What It Takes to Retain Radiologists: Insights from Dr. Elizabeth Bergey
Radiology leaders are navigating a moment where their enterprise and its patients are threatened by radiologist burnout, retention, and increasing operational complexity, while rapid technological advancements offer ever-evolving solutions that require careful assessment.
In this Q&A, Elizabeth Bergey, MD, Chief Medical Information Officer at Rad AI, shares her perspective on what it takes to retain top performers, how she evaluates whether operational changes are truly helping practices and patients, and how to decide whether to build or buy new technology.
Radiologist burnout and turnover are huge issues. What unconventional retention or incentive strategies have you tried that worked (or didn’t) in keeping top performers?
Top producers now demand financial recognition for their contributions, especially in environments where they feel they’re carrying more than their share of the workload, and frankly, sometimes they are. They’re acutely aware that there are plenty of “eat-what-you-kill” jobs out there where faster radiologists proportionally earn more and that reality shapes expectations. Further, some radiologists disproportionately take on non-revenue producing activities, including conferences, training, and even speaking to technologists, patients, and clinicians. They must be properly credited for that time, or bad things happen to the enterprise–no one picks up the phone. If you want to retain top talent, recognition has to be tangible.
That recognition starts with developing a fair way to measure work performed, both revenue-generating and non-revenue generating. Often, those metrics start with a modification of the CMS wRVU. There are some procedure codes that offer disproportionate credit payment for the time it takes to read the exam. PET/CT procedure codes are, for example, grossly undercredited. These time-based adjustments to the RVU to improve concordance are frequently part of the management answer.
To fairly compensate their radiologist, practices must also recognize the fact that not all roles and not all people contribute to the very high percentage of non-compensated work required of radiologists. This could mean adjusting RVU targets based on a radiologist’s daily assignment, since some assignments require more time away from reading to manage technologists, patients, and clinicians, or offering a time-adjusted RVU equivalent for each non-revenue-producing minute worked.
Finally, financial recognition can well include per-click bonuses, either on top of their usual hourly pay, applied only beyond a threshold with caps on daytime work, or targeted to off-hours using modified RVUs.
With so many technologies and metrics now available, what’s your own ‘north star’ measure of whether a change is really helping both the practice and patients?
My north star is simple: anything that keeps radiologists in your market working optimally to serve your patients helps both radiologists and patients. If the radiologist workforce is sustainable, that’s a strong signal you’re doing something right.
To understand whether that’s actually happening, I look at a few key measures. These include total modified RVUs per unit time, further broken down by radiologist and by assignment, which helps me understand workload and efficiency. But just as important is retention data, especially when compared to national trends. If productivity looks good but retention is lagging, then the changes aren’t really helping.
How do you weigh the decision to buy a vendor AI solution versus building something with your own team or an academic partner? Are there clear signals that favor one approach over another?
I actually did this! Our complex group – multiple hospitals, PACS, RIS, and EMRs – was in the market for a workflow orchestrator. We shopped around at RSNA, but every option we found was essentially a highly filtered conventional worklist. The "buy" option simply didn’t meet our needs, so we did a “make analysis” to estimate the cost of building a novel product.
We needed a dynamic solution that could adapt to the craziness of everyday radiology: knowing which emergency departments were busy, and which doctors were unavailable. We built it with a very important goal of keeping radiologists in their flow state. After running the numbers, we realized that building our own tool would deliver ROI by keeping radiologists away from clerical and repetitive case management duties. It returned its investment in under two years. Part of that ROI was related to the novel compensation programs we developed and the orchestrator administered.
The benefits went beyond efficiency. Our custom solution delivered uniquely robust analytics on performance and margins across sites and procedures, insights we could never have gained from an off-the-shelf product. I knew which of our procedures, sites, and assignments were profitable. We based many decisions, including contract negotiations, on insights gained with this novel data.
Continuing the Conversation
Dr. Bergey’s insights add another important dimension to our CMIO Q&A series, reinforcing that sustainable progress in radiology depends on fairness, transparency, and long-term thinking.
To see how these conversations build on one another, revisit our previous installment, AI That Works for Radiologists: Dr. Andrew Del Gaizo’s Perspectives, where Dr. Andrew Del Gaizo explores the distinction between operational and diagnostic AI, strategies for avoiding automation bias, and what AI success could look like five years from now.
