When I was a medical student in the 1980s, the beating heart of the Hospital of the University of Pennsylvania was not the mahogany-lined executive suite, nor the dazzling operating room of L. Henry Edmunds, Jr., HUP's most famed cardiac surgeon. No, it was in the decidedly unglamorous, dimly lit Chest Reading room, where all the X-rays were hung on a moving contraption called an alternator that resembled the one on which the clothes hang at your local dry cleaner. Controlled by a seated radiologist operating a foot pedal, the machine would cycle through panel after panel until it arrived at your films. The radiologist took his foot off the pedal, the machine ground to a halt, and the dark X-ray sheets were brought to life by intense backlighting.
At Penn in the 1980s, everybody — and I mean everybody, from the lowliest student to the loftiest transplant surgeon — brought films for deciphering to the late Wallace Miller, Sr., a crusty but endearing professor of radiology and one of the best teachers I've ever known. For students like me, time spent with him was at once exhilarating and terrifying. "What's this opacity?" he asked me once, the memory burned into my hippocampus by that cognitive curing process known as overwhelming anxiety. "A ... a pneumonia?" I stammered.
"Mooiaaa," retorted The Oracle, an unforgettable signature sound uttered as Miller smartly turned his head away in mock disgust. I loved it. We all did.
Today, many of my internal medicine trainees barely know where the radiology department is. Just as your record player and LPs are now long gone, in your local hospital today, the films, the analog X-ray machines, and even those charming film conveyor belts have left the building.
Why? In 2000, only 8 percent of U.S. hospitals had some version of a game-changing computer technology called the Picture Archiving and Communications System, or PACS. By 2008, more than three out of four did.
Because radiology was the first medical specialty to computerize, what has happened to it — at once shocking and, in retrospect, entirely predictable — is our canary in the digital coal mine, its experience offering important lessons for patients, clinicians and health care systems.
The Beauty of PACS
While the main catalyst for PACS was economic, the quality of the images and the ability to manipulate them were also important. Unlike regular films, CT scans need to be viewed at various contrast levels: One setting is best to look at bones, another to look at lungs, and still another to look at soft tissue like muscle.
PACS allowed radiologists to toggle through these views, in the same way that Instagram lets you play with your photos. You can also use a nifty magnifying glass to zoom in on a part of the image. An unexpected benefit was “stacking”: rather than looking at 100 images arrayed in a 10 × 10 grid on a one-dimensional page, the images could be digitally stacked, one on top of another, allowing the radiologist to scroll through them swiftly by rolling a mouse ball. Moreover, computerization let the radiologist look at the images from home, enabling senior experts to weigh in on subtle findings that trainees might flub. And while the images were fuzzy at first, today they’re as crisp as high-definition television.
Perhaps most important, PACS obviated the need for maddening searches for prior X-rays. Twenty years ago, when a chest X-ray revealed a lung nodule, the first commandment on the radiologist’s report was to “obtain old films.” The rationale: If the nodule had been unchanged for many years, it could safely be ignored—such stability simply wasn’t consistent with a diagnosis of cancer. But searching for old films was often an exercise in frustration: They were lost, or locked up, or at another institution, or in a filing cabinet in the thoracic surgeon’s garage, behind the golf clubs. When my colleagues and I came up empty-handed, which was more often than not, the patient frequently paid the price in the form of an unnecessary biopsy. But PACS made finding old films a breeze (assuming that they were done at the same hospital or had been scanned into the system); they’re usually just a click away.
While PACS was widely anticipated and generally accepted by radiologists, some prescient observers worried that computerization might lead to unbidden effects on the field. In 1999, Stephen Baker, chair of the Department of Radiology at New Jersey Medical School, fretted that PACS might turn radiologists into “disembodied functionaries, more akin to servicing technicians than professional colleagues.” Paul Chang, professor of radiology at the University of Chicago and an early leader in digital radiology, describes the day his father, a retired radiologist, took him to task.
“Before PACS, we were the doctor’s doctor,” his father berated him. “Medicine and surgery rounds started in radiology. . . . Every morning the clinicians and the radiologists collaborated.”
