Among the many, many (really many) doctors who have written in to berate me for my column in this week’s magazine claiming that “doctors hate science” (which was shorthand and headline-speak for “why doctors are so reluctant to embrace evidence-based medicine and comparative-effectiveness research”), quite a few made a crucial point. Doctors may be paragons when it comes to using only treatments that have been proved to work. Patients are a whole ‘nother story.

Robert M. Kaplan of UCLA warned me about that when I spoke to him last week about his brilliant book, Disease, Diagnoses, and Dollars, in which he lays out proposals for using our health-care dollars a lot more intelligently than we do today. (He is particularly thought-provoking in explaining the dubious benefits we get from cancer screening.) I mentioned the 2004 study finding that something like 10 million women who had had total hysterectomies for a condition other than cancer were still getting regular Pap tests even though they did not have a cervix. Kaplan went me one better. After medical groups concluded that women who have had several clear Pap tests in a row (and met a few other criteria) can get the test every three years rather than annually, a California clinic began to implement that recommendation. But when it told its low-income patients that they could skip the Pap test this year, the women rose up in protest. How come those rich women going to private doctors get an annual Pap test, and you’re letting me have one only every third year?, they demanded.

Which brings me to some of the points the unhappy doctors have been making to me via email. What are they supposed to do when a patient demands antibiotics for a cold? for a child’s ear infection? when a patient demands an MRI for back pain or knee pain? If the y refuse, several doctors told me, they can expect a call from the patient’s lawyer that afternoon.

As my former colleague (at The Wall Street Journal) Tara Parker-Pope writes in today’s New York Times, as long as patients demand the most expensive, newest, high-tech pill, scan or treatment, we’ll never implement good medical practices. Doctors just can’t be expected to stand up to this onslaught themselves. Which is why evidence-based medicine needs to have teeth in it, and those teeth have to do with insurance coverage. Simply put, if Medicare and private insurers refuse to pay for things that are not needed or that do not work, then patients will stop demanding them and doctors can stop acquiescing in this insanity. Just to be clear, this is about more than saving money. It is also about giving patients the best treatment: prescribing something that doesn't work exposes a patient to side effects with no attendant benefit.

The American Medical Association issued a statement on Feb. 20 supporting comparative-effectiveness research but, curiously, insisted that whichever government entity conducts or disseminates that research “not have a role in making or recommending coverage or payment decisions. . . . Physician discretion in the treatment of individual patients remains central to the practice of medicine.” In other words, it’s fine to disseminate research showing that antibiotics for colds are a waste of money and an excellent way to spread antibiotic resistance, but for God’s sake don’t let insurers refuse to pay for the prescription.

As I said in the column, of course patients are individuals, and whatever works for the majority might not work for some; doctors must be free to customize treatment. But really—aren’t there some dumb practices we can agree should not be covered, especially since that will arm doctors against the ridiculous demands of their patients? (See Pap test example above.)