A Sustainable Post-Pandemic Health Care System Needs Adjustable Payment Models

[Chernew:] I personally don’t like thinking of the payment mix as a portfolio problem because I see a whole range of broader [problems], with the fee-for-service payment writ large. Of course, I see a lot of problems with population-based payments or episode-based payments, as well. But given what happened, delivery systems are understanding that they are facing this service risk and that [..] moving to population-based payment models will give them both a little more mobility and I would argue a little more flexibility and a little more incentive to maintain and build an efficient health care system when we move past, and hopefully soon move past, the Covid-19 pandemic.

[Lee:] Risk adjustment scores are used to influence next year’s budgets for commercial risk contracts and Medicare Advantage. Any insights into how big a problem this is and what might be done about it?

[Chernew:] The quick answer is, it is a big problem. The part that I’d like to highlight is, we spent a lot of time in the article [earlier publication in NEJM Catalyst] criticizing aspects of fee-for-service. The issue you raise points out concern with broader payment models, like population-based payment models, and, as you mentioned, even in Medicare Advantage, which is essentially a type of population-based payment model where you have to deal with issues like risk adjustment. While these models might be able to smooth some of the impact of volume decline, you do have to figure out how to deal with the risk adjustment issues. You have to figure out to deal with the attribution issues. You have to figure out how to deal with the quality measurement issues.

[..] The important thing is that when we design a payment model, it’s hard to design one that will work perfectly when we have an event like Covid. In addition to the payment model the way that it’s written, you’re always going to need some amount of ability to adjust. The biggest problem in the fee-for-service system was not simply it was fee-for-service. It’s that we hadn’t built the infrastructure to figure out how we were going to do things like allocate government support. Going forward, we’ll have a better plan.

[Lee:] could you give a quick explanation of PPS [provider premium support] and how you think it might be able or unable to respond to the kind of turmoil that we’ve had this year?

[Chernew:] Provider premium support is basically a different way of setting the benchmarks in population-based payment models. The benchmarks are the crucial targets that organizations need to hit to define savings or losses. The idea behind provider premium support is you essentially set them administratively, like with a growth rate, the way we set almost every other price in the Medicare system.

There are several advantages to versions like that. One of them is, in a situation like Covid, you don’t have a huge drop off in fee-for-service spending driving down benchmarks. [..] It gets rid of some of the budgetary concerns that we have, and Covid’s going to create huge budgetary concerns for the Medicare program or for the nation as a whole. It avoids other problems. One of the biggest ones, for example, in existing ACO [accountable care organization] programs I call the ratchet effect, which is if you perform well, they then lower your benchmark going forward. You basically pay a price for past successes.

Provider premium support solves that problem, and in some ways it’s analogous to what we normally think of premium support when people are buying, say, insurance. The difference is, the risk is borne by the providers, not by the beneficiary.

[..] a number of other issues that Covid has brought up. A good example would be telemedicine, where we’re worried that if we allow broad access, there’ll be a dramatic increase in use in the fee-for-service system. All types of population-based payment models can help with that, but things like provider premium support can, perhaps, provide the most robust benchmark-setting system to get around that problem and allow the most flexibility for things like telemedicine or hospital-at-home or a number of other innovations.

[Lee:] there are three basic questions about telemedicine. It could end up being just more care, because I know many institutions for whom in-person care has gone back to normal; we could just end up with what we were doing before plus telemedicine. It might be better care, that more care might lead to better control of risk factors. But then, it might be more efficient care, too. It could replace care that’s higher cost. Do you have a sense of how incentives might drive the third, which is more efficient care that’s also good?

[Chernew:] many folks who think about designing the incentives worry much more about the actions of the least well-motivated providers in the world as opposed to what the best providers in the world will do. The concern is that if you allow access very, very broadly to services like telemedicine, that would be opening an opportunity for abuse of the system, overuse, without getting the comparable quality that you need.

The advantage of moving to a different payment model, like a population-based payment model — you could think about it as an episode-based payment model — is that the system doesn’t have to put in place all of the various fences that people are talking about around telemedicine: the originating provider, what type of services are getting paid, a whole bunch of other things. You don’t have to put as much attention on that, and you can allow providers to be flexible, to use it when they want to use it, when they think they can get value. And the payers — Medicare, the commercial payers — are not at as much financial risk.

The problem with all of these innovations that are great for a large number of people, but can be overused and in some cases dramatically overused, is that the fee-for-service system forces a bunch of administrative fences, caps, limits, certifications, administrative burdens to justify use, that will drive everybody crazy. You can get around some of that administrative complexity if you just give the providers the flexibility to use it when they want to use it without having to worry as much that there will be some providers that will overuse the service.

[..] understanding that no payment model is perfect, the right payment model will vary for different provider types, and that every payment model has some blend of a lot of these ideas. Even provider premium support operates with an underlying fee-for-service chassis, for example. We need to think about how to blend these ideas, how to blend these incentives, and how to make them robust, how to react should we have a Covid pandemic.

[..] I happen to believe that a broader payment model, like a population-based payment model, gives the delivery system, the providers, the most flexibility to provide efficient care at a price that can be both afforded by the payers and sufficiently profitable or sufficiently valuable to the providers. And that’s going to be hard to do in a fee-for-service system that we have facing all the budget pressures that we’re [facing].

[..] Going forward, the health care system is going to have to become more efficient. What these alternative payment models do is they allow the provider system to capture the financial gains associated with that efficiency in a way that is going to be particularly important over the next 5 to 10 years.”

Full interview, Chernew ME and Lee TH. New England Journal of Medicine Catalyst: Innovations in Care Delivery. 2020.10.7