CMS Grades 21 Payment Models

When CMS released the dryly titled “Synthesis of Evaluation Results across 21 Medicare Models (2012-2020)” last year, the 49-page report landed without much fanfare or comment. The goal was to analyze data from 21 payment models that CMS’ Innovation Center has experimented with to “inform future model development.” Two years ago, the Innovation Center set out to pare its portfolio of payment models to concentrate on designs that were effective in reducing cost and improving quality while also addressing healthcare inequalities.

The synthesis “was more of a ‘what did we learn’ and less of a ‘what can we do about it’ perspective,” explains David Muhlestein, Ph.D., J.D., chief strategy and chief research officer for Leavitt Partners, a healthcare consulting firm in Salt Lake City. “I’m glad they did it and kind of looked at everything across the board and tried to do an assessment of what we have learned so far.”

Lawton R. Burns, Ph.D., MBA, chair of the Health Care Management Department at the Wharton School at the University of Pennsylvania, has a harsher assessment: “I can sum it up, and this is my take: It’s a nothing burger.”

The evaluation looked at the gross and net effects in Medicare spending and measures of utilization in areas of inpatient admissions, emergency department visits, post-acute care, inpatient readmissions, quality of care, self-reported satisfaction with care, and mortality. Gross savings to Medicare are calculated before incentives are paid to participating providers. Net savings are what is left for Medicare after incentives are paid.

Of the 21 programs, 14 showed gross savings in spending as the result of improvements in inpatient admissions (10 of the models) and/or post-acute care (14 of the models). Eighteen of the payment models included incentive payments and the results were evenly split: six showed net savings for Medicare, six showed net losses, and six had no effect.

The CMS researchers separated the models into two groups: nine models that targeted acute or specialty care — joint replacement, for example — and 12 focused on primary care and population health management.

In broad strokes, the results from the acute and specialty care models were more favorable than those for the primary care and population health. The difference is partly a reflection of the fact that acute and specialty care is more costly, so there are more opportunities to find savings by reducing spending on big-ticket items, such as hospital admissions. Indeed, all nine of the acute and specialty care models had a positive effect on gross savings, driven by reductions in inpatient admissions and spending on post-acute care. Eight of the nine had a significant effect on post-acute care utilization or spending. Five models reduced inpatient readmissions (one of the nine did not include inpatient readmissions among its outcome measures). It was not all good news. Just two of the models reduced emergency department visits. Of the six models that included incentives, only two produced net savings: the Bundled Payments for Care Improvement Advanced Model and the Medical Care Choices Model.

The results for gross savings among the 12 primary care and population models were a mixed bag: five of the models produced significant gross savings; six produced nonsignificant savings; one model, the Advance Payment ACO (accountable care organization) Model, resulted in an increase in gross spending. Just three of the 11 models with incentive payments produced net savings, and three of the models resulted in net losses to CMS: the Advance Payment ACO Model, the Comprehensive Primary Care Plus Model and the Next Generation ACO Model. Four of the primary and population models resulted in significant reductions in inpatient admissions or spending on them; five reduced emergency department visits. The results for hospital readmissions did not show the 12 models, as a group, as holding great promise. Three models did produce reductions in hospital readmissions, but the other nine produced nonsignificant changes.

‘Sucking wind’

Gathering and analyzing data from CMS Innovation Center’s 21 models was important, concedes Burns, because, he says, “once it’s clear what doesn’t work, what’s left over has a greater chance of working. That’s a good thing, unless what is left over only kind of works.”

“I’m not a crank or the curmudgeon out in the peanut gallery,” Burns continues. He says he has looked at the results dispassionately: “If those are all the savings we are going to get, that’s just a little, tiny pinprick in the bucket. It is just not going to do it. We had better look for some other things.”

Burns has been skeptical of the ACO models’ ability to rein in healthcare costs since they were launched as part of the Affordable Care Act. He and his Wharton colleague, Mark Pauly, Ph.D., wrote that ACOs were not the silver bullet for controlling healthcare costs. Just three of the models that CMS looked at in this review were ACO models. The Medicare Shared Savings Program, which is CMS’s largest ACO, was not included in the synthesis.

“To be honest, and here again I don’t mean to rain on any parades, but ACOs are kind of sucking wind,” Burns says.

Using disincentives

Muhlestein is a bit more optimistic about ACOs, and is holding out hope that CMS can still find “big-picture” models that will reel in costs. But his hope is wavering. “Either we really need to keep looking for options, or there simply isn’t a way that you can use any of these fee-for-service models to radically lower the cost of care,” he says. “I’m not quite ready to do that second one and say there isn’t one. But I am increasingly pessimistic that there is.”

A survey conducted by The Chartis Group, a healthcare advisory firm, found that 71% of primary care physicians believe their capabilities to deliver on value-based care will improve in the next five years. However, 20% of the doctors surveyed did not know whether their practice was enrolled in a
value-based care payment model.

“OK, I’m willing to wait,” Burns says. “But we’ve been waiting
10 years, and the ship really hasn’t set sail.”

One way to bolster ACOs, he says, is to increase the incentives for primary care practices to enroll in value-based models. Right now, Burns says, the rewards are “just not worth it.”

Or, counters Muhlestein, CMS should make fee-for-service less appealing. CMS “is not making the disincentives strong enough,” Muhlestein says. “Fee-for-service is comfortable. As long as it continues to be a viable option, most people will just stick with what they know.”

Along with disincentivizing fee-for-service, Muhlestein says CMS needs to design ACO models for providers like audiologists, physical therapists, addiction medicine and surgical specialists — services where Medicare spends billions of dollars in care.

“There is just a need to continue to create more models that cover a broader range of things,” he says. “If you are not going to solve it in one fell swoop, you can at least make a lot of little chips at it and get better. It’s that whole idea of ‘small strokes fell great oaks,’ where you need a whole lot of small strokes.”

Focusing on sicker patients makes sense, says Muhlestein, because it is harder to do effective population management if that population is healthy.

“I think you can see meaningful savings for those (sick) populations,” he says. “Maybe in a nutshell, it really is that Medicare needs to be doing more models for sick patients than for healthy patients.”

Robert Calandra is an independent journalist in the Philadelphia area.