• How should physicians respond to requests for nonbeneficial interventions?

    In JAMA, Allan Brett and Laurence McCullough ask us to

    consider an insured man with new-onset low back pain and no findings suggesting a serious disorder. He requests a magnetic resonance imaging (MRI) scan “to see what’s going on.” The physician should explain that MRI will not change initial management and that standards of care do not endorse MRI for the patient’s problem. Suppose the patient counters that he wants to get his money’s worth for his insurance premiums. The physician should reply that medical professionals must utilize health care resources responsibly and that the insured patient’s health care costs ultimately affect other patients. What if the patient offers to pay out of pocket? Even then, the physician should not order the scan: ordering it reinforces a habit of nondeliberative clinical reasoning and the mistaken idea that costs are irrelevant.

    Is the physician exercising an appropriate role to promote parsimonious care? Brett and McCullough think so. This is obviously in conflict with consumer-directedness or patient-centeredness or shopper-empowerment or I-will-do-what-I-damn-well-please-with-my-money-ness.

    Patient autonomy is often invoked to support patients’ requests for specific interventions. According to this perspective, patients’ preferences are always decisive because medical decisions reflect value judgments, and patients are always better suited to choose interventions consistent with their personal values than are physicians. However, this rationale is flawed. Distorting biases may influence a patient’s clinical judgment, and autonomous patients sometimes make decisions that confer no benefit or put their health at risk.

    Using patient autonomy to justify acquiescence to patients’ requests for nonbeneficial services violates professional integrity. Professional integrity requires physicians to adhere to standards of intellectual and moral excellence. Physicians achieve intellectual excellence by submitting clinical judgment to disciplined, evidence-based reasoning. Physicians achieve moral excellence by protecting patients’ health-related interests as a primary concern, keeping self-interests systematically secondary. Commitment to professional integrity requires that physicians challenge requests for nonbeneficial interventions. For example, patients may derive subjective value from taking antibiotics for viral infections; however, such value is not decisive in the absence of benefit from the medical perspective. Over time, allowing patients’ demands for unnecessary interventions to trump careful clinical reasoning results in a nondeliberative, rote practice style that undermines clinical excellence.

    Many will not find this argument compelling. Either you think medical science trumps patient autonomy or not. Some will think not. But there’s another perspective, not articulated by the authors. It’s most salient in the self-pay case. Imagine the patient, insured or not, agrees to pay the full cost out of pocket. In this case, only two entities matter, presuming externalities don’t exist or are insignificant. We have a buyer (the patient) and a seller (the physician). The transaction is similar to any other exchange of good or service for money. The exchange should be entered into by both parties willingly, voluntarily.

    Now whose views should dominate? Does the patient have the right to the (nonbeneficial) care? Or does the physician have the right to refuse it? I think this case is perfectly clear. Each has the right to a veto. The physician need not substitute the patient’s values for his own. It’s his service and he has the right not to sell it. The patient has the right to seek another provider. He does not have the right to compel a sale.

    If all nonbeneficial care were excluded from coverage, this is precisely the situation we would be in. It would be no different than many other non-essential transactions. There are many professionals who probably refuse work in their field they find objectionable or unethical for one reason or another. A construction contractor might not wish to build a dungeon. A tax professional my not want to perform certain “borderline” tax shelter maneuvers. I do not wish to sell my hair.

    The customer is only always right if you want to make the sale. I see no reason why a physician must make every sale and especially when the intervention is not beneficial to health.

    UPDATE: I see there is some confusion over what question I’m asking/addressing in this post. I can put it more simply this way: If a physician (reasonably) believes a treatment to be nonbeneficial, must she provide it anyway on request by a patient? This is a slightly different question than the authors I quoted address, and it has nothing to do with the precise treatment or the characteristics of the patient beyond those that inform whether the treatment is or is not beneficial in the mind of the physician.


    • As I read your post, a Berwick passage immediately came to mind, and one I will never forget. Its from a particularly memorable piece that should be essential reading for all. Hardly something you would thinks emanated from our ex-CMS chief:

      In this form of truly patient-centered design, many, if not most, classically trained health care professionals will find cause for alarm. Let me anticipate three objections.

