\nthat artificial intelligence has already made it unnecessary to hire more radiologists in the United
\nStates. The team is led by Elizabeth Rula, executive director of the Harvey L. Neiman Health Policy
\nInstitute (HPI) in Reston, Virginia, and responds to comments published in 2026 in JAMA Health
\nForum<\/em> by Harvard economist David Cutler and Lev Klarnet.<\/p>\n
Cutler and Klarnet argued that the decrease in medical imaging had “allowed the U.S. to meet the The central problem Rula and colleagues identify is the empirical base. The key study the The second problem is the metric. Counting exams ignores actual radiologist workload, which The rebuttal rests on peer-reviewed work from the institute itself. In 2025, the group projected Another HPI study measured workload in relative value units (RVUs), which weight the time, effort Cutler and Klarnet also argued that electronic health records have reduced repeat imaging. HPI The editorial spends time on historical context, and rightly so. In 2016, Nobel laureate Geoffrey Rula and colleagues note that the concern “remains newsworthy and radiologists remain defensive, The dispute concerns U.S. data, but the methodological lesson travels. Any discussion of whether There is also the effect on people already in the profession. Turnover research consistently Scepticism is warranted in both directions. HPI is an institute affiliated with the American The picture the evidence supports is neither replacement nor automatic shortage. It is a system
\nneed for imaging without an increase in radiologists.” The HPI answer is blunt: “The assertion, we
\nthink, is counter to recent collective experience of the radiology community and ignores a decade of
\nevidence.”<\/p>\n![\"Horizontal](\"https:\/\/rtmedical.com.br\/wp-content\/uploads\/2026\/10\/radiology-workload-vs-reimbursement-hpi.png\")
Where the reading of the data goes wrong<\/h2>\n
\neconomists cite examined per capita imaging use through 2016 and attributed part of the apparent
\ndecline to Medicare code bundling \u2014 a billing change, not a change in exam volume. When a payer
\nmerges two codes into one, the count of “procedures” falls in the administrative record without a
\nsingle exam going undone.<\/p>\n
\ndepends on volume and complexity together. A whole-body CT with multiplanar reconstructions and
\ncomparison against three priors is not equivalent to a chest radiograph, yet both register as “one
\nexam.”<\/p>\nThe numbers HPI put on the table<\/h2>\n
\n17% to 27% growth in U.S. medical imaging use through 2055, driven by population growth and ageing.
\nMedicare data compiled in the Neiman Almanac shows meaningful Part B volume increases in CT and MRI
\nbetween 2016 and 2024 \u2014 precisely the period after the series the economists relied on ends.<\/p>\n
\nand risk of each procedure. Imaging RVUs per beneficiary reached their highest level in 2021, up 13%
\nfrom 2005. Over the same span, Medicare’s budget neutrality mandate cut inflation-adjusted
\nradiologist reimbursement by 48.3% between 2005 and 2025. More work, less payment per unit of
\nwork.<\/p>\nThe electronic record argument does not hold either<\/h2>\n
\ncounters with a recent analysis finding that 12.5% of Medicare office-based imaging studies were
\nrepeated between 2013 and 2022. That is not a marginal detail: duplicate imaging is a leading
\nsource of waste and of avoidable radiation exposure, and it persists even where interoperability is
\nformally available \u2014 a process failure that echoes what we have written about
\ntechnology governance in radiology<\/a>.<\/p>\nThe ghost of 2016 is still in the room<\/h2>\n
\nHinton said deep learning would outperform radiologists at image analysis and that training more of
\nthem made no sense. The prediction did not hold, and Hinton himself has walked it back: in a
\nNew York Times<\/em> interview in September 2026, he said a combination of AI and radiologists
\nwill make radiologists more efficient while improving accuracy.<\/p>\n
\nindicating the longstanding ripple effect of Hinton’s prediction.” The practical argument is that
\nbroad declarations from publicly authoritative figures have real consequences for specialty choice,
\nresidency sizing and investment decisions \u2014 and take years for evidence to correct.<\/p>\nWhat it means outside the United States<\/h2>\n
\n“AI will solve the radiologist shortage” has to separate three things that routinely get conflated:
\nexam volume, complexity per exam and available reading capacity. In health systems where specialist
\ndistribution is heavily concentrated in large cities and teleradiology absorbs part of the demand
\nfrom smaller regions, the risk of reading an aggregate indicator as proof of sufficiency is higher
\nstill \u2014 exactly the point we raised when we asked
\nwhether the radiologist shortage is regional<\/a>.<\/p>\n
\nshows perceived workload predicting exits, and public narratives of obsolescence help neither
\nretention nor recruitment \u2014 a theme we covered in
\nwhy radiologists quit<\/a>.<\/p>\nHow to read this without picking a side by reflex<\/h2>\n
\nCollege of Radiology, and defending the economic relevance of the specialty sits inside its
\ninstitutional remit; that does not invalidate the data it cites, but it does ask for attention to
\nwhich metrics were selected. On the other side, the criticised comments were published as opinion,
\nnot as original research. What the rebuttal demonstrates solidly is narrower and more useful than
\nthe polarised debate suggests: the data series used to argue that demand fell ends in 2016, is
\nconfounded by a coding change, and does not measure complexity.<\/p>\n
\nwhere demand grows, complexity per exam grows faster, payment per unit falls, and AI arrives as a
\nproductivity tool whose gains are still being measured case by case \u2014 as the
\nFDA radiology AI authorisation figures<\/a>
\nillustrate.<\/p>\n