Choosing Wisely

Abstract:

The American College of Physicians in its ethics manual endorsed the idea that physicians ought to improve their ability to provide care to their patients more parsimoniously. This elicited a critical backlash; critics essentially claimed that what was being endorsed was a renamed form of rationing. In a recent article, Tilburt and Cassel argued that parsimonious care and rationing are ethically distinct practices. In this essay I critically assess that claim. I argue that in practice there is considerable overlap between what they term parsimonious care and what they define as rationing. The same is true of the Choosing Wisely campaign endorsed by the American Board of Internal Medicine. In both cases, if the goal is to control healthcare costs by reducing the use of marginally beneficial care that is not cost effective, then a public conversation about the justness of specific choices is essential.

The following passage from the ethics manual of the American College of Physicians triggered something of a public and professional firestorm in the United States. “Physicians have a responsibility to practice effective and efficient health care and to use health care resources responsibly. Parsimonious care that utilizes the most efficient means to effectively diagnose a condition and treat a patient respects the need to use resources wisely and to help ensure that resources are equitably available.”¹ In the English language (for the benefit of my international readers) the term “parsimonious” typically carries a negative connotation that suggests being stingy. In a medical context that suggests skimping on needed care for the sake of saving money.

Patients may freely choose parsimonious care for themselves; and, if such a choice is not a product of impoverished financial circumstances, it will generally not be seen as being morally problematic. But if physicians choose to provide parsimonious care, and if physicians are supposed to be trusted advocates for the best interests of their patients, this will be presumptively morally problematic. To be sure, this was not the intent of the authors of this passage. They were advocating for a more prudent, less lavish, more judicious use of healthcare resources that would still yield effective, high-quality healthcare. Their ultimate intent was presumptively morally laudable—namely, a more judicious use of healthcare resources so that the savings thereby generated would result in a more equitable distribution of needed and effective healthcare. Still, the obvious implication of this directive is that some patients would get less care than they might otherwise have received in an environment unconstrained by the need for a more parsimonious approach to providing healthcare. Thus it looks like parsimonious care is really a form of healthcare rationing, even if the ultimate goal is more equitable access to needed healthcare. Proponents of parsimonious care have been quick to contend that it is not some alternate language for healthcare rationing. On the contrary, their contention is that the two practices are entirely distinct from each other.² The purpose of this essay is to critically assess that contention.

Is Parsimonious Care Subtle Rationing?

Tilburt and Cassel want to draw a sharp distinction between parsimonious medicine and rationing. I concede that the two notions are not synonymous with each other, nor do they perfectly overlap with each other in practice. But I argue that there is considerable overlap in practice between what they would describe as parsimonious care and what others would characterize as instances of healthcare rationing. Tilburt and Cassel define rationing as “explicit or implicit withholding and allocation of beneficial resources for some patients for the benefit of others.”³ They point out that the ethical rationale for rationing rests on concerns regarding distributive justice. They contend, however, that parsimonious care “means delivering appropriate health care that fits the needs and circumstances of patients and that actively avoids wasteful care—care that does not benefit patients.”⁴ In other words, if care is not going to benefit a patient, then it is really wasteful. And if it is really wasteful, there can be no disagreement about whether or not that patient has a just claim to that care. No one has a just claim to care that is wasteful. Further, if this is an accurate characterization of parsimonious care, then physicians cannot be justly accused of being less than loyal advocates of the best interests of their patients because patients cannot have a reasonable interest in nonbeneficial care.

