New system improves on CMS subspecialty classification, carries potential in transition to performance-based payments

A new system accurately identifying the subspecialties of practicing radiologists using Medicare data represents a considerable improvement over the system in place, which only differentiates between diagnostic radiologists, nuclear medicine physicians and interventional radiologists.

These more specific classifications could have a large impact; many current Medicare metrics reward treatments rather than diagnosis, putting radiologists whose work is centered on diagnosis at a disadvantage. Additionally, CMS continues to implement additional performance-based reimbursement methods, emphasizing the importance of metrics that accurately measure a physician’s success.

Details on this new system and a study measuring its effectiveness were published in the American Journal of Roentgenology.

“Emerging payment models are designed to more tightly link payment to quality metrics, thereby rewarding physicians who demonstrate the greatest value for their patients,” lead author Andrew Rosenkrantz, MD, affiliate research fellow at the Harvey L. Neiman Health Policy Institute, said in a press release. “Metrics developed by the Centers for Medicare and Medicaid Services (CMS) have been the subject of considerable criticism, particularly as they apply to radiologists.”

For example, neuropathologists, cardiothoracic radiologists and breast imagers are all under the “diagnostic radiologist” umbrella, even as optimal performance metrics for these subspecialties would look much different.

To mitigate these negative effects, researchers from the Neiman Institute developed a system called Neiman Imaging Types of Service (NITOS). The NITOS ties all Health Common Procedure Coding System codes to a modality, body region and, when appropriate, one of seven subspecialties. They assigned a subspecialty to a radiologist when more than half of their work Relative Value Units (RVUs) came from a single specialty.

“Using Medicare public use files, we identified 33,118 self-designated radiologists, and through a manual and laborious search, we further identified 1,860 of those working at the top 20 National Institutes of Health-funded academic radiology departments across the country,” said Richard Duszak, MD, at Emory University and affiliate senior research fellow at the Neiman Institute. “Medicare claims for those radiologists were NITOS-mapped by subspecialty, and that mapping was compared to their departmental website self-designated subspecialty area.”

Using this system, they correctly identified the radiologist’s subspecialty almost 90 percent of the time, with only 4.2 percent being incorrectly classified—meaning they exceeded the 50 percent threshold, but for a different specialty than what was listed on their departmental website. While not perfect, it vastly outperforms the system used by CMS in two ways, according to the authors.

“The CMS system correctly identified fewer than 50 percent of interventional radiologists and nuclear medicine physicians (which is much lower than the overall rate of correct subspecialty identification noted by our system), and the CMS system does not further subclassify the remaining diagnostic radiologists,” they wrote. “We are unaware of any other system currently in place that a payer could use to reasonably and reliably identify the subspecialty of radiologists in an objective and transparent manner.”

This system has real-world policy implications as the industry increases the amount of reimbursement dollars tied to performance metrics. CMS will begin to use metrics down the subspecialty level over the coming years, and they need a reliable method to sift through thousands of subspecialists.

“For example, it makes little sense for dedicated neuroradiologists to be evaluated on the basis of their barium enema fluoroscopy times or rates of recommendation for additional imaging for incidental findings in the liver or kidney,” the authors wrote. “If this concept could be validated more broadly, it would permit the development of subspecialty-focused quality metrics, thus broadening opportunities for radiologist subspecialty work to be appropriately acknowledged in future payment models,” added Duszak.