Health & wellness

Study: Pay-for-performance did not reduce deaths in Medicare pilot program

A large Medicare pilot program that paid hospitals more if they consistently administered certain medications and vaccinations, provided appropriate counseling for people with heart conditions, or hit other quality targets did not reduce the number of patients who died within 30 days of admission to the hospital, a study published online Wednesday by the New England Journal of Medicine found.

The results are “sobering,” the authors wrote. The program served as a model for a major national initiative being rolled out this year.

Such pay-for-performance programs have been central in efforts to change how health care is paid for, shifting from a system that pays doctors for each test or treatment to one that rewards them for keeping their patients healthy.


These kinds of incentives are the right path forward, but the formula needs tweaking, said Dr. Ashish Jha, associate professor of health policy at Harvard School of Public Health and lead author of the study.

Get The Weekender in your inbox:
The Globe's top picks for what to see and do each weekend, in Boston and beyond.
Thank you for signing up! Sign up for more newsletters here

“The question is, what do you pay for?” he said. “What are the performance measures? That part we haven’t figured out ... We have not come up with the right set of metrics to focus on.”

The study looked at mortality rates among more than 6 million patients treated over six years at 252 hospitals involved in the Premier Hospital Quality Incentive Demonstration. The program tied up to 2 percent of Medicare payments to performance on 33 quality measures, including two related to mortality. Most assessed how consistently hospitals carried out recommended treatments, tests, and preventive care -- so-called process measures. The patients were treated for heart attack, heart failure, or pneumonia, or had bypass surgery.

The mortality rates were compared with those at thousands of other hospitals that publicly reported performance on the same measures but were not part of the payment program. The authors found that deaths declined in both groups but at similar rates, even among those hospitals considered poor performers at the start of the program.

Participation in the Premier program improved process measures at Springfield-based Baystate Health, and that was the central focus of the program, not outcomes, said Dr. Evan Benjamin, chief quality officer. Baystate, one of three Massachusetts hospitals that participated, did see a drop in deaths from heart attacks during the program, he said.


The national program and the Premier program are not identical. But, starting this fall, the Centers for Medicare & Medicaid Services will use 13 measures to determine distribution of about 1 percent of hospital payments. Many were pulled from the Premier program, including whether discharge instructions are given to people who have heart failure, a blood culture is performed in the emergency department before antibiotics are administered to patients with pneumonia, and clot-busting drugs are quickly delivered to patients having a heart attack. The national program also includes surveys of patient satisfaction.

The federal agency has acknowledged that the system is a work in progress.

“I think we would all be surprised if they hit the perfect design and formula the first time out of the gate,” said David Smith, senior director for data analysis and research at the Massachusetts Hospital Association.

Smith said the study should prompt some concern, given that “there will be winners and losers financially” under the new system. But, he said, there’s more to consider besides mortality rates.

The process measures “weren’t just pulled out of thin air,” he said. “There’s evidence that these things, in the end, are helpful to people’s health.”


Jha said patients care less about whether a process is carried out consistently than they do about whether it improves their health in the end. And some process measures can be distracting, he said. He pointed to the one tracking whether patients are given discharge instructions. That can be a subjective process. Do patients understand the instructions? Do they follow them at home?

The best way to improve people’s health is to focus on a small set of metrics with a big effect, he said.

“There is no higher-value measure for hospital care than mortality rates,” he said.

Chelsea Conaboy can be reached at Follow her on Twitter @cconaboy.