Why are Tilburt and Cassel so insistent that there must be this sharp ethical distinction between parsimonious care and healthcare rationing? The short answer is that there is considerable political pressure in the United States to control healthcare costs (this is largely true in European countries as well, even though those countries spend a much smaller fraction of their GDP on healthcare than does the United States). In 2014 healthcare spending in the United States reached $3.1 trillion, roughly 17.5% of the GDP. More troubling are projections to 2024, when it is predicted that we will be spending $5.4 trillion on healthcare, or about 19.6% of the projected GDP.⁵ The problem is that roughly 80% of those healthcare dollars are “allocated” by physicians because only physicians have the legal authority to authorize surgery for various medical conditions, or certain costly drugs, or some number of days in the hospital, or home care, or rehabilitation, and so on. Consequently, if there is going to be a focal point for controlling healthcare spending, it will be physicians and their judgment of what constitutes genuine medical need and the care that is necessary to address that need effectively.⁶

If physicians are the real allocators of healthcare resources, and if healthcare spending must be controlled, then physicians are at risk of having to do “bedside rationing,” which is viewed by Tilburt and Cassel as ethically problematic because physicians are compromising their commitment to the best interests of their patients. However, if the goal of adequate control of healthcare spending can be achieved by simply making parsimonious (nonrationing) choices, then neither the integrity of physicians nor the just claims of patients would be at risk of being compromised. Tilburt and Cassel see the Choosing Wisely campaign as the professionally appropriate mechanism for achieving a more parsimonious healthcare system without the moral risks of rationing.⁷

What Counts as “Waste”?

How should we correctly identify what should count as “wasteful” healthcare? Here is the story of Mr. Krieger, whose last 10 days of life cost $323,000. He was 88 years old and was in the late stages of Alzheimer’s, with brittle bones and a serious cardiac condition. Three months before his death he broke his hip in a fall and it was surgically repaired. Ten days before his death his daughter found him shaking, dehydrated, and speaking gibberish. She had him rushed to the emergency room, where he was diagnosed with septicemia. He suffered a large drop in blood pressure; was placed on a vent, transferred to the ICU, given very powerful antibiotics, diagnosed with necrotizing fasciitis, and given Linezolid (an extraordinarily expensive drug) and immunoglobulin; and surgery was discussed. Though his daughter had durable power of attorney for health and knew from years earlier his preference for nontreatment, she was quoted as saying, “This was the man who gave me life. Who was I to summon his death?”⁸

But for our purposes the question we want to ask is whether any of the care provided to Mr. Krieger was wasteful or medically unnecessary. It seems that everything done to Mr. Krieger was medically necessary, at least in the sense that these were things physicians would typically do for a patient who was septic with life-threatening hypotension. Hence, in that respect the care provided to him was not wasteful. After all, it did provide him with 10 extra days of life (which may be regarded as a prima facie medical benefit). So, could an advocate for parsimonious care deny Mr. Krieger any of the $323,000 worth of care he received?

It might be argued that Mr. Krieger himself would have refused all this care, had he been conscious and sufficiently competent, because he would not have regarded it as a net benefit. He might have regarded it as wasteful of social resources as well, which would seem to justify a parsimonious judgment by his physicians. However, imagine a Mr. Krueger whose medical condition is just like Mr. Krieger, except that he is in the early stages of dementia. He comes into the emergency room conscious and competent, and demanding everything that might prolong his life, because he does not want to die. Could his physicians deny him such care on the basis of parsimony when he (Mr. Krueger) would regard such care as potentially beneficial for him (even if he gets only 10 extra days of life)? If the legitimacy of what counts as a medical benefit depends on the subjective judgment of the patient, that takes judgments of parsimony out of the hands of physicians (and undermines presumptively just efforts to control healthcare costs). But if physicians are the ultimate judges of what counts as wasteful, nonbeneficial healthcare, then they would have the moral authority to deny both Mr. Krieger and Mr. Krueger that $323,000 worth of care. However, such a judgment could be correctly described as being as much a matter of rationing as parsimonious care. The sharp distinction that Tilburt and Cassel insist on between these two sorts of practices would effectively have disappeared.

Parsimony and Diagnostic Uncertainty

It might be argued that one example is not sufficient to undermine the distinction that Tilburt and Cassel wish to defend. But then there are the challenges of diagnostic uncertainty associated with very complex presentations of symptoms. What is a loyal but parsimonious physician supposed to do with regard to diagnostic testing in the face of potentially misleading or ambiguous diagnostic clues? Should a parsimonious physician cease diagnostic testing when she is 80% confident that she has a correct diagnosis? Should she continue testing until she is 90% certain, or 95% certain, or 99% certain? At what point does diagnostic testing become wasteful and inefficient—that is, something that a parsimonious physician could justifiably forego (even though such delay in some small number of cases could have disastrous consequences for a patient)?