      Evidence-based medicine sometimes must take a back seat.
      First, leaving choice ultimately up to the patient and family means that evidence-based medicine may sometimes take a back seat. One e-mail correspondent asked me, “Should patient ‘wants’ override professional judgment about whether an MRI is needed?” My answer is, basically, “Yes.” On the whole, I prefer that we take the risk of overuse along with the burden of giving real meaning to the phrase “a fully informed patient.” I contemplate in this a mature dialogue, in which an informed professional engages in a full conversation about why he or she—the professional—disagrees with a patient’s choice. If, over time, a pattern emerges of scientifically unwise or unsubstantiated choices—like lots and lots of patients’ choosing scientifically needless MRIs—then we should seek to improve our messages, instructions, educational processes, and dialogue to understand and seek to remedy the mismatch. For the same reason, I wish we would abandon the word “noncompliance.” In failing to abide by our advice or the technical evidence, the patient is telling us something that we need to hear and learn from. Honestly, how many of us have ever faithfully taken a full ten-day course of a prescribed antibiotic or never consciously skipped a statin dose? Are we fools who did that? Or did we choose that because of some sensible, local considerations of balance, convenience, or even symptom information that the doctor never had?

      I can imagine just as easily as my critics can a crazy patient request—one so clearly unreasonable that it is time to say, “No.” A purely foolish, crazy, or venal patient “want” should be declined. But my wife, a lawyer, told me long ago the aphorism in her field: “Hard cases make bad law.” So it is in medicine: “Exceptional cases make bad rules.” You do not successfully rebut my plea for extreme patient-centeredness by telling me that, on rare occasions, we ought to say, “No.” I say, “Your ‘rare occasions’ make for very bad rules for the usual occasions.”


      • “For the same reason, I wish we would abandon the word ‘noncompliance’. In failing to abide by our advice or the technical evidence, the patient is telling us something that we need to hear and learn from. ”

        I think it’s incorrect to think assume that the patient is ‘noncompliant’ for intentional/self-interested reasons. In Baicker et al.’s article about “Choice Hazard in Health Insurance” she wrote:
        Finally, the observed reactions to “nudges” further undermine the credibility of the standard rational model applying. If people are simply rationally trading off current costs and future benefits, then a text message reminder should not produce a substantial change in adherence – but there is ample evidence that such nudges do produce substantial changes, especially in the case of the management of chronic diseases where neither the symptoms of the disease nor the immediate consequences of failure to comply with recommended care are salient. For example, adherence to statins is far from perfect and is responsive to nudges and small copayment changes. Schedlbauer et al. (2010) note that poor patient adherence “is a major factor in the lack of success in treating hyperlipidaemia.”
        They find that nudges focusing on reminders and reinforcements were particularly promising, with four out of six trials reviewed producing significant increases in adherence ranging
        from 6-24 percentage points. Similarly, simplifying the dosing schedules for blood pressure medication can lead to a 10-20% increase in adherence (Schroeder et al., 2004) and text reminders
        increased adherence to asthma medication by almost 20% (Strandbygaard et al., 2010). These are all medications with known but hard-to-observe efficacy. Nudges may be particularly effective in
        cases where there is more active resistence to treatments whose costs are particularly salient: reminders increase compliance with mammography by 4 percentage points among high-risk patients
        for whom the procedure had particularly high probability of health benefit (Parkington, 2009; DeFrank et al., 2009), while on the flip side, small copayments reduced mammography rates by more
        than 9% (Solanski, 2000). ”

        A great deal (probably the majority…) of noncompliance has to do with making the imperative of correct care not salient enough for the patient, and the ease of following correct care not simple enough for the patient.

    • I am not seeing why the MRI is characterized throughout the post as nonbeneficial. Presumably if the concern were that the MRI would actually be harmful, then “nonbeneficial” wouldn’t be the right word. So I take it harm is not the concern. But then is the MRI really nonbeneficial? Wouldn’t it help rule out (or maybe even confirm) cancerous activity in the lower back? Isn’t it a benefit to have this information? The reasoning offered for thinking it’s nonbeneficial was that it would’t change initial treatment anyway. But it surely would have if cancer were found, right?

      I’m not saying this should change your ultimate conclusion. Maybe yes, maybe no. But it does seem relevant whether a demanded service is marginally beneficial, nonbeneficial, or (most obviously) harmful.

    • Seems to me the issue is trust. The patient doesn’t trust the doctor’s clinical judgement. The doctor needs to address that or suggest the patient see a doctor he can trust.

      I don’t see why a patient willing to pay out of pocket to get a picture of his back shouldn’t be able to do it, as long as he isn’t causing patients with a real need to be delayed.

      • “I don’t see why a patient willing to pay out of pocket to get a picture of his back shouldn’t be able to do it, as long as he isn’t causing patients with a real need to be delayed.”

        Note that my post is consistent with this view. My point is that the physician should not be obligated to practice in a way he or she finds objectionable, provided he or she is not causing patient harm.

    • “My point is that the physician should not be obligated to practice in a way he or she finds objectionable, provided he or she is not causing patient harm.”

