Conscripting hospitals to test new payment models

This is a TIE-U post associated with Nick Bagley’s Health Reform and Its Legal Controversies (Michigan Law 866, Fall 2015). For related posts, see the course intro.

A few weeks ago, the Center for Medicare and Medicaid Innovation put the finishing touches on a new demonstration project to test a bundled payment model for hip and knee replacements. With its previous projects, the Innovation Center had asked for volunteers. This time, however, the Center said that it would require hospitals in 67 geographic areas to participate. Critics pounced, saying that providers can’t be forced into a demonstration project. Is that right?

To answer that question, some background might be helpful. The big problem with voluntary demonstration projects is that volunteering organizations could be—indeed, they likely are—systematically different from organizations that don’t sign up. What works for volunteers may not work for non-volunteers. Heck, we won’t even be able to say whether the model works for the volunteers. Who do you compare them to? What’s the right baseline against which to measure success?

In other words, voluntary demonstration projects raise all the problems of causal inference that you get when you run an observational study. You can’t really learn from them what would happen if you scaled up a payment model.

One alternative would be to force randomly selected organizations to adopt the model. But that’s still not ideal. Forcing an organization into a program and not its competitor could put one or the other at a disadvantage. Worse, there could be weird intra-market effects. If the model sucks, some patients—perhaps the sickest patients—might shun the participating hospital. You could get screwy selection at the patient level.

A better approach would be to force every organization in certain geographic areas to adopt the model. You’d then compare organizations in the tested areas to organizations in similar areas that didn’t try out the model. You’d then have some confidence about what would happen if the model was rolled out nationally.

That’s exactly how the Innovation Center’s new bundled payment program works. It’s designed, as the Center explained, “to avoid the selection bias inherent to any model in which providers may choose whether to participate.” The program is controversial; among other things, the critics say that it’s not really a bundled payment program at all, but instead a disguised version of pay-for-performance.

But what about the legal objections? The critics accurately note that there’s nothing explicit in the Innovation Center’s statute that speaks to whether the Center can conscript providers. Given Medicare’s traditional reluctance to mandate participation in demonstration projects, they argue that Congress would have spoken clearly if it meant to confer that power on the Center.

But the critics are making too much of Congress’s silence. Read against the backdrop of what Congress hoped to accomplish with the Innovation Center, the best inference to draw from the statutory text is that the Center can mandate participation where appropriate.

Consider first that Congress vested the Center with sweeping authority “to test innovative payment and service delivery models to reduce program expenditures … while preserving or enhancing the quality of care furnished to individuals.” The whole point of the testing is “to determine the effects of applying such models … on program expenditures … [and] quality of care.” Why would Congress create a Center to measure the effects of new delivery models—but then force it to run really crappy tests?

Bear in mind, too, that the Center has the power to adopt payment models “on a nationwide basis,” but only if CMS’s Chief Actuary certifies that the new approach won’t increase spending. The Chief Actuary couldn’t responsibly make that certification if he could only look to voluntary demonstration projects.

More importantly, Congress would never have armed the Center with the extraordinary power to introduce a delivery model nationwide if it couldn’t rigorously test that model. To the contrary, Congress specifically told the Center that it “may elect to limit testing of a model to certain geographic areas,” suggesting that Congress anticipated the very kind of area-limited study that the Center now means to run.

In short, the Innovation Center was right to reject bogus legal objections to mandated participation. We’ll only ever find out which models work—and which ones don’t—if the Center can force hospitals and physicians to try them out.


Hidden information below


Email Address*