Saturday, February 13, 2010

Comparative Effectiveness---- How effective is it really?

The must read health care policy piece so far this year is this one from Dr. Jerome Groopman in the NY Review of Books. I'm not kidding. You have to read the whole thing.

The article is about "comparative effectiveness research" (CER from here on out) as applied to the field of medicine. Using best available evidence, accrued from rigorous scientific studies, ought to guide our decision making in clinical practice. No reason to give patients treatments that have been proven to be clinically ineffective. It's a waste of money and potentially dangerous to patients. So the utility of CER is not controversial. What Groopman is getting at in the piece is more subtle, and much more difficult to elucidate. He goes beyond the trope of "CER is good" to find a much trickier question, i.e. "to what extent ought CER to guide clinical decision making." In other words, how should doctors use CER in everyday practice?

Two schools of thought have arisen to answer this question. One stance, led by Budget Director Peter Orszag, is that CER ought to be a coercive force in clinical decision making. Doctors and hospitals who veer from "best practice" guidelines would face negative consequences for failure to adhere to algorithmic guidelines as determined by the latest CER. These consequences could come in the form of withheld reimbursements or actual fines. (Dr. Parks, we are writing to inform you that you owe the Sate of Ohio $1500 for failing to remove the Foley from Patient #1234646 in a timely manner.) This represents a more inflexible, authoritarian application of CER but in theory, standardization of practice allows one to control costs while providing optimal treatment strategies to patients.

The other school of thought is led by Cass Sunstein. This one is much more flexible, allowing physicians the freedom to "opt out", as it were, recommended CER paradigms if said physician deems the default pattern is not appropriate for his individual patient. Most doctors will utilize the CER default pathways, just out of expediency or inertia, but the opportunity for individualized practice is left open.

CER represents the ultimate fusion of the scientific method with medicine. No longer is it appropriate to practice medicine in the old-school, paternalistic way of deference to previous patterns. We interrogate our options in rigorous scientific models to determine which treatments seem to work better than others. But it's not an infallible, all-powerful mode of scientific inquiry. Rather than interpreting the data gleaned from some particular randomized controlled trial as insight into some sort of Hegelian, transcendental Truth we need to use the CER data in a more modest, pragmatic way. We shouldn't use the data to "define Reality" but instead just to figure out what works somewhat, sort of better than the other options under certain circumstances. Groopman gives a few examples where CER can lead to opposite conclusions, over short periods of time:
For example, Medicare specified that it was a "best practice" to tightly control blood sugar levels in critically ill patients in intensive care. That measure of quality was not only shown to be wrong but resulted in a higher likelihood of death when compared to measures allowing a more flexible treatment and higher blood sugar. Similarly, government officials directed that normal blood sugar levels should be maintained in ambulatory diabetics with cardiovascular disease. Studies in Canada and the United States showed that this "best practice" was misconceived. There were more deaths when doctors obeyed this rule than when patients received what the government had designated as subpar treatment (in which sugar levels were allowed to vary).

There are many other such failures of allegedly "best" practices. An analysis of Medicare's recommendations for hip and knee replacement by orthopedic surgeons revealed that conforming to, or deviating from, the "quality metrics"—i.e., the supposedly superior procedure—had no effect on the rate of complications from the operation or on the clinical outcomes of cases treated. A study of patients with congestive heart failure concluded that most of the measures prescribed by federal authorities for "quality" treatment had no major impact on the disorder. In another example, government standards required that patients with renal failure who were on dialysis had to receive statin drugs to prevent stroke and heart attack; a major study published last year disproved the value of this treatment.


Obviously, I'm going to advocate for the Sunstein school of CER thought. Any independent minded physician ought to. The loss of autonomy implied in Orszag-style authoritarian use of CER data isn't just an ego-motivated slight. There is an art and a mystery to medicine even in this pax romana of evidence based clinical practice. A few weeks ago I received consults for two patients who were in extremis. The one was a previously active, independent grandmother who suddenly developed abdominal pain and presented to the ER hypotensive with peritonitis. The other patient was a 88 year old guy with an acute abdomen, but he had been in the ICU for a few days and was already intubated and several organ systems were failing. Both patients ended up having ischemic colon. I operated on the woman and found gangrenous colon from cecum to rectum. After a rocky initial course she improved and is now in a rehab facility. I sat down with the family of the other guy and told them that his condition was poor and that anything I did had a very small chance of making him better; if anything the surgery would likely expedite his deterioration. I told them I would support their decision either way. They decided to just make him comfortable; he died the next day.

Two similar patients, but entirely different in all the important ways. How do you account for those subtle differences in some sort of CER-determined National Algorith? You can't. Try as people like Peter Orszag might, you can't eliminate the role of an asute doctor in clinical decision making. We go to school a long time. We see thousands of patients during our training. You can't replace that human intuition entirely with cookie cutter models.

The ultimate idea behind the Orszag paradigm is, not to sound too conspiratorial, to marginalize physicians in the provision of health care. If doctors are just carrying out orders from above, then why do we need to pay them what they make? Why couldn't a nurse practitioner or physician assistant do the exact same thing? It seems rational enough, right?

There's an arrogance implicit in such thinking. The limits of human reason ought to be clear enough after two millenia. It's futile and presumptuous to think we have determined "best practices" with the knowledge so far gained. A collective humility ought to prod us into admitting that all the best CER and evidence acquired can only make us a little better than before, with the modest hope that the future will be just a little better than today....

1 comment:

#1 Dinosaur said...

First of all, I agree with you.

However this post actually discusses apples and oranges. Medicare's use of untested parameters (eventually found to be flawed) such as tight glucose control in ICUs has nothing to do with CER. It's really an indication of CMS's laziness in trying to devise P4P schemes.

The most important part of CER is to get it right. As Groopman says, "best available evidence, accrued from rigorous scientific studies, ... to guide our decision making in clinical practice."

Well done CER would be, IMHO, things reasonable physicians (like you and me) would advocate for our patients. Wouldn't you appreciate having a guideline that says end-stage dementia patients should not have PEG tubes inserted? Evidence is building that they do not benefit from them. And how many of your ICU beds are occupied by people whose suffering they are only prolonging? Might not well-studied guidelines offer support for YOUR style of practice, ie, letting terminal patients go, rather than "doing everything", just because we can?

That said, I agree wholeheartedly that the government doesn't have a snowball's chance in hell of actually producing this kind of useful CER. Which means I agree with your concerns about being coerced to provide care that may ultimately be found harmful. Think there's any way we can avoid getting sued for it too? Me neither.