Financial Incentives and the Art of Payment Reform

During the past 2 decades, physicians have expanded the scope of care provided in their offices to encompass a variety of services including advanced imaging that were traditionally performed in hospital-based settings. In this issue of JAMA, Shah and colleagues¹ describe a well-recognized consequence of this shift; namely, that physicians who provide and bill for a service, in this case cardiac stress imaging, tend to do more of it. The authors explored this relationship by linking physician billing patterns to the routine use of cardiac stress imaging after coronary revascularization—a practice with little supporting evidence.² The main finding was that the use of cardiac stress imaging for this typically discretionary indication was more common among patients evaluated by physicians who billed for the service, particularly physicians whose billing included technical fees in addition to professional fees. At first glance, the solution would appear clear—dampen the incentive to do more by additional regulatory and administrative levers and unnecessary services will be reduced.

At issue here is the oft-debated controversy surrounding physician self-referral and its associated financial incentives, which are governed by the Stark laws. Implemented in the 1990s, these laws were designed to remove the financial conflicts of interest from physician decision making for clinical laboratory tests (Stark I) and a variety of other ancillary services, including imaging (Stark II).³ To preserve the potential efficiency advantages of legitimate business arrangements, however, numerous exceptions were established. The most commonly cited of these is the “in-office ancillary services exception,” which permits self-referral to a physician-owned entity for certain services performed in the office. Although office-based care was initially designed for simple services, such as laboratory tests and chest radiography, this care setting has evolved to include expensive, high-end services, such as magnetic resonance imaging, computed tomography, and cardiac stress imaging. The financial benefits of such arrangements are clear as evidenced by their popularity—almost 1 in 5 physician practices report owning or leasing equipment for advanced imaging.⁴

As a result of such data, there are concerns that these exceptions have made the Stark laws ineffective at constraining imaging use, which increased by 70% during the last decade.⁵ Increased use of imaging was particularly fast-paced among cardiologists—a group in which payments for imaging increased by approximately 200% over the same period.⁶ However, these numbers must be viewed cautiously. Such trends occurred in the setting of considerable transitions in cardiac care from inpatient to outpatient settings,⁷ and more importantly have been linked to substantial declines in mortality related to coronary disease.⁸ Thus, it is uncertain whether the observed increase in imaging utilization is entirely a bad thing. For example, office-based imaging provides numerous potential advantages that may improve patient care and satisfaction. Chief among these are the possibility for enhanced and earlier diagnosis by physicians who are directly involved with the patient, as well as improvements in the continuity and coordination of care—all of which may result in more timely and more appropriate intervention.

The study by Shah et al¹ highlights the principal risk of in-office imaging. By examining this phenomenon in a clinical context generally considered to be “inappropriate”—namely, routine cardiac stress imaging after coronary revascularization—the investigators have demonstrated the persistence of financial conflicts of interest as a driver of utilization. The truism “if you provide a service, you're more likely to provide a service” apparently has not changed over the years.

However, several limitations of the study may potentially affect its interpretation and overall implications. First, the absence of clinical information, which is carefully acknowledged by the authors, prevents understanding physician intent for performing the imaging study. Second, due to the observational nature of the data, unmeasured confounding poses an important threat to the validity of the findings. Third, the possibility of selective referral of patients looms large. It seems highly plausible, if not likely, that patients who were thought to be in need of cardiac stress imaging would be preferentially referred to physicians who perform the service. Collectively, these limitations might explain much of the differences in use according to physician billing status, especially given that the absolute magnitudes in differences were small (approximately 7% for nuclear stress testing and 2% for stress echocardiography), although the relative differences were large.

