Dr. Timothy Wilt, a member of the United States Preventive Services Task Force, stood in front of the American Urological Association audience and explained why the task force could not recommend that men undergo routine PSA screening. At most, he explained, the test had been shown to benefit one out of 1000 men. Meanwhile, the test would cause hundreds of men to experience anxiety, and scores of them to experience impotence and incontinence from unnecessary treatments.
Twenty minutes later, I stood behind the same podium and asked the audience members to raise their hands if they disagreed with the task force's conclusion. Ninety percent expressed their skepticism. What happened in the time between Wilt's presentation and mine reveals a great deal about why experts cannot agree whether screening tests, like the PSA in middle-age men or mammograms in 40-year-old women, bring more benefit than harm, and about what psychological forces impede our ability, as a society, to figure out what basic bundle of healthcare services all insurance companies ought to pay for.
Wilt’s presentation was a model of scientific clarity. He explained that only two randomized clinical trials were conducted with enough scientific rigor to provide useful estimates of whether the PSA test saves lives. One trial showed no benefit and the other revealed the one in 1000 number which the task force took as the best case scenario. Wilt was followed on stage by Ruth Etzioni, a biostatistician at the Fred Hutchinson Cancer Research Center in Seattle. Etzioni presented a statistical model suggesting that the PSA test benefited many more than one in 1000 men.
Psychology was already at work. After hearing two scientific presentations, one emphasizing randomized trials (which physicians typically considered to be the gold standard of medical evidence) and another focusing on statistical models (which physicians typically consider too abstract to capture clinical reality), urologists found themselves embracing the scientific merits of the statistical models.
In a famous1979 study, 3 Stanford psychologists recruited students who held strong opinions about the death penalty, and exposed each of them to two hypothetical research studies, one suggesting that capital punishment deters crimes and one pointing to the opposite conclusion. All saw critiques of each study, as well as responses to the critiques. The result? The more information students saw, the more polarized their opinions became. Pro-death penalty students became even more in favor of the death penalty, while those previously opposed to the death penalty became even more opposed to it.
Not liking the task force's conclusion, urologists viewed the randomized clinical trials with deep skepticism, a skepticism they did not direct towards the equally shaky statistical model. We see what we hope to see.
One more presenter stepped behind the podium before my presentation: Capt. Mark Kelly, retired NASA astronaut and, just as famously, supportive spouse of Rep. Gabby Giffords. Kelly charmed the audience with the story of how he was diagnosed with prostate cancer four years earlier, at the age of 44, and was then cured by a urologist who removed his cancerous gland. Looking no worse for the wear, and joking about the benefits of not having to go to the bathroom in the middle of the night, Kelly was living proof of the benefits of PSA testing.
But did the PSA test save Kelly's life? Psychology was at work again. The task force estimate of one in 1000 lives saved was not established by measuring how many men, out of 1000, are diagnosed with prostate cancer by the PSA test but, instead, by estimating how many of those 1000 men are not only diagnosed but are also cured of a cancer that would otherwise have killed them. When a 71-year-old man has his prostate removed to treat localized cancer, it is impossible to know whether his life was saved by the procedure. If he dies at age 78 of a heart attack, for instance, no one can say for certain whether his prostate cancer, if untreated, would have killed him before then. The vast majority of men who undergo prostatectomy, in fact, do not live longer because of the operation. But here is where psychology comes into play: virtually every man who undergoes this arduous procedure is convinced that the operation has saved his life. Those who experience impotence or incontinence from the treatment are likely even more convinced of its benefits. How else can they overcome the cognitive dissonance of willingly undergoing a procedure with such significant side effects?
We may never know the true benefits of PSA screening. It is almost impossible to conduct a clinical trial in which those men randomized to the “control group” are willing to forego screening tests for the duration of the trial, meaning that the trials will inevitably underestimate the benefits and harms of PSA screening. That leaves us with statistical models that will only confirm people’s pre-existing biases.
Our best hope is to conduct high-quality randomized trials of newer, potentially better screening tests, trials in which men in the control group are not allowed to obtain the new screening test until the trial is complete. No matter how promising the new test appears to be, physicians and insurance companies need to agree not to offer, or pay for, the test outside of the trial.
Meanwhile, men need to decide whether the unproven benefits of the PSA test outweigh its proven harms. This is no simple decision, as Capt. Kelly's story makes clear.
At a 50-year-old physician myself, now six years older than Kelly was at the time of diagnosis, I have decided to hold off on PSA testing. Meanwhile, I will stick with tests that yield more certain benefits. That means the word colonoscopy will be taking on new relevance in my life very soon.