If you’re at ASNR this year, I’m doing part of the session on Choosing & Navigating Your First Job tomorrow (Wednesday) at 1:15pm. Come say hi!
Clinicians can bill, at least to an extent, to account for complexity. When a patient walks into a clinic for an annual physical, an acute upper respiratory tract infection, or an endless litany of chronic complaints including uncontrolled hypertension, diabetes, and hypercholesterolemia, and an acute complaint, their documentation and the codes they use can differ between a brief med check and some more demanding undertaking.
Modality ≠ Complexity
In radiology, we don’t have complexity. We have modality. MRIs earn more than CTs, which earn more than ultrasounds, which earn more than radiographs. There is no distinction between a unindicated pre-operative screening CXR on a healthy adult and an ICU plain film. There is no distinction between a negative trauma pan-scan in an 18-year-old and a grossly abnormal pan-scan in a 98-year-old with ankylosing spondylitis, multiple fractures, and a few incidental cancers.
Leave aside adjusting actual reimbursement RVUs from payors and the government, which is beyond the scope of this essay and would require changes that are likely ultimately unhelpful in the sense that assigning RVUs for reimbursement is a zero-sum game: paying more for one thing will mean paying less for others. Yes, the reality is that some groups and some locations do have more complex cases than others, but capturing that in a fair way by a third party would be a substantial challenge and one with clear winners and losers. Reimbursement has never been fair: between the wide range of complexity and payor contracts, some doctors (or at least institutions) are simply paid more on a per-effort basis.
Internally, however, a group has limitless wiggle room to adjust internal accounting to reward effort and pursue fairness. Again, on the whole, in an ideal world, everyone receives a combination of easy and hard cases, and therefore everyone’s efforts will, on the whole, be comparable. In practice, this may not be the case in many contexts.
For example, a community division in a large university practice may not be reading the same kinds of cases as their counterparts working in the hospital. Some attendings work in rotations with junior residents, and some don’t. Different shifts and different silos across practices that involve different hospitals or different centers can vary widely, and even imaging centers in strip malls may draw different kinds of pathology by zip code and referral patterns. Even covering the ER may yield different sorts of cases with different issues, depending on the time of day. Deep night, high-speed MVCs at 3 am at your hospital may be different from the parade of elderly falls that come during the late morning. If all radiologists share in the differing kinds of work equally, no biggie, but especially in larger practices, that is not always the case.
Across modalities and divisions, it can be relatively straightforward to account for an internal work unit according to generalized desirability or typical time spent. A group might choose to bump the internal RVUs of radiographs and decrease them for some varieties of MRI. A group might decrease single-phase abdomen/pelvis CTs and increase chest CTs. A group might bump thyroid ultrasounds but decrease right upper quadrant ultrasounds. These sorts of customized “work units” based on average time-to-dictation are common.
But the problem of variable challenge within an exam type is thornier. Complexity varies, and preventing the peek-and-shriek cherry pick is a nontrivial task. A normal MRI of the brain for a 20-year-old with migraines is a different diagnostic challenge than a case of recurrent glioblastoma with radiation necrosis or progression.
Most of the metrics one could use to attempt this feat on a case-by-case basis are gameable and ultimately not tractable. If you use time spent to read the case, it’s very challenging to normalize across individuals with varying intrinsic speed, let alone the fact that someone can open an easy case and leave it open while dropping a deuce. I don’t think anyone wants to live in a world where Big Brother is tracking their mouse movements or other invasive surveillance. Radiologists have a hard enough time fighting the widget factory worker mentality enough as it is.
But even when everyone behaves nicely, having a system that accounts for tough cases would help with frustration, burnout, blah blah. No one likes to get bogged down in a complex case and feel behind. What constitutes a solid day’s work depends on how hard the work is.
Enter the eRVU
Here’s an example of how the scalability of AI could make an intractable problem potentially tractable: the use and application of LLMs on reports after the fact to create a complexity grid to help account for case difficulty. Such a model (or wrapper) could be trained to create a holistic score using a variety of factors like patient age, patient location, indication, diagnosis codes, ordering provider, prior exams and arduousness vs ease of comparison, number of positive or negative findings, and the ultimate diagnosis.
Now, obviously, such solutions—like all things—would be imperfect. It may even be a terrible idea.
For one, you’d have to determine how to weigh and differentiate between essential findings or extraneous details such that dumping word salad into a report does not increase the complexity score when it does not meaningfully add value. We all know that longer is not better. But I think if people are creative, viable experiments could be run to figure out what feels fair within a practice and how to drive desired behavior. It’s possible a big powerscribe dump with report data could yield a pretty robust solution that takes into account what “hard” and “easy” work looks like historically based on report content and the time it took to make it. Or maybe you need to wait for vision-language models that can actually look at pictures.