His father’s less-than-endearing nickname for his famous son: “The Man Who Ruined Radiology.”
The advantages of PACS are so vast that few would want to turn back the clock. Yet the effects on those of us who order X-rays and the radiologists who read them have been profound, and they’re not all positive. The fact that we can now review our images without trekking down to radiology means that we rarely do make the trip.
An Awkward Trip to Radiology
A few years ago, when I asked my interns and students to visit the radiology department to review the key films, they looked at me as if I had grown a second head. After my team humored me by accompanying me to the radiology department, I conducted a little sociology experiment. Standing outside my hospital’s chest reading room, I delivered a brief speech:
"Watch what happens when we enter. Does anybody turn around and welcome us, ask, 'How can I help you?' and seem genuinely enthusiastic? When they go over the X-ray, do they delve a layer deeper than what they said in the formal report? Do they make any teaching points? Does the radiologist suggest courses of action or ask provocative questions?"
I did this because I am deeply concerned that mine is the last generation to have learned the habit of going to the radiology department. Nostalgic for my interactions with Wally Miller and his like, it saddens me that our current trainees will never know how much they can learn from a great radiology teacher, and how much their patients’ care can be improved by actually talking to a real live radiologist. Yet I know that even if I bring my young horses to water, whether they visit the radiology department after I am no longer their wrangler will be determined by the quality of their experience.
We entered the chest reading room and were greeted by a wall of radiologists’ backs, their faces trained like lasers on the computer screens in front of them. Not a single head—located atop the shoulders of about eight different radiologists—turned to greet us.
After a couple of awkward minutes of crescendo throat-clearing, one of the radiologists grudgingly swiveled around to face my team and me. “Oh, do you need something?” he asked.
“Sure; can you help us look at a few films?”
He did, kind of, but offered his help in a whisper animated mostly by passive aggressiveness.
I thought it couldn’t get any worse, but it did.
“What do you think of this area?” I asked him, pointing to a confusing patch of whiteness on one patient’s chest CT scan.
“Did you look at the official report?” he hissed. (In other words: "Perhaps you don’t know how to turn on your computer?")
The unspoken message was clear: Get out of my space; I’m busy.
Now, I understand that he might well be busy, and that it has to be annoying having clinicians interrupt you every few minutes to go over images, particularly after you’ve just reviewed them with a different set of specialists and dictated a report. But that is the radiologist’s job. Or at least it used to be.
Allison Tillack, a young radiologist and a medical anthropologist whose Ph.D. thesis involved observing the world of radiologists for a year at a prominent academic hospital, has explored how the computerization of radiology has transformed the worlds of radiologists and those who use their services.
"The ability of PACS to alter the accessibility and tempo of medical imaging has resulted in visits to the reading room being viewed now by non-radiology clinicians as a ‘waste of time’ and by radiologists as an ‘interruption,’” she wrote.
In a Funk
While I was well aware of the changing perceptions of radiology by nonradiologists, I had not, until I met Tillack, appreciated the degree to which the field of radiology is itself in a PACS-fueled funk.
After all, the field remains extremely popular among medical students, as many perceive it as offering the perfect blend of “great lifestyle” (that is, banker’s hours and limited overnight call) and high income, which averaged $340,000 in 2013. In fact, it’s often said that today’s medical students are attracted to the “ROAD specialties”: Radiology, Ophthalmology, Anesthesiology, and Dermatology, all of which are lucrative and none of which involves a lot of contact with those pesky sick people. In her research, Tillack found that the vast majority of radiologists and radiology residents identified the lack of direct patient contact as one of the main attractions of the field.
Given all these pluses, many frontline clinicians think of radiologists as having “won the game.” Yet I should have gotten a hint of the field’s handwringing in 2005, when I saw the results of a survey of physicians regarding their satisfaction with their chosen specialty. The happiest doctors were radiation oncologists (the folks who deliver radiation therapy to cancer patients), who do satisfying work, earn a good income, and have predictable hours. The least happy were cardiac surgeons, who train forever and, in recent years, have seen much of their business eroded by stents and other nonsurgical approaches to heart disease.
