Waiting seven days or more to get a report of your recent CT scan? Welcome to today’s new normal. Until a couple of years ago, the digital transformation of radiology dramatically shortened turnaround times of report delivery. The last-century process of audio dictation, manual transcription, editing, and signing final reports was replaced by direct voice recognition software. Radiologists became the transcribers and editors. Turnaround of reports from study time shortened to minutes.
Why in this day of AI have lag times in report delivery grown into the week levels even at the top academic centers? The answer lies in the first lesson of economics coined by the famed Stanford economist Thomas Sowell, namely, scarcity. “There is never enough of anything to satisfy all those who want it,” he wrote.
The unbridled demand for imaging is rapidly outpacing the supply of radiologists. In a Journal of the American College of Radiology paper by Eric Christensen, the projected imaging utilization based on past trends will be approximately 20 percent higher by 2055. I suspect that is a gross underestimate, as continuation of recent per-person utilization trends dramatically broadens this range.
The demand for imaging is influenced by a variety of factors. Population growth, particularly those over 65 where utilization of health care triples per capita, is a main one. Another is that imaging can expedite patient throughput for busy emergency room doctors, urgent care centers, and even primary docs who have 15 minutes to figure out patient disposition. Got a rattling cough or belly pain? Why auscultate when you can image. Part of that is the clinician’s fear of missing a serious condition and subsequent litigation exposure.
Real physician voices, twice a week
Join 100,000+ subscribers. Free, and one click to unsubscribe.
Next, facilities like imaging centers and hospitals are incentivized by volume as reimbursement diminishes. The U.S. imaging services market is poised for significant growth, projected to expand from $104.25 billion in 2025 to $121.32 billion by 2033. Though only approximately 10 percent of all U.S. health care services expenditures, it is real money, especially in terms of profit margins, which average 20 percent in independent imaging centers.
Patients also drive demand. Dr. Google can influence patients to pressure their physicians to order scans, and the worried well have access to out-of-pocket imaging screening centers that find incidentalomas, driving targeted imaging reimbursed by third-party payers.
The U.S. population was 334.9 million in 2023, and the U.S. Census Bureau projected it to grow 8.2 percent to 362.4 million in 2055. By age, the population under 35 years of age is projected to decline 8.4 percent by 2055, and those 35 and older is projected to grow 21.6 percent. More specifically, growth is concentrated in the 75 to 84, 85 to 94, and 95 and older age groups, which are projected to increase 51.5 percent, 149.6 percent, and 282.1 percent by 2055, respectively. This is a macro view of what will fuel imaging demand.
The rapidly growing demand is outpacing the supply of radiologists. There is a current shortage of radiologists in practices and academic departments in the U.S., as reflected by the more than 1,400 physician positions posted recently on the ACR Career Center.
The number of students matching into radiology residency positions through the National Resident Matching Program was just over 1,000 in 2010, and in 2023 there were still only 1,006 matched into diagnostic radiology.
Now that AI is here, excellent at machine learning of pattern recognition and even formulating a differential diagnosis, medical school graduates are questioning a previously highly desirable training and career in radiology. This quandary magnifies when the CEO of NYC Health and Hospitals said his network could already replace many radiologists with AI if not for government rules, like how do you license a bot to practice medicine, and who provides bots with malpractice coverage?
Imaging, along with laboratory and pathology results, is the IT of clinical medicine, crucial to optimal and timely health care management. Satisfying all those who want it will depend on the difference between want and need. That is where AI may play a bigger role than just pattern recognition of the images themselves. Otherwise, the first lesson of economics will prevail.
Michael Brant-Zawadzki is a radiologist and physician executive.