Again, a non-terrible version of such a product would be for internal workload and efficiency accounting, not for reimbursement. Think of it like the customized wRVU tables already in use but with an added layer that it would work across all exam types instead of just modality.
With an effortRVU, we could account for the relative complexity of certain kinds of cases within any modality. We could account for the relative ease of an unchanged follow-up exam for a single finding, and we could account for the very heavy lift that sometimes drives certain types of cases to be ignored on the list, like temporal bone CTs or postoperative CTs of extensive spinal fusions with hardware complications.
Providing good care for the most challenging cases should never be a punishment for good citizens.
(Yes I’m aware some institutions already use an “eRVU” for educational activities, meetings, tumor boards, etc. Accounting for non-renumerative time is also a defensible approach, but that’s not related to the challenges associated with variable case complexity itself.)
((It’s also worth noting that it’s also not hard to imagine a world where payors try to do things like this without your permission. Long term, how reimbursement changes for work in a post-AI world is anyone’s guess because all the current tools suck.))
Infighting & Fun
Any attempt to differ from the status quo, any variation of customization—whether simple wRVU tweaks or something more dramatic like this—is inevitably fraught. The more such a solution is based on messy human opinions, the more contentious the discussions would likely be. Everyone has an opinion about RVUs, and no one wants to see their efforts undervalued. Every tool is just a reusable manifestation of the opinions that go into it. For example, historically common complaints about the RVUs of interventionalists (often ignoring the critical clinical role our IR colleagues play and the physical presence it requires) are a cultural and financial problem, but probably not an AI one.
It’s worth noting that the desire to not “downgrade” work or deal with infighting is probably why many practices choose to change daily targets and bonus thresholds based on subspeciality, shift-type, etc instead of creating/adjusting work units. It’s the same idea tackled less dramatically from a greater distance.
Counting every activity (phone calls, conferences, etc) is also something that’s been deployed in some settings, but it’s easy to see how taken to extremes such efforts to reward behaviors can veer too far into counterproductive counting games and even tokenizing just being a decent person.
If there is a “right” answer, it may be specific to the company and the people in it, and adding complexity to the system has its own very real costs. Nonetheless, there is a strong argument to be made that some degree of practice effort to make sure that everyone’s work goes noticed and appreciated in a “fair” way is a step in the right direction for subspecialized practices.
Internal productivity metrics help prevent low effort output while smart worklists and other guardrails can ensure largely ethical behavior within the list. (But sure, theoretically, if you can solve case assignment, most everything else that matters should just even out in the long run.)
Ultimately, radiology is a field where, especially in large organizations, it can become easy to feel like an anonymous cog. Individualizing productivity accounting to truly recognize the hard, challenging work many radiologists do—and reward those who are willing to develop expertise and do a good job reading complicated cases—might help humanize the work.
(Or…maybe it would just be more stressful and counter-productive to get less credit for those easy palate-cleansers, I don’t actually know. I do know that this particular food-for-thought is bound to make some people very uncomfortable. You can tell me how far you think the gulf between possible and desirable is.)
I was at the ACR last weekend (giving a talk about the current radiology practice landscape) and had the chance to enjoy one of the American Board of Radiology’s periodic update sessions, this time given by Executive Director Dr. Wagner (who, for the record, is a nice guy). There were several things that I would have loved to ask (or push back on) but wasn’t able to due to time. I thought I’d share some of that here.
On Representation
Dr. Wagner said that the reason they don’t have specific stakeholder representation slots—like a member from the ACR, trainees (RFS), or young professionals (YPS)—is that they believe those people would advocate for their specific stakeholders instead of patients, who are the true mission of the ABR.
The mission, for those who don’t know:
Our mission is to certify that our diplomates demonstrate the requisite knowledge, skill, and understanding of their disciplines to the benefit of patients.
Leave aside that such stakeholder advocacy/perspective would be perfectly reasonable and not necessarily at odds with the ABR’s mission, but if you happened to read my article on the ABR’s bylaws from back in 2019, you may recall how the election of the board of governors works. From the updated November 2024 bylaws:
Section 4.3. Election of Governors. Nominees shall be solicited from the Board of Trustees and the Board of Governors, and may be solicited from any appropriate professional organization. Professional organizations shall provide such nominations in writing. An affirmative vote of at least three-fourths (3/4ths) of the entire Board of Governors shall be necessary for the election of any nominee to the Board of Governors.
I would say this process is insular and suspect, and I don’t see any plausible argument for how this method ensures a board that is more “mission-driven” than any other method. Company boards are usually elected by shareholders and not the boards themselves because external accountability is critical for governance.
Dr. Wagner admits the average age of the board is, in his words, “very old.” But since the board leadership itself selects the future board leadership without external stakeholder accountability or vote, the control and trajectory rest entirely internally. Dissident voices have no power and no recourse to engender change.
