Politicians: Keep Out

This past week, 150 Congressmen sent a letter to the Centers for Medicare and Medicaid Services (CMS), urging that it approve reimbursement for lung cancer screening with lowdose CT scans. These legislators are trying to exercise their political muscle in an arena where they have no business intervening. Medicare has a fair, transparent, and extremely thoughtful process for deciding what tests to cover. The attempt to destroy this honest, objective, and time-tested process by injecting political pressure is reprehensible.

The Medicare program is required by statute to limit coverage to tests and services that are “reasonable and necessary” for the treatment of illness.  Most “coverage decisions” continue to be made by local intermediaries, the regional contractors that function as Medicare’s agents throughout the country. But occasionally, for particularly important decisions, Medicare issues a National Coverage Determination which is then binding on all its intermediaries.  A few months ago, CMS was asked to make a decision about paying for lung cancer screening using low-dose CT scans. It has diligently been conducting a thorough, comprehensive assessment.

What Medicare often does is to ask MEDCAC, the Medicare Evidence Development and Coverage Advisory Committee, to collect information about the procedure it is supposed to evaluate and to discuss, publicly, its evaluation of the information. MEDCAC is an independent panel of 100 people, drawn from medicine, industry, science, ethics, public health, economics, and the public, from whom up to 15 people are chosen to address any particular issue that comes up. Medicare asked MEDCAC to review low-dose CT screening and on April 30,  the committee had an all day meeting. The agenda is available on line. The evidence was presented and discussed. The committee voted—each person’s vote is public and each person was asked to explain the rationale for his vote. What the committee concluded was that the evidence supporting the use of low-dose CT scanning to screen for lung cancer in the Medicare population just wasn’t there. The transcript of the entire meeting is on line. It runs to 310 pages.

To be precise, the committee members were asked “how confident are you there is adequate evidence to determine if the benefits are greater than the harms” for Medicare enrollees. They could vote  from “1” (little confidence) to “5” (high confidence). The average vote was 2.33. But the US Preventive Services Task Force, another independent body of experts, had recently given low dose CT scanning a “B” grade, recommending that it be used in people ages 55-79 who have a 30 pack-year smoking history and are currently smoking or have quit within the last 15 years. How could MEDCAC vote no and USPSTF vote yes?

It turns out that the members of the USPSTF didn’t exactly vote yes. They suggested excluding people with “a health problem that substantially limits life expectancy or the ability or willingness to have curative lung surgery.” The reason for this caveat is that screening for lung cancer, like screening tests in general, only makes sense if early detection leads to cure or at least more effective treatment. And the only truly effective treatment for the vast majority of cases of lung cancer in smokers is surgery. Major surgery: removal of all or part of a lung. So the question for Medicare is whether doing major surgery in older people with lung cancer is a good idea. 

The single study on which the USPSTF recommendation was principally based, the National Lung Screening Trial, though it included 52,000 high risk individuals randomized to screening with low-dose CT or screening with an old fashioned chest x-ray, included relatively few people over 65 (26%), very few people over 70 (9%), and few individuals with other health conditions. So when this study, which by the way was funded by the National Cancer Institute, part of the National Institutes of Health, reported that the death rate was 20% higher in people screened with a conventional x-ray than in those screened with low-dose CT scan (based on a reduction in the death rate from lung cancer from 309/100,000 to 247/100,000 in 6.5 years), its conclusion rested on the 79 excess lung cancer deaths (425 vs 346) in those getting regular x-rays. In other words, 320 people had to be screened to prevent one death. We do not know how many of these 79 deaths were in older people; we do not know how many of these 79 deaths were in people with other serious illnesses such as heart disease or diabetes; and we do not know, for those who survived their lung cancer, how many would go on to die of other illnesses in the near future.

It was based on this kind of analysis that MEDCAC determined there just wasn’t enough evidence to justify ordering Medicare to reimburse for the screening of high risk older individuals with low-dose CT scanning. It didn’t say low-dose CT scanning doesn’t work at all: it said we don’t have enough information about older, sicker patients. It didn’t say low dose CT scanning won’t work in older people: it said there isn’t compelling enough reason to mandate reimbursement for this test and treating such patients for lung cancer with surgery, the only treatment associated with a high cure rate, might in fact do more harm than good to such individuals.