Again, only a small risk of serious harm to the patient might be associated with settling for 95% diagnostic certainty, but that still represents a denial of potential (albeit marginal) benefit to the patient. Then that looks more like rationing rather than an unequivocally pure parsimonious choice. No doubt some physicians are promiscuous users of diagnostic tests, often failing to get any useful information from many of the tests that they order. Tilburt and Cassel would say these are practices that should be ethically curtailed because they are wasteful and inefficient. No doubt there are clear circumstances in which this is sometimes true. But it is also the case that it is only after the fact that physicians often know that a particular diagnostic test yielded no useful information. Making thoughtful, ethically acceptable parsimonious judgments before the fact may be more difficult in most cases than Tilburt and Cassel would concede.

Parsimony and Therapeutic Uncertainty

In addition to the problem of diagnostic uncertainty there is the problem of therapeutic uncertainty. In the earlier passage quoted from the American College of Physicians, parsimonious care is defined as care that will “efficiently and effectively” treat the patient’s condition. But what exactly does “effectively treat” mean when the patient being cared for has a chronic degenerative condition (congestive heart failure or metastatic cancer or type 2 diabetes) that is not going to “get better”? Does it mean “try anything and everything that might work,” no matter how small the benefit, no matter how unlikely the benefit, no matter how large the cost? That would not seem to fit any reasonable understanding of what is meant by parsimonious care. But if patients are going to be denied extremely costly care that has only a small probability of yielding more than a marginal benefit on grounds of parsimony, then that looks very much like what Tilburt and Cassel would describe as rationing, because patients are being denied a benefit.

To be fair, we cannot conclude that all instances of what Tilburt and Cassel would refer to as parsimonious care are also instances of healthcare rationing. There are examples of purely parsimonious care that fit what they are advocating for. For example, if an individual has had a radical prostatectomy for his prostate cancer, follow-up care will usually include radiation treatment to kill any remaining cancer cells. The two most effective forms of radiation treatment for this condition are intensity modulated radiation therapy (IMRT) or proton beam therapy. The latter costs roughly twice as much as the former ($36,000 vs. $18,000) for exactly the same outcome in terms of quality of life and life prolongation.⁹ Assuming that the research that justifies this conclusion is correct, that means there is no additional benefit that is gained for that extra $18,000. This would represent an instance of pure waste. Providing a patient with IMRT instead of proton beam therapy (which the patient might want) would be a clear example of a purely parsimonious judgment. Given Tilburt and Cassel’s definition of rationing, it would not be correctly described as rationing. The problem, however, is the paucity of such clear examples of parsimonious care in the real world of medicine.

Berwick and Hackbarth would disagree with this last conclusion. They claim that 25% of all healthcare spending in the United States ($750 billion) is wasteful.¹⁰ They include in that figure administrative complexity, fraud and abuse, and pricing failures. We need to pass over those items. Instead, we focus on their claim that $158 to $226 billion of that total represents the wasteful overtreatment of patients, in particular intensive care at the end of life. However, as the vast majority of physicians who work in intensive care would point out, considerable uncertainty exists regarding which patients are likely to die during their admission to the ICU.¹¹ Again, after the fact it is easy to judge that it was not worth it to have invested so many life-prolonging resources in a particular patient, because he or she died anyway. Before the fact physicians will often think to themselves that they would not be surprised if this patient were to die in the ICU, but some of those patients rally and survive another few months. Many of those patients will see that additional measure of life as a benefit, albeit a not-unqualified benefit. If those patients were denied those marginal gains in life expectancy, that would have to be described as a rationing decision, not a parsimonious choice. Under the right circumstances such decisions would be just and reasonable, but it would certainly not be a matter of eliminating waste from the healthcare system. That would be a disingenuous conclusion intended to hide from more careful moral scrutiny the question of whether such decisions in particular circumstances were right and just.

