Arnold Kling  

Against Moneyball Medicine

The Climate Skeptic... Localized GDP...

Richard Dolinar, M.D., writes,

The physician has taken the history, performed the physical, reviewed the labs, and discussed the illness with the patient and family. He knows the patient's wishes, desires, and values. All this critical information must be considered when treating patients.

...It is absurd to think that a third party, operating at a distance in time and space from the patient being treated, is able to make a better medical decision than the treating physician and therefore should be allowed to preempt the treating physician's decisions.

It is absurd to think that a baseball statistician, operating at a distance in time and space from a player being scouted, is able to make a better decision about the player's likelihood of major league success than a local scout. Except that it happens.

It is absurd to think that credit scoring models are better predictors of default than human underwriters. Except they are.

Today, we are in between two images of the doctor. One image is the heroic personal savior, who uses his own experience and intimate knowledge of the patient to make the best decisions. The other image is the trained technician, who gathers data, feeds it into a decision tree, and implements that recommended course of action.

I think we need to combine the two images, rather than treat doctors solely as heroic personal saviors. Too often, doctors' personal experience provides nothing but availability bias. As for intimate knowledge of patients, often you're kidding yourself if you think that your doctor has even digested what's in your folder.

The odd thing about "pay for performance" is that it mixes up the two images. It uses the trained-technician image to determine what doctors ought to be doing. But then it gears the bonuses and penalties to individual doctors, as if they were heroic personal saviors.

My own view is that a remote third party probably can use statistical evidence to make good recommendations for a course of treatment. Patients should be aware of those recommendations, and doctors should allow patients to choose between the statistical recommendations and the doctor's personal recommendation. I would like to see consumers responsible for more of their own health care spending, with access to the third-party information, and with doctors not being so ego-driven in claiming that they know best. In that case, consumers would have the means, the motive, and the opportunity to make cost-effective decisions.

Comments and Sharing

COMMENTS (8 to date)
conchis writes:

Malcolm Gladwell discusses similar issues in Blink - recounting an example where a shift to a basic checklist approach for treatment decisions (I think it was some sort of cancer diagnosis, but I can't remember exactly) significantly outperformed the "more-informed" and "more-personalised" judgements of doctors.

Acad Ronin writes:

Atul Gawande has an article in the New Yorker on the health benefits of having doctors and other health professionals administer treatments by following checklists. He makes your point about mixing up the images of doctor as solo hero and the doctor as technician.

Turns out that following checklists delivers better outcomes.

Rafal Smigrodzki writes:

I am a physician myself and I almost totally disagree with Dr Dolinar. *Of course* the statistician looking at the results of a double-blind clinical study has a better idea of what works and what doesn't, than a doctor writing scripts and talking to patients afterwards.

The only place where a physician's judgment applies is when there is conflicting data, data of uneven quality, and when the patient herself is different from test subjects and not sure what she wants. Many, especially older or less intelligent patients have significant difficulty in making sense of even relatively clear research results - ideally the physician should be able to provide a better interpretation. If only more physicians knew how to calculate the PPV from test specificity and disease prevalence....

In general, the vast majority of input in a physician's recommendation should come from peer-reviewed research results, and only a small fraction may be personal guesswork based on experience.

Dr. T writes:

I agree with Dr. Smigrodzki. Dr. Dolinar's viewpoint is accurate only in limited situations. Those are: an excellent physician, an unusual problem or a strange presentation of a common problem, and sufficient time to take a detailed history, perform a thorough exam, review all the lab and imaging studies, and discuss the findings with the patient at a second visit. I estimate that this type of scenario comprises far less than 1% of patient-physician encounters, and that less than 5% of physicians are good enough to beat the 3rd party's diagnosis and treatment plan even in these types of scenarios.

As a salaried pathologist, my viewpoint of medicine is quite different from the majority of physicians. Healthcare is a mess today, and I believe that physicians are the number 2 cause, with governments (often responding to the demands of the citizens) being number 1.

Rich MD writes:
Patients should be aware of those recommendations, and doctors should allow patients to choose between the statistical recommendations and the doctor's personal recommendation.

I wonder what Dr. Merenstein would have to say about your proposal.

fundamentalist writes:

Computer programs that diagnose illnesses and suggest treatments more accurately than can most physicians have been around for a long time. These programs should be made available on the internet with a fee for access. Then people should decide for themselves which drugs to take or not and they shouldn't need a prescription. I have traveled outside the US several times and have found that many countries don't require prescriptions for medicine, even antibiotics. I can't understand why American doctors think Americans are more stupid than Mexicans or Moroccons about medicine. I can't think of any reason to require a prescription for medicine, except maybe for narcotics.

Leanne writes:

Sometimes researchers, in their process to reduce the number of variables and homongenize standards of practice, lose clinical reality and common sense. Evidence based practice drives us....but don't forget reality may alter the steering. Evidence based practice needs to be blended with value based practice.

Rich writes:
I can't understand why American doctors think Americans are more stupid than Mexicans or Moroccons about medicine.


Doctors do not decide what medications require a prescription. Your federal government (via the FDA) does.

Comments for this entry have been closed
Return to top