Doug Robertson is the kind of doctor who eats his own dog food. As a gastroenterologist in the Department of Veterans Affairs health-care system, he is overseeing a 50,000-person study comparing two different ways to screen for colon cancer: Patients aged 50 to 75 are randomly assigned to receive either a colonoscopy or a fecal immunochemical test, which can be conducted at home and detects tiny amounts of blood in a patient’s poop. As a way to honor the veterans who volunteered to have their medical care determined by chance, Robertson, when he turned 50, decided to randomize himself too. A flip of an oversize novelty coin determined the (informal) outcome: tails—the doctor would get at-home fecal tests for his colon-cancer screenings in the years to come.
Most of us aren’t a Doug Robertson; we’re not willing to let important decisions regarding our medical care be decided by dumb luck. As a result, much of our medical knowledge is left with surprising and uncomfortable gaps. The National Cancer Institute has spent $100 million on a breast-cancer screening trial comparing “3-D mammograms” with traditional mammograms, for instance, yet it’s still struggling to recruit enough participants. Radiologists are already convinced that 3-D mammography is better, and that belief has trickled down to patients. Answers to even more fundamental questions—such as whether any kind of screening is better than no screening at all—are simply unattainable, because, ethically and logistically, it’s all but impossible to run a study in which one group of patients doesn’t get the standard suite of cancer tests. Could modern chemotherapies, surgeries, and diagnostic tests have made early detection of a tumor less important than it was before? We just don’t know. Does getting screened for cancer really prolong your life? We aren’t sure. Even screening tests with proven benefits, such as pap smears for cervical cancer, also have potential harms to consider.
Scientific brawls over these issues have been known to tear professional friendships apart, pitting doctor against doctor. It’s hard to exaggerate how acrimonious these debates become. Anti-mammography advocates have referred to messaging associated with Breast Cancer Awareness Month as “pink month lies,” while pro-mammography advocates have said that undermining the technology is “going to get women killed.” Such arguments, however, were always theoretical. No one could imagine a modern health-care system in which screening didn’t play a central role—in which millions of people simply skipped their annual mammograms and prostate-cancer tests. Certainly no one could imagine that the American Cancer Society itself might one day suddenly declare, “No one should go to a health care facility for routine cancer screening.”
At least, no one could until last year.
Think back to March of 2020. Hospitals around the country were bracing for a wave of patients as COVID-19 took hold in the United States. The disease had overwhelmed Italy’s health-care system, and New York City’s was beginning to seem fragile. In an effort to triage hospital beds, save scarce personal protective equipment, and reduce patients’ exposure to infection, hospitals canceled all elective surgeries and procedures. Other medical conditions didn’t cease to exist, of course, but physicians and patients were focused on the looming dangers of the pandemic. Even supposed emergencies such as heart attacks and strokes suddenly went missing from the health-care system.
Cancer care, in particular, was quickly transformed. Although hospitals still provided the most urgent forms of cancer treatment such as chemotherapy infusions, routine screenings were postponed. No more mammograms, pap smears, or colonoscopies were put on doctors’ schedules. The Early-Detection Empire, which the health-care system had been building over decades, crumbled in an instant.
At first this looked like another tragedy, layered over the one delivered by the coronavirus. Cancer experts started fretting about a second pandemic of advanced disease that would result from all the diagnostic delays. In June 2020, Norman Sharpless, the director of the National Cancer Institute, warned that “ignoring life-threatening non–COVID-19 conditions such as cancer for too long may turn one public health crisis into many others.” His agency’s own calculations suggested that, as a result of the pandemic, there could be 10,000 additional breast- and colon-cancer deaths over the next decade.
Yet it also seemed plausible, at least, that the suspension of cancer screenings might do less harm than anticipated. In April 2020, I published a commentary suggesting that the changes brought on by the pandemic would amount to a very rare, and perhaps massively informative, perturbation of standard health-care practice, and that it might even teach us whether cutting back on cancer screening really is a mortal threat. Other doctors said more or less the same in The New York Times, CNN, and Bloomberg. We’d all converged on an alluring hypothesis: The pandemic would provide medicine with the greatest natural experiment on the efficacy of screening in modern history.
