Editor’s Note: This is the third post in a Health Affairs Blog series on Medicare physician compensation and the Sustainable Growth Rate mechanism. Paul Ginsburg and Robert Berenson have contributed earlier posts, and in the coming days the series will feature posts by Jay CrossonFrank Opelka, Eugene Rich, and Gail Wilensky.

Mina Matin is the lead author of this post; she is a third-year resident in family and community medicine at the University of California, San Francisco. Thomas Bodenheimer and Kevin Grumbach are coauthors. Thomas Bodenheimer is a general internist who received his medical degree at Harvard, spent 32 years in primary care practice, and is currently Adjunct Professor of Family and Community Medicine at the University of California, San Francisco. Kevin Grumbach is chair of the UCSF Department of Family and Community Medicine. He is also Chief of Family and Community Medicine at San Francisco General Hospital and Director of the UCSF Center for California Health Workforce Studies.

In 1997, the Balanced Budget Act created the Sustainable Growth Rate (SGR) system to control Medicare physician spending. The SGR system limits the amount by which total national Medicare physician spending is allowed to increase each year by setting yearly spending targets. For fee-for-service physician payment, Medicare assigns each type of service a relative value unit (RVU) and multiplies this RVU by a monetary conversion factor (CF) to arrive at the actual Medicare-approved fee for each type of service. By adjusting the conversion factor, Congress can manipulate total Medicare physician expenditures on an annual basis.

The SGR policy calls upon Congress to keep the increase in total physician expenditures within a targeted range of spending. Each year, if the total amount of money that physicians bill to Medicare exceeds the SGR target, Congress is supposed to reduce the conversion factor for the following year to keep Medicare physician spending at the targeted amount.

However, pressured by the physician lobby, Congress has repeatedly overridden the SGR targets and has not allowed the conversion factor to decline as much as the SGR formula stipulated. Between 1997 and 2006, annual Medicare expenditures for physician services nearly doubled, increasing from $49.2 billion to $93.7 billion. If Congress had held Medicare physician expenditures to the SGR targets, Medicare physician spending from 1997 to 2006 would have grown by 66 percent rather than 90 percent. Instead, the actual 2006 Medicare spending of $93.7 billion far surpasses the $81.8 billion specified by the SGR approach.

A political storm over Medicare physician payment policies is intensifying as advocates of fiscal restraint argue for more stringent enforcement of SGR targets while a broad coalition of physician organizations, seeking higher Medicare fees, demands that the entire SGR system be scrapped. Lost in this debate is an appreciation of how the SGR approach has contributed to a large and widening gap in earnings between specialists and primary care physicians, an income gap that is a major contributor to the dwindling number of physicians entering careers in primary care, putting the foundation of the American health care system at risk. In 2004, specialists earned 180 percent the income of primary care physicians; between 1995 and 2004, primary care income rose by 21 percent, while specialist income increased by 38 percent.

In this essay, we explore a different approach to the SGR: splitting the SGR into two pools. The objective of this modeling exercise for a split SGR is to demonstrate how Congress might have enacted an SGR policy in 1997 that would have (1) prevented the widening gap in payments to primary care and specialist physicians, while at the same time (2) controlling Medicare physician expenditures. After considering how the past ten years of Medicare physician payment might have looked under a split SGR, we then propose how the SGR might be reformed now, to control Medicare physician spending while also addressing the adverse effects of the current SGR on primary care.

Why Does The SGR System Penalize Primary Care?

Although the SGR methodology applies a single conversion factor to all services, the volume growth in physician services has differed markedly by category of service. Physician services may be divided into two large categories. Evaluation and management (E&M) services include office visits and are the most common billing codes for primary care physicians. Non-E&M services include surgical procedures, diagnostic procedures (such as colonoscopy), and imaging (such as CT or MRI scans), most of which are provided by specialists.

The volume of many non-E&M services has recently grown at a much more rapid pace than the volume of E&M services. For example, the number of office visits billed to Medicare for established patients increased by 12 percent from 2000 to 2005, while the number of billed colonoscopies increased by 40 percent, CT scans by 65 percent, and MRI scans by 94 percent. Non-E&M services are mainly responsible for Medicare physician spending exceeding SGR targets.

This rapid growth in non-E&M services has resulted in specialists’ generating steadily higher Medicare billing revenues, even in years when the Medicare conversion factor has not increased. According to the Office of the Actuary at the Centers for Medicare and Medicaid Services, from 1997 to 2006, although overall Medicare physician spending grew by 90 percent, Medicare expenditures for E&M services grew by 74 percent (from $19.7 billion to $34.4 billion) while non-E&M expenditures grew by 101 percent (from $29.5 billion to $59.3 billion). As a result, in 2006, non-E&M services accounted for 86 percent of the overage in Medicare physician expenditures. The “overage” is the difference between actual Medicare physician expenditures in 2006 ($93.7 billion) and the $81.8 billion that would have been spent had the SGR been enforced.

Because there is one conversion factor for all services, primary care physicians are essentially penalized when large increases in expenditures for specialized services drive down the conversion factor that is applied to E&M and non-E&M services alike. In 1997, E&M expenditures made up 40 percent of total Medicare expenditures, but by 2006, E&M services had decreased to 36.7 percent of total Medicare physician expenditures.