Gilbert Welch, a physician, in his book Overdiagnosed: Making People Sick in the Pursuit of Health calls our attention to yet another problem for defenders of parsimonious care.¹² His primary message is that all sorts of standards for being “ill”—having a healthcare need—have been altered over the past several decades; for example, the total cholesterol standards for deciding when patients should be prescribed statins have been lowered in recent years. If someone has a total cholesterol level of 350 mg/dL, his or her mortality and morbidity risks will be significantly reduced if he or she is prescribed statins. This is largely true for patients with a total cholesterol of 300 mg/dL. It becomes less true (but not false) for total cholesterol levels of 240, 200, and 180 mg/dL. This is again a matter of marginal benefits, not no benefit. The question we would raise for Tilburt and Cassel is this: If physicians no longer offered statins to patients with cholesterol levels below 220 mg/dL, would that be properly described as a matter of parsimonious care? Or should such choices be thought of as instances of rationing (because patients would not be offered the opportunity to access that marginal benefit)?

In 2015 a new class of drugs for lowering LDL, the so-called bad cholesterol, received FDA approval. They had been referred to as PCSK9 inhibitors (alirocumab and evolocumab). LDL levels above 100 mg/dL are generally a point at which a patient would be offered an LDL-lowering statin. These statins are able to lower LDL by about 17%, but the PCSK9s are able to reduce LDL by as much as 70%, often to levels of 50 or 60 mg/dL.¹³ No one can doubt that this represents a dramatic improvement in lowering LDL levels. Roughly 75 million Americans have LDL levels above 100 mg/dL. In theory they would all be candidates for these drugs. But these drugs have been priced at $14,000 per year, and patients would have to be on these drugs for the rest of their lives. If these drugs were offered only to the patients who could benefit the most, about 11 million Americans would qualify as candidates. That would add $150 billion per year to healthcare costs in America. Given this background information, what would we imagine should be the appropriate response of parsimonious physicians concerned about the prudent and equitable expenditure of limited healthcare resources?

If Tilburt-like physicians were unwilling to deny access to these drugs to the vast majority of patients with very elevated levels of LDL because this would represent a form of rationing, then the consequences would appear to be both imprudent and inequitable, given the need to control healthcare costs. On the other hand, if they were to say (for parsimonious reasons) that some of the current statins are good enough in reducing LDL, then they would be giving up possible benefits for their cardiac patients, which would be fairly described as a rationing decision. The strategy that is not available to them is to say that this situation is comparable to the proton beam therapy–versus–IMRT situation for prostate cancer. The research cited makes abundantly clear that PCSK9s are much more effective in reducing LDL levels than any current statins. What is unclear (because longer trials are needed) is whether reductions in mortality and morbidity are comparable to the very large reductions in LDL achieved by these drugs.

Parsimony and Choosing Wisely

Though Tilburt and Cassel insist there is a very sharp distinction between parsimonious care and healthcare rationing, they muddle that distinction when they endorse the Choosing Wisely campaign as an example of physicians embracing parsimonious care as a morally legitimate part of medical practice. The reason for this is that the Choosing Wisely campaign is about drastically reducing “low-value” care, which is not the same as no-value care—care that is wasteful and inefficient. Low-value care would largely include what is referred to as “marginally beneficial” care—therapies that tend to yield only a low likelihood of benefit at a very high cost.¹⁴ This is precisely what Tilburt and Cassel had associated with rationing.

The goal of the Choosing Wisely campaign is to identify diagnostic or therapeutic interventions that are substantially overused and that tend to yield relatively little benefit. Moriates et al., for example, call attention to the use of computed tomography angiography (CTA) for diagnosing a pulmonary embolism (PE), which can be life threatening. As they note, a PE can be difficult to diagnose on the basis of symptoms alone. In that respect CTA represents a very useful diagnostic technology. As a consequence of the availability of that technology, over an eight-year period of time the diagnosis of PE nearly doubled. “However, despite these rapidly increasing number of PEs found and treated, the effect on mortality was very small.”¹⁵ What should a morally sensitive, parsimonious physician do under these circumstances? Should clinical practice revert to a pre-CTA style of practice? I doubt anyone would advocate that. Elshaug et al. note, “Clinical heterogeneity makes it difficult to develop simple approaches for identifying low-value care. Services that are ineffective, unsafe, or both for all patients and all indications are rare.”¹⁶ The point of the passage is that identifying low-value care before the fact in specific clinical circumstances is often very difficult. Our example of using CTA to identify PEs perfectly illustrates that point.

