Hospitals participating in a Medicare-sponsored, pay-for-performance demonstration project have sustained initial gains in quality improvement and seen a decline in mortality and costs for selected conditions over 3 years, according to data released by Premier Inc., a hospital performance improvement alliance.
The median hospital cost per patient dropped by $1,000, and the median mortality dropped by 2%. The project has 250 participating hospitals, and more than 1 million patient records were analyzed.
Premier, which manages the Centers for Medicare and Medicaid Services-funded Hospital Quality Incentive Demonstration project, estimated that if every hospital in the U.S. achieved the same benchmarks, there would be 70,000 fewer deaths each year and hospital costs would drop by $4.5 billion.
At a briefing, Mark Wynn, Ph.D., director of payment policy demonstrations at CMS, said the project is one of the agency's primary arguments in favor of value-based purchasing, a policy CMS regards as the most effective way to reward efficiency and value. “Relatively modest dollars can have huge impacts,” he said.
Dr. Evan Benjamin, chief quality officer for Baystate Health System in Springfield, Mass., agreed. He was the lead author of a study looking at earlier data which found quality was higher among the 250 incentivized hospitals than it was in control hospitals that reported data publicly but were not given incentives (N. Engl. J. Med. 2007;356:486-96).
The demonstration project began in October 2003; data covered every quarter through June 2007.
Hospitals were given aggregate scores for each of five conditions–acute myocardial infarction, heart failure, coronary artery bypass graft, pneumonia, and hip and knee replacement–based on reporting for 27 process measures. Hospitals with fewer than eight cases per quarter were excluded, and all data were adjusted using the All Patient Refined-Diagnostic Related Groups (APR-DRG) methodology created by 3M Information Systems.
Overall, hospitals improved by an average 17% on a composite quality score used by the project.
There was a continuing downward trend in performance variation among the hospitals, with all moving toward the ideal, said Richard Norling, president and CEO of Premier Inc. Costs and mortality were lowest for the hospitals that were on target 100% of the time with 100% of patients, he said.
For instance, the mortality for coronary artery bypass graft patients was close to 6% at hospitals that met appropriate care benchmarks in only half the patients or fewer. Mortality was just under 2% for facilities that met those benchmarks in 75%-100% of the patients, Mr. Norling told reporters.
Attaining the goals of the demonstration project required cultural shifts and investments in information systems. Before the project, the Aurora Health Care system was reactive and was achieving only incremental quality improvement, said Dr. Nick Turkal, president and CEO of the Milwaukee-based nonprofit system. The system's 13 hospitals have 100,000 admissions annually. Data on meeting the pay-for-performance goals are given to employees every 60 days, and are updated regularly on the system's Web site for the public to see. Mortality and costs are down significantly, but “we're not done yet,” he said.