Cover Story—Pay-for-Performance
Good is Never Enough for P4PAs CMS–Premier quality standards get tougher and tougher, hospitals must push beyond their own limits
The hospitals themselves keep raising the bar. Quarter after quarter, the 250-plus hospitals participating in Medicare’s pay-for-performance demonstration project have discovered that merely maintaining today’s quality scores can risk dropping them to the bottom of tomorrow’s hospital rankings.
Even in the middle of the pack, the competition remains fierce. During the first 15 quarters of the Hospital Quality Incentive Demonstration project, the median composite quality score—a combination of process and clinical outcome measures—increased an average of 17.3 percent across five clinical areas studied, according to a Premier analysis of data from October 2003 through June 2007.
In pneumonia, for example, a hospital had to achieve a CQS above 70 percent to rank among the top half of participating facilities in the first quarter, which began in October 2003. Nearly four years later (15 quarters), that same hospital must exceed 93.5 percent to remain in the top 50 percent of facilities.
Initially, hospitals could significantly boost their quality scores by implementing basic steps, such as improving clinical documentation. Now the demonstration project requires constant vigilance; missing even a few smoking cessation opportunities, for example, can drop a hospital’s score several deciles because so many other hospitals are achieving 100 percent compliance. Achieving high marks one year doesn’t guarantee a repeat performance the next year.
Officials at New Jersey’s Hackensack University Medical Center have learned that lesson firsthand. During the first two years, only Hackensack ranked in the top 20 percent for all five clinical areas—acute myocardial infarction, congestive heart failure, coronary artery bypass graft, pneumonia, and hip and knee replacement—according to a Premier spokesman. By year three, the medical center’s scores for two of those five areas—heart bypass, and hip and knee replacement—had dropped into the fourth decile, a decline that Hackensack’s chief medical officer attributes to leadership turnover, combined with a shift in focus.
“Nothing fails like success,” says Peter Gross, M.D., senior vice president and chief medical officer. “You think you’re really good. You get smug. You relax. And you’re no longer as good as you thought you were.”
Officials at the 781-bed academic medical center were among leaders of four hospitals or health systems who volunteered to share their experiences, lessons learned and three years of CQS results with Hospitals & Health Networks. The pay-for-performance project is a joint effort between Premier and the Centers for Medicare & Medicaid Services. It could serve as the foundation—or at least a benchmark—for CMS’ plan to more broadly link reimbursement to quality, a proposal that is currently awaiting congressional action.
Officials from the four hospitals talked about the importance of developing medical order sets and other processes to make clinical measures routine. They described the benefits of designating key clinicians, essentially quality champions, to provide oversight. And they discussed the value of sharing data in order to foster healthy competition among hospitals in a system or among units.
Penalties Loom
This summer, CMS will release bonus data for the demonstration project’s third year, October 2005 to September 2006. It’s also the first time that it will impose penalties for hospitals that fail to meet certain markers. At press time, Mark Wynn, director of the division of payment policy demonstrations at CMS, declined to specify a penalty amount, saying only that he hoped it would be “a relatively small number of hospitals with a relatively modest number of penalties.”
The demonstration project, initially slated to wrap up in the third year, was extended through September 2009. For the project’s second phase, federal officials added some quality measures and revamped the bonus structure, providing a way for hospitals that rapidly improve to receive bonus money. Through the project, CMS officials are testing the collection and use of quality data, with the goal of transforming the federal system from a simple claims processor to one day driving quality improvements through value-based purchasing.
“We are very happy with what we’ve seen in the Premier demonstration so far,” Wynn says. “And we do believe that we’re seeing an overall improvement in quality rather than simply an improvement in the three dozen measures in the demonstration [project] per se. There’s a spillover to the entire hospital’s quality environment.”
As the demonstration project continues, though, it’s important to remember that the participants represent a highly motivated, self-selected group, cautions Nancy Foster, the American Hospital Association’s vice president for quality and patient safety policy.
Foster also worries about the risk of “potential unintended consequences,” given that Medicare patients are the focus of this project. “I worry about the diversion of attention away from other patient populations,” she says.
P4P-Quality Debate
Studies to date analyzing the relationship between various pay-for-performance projects and patient quality remain limited, according to RAND Corp., which recently conducted an analysis for federal health officials.