This last example might be very frustrating for physicians who wish to use healthcare resources wisely. But even more challenging, morally speaking, is when we have the ability to distinguish high-value from low-value care in specific clinical circumstances but physicians (or society) lack the moral courage or moral legitimacy that would allow them to act on the basis of that knowledge.

A recent article in the Wall Street Journal told the stories of nine patients who achieved remarkable life prolongation (4–9 years) as a result of novel immunotherapies for their cancer.¹⁷ These patients, typically a small percentage of cancer patients, are referred to as “super responders.” The obvious research question is, What made them so therapeutically distinctive that they were able to achieve this sort of success when other patients with the “same” cancer may have gained only extra weeks or extra months of life? The generic answer to that question seems to be that genetic features of their cancer or their personal genome likely explain that success, though researchers for the most part do not know right now what those features might be. But that just raises the ethical question of whether we ought to spend substantial research dollars to discover those answers, which in turn raises the additional ethical question of whether we would expect oncologists to deny their patients these $100,000 cancer drugs if they knew before the fact that genetic features of their cancer confidently predicted they would be low responders (less than six months of life gained) rather than high responders (more than two years of life gained).¹⁸

Illustrative of this last point is one of my colleagues, Randy Hillard, a psychiatrist who was diagnosed with stomach cancer in 2010 and was given a life expectancy of one year.¹⁹ But his cancer was genetically analyzed and found to be HER2-positive. That meant he was a candidate for trastuzumab, which targets HER2. He has received an infusion of that drug every three weeks for the past five and a half years, and he feels fine for the most part. He estimates his cancer care has cost about $1 million.

Roughly 20% of patients with Hillard’s cancer are HER2-positive, though many of them will receive no benefit or very little benefit from trastuzumab. A likely reason for this less beneficial outcome is that their tumors are genetically complex, which is to say that their cancers harbor multiple genetic mutations. Trastuzumab may kill those cancer cells driven by HER2 but then allow other genetic drivers of that cancer to become dominant. This is what is known as cancer drug resistance. If Hillard’s cancer does not have multiple genetic drivers, and if others with stomach cancer can be identified as being equally genetically fortunate, then they would have a presumptive, just claim to this therapy, even though it is high cost, because it is also high value.²⁰ For the rest of those patients with advanced stomach cancer, their only treatment option would be ramucirumab, which costs $15,500 per month and yields median gains in life expectancy of 6–12 weeks. This may be fairly characterized as low-value care. If that is low-value care, then what should be the response of parsimonious physicians committed to making wise choices? This is a rhetorical question because, as things are now, physicians (certainly oncologists) are not going to deny these patients access to ramucirumab. This would be too clearly an instance of rationing.

Rhetoric, Rationing, and Medical Reality

Peter Ubel assessed the article by Tilburt and Cassel and came to the same conclusion for which I have argued: namely, that in reality there is considerable overlap between choices that would reflect parsimonious, low-value care judgments and rationing.²¹ However, instead of concluding that this bit of linguistic legerdemain might be ethically problematic (as I argue), he contends that it was the smart thing to do. He notes (correctly) that the language of rationing and cost-effectiveness is so negatively emotionally laden that it is impossible to have a productive, rational conversation about healthcare cost containment if those words are part of the conversation. He prefers instead to speak of parsimonious care and “financial stewardship.”²² If these linguistic adjustments were all that was necessary to have more productive social conversations about making painful, ethically controversial rationing decisions, then it would be petty of me to object. However, the problem that I see is that these linguistic changes have the morally controversial effect of obviating the need for these painful social conversations, thereby making rationing practices socially invisible, unavailable for critical ethical scrutiny.

