In a recent study by the Harvey L. Neiman Health Policy Institute (HPI), backed by the Society of Interventional Radiology, it was found that 76% of interventional radiologists (IRs) were wrongly labeled as diagnostic radiologists in Medicare data. This highlights the difficulty in identifying interventional radiologists in healthcare claims data, crucial for population health and research.

“Without a method that provides more complete and accurate inclusion of IRs in research, studies may produce non-generalizable or biased results and conclusions,” says author C. Matthew Hawkins, MD, a professor of radiology and imaging sciences at Emory University School of Medicine in Atlanta. “This gap hinders researchers’ ability to study the value of interventional radiology in healthcare at a population scale.”

In the study published in the Journal of Vascular and Interventional Radiology, the HPI and IR research team developed and tested a method to identify IRs in claims data, bypassing the limitations of the specialty variable. They identified all diagnostic radiologists and IRs, evaluated the percentage of clinical work requiring interventional radiology fellowship training for each radiologist, and compared their findings with sources like SIR membership data to ensure accurate IR identification without misclassifying diagnostic radiologists.

A statistical validated threshold to maximize IR identification was 10% or more IR work; higher thresholds were also evaluated as options for selecting IRs for research. “When we grouped IRs by their percentage of IR work, we found that a majority of IRs were at each end of the spectrum,” says Elizabeth Y. Rula, PhD, Executive Director, Harvey L. Neiman Health Policy Institute.  

Approximately one-third of IRs performed 10%-20% of their work in IR, and a similar proportion performed 90% or more of their work in IR.  The other one-third was distributed across the entire 20%-90% range of IR work.  At all thresholds tested in the study, IRs were more likely to be male, metropolitan, and earlier in their careers than other radiologists. 

“At a 10% threshold we identified 5,453 unique IRs in 2019, 3.6-times more IRs than would have been identified using the specialty information in the Medicare data,” says lead author Mikki Waid, PhD, a senior research fellow at the Neiman HPI. “When we tested higher threshold percentages of IR work, 50% and 90%, each resulted in the identification of fewer IRs [2,494 and 1,239, respectively], a higher proportion of which were labeled as an IR in the data, compared to the 10% threshold.  Our study provides information that will aid selection of the appropriate threshold for a specific research objective.” 

The primary results from 100% Medicare data, with 32,131 radiologists, were verified using a large national commercial and Medicare Advantage claims database that included 26,270 radiologists. “Both datasets produced similar results, further validating our findings,” according to Waid.

The researchers also conducted a deeper evaluation of the types of clinical work IRs perform.  In the Medicare data, IRs dedicate more time to procedural work, whereas in the commercial and Medicare Advantage dataset, a higher proportion of effort is dedicated to evaluation and management.  

“The smaller proportion of evaluation and management work billed to Medicare Part B versus private insurers may reflect more complex requirements by Medicare for evaluation and management reimbursement, rather than actual differences in the care delivered to Medicare patients,” says author Raymond Liu, MD, vice president at Mass General Brigham Global Advisory. “Medicare has made efforts to simplify these requirements, so future research may find greater parity for IRs between payers.”

“The HPI and SIR joined forces for this research because of the need for methods that improve the generalizability and comparability across claims-based IR studies, and it is our hope that this work leads to growth in research investigating the value of this field of medicine to patients and populations,” adds author Laura Findeiss, MD.