Peter B. Bach

Does the Doctor Know Best?

VTR Date: December 29, 2012

Dr. Peter Bach B. Bach discusses the issue of terminally ill patients.


HOST: Richard D. Heffner
GUEST: Peter B. Bach, MD, MAPP
VTR: 06/14/12

I’m Richard Heffner, your host on The Open Mind. And both times today’s guest has joined me here to discuss matters of real medical concern, the response to his provocative comments both on whether – as much current wisdom would have it today – we as a nation actually “waste” money on terminal patients… and on the inevitability of a now rapidly developing but not yet fully comprehended new medical specialty, that of the full-time, salaried professional “hospitalist” … each time he has elicited numerous viewer responses, including those from his medical colleagues.

Well, so be it this time…for once again I’ve asked Dr. Peter Bach, the director of the Center for Health Policy and Outcomes at Memorial Sloan-Kettering Cancer Center, to elaborate here upon a fascinating recent opinion piece in the New York Times that he obviously directed at readers and viewers like you and me. Its intriguing title: “The Trouble With
‘Doctor Knows Best'”.

And I, of course, would ask my medical guest just what is that trouble?

Didn’t my doctor and your doctor spend all those years studying everything one could possibly learn about human health and illness, didn’t he or she slave impossibly long hours in college over physics, chemistry and calculus, then labor hard, harder, hardest in medical school, then as a sleepless Intern, then as a Resident, then whatever, wherever else just precisely TO know best?

And don’t we — his or her patients — deserve the comfort of darn well believing that the Doctor does know best? And that’s the question I would put to you first, Dr. Bach.

BACH: Well, thank you for the question and thank you again for having me. As you set it up, it makes me realize the pressure to … if you will … assume that doctors know best in virtually all medical things.

And as a doctor I grew up in exactly the environment you talked about. I studied physics, too, and I slaved long hours, too. And much of what’s taught during that time is extremely important and helps doctors make excellent decisions.

I suspect some of your viewers may think my punch line will be not that doctors know best, but instead that patients know best.

Certainly your set-up suggests that. But that’s actually not what I’m arguing at all in this piece. I’m actually arguing that statisticians know best.

HEFFNER: Statisticians?

BACH: And I don’t use that term …

HEFFNER: Numbers people?

BACH: Numbers people. Other nerdy quantitative types. And I don’t really mean that and I don’t envision a system some day where you’re sitting in the waiting room and they say, “The statistician will see you now”. Or anything like that. But what I’m trying to argue and make clear is that in the world of disease prevention and in disease screening, where what we’re doing as doctors is sending patients … actually not patients but people, with nothing wrong with them, for medical testing that the decision to do that or not is actually made or shouldn’t be made in the same way we, as doctors, make decisions about how to take care of people who are actually sick.

And here’s the rationale. Doctors are trained, as you suggested through long hours, through repetition, through exposure. They essentially are apprentices, they learn a craft, they become artisans. And we all talk about the “art” of medicine and its intersection with science.

And that experience makes us good at what we do. But that experience comes from taking care of people who are actively ill in front of us, whom we then order tests on or make interventions on and we can see the results.

And in many areas we’re doing that without a solid base of scientific research … experience driven. And it’s community experience driven, too, we go to meetings, we talk about our collective experience … we meet in the mornings and talk about patients and share our ideas and experience.

But all of that applies to taking care of people who have a problem. And once they have a problem, then we can judge if what we do resolves that problem.

In disease prevention … we’re in screening and I talk mostly about cancer screening in this piece, there’s nothing wrong with the person. There may be something wrong that we don’t know about, but you can’t then do the intervention and see what happened.

And that’s essentially my argument. That we’ve tried to apply the same cognitive structure … a sort of experiential learning to cancer screening tests that we apply to clinical medicine and it doesn’t work.

And the way you know it doesn’t work is that we have decades of evidence that we screened the wrong people, we screened them in a way that’s not optimal for them, we over screen people we shouldn’t be screening, we screen … we fail to screen people we should be screening. And each time a screening test then gets adopted … as we learn more and more and we refine what should be done and how it should be used … the resistance is very strong.

Stronger than it would be if we had an evidence base or a scientific basis for doing things. And it’s because of this cognitive space that most of us occupy which is “doctor knows best” that we can figure it out through experience … the best way to approach things. And it just doesn’t work for screening.

HEFFNER: What does work? What would work … if the doctor doesn’t know best … who does? Who could?

BACH: Well, so … and I mentioned in the piece a number of times … we actually do research studies and we involve tens of thousands, and sometimes even hundreds of thousands of patients in the aggregate, to figure out which screening tests work and what they do and what they’re complications are.

And screening is something you do to populations of people. And the outcomes you affect. The good stuff and the bad stuff is measured at the population level.

Now individual doctors, as I order a particular test, I’m ordering it only for the person in front of me.