(The ABR would tell you they take oodles of feedback via various stakeholder panels and work groups. I really, really, don’t think that’s sufficient. Of course the leadership is old when the process of getting on the board has often meant paying your dues through years of volunteering and ultimately being selected by the board itself to fill its own vacancies.)
If you’ve read the bylaws of the ABR, you would also know that the fiduciary duty of the board members is to the board itself—not to the field of radiology, not to patients, and not even to its stated mission. From Section 6.1, Conflicts of Interest:
It is the policy of this Corporation that the legal duty of loyalty owed to this Corporation by an individual while serving in the capacity as a member of the Board of Governors or Board of Trustees requires the individual to act in the best interests of this Corporation. Consistent with the duty of loyalty, a person serving as a member of the Board of Governors or Board of Trustees does not serve or act as the “representative” of any other organization, and said member’s “constituency” as a member of the Board of Governors or Board of Trustees of this Corporation is solely this Corporation and is not any other organization or its members.
Arthur Jones of Proctor & Gamble once remarked that “all organizations are perfectly designed to get the results they get.” The reality is that this composition, process, and mandate create an unnecessarily insular crowd of well-meaning people designed to make exactly the kind of system we have. Until this process is deliberately opened to demand outside perspectives, they will simply not occur with enough frequency to make a difference.
On Fees
The ABR was also asked about fees, particularly the fact that the fees for initial certification hurt when you are a trainee (here and here and here are a few posts about fees).
The answer was a bit of self-love about how fees have stayed flat despite inflation (without any consideration to whether they could ever, you know, go down).
I would point out, however, that enrollment in MOC has increased over the years as new diplomates graduate while many of those retiring had been grandfathered—such that revenues from MOC have increased over time.
According to the ABR’s 2023 Form 990, there were approximately 35,200 diplomates enrolled in MOC to the tune of $340 a pop. In 2017, that number was 27,000. The increases won’t continue as grandfathered radiologists eventually all retire, but the revenues from MOC have obviously increased. In 2017, MOC revenues were $9.2 million and total revenues were $17.4 million (a previous deep dive is here if you’re curious). In 2023, it was almost $12 million on total revenues of $18.36 million (this after, to their credit, CAQ exam fees were finally slashed from their ludicrous $3k+ price tag).
This means that MOC accounts for an increasingly large amount of money and a larger fraction of revenues over time. I believe ABMS fees should be decreasing, not increasing—especially for a small group like trainees, who make less, usually do not have assets, and should not be required to pay so much for initial certification, especially as MOC for attending physicians with higher income is now essentially an involuntary permanent revenue stream. Initial certification still costs $3,200—no small sum for a bunch of multiple-choice questions written by volunteers and administered over the internet.
Any inflationary increases in expenses are presumably mostly related to the salaries of their employees. The ABR spent $1.66 million on executive salaries, $7.67 million on other salaries and wages (+$1.1 million on retirement contributions and other benefits), and $1 million on travel.
If the ABR takes its fiduciary responsibilities seriously, I think we can make that math work a little better on behalf of our youngest colleagues.
The Return of Oral Boards
There was frank acknowledgement that the Certifying Exam is duplicative and useless (my words of course, not the way they said it), with a caveat that it was a well-intentioned mistake that seemed like a good idea at the time. I think it was pretty clear from the outset that the current Certifying Exam had no value after the very similar Core Exam.
So, Orals are back (my initial thoughts here). Whatever the downsides, there is no denying that oral boards will provide more value than the current Certifying Exam.
There was lip service paid to the consideration of alternative ideas, like perhaps AI could provide value in the certification process? But this was dismissed with a handwave that “If we can’t all agree on what a good report would look like, how could we have an AI do so?”—which is bizarre in the sense that the same problem arises when assessing communication and critical thinking in the context of an oral examination.
What was not acknowledged is that a knowledge-based MCQ assessment doesn’t test skill (a stated important part of that ABR mission). A small number of contrived oral board cases unquestionably adds a different and useful kind of information, but even this is a skills assessment in only a very limited fashion.
So the critical question remains: are we sufficiently assessing skill with any of our current or proposed exams?
If we are being forward-thinking, we could just acknowledge that multiple-choice knowledge assessments are suboptimal. They have been historically necessary and remain efficient but are increasingly inappropriate in our modern connected world to provide actionable information. Knowledge is simply only part of the game.
Simulation is the only way the ABR can really fulfill its mission, and we are clearly living in a world where we could employ batch grading of dictated or typed answers, forcing residents to make findings from real cases and then provide diagnoses or differentials instead of just picking one from a list.
We could do this without AI, but yes, we certainly could employ AI to help grade answers (if not necessarily full reports). This would be a much more meaningful assessment.