      Strictly speaking, this can’t be right. Suppose a physician is quite happy to provide requested nonbeneficial care–except to racial minorities. I suspect you would not support her right to refuse to treat black patients, which she finds objectionable, even if in doing so she was not harming them (medically).

      • As I hope is clear from the post, I am thinking in the dimension of “Do I provide this service or not (to patients for whom it would provide no medical benefit)?” I can’t qualify every sentence.

        But, going with what you raised, here’s the general, and interesting, question (to me): What are one’s obligations as a potential seller of a good or service? Contrast those with the freedom we normally afford to the potential purchaser. Usually it would not be regarded as odd if the purchaser did not wish transact with a class of individuals (e.g., “I don’t buy from white people.”). But if the potential seller made the same restriction (“I don’t sell to white people.”) we have a problem. There seems to be an asymmetry in what many suggest should be entered into completely voluntarily. Why?

        Now, I may have a problem with sellers or buyers who hold such discriminatory views, and therefore I may choose to not transact with them (raising the very same issue sketched above, discriminating by class, but now of a different sort). That’s my choice. Does that imply a moral claim? Not sure.

    • Not sure which of my comments your first paragraph responds to… If it’s to my comment about “nonbeneficial” services, I would just reiterate that I do think it’s important to know whether it is only nonbeneficial services that can be denied or whether (very marginally) beneficial services can too. I don’t want the harder issue (of refusing beneficial services) to be swept under the rug by stipulating (the potential falsehood) that the MRI was nonbeneficial.

      If you are responding to my thought experiment involving the racist, I don’t think I was being unfair by unreasonably requiring you to qualify every sentence. When you raised the thought experiment of the construction worker who refuses to build a dungeon, I took that as a signal that you were interested in the nuances of this tough issue. In light of your dungeon-example, I don’t think my racist-example comes out of left field.

      As to the asymmetry you raise: I agree, that’s something to think more about. I don’t have a good answer.

      • Sheesh! Who said anything about unfair? I was just clarifying what I thought was clear. Then I actually address your concerns, though broaden the question, which you seem to think is worth thinking about. However, I did not say that a contractor may not want to design a dungeon for black people.

        I didn’t notice your comment about MRIs. As to that, I was abstracting from the specific example to a thought experiment about a situation in which the physician believed the requested service was nonbeneficial. Under what circumstances should we insist that a professional provide a service like that? Why or when must the potential seller substitute his values for the potential buyer’s? Only in medical care? Never? Sometimes? When?

        UPDATE: I realized that you, Paul, are making a sale/no sale distinction on the basis of the customer’s characteristics (black, not black). Though I didn’t state it explicitly, my intention was to make a distinction based on the nature of the product (potentially) for sale (medically beneficial, not medically beneficial). Beyond the question of benefit to the patient (which may vary by patient characteristics), patient factors are not relevant. The point of raising the contractor or the tax lawyer was just to point out that other professionals may make transaction decisions based on ethical considerations. My assertion is that, similarly, a doctor who feels providing nonbeneficial treatment is unethical may reasonably refuse to do so. But this has nothing to do with patient characteristics beyond those relevant to the benefit of the treatment. So, now I do think you’ve raised something out of bounds. It may still be a worthwhile and interesting question, but it is a different one and not one I was raising. Hope that’s clear.

    • This is indeed a tricky issue, since it hinges on deeply held beliefs about the patient/physician relationship. But I think it’s actually much more difficult than this discussion indicates, because the example is a little artificial: we have pretty solid evidence that using an MRI for nonspecific lower back is basically useless, but it’s relatively rare when things are so straightforward. What should the physician do when there isn’t particularly strong evidence one way or the other, or when there’s just no evidence at all? And what happens if the evidence indicates that the treatment is indeed beneficial, but perhaps not enough to justify the costs? I propose that these situations arise much more often in daily clinical practice than the clear-cut example of an MRI for lower back pain.

      Physicians and other providers operate in a culture in which their primary duty is to do everything they can to help the patient. Asking them to balance that duty to the patient with a duty to society is a big, big change: now there are three actors in every consultation, the doc, the patient, and Society, and it’s not really clear to the patient whose side the doc is on. I’m not saying that it’s unreasonable to ask providers to balance patients’ demands with societal costs, but we should recognize that such an approach constitutes a deep and radical shift in the basic philosophy of care.

      • If you read all the comments, I think, in fact, the discussion is not as complex as you make it out to be. That’s not to say you don’t raise important (and complex) issues. You do! But I am trying to stick to a simpler one, as articulated in the post and, perhaps better, in some of my comments.