Nonetheless, these findings remained robust in a broad range of carefully performed subgroup analyses that focused on symptom status and physician specialty. However, insofar as the current study provides a basis for action, the Centers for Medicare & Medicaid Services (CMS) has already responded to concerns related to overuse of cardiac stress imaging. Between 2005 and 2010, the CMS substantially reduced reimbursement for several imaging services, including cardiac stress imaging, to better reflect resource utilization and financial data.⁹ Following these changes, there were significant declines in the rate of increases in use of imaging services among Medicare beneficiaries¹⁰ (ie, this was after the time period examined in the study by Shah et al¹ ). Beyond payment reform, the CMS also initiated demonstration projects to examine how appropriate use criteria produced by professional organizations and radiology benefits managers may more carefully delineate the use of advanced imaging.¹¹ Although these approaches are likely to result in reducing spending related to cardiac imaging in office-based settings (at least temporarily), they also have had the less obvious and perhaps unintended effect of changing the way cardiologists are organized in clinical practice.

For many specialists, particularly those who are procedure-based, the single-specialty group model has become the hallmark of physician organization.¹² One potential motivating factor for the formation of these groups was the desire to procure indirect revenue streams through facility and equipment ownership, which could only be accomplished by leveraging their collective financial capability to acquire the necessary infrastructure.¹³ Admittedly, the fact that imaging services were possibly reimbursed too well made it attractive for practices to focus on these services. However, due to the anticipated substantial changes in reimbursement for cardiac imaging, at least in part, many cardiologists have left single-specialty groups, joining larger multispecialty and hospital-based practices. In the American College of Cardiology's 2010 Practice Census, more than 50% of the 2413 practices surveyed across the United States were at least contemplating hospital integration or practice mergers.¹⁴

Although there are no empirical data to quantify the effects of these changes on the delivery of cardiac care, there are reasons to believe they may affect stakeholders adversely. For example, it remains unclear whether physician groups purchased by hospitals will perform fewer imaging studies or not. Many of these arrangements will continue to depend on physicians meeting productivity targets for their new employers—now the hospital. Unless there is substantial reduction in the number of studies performed, this shift could actually increase costs by leading patients to outpatient facilities owned by hospitals, where care is more expensive and often less efficient. The study by Shah et al¹ has an important limitation in being unable to detect the extent to which this may be occurring. Their use of physician billing codes within claims data cannot distinguish individual practitioners from groups or hospitals.

Changes in the single-specialty group practice model also may have important implications for the quality of care patients receive. For instance, by focusing their human and financial resources on 1 service line (including ownership of relevant imaging facilities), large cardiology groups may optimize efficiency by a number of mechanisms. These include subspecialization, development of specialized support staff, enhanced patient access to necessary diagnostic and therapeutic services, and implementation of various quality-improving and cost-reducing processes.^{12 ,15} Ultimately, this “one-stop,” integrated approach likely improves patient access, satisfaction, and adherence with physician recommendations, at least as they relate to specialty care. These benefits are difficult to estimate and may be an important reason that the Medicare Payment Advisory Commission (commonly referred to as MedPAC) recently decided to maintain the in-office ancillary services exception for advanced imaging, and instead chose to focus on ongoing inefficiencies in payment accuracy.⁹

These decisions by the CMS are not only important for cardiologists, but will have significant implications for how other specialists, such as urologists and orthopedic surgeons, deliver care in the United States. Understanding the broader implications of these levers warrants careful consideration to avoid unintended consequences that ultimately affect patient care. This is particularly relevant given other initiatives (eg, accountable care organizations, payment bundling) under evaluation by the CMS that will mitigate some physician incentives. If implemented widely, such reforms are destined to affect the organization of physician practices, particularly among specialists, as well as self-referral relationships. Although these and other policies may directly lead to lower utilization of some services, they ultimately may have spillover effects that negatively influence quality and overall cost. At this time, the potential net effects are simply unclear.

At the end of the day, this debate is about “moneycare”—who bills for what and how that billing pattern influences utilization. Controversies surrounding physician self-referral and associated incentives wax and wane, and are seemingly repeated each decade. Moving forward in the current era of health care reform, the focus should be less about eliminating incentives altogether, and more about getting the price right in the first place. Regulation is imperfect—all of the nuances involved in the art of clinical care cannot be regulated (which has led to many disappointments associated with the exceptions to the Stark laws). Although most physicians delivering patient care recognize the limits of regulation, perhaps it is time for policymakers to do so as well.