After more than a year, some results from this experiment—or rather, some illuminating non-results—have finally become clear. We’ve learned, first and foremost, that the modern colossus of routine screenings is so fundamental to American health care—so central to its functioning—that even cataclysmic world events could not do much to knock it off its perch. Consider the estimated toll of delayed screening, as laid out by the National Cancer Institute in the summer of 2020. That was based on the assumption, which seemed reasonable at the time, that 75 percent of all screenings for breast and colon cancer would end up being skipped over a period of six months. We now know that COVID-19’s disturbance of the system lasted far less time than that. According to a review of health-insurance databases published in April 2021, screening rates for breast and colon cancer dipped by that amount only during one month of the pandemic (April 2020), before bouncing back to near-normal levels by June. In other words, our historic natural experiment on cancer screening was over shortly after it began.
The speed of this rebound, even in the midst of COVID-19 outbreaks, is a bit surprising in itself. With so many other services and businesses still closed, the health-care system worked relentlessly to reschedule patients for all of their mammograms, colonoscopies, PSA tests, pap smears, lung-cancer scans, and skin checks. Hospitals suffered immediate financial distress when elective procedures were canceled at the start of the pandemic, and they had to make the pause as brief as possible. “It was bad for business,” says Laura Esserman, a breast-cancer surgeon and screening expert at UC San Francisco. “So they reached out to get everybody back in.”
A huge amount of money was at stake. The annual cost to patients and insurance providers of screening exams has been estimated at $447 million for prostate cancer, $6.6 billion for cervical cancer, $7.8 billion for breast cancer, and $23 billion for colon cancer. That amounts to at least $38 billion that hospitals and clinics might have lost if a full pause had gone on for a year. This alone may represent a fraction of the health-care system’s overall revenue, but the financial impact extends far beyond the price of these procedures: Each positive diagnosis corresponds to another cancer patient, who will then come in for additional clinic visits and treatments. For doctors who specialize in identifying or treating these cancers, screening could make up a large portion of their livelihood.
Doctors like me saw the money crunch that resulted from just a two-month pause in screenings. A number of my physician colleagues who practice pathology, radiology, urology, gastroenterology, and dermatology—all screening-heavy specialties—faced pay cuts or had job offers revoked. One survey found that 62 percent of physicians had seen their salary reduced by May 2020. Some practices resorted to begging for donations on GoFundMe. With screening and other forms of elective care now back in action, physician hours and salaries are also starting to bounce back.
Whether you consider cancer screening’s rapid return to be a sign of health care’s resilience or its intransigence depends on what you thought about the practice before the pandemic. Either way, the status quo has proved to be astonishingly durable. Last summer, when many people were struggling to get timely COVID-19 test results from dirty parking-lot swab stations, some health-care systems were opening spa-like mammography suites featuring the “sights, sounds, and smells of the seaside, garden or waterfall.”
“Once stuff gets into the business of medicine, it makes it so hard to change,” Esserman told me.
The natural experiment on screening may have been cut short, but it was still unprecedented. For about two months in 2020, hospitals and clinics really did give up billions of dollars to keep people safe, and patients still haven’t fully caught up on millions of missed exams. It’s important to take stock of the effects this had on people’s health.
Some doctors have already drawn dire conclusions. Two-thirds of radiation oncologists say their patients are showing up with more advanced cancers than before the pandemic, according to a survey released in March. An analysis out of Modena, Italy, suggested that a two-month delay in mammography, from mid-March to mid-May 2020, caused an 11 percent rise in metastatic breast cancer. Doctors at the University of Cincinnati reported seeing the proportion of lung nodules that were likely cancerous more than triple when screening there resumed, from 8 percent to 29 percent.