    • “I did not say that a contractor may not want to design a dungeon for black people.”

      Didn’t suggest you did. I introduced race. And only as a counterexample to your claim that “The physician should not be obligated to practice in a way he or she finds objectionable, provided he or she is not causing patient harm.” As I say, it only introduces some nuance (that many might not wish to entertain). But who knows, maybe examining this nuance will lead to many more non-race-related examples where it is not permissible for a physician to refuse nonbeneficial services. I haven’t thought enough about it, but I wanted to flag the possibility.

      Re “nonbeneficial,” I raise it because I personally am always on guard against those who would accuse me pulling a fast one by classifying marginally beneficial services as nonbeneficial. If MRIs of the sort you write about in the post are in fact nonbeneficial, then I’d be glad to learn that. It might make it easier to argue against those who think people are entitled to those MRIs if they want them. It’s harder (but not impossible) to argue against them when the MRI they request might do them some good.

      • I speculate you wrote this before I wrote my update. I think I’m addressing a different question and one pure (hypothetical) enough to abstract from MRIs or any specific treatment. This is really about what a physician (reasonably) believes about a treatment and whether she must provide it independent of that belief when requested by a patient. I’d like to stick to that issue.

    • Austin-

      Thank you for another provocative post. I agree with your concluding sentence: “The customer is only always right if you want to make the sale. I see no reason why a physician must make every sale and especially when the intervention is not beneficial to health.”

      The clinical vignette you quote is even more interesting because, in the case of LBP & MRI imaging, there is an additional concern, as yet unmentioned in the discussion. The concept I will describe is not easily understood, requires careful explanation, frequently leads to more questions, and thus a non-trivial amount of time.

      A false-positive finding is the largest risk from an MRI for a patient with acute, non-complicated LBP. Numerous studies have demonstrated that MRI’s are overly sensitive and unable to differentiate between a painful pathology and non-painful age-related changes (here, here, here, & here for a brief intro into this topic).

      Approximately 30% of individuals having an MRI, with or without a painful back, will have a potentially painful ‘abnormality’. The suffering individual is likely to seek further treatment based on the finding.
      Invasive care for LBP can include various injections and surgical procedures, each of which incurs some risk. For the properly diagnoses patient, the risks can be balanced by the potential benefits.

      In the case of a patient with a false-positive MRI finding, there can only be risks.

      Explaining this to a patient in pain and seeking help is not easy. Nor is it satisfying for either of us. I like to be helpful and my explanation is not likely to be viewed as any form of relief. My thoughtful, evidence-based approach takes up more of my time, delays focusing my attention on my next patient, and will not be considered billable work. I will miss any additional revenue from procedures that may have resulted following the imaging study.

      Finally, the unsatisfied patient is likely to take his painful back down the street to my competitor who may not be willing to pass up this revenue stream.

      This vignette is not isolated to spine care. Throughout healthcare, testing and treatment procedures are incentivised while discussion and explanation are not.

      A patriotic desire to save our country from financial ruin is the only incentive for me to have this difficult conversation. Every other incentive is perfectly aligned for me to do the MRI.

      Reason # 17 why I went into health policy.


      • In 1995, I was one of those low-back pain patients. I had severe sciatica, actually. I could barely walk. I was 23. I received an MRI. My orthopedist showed me the bulging disk. Then he very kindly told me that I didn’t require surgery. I could do PT and load up on anti-inflammatory drugs. It would take time, but I would get better. I am not kidding when I say that it took about five years to fully recover. I worked very hard and had many relapses. The journey was worth it.

        If that doctor had told me I needed surgery, I might have believed him. To date, that may have been the most important interaction with a physician I ever had.

        PS: He trained in the UK.

        • Part of my training was in NZL where there were no MRI’s until 1997, one year prior to my stint there. Patients suffering from LBP were not suffering from under-treated LBP.

          Your story is very common. In 15+ yrs of clinical practice I have met 2 patients who wished they had surgery. Most are like you. I have met many patients who are unhappy with their decision to have surgery. This is a biased sample because those that are happy have surgery did not come to see me….

          For the record, the most surgeries on a single patient with no congenital malformations during my clinical career – 23. She was 31 when we met.

          I’m glad you are doing well. Thanks, again, for all you do with your blog. Excellent work…

    • I skimmed through the comments, and I don’t think this was mentioned. You turn things around for buyers and sellers of goods and mentioned the asymmetry in their buying biases, but there’s also asymmetry in the health care arena.

      For example, if a doctor believes the patient should undergo a procedure, but the patient doesn’t want it, of course, the doctor can’t force him to, and this is not even debated by anyone (except maybe Dr. House).