Editor’s Note: In the post below, Jeff Goldsmith argues that the concept of accountable care organizations (ACOs) is “not ready for prime time.” In a response, Aaron McKethan, Mark McClellan, Elliott Fisher, and Jonathan Skinner state that ACOs represent a critical step away from volume-based health care payment and toward better health and better care at lower cost.

Everyone from President Obama to Rush Limbaugh seems to agree on one thing about health reform:  the vital importance of containing health costs. Policymakers seem finally to have learned that merely reducing Medicare’s unit payments to hospitals and doctors backfires.  It increases  “moral hazard”–driven volumes and further shifts costs from public programs onto drowning businesses and cash-strapped households.  So an intensive search is on for policy alternatives for saving money.

There were lots of cost containment ideas bandied about during the presidential campaign: disease management, medical homes, the electronic health record, prevention and wellness, bundling provider payments, “consumer empowerment.”  Many are cropping up in health reform legislation. All these ideas suffer from a common problem: it’s depressing how little credible evidence there is that any of them actually save money.

In this yawning policy vacuum, people, including congresspeople,  are flocking to what seems like a bright idea that originated the Dartmouth health policy factory- the accountable care organization (ACO). The ACO grew out of the important Wennberg/Fisher work on market variation in health cost, and the idea has been studied and developed by the Dartmouth group and others, notably researchers at the Brookings Institution’s Engelberg Center for Health Care Reform. It was built on what is (to my consulting colleagues, anyway) a stunningly obvious discovery: Medicare spending for physician services tends to cluster around hospital service areas.

The policy leap that led to the ACO idea was that since these “communities” already use hospitals,  let’s assume that they and their hospitals are actually “virtual organizations” and give them a global budget. Consumers would not be aware that they were being treated by ACOs. Rather, they would be “attributed” to them: virtual patients of virtual organizations. Aggregate health spending for attributed patients would be tracked, and increases in that spending would be capped using a form of “shadow capitation.” ACOs that lived within the caps would get their fees increased.  Those that overspent would see their fees reduced or frozen.

Some policy types on the Medicare Payment Advisory Commission (MedPAC) became intrigued with the ACO idea and saw it as a solution to the Sustainable Growth Rate (SGR) problem, a durable policy headache bequeathed by Congress to physicians in the Balanced Budget Act.  The SGR “global budget” for Part B physician services failed because there were no workable feedback loops to alter physicians’ behavior. In MedPAC’s vision,  the ACO  was a mechanism for salvaging the SGR. If a national SGR didn’t work to constrain physician behavior, why not create a hospital service area–specific SGR?

Fisher and colleagues theorize that when you cap the total resources coming into a specific “community,” hospitals and physicians would form organizations to accept and manage the global payments. This would create an economically motivated community lobby for not building more hospital beds, not recruiting additional cardiologists, not putting CT scanners in physician offices. Little delegations from the ACO would visit the high utilizers and work with them to get them to give up their diagnostic equipment and reduce their incomes. (As Dave Barry would say, “I am not making this up.”)

The problem with this movie is that we’ve actually seen it before, and it was a colossal and expensive failure. During the 1990s, many hospitals and physicians believed that the Clinton health reforms would force them into capitated contracts with health plans.  This catalyzed a flurry of mergers and physician practice acquisitions, all motivated by a desire to control the stream of payments from health plans, rather than being subcontractors to those who did. Many system builders assumed that they would increase their market share through selective contracting at the expense of docs and hospitals that remained unorganized.

Risk-bearing physician/hospital organizations and hospital-sponsored preferred provider organizations (PPOs) sprang up all over the country. They typically paid their docs a discounted fee based on prevailing rates in the community (platinum for proceduralists, battered copper for primary care docs), in anticipation of capitated health plan payments. Some of these hospital/physician efforts actually succeeded and survive today. AultCare in Canton, OH, and Sanford Health Plan in Sioux Falls, SD, are examples.

However, these were outliers in an expensive failure.  Employers and patients preferred open panels managed by health insurers to closed panels managed by providers. Billions of dollars were lost.  After rivers of red ink, most bold 1990s hospital risk-sharing experiments were terminated, along with the CEOs and physician leaders who created them.  Many of the practice acquisitions were reversed, as hospital systems sought to rein in their expenses and adjust to an open-panel world dominated by point-of-service style health plans

However, the 1990s left behind an expensive legacy:  highly concentrated local provider markets. All but a handful of American cities now have at most three, and sometimes only one, providers of hospital services, who can extract monopoly rents from health plans. In many communities, high-income specialists like cardiologists, radiologists, or anesthesiologists formed powerful single-specialty groups, with authoritative bargaining power not only with local hospitals, but with health plans. Accountability was not the agenda, but rather unavoidability.