These kinds of alarming results don’t make sense to everyone. “I don’t think you could credibly know whether or not a delay of one or two or three months could have a causal effect,” Anupam Jena, a physician and economist at Harvard, who studies natural experiments, told me. Cancers we screen for tend to be slow-growing, he said, so we wouldn’t expect to see an enormous rise in cancer rates. The average screen-detected breast cancer, for example, might be expected to grow only a few millimeters in a couple of months. According to Esserman, it’s most important to find breast cancers before they get to about two centimeters, or before they’ve spread. But doctors and the public “hold on to this very fervent belief that millimeters make a difference.”
Interruptions to standard medical care posed the greatest risk for patients with concerning symptoms, Esserman said, or those who had gotten recent cancer diagnoses. “That was a real tragedy of the pandemic,” she told me. “People who had a mass, even if they called, were told, ‘Well, why don’t you just wait.’ Those are the people who had consequential cancers, who had real delay.” The full effects of a screening pause, and thus of doctors finding fewer cancers in patients without symptoms, remains uncertain. But the data are clear on the importance of the care that’s given after diagnosis. Once a cancer has been identified, patients have a greater chance of dying the longer they wait to start treatment.
We have no way of knowing yet whether the pandemic has led to a higher overall number of cancer cases or deaths, whether on account of last year’s screening pause or for another reason. (The most reliable annual statistics, produced by the American Cancer Society, are not yet available for 2020 and 2021.) Disturbing early observations in Modena and Cincinnati may be partly the result of a statistical illusion. Individual physicians and researchers could be treating a different population of cancer patients—with different symptoms or risk factors—than they were before the pandemic. If that’s the case, then they might end up seeing worse outcomes even where the total number of serious cancers across all people in their community stayed the same. The Modena and Cincinnati studies each looked at cancer rates in a single hospital at the very beginning of the screening rebound. Their findings may not hold over time or in other locations. Researchers in the Netherlands were able to track all breast-cancer diagnoses throughout their country, for example, and found no increase in advanced cancers during the same period.
There is a better way to draw useful information from last year’s screening pause. Instead of looking only at the people who happen to show up to a single clinic, researchers could follow a larger, more diverse group of patients over time. In particular, experts told me, we should focus on the patients who were (or weren’t) seen by their provider just before and after shutdowns began. Say a hospital decided to cancel all elective procedures on March 19. Patients who had their mammogram or pap smear or lung-cancer scan scheduled on March 18 would have gotten their exam, while those who happened to have made their appointment for March 20 would have been out of luck. Appointment dates are essentially arbitrary, so the patients who got their screening just before the pause should be very similar, in general, to those who waited months to resume their medical care. Voilà, a pseudo-randomized trial is born! A natural experiment like this could provide information well into the future. Although some patients who missed an exam rescheduled it as soon as possible, others might have decided to skip that year’s test altogether, and catch up in 2021. Still others may end up never getting screened again. Scientists could track screening’s “pandemic generation” for the rest of their life and see what happens.
The public is going to be inundated with analyses like these, says Aaron Schwartz, a physician and health-policy researcher at the University of Pennsylvania, but “it gets messy very fast.” The pandemic disrupted so much of our daily lives at once that we may not be able to tease out the effect of screening delays. What if a patient who skipped his colonoscopy also lost his health insurance and started eating more junk food? We couldn’t say for sure which factor was to blame for a bad outcome. Because of these simultaneous changes, “it’s a very tall order to measure the effectiveness of cancer screening,” Schwartz told me.
Perhaps I was naive in spring 2020 for suggesting that new data would settle one of medicine’s great controversies. Doctors aren’t number-crunching automatons; we’re just as often swayed by our experiences. What the natural experiment of the pandemic did do was show millions of people that there are various ways of receiving (or not receiving) care. “A patient that skips their cancer screening because of COVID might think differently about how important it is,” Schwartz said. Doctors, too, have been shaken out of their routine: Some physicians who started offering home-based colon-cancer screening, for example, will keep doing so after the pandemic—long before we get definitive results about the method from Robertson’s trial at the VA. In the end, some important decisions regarding our medical care do get decided by chance and circumstance.