Editor’s Note: This is the second post in a Health Affairs Blog series on Medicare physician compensation and the Sustainable Growth Rate mechanism. The first post by Paul Ginsburg appeared yesterday. In the coming days, the series will feature posts by Jay CrossonMina Matin, Frank Opelka, Eugene Rich, and Gail Wilensky. 

To paraphrase Rodney Dangerfield, the Sustainable Growth Rate (SGR) gets no respect. In this decade, SGR-mandated fee cuts for physicians under the Medicare fee schedule (MFS) has occurred only once, in 2002. In other years, would-be cuts have been overridden by Congress, caught between the budget scoring rules that assume substantial pay cuts resulting from inexorably increasing quantity of physician services and the programmatic and political reality that physicians cannot sustain fee cuts accomplished over many years, which in aggregate would exceed 40 percent.

Each year, Congress bestows a late Christmas present of a 0-1 percent increase in fees for the following year. However, this past December the White House threatened to veto any legislation that would reduce the substantial overpayments bestowed on Medicare Advantage plans. Such a reduction would have had the twin virtues of easing the SGR funding problem and moving toward level-playing-field competition between private plans and traditional Medicare. Lacking the will to take on the White House, Congress bought only a six-month reprieve. We’ll be back here in June.

Given the manifest problems created by the spectre of the SGR trigger, it would be tempting to just throw it overboard (and bite the political bullets on how to fund its demise). Some have suggested alternative payment approaches to fee-for-service (FFS) with expenditure targets — the current approach. Alternatives include bundled Parts A and B payments, payments for episodes of care, even capitation redux. 

Unfortunately, while perhaps conceptually preferable, these are not nearly ready for broad adoption in Medicare. The unfortunate reality is that how physicians are paid — by Medicare and most private health plans — is the last bastion of FFS in health care. Other payment systems in Medicare range on a spectrum of “prospective payment” approaches that differ importantly from FFS in that they all, more or less, reward providers for economical provision of services by not paying for each and every activity the provider carries out. 

For physicians, payment is made for services described in more than 6,000 codes, profitably catalogued by the American Medical Association in its Common Procedural Terminology (CPT) coding system. Under FFS, the more billable services physicians provide, the more reimbursement they get, and over many decades the quantity of physician services has been increasing much faster than overall health spending. In this environment, there is an understandable policy impulse to limit expenditures for physician services by establishing a formal payment relationship between the quantity of reimbursable services and the unit price for those services. As I will discuss in elaborate detail later, there is economic logic to such a relationship because, as a general economic principle, as the quantity of goods and services produced increases, the unit cost decreases because some part of those costs is fixed and can be spread over a larger volume. 

Other FFS countries, particularly Germany, provided models of how to set expenditure targets that constrained total spending by changing unit prices. Medicare adopted the resource-based relative value scale (RBRVS) and an initial expenditure control mechanism, called the Volume Performance Standard (VPS), in 1992. And for most of the first decade of their use, they actually seemed to be working reasonably well — at least to adjust payments for changes in quantity of services provided.(1) During this period, the quantity of physician services did not increase wildly; in most years, physicians therefore received reasonable fee schedule increases, generally in line with their practices’ increases in the cost of doing business. 

Different services increase at different rates 

In this decade, however, the situation has changed dramatically, with explosive increases in quantity for particular services, most notably for so-called advanced imaging, including MRI and CT scans. The quantity of advanced imaging tests more than doubled between 2000 and 2005 and continues to increase annually at double-digit percentages. More generally, the data demonstrate that quantity increases do not apply equally to all categories of services. Importantly, there have not been extraordinary increases in the provision of major surgical procedures or of evaluation and management (E&M) services, the latter category the catch-all for hospital and office visits and consultations. Which is fortunate: Because E&M services constitute about 40 percent of Medicare physician spending, the SGR funding problem, which already represents a $300 billion burden for budgeters, is merely difficult and not yet impossible. 

Different rates of service growth are not all that surprising if one considers the practical and ethically based limits on physicians’ inducement of patient services. As emphasized by Jack Wennberg, area variations in frequency of particular surgical procedures reflect “surgical signature” — that is, the combined effects of training experiences, the role of mentors, and other mostly professionally related factors. Baseline rates of surgical procedures therefore can and do vary significantly across geographic areas. However, ethical surgeons would not and, apparently, do not violate their baseline signature by responding to negligible fee increases by increasing the quantity of invasive, potentially dangerous procedures they recommend for their patients. For E&M services, which are time-based, there are natural limits on practice capacity to see more patients in response to the impact of fee limits. E&M physicians often are already stretched nearly to the breaking point — and would rather see additional non-Medicare patients, who are easier to care for. 