The goal of the ABR exam certification process is to differentiate between competent and incompetent radiologists, and I don’t think we’re there yet. As always, I am disappointed by the lack of imagination at play. We can do better.
The current welcome bonus landscape:
- ENOS is a new healthcare panel offering instant payments, (including $5 when screened out) and a $25 sign-up bonus.
- M3 has three panels–M3 Global Research, M-Panels, and All Global Circle–and you can earn $25 for joining one panel, $40 for two, and $60 for joining all three (for the following specialties: Hematology-Oncology, Neurology, Gastroenterology, Nephrology, Cardiology, Urology, Surgery, Rheumatology, Obstetrics and Gynecology, Pulmonology, Allergy and Immunology, Family Medicine, Psychiatry, Dermatology, Ophthalmology, Endocrinology/Diabetes, and Pediatrics).
- Sermo is again (finally, after years) offering readers of this site an exclusive $10 welcome bonus.
- InCrowd is offering $10.
My complete list and more thorough descriptions can be found here.
In addition to being a way to earn extra money (and start a side business that enables you to take some business deductions or start a solo 401k), signing up through these links also helps support my writing. Thank you!
A struggling post-bankruptcy Envision was so desperate to get out of their radiology business—presumably due to the impossibility of recruiting/retaining, meeting clinical obligations, navigating the general tumult, etc.—that they have agreed to “transition” their remaining practices/contracts to RadPartners. Sounds like they already shed the salable groups/assets/non-competes, so this is most likely a final liability dump/hand-washing endeavor. It’s unclear how much money exchanged hands for the remaining 400 rads and some desperate/unhappy hospitals.
We discussed “Choosing Rocks” earlier this year, and I wanted to return to Four Thousand Weeks again to discuss distraction and control.
On the true nature of saying “no”:
Elizabeth Gilbert points out, it’s all too easy to assume that this merely entails finding the courage to decline various tedious things you never wanted to do in the first place. In fact, she explains, “it’s much harder than that. You need to learn how to start saying no to things you do want to do, with the recognition that you have only one life.”
This is the principle of the popular “One Thing” argument: with finite time and energy, we are always saying no to things inadvertently by omission. Saying yes to reasonable or even awesome things can be a mistake if it distracts from your true priorities. If you really want to coach your kid’s sports team, even the most engaging opportunities may be a no to your priority as a parent.
French philosopher Henri Bergson tunneled to the heart of Kafka’s problem in his book Time and Free Will. We invariably prefer indecision over-committing ourselves to a single path, Bergson wrote, because “the future, which we dispose of to our liking, appears to us at the same time under a multitude of forms, equally attractive and equally possible.” In other words, it’s easy for me to fantasize about, say, a life spent achieving stellar professional success, while also excelling as a parent and partner, while also dedicating myself to training for marathons or lengthy meditation retreats or volunteering in my community—because so long as I’m only fantasizing, I get to imagine all of them unfolding simultaneously and flawlessly. As soon as I start trying to live any of those lives, though, I’ll be forced to make trade-offs—to put less time than I’d like into one of those domains, so as to make space for another—and to accept that nothing I do will go perfectly anyway, with the result that my actual life will inevitably prove disappointing by comparison with the fantasy. “The idea of the future, pregnant with an infinity of possibilities, is thus more fruitful than the future itself,” Bergson wrote, “and this is why we find more charm in hope than in possession, in dreams than in reality.” Once again, the seemingly dispiriting message here is actually a liberating one. Since every real-world choice about how to live entails the loss of countless alternative ways of living, there’s no reason to procrastinate, or to resist making commitments, in the anxious hope that you might somehow be able to avoid those losses. Loss is a given. That ship has sailed—and what a relief.
The liberation is perhaps a bit more bittersweet than Burkeman suggests, but this is a fantastic paragraph.
Distraction from Without and Within
So it’s not simply that our devices distract us from more important matters. It’s that they change how we’re defining “important matters” in the first place. In the words of the philosopher Harry Frankfurt, they sabotage our capacity to “want what we want to want.”
Frankfurt also wrote the delightfully-titled short book On Bullshit.
In T. S. Eliot’s words, we are “distracted from distraction by distraction.”
It’s not usually that you’re sitting there, concentrating rapturously, when your attention is dragged away against your will. In truth, you’re eager for the slightest excuse to turn away from what you’re doing, in order to escape how disagreeable it feels to be doing it; you slide away to the Twitter pile-on or the celebrity gossip site with a feeling not of reluctance but of relief.
Ugh. How many times have people harped on the advice to change the notification settings on your phone? Yes, of course that helps to an extent. But I am perfectly capable of distracting myself thank you very much.
