I had a patient once who wanted an exercise stress test even though he had neither symptoms nor risk factors to suggest the presence of coronary artery disease (such as chest pain with exertion). I argued vociferously against it. However, extenuating circumstances (not relevant here) prompted our mutual decision to go ahead with it anyway. To our surprise and dismay, it came back positive.
When to screen
Given the amazing advances in medical technology in the last four decades or so, the American public has come to expect and believe that if a test exists for a disease, it should always be done. Nothing, however, could be further from the truth. The decision about which tests to perform on which types of patients and when to perform them actually requires a complex calculation.
First, we must distinguish between a screening test and a diagnostic test. Often a single test can be used as either, the difference arising from the circumstances under which it’s performed. So, in my patient who had no symptoms or risk factors to suggest the presence of coronary artery disease, putting him on an exercise treadmill by definition represented a screening test: that is, an attempt to identify the presence of a disease before it produced symptoms. Were he to have complained of chest pressure while climbing stairs, in contrast, putting him on a treadmill would have represented a diagnostic test: that is, an attempt to confirm or exclude a disease that had already manifested symptoms.
Why don’t we screen everyone for every disease we can? First:
- The disease must represent an important health problem. We could, for example, screen everyone for mononucleosis. But given that 90% of the population has already been infected without ever knowing it by the time they leave adolescence (and having had it are now immune to having it again), no reason exists to screen for it.
- An effective treatment for the disease must exist. Scientists are working on tests to detect Alzheimer’s dementia in its early stages. But because we still have no effective treatment for it, screening for Alzheimer’s at this point makes little sense. In fact, doing so may cause real harm—the harm that comes from knowing you’re highly likely to develop a fatal disease for which there exists no cure.
- A latent stage of the disease must exist. If no such stage exists (e.g., asthma), no opportunity to identify the disease before it becomes symptomatic exists. If the first time it becomes possible to identify the presence of a disease is only once it becomes symptomatic, why bother attempting to screen for it?
Second, putting aside the issue of affordability, specific criteria exist that must be met before any test can be considered appropriate for use in screening effectively and safely.
- The screening test must be able to detect the disease in the latent stage early enough to affect outcomes. The key isn’t just catching the disease in the latent stage but catching it before what’s known as the critical point—the point beyond which, even though the disease hasn’t yet produced any symptoms, it’s no longer curable. A case in point: studies have been done showing that obtaining screening chest x-rays in patients lacking signs or symptoms of lung cancer does indeed identify asymptomatic cancers earlier than if the x-rays weren’t performed—but still too late to affect the rate of cure. Interestingly, a recent study suggested that performing CT scans as a screening test does catch at least some lung cancers early enough to increase the likelihood of cure (by about 20%). This is probably because CT scans can pick up much smaller tumors than chest x-rays can, presumably increasing the likelihood of identifying the presence of lung cancer prior to the critical point.
- The risk of false positives must be acceptably low. A false positive result occurs when the test says you have a disease that you really don’t. The likelihood that a test is right when it says a patient has lung cancer (what’s called the positive predictive value of a test) turns out to be greater in smokers because smokers have a higher risk than non-smokers of getting lung cancer. (This is because the more prevalent a disease is in a given population, the more times the screening test will actually find the disease it’s designed to detect.) This matters a great deal because once a screening test says a patient has a disease, to confirm the presence of the disease progressively more invasive tests and procedures are often required, which exposes patients to progressively greater risks of complications. For example, a single chest CT poses little risk to anyone (the rate of contrast reactions is quite low in the general population), but the lung biopsy or wedge resection that will likely follow a positive scan are invasive enough to pose significant risk. And though uncommon, the following scenario can occur: a patient has a positive CT, gets a biopsy or wedge resection of her lung, and develops a complication that ultimately leads to her death—only to have the pathology on the tissue taken from her lung turn out to be benign. The risk of a similar scenario prevents us from recommending exercise treadmill tests, like the one my patient had, for people without symptoms or risk factors for coronary disease. False positive treadmill tests happen frequently in those patients, often leading to cardiac catheterizations, which carry a small but definite risk of death—a risk that’s only justified when the suspicion of a false positive result is sufficiently low.
- The risk of the test itself must be acceptably low. Blood tests represent almost the lowest-risk tests we have (the needle stings and some people get woozy or faint, but that’s about it) other than physical exam maneuvers. A colonoscopy is slightly more invasive and carries a slightly higher risk (colonic perforation is a catastrophic event but happens rarely). A cardiac catheterization, on the other hand, is distinctly more invasive (the puncture is arterial rather than venous) and therefore carries a distinctly higher risk, one high enough to disqualify it as a screening test altogether.
The number of tests that meet these criteria are far fewer than most people realize. Further, the utility of many that do satisfy them remains controversial. Consider the PSA test, a blood test used to identify asymptomatic prostate cancer. While it satisfies the three criteria listed above, the more studies that are done on it the less clear we become about what number actually represents a positive result. It used to be a PSA below 4.0 was thought to effectively rule out the presence of prostate cancer. Then we found a small but significant minority of patients with PSAs between 2.0 and 4.0 actually had it. Now we’re thinking this may also be true for patients with PSAs between 1.0 and 2.0. (To explain how researchers figure out what “normal” is for any given test would take an entire post in itself.)
In the end, convinced my patient’s treadmill test represented a false positive result, rather than take him to cardiac catheterization, I put him on a treadmill test again, this time using nuclear imaging (an addition that conveys a negligibly increased risk but far greater sensitivity), which, thankfully, was negative. We both heaved sighs of relief, and he became convinced that just because we can test for something doesn’t mean we always should.
Alex Lickerman is an internal medicine physician at the University of Chicago who blogs at Happiness in this World. He is the author of The Undefeated Mind: On the Science of Constructing an Indestructible Self.