Radiologists show up a bit below the mean on the satisfaction scale—just behind the perennially overwhelmed and undercompensated primary care doctors.
In a 2012 paper, Tillack and a colleague described “the loneliness of the long distance radiologist.” One radiologist told them, “Before, I knew the face, name, wife’s name, and kids’ names of all the clinicians, but now I don’t know who you are if you joined the medical staff after we got PACS. . . . Before, when a clinician showed up, I could ask them and find out what’s really going on with the patient.”
I hear similar stories from every radiologist I meet. Patrick Luetmer, a Mayo Clinic neuroradiologist, described what happened when his MRI suite was remodeled. The suite was originally configured with two MRI “donuts” (the huge magnets that are responsible for the image) on either side of a central workstation in which Luetmer sat. There, he could monitor the scans as they were being performed, and talk to both the patients and the radiology assistants. Clinicians sometimes wandered down to look at the scans with him.
A few years ago, as part of a big efficiency push, Mayo decided that a third MRI machine was a better use of that central area than the radiologist’s air traffic control desk. Luetmer’s workstation was moved to an office a few hundred feet away, where he could follow the scans on his computer monitor and communicate with the techs via a special text messaging system.
“One day I tried to see if I could go the whole day without speaking to anyone. And that’s what happened—I didn’t speak to a single person. It was incredibly isolating.”
The radiologists were lonely, sure, but their situation involves something far deeper. Hari Tsoukas, an expert in organizational studies at the University of Cyprus, highlights the distinction between “information” and “knowledge.” Information, he wrote, “consists of objectified, decontextualized, time-less, impersonal, value-free representations,” whereas knowledge is “context-dependent, personalized, time-bound, and infused with values.”
Adds Tillack, “Hunches, hypotheses, frustrations with patients or their families, second guessing, judging of colleagues, and similar activities that mark how uncertainty is negotiated on a daily basis in medical practice are rarely reflected in the medical record . . . this knowledge can only circulate in private or semi-private contexts—by its very definition, this knowledge cannot be made a matter of public record.”
By purging the informal conversations during which such information was exchanged, the emergence of PACS left radiologists information-rich (Think of all those pixels! And old films just a click away!), but knowledge-poor.
'Great Case. Next Case.'
Radiologists’ alienation runs deeper than the lack of collegial exchange and the inability to find out what’s really going on with the patients. It’s also about power, status, and expertise. The fact that the traditional film lived only in the radiology reading room gave radiologists a monopoly over their entire ecosystem. PACS, observes Tillack, created a new normal in which “the ‘right’ to see [the image] is no longer mediated by radiologists, as it was in the reading room,” and has thus “eroded radiologists’ claims for authoritative knowledge over the interpretation of medical images.”
Once the radiology department no longer housed the films, the impact was immediate and dramatic. Without any changes in policy or very much forethought, the mid-1990s transition to filmless operations at the Baltimore VA hospital led to an 82 percent decrease in in-person consultation rates for general radiology studies. Today, many clinicians—particularly specialists like neurologists, pulmonologists, and surgeons—look at images themselves and act on their own interpretations; Many don’t even bother to read the radiologist’s formal report (which usually takes several hours, sometimes even a day, to reach the chart) unless they have unanswered questions or judge the study to be particularly challenging.
PACS was to increase efficiency, but that virtue has also become a curse, as radiologists increasingly feel like Lucy and Ethel on the assembly line of the chocolate factory. Among teleradiologists (radiologists reading x-rays from a distant site, often covering emergency departments at night while the hospital’s own radiologists are sleeping), there’s a well-known adage that captures the relentless objectification of their modern predicament: “Great case. Next case." As with so many other aspects of our modern digital lives, PACS sped up the clock, and did so without mercy.
That clock is constantly ticking. “Instead of waiting for films to be acquired, printed, sorted, and hung, radiologists now are always playing catch-up, looking at more ‘stuff’ in less time,” observed Tillack. That miraculous access to old films also creates an obligation for the radiologist to actually review them.