Mary Oliver calls this inner urge toward distraction “the intimate interrupter”—that “self within the self, that whistles and pounds upon the door panels,” promising an easier life if only you’d redirect your attention away from the meaningful but challenging task at hand, to whatever’s unfolding one browser tab away. “One of the puzzling lessons I have learned,” observes the author Gregg Krech, describing his own experience of the same urge, “is that, more often than not, I do not feel like doing most of the things that need doing. I’m not just speaking about cleaning the toilet bowl or doing my tax returns. I’m referring to those things I genuinely desire to accomplish.”
Burkeman suggests an extension of the timeless classic Pascal quote (from—get this—1654!): ‘All of humanity’s problems stem from man’s inability to sit quietly in a room alone.’
No wonder we seek out distractions online, where it feels as though no limits apply—where you can update yourself instantaneously on events taking place a continent away, present yourself however you like, and keep scrolling forever through infinite newsfeeds, drifting through “a realm in which space doesn’t matter and time spreads out into an endless present,” to quote the critic James Duesterberg. It’s true that killing time on the internet often doesn’t feel especially fun, these days. But it doesn’t need to feel fun. In order to dull the pain of finitude, it just needs to make you feel unconstrained.
The overarching point is that what we think of as “distractions” aren’t the ultimate cause of our being distracted. They’re just the places we go to seek relief from the discomfort of confronting limitation. The reason it’s hard to focus on a conversation with your spouse isn’t that you’re surreptitiously checking your phone beneath the dinner table. On the contrary, “surreptitiously checking your phone beneath the dinner table” is what you do because it’s hard to focus on the conversation.
Satisfaction = reality minus expectation:
The most effective way to sap distraction of its power is just to stop expecting things to be otherwise—to accept that this unpleasantness is simply what it feels like for finite humans to commit ourselves to the kinds of demanding and valuable tasks that force us to confront our limited control over how our lives unfold.
Control Is an Illusion
The cognitive scientist Douglas Hofstadter is famous, among other reasons, for coining “Hofstadter’s law,” which states that any task you’re planning to tackle will always take longer than you expect, “even when you take into account Hofstadter’s Law.”
So a surprisingly effective antidote to anxiety can be to simply realize that this demand for reassurance from the future is one that will definitely never be satisfied—no matter how much you plan or fret, or how much extra time you leave to get to the airport.
I remember struggling to get through Hofstadter’s Pulitzer Prize-winning Gödel, Escher, Bach: An Eternal Golden Braid in high school (my father kept telling me to read it, I wasn’t quite that big of a dork). Some of the references and predictions have aged out, but it’s still something else (really been meaning to re-read it sometime).
What we forget, or can’t bear to confront, is that, in the words of the American meditation teacher Joseph Goldstein, “a plan is just a thought.” We treat our plans as though they are a lasso, thrown from the present around the future, in order to bring it under our command. But all a plan is—all it could ever possibly be—is a present-moment statement of intent. It’s an expression of your current thoughts about how you’d ideally like to deploy your modest influence over the future. The future, of course, is under no obligation to comply.
Yet it turns out to be perilously easy to overinvest in this instrumental relationship to time—to focus exclusively on where you’re headed, at the expense of focusing on where you are—with the result that you find yourself living mentally in the future, locating the “real” value of your life at some time that you haven’t yet reached, and never will.
Always beware the arrival fallacy.
Last year, we also discussed Inescapable Finitude and the Productivity Trap.
I don’t think I was ever more uncertain about my chosen field than during the first couple of months of my R1 year. Coming off my intern year, I had gained in skill and responsibility, and I wouldn’t have been unhappy taking on a role as an internist during my PGY2 year.
I didn’t read all that much radiology material during my intern year and had no radiology electives because there was no radiology residency where I did my transitional year (an ACGME requirement). So when I began radiology at a new institution—with new people and a new hospital—it was a complete reset.
The first lecture I attended as a radiology resident was GU #3, the third part in a series of case conferences on genitourinary imaging, covering topics like intravenous pyelograms. I had absolutely no idea what was going on. That feeling—of being completely lost—defined much of my early experience in radiology. I lacked the foundation to get anything meaningful out of the lectures.
In the reading room, I spent a lot of time transcribing and editing reports—often repeating words I didn’t understand about anatomy I barely knew. We had a weekly first-year radiology resident interactive conference (a 2-hour pimp-session) based on chapters from Brant and Helms, but this meant I had to do additional reading on my own time, which didn’t always align with what I needed to learn for my rotation. The questions were always challenging and got harder until you failed. There was no escape.
Of course, in the end, it all worked out. At the time, I benefited from some slower rotations at the VA, which gave me some extra time to shore up my reading. And I kept plugging away, day after day on service, doing my best to understand what I was looking at and awkwardly dictate that into comprehensible English (hopefully quietly enough that no one could hear me).