HEFFNER: Indeed.

BACH: But I’m participating in a population based intervention, which is not what any of us are trained to do. And so, my … I, I revert to thinking about just this individual patient. Could I find something in this person that could help them, and think about well, let’s apply the screening test because I could help them in that domain.

HEFFNER: Yes, but the question I would put to you is how could it be otherwise?

BACH: You mean imagine a different world … or …


BACH: … then … I’m not sure, you know, the joke, of course, is that, you know, maybe everyone should go see a statistician and see if they should be screened.

Ah, or maybe my colleagues and I can get better at trying to use the sort of metrics … an approach that … about population health that screening tests had better fall into.

HEFFNER: Explain yourself … please. For instance.

BACH: Well, for instance … so I wrote … one of the things I mentioned is that I worked on a recent set of reports around screening for lung cancer.

And one of the interesting things about screening for lung cancer is that we can tell … with a good degree of, of accuracy what the chances that someone will get lung cancer … and it’s really based on their age and their smoking history.

And a couple of other things. There’s a few occupations that put people at risk. There’s a few genetic components that you can get from family history.

HEFFNER: But mostly smoking history.

BACH: It’s mostly smoking and age. Right, that drives it. Right. And there are models that are available that statisticians have developed. I was involved with one of those models, but there’s multiple models that actually can put people in buckets of … you know they’re at high enough risk that screening makes sense. And they’re not at high enough risk that screening makes sense.

And screening makes sense because it prevents about one in five lung cancer deaths amongst people at high enough risk.

But it has downsides, too. It uncovers, it opens up Pandora’s Box of other things. The way we screen for lung cancer is we use a CAT scan.

And we take pictures of the entire chest … the lungs and everything else you see … the heart and the architecture of the lung itself. When you do that you find things that aren’t dangerous and sometimes you find cancer also.

And when you find the things that aren’t dangerous, which is about 95 out of 100 times, other sorts of tests get done, and sometimes procedures get done to patients.

And all those things harm them unless you happen to find a lung cancer and then, within that, a lung cancer that you can change how it will behave.

So even in the highest risk people we only prevent about one in five lung cancer deaths when we screen people. Four out of five people, even if we find them, we don’t alter the outcome and there’s a whole bunch of harm.

So, you take a statistician’s view of that … we need to focus our energy only on the people high enough risk that exposing them to all of those risks is worth it.

HEFFNER: Do you think for a moment that the American patient … what do I mean by that … I don’t know what I mean by that, but you know what I mean by that … you take it the generality.

BACH: Okay.

HEFFNER: … would accept that approach?

BACH: HmmMmm. It’s a challenge. We’re living in an increasingly consumer directed world. And obviously if you go on the Internet it’s not hard … in fact, right before I came on your program I went on Google and I started looking for lung cancer screening advertisements. And I was able to find places all over the country saying, you know, “Come in if you’ve ever smoked. Come in if you’re over 40 and you have a relative who smoked. You know … for your lung cancer tests.”

And that makes absolutely no sense, that would just be harm without benefit. And I … in the consumer directed context people may start demanding it.

But, let me, let me present an alternative view of how my profession would behave in an ideal way. For whatever reason doctors are the prescribers in health care … right … most medications you can’t just go get yourself. There’s over the counter, of course, but most things, you need me to write you a prescription.

Most tests … you can’t just walk in and say, “Here’s my American Express card, do an MRI on me”. You need a doctor to prescribe. So we are, in a sense the people responsible for figuring out whether or not you should have the test.

And because a patient demands it isn’t a great reason, actually, for ordering something. Now if you demanded OxyContin, I would say “Well, you know, if that’s not appropriate, I’m not going to prescribe it. That’s harmful. Right, it’s addictive, it has all these bad things”.

So I would never feel sort of … you wouldn’t be able to “twist my arm” if you will.

But a CAT scan doesn’t seem like a big deal … right. Similarly if you came to me and said, “I want by-pass surgery … do by-pass surgery on me” … hopefully all doctors would say “You don’t need it, I’m not doing it.”

And so I actually think the important question is whether or not doctors abdicate responsibilities that they shouldn’t. Because things like screening tests seem like their so low risk. Less risk than OxyContin prescription, less risk than by-pass surgery.

But I actually think we shouldn’t look at this as a continuum … we should be prescribing what we think is in the net best interest of patients, based on our judgment and, in this case, happens to be based on clinical research that patients have volunteered to participate in so we would make better decisions.

HEFFNER: I’m puzzled how the notion … “doctor knows best” comes into all this. You’re saying “for crying out loud, doctor doesn’t know best” because he or she is influenced by mythology.

BACH: No … that, that’s too strong. And, of course, it’s an opinion piece so it has a generalization in there which is unfair … right, there’s about 750,000 of us in the United States … I don’t pretend to really know how all 750,000 of us think.

