Saturday, December 10, 2022

Do we want hospitals and doctors to take on risk?

See also on Substack

Getting incentives right is always hard

Merrill Goozner, the former editor of Modern Healthcare, has posted a three-part series surveying twelve years of healthcare payment reform projects seeded by the Affordable Care Act and overseen by the federal Center for Medicare and Medicaid Innovation (CMMI). Part 2 focuses on the one notable success among those projects: The welding in 2014 of Maryland’s state-wide all-payer system to a global budget for each hospital that rises slightly below the inflation rate annually. In 2019 the state further expanded the global budget sphere with a Total Cost of Care Model, which sets a per capita limit on Medicare total cost of care in Maryland and provides incentives for hospitals and doctors to coordinate care, reduce hospitalization rates and meet other quality measures.

Goozner notes that according to CMS analysis, “In the eight years since Maryland began global budgeting, the program has saved the federal government well over $1 billion compared to Medicare spending in other states.”

In Part 3, Goozner leans into the premise that not only uniform payment rates and global budgets are essential to effective healthcare cost control, but also “convincing providers to take on risk is key.” That is, providers must be on the hook if per-patient costs exceed targets.

I have a question about that.

Goozman writes:

Under provider-led capitation, the hospital becomes a cost center, not a revenue source. Indeed, the decision on where to send patients (for hospitalization or highly complex cancer care, for instance) is placed in the hands of primary care physicians and their support staff, who are best suited for referring patients to high quality, unwasteful, and appropriately priced outside providers. They also are far more trusted by patients than insurance companies, who use blunt instruments like prior authorization and narrow networks to achieve their own version of those goals.

Hospital systems also should take on risk:

Managers of vertically integrated systems, if given financial responsibility for all patient care, would quickly see that the best way to maintain positive margins under global budgets that grew slightly slower than the rest of the economy would be to reduce severe sickness, reduce hospital use and promote better health.

My question: When providers are incentivized to hold per-patient costs down, rather than ratchet them up, will they behave like insurers? Will they be inclined to avoid more expensive treatments that may be optimal, whether via insurers’ current “blunt instruments” or more subtly, with scalpels?

Medicare Advantage plans hold per-patient costs down — that is, they pay less per patient for medical care than does fee-for-Service Medicare, though they receive more per patient from the federal government than does FFS (and must use most of the excess payment on extra services or reductions in patients’ out-of-pocket costs). In 2022, according to MedPAC, MA plans’ bids to CMS averaged just 85% of FFS spending, though Medicare’s payments to the plans were 104% of what their patients would have cost in FFS Medicare. While MA plan practices vary, the industry is notorious for unjustified coverage denials, onerous prior authorization protocols, and payments padded by risk adjustment upcoding.

Do we not trust providers more than we trust insurers? Individually, perhaps. But American healthcare businesses — hospital systems, large provider practices owned by hospitals or private equity, nursing home and hospice chains — are profit optimization machines (regardless of whether they are organized as nonprofits). Tales are legion of business models that depend on predatory pricing (e.g., in specialties given to balance billing, now mostly banned), pressure to use expensive equipment or expensive procedures, and pressure or incentives (e.g., from Medicare Advantage insurers) to upcode patient diagnoses. In care industries such as nursing homes and hospice, where payment is a fixed rate per patient, the incentives are reversed, and cutting staff and sometimes treatment to the bone is common.

On balance, providers are probably less likely to deny needed care than insurers. They are closer to the patient. They are trained by the current system to fight coverage denials. But incentives are powerful forces. Today they bend providers toward providing unneeded care, or unduly expensive forms of care. It’s not inconceivable that they could do the opposite.

In various niches — certain Medicare Advantage and managed Medicare plans, pilot Account Care Organization programs, Maryland’s all-payer system — providers have been and currently are paid on a per capita basis by some payers. I’m not aware of studies indicating that doctors and hospitals incentivized to reduce per-patient costs do in fact skimp on care. But there’s not much evidence regarding full capitation from all payers — except perhaps in Maryland.