It’s not weird to find radiology disorienting when you first start—it should be expected. The medical school process trains you for clinical medicine. Especially between third year, fourth year, and the intern year, you develop along a continuum that doesn’t naturally lead toward a career in diagnostic radiology.
Becoming a radiology resident is a step backward in personal efficacy. For someone who has done well in school, met expectations across multiple clerkships, and excelled on tests, it’s frustrating to suddenly feel useless.
Some people struggle with feeling like they’re not a “real doctor” in radiology because they are removed from direct clinical care for a large portion of their time. But that sense of detachment is even more profound when you can’t even do your job yet. You can only watch an attending highlight your entire report, delete it en bloc, and start from scratch so many times before your ego takes a hit.
Some attendings even dictate reports to you word for word as though you’re very slow, inaccurate, fleshy dictation software, and then judge your performance by how well you parrot everything back. This process can feel infantilizing.
But, as I’ve previously discussed in the craftsmanship mentality of residency training, I believe we can find satisfaction in our work by taking pride in doing it well.
Reading books is important. Doing practice cases and questions is important. Watching videos can be helpful. You absolutely must do the extra work to become proficient in radiology. You can’t just rely on the list gods to expose you to the full spectrum of pathology needed to adequately learn radiology and provide high-quality diagnostic care.
When everything feels overwhelming—the sheer volume of material, the anatomical complexity, the endless variations in pathology—the answer is to take it one scan at a time.
From the titular reference of Ann Lamott’s beloved Bird by Bird: Some Instructions on Writing and Life:
Thirty years ago my older brother, who was ten years old at the time, was trying to get a report on birds written that he’d had three months to write, which was due the next day. We were out at our family cabin in Bolinas, and he was at the kitchen table, close to tears, surrounded by binder paper and pencils and unopened books on birds, immobilized by the hugeness of the task ahead. Then my father sat down beside him, put his arm around my brother’s shoulder, and said, ‘Bird by bird, buddy. Just take it bird by bird.’”
You learn by doing. Every day is a learning experience. Every scan is a chance to learn a new anatomical structure or detail. Every pathology is an opportunity to expand your internal library of normal versus abnormal. Every case is a lesson—not just in recognizing the pathology present but also in differentiating it from other possible diagnoses. Yes, the work has to get done, but it can’t just be about getting through the work.
The key to being good at radiology—beyond hard work, attention to detail, and sustained focus—is realizing that taking it scan by scan isn’t just a temporary survival strategy for residency:
It’s the way we learn—when we’re right, actively reinforcing our knowledge, and when we’re wrong, absorbing the painful but essential lessons that only come from making mistakes over and over and over again.
I made the mistake of procrastinating on something more meaningful by reading a variety of random commenters on issues related to radiology. One type of flawed thinking stuck out: the all-or-nothing fallacy.
For example, as it pertains to artificial intelligence, the argument often goes, “AI will never replace a human in doing what I can do, and therefore I can ignore it.” Or, “I put a radiology screenshot into a general-purpose LLM and it was wrong,” or “our current commercially available pixel-based AI is wrong a lot,” and therefore, “I can ignore the entire industry indefinitely based on the current commercially available products.”
Leave aside the potentially short-sighted disregard for this growing sector because of its obvious and glaring current shortcomings. Even the current state of the art can have an impact without actually replacing a human being in a high-level, high-training, high-stakes cognitive task.
For instance, let’s say the current radiologist market is short a few thousand radiologists—roughly 10% of the workforce. Basic math says we could:
- Hire 10% more human beings to fill the gap (difficult in the short term)
- Reduce the overall workload by 10% (highly unlikely)
- Increase efficiency by 10%
The reality is, it doesn’t take that much magic to make radiologists 10–20% more efficient, even with just streamlining non-interpretive, non-pixel-based tasks. If only enterprise software just sucked less…
We don’t need to reach the point of pre-dictated draft reports for that to happen. There’s plenty of low-hanging fruit. Rapid efficiency gains can come from relatively small improvements, such as:
- Better dictation and information transfer. When dictation software is able to transcribe your verbal shorthand easily (like a good resident), radiology is a whole different world.
- Real summaries of patient histories.
- Automated contrast dose reporting in reports.
- Summaries of prior reports and follow-up issues (e.g., “no change” reports where previous findings are reframed in the customizable style and depth).
- Automated transfer of measurements from PACS into reports with series/image numbers.
- Automated pre-filling of certain report styles (e.g., ultrasound or DEXA) based on OCR of handwritten or otherwise untransferable PDFs scanned into DICOM.
These tasks, as currently performed by expensive radiologists, do not require high-level training but instead demand tedious effort. Addressing them would reduce inefficiency and alleviate a substantial contribution to the tedium and frustration of the job.