HEFFNER: But you … excuse me … but you …

BACH: Yeah?

HEFFNER: … do know from the statistics you use and the, the point you make here about the numbers of tests that have recently been challenged in terms of their validity in their use on such a large scale.

We’re being told, it looks to me, every week … stay away from this test, unless … stay away from that test, unless … so you’re, you’ve got a lot of substance to stand on here.

BACH: Oh, right … I’m not … I’m not backing away from what I said, I just don’t want to suggest that all doctors make the same set of cognitive mistakes. But I think collectively the evidence is that we are making this particular cognitive error, which is that we are … we trust the “seat of the pants”, we trust our instincts, and we’re a little bit shy about trusting research and science that tells us something difference from our instincts.

And that’s not an intellectual failure … doctors are smart people, it’s a … it’s a problem with how we’ve been trained in this particular context. We’ve been trained to follow our instincts and our experience because that actually has to guide us in almost everything we do. There’s not a scientific report on every decision I make. Most of them I make because I’ve learned to behave in a certain way, from experience, an apprenticeship. This is a place which is really very, very different from what I do … taking care of actual sick people.

HEFFNER: And the question that’s raised about financial rewards? How much does that play in this … in your estimation, your educated estimation?

BACH: It, it probably varies. There’s no question that there are specialists in some areas who make a meaningful fraction of their income, in some cases, as a result of cancer screening.

So there are doctors, for example, urologists who mostly take out prostates for prostate cancer. They are … their business boomed with the creation of the PSA test.

That said, they’re not usually the ones ordering the PSA test. Right? You only get to the urologist after somebody else ordered it.


BACH: So …

HEFFNER: … interesting point.

BACH: … it’s difficult to, to trace it back. Right? That doesn’t mean they aren’t strong advocates for continued testing. I think in many cases because they believe they’re helping people and the PSA test is how that happened to occur.

But it’s not quite the scenario that I think it’s been portrayed. Same with gastroenterologists … there’s some who … you know, really a lot of their business is screening colonoscopy. But, again, they don’t self-refer … right. Patients are sent to them by primary care doctors, or by other colleagues. And that doesn’t mean they don’t believe they’re helping people. But they’re not generating their own business.

HEFFNER: Let’s look at the other side of it. The question that’s … the matter that always comes up. Fear of lawsuits if you don’t say … get thee to a urologist … general practitioners …

BACH: After the PSA test is sent?


BACH: Yeah. It’s … these are … that’s a real challenge and I don’t think … I’m very familiar with the studies that have looked at this question more generally … and I don’t think it’s convincing in either direction.

You will hear about the defensive practice of medicine … in fact one of the letters in response to my piece said that I downplayed malpractice too much.


BACH: I actually don’t think it’s clear. The best …

HEFFNER: Wait a minute … don’t think what is clear? That there is that connection?

BACH: … how impor … how important malpractice in, overall, in terms of its effect on health care, the provision of health care services. It may linger at all in the back of our minds … but a couple things.

One is, if you follow practice guidelines in a way that your colleagues follow them, then malpractice cases really can’t be successful. Because the first test of a malpractice suit is that you deviated from the standard.

So, it should be … if there are widely held, widely propagated, endorsed guidelines that say, “Do not screen these people” … do not screen this individual. It doesn’t mean you can’t be sued … you can be sued for anything … but it means it’s unlikely to prevail.

But more generally, when you look in states that, for example, have caps on liability …tort caps …


BACH: … they’re essentially low malpractice states, for argument’s sake. You don’t see much less of this sort of testing and ordering than you do in states that don’t have caps. That’s the closest we can get to understanding whether or not malpractice drives things.

HEFFNER: Would you, would you forgive me if I said that’s not very close.

BACH: As … proof?


BACH: I, I think I prefaced my comment by saying “It’s not clear”. You can also look in other countries which have fundamentally different tort systems, like Germany. And they use imaging and things like that just as much as we do.

HEFFNER: They do?

BACH: Yes.

HEFFNER: So the medical idea or ideal prevails throughout the world, throughout the Western civilization … what-have-you. It’s not just our doctors, if you wrote this and published it in a German publication …

BACH: Ah. For cancer screening … I was talking more generally about imaging and diagnostic testing used in hospitals for surgery or surgical procedures. You can’t convince yourself that German is substantively different from us, even though the malpractice structures are totally different.

On the cancer screening front the Europeans are different from us … I don’t want to say they’re ahead or behind, and obviously Europe is heterogeneous.

But if you look in the UK or you look in Germany, look in France, their approach to screening much more closely mirrors a public health approach where they have systems in place for screening populations that are defined in a, in a prospective way and they have outreach and they follow-up and they have quality standards and they have dedicated centers in most places.

