When screening becomes personal: Why Biostatistics should matter to every physician
I introduced in an earlier post the idea that statistics can be both deceiving and revealing about the complex world around us, especially as they relate to screening and medical care. Notably, I believe that we might approach these discussions differently if the numbers represented individual people and their stories. Then we could more personally see the harms of misdiagnosis, overdiagnosis, and overtreatment.
Misdiagnosis is the concept described in my prior post. If a test isn’t sensitive enough, and the prevalence of a certain condition is low enough, then a test may return a positive result when the patient does not in fact have a certain condition. This is a “false positive” result and is an example of misdiagnosis. More subtle is when a test comes back abnormal, and the abnormality is indeed present in an individual, but was not causing symptoms and was never actually going to cause symptoms or death. This is not a false positive or a misdiagnosis, since the problem does not lie with the accuracy of the diagnostic test itself. Rather, it is an example of “overdiagnosis” – a diagnosis given to a condition which would never impact a patient. As we screen asymptomatic patients for earlier and earlier indicators of disease, we run the risk of identifying potentially innocent abnormalities and subjecting patients not just to overdiagnosis but also to the harms of further testing and overtreatment.
Halfway through medical school, home visiting my parents over the holidays, I found a card in the kitchen advertising “life-saving” local health screenings that cost a modest fee and were to take place at a local church. I found out that my mother had already made an appointment so I took a closer look at the offerings. The first was ultrasonography of the neck for the detection of asymptomatic narrowing of the arteries that supply blood to the brain (carotid stenosis). The test itself carries no health risks, and management of carotid stenosis through measures such as blood pressure control, antiplatelet therapy, and surgery have the potential to avoid life-threatening conditions such as strokes. To my mother, this tradeoff clearly favored screening. After all, to her, more medical care seemed to be better medical care.
It may seem surprising, then, that the United States Preventive Services Task Force (USPSTF), an independent panel of experts in prevention and evidence-based medicine, has published guidelines recommending against screening for asymptomatic carotid artery stenosis in the general adult population. They assigned a grade of “D,” noting, “There is moderate or high certainty that the service has no net benefit or that the harms outweigh the benefits.” They reason that screening would result in “many false-positive results,” which would “lead either to surgeries that are not indicated or to confirmatory angiography. As a result of these procedures, some people would have serious harms (death, stroke, and myocardial infarction) that outweigh the potential benefit that surgical treatment may have in preventing stroke.”
The key here is that though the ultrasonography itself may be harmless, physicians must consider how they will use the information from such a test in future diagnostic or treatment decisions. It is one decision to intervene surgically upon symptomatic patients who are having strokes as a result of their carotid artery stenosis, but screening broadly in the population would likely cause more harm than benefit as a significant number of people would have false positive results. Even some of the “true positives” (meaning that some degree of stenosis is in fact present) might be instances of overdiagnosis – diagnosis of a “disease” that will never cause symptoms or death in a patient’s lifetime. In such cases, by definition there is no potential for any benefit from treatment since the condition itself was never going to cause symptoms or death for the patient.
My mother had very few of the risk factors associated with carotid artery stenosis (these include older age, male sex, hypertension, smoking, hypercholesterolemia, and heart disease), so her probability of having carotid artery stenosis was low and any positive result on ultrasonography had a reasonable chance of being a false positive. If we would be inclined to not trust a hypothetical abnormal test result, why should we feel compelled to test each patient and then work-up and treat those abnormal findings which inevitably appear? As the USPSTF described, such interventions carry risks for each patient that may outweigh the benefits to the few patients who would have had clinically-significant disease resulting from true untreated carotid artery stenosis.
Yet, companies such as Lifeline Screening recommend that carotid artery screening be performed annually on “anyone over age 50” and “anyone over age 40 with risk factors”. They support this guideline by reasoning that “there are often no warning signs or symptoms of carotid artery disease” and that “9 out of 10 cardiovascular doctors support preventive health screenings for cardiovascular disease (plaque in the arteries) among patients with key risk factors.” Their website is littered with testimonials from patients who feel that their lives were saved alongside statistics supporting how many cases of “potential atherosclerosis and plaque buildup,” “elevated heart attack and stroke risk,” and “possible cases of critical vascular disease” have been identified among the 11 million patients that they have screened since 1996. They do not comment on whether all individuals over 50 truly fall into the high risk groups that 9 out of 10 cardiovascular doctors would support screening, nor do they question whether such “potential” plaque buildups or “possible cases” of vascular disease could be overdiagnosis in the setting of overzealous screening.
A logo from the careers page of Lifeline screening.
The “life-saving” health screening advertised to my mother also promised to search for abdominal aortic aneurysms (AAAs), which are when the wall of the largest blood vessel of the body gets weakened and bulges. This is again a condition for which the diagnostic modality (ultrasonography) carries few risks, the dreaded condition can have fatal complications (aneurysm rupture), and interventions exist to modify this risk of death. One-time screening is, in fact, recommended by the USPSTF for men ages 65-75 who have ever smoked given their higher risks of AAAs. But, in populations with a lower incidence such as women who have never smoked (including my mother), USPSTF again gives screening a “D” recommendation.
Equipped with this knowledge, I sought to persuade my mother that screening would be ill-advised. “But it’s quick, painless, and not too expensive,” she reasoned with me. “And what if they happen to find something?” I had to reframe the discussion for my mother. If all of the testing was within normal limits, then she would have been reassured and would also have averted any harms from misdiagnosis, overdiagnosis, and overtreatment. It was precisely if “something” was identified that she might have fallen into the group of false positives and suffered from further testing and interventions. She never would have known if she had truly found a latent life-threatening condition or if she had been forced to undergo testing, medication changes, and possibly surgery for a condition which might not have ever impacted her life.
The reminder for physicians is, as I’ve often been told as a medical student, to order tests only if they would alter some aspect of diagnosis and treatment. A “harmless test” may produce false-positive results if implemented in the wrong population and could lead to further diagnostic tests and treatment which could harm a patient. And this is not to say that ultrasonography for carotid artery stenosis or abdominal aortic aneurysm is a poor test. Rather, the diagnostic test is just one component in the axiom underlying the push towards smarter medical diagnostics – the ideal is the right test for the right patient at the right time. My mother, after discussing these possible outcomes with her primary care physician, elected not to get screened.
An advertisement for a one Lifeline Screening event in New York at which patients could pay $149 for carotid artery screening, peripheral arterial disease testing, abdominal aortic aneurysm ultrasound, an electrocardiogram to check for atrial fibrillation, and heel bone ultrasound to assess bone mass density and osteoporosis risk.
As I was writing this article, my mother found out that one of her college classmates had passed away suddenly in his late 50s from complications of an aortic aneurysm while visiting his family for Thanksgiving. The movement to recognize the harms of overdiagnosis and overtreatment does not aim to minimize the suffering experienced as a result of these conditions, nor does it seek to dissuade people from seeking medical care. The opposite is true — we have gotten into the habit of screening asymptomatic individuals precisely because we recognize the potential morbidity and mortality of certain diseases and we hope to prevent that suffering before it occurs. In doing so, we should continue to strive to identify those at highest risk, such that we can intervene where appropriate. At the same time, we must characterize those at the lowest risk so that they may avoid unnecessary testing and treatment. These risks are highly personal, taking into account medical history, family history, and other demographic and social factors, so patients should engage in their own risk discussions with their medical professionals and understand that more care is not always better care.
Do you have questions or comments on this article, or feedback for future columns in this epidemiology series? I welcome your thoughts at nschnure@mail.med.upenn.edu!