We noted earlier that Berwick and Hackbarth asserted that there was $750 billion worth of waste and inefficiency annually in the U.S. healthcare system. If there is all that “waste and inefficiency” in our healthcare system, there is no need to get embroiled in difficult and divisive societal conversations about healthcare justice and rationing. Justice concerns are not at stake if all our efforts at cost containment are about waste and inefficiency. No moral arguments of any kind would support protecting waste and inefficiency when the domain of healthcare needs is expanding and demanding resources to meet those needs. Nor would there be any need for irate political argument. However, the fact of the matter is that what some economists might identify as wasteful and inefficient (non-cost-effective) healthcare may be viewed by patients needing that care as necessary, life-sustaining care. Those patients may not have a just claim to that care. Certainly that is arguable in the case of Mr. Krieger. But justifying that conclusion requires a separate conversation about healthcare justice.

That a certain type of care in specific clinical circumstances is either cost effective or not does not by itself settle the question of whether patients needing such care have a just claim to that care. The same is true when care is correctly described as being either high value or low value. When societal funders of healthcare decide that they will not pay for a $40,000 implantable cardioverter defibrillator for an 80-year-old patient in the advanced stages of Alzheimer’s with a potentially life-threatening arrhythmia because that would be “wasteful,” they are effectively closing off any discussion of whether such a decision is just.

Calabresi and Bobbitt have described such situations as “tragic choices” because any decision we might make will be a violation of a deep societal value.²³ The proliferation of very costly life-prolonging technologies in our society makes such tragic choices ubiquitous and the opportunities for divisive and destructive societal argument endless. Their recommendation for addressing such tragic choices is political dishonesty, finding clever ways to hide from social visibility such choices. The language of wastefulness and inefficiency accomplishes that objective when it comes to healthcare cost containment. The introduction of diagnosis-related groupings (DRGs) in 1984 as a payment mechanism for hospitals in the Medicare program also has the same effect.²⁴ DRGs represent a fixed budget given to a hospital for a patient with a specific diagnosis. The implicit message to hospitals was that too much care (too many hospital days) was being lavished on patients, and this was wasteful. This was largely true. But equally true (but not publicly spoken) was that some small number of patients might not have suffered avoidable morbidity and mortality if they had been allowed extra days in the hospital. Those tragic outcomes might often be just and reasonable. But no one (certainly not patients affected by such choices) could confidently know that, because those conversations never occurred at a socially visible level. Instead, decisions were made invisibly for unknown reasons in the bowels of hospital bureaucracy. This is presumptively unjust. Just choices and their justifying reasons ought to be transparent.

In my own work I have argued that a fair, reasonable, and responsible approach to healthcare rationing can be carried out through inclusive processes of rational democratic deliberation.²⁵ What we seek to achieve through such processes are shared social understandings of what rationing protocols we (all of us) are willing to impose on our future possible selves in specific clinical circumstances because we judge these protocols to be just enough. What we want in any society as the core of justice is reciprocity and fair terms of cooperation. If I want Medicare to save my taxpayer dollars by not providing $40,000 defibrillators to patients in the advanced stages of Alzheimer’s who are strangers to me, then I must endorse my being treated that way should I find myself in those circumstances in the future.

Perhaps Ubel is correct, and we ought to be more parsimonious in our use of the term “rationing” in these public discussions. We can talk instead about the need to make more prudent, parsimonious choices. But if we do that, then we must be absolutely clear and explicit that efforts to control healthcare costs by trimming marginally beneficial care in a range of clinical circumstances will often result in sacrificing some “statistical” lives and various degrees of well-being for the sake of a more just and efficient healthcare system. Those outcomes may be tragic and socially painful to behold, but they must be beheld. They must be recognized. In the final analysis it needs to be the case, contrary to Tilburt and Cassel, that the ethic of rationing is the same as the ethic of parsimonious care. Finally, if various forms of parsimonious care have been endorsed as being just enough by future possible patients through a fair process of rational democratic deliberation, then physicians can in good conscience incorporate those parsimonious practices into their care and remain loyal advocates of their patients’ interests.