Anyone who thinks these growing capabilities—while not all here yet, nor evenly distributed as they arrive—can’t in aggregate have an impact on the job market is mistaken. And if AI isn’t implemented quickly enough to prevent the continued expansion of imaging interpretation by non-physician providers, the radiology job market will be forced to contend with a combination of both factors, potentially leading to even more drastic consequences.
When you extrapolate a line or curve based on just two data points, you have no real idea where you started, where you’re headed, or where you’re going to end up. Just because you can draw a slope doesn’t make the line of best fit meaningfully reflect reality or extrapolate to a correct conclusion.
Don’t fall prey to simple black and white thinking.
What is quality care, and how do you define it?
I suspect for most people that quality is like pornography in the classic Supreme Court sense—you know it when you see it. But quality is almost assuredly not viewed that way when zoomed out to look at care delivery in a broad, collective sense. Instead, it’s often reduced to meeting a handful of predetermined outcome or compliance metrics, like pneumonia readmission rates or those markers of a job-well-done as defined in the MIPS program.
The reality is that authoritative, top-down central planning in something as variable and complicated as healthcare is extremely challenging, even if well-intentioned. As Goodhart’s law says, “When a measure becomes a target, it ceases to be a good measure.”
In the real world, I would argue a quality radiology report is one that is accurate in its interpretation and clear in its communication. But without universal peer review, double reading, or an AI overlord monitoring everyone’s output, there is no way to actually assess real quality at scale. You can’t even tell from the report if someone is correct in what they’re saying without looking at the pictures. Even “objectively” assessing whether they are clear or helpful in just their written communication requires either a human reviewer or AI to grade language and organization by some sort of mutually agreed-on rubric. It’s simply not feasible without a significant change to how healthcare is practiced.
And so, we resort to proxy metrics—like whether the appropriate follow-up recommendation for a handful of incidental findings was made. The irony, of course, is that many of these quality metrics are a combination of consensus guidelines and healthcare gamesmanship developed by non-impartial participants with no proof they reflect or are even associated with meaningful quality at all.
We should all want the quality of radiology reporting to improve, both in accuracy and in clarity. Many of these problems have been intractable because potential solutions are not scalable with current tools and current manpower—which is why soon you’ll be hearing about AI for everything, because AI solves the scaling problem, and even imperfect tools over the coming years will rapidly eclipse our current methods like cursory peer review.
Everyone would rather have automated incidental finding tracking than what most of us are still using for MIPS compliance. Right now, it’s still easy to get dinged and lose real money because you or your colleagues omitted some BS footer bloat about the source of your follow-up recommendations for pulmonary nodules too often. Increased quality without increased effort is hard to complain about.
But even just imagine you have a cheap LLM-derived tool that catches sidedness errors (e.g. right abnormality in the findings and left in the impression) or missing clarity words like forgetting the word “No” in the impression (or, hey, even just the phrase “correlate clinically”). This already exists: it’s trivial, requires zero pixel-based AI, (and—I know—is rapidly becoming table stakes for updated dictation software), but widespread adoption would likely have a more meaningful impact on real report quality than most of the box checking we do currently. A company could easily create a wrapper for one of several current commercial products and sell it tomorrow for those of us stuck on legacy systems. It might even be purchased and run by third parties (hospitals, payors, Covera Health, whatever) to decide which groups have “better” radiologists.
But now, take it that one step further. We’ve all gotten phone calls from clinicians asking us to translate a colleague’s confusing report. Would a bad “clarity score” get some radiologists to start dictating comprehensible reports?
It’s not hard to leap from an obviously good idea (catching dictation slips) to more dramatic oversight (official grammar police).
Changes to information processing and development costs mean the gap between notion and execution is narrowing. As scalable solutions proliferate, the question then becomes: who will be the radiology quality police, and who is going to pay for it?
From the excellent Alchemy: The Dark Art and Curious Science of Creating Magic in Brands, Business, and Life by Rory Sutherland:
In theory, you can’t be too logical, but in practice, you can. Yet we never seem to believe that it is possible for logical solutions to fail. After all, if it makes sense, how can it possibly be wrong?
[…]
If you are a technocrat, you’ll generally have achieved your status by explaining things in reverse; the plausible post-rationalisation is the stock-in-trade of the commentariat. Unfortunately, it is difficult for such people to avoid the trap of assuming that the same skills that can explain the past can be used to predict the future.
The world trades in stories, but compelling stories aren’t necessarily true. See also: hindsight bias. The post hoc seeming inevitability of where we are now is a mirage.
