In a study published in the February 2018 issue of Academic Radiology, Dr. Andrew Rosenkrantz of NYU Langone Medical Center proposes that measuring physician outcomes and resource utilization requires appropriate patient risk adjustment. Using publicly available data from CMS, he and his fellow researchers compared one year of risk-adjustment scores among fifty-four physician specialties to determine the relative complexity of their attributed patient populations. The team analyzed the results for radiologists (31,175 of 549,194 total) based on a range of practice characteristics: teaching affiliations, practice size, geography, and subspecialty.
Using the Medicare Hierarchal Condition Code (HCC) risk-adjustment factor (RAF) weighting system, interventional radiology ranked 4th (2.60 ± 1.29), nuclear medicine ranked 16th (1.87 ± 0.45), and diagnostic radiology ranked 21st (1.75 ± 0.61) in relative patient complexity. Risk scores were higher for radiologists with teaching affiliations, larger practice size, urban, and subspecialized practices. Teaching affiliation was the strongest independent predictor of patient complexity for both interventional and noninterventional radiologists.
The researchers concluded that radiologists on average serve more clinically complex Medicare patients than most physicians nationally. When interviewed by AuntMinnie.com at the time of the study’s publication, Dr. Rosenkrantz identified risk-adjustment as an under-investigated aspect of practice that will demand greater attention as value-based payment contracts proliferate. Judging by the widespread interest in this study, it would seem that many others agree.
In the JACR Radiology Firing Line podcast entitled “Do Radiologists Care for Sicker Patients?”, Dr. Saurabh Jha and Danny R. Hughes, PhD, of the Neiman Health Policy Institute review the results of this study as a point of departure to examine how risk adjustment is designed, how physicians understand patient complexity, and how to integrate relative risk into payment models that ensure a fair deal for those physicians who must treat the sickest patients.
To establish context, Dr. Hughes notes that the HCC methodology is one of several patient complexity algorithms that are currently utilized. He distinguishes between prospective risk models, such as HCC, that use diagnoses collected in a base year to predict expenditures in the following year, and retrospective (“concurrent”, or “contemporaneous”) models that have no predictive function but may be preferable for comparative reporting of performance measurement results.
Risk adjustment scores are not used for individual patient treatment decision making, only for population health management. Dr. Hughes argues that there is imprecision in prospective risk adjustment. For example, risk scores may be based on patients’ similar ICD and CPT code information, but not on specific clinical details contained in the EHR. Additionally, historical claims data is by its definition not up-to-date.
Dr. Jha points out that physicians have an intrinsic sense of patient severity by the comorbidities they present. He also comments that the most resource consuming patients are those with lower mortality rates who may have multiple post-operative complications and who do not receive follow-up care coordination, and that outcomes are related to comorbidity. He wonders if this incentivizes physicians to take on lower-risk patients (“cherry picking”), and suggests making risk scores a quality metric, i.e. a reward for risk assumption.
Dr. Jha notes that the HCC score for diagnostic radiology is not much higher than average, with a mean score of 1.79 and median score of 1.69. This presents as a tail in a distribution curve (see fig. 1 in paper) meaning that most are seeing an average level of comorbity, but some radiologists are seeing much sicker patients. He asks, how do we ensure that clinicians are getting paid fairly for the complexity they are incurring?
This raises the question: what defines fair payment? Dr. Hughes argues that, despite the perceived primacy of clinical decision making, it is policy making – governmental and organizational – that determines most payment decisions. Both commenters suggest that physician revenue should include compensation for investment in human capital plus the demands of working with resource intensive patient populations. Dr. Hughes concludes by introducing Social Determinants of Health (SDOH) as a critical yet uncalculated factor in most risk adjustment methodologies.
Risk adjustment will be featured in all value-based payment contracts, regardless of payer. The HCC system is currently being utilized by CMS in Medicare Advantage plans, Alternative Payment Models (APMs), and MIPS, and will factor into future iterations of episode-based Cost measures.
As this worthwhile debate on the future of physician compensation models continues, radiology practices, like all providers, are challenged to deal with the realities of the present. That means keeping MIPS participation squarely in focus for now, considering if APMs become a viable option in the near future, and working with business partners that can support your staff and compliance needs. Subscribe to our blog now for more on this topic and expert advice on a variety radiology practice revenue cycle management issues.
Richard Morris is the Director of Value-Bases Strategy at Healthcare Administrative Partners.
Recent Articles:
2018 Budget Bill Makes Changes to MIPS
How Removing IVC Filters Benefits Patients and Radiology Practices Alike
How Radiology Practices Can Drive True Quality of Care
