The recent Harvard study that debunked Medicare's largest pay-for-performance program with the "sobering" conclusion that the project's process measures failed to reduce 30-day mortality has prompted many thoughtful healthcare providers to wonder anew:
Are we using the best process measures of care to save lives? Or are we just measuring how well hospital teams perform on relatively less important tasks? Are we forgetting or ignoring much better measures, such as in heart care, that are more solidly linked to better outcomes? Or are we just slow to adapt what we know about good care to public policy?
Those are pretty darn important questions because a six-year, 252-hospital pay-for-performance program, the Hospital Quality Incentive Demonstration that ran under Medicare's contract with Premier Inc., was the model for the value-based purchasing quality incentive program, which soon will redistribute reward money to hospitals from a pool of $850 million.
Cardiac Care Concerns
For heart care especially, the answer to the questions above is an adamant "yes," says Gregg Fonarow, MD, Director of the Ahmanson-UCLA Cardiomyopathy Center. In an interview last week, Fonarow listed six measures proven to reduce mortality, which the Centers for Medicare & Medicaid Services and the Joint Commission are not using to evaluate hospital care today.
As a result, "there are heart failure patients and patients with acute myocardial infarction that are paying the price."
Take statins, the cholesterol-lowering medication that also reduces coronary artery occlusion, he says. "There's unequivocal, overwhelming evidence that (giving) statins to patients who have had heart attacks reduces all-cause mortality," he says.
"We've known this since 1994; it was reinforced in 2001. It was not until 2005 when it was made into any (pay for reporting) measure set, and still the Centers for Medicare & Medicaid Services does not have this as a measure. You can be classified as having received outstanding care by CMS measures without having been treated with this essential, life-saving medication."
Under VBP, hospitals count their heart attack patients given a fibrinolytic medication within 30 minutes of arrival, the number of patients receiving percutaneous coronary intervention within 90 minutes, and the percent of heart failure patients who are given discharge instructions. Those aren't bad measures, Fonarow says, but they should be supplemented with others that have solid scientific correlation to reduced mortality.
He adds that across the country, "all of these hospitals have spent millions of dollars collecting and reporting data [on the 33 Premier and 12 CMS-approved process measures. It's been mandated for heart care. The entire country was being told that these measures, if you do a good job with them, these are hospitals providing better care. But it has not translated to any clinical benefit in patients."
Problem Extends Beyond CMS
The problem isn't just with CMS. It rests with other quality organizations, as well, that have dragged their feet in giving more reliable heart care measure sets credence. The National Quality Forum, the organization that endorses measures that CMS and other payers use to measure quality, "is behind the times as well," Fonarow says.
"In an ideal world, you'd have a harmonized set of measures that are reflective of the current state of knowledge. They'd be most focused on those areas where there are the greatest treatment gaps and variation and disparities in care," he says.
Fonarow acknowledges that it was important for payers, especially CMS, to start measuring quality when it did. Hospitals needed to be accountable. And the movement had to start somewhere. But time has passed, and CMS needs to be clear that longer life, not process measure scores, is the more desired result.
Fonarow and Eric Peterson, MD, of Duke University School of Medicine, have been pointing out the disconnect between performance measures and outcomes since 2009, when they wrote a Commentary in the Journal of the American Medical Association.
5 Measures Proven to Reduce Mortality in Heart Failure Patients
Besides statin medications, Fonarow lists five other treatments that are proven to reduce mortality in appropriate heart failure patients, but aren't in the CMS measure sets.
- Giving beta blockers at discharge.
- Giving aldosterone antagonist medication at discharge.
- Offering cardiac resyncronization therapy or CRT at discharge.
- Implanting cardioverter defibrillators to patients at risk for sudden arrythmia or tachycardia.
- Giving hydralazine-nitrate combination at hospital discharge for eligible patients who are African-American.
And by the way, the first performance payments—some might call them penalties for poor performance—will be reflected in federal reimbursement checks effective for patients discharged as of October 1. Hospitals will soon found out how they did for the first performance period, which began July 1, 2011 and ended on March 31, and whether they'll be rewarded from the $850 million pool.
30-Day Window 'Completely Arbitrary'
By October 1, 2014, the VBP formula will begin giving some weight to 30-day mortality measures. How much remains to be seen.
Clyde Yancy, MD, chief of cardiology at Northwestern University and former president of the American Heart Association, agrees with Fonarow, saying "there are in fact, probably a better collection of process measures that should be incorporated."
But he takes things one step further, questioning the reliance on 30-day mortality as a sharp cut-off for what constitutes better care.
"Thirty days is completely arbitrary," he says. "And it concerns me that so many systems, so many practitioners, are managing for the 30-day window...It's causing us, I think, to make some unwise decisions in the way we allocate our own internal resources, equipping hospitals with transition of care coordinators and nurse managers and a whole field of personnel to focus with high intensity care from discharge day zero to day 30.
Yancy emphasizes that while there is some data that the process measures in use improve outcomes, such as reducing mortality or readmissions, the window of time that improvement is seen isn't usually 30 days; it may be more like 90 days, or six months or a year.
"Well what happens on day 31?" he asks. "What it will require is more of a longitudinal commitment, to improve outcomes for patients so they are free of the need and the burden of rehospitalization."
In defense of the HQID result Premier issued a statement from vice president Blair Childs that said in part, "It's important to remember that the HQID was a first of its kind effort, predating most measures and public reporting programs in place today.
"As such, it's simply not fair to judge the measures selected 10 years ago against today's standards. Hindsight is 20/20—we won't ever be happy looking backward because the evidence will always evolve. If we were to design something today, using today's evidence, the measure set might look very different."
And Premier reiterated some of the points made by Yancy regarding the arbitrary cutoff of 30-day mortality.
"Too often, people have behaved as if the 30-day measurement of mortality is the gold standard in determining effectiveness; it is not. In many cases 30-day mortality is a very blunt measure of quality.
"First, it is in many conditions a relatively rare event. Second it comprises an extremely narrow time frame. Most studies set up to evaluate interventions, like the many of the evidence-based care processes we currently measure, evaluate long-term outcomes. No one expects these interventions will necessarily impact 30-day mortality,"
The next few years should help us know more about how improve the ultimate measure: the number of potentially preventable patient deaths.