Making the Call

A recommendation against PSA screening?¹ I’ve been raising questions about the value of cancer screening for more than a decade,² yet I’ve never recommended against it. But then again, I’ve never recommended for screening either. I guess that's why I’m not a member of the US Preventive Services Task Force.

As you can probably tell, I am a fence-sitter (and, by the end of this essay, maybe even a flip-flopper). While I believe that media campaigns, advocacy groups, and physicians have systematically exaggerated the benefits of cancer screening and have downplayed—or, worse, ignored—its harms, I also believe that both benefits and harms exist. And no matter how well researchers can quantify this trade-off (and there is much room for improvement here), the metrics are different: small changes in cancer-specific mortality vs a boatload of hassle factors and fear, some unnecessary treatment, some resulting complications, and even a very few deaths (that can be missed in the measurement of cancer-specific mortality³ ).

That leaves me with an uncomfortable reality: there is no calculus or decision model that will produce a single “right” answer. Instead, physicians and patients are stuck comparing apples and oranges—in other words, we are left making value judgments about the net effect of cancer screening.

I’m happy to make this value judgment about prostate cancer screening for myself. I’m happy to accept the most favorable randomized trial as an estimate of benefit—on the order of one man avoiding a prostate cancer death out of 1000 men aged 55 to 69 years screened over 10 years⁴ —and ignore the less favorable data.^{5 - 6} Simply knowing that the overdiagnosis harm is somewhere around 30 to 100 times the estimated benefit⁷ —and knowing what treatment and its complications entail—is enough for me. I don't need any more data.

My value judgment is simple: It's an awful deal. And that's without even considering the unnecessary biopsies and their related complications.⁸ Or the craziness of wondering what PSA level I would agree to act on (in an effort to mitigate overdiagnosis)—and whether I would have second thoughts about that decision. I’m happy to forgo the test and accept that I may be the one man out of 1000 who could have benefited, simply to avoid these much more common harms.

I’m less happy, however, to make this value judgment for others. In clinic, I never begin the discussion about PSA screening with my recommendation; instead I default to my pragmatic version of informed decision-making. Some of my patients are sure they want PSA testing. I don't spend a lot of time trying to determine how informed their decision is; they get the test. Others are sure they don't want PSA testing. Again, I don't spend a lot of time trying to determine how informed their decision is; they don't get the test. Some patients want to know more (and, frankly, they probably get more information than they want).

But the Task Force is charged to make a recommendation. I would have advised them to recommend that the PSA decision should be an informed decision. I would have advised them that all screening decisions should be informed. Who could argue with that?

What a cop-out. What a fence-sitter.

The Task Force's decision to recommend against PSA screening challenges me to rethink informed decision-making in cancer screening. It can't always be the right answer. To truly inform decisions requires a tremendous amount of time and effort: not only in creating and updating decision-making materials, but also in participating in subsequent discussions with patients and assessing their understanding (as you see above, I don't do that very carefully). Such effort is warranted when the stakes are high—decisions about major surgical or medical interventions that can alleviate problems patients currently face, yet also pose substantial harms.

But the effort is not always warranted. In cancer screening the stakes for benefit are necessarily low—the “patients” are asymptomatic, and the vast majority are not destined to die of the target cancer. A 3% lifetime risk of prostate cancer death⁹ means that 97% of men have to be informed about a test that cannot help them—it can only harm them. And the benefit only accrues to a small fraction of those destined to die of the disease. How well can we physicians communicate this tiny probability of benefit (<0.1% reduction in prostate cancer mortality over 10 years) or our uncertainty about any effect on all-cause mortality?

If there are high stakes in prostate cancer screening, they involve the harms. How well can we physicians communicate what it's like to have an infection following biopsy, incontinence and erectile dysfunction following surgery, or proctitis following radiation? How well can we communicate that undergoing screening is the fastest way to be diagnosed with prostate cancer and more than doubles the lifetime risk of being diagnosed with the disease?¹⁰ And do we physicians even acknowledge that the entire process of raising awareness to promote screening and sharing decisions may itself introduce new “dis”-ease into the population?

What does it take for the medical community to simply make a recommendation?

I worry that excessive fealty to informed decision-making in cancer screening may distract us from more important issues—like what is bothering patients right now. It may also lead physicians to abdicate our role as advisor—physicians simply transfer responsibility to their patients. And the problem appears to be asymmetric: we seem less inclined to recommend against than to recommend for. No one suggests that physicians should engage patients in the decision of prescribing aspirin following myocardial infarction. We physicians routinely recommend it.

Why are we so reluctant to make value judgments about things not to do to patients? Is it because we are so fearful of potential litigation? Is it that we have become so financially conflicted? (The chief medical officer of the American Cancer Society once remarked that one PSA test has an expected Medicare return to our hospitals of about $5000, from ensuing biopsies and treatment.¹¹ ) Or is it that we fear being accused of caring about cost? (Even if the United States had all the money in the world to spend on medical care, I’d still argue that PSA screening is bad deal.)

Maybe clinicians need the Task Force to make the call for us. And, for PSA testing, they did. If I was pressed to get off the fence, it's the call I would have made.

The United States now needs the medical profession to make more calls like this. There have always been plenty of calls to do more. But the nation also needs the profession to communicate the nuances of medicine. Most tests and treatments are neither unambiguously good nor unambiguously bad. Instead, their effect is modified by how they are used.

Nowhere is nuance more relevant than in cancer screening. The trade-off of benefit and harm is sensitive to who is tested and what treatment is employed for detected abnormalities. But it is most sensitive to what constitutes an abnormal test—the question of how hard to look for cancer.

Let's be honest: in the case of the prostate, for the past two decades we’ve been looking too damn hard. That's what's led to so many biopsies and so much overdiagnosis. Maybe those of us who have worked to inform the decision should have spent our time working to make the trade-off more favorable. How? By looking less hard for prostate cancer.

We physicians have to be willing to ask some heretical questions. These questions must address the basic parameters of screening: the test threshold (what if the PSA had to be higher than 10 ng/mL to be considered abnormal?), the biopsy threshold (what if only nodules larger than 1 cm were biopsied?), and the role of dynamic observation (what if treatment was only contemplated after a threshold doubling time was exceeded?). Strategies like these might well preserve whatever mortality benefit exists for PSA screening; they would certainly dramatically reduce its harms.

But such strategies will never be seriously considered and tested until the public understands the nuances of cancer screening—in particular, why the goal can't be simply to find as many early cancers as possible. Another task for the Task Force?