Of the seven published studies RAND identified by early 2007, only two included a control group. Those two studies identified only “very modest improvements in performance,” according to the RAND analysis, published in a November 2007 report to Congress discussing CMS’ broader concept for value-based purchasing. One of the cited studies, published in 2007 in the New England Journal of Medicine, identified quality improvements ranging from 2.6 percent to 4.1 percent over a two-year period, when comparing Premier hospitals with nonparticipating hospitals.
A study published in May in the Journal of the American Medical Association raised questions about the impact of pay-for-performance initiatives—and related financial incentives—on safety net hospitals. Through their analysis involving nearly 3,700 hospitals, researchers found that those with a high percentage of Medicaid patients demonstrated smaller performance improvements from 2004 to 2006 than those with low percentages. Thus, financial incentives may, over time, exacerbate health care disparities in lower-income patients and other vulnerable populations, the researchers wrote.
To counter such lukewarm findings, Premier officials unveiled a wealth of statistics in late 2007 covering the first 15 quarters of the project. (The data from year four is still considered preliminary.)
Along with the composite quality scores, the analysis also scrutinized appropriate care scores. Also called process scores, they assess a hospital’s ability to complete all treatment measures in a particular clinical area, such as pneumonia. During the first 15 quarters, the appropriate care score across all five clinical areas increased by nearly 53 percent.
“The process scores are getting really close to 100 percent, particularly in areas like heart attack. And that’s driving better outcomes,” says Stephanie Alexander, Premier’s senior vice president of health care informatics.
Both mortality and cost declined during the first three years of the demonstration project, according to the Premier analysis. On average, the cost per patient—across all five clinical areas—declined $1,063. Mortality also declined by 1.87 percent overall, with the greatest improvements found in heart attack (2.27 percent) and pneumonia (2.39 percent).
Transparent Incentives
In terms of motivation, though, the bonus dollars don’t seem to be highly influential. Hackensack received a total of about $1.6 million in bonuses the first two years, nearly 10 percent of the total amount distributed during that time span. But when CMS officials asked if the money made a big difference, the hospital’s answer was not really. “We like to compete for quality, and this is what our medical center is all about,” Gross says, echoing a point made by other hospital officials.
The specter of penalties also doesn’t appear to have much leverage. At this point, the bar is set so low that it doesn’t pose a significant threat, several hospital officials said privately. Under the rules, a hospital must not have improved beyond the ninth decile score, as set two years previously; year one in the case of year-three penalties.
However, CMS’ Wynn says the potential for penalties “does provide some incentive for hospitals to take the project seriously.”
Transparency is apparently an altogether different story. Hospital officials frequently say they use comparison and competition to help spur and maintain improvements.
At Aurora Health Care, a not-for-profit hospital system in eastern Wisconsin, quality scores are posted on the employee-accessible intranet in a bingo-style form, so staffers can easily track how their facility fares versus others in specific clinical areas. At Fairview Health Services in Minnesota, officials slice and dice data internally and publish data on their Web site, easily accessible to the public by clicking on the “about Fairview” button from the home page.
Raising the Bar
But consistent top-level results can’t be achieved without the oversight and enthusiasm of quality champions, engaged clinicians with responsibility for the effort. Gross credits the efforts of two nurse practitioners with keeping the New Jersey teaching hospital in the top 10 percent for pneumonia and heart attack—three years straight.
To boost its heart failure results above the fourth decile, officials at St. Vincent Health Center in Erie, Pa., decided to rely on more than physicians to quickly identify and treat those patients. By involving other hospital staffers—from nurses to coders—St. Vincent’s composite quality score increased to the top 20 percent nationally by the third year.
The demonstration project also promotes clinical improvements that can spawn a trickle-down effect, benefiting other patients, says Bradley Beard, president of Fairview Southdale Hospital. At the 390-bed facility, smoking cessation counseling—requisite for heart attack and a few other heart-related diagnoses—became a routine practice for all patients.
“It’s the right thing to do,” Beard says. By the third year, Fairview Southdale ranked in the top 20 percent in all five clinical areas, including heart treatment.
Improvement Matters
As the demonstration project progresses, keeping all hospital officials engaged—even if the facility’s performance remains subpar nationally—is of primary importance to federal officials, Wynn says. Beginning with the fourth year (October 2006 to September 2007), CMS officials added a bonus category for hospitals that attain median level performance and rank among the top 20 percent of most improved hospitals in a particular clinical area. Fourth-year data had not been released as of late May.
