Payment reform in the United States has been going on for years, and a panel at the AcademyHealth National Health Policy Conference analyzed how much progress has been made to move away from fee-for-service, and what the evidence on alternative payment models has found.
Payment reform in the United States has been going on for years, and a panel at the AcademyHealth National Health Policy Conference analyzed how much progress has been made to move away from fee-for-service, and what the evidence on alternative payment models (APMs) has found.
Kicking off the session, Michael E. Chernew, PhD, of Harvard Medical School and co-editor-in-chief of The American Journal of Managed Care, declared that to determine the success of payment reform, it has to be compared to something else.
“No matter how much you hate it, you just have to hate it less than the other things,” he said.
He focused on 2 types of payment reform: population-based payment and episode payment. The research on population-based payment models is in consensus that they save money, but how much varies a lot, and that the savings tend to grow over time. In addition, quality is typically the same. For episode payments, the literature is “all over the board” with some showing huge savings, but there is concern about expansion of episodes.
Ashish K. Jha, MD, of the Harvard T.H. Chan School of Public Health, picked up the conversation to discuss findings on quality improvements with new payment models. Before the current APMs being implemented, the primary method of trying to improve quality had been pay-for-performance. The evidence has shown that pay-for-performance doesn’t really work.
“Pay-for-performance has mostly been a bust if the goal is to improve outcomes of patients who get care under those models,” said Jha, who had been a fan of pay-for-performance for a long time.
Looking at episode-based payments, the overwhelming evidence has shown little to no impact on quality. Jha explained that these models seem to be saving money, and since they don’t actually hurt quality, “maybe that’s a success.”
These models really need ongoing experimentation and changes, particularly around the timing of the bundles. Currently, timing is for 30 days and 90 days, but clinically those time periods don’t really make sense, Jha said.
“There is a problem if you get the time period wrong,” he said.
There is also a need for better quality measures that closely align with where there will be potential cuts to ensure providers aren’t skimping on care and hurting people on the way.
Looking at accountable care organizations (ACOs), as a population-based model, the evidence on whether they have improved quality is “pretty thin gruel.” There isn’t any data to show that they have meaningfully improved quality. While quality has improved year over year, quality is also improving year over year outside of ACOs, Jha pointed out.
“I think we’re at version 1.0,” Jha said. “One of the big questions we have to ask ourselves is are we doing this to save money or are we doing this to improve quality? If we’re doing this to save money and we don’t want to harm quality—that’s fine.” And this is what most policy people admit they are looking to do, he added.
However, he would love for improving quality to be a goal, and he believes that will only happen if it is made a more substantial focus instead of it being an afterthought of a program designed to save money.
Gail Wilenksy, PhD, of Project HOPE, followed up Jha’s talk and said that one of the reasons there have been no results or small effects on quality improvement is that there have been so many attempts to have pilots and demonstrations. The problem is that some physicians and institutions are involved in multiple experiments and strategies, and they know that not all of these pilots and demonstrations will be a part of the future. As a result, they are not inclined to change their behavior to find success in these models.
Despite the fact that there has been so much talk of change going on in healthcare, a recent Deloitte survey has shown that more than 80% of physicians are still paid on a fee-for-service basis, Wilensky said. This is because most of the work of the Center for Medicare and Medicaid Innovation has been focused on hospitals, when Wilensky said she would prefer to see more models based on physicians or health systems where they are rewarded for keeping people healthy and out of the hospital.
“I just think it’s about time to move away [from fee-for-service],” she said. “MACRA [Medicare Access and CHIP Reauthorization Act] in some ways is going to force our hand. It’s not perfect by any means, but relative to what we’ve had … it’s clearly a step forward.”
Later, the panelists all discussed the idea of doing away entirely with pilots and experiments for payment models. Wilensky was the first to advocate the idea: “Just do the bundle,” she said, referring to the bundled payment model.
Mandatory payment models give the designer of the benchmarks more flexibility, explained Chernew. He added that making a model mandatory would also have more favorable fiscal ramifications.
Jha agreed that there is enough evidence that bundles, for example, don’t do any harm and likely save some money if the government wanted to push forward and make the model mandatory.
“The point is: we have a system right now, it’s not working very well; our threshold for trying something new at a large scale should be lower... We don’t need a randomized trial,” he said.