Wednesday, December 13, 2006

The Dark Side of Pay for Performance

The pay-for-performance nerds are in a tailspin.

Yesterday’s JAMA article, co-authored by Drs. Rachael Werner and Eric Bradlow, compared the World’s Best Hospitals (top “75th percentile”) to Johnny Q Public hospital (bottom 25th) and found that, gee, people die at about the same rate at hospitals in the US. Amazing. But what was beautiful about this study, was the data mining of the data came from The Centers for Medicare and Medicaid Services (CMS)’s own Hospital Compare website. Hospital performance measures that were supposed to ferret out the good boys from the bad boys were found to be crummy measures and only predicted small differences in hospital risk-adjusted mortality rates.

But to the quality nerds that want to use such statistics to form pay incentives for physicians and hospitals, this presents a dilemma. What do you do when everyone does a good job? Or, as the quality nerds would like to say, what do you do everyone in the United States is performing in the same mediocre fashion?

Why, it’s easy! Make more measures! The quality nerds responded:
… Michael Rapp, director of CMS's quality measurement and health assessment group, said the researchers most likely would have found bigger differences between hospitals if they'd examined all 22 quality measures used on Hospital Compare. Finding only slight differences when using a few measures is not surprising, Rapp said.
Whoa there, Mr. Rapp. If I have a heart condition and want to find out about the World’s Best Hospital caring for heart disease, why do I need 21 or more other measures? But I know how you will clarify it for me:
Still, Rapp said he agrees that more quality measures are needed to evaluate hospitals. "CMS is actively working to expand quality measures used on Hospital Compare," Rapp said.
Please, Mr. Rapp, give me more data do I can be even more confused. Give our patients more measure to make this “clear.” If the public can’t figure out Medicare Part D, how the heck are they going to decipher the 22 measures you already have, or 100, 200, or even 1000 measures? Is this how we're going to give “power to the people?”

While carefully-controlled drug trials have demonstrated the effectiveness of aspirin or a beta-blocker therapies at reducing mortality after a heart attack, to suggest that measuring compliance with a medical regimen will translate to improved patient mortality outcomes after heart attack in the uncontrolled real world is a leap. Patients are not homogenously selected like they are for such trials. Every patient is unique and every patient’s problem list different. Medicine is complicated, not cookbook.

When a good researcher stops and wonders why his experiment failed, he gains valuable information to steer him in the right direction to test his next hypothesis. CMS does not seem capable of this. Rather, their answer is to develop still more convoluted “measures” rather than focusing on other, more urgent matters that might save the health care system.

I would suggest that CMS cut costs by focusing stricter guidelines for insurers dealing with Medicare patients by restricting overpaid insurance CEO’s and board members and require liability reform nationwide for any state desiring Medicare or Medicaid funds, rather than leaving the insurance, regulatory and legal interests to cripple our health care system further and price our patients out of the health care market.

You see, measuring performance measures by its very nature has a more sinister side, especially if one gets the evaluation measure wrong. Tacitly stated, measuring “performance” differentials implies one must also measure ”non-performance.” And you might as well call it “incompetence.” Doctors, hospital administrators, and people in general don’t like being called, or even considered, incompetent - especially by a governmental body that demonstrates its own inability to get the measure(s) right.

-Wes

3 comments:

Anonymous said...

It really depends who does the measuring. We have been on an EMR for 10 years and have better data than our insurers (and the feds). They are very interested in our good data because it makes them look better. Information is power. P4P is going to happen (it already is), so the question is simply, who is going to be in charge. I would personally want to give my own data.

By the way, I tag you back.

Rob

DrWes said...

Rob-
Here's some data to browse. Insurance CEO's salaries averaged over $18 million in 2002 (not including stock options, benefits, etc). How our society measures "performance" is skewed. In my estimation, if we thought about REAL performance, then our CEO's of iinsurance comapanies would be making $35K a year, while our teachers, firefighters, EMT's and social workers, etc would be making millions annually. Pay for "Performance" is in the eyes of the beholder. Financial return, my friend, should never be the incentive, but sadly, it has become that in medicine.

And thanks for the tag - I'll throw my Christmas traditions up here soon enough...

Anonymous said...

Counterpoint:

I am paid less to do good work in my practice. I am paid for poor performance, since I am paid mainly on volume. If I spend time with patients to do a better job, then I make less money (even with proper coding). It is very frustrating to have to pit my income against my conscience.

I agree with the travesty of Insurance CEO's salaries. That is why physicians need to collect their own data and take back the concept of "managed care." We are better care managers than insurance companies. I know this is not likely to happen any time soon, but it is far more likely if we actually do something about it rather than just sit around and wait for change.

Another travesty is the income of the device manufacturers and drug companies. They have billions at stake to keep the status quo and keep people sick. I am sure that their lobby has a lot to do with the drops is primary care reimbursement while expensive procedures continue to be approved by Medicare.

Sorry for the "bah, humbug," but it is a pet peeve.

Rob