Modeling a Split SGR

What if Medicare used separate SGR systems for E&M and non-E&M services, such that the conversion factor for each category of service rose or fell based on expenditure trends within that category of service? This was the question we modeled — using information from the Congressional Budget Office and the CMS Office of the Actuary — for the period of 1997-2006 by splitting the SGR into one pool dedicated to E&M services and a second pool paying for non-E&M services. To illustrate the effects of such a policy, we developed three hypothetical scenarios for modeling the outcomes of a split SGR policy. (Detailed calculations and tables explaining the split SGR proposal are available from the Center for Excellence in Primary Care website.)

In the first model, which we refer to as the “high growth” scenario, we allow total Medicare physician expenditures to increase to the actual 2006 level of $93.7 billion, but keep the 1997-2006 rate of expenditure increase (90 percent) equivalent within each of the E&M and non-E&M pools. In the second model, the “controlled growth” scenario, we specify that the 2006 level of overall Medicare physician expenditures would be at the $81.8 million level targeted for that year under the original SGR plan and that the expenditure growth rate in each category should be held to the same sustainable growth rate of 66 percent between 1997 and 2006. A third scenario models a “moderate growth” policy, in which overall physician expenditures for 2006 are set at $87.8 billion, the midway point between the actual and targeted 2006 spending level.

By stipulating in all three scenarios that E&M and non-E&M services have equivalent rates of growth, we attempted to hold each of these categories of services “accountable” for the increases in the volume and intensity of billings in that category that drive expenditures above the SGR target, with such cost overruns in turn triggering a reduction in the conversion factor for services in that category to keep expenditures within the target.


The High Growth Scenario. Under this scenario, E&M spending in 2006 would have been $37.5 billion rather than $34.4 billion, and fees for E&M services would have been 9 percent greater in 2006 than they actually were. Non-E&M spending in 2006 would have been $56.2 billion rather than the actual $59.3 billion. The conversion factors in 2006 under the high growth scenario would have been 41.3 for E&M services and 35.9 for non-E&M services. These compare with the actual 2006 conversion factor of 37.9 for both E&M and non-E&M services.

The Controlled Growth Scenario. Under this scenario, the conversion factor for non-E&M services would decrease considerably from 37.9 to 31.3, while the conversion factor for E&M services would decrease more modestly from 37.9 to 36.0. Under the controlled growth scenario, 2006 expenditures for E&M services would have been $1.6 billion less than actual, and expenditures for non-E&M services would have been $10.3 billion less.

The Moderate Growth Scenario. Under this scenario, Medicare physician spending would have increased by 78 percent rather than 90 percent from 1997 to 2006. The conversion factors in 2006 would have been 38.7 for E&M and 33.5 for non-E&M. The E&M conversion factor under this scenario would be somewhat greater than the actual 2006 conversion factor value of 37.9, while the conversion factor for non-E&M codes would be more than 10 percent lower than the actual 2006 value.

A Split SGR: Holding Each Category Of Service Accountable

With a split SGR, E&M and non-E&M services could be held accountable for their approximate contributions to the Medicare physician expenditure cost overrun. The split SGR might have allowed Medicare to save money while improving payment for primary care and providing a financial incentive for both E&M and non-E&M physicians to contain costs — the primary objective of the SGR formula.

In this post, we have proposed an exercise of “rewriting history” by developing alternative models for SGR policy since its inception in 1997. Of necessity, such an exercise oversimplifies reality to some extent, thereby creating limitations to the data presented. One difference between our simulation models and actual SGR policy is that in our models, the conversion factor is adjusted in the same year that the expenditures occur, rather than in subsequent years. We used this concurrent year adjustment to simplify the modeling and its interpretation.

In another simplification, we assumed that the volume and intensity of services in the E&M and non-E&M pools were not affected by changes in the conversion factors. In reality, physicians may increase or decrease the volume and intensity of service delivery as fees move upward or downward. We do not think that these limitations change the main conclusion of these simulations: that tying physician Medicare fees (through changes in the conversion factor) to volume growth in E&M vs. non-E&M services has the potential to improve primary care reimbursement while putting a brake on Medicare physician expenditure growth.

At present, a major debate is taking place regarding the future of Medicare physician spending and the SGR. In its March 2007 Report to the Congress, the Medicare Payment Advisory Commission (MedPAC) was not able to come to consensus and recommended a number of alternative fixes to the SGR: separating the SGR into a number of pools based on geography; separating the SGR into a number of pools based on type of service (E&M, surgical procedures, imaging, etc.); exempting multispecialty group practices from the SGR pool and providing separate conversion factors to those lower-cost organizations; creating many SGR pools from emerging organizations created by a hospital and its medical staff; or reducing conversion factors for outlier physicians who practice unusually expensive styles of medicine.

The strategy proposed here, splitting the SGR into E&M vs. non-E&M pools, could make a contribution to resolving this debate. A split SGR, which could involve two pools (E&M and nonE&M) or more pools (E&M, major procedures, minor procedures, imaging, and other services), could reward primary care physicians who have been minimally responsible for Medicare expenditure growth by bolstering the E&M conversion factor. A split SGR could be combined with a policy to exempt multispecialty groups from the major SGR pools and to reward them if they continue to keep Medicare costs under control.

At the same time, the split SGR could reduce conversion factors for services whose volumes markedly increase. Such a policy could help to rescue primary care from its impending severe workforce shortage while helping to control Medicare spending increases by limiting expenditures for services with high volume growth.