Adam Smith, the father of economics – but also, in a way, the father of behavioural economics – clearly spotted this fallacy over two centuries ago. He warned against the ‘man of system’, who: ‘is apt to be very wise in his own conceit; and is often so enamoured with the supposed beauty of his own ideal plan of government, that he cannot suffer the smallest deviation from any part of it. He goes on to establish it completely and in all its parts, without any regard either to the great interests, or to the strong prejudices which may oppose it . . . He seems to imagine that he can arrange the different members of a great society with as much ease as the hand arranges the different pieces upon a chess-board. He does not consider that the pieces upon the chess-board have no other principle of motion besides that which the hand impresses upon them; but that, in the great chess-board of human society, every single piece has a principle of motion of its own, altogether different from that which the legislature might chuse [sic] to impress upon it. If those two principles coincide and act in the same direction, the game of human society will go on easily and harmoniously, and is very likely to be happy and successful. If they are opposite or different, the game will go on miserably, and the society must be at all times in the highest degree of disorder.’
This chess piece argument is also a metaphor favored by conservative economist Thomas Sowell as the key failing conceit of central planning.
The problem that bedevils organisations once they reach a certain size is that narrow, conventional logic is the natural mode of thinking for the risk-averse bureaucrat or executive. There is a simple reason for this: you can never be fired for being logical. If your reasoning is sound and unimaginative, even if you fail, it is unlikely you will attract much blame. It is much easier to be fired for being illogical than it is for being unimaginative.
[…]
The fatal issue is that logic always gets you to exactly the same place as your competitors.
The primary function of managers is to preserve their position within management. The second function is to be promoted. The distant third is to actually manage people well or improve their organizations. (Further Reading: Academic Medicine and the Peter Principle)
The late David Ogilvy, one of the greats of the American advertising industry and the founder of the company I work for, apparently once said, ‘The trouble with market research is that people don’t think what they feel, they don’t say what they think, and they don’t do what they say.’*
[…]
It is fine to provide up-to-date magazines in reception to show that you care, but when the urge to show commitment to patients involves performing unnecessary tests and invasive surgery, it probably needs to be reined back.
Yes. I am reminded of Patient Satisfaction: A Danger to be Avoided.
If you want to change people’s behaviour, listening to their rational explanation for their behaviour may be misleading, because it isn’t ‘the real why’. This means that attempting to change behaviour through rational argument may be ineffective, and even counterproductive. There are many spheres of human action in which reason plays a very small part. Understanding the unconscious obstacle to a new behaviour and then removing it, or else creating a new context for a decision, will generally work much more effectively.
Behavior change is hard. I can barely control a whole host of my own impulses, let alone guide others.
The self-regarding delusions of people in high-status professions lie behind much of this denial of unconscious motivation. Would you prefer to think of yourself as a medical scientist pushing the frontiers of human knowledge, or as a kind of modern-day fortune teller, doling out soothing remedies to worried patients? A modern doctor is both of these things, though is probably employed more for the latter than the former. Even if no one – patient or doctor – wants to believe this, it will be hard to understand and improve the provision of medical care unless we sometimes acknowledge it.
[…]
To put it crudely, when you multiply bullshit with bullshit, you don’t get a bit more bullshit – you get bullshit squared.
[…]
Nassim Nicholas Taleb applies this rule to choosing a doctor: you don’t want the smooth, silver-haired patrician who looks straight out of central casting – you want his slightly overweight, less patrician but equally senior colleague in the ill-fitting suit. The former has become successful partly as a result of his appearance, the latter despite it.
The Taleb reference is commonly referenced even if it may not always work in real life. It’s one of those lightbulb-generating remarks that strikes that magic of being surprisingly intuitive after seeming counterintuitive.
There is an important corollary: proxy metrics (patrician manner, various training factors) don’t actually mean what we want them to mean. You want to see a Harvard-trained doctor because you assume they are better through the implied meritocratic scarcity of an elite institution and the presumption that therefore, surrounded by other geniuses and some presumably fancy digs, their training was uniquely better—but there is no actual basis for this belief.
In making decisions, we should at times be wary of paying too much attention to numerical metrics. When buying a house, numbers (such as number of rooms, floor space or journey time to work) are easy to compare, and tend to monopolise our attention. Architectural quality does not have a numerical score, and tends to sink lower in our priorities as a result, but there is no reason to assume that something is more important just because it is numerically expressible.
Measurability does not equal importance. See “Overweighing what can be counted” in Munger’s Incorrect Approaches to Medicine.
The more data you have, the easier it is to find support for some spurious, self-serving narrative. The profusion of data in future will not settle arguments: it will make them worse.
…I naively thought Covid would bring people together. And it did, for maybe a week or two. Then the dueling data wars began.
We are flush with data. Absolutely awash in data. If the past few years of social media have taught us anything, it’s that information isn’t truth. It’s raw material for storytelling. Yoval Noah Harari does a nice if depressing job discussing information networks in his recent book, Nexus.