I would note that a 2021 evaluation of Maryland’s Total Cost of Care Model highlighted reduction in the use of post-acute care as a goal. Minimizing post-acute care is a major source of cost savings for Medicare Advantage plans — and probably the single most intense focus of complaints from patients denied care in a desired facility, or subjected to sharp limits on length of stay.

I posed the question addressed in this post in the comment section of Goozner’s last post. His response:

This is a valid concern, but I would prefer providers making those decisions rather than insurers. Regulations could create mechanisms for dealing with outlier situations like temporary price increases (this happened in Maryland in the early days of the pandemic when discretionary use collapsed) or catastrophic reinsurance pools…which have proven successful in several states in reducing the cost of ACA plans.

Providers paid on a per capita basis can be compensated for exceptional cases. The most common means of adjusting for patient need, used in insurance, is risk adjustment, which assigns a “risk score” based on diagnoses to each patient and pays proportionately more for patients with higher scores. MA plans are notorious for gaming risk adjustment — sometimes enlisting providers they contract with in the “upcoding” of their enrollees. In an email, Goozner notes that oversight of MA plans’ self-reported risk scores is hindered by incomplete “encounter data” — data as to actual treatment — whereas state regulators in Maryland are in a position to demand more complete data. Also, the “outlier” payments he references would be from a separate pool for special cases, not integrated in a risk adjustment program.

It’s often said that those who claim to be non-ideological have ideologies — they just don’t recognize them as such. Perhaps something similar can be said about incentives: all payment systems create them. We can recognize that any incentive system can be gamed — but we still need to do our utmost to get the incentives right.

To my mind, international evidence suggests that the sine qua non of effective cost control in healthcare is rate-setting, whether on a single-payer or all-payer basis. 

Among wealthy countries, the U.S. stands alone in not ensuring that all payers pay the same rates to providers - -and lo, we spend about twice as much per capita as peer nations. It’s also true that all countries struggle to balance cost and quality, and to control cost growth. Going forward, rate-setting may not be enough; experimentation with pay-for-performance is global.

But I don’t regard U.S. providers, institutionally speaking, as less rapacious than insurers.


  1. Thanks for raising an important issue. Here are a couple of comments:

    1. I looked up the data on annual Medicare spending. Hospital care only accounts for about 30 per cent of all Medicare spending -- doctors, equipment, drugs, and home care are also very large spenders.
    This surprised me. I have been reading about Part A trust funds etc for so long, I had the impression that this was the main part of Medicare.

    2. According to govt data, Maryland Medicare spends about $10 billion a year on hospital care. Thus, saving $1 billion over 8 years is not earth shaking.

    3. The fighting over post-acute care is really a nursing home issue. And it is very important, but I don't think doctors have much to do with it.

    1. Bob: 1) Right, the hospital care savings (~ $800 million) amounts to 4% of Medicare hospital spending in Maryland. That's according to this independent evaluation of the program (p. 12). That's a really significant savings, much bigger than in other CMMI pilot programs. 2) the Maryland program "adjusts Medicare payments to hospitals based on total cost of care (not just hospital care) over the year for each hospital’s attributed Medicare FFS beneficiaries." "Lower use of post-acute care facilities is a stated goal.

  2. I am not an expert here, but I read the numbers to say that Medicare hospital spending in MD was about $80 billion over 10 years. If the caps saved $800 million, that is one per cent not four per cent.
    That is better than the out of control growth we saw in the early years of Medicare, I suppose.

    1. The $796 million in hospital savings and $179 million in non-hospital savings were over 4.5 years, 2014-2018. Not sure from what baseline the 4% claim in the independent report was derived. It would seem to suggest $20 billion in hospital Medicare spending in that time frame.

  3. Excellent post exploring one of the key issues with Value Based Payment models. I think the answer ultimately lies in robust quality measurement. Without being able to hold providers responsible for the quality of the care delivered and the outcomes of their patients, the incentives that you outline here could cause real problems.