Accordingly, it is not surprising that increases in services, often of marginal usefulness at the “flat of the curve” but which ostensibly cause no harm to patients, occur more in the other categories of services, including imaging, minor procedures, and tests.(2) Another stimulus to overprovision of discretionary, “harmless” diagnostic and therapeutic services is the burgeoning phenomenon of physicians’ navigating their way around self-referral restrictions in order to directly benefit as owners from increasing the quantity of such services. 

In short, FFS, discretionary “harmless” services, self-referral, and, sometimes, patient demand all combine to produce the explosive growth in quantity of physician services — but only for some kinds of services, provided by some kinds of physicians. In policy shorthand, the resultant disproportionate reimbursements and incomes (as private plans emulate Medicare payment approaches) are often described as primary care losing out to proceduralists. The more accurate summary would describe the winners as niche specialists and the losers as generalists — across and within specialties. So, general surgeons are compensated less well than many surgical specialists, and general orthopedists and ophthalmologists less well than spine surgeons and retina specialists, respectively. General internists (and “cognitive” specialists like endocrinologists) are near the bottom of the income totem pole, while gastroenterologists and, particularly, invasive cardiologists are near the top, even if they once trained together in internal medicine residency training programs. 

By applying to all services equally, an SGR-imposed fee limit further accentuates income and service disparities originally created by misvaluations within the RBRVS system and the differential opportunity physicians have to increase the volume of services they provide. A root problem is that the Medicare fee schedule based on “relative values” does not permit consideration of the value of services to beneficiaries or the Medicare program. Rather, it relies on attempts to determine the relative resource costs of producing the thousands of services for which Medicare reimburses physicians and a sophisticated, but inherently subjective and, as it has been allowed to develop, highly political process. 

Medicare produces hourly reimbursement variations that likely approach the well-documented income variations whereby some specialties — disproportionately the niche specialties — earn three to four times more than primary care physicians earn. Medical students quickly learn to pick careers by “following the ROAD” — namely, Radiology, Orthopedics, Anesthesiology, and Dermatology.(3) Not only are the service commitments associated with these specialties more compatible with predictable working hours, but their handsome incomes allow them to truly enjoy their free time.

So what does the foregoing have to do with reforming the SGR mechanism? First, if Medicare were allowed to be a value-based purchaser — that is, able to adapt policies to best achieve an improved balance of access, cost, and quality on behalf of beneficiaries and taxpayers — it would straightforwardly alter fees to produce a better mix of services provided — for example, more E&M services, including greater emphasis on care coordination, and fewer imaging and other tests, as well as a higher ratio of generalists to niche specialists. Plausibly, these changes would also reduce spending. 

How quantity changes affect relative values 

But even lacking authority to alter fee levels to influence physician behavior, within the RBRVS construct, payment levels can be altered to reduce spending on physician services. Relative value units reflect relative resource costs, representing both practice expenses (rent and utilities, staff salaries, equipment and supplies, etc.) and the value of physicians’ time and effort — their “work.” Overall, physician work represents about half of the total resource costs, but the contribution of work varies significantly across the range of CPT codes. At one end of the spectrum, physician work represents most of the resource costs associated with hospital-based surgeries because the costs of the operating room, etc., are not borne by the surgeon’s practice. At the other end, practice expenses make up most of the relative value units (RVUs) of practice-based imaging services because amortization and depreciation of very expensive equipment exerts a disproportionate impact on total resource costs for imaging codes. 

Stay with me here. The per unit resource costs of services that mostly represent physician work should not be affected much by the quantity of services provided — the physician’s work component is a variable expense; that is, it is associated fully with each additional service provided.(4) In contrast, the per unit resource costs of heavily practice expense concentrated codes do vary with quantity because much of these practice expense components are fixed costs and, therefore, do not vary with each additional service provided. For example, the unit cost of imaging equipment declines as it is used more. 

Yet the Centers for Medicare and Medicaid Services (CMS) does not try to alter RBRVS values for quantity changes, even when the unit cost of fixed office expenses predictably drops as the volume of services increases. Thus, even though advanced imaging services double over five years, fees are not adjusted downward; instead, the per unit profit keeps increasing as the quantity of services delivered increases, raising expenditures and further promoting the incentive for physicians to purchase imaging equipment to self-refer imaging services. 