And it’s not just the old films that need to be examined; PACS makes vast amounts of information available with every study. In the early days of CT, the output of a scan might have been about 12 “slices,” each one representing a ¼-inch section through the thorax or abdomen, akin to a thick slice of deli-cut salami. But today’s ultra-fast CT scanners can produce images of more than 50 slices per inch of the human body, more like ultra-thin cuts of prosciutto. And PACS, with its massive memory bank and blazingly fast transmission speeds,can easily display every slice, which means that the radiologist has to scroll through hundreds of images in order to read a single CT study. This combination of more information in each scan, more old studies to compare, and more time pressure is unremitting.
The clock is ticking for other reasons as well. Since the image is available to the ordering clinician the moment it is created, radiologists feel obliged to perform their review quickly lest their reading seem like old news, like an afternoon newspaper in the Age of Twitter. Piling on, after recognizing the efficiency of PACS, insurance companies and Medicare slashed the reimbursement for each interpretation, pushing radiologists to read more films in less time in order to maintain their incomes. Said one radiologist, “With PACS, work is busier now. We have 70 percent more cases to read than 10 years ago. . . . At the end of the day . . . I’m fried.”
On top of this, there are even greater threats to radiologists’ livelihoods and happiness. One of them flows from the growing pressure on health care systems to slash their costs. Currently, virtually every X-ray performed at a U.S. hospital is sent for a formal reading by a radiologist, who is paid a fee by an insurance company. In today’s cost-cutting environment, it’s probably only a matter of time before some health care systems permit their frontline specialists to officially read certain films, reserving radiologist “overreads” for those images that the clinicians have questions about or the ones with super-high malpractice risk if they are misread. Radiologists can be counted on to fight such a move by frantically waving the banner of quality, but they will need to demonstrate that the value of having them review every film is worth the considerable expense.
Moreover, a major theme of Obama-era health reform is a shift from our historical fee-for-service, piecework payment model to one that dispenses a single payment to a hospital and doctors to manage all the care for a group of patients (“accountable care organizations,” ACOs for short) or a given episode of disease (“bundled payments”). Under such systems, the risk for the cost of care shifts from the insurer to the providers, and it’s up to the latter to decide how to divvy up the cash. Ron Arenson, chairman of the department of radiology at the University of California, San Francisco, sees this as the greatest threat to his field.
“If the world moves to bundled payments, we won’t do well,” he said. “We’re not very high in the pecking order.”
Some nonradiologists, particularly ER doctors working nights and weekends, have little sympathy for their colleagues’ new predicament. In fact, they have begun to wonder why radiologists should be compensated for next-day readings when they’ve already looked at the images themselves, acted on their interpretations, and assumed the risk of being sued if anything goes wrong. In a 2011 editorial entitled “The Life Cycle of a Parasitic Specialist,” ER physician William Mallon took off the gloves.
[On Monday morning] these parasites will commence to feed on the financial juices of the lowly unfortunate emergency physicians, who had to work the entire weekend without radiologic support or backup. . . . The radiologist arrives well rested, café latte in hand, and promptly installs himself in a dark room to re-read and bill for all the films the emergency physicians read over the weekend. . . . Never has a specialty done so little for so many and been paid so much.
Another challenge to radiology made possible by the death of film has come in the form of teleradiology. Once X-rays went digital, it was no longer crucial for radiologists to be in the same building as the patient or the treating clinicians. As a result, many multihospital systems consolidated their reading rooms, particularly on weekends and nights, with centralized radiologists supporting multiple sites. Predictably, once the technical challenges of connectivity were solved, teleradiology companies emerged to fill this need. As is often the case with contented “legacy” providers (in health care and other industries), traditional radiologists were only too happy to have their colleagues read their films during off-hours. Who wouldn’t be?
The playing field soon expanded across national borders, as radiologists in Zurich, Israel, and Singapore began to read nighttime X-rays for American hospitals during their own local daytimes. Hundreds of hospitals now use these "nighthawks," and everybody seems happy about it, including the domestic radiologists, who are sleeping soundly while the overnight images are read half a world away.