HEFFNER: Do you make a connection between that and the financing of European systems … and the financing of our system of medical care.

BACH: I’d be hesitant to do, because I don’t feel like I have the expertise. But … you would never say that the German health system, the UK are similarly financed either.

Germany looks much more like we do, a mixture of different public programs, both international and regional. Private programs, self-pay programs, different kinds of co-pays.

It’s starts to look a lot like the US, but they don’t have uninsured … right. And, and the UK is, is dominated by a single … a nationwide system, called the National Health System.

HEFFNER: Where do you want us to go? And how would you begin to think about getting us there?

BACH: So, hopefully, you’re talking beyond just this conversation (laugh) … the …

HEFFNER: No, I’m not …

BACH: Yeah …

HEFFNER: … being cosmic … I’m talking about what you wrote …

BACH: Ahemm, it would be nice in all areas of medicine if we were better at applying what we already know how to do in an effective and efficient way.

And we fall down over and over again … it doesn’t matter if it’s cancer screening or it’s the newest, most intriguing technological operation … or some expensive drug or basic preventive services like well-child care.

We know the things that we should be doing from a scientific basis and we fail to do them over and over again. And you can’t name an example where there aren’t important short-falls that lead to harms either because beneficial services do get to patients … I give it the example of the Pap smear here, or because services that are over-exuberant and harmful are given to too many patients. And there’s piles of examples of that.

So, the ideas for solving this have to do with better integration of care systems, quality reporting and monitoring, better cohesion and care coordination, use of clinical decision, supporting electronic health records, more guideline driven care. These are all the ideas. That whole laundry list of ideas has yet to convincingly, in part or in whole, show that we’re getting the system improvements.

But my biggest disappointment in our system is captured by the last line of this piece. And I don’t remember exactly what I wrote, but basically that, you know, we’re very good at inventing things that can help people and very bad at finding the people to use them on.

HEFFNER: “It is time for us to own up to our shortcomings in cancer screening and we must start by acknowledging a hard fact. Doctors sometimes don’t know best. We are terrific at inventing new tests that can be performed on people, but we have less good at figuring out which people should have them.” Philosophic statement. Not a statistical statement. But you do approach this … as we began … from a statistical point of view. Is it a scientific point of view as opposed to what we find generally?

BACH: No, it is, it’s a statistical point. It’s not a philosophical one. The overriding philosophy is one that no one could disagree with … we should be getting the most out of the bio-medical advances that we already have, even as we seek new ones. But the point is a statistical one.

When we started looking … not “we, me”, but “we” our health care system …


BACH: … we started looking at care of patients after a heart attack in the US. We found that half the patients weren’t going home on aspirin … something we had understood for decades was critical in preventing recurrent heart attacks and death.

So we looked further, we found they weren’t going home on beta-blockers either. That’s not because doctors did know about it, it’s because we didn’t have systems in place to make sure that every patient, when they went home, knew to go buy an aspirin … right … and take one every day.

Or got a prescription for a beta-blocker and knew the pharmacy to go to. And every single thing we do, you can find these sorts of short-falls. The example I give of the Pap smear … Pap smear stands alone as the most effective cancer screening test we have ever come up. And you know, it’s more than 50 years ago.

And today, or this year … 4,000 women, plus, will die of cervical cancer in the US. And the vast majority of them will have either never had a Pap smear or not had a Pap smear in a decade or two and this is a test we’ve had around for longer than most things that we encounter in the health care system and we’re failing to reach those women, even as many more women have too many Pap smears, Pap smears they don’t need, Pap smear every year even though it’s safe to do it every 3 years or even longer in some women.

You know, so … to me that’s an opportunity to save lives just left by the wayside because our system doesn’t act like a system.

HEFFNER: You’re talking about a systematic approach then? You’re talking about putting systems in place.

BACH: That’s right. So, most European countries and I … I’m aware enough of the political discussion around this to never suggest we should be like Europe, although some things about Europe I like a lot … like the food.

But they have population based approaches to cancer screening. And we have population based approaches to things like pediatric immunization … right … that’s why we use the schools … right … to make sure kids get their shots … right … and that they have well-child-care.

But we don’t to cancer screening and I don’t really know why.

HEFFNER: And that, I suppose is the place we should end, besides the fact that I’m getting the signal that our time is up. We don’t know why …

BACH: I’m sorry it’s ended so quickly.

HEFFNER: But, you’re going to come back … I hope …

BACH: It would be my pleasure.

HEFFNER: Thanks, Dr. Bach, for joining me today again.

BACH: Thank you very much.

HEFFNER: And thanks, too, to you in the audience. I hope you join us again next time. Meanwhile, as an old friend used to say, “Good night and good luck.”

And do visit the Open Mind Website at to reprise this program online right now or to draw upon our Archive of 1,500 or so other Open Mind and related programs. That’s