“I would argue that, in fact, we might be doing more for the hospitals and for our Medicare beneficiaries to take a hospital that’s doing a poor job and bring it up to a satisfactory job, than we would be doing if we took one of our very top hospitals that’s in the 98th percentile and get it to the 99 percentile,” Wynn says. “Most people wouldn’t really perceive much of a difference there.”—Charlotte Huff is a freelance writer in Fort Worth, Texas.
1. Case Study
Fairview Health Services I Minneapolis
Hard-Wiring Best Practices
The clinicians at Fairview Northland Medical Center worked their way through several stages of processing as they absorbed the hospital’s low rankings for the first year of the CMS–Premier demonstration project, says Greg Schoen, M.D., chief medical officer at the 54-bed hospital.
Some clinicians exhibited signs of denial, followed by questioning of the data, attributing the low rankings to documentation issues. Then the next quarter’s data didn’t look much better. “We finally came to the acceptance that it was not just data collection—that we actually weren’t doing some things,” he says.
That first year of results also revealed dramatic differences between similarly sized facilities within the Minnesota hospital system. While Fairview Northland ranked in the bottom two deciles for all clinical areas they participated in, 70-bed Fairview Lakes Medical Center ranked in the top 20 percent. In some areas, such as heart failure, Fairview Lakes clinicians benefited from a leg up initially because they were involved in another quality project prior to the CMS–Premier demonstration, says Alison Page, the Fairview system’s chief safety officer.
Northland clinicians, galvanized to boost their hospital’s results, quickly applied some of the medical order sets and other processes Fairview Lakes already had implemented. Whereas only a few order sets had been used previously, now Fairview Northland clinicians rely on roughly a dozen for common hospital admissions, Schoen says. A nurse also is involved with daily monitoring of patient charts to make sure that diagnoses, such as heart failure, are not missed.
Fast forward two years. By year three, the hospitals had switched places in several clinical areas, including heart failure. Fairview Northland ranked in the top 10 percent of participating hospitals nationally; Fairview Lakes had dropped to the ninth decile. “What it tells me is, in areas where we’ve seen slippage, we were relying on the vigilance of individual human beings and we hadn’t hard-wired [the processes] into the system well enough,” Page says.
The dip in Fairview Lakes’ numbers coincided with a hospital expansion effort and are now on the upswing. Along with emphasizing the importance of processes, officials there have made an effort to involve a broader array of clinicians, including doctors and pharmacists, along with nurses, in heart failure treatment.
Still, Page doesn’t only focus on individual decile rankings. She also pays attention to trends across the Fairview system. To achieve that, she compiles all of the measures for a single clinical area, such as pneumonia, into a single average that can be tracked from year to year. At the same time, she keeps an eye on the spread between the best- and worst-performing facilities in the system. “We want that bar to get shorter and shorter,” she says.
2. Case Study
Saint Vincent Health System I Erie, Pa.
The More Eyes the Better
During the first two years of the CMS–Premier demonstration project, Saint Vincent’s heart failure scores lagged slightly. While the rest of the hospital’s clinical areas fell within the top 30 percent nationally, its composite quality scores for heart failure didn’t budge above the fourth decile.
One factor was the difficulty involved in identifying those patients so the best treatment could be started early, says Joseph Cacchione, M.D., chief of quality and executive vice president at the 436-bed, not-for-profit hospital. The system implemented a series of clinical indicators to more quickly flag a potential heart failure patient, including any previous diagnosis of heart failure and a pharmacy order for an intravenous diuretic, among others.
Just as importantly, the hospital no longer relies only on physicians. “The physicians need to be part of it,” Cacchione says. “But if they are the only people we target as being responsible for making this happen, it will fail all of the time. It’s only one set of eyes.”
Now at least four sets of eyes are searching for potential heart failure patients and making sure they receive all of the recommended care. Besides the physician and the bedside nurse, other hospital staffers are involved, including the outcomes care manager and even the medical coders involved in billing, if they notice anything unusual on the chart.
Those steps, which started being implemented in 2004, began to pay off by the third year of data. “We know we are finding [heart failure patients] more often,” Cacchione says.
In year three, Saint Vincent’s heart failure scores ranked in the top 20 percent nationally. In fact, the hospital was one of only a very few in year three to rank in the top 20 percent for all five clinical areas, according to a Premier spokesman.
3. Case Study
Aurora Health Care I Milwaukee
Tapping That Competitive Spirit
Aurora’s experience illustrates that the CMS–Premier demonstration project doesn’t just attract leading hospitals that compete with each other at high-altitude levels, says Patrick Falvey, senior vice president and chief integration officer of the not-for-profit hospital system in eastern Wisconsin.