But one wonders whether this is the start of so-called disruptive innovation, the concept made famous by Harvard’s Clay Christensen. Disruption often begins with a fat and happy incumbent content to preserve its existing enviable position in a market. In industries ranging from commercial aviation to steel manufacturing, an upstart comes in and grabs an unattractive part of the market (in this case, nights and weekends). But once a low-cost company has squeezed through a crack to capture a slice of a previously locked franchise, it is rarely content to stay put. With the average U.S. radiologist earning about $350,000 per year and the average Indian radiologist earning less than one-tenth of that, one wonders whether the same World-is-Flat forces that have revolutionized other industries but mostly bypassed health care will be unleashed.
This is where radiologists’ loss of trust and collegiality with other clinicians may exact its heaviest toll. “Some people see teleradiology as a big threat, but I don’t,” UCSF’s Arenson told me. “I think that relationships with radiologists are important.” I do too, which is why I believe he may have his head in the sand: If physicians don’t get much out of visiting the radiology department or have even forgotten where it is located, we have little reason to fight to keep it in our buildings. Or, for that matter, our country.
Like all legacy providers faced with a technological or global workforce threat, radiologists can be counted on to argue that quality would take a huge hit if we outsourced their work to less expensive providers, domestic or foreign. The degree to which the field has accepted nighttime readings from non-U.S. radiologists will, of course, undermine this argument. It’s hard to make the claim that a Bangalore-based teleradiologist is sufficiently competent to read an image for your hospital at 3 a.m., but not at 3 p.m.
The Ultimate Threat
Finally, there is the ultimate threat: replacement by the machine. Of course, this issue is marbled throughout health care as we enter the digital age. To date, most claims that “this technology will replace doctors” (in areas ranging from diagnostic reasoning to robotic surgery) have proven to be hype.
However, in fields that are primarily about visual pattern recognition, the promise (or, if you’re a radiologist, the threat) is much more real. Studies have shown that computers can detect significant numbers of breast cancers and pulmonary emboli missed by radiologists, although nobody has yet taken the bold step of having the computers completely supplant the humans, partly because there are armadas of malpractice attorneys waiting to pounce, and partly because, at least for now, the combination of human and machine seems to perform better than either alone.
But over the long haul, I wouldn’t bet on the humans here, particularly since one of the hottest areas in artificial intelligence research is “deep learning”—research that has created computers that are reasonably skilled at “reading,” “hearing,” and, yes, “seeing.” The same kind of software that now allows Facebook to guess that a certain collection of pixels is a picture of you, or that alerts the casino’s security guards to keep an eye on that guy, is likely to eventually crack the code in radiology, and in similar areas such as dermatology and pathology.
Slowly, radiologists are waking up to their peril. Rather than isolating themselves from clinical care, some are now relocating their reading stations in clinical areas, such as the ER and the ICU, to be in the line of sight of their clinician colleagues. Others are resurrecting interdisciplinary conferences and training their staff in customer service. Technological solutions that allow radiologists and frontline clinicians to communicate through PACS and the electronic health record are springing up (through programs that create a mash-up of a Skype-like communication tool and a John Madden–style telestrator).
Said Paul Chang, the University of Chicago radiologist whose advocacy of PACS so upset his father, “We have to go beyond isolating ourselves and concentrating on messages in a bottle, where we just write a report and are done with it, but instead fostering collaboration.”
Since radiology represents such a large segment of health care expenditures (and also a source of harm through the risk of radiation), some health care systems are even putting radiologists in the position of being gatekeepers to their own technology. It’s a role they might not have accepted in the past, but, faced with an existential threat, many now welcome it. Said David Levin, a radiologist at Thomas Jefferson University, “We have to act more like consulting physicians . . . to look at the appropriateness of the requests for advanced imaging studies . . . rather than just going ahead and doing the study.”
Radiology’s experience over the past 15 years offers a crystal ball for the rest of the health care system. The speed with which computerization unleashed a series of forces that completely transformed an established field would be all too familiar to travel agents, journalists, and others who have been run over by the digital bulldozer, but it has shocked many health care observers, even astute ones.
Will the computerization of the rest of medicine similarly upend the lives of other kinds of doctors, as well as their patients? The early returns are in, and the answer is yes.
Get the best of KQED’s science coverage in your inbox weekly.