EBM and P4P: Uniforms don't make the man

There is a push afoot to use evidence based medicine (EBM) to generate standards of care and then use pay-for-performance  (P4P) to Nudge people towards better behavior. The claim is that this will improve the quality of care, and cut costs by eliminating ineffective care. We should obviously use the best evidence available to inform the decision making process. And there is certainly evidence to support the belief that incentives can change behavior. But from where I sit (facing a patient in an exam room), the Laws of Unintended Consequences are hard at work. I doubt this approach will either improve care or reduce costs. In fact, I think it is more likely to cause harm.

Let’s take diabetes as our illustrative example. Markers for high quality care are derived from a combination of three sources:

  • Prospective studies done comparing groups of patients who get ‘usual care’ in the community to matched groups who get the care being tested. This usually happens in a research setting where there is considerable additional support for the patient and many of the time and revenue constraints seen in the usual care setting are not present. These studies tend to focus on patients with only one medical problem and exclude patients like those in the real world in order to make the research easier and cleaner.
  • Diabetics with good outcomes are compared with similar diabetics with poor outcomes, looking for differences. 
  • A ‘consensus’ report from a group of (usually mostly academic) experts who create a recommendations based on a review of the limited literature and what they feel works well for them

The evidence tells us that high quality care for adult diabetics has certain characteristics:

  • A1c levels are less than 7
  • A1c levels are checked every 6 months if over 7 and every 3 months if under 7.
  • LDL < 100, with an ideal goal of 70
  • Systolic BP is < 130
  • Screening for early kidney disease is done with a measure of microalbumin on a urine sample
  • There is an annual foot check for neuropathy, circulatory integrity and skin integrity.
  • The patient as a dilated retinal exam (DRE) to look for diabetic microvascular disease annually
  • The patient is current on pneumonia and influenza vaccine status

However, while these markers are all associated with good outcomes, there is no credible evidence that they CAUSE good outcomes. It is equally likely they are the RESULT of whatever has caused the good outcomes. Think of it this way. Uniforms are highly correlated with the police - if you compared 10,000 policemen to 10,000 bank clerks, you would find that a uniform is a reliable marker for the policemen. But putting a uniform on a bank clerk doesn’t turn the bank clerk into a policeman. Correlation is not causation.

One could easily argue that the cause of good outcomes is skilled clinicians who have access to current best evidence and adequate time to spend with motivated patients, allowing them to work together on the diabetes. As a result of the skill and time, good care occurs. The quality markers are byproducts of the good care, not the cause of good care.

Now imagine a typical pay for performance system. The payors (Medicare, Medicaid, private insurers) pay medical organizations more if they can document compliance with the quality markers (or dock them if they fail). The organizations set up systems to compensate their employed clinicians more if they can document compliance with the quality markers, and withhold income if they fail to comply. The clinician is now heavily focused on documenting markers. He or she is focused on actually performing the actions only secondarily to the need to document them. Using the results in a collaborative manner with patients is less likely because the focus has changed from treating the patient to treating the system.

The patient is now like the bank clerk dressed up in a policeman’s uniform. He may look like a law enforcement officer, but he is still a bank clerk. He is the same patient with the same care. Only the chart looks different. Just because the cat has kittens in the oven, it doesn’t make them bread. Put lipstick on a pig and you have - a pig with lipstick.

Thinking that financial incentives tied to EBM standards will improve care makes the unfounded and very offensive assumption that sub-standard care occurs because clinicians are ignorant and not motivated to provide good care. In fact, the overwhelming majority of clinicians are well aware of the guidelines and standards are are fiercely dedicated to providing the best possible care. But every system is designed to give exactly the results it gives. Clinicians currently work in a system that is NOT designed to provide quality.  Quality incentives for clinicians will not work unless the systems issues are addressed. 


Links to more on this topic::