• The state of shared decision-making

    In a post last week I raised the issue of when shared decision-making is appropriate. Here I want to explore the state of state of shared decision-making (SDM). How well does it work? Is it the answer to the problems of over-, under-, and/or misuse? First, let’s get straight what SDM is. Maybe it means different things to different people, but the Institute of Medicine (IOM) yokes it to patient-centered care:

    The IOM defined patient centered care as “care that is respectful of and responsive to individual patient preferences, needs, and values” and that ensures “that patient values guide all clinical decisions.”

    That’s from a new NEJM Perspective piece by Michael Barry and Susan Edgman-Levitan. They go on to write,

    For some decisions, there is one clearly superior path, and patient preferences play little or no role — a fractured hip needs repair, acute appendicitis necessitates surgery, and bacterial meningitis requires antibiotics. For most medical decisions, however, more than one reasonable path forward exists (including the option of doing nothing, when appropriate), and different paths entail different combinations of possible therapeutic effects and side effects. Decisions about therapy for early stage breast cancer or prostate cancer, lipid-lowering medication for the primary prevention of coronary heart disease, and genetic and cancer screening tests are good examples. In such cases, patient involvement in decision making adds substantial value. […]

    The process by which the optimal decision may be reached for a patient at a fateful health crossroads is called shared decision making and involves, at minimum, a clinician and the patient, although other members of the health care team or friends and family members may be invited to participate. In shared decision making, both parties share information: the clinician offers options and describes their risks and benefits, and the patient expresses his or her preferences and values. Each participant is thus armed with a better understanding of the relevant factors and shares responsibility in the decision about how to proceed.

    It’s clear that shared decision-making leans heavily on patient education. I’m all for that, but how well does it work? A Cochrane review (pdf) of 86 trials of patient decision aids concluded that patients who use them

    a) improve their knowledge of the options; b) are helped to have more accurate expectations of possible benefits and harms; c) reach choices that are more consistent with their informed values; and d) participate more in decision making. Decision aids have a variable effect on actual choices but they reduce the choice of elective surgery when patients consider other options. When patients use decision aids, there appears to be a positive effect on communication with their health practitioner, and a variable effect on the time required for this consultation. Although there is an improvement with more detailed decision aids compared to simpler decision aid, the amount of improvement is smaller than that seen when decision aids are compared to usual care. There are no apparent adverse effects on health outcomes or satisfaction.

    However, decision aids are variable (pamphlets, videos, web-based tools) and variably applied. Though I don’t have any statistics handy, my guess is that a large proportion of decisions are made with nothing more than patient-physician conversations. In this messy, real-world, outside of a randomized trial, how well does this communication work, whether enhanced with a decision aid or not?

    Richard Hoffman and colleagues conducted a telephone survey in 2006-2007 that included 375 men who had undergone or discussed with providers PSA testing. Here’s what they learned:

    Overall, 69.9% of subjects discussed screening before making a testing decision, including 14.4% who were not tested. Health care providers most often (64.6%) raised the idea of screening, and 73.4% recommended PSA testing. Health care providers emphasized the pros of testing in 71.4% of discussions but infrequently addressed the cons (32.0%). Although 58.0% of subjects felt well-informed about PSA testing, 47.8% failed to correctly answer any of the 3 knowledge questions. Only 54.8% of subjects reported being asked for their screening preferences. An HCP [health care provider] recommendation (odds ratio, 2.67; 95% confidence interval, 1.08-6.58) was the only discussion characteristic associated with testing. Valuing HCP information was also associated with testing (odds ratio, 1.26; 95% confidence interval, 1.04-1.54).

    If one believes that patients should be making their own, well-informed decisions about the PSA test (and that itself is debatable), it’s hard to interpret this as anything but evidence of the need for the SDM paradigm to penetrate more deeply into the US health system. Clearly much more work must be done.

    In an editorial that accompanied publication of Hoffman et al.’s study, Steven Woolf and Alex Krist survey the considerable challenges facing more thorough implementation of SDM for the PSA test. They point to deep questions about what information should be communicated to patients, how it should be done so, and what metrics should be used to measure comprehension and progress of the SDM model. Beyond that, Woolf and Krist point out what is all too obvious to seasoned observers of the health care delivery system. There is tremendous inertia from top to bottom, both at the sharp end and among policymakers and managers. Even the best ideas can take years to develop and implement. Even terrible ideas take years or decades to extinguish. Woolf and Krist conclude,

    Without these efforts by researchers, health systems, public health leaders, payers, and the courts, SDM is unlikely to gain its footing in routine patient care. In the United States, where medical technologies are often adopted long before their effectiveness and safety are confirmed, the difficulties of implementing SDM for prostate cancer screening will likely recur with other modalities of care. What is ultimately required is a deeper change in culture among providers and consumers of health care to delay dissemination, resist the assumption that newer is better, wait for evidence, tolerate observation over intervention, and accept uncertainty.

    SDM may be an appropriate and important tool to improve the US health system. But, like just about any change to the health system, translating the vision into reality is a heavy lift.


    • “If one believes that patients should be making their own, well-informed decisions about the PSA test (and that itself is debatable)…”

      How is this debatable? I think I must be missing something, because the alternatives seem to be (1) coercive screening or (2) poorly-informed decisions, neither of which I suspect you are not proposing.

      • Some people, the members of the USPSTF among them, think that the PSA test should not be offered. That’s the sense in which it is arguable that patients should not be making their own decisions. Not being offered the option is pretty much the definition of uninformed (on the assumption most people would learn about health treatments from their doctor).

    • The problem with bringing the SDM framework to bear on PSA screening is that there is no easy inference from what we know about PSA to what any patient ought to do; physicians themselves cannot draw a pro- or con- conclusion (re the desirability of getting the test done) in any straightforward way from any patient’s expressed preferences, it seems to me. Its a lousy test (per its specificity and sensitivity); it leads to lots of biopsies. And pos results on those biopsies for an important range of gleason scores simply do not tell us whether therapy will extend life or not, while it is certain that therapy will lead to lots of complications.

      The SDM model more generally is, I think, a flawed attempt to paper over the reality that physicians shape the decisions that patients make to a great extent no matter what physicians do or say. I’m all for discussing options, etc, but I think a limited paternalism (offer information, persuade, do not coerce) is better than SDM both as a description of what physicians do and as a normative model for what we ought to do.

    • -Seems like it’d be reasonable to strike a balance between coercive and shared decision making via payment mechanisms. E.g. If you want something that doesn’t pass the cost-benefit test, you can have it, but you have to pay for all or part of it out of pocket.

      Having said that, Implementing those payment mechanisms in a manner that could be seamlessly integrated into discussions and decisions in a clinical setting would present its own challenges.

      For those of us in high-deductible plans, that’s already the default. If I’m understanding VBD correctly, then it’d be possible to engineer positive incentives into even high-deductible plans that would use positive incentives to promote practices that pass a given cost-benefit threshold.

    • I think that there are also cognitive issues here which are being ignored. There are very large groups of patients who just say “do whatever you think is best doc”. Many people do not know their own medical history well enough, or the implications of their illness to make anything near an informed decision.

      As a physician I think the best I can do in these kinds of situations is to get a general feel for the patient’s values or beliefs, then recommend care in line with what I perceive those to be. Not always easy to do.


    • The reality is that few doctors have the time to thoroughly discuss the options and pros and cons with a patient, unless it is for a major issue.

      You rarely get more than 15 minutes with your doctor and its far easier just to get the screening and put off the issue of what you will do if you have a problem.