Aurora Health Care ranked as “a mid-tier performer” when the first round of results were released, Falvey says. Officials quickly recognized that they needed to ratchet up the improvement pace to have any hope of breaking from the middle of the pack. “The group of hospitals that had volunteered to participate were improving fast and soon getting close to perfect scores,” he says.
Better documentation helped. Clinicians became more diligent about writing down care a patient had received, such as the aspirin administered en route to the hospital. Even so, improved paperwork only accounted for 25 percent to 30 percent of the quality equation, Falvey estimates.
To address the remaining gap, Aurora leaders relied on human nature—that innate competitive instinct. They developed a bingo-style format to present data for all 10 participating facilities and highlight variation in performance among them. The two lowest (worst-performing) deciles were shaded in red; the highest performing deciles were spotlighted in a bright teal color.
Officials of each Aurora facility also were asked to design action plans to boost their results, whether their data placed them at the bottom of the curve or significantly higher. The action plans were posted on the employee-accessible intranet, allowing other hospitals to weigh in, sharing their own strategies and lessons learned.
By year three, a larger number of Aurora facilities appeared in the top 20 percent of hospitals nationwide. Pneumonia treatment was a particular standout; only one of the 10 participating Aurora facilities fell below the top two deciles.
Falvey likes to pull out a master graph to demonstrate how far Aurora facilities climbed during the nearly three years. In late 2003, Aurora’s overall appropriate care score—the percentage of times the hospital system administered the right care across all five clinical areas—fell slightly below the median score for all hospitals (78.9 percent) and far below the score (87.1 percent) for the top 20 percent.
By early 2006, Aurora’s score was 96.9 percent—exactly the same as the top 20 percent of hospitals. It also placed Aurora above the median for all 250-plus participating hospitals, which was 93.2 percent. “We had to really move and do a lot of quality improvements to get to that level,” Falvey says.
4. Case Study
Hackensack University Medical Center I Hackensack, N.J.
Data and Rewards That Make A Difference
To motivate clinicians to provide the right care each and every time, Denise Patriaco, R.N., circulates on the units. She reviews patient data not from several quarters ago, but from yesterday and the day before. And she provides rewards to clinicians who strive for optimal care, even if they don’t always succeed.
“I’m trying to get them to buy in,” Patriaco says. “And not just reward when everything is done right, but reward the learning experience. I know they’re going to try harder the next time.”
Peter Gross, M.D., Hackensack’s chief medical officer, credits the vigilance of Patriaco and another nurse practitioner, Geri Vargas, R.N., in helping the 781-bed teaching hospital to consistently rank in the top 10 percent for heart attack and pneumonia treatment during the first three years of the CMS-Premier demonstration project. Two other clinical areas—heart bypass and hip and knee replacement—haven’t received a similar level of scrutiny during changes in leadership, he says.
In the third year, the ratings for those clinical areas placed Hackensack in the top 40 percent, down from the top 20 percent. “These numbers require constant vigilance,” Gross says. “The competition has gotten more and more intense since the first year.”
For taking the right steps, or even attempting them, Patriaco hands out coupons to be redeemed for a free cup of gourmet coffee or, as they’re accumulated, rewards like lunch or clothing items. She’s educated a wide swath of staffers, including technicians and clerical people, on the typical and atypical signs of a heart attack, so patients can be screened as soon as possible.
And she’s trained nurses on a low-key way to approach physicians if a patient hasn’t received a recommended standard of care, such as aspirin for a heart attack patient. “That was very difficult to start,” she says. “And the doctors did not take kindly at first to feeling like their care was being questioned.”
Patriaco also feels strongly that, for change to occur, reviewing data from several months back is nearly irrelevant. Clinicians have moved on, and they’ve forgotten the relevant details. “It’s old data they don’t care about,” she says.
Instead, she reviews patient charts in real time, even going to the emergency department early enough to talk to the night staff. She recently noticed that a patient hadn’t received an antibiotic when required. Backtracking, she learned that the doctor had ordered the drug, but the nurse hadn’t been able to administer it. The patient, because of kidney issues, only required half a dose; the nurse couldn’t get the specially prepared antibiotic from the pharmacy within the optimal timeframe.
“That’s something that three months later they [the clinicians] wouldn’t have remembered,” Patriaco says. “They would have seen thousands of patients since then.”