There were numerous reasons for the 1990s collapse of at-risk hospital/physician partnerships, besides the failure to find willing buyers of their services. These efforts lacked infrastructure, experienced management, as well as reliable and timely cost information to support cost management. They assumed global risk but paid for care on a fee basis, just as Fisher and colleagues propose.  But these hospital-sponsored organizations could neither redistribute income nor exclude their high-cost providers (who inconveniently generate most hospital profits).

Some things have clearly changed in the ensuing decade. Hospital management has markedly improved, as have the information technology (IT) systems at their disposal. Private practice is also collapsing in many communities. A rapidly increasing percentage of physicians, particularly primary care physicians, are now hospital employees.  A larger percentage of the physician community receives hospital subsidies for call coverage. Many of these subsidies are, in fact, extorted from the hospital by specialists in scarce supply, destined to become scarcer.  An entire generation of 80-hour-a-week baby-boomer physicians are retiring and being replaced by younger physicians who want to work 30 hours a week.  You are not going to see a lot of these younger physicians in utilization review committee meetings after hours; they are going to be at their kids’ soccer practices.

What hasn’t changed is the fragmentation of care, the huge disparities in income and political power inside physician communities, and also the level of suspicion that physicians have of their now much more powerful local hospitals.  There is also, sadly, a thundering absence of collegiality — in my view, the central precondition of assuming risk and managing care. This absence is palpable in suburbs and even more pronounced in many “lifestyle dominated” resort communities in the sunbelt. All that physicians in these areas seem to have in common is a desire to live comfortably in a nice place.  They are “collections” of physicians, not communities.

The hospitals in these areas appear formidable:  they have beautiful campuses,  prestigious boards, and deep financial reserves (until last fall, anyway).  But these hospitals have been picked clean of vital outpatient services by their medical “communities.” See below. (The diagram represents an actual but deidentified Sunbelt Hospital.)


Entire disciplines have disappeared from hospitals: ophthalmology, cosmetic surgery, gastroenterology, urology. Even community-based internists and family practitioners have stopped coming to the hospital; their patients are cared for by hospitalists who work full time inside the hospital.

In other words, while the hospital has become more involved in subsidizing physician practice, physician communities have drawn away from the hospital and function increasingly independently on a day-to-day basis. Wennberg’s own data show that something like 40% of physicians no longer have any Medicare hospital-related fee income. So squashing hospitals and physicians back together into economic interdependence in a joint hospital/physician economic pool makes no real-world sense.

However, many hospital systems are creating large formal multispecialty groups. These are not “virtual organizations” but real organizations with P+Ls, medical directors, and management infrastructure. Prominent examples in my home region include Carilion Health System in Roanoke and the Bon Secours Health System in Richmond. Voluntary ACO arrangements, with Medicare and with private insurers, may find enthusiastic partnerships with many of these hospital-sponsored physician groups.

Medicare beneficiaries should not be “assigned” without their knowledge to ACOs, as Fisher and his colleagues have discussed. Rationalizing how care is provided will not work without patient cooperation. Rather, patients should be given rich comparative cost and value information. They should also be given the powerful incentive of reduced cost sharing and the opportunity to forgo costly Medigap coverage if they choose to receive care in an ACO.  It is vital that both local health systems and patients consent to participate in sharing risk.

The State of Massachusetts seems headed toward imposing an all-payer mandatory, communitywide ACO model for cost containment purposes. There is time to reconsider what I think is a reckless decision. Unless they are very careful, and build on existing risk-sharing arrangements, the Massachusetts ACO experiment is likely to be a gory and comprehensive failure. Virtual assignment of patients, virtual organizations, “shadow capitation” superimposed on fee-for-service based economically independent docs, further consolidation of local hospital monopolies: we really ought to know better.  Those single-specialty monopolies and scarce specialists mentioned above will be able to name their price (just as the big teaching hospitals in Boston do today) and grab a big piece of the virtual pie.

The sad reality is that most hospitals, even the well-managed ones, simply lack the tools, leadership, and leverage to enable them to bear and manage global risk.  Many will not possess them in a decade. The mandatory ACO (apparently still a live option in the June 2009 MedPAC report) is one of the worst health system reform ideas since the Health Systems Agency. Fisher and his colleagues are attempting to broaden the idea to encompass independent practice associations (IPAs), existing multispecialty groups, even academic health centers.  But the core idea remains that physician communities and hospitals in defined geographies are viable economic units.  They are not. The ACO idea may be worth experimenting with, but if recent history is any guide, it is not an idea worth betting the Medicare programs’s future on. As a vehicle for reorganizing medical practice at the community level, ACOs are not ready for prime time.