Ideally, the CMS would develop a technical solution to factor volume into consideration when setting relative units based on resource costs instead of setting and maintaining fixed RVUs. However, so far, the CMS has been unable even to determine the actual hours that advanced imaging equipment typically is used, thereby inflating the unit fee for these services. 

In the short term, the expenditure target mechanism can be altered to better conform practice expenses to reality. This approach might be done with categories of similar services, such as advanced imaging services, or with individual families of services; for example, the fees for various CPT codes that all rely on MRI equipment could be adjusted downward by formula if their volume increases exceed a predetermined threshold, perhaps 8 percent. Specifically targeting exploding services, such as MRIs, for payment reductions takes place in Japan. 

A few other thoughts 

The SGR is an admittedly crude approach and creates perverse incentives for individual physicians to increase their own quantity of services in a race to beat the impact of fee cuts. As mentioned earlier, there are other approaches that embed cost-consciousness within the payment method, rather than imposing a crude expenditure target on top of cost-unconsciousness. Capitation — which needs a new name — can be the ultimate cost-conscious approach, and it does not deserve the disrepute it has attained. Lots of California, Denver, and other primary care physicians (PCPs) actually prefer capitation to FFS. Learning from the execution mistakes health plans made in imposing capitation on physicians in the past, Medicare could move to per person per month payments, perhaps combined with discounted FFS payments for some CPT-based services, for PCPs able and willing to assume professional risk for their own services and be evaluated on how well they manage costs throughout the system. This would represent one approach to paying for the “patient-centered medical home.” Practices that accept this form of financial risk should be exempted from expenditure target-imposed payment cuts. 

A similar rationale applies to exempting multispecialty groups from the expenditure targets if they were otherwise held accountable for costs and quality, as recommended in the piece by Jay Crosson. Public policy should tilt — a little — to favor the formation of actual accountable care organizations. Small FFS practices don’t have to participate, but if they reject alternatives, such as forming a recognized medical home or joining a multispecialty group practice, they would remain subject to an expenditure target system, albeit an improved one. 

An important but rarely discussed SGR-related problem lies in how the size of the dollar pool on which the expenditure targets mechanism is based is funded. The emphasis has been on how much the target should increase annually, with volume of services and the health of national economy, as reflected in gross domestic product (GDP), serving as primary determinants of how expenditure targets are increased. As we are likely entering a recession — for reasons most would acknowledge are not related to physicians’ performance — it seems reasonable to ask why physician services are the only provider sector in Medicare where allowed spending is tied to state of the economy. 

Even more fundamentally, there needs to be a policy construct for reallocating resources across the various provider payment silos as enhanced technology and management approaches permit the actual patient care to move — from inpatient to outpatient, from outpatient to physicians’ offices, from acute care to postacute and long-term care. Surely some of the volume growth in some physician services reflects shifts from provision in hospitals to physician sites of service. There should be a process, for example, for reducing Medicare expenditures associated with the inpatient prospective payment system (PPS), shifting them to ambulatory care providers, including physicians. The shift would increase the budgeted pool on which the expenditure target is based. 

When physicians receive less than 1 percent fee increases year after year, we can expect physicians increasingly to stop seeing Medicare patients, at least those whose clinical expertise does not depend inordinately on the disabled and seniors. Already many PCPs have stopped accepting new Medicare patients, whether or not national surveys have detected the phenomenon. Many physicians who continue to serve Medicare patients are themselves approaching Medicare age and will soon retire, leaving patients without a personal physician and little likelihood that younger physicians will fill the void. 

In conclusion, payment for generalist physicians needs to increase. Payment for niche specialists can safely be reduced, perhaps with a redesigned expenditure target approach. Additional funding sources will need to be found to get out of the SGR budget hole. And there needs to be a process for shifting funds across provider silos.


1) In 1997 the VPS mechanism was changed in a number of ways, including tying allowed increases in expenditure targets to increases in the gross domestic product, and renamed the SGR.

2) Only recently has there been attention paid to the very real safety threat posed by repeated exposure to imaging services involving radiation, especially CT scans.

3) Medicare actually pays relatively little for anesthesia services. The specialty more than makes it up from overly generous health plan payments, taking advantage of the hospital-bestowed monopolies they enjoy when negotiating with health plans over reimbursement rates. Private plans typically pay 3-4 times more for anesthesia services than Medicare does.

4) Plausibly work does decrease as volume increases as physicians gain greater proficiency with providing a service. Under the RBRVS system, that should be reflected in periodic revaluations downward of work values, as recommended by MedPAC. For the most part, work is rarely reduced for greater “learning curve” efficiencies.