What the Medicare data shows
Researchers from the University of Virginia and the Harvey L. Neiman Health Policy Institute ran a retrospective, repeated cross-sectional analysis of a 5% sample of fee-for-service Medicare claims from 2008 through 2023 — 476,688 radiologist-years covering 46,533 unique radiologists. The study, led by Luke Wilkins, MD, with senior author Eric Christensen, PhD, was published in the Journal of Vascular and Interventional Radiology in June 2026.
The headline trend runs in two directions at once. Fewer radiologists touch IR work at all — the share doing any interventional radiology fell from 66.9% in 2008 to 49.6% in 2023 (equivalently, radiologists doing zero IR work rose from 33% to 50%). But among those who still do IR work, they're doing more of it: the share with a majority of their work in IR climbed from 12.6% to 18.5%. And across all radiologists — not just those who practice IR — the share with a super-majority workload (more than 90% IR) more than doubled, from 4.1% to 8.8%.
In short: interventional radiology isn't disappearing or shrinking as a specialty — it's concentrating. A smaller, more committed subset of radiologists is doing a larger share of the procedural work, as Radiology Business and the AuntMinnie coverage of the study both note.
Why this is happening — and how it varies by age
The study's authors point to a structural cause: the 2016 introduction of integrated IR/diagnostic-radiology residency pathways, which let trainees commit to an interventional-heavy career earlier instead of building a general diagnostic practice first and specializing later. The age breakdown in the 2023 data is consistent with that: among radiologists doing any IR work that year, 24.7% of those aged 25–34 had a majority-IR practice and 9.2% a super-majority — versus 11.9% and 6.8% respectively for radiologists 65 and older.
That age gap is a snapshot of where each cohort stands today, not a sign that the shift is concentrated in younger radiologists. The study's senior author, Eric Christensen, PhD, told the Neiman Health Policy Institute that the rate of change toward greater IR concentration over the study period was not greater among younger radiologists — if anything, the increase in super-majority IR work was steeper among older radiologists. The residency pathway helps explain today's age pattern; it isn't the whole story of how the trend built over 15 years.
A supply-side squeeze distinct from the "radiologist shortage" headline
Most workforce coverage in radiology focuses on headcount: how many radiologists are being trained, how many residency spots go unfilled, how many jobs sit open. That framing misses a second, quieter mechanism this data isolates — reallocation within an existing headcount.
Even in a year where the total number of practicing radiologists doesn't change, a rising super-majority-IR cohort means a shrinking share of that same pool is available to read diagnostic CTs, MRIs, and X-rays. A radiologist who moves from a 20%-IR practice to a 90%-IR practice hasn't left the profession and won't show up in an attrition statistic — but for diagnostic reporting purposes, they're functionally gone. Scaled across tens of thousands of radiologists over 15 years, that's a meaningful and largely invisible drain on the pool responsible for the reporting queue that most imaging volume still runs through.
The trend, in numbers
| Metric | 2008 | 2023 |
|---|---|---|
| Radiologists doing any IR work | 66.9% | 49.6% |
| Radiologists doing zero IR work | 33% | 50% |
| Majority IR work (>50%), among IR practitioners | 12.6% | 18.5% |
| Super-majority IR work (>90%), of all radiologists | 4.1% | 8.8% |
Source: Wilkins et al., "Trends in the Concentration of Interventional Radiology Work among Radiologists in the United States: Analysis of Medicare Claims Data, 2008–2023," Journal of Vascular and Interventional Radiology (2026).
Where this leaves imaging providers
This trend doesn't operate in isolation. It compounds existing pressure points already documented elsewhere in radiology's workforce data — from the technologist-side staffing gaps that feed the same reporting bottleneck to the uneven geographic distribution covered in analyses of where radiologists choose to practice. None of these forces show up as a single dramatic headline. Each one quietly narrows the number of people available to close out a diagnostic report on any given day.
For imaging centers, hospital departments, and teleradiology providers, the practical takeaway is that diagnostic reporting capacity should be planned for as its own constraint — separate from overall radiologist headcount and separate from equipment or scan-volume capacity. A department can add CT scanners and still bottleneck on reads, because the radiologists behind the reads are increasingly split across two different jobs.
Where AI-drafted reporting fits
Since the constraint this data points to is reading capacity — not scanner throughput or patient volume — the tools that help most are the ones that extend what a fixed diagnostic workforce can get through, rather than tools aimed at acquiring or triaging more images. That's the premise behind AI-drafted CT reporting: the AI produces a structured draft, xAID's in-house radiologist reviews every preliminary, and the report reaches the client ready-to-sign, so the reading radiologist's time goes toward review and judgment rather than drafting every report from a blank page. As more radiologists shift a larger share of their practice into IR, that kind of leverage on the remaining diagnostic pool becomes less of a convenience and more of an operational necessity.
Frequently asked questions
What does the new Medicare data show about interventional radiology workforce trends?
A 2008–2023 analysis of Medicare claims covering 46,533 radiologists and 476,688 radiologist-years, published in the Journal of Vascular and Interventional Radiology, found that the share of radiologists doing any interventional radiology (IR) work fell from 66.9% to 49.6%. Among those who do practice IR, the share concentrating a majority of their work in it rose from 12.6% to 18.5%. And across all radiologists, the share with a super-majority (more than 90% IR) more than doubled, from 4.1% to 8.8%.
Why are more radiologists concentrating on interventional work instead of diagnostic reading?
The study's authors point to the 2016 introduction of integrated IR/diagnostic-radiology residency pathways, which let trainees commit to interventional practice earlier in their careers. Younger radiologists reflect this: among 25–34-year-olds doing IR work in 2023, 24.7% had a majority-IR practice and 9.2% a super-majority, compared with 11.9% and 6.8% for radiologists 65 and older.
How does this compound the radiologist shortage for diagnostic CT reporting?
It is a separate mechanism from headcount attrition. Even if the total number of practicing radiologists holds steady, every radiologist who shifts a larger share of their time toward procedural IR work is, in effect, a partial subtraction from the pool reading diagnostic CTs — because a workday spent in the IR suite is a workday not spent on the reporting queue. The Medicare data shows that pattern continuing, not reversing, across 15 years of claims.
How can imaging providers offset a shrinking diagnostic reading pool?
Since the constraint is reading capacity rather than image acquisition capacity, tools that extend the output of remaining diagnostic radiologists — rather than tools that acquire or triage more images — address it most directly. AI-drafted, ready-to-sign CT reports are one example: the radiologist's role shifts from producing every report from scratch to reviewing and signing a structured draft, which changes how many studies a fixed diagnostic workforce can move through the queue.
Source: Wilkins LR, Drake AR, Rula EY, Duszak R Jr, Lamparello NA, Khaja M, Christensen EW, "Trends in the Concentration of Interventional Radiology Work among Radiologists in the United States: Analysis of Medicare Claims Data, 2008–2023," Journal of Vascular and Interventional Radiology (2026), doi.org/10.1016/j.jvir.2026.108913, as reported by the Harvey L. Neiman Health Policy Institute, Radiology Business, AuntMinnie, and Imaging Technology News. Figures are rounded as reported.