significant risks of bleeding. If treatment is instituted for a trivial embolism which had little risk of causing harm, the likelihood of harm from treatment may vastly outweigh the risk of the underlying disease.

I colleague of mine who I met at the Society for Improved Diagnosis in Medicine (SIDM) has recently published an iBook entitled "The science and art of medicine" (link). It is a masterful work which should be required reading for all those who order or do diagnostic tests. He presents a framework to understand why diagnostic tests MAY be useful, based upon Bayesian probabilities.  All tests are validated within particular contexts and are ultimately useful because they provide prognostic information regarding endpoints relevant to patients. Increasing sensitivity of diagnostic approaches does not always lead to better diagnostic acumen.

The general public tends to view diagnostic tests as very binary. In their minds the test either shows you have a disease state or not. Unfortunately, the same mindset is not so rare among practicing physicians. This is a problem and it is a problem which is aggravated by both financial incentives as well as heavy reliance of faulty heuristics and reliance of intuitive system one based decision making. Reflection and appreciation of nuance takes time and the desire to move rapidly through one's work results in quickly moving to framing decisions as all or nothing. Tests are viewed as either positive or negative with a positive test meaning the patients has the "disease".

As we begin to realize that we are making decisions based upon flawed tests and flawed assumptions, the timing could not be worse. Health care is being stretched by cost cutting, changing expectations of the public, and massive disruptions of workflow by deployments of electronic health records (EHR's). How do we provide feedback to our peers and choices to our patients in an environment which leaves little or no time for reflection and discussion?