The specific reasons that MEDCAC chose to vote no to Medicare reimbursement are not actually terribly important—though I’ve included some to give a sense of the reasons that were invoked, and all the reasons are publicly available—what is most important is that the rigorous, evidence-based process on which the decision was based be honored. Medicare has yet to issue an “NCD” (National Coverage Decision). It could still be swayed by political pressure, by lobbyists, by emotional personal stories of individuals whose lung cancers were detected by low-dose CT scanning and who believe their survival hinged on this. The 150 lawmakers who are pressuring Medicare are hoping to achieve exactly this end. If Medicare is to remain the excellent insurance program that it in many ways is, it must do what all third party insurers have to do: decide what to cover and what not to cover. And it should make that decision based on the facts, not on the ignorant screed of politicians.

A POLST Post

For 40 years, doctors, philosophers, and patients have been trying to figure out how to make sure that people can have a say in decisions about their medical care near the end of life, even if they are no longer able to speak for themselves. We’re still trying.

First we had the “living will.” A conventional will allows a person to specify what should happen to his property after he dies; why couldn’t a similar document indicate what should happen to his person while he was still alive but unable to articulate his treatment preferences? The problem with living wills, which were typically of the form “if I am terminally ill and there is no reasonable chance of my getting better, I do not want extraordinary measures to prolong my life,” is that they were too vague. Just exactly when is a person terminally ill? In the last days of life? Months? What about someone with a progressive, ultimately fatal disease such as Alzheimer’s, which can last years? At what point does someone with dementia become “terminally ill?” And what are extraordinary measures? A breathing tube? Attempted cardiopulmonary resuscitation? Hospitalization?

Because living wills were so imprecise, two physician-ethicists came up with the idea of an “instructional directive,” a document that spells out what the doctor should do in each of various situations. Called “The Medical Directive,” it took the form of a complicated grid and asked patients, when they were well enough to make medical decisions for themselves, what approach to treatment they would want if they were irreversibly comatose, if they were comatose with a very small chance of regaining some degree of function, if they had advanced cognitive impairment and terminal illness, or if they had advanced cognitive impairment but no (other) terminal illness. Prospective patients were offered 12 possible procedures or treatments, ranging from CPR to dialysis to antibiotics. For each of the 4 scenarios, they were asked to indicate if they would want (perhaps a better word is “accept”) the procedure if recommended by a doctor, if they would refuse the procedure, if they would want a limited trial, or if they were unsure. This document seemed like the doctor’s dream: it would tell the physician exactly what to do when faced with a very sick but mentally incapacitated patient. Considering a breathing tube? Chemotherapy? Blood tests? Just scan the Medical Directive and see if it’s on the approved list. But there were problems. It was a complicated document to fill out. It was not clear if patients understood what they were saying yes or no to when they filled it out. And it only included 4 clinical scenarios. What about the 85-year-old with heart disease, kidney failure, diabetes, and emphysema who was hospitalized with a severe pneumonia? That patient doesn’t show up in any of the 4 scenarios—but might not want some or all the proposed interventions.

Some people despaired of finding a way to record in advance what a patient would or would not want. Instead of telling the doctor what to do, they proposed telling the doctor whom to speak with to make medical decisions. With this approach, all a patient has to do is designate who that person should be—variously called a health care proxy or a durable power of attorney for health care or a surrogate decision-maker. But this is problematic, too. Often prospective patients choose a health care proxy without talking to that person about their preferences. Sometimes the proxy doesn’t even know he is the proxy.

In recent years, many physicians and ethicists have advocated “combination directives,” which include short versions of all the above documents: they name a surrogate, they summarize the patient’s general preferences, and they list which of a handful of procedures would be acceptable in each of several clinical situations. But even these combination directives were not a great solution, in part because only a few people filled them out, and in part because physicians had difficulty figuring out how to translate them into practice.

To solve the translation problem, physicians came up with a new approach: instead of relying on an advance directive, create a doctor’s order, binding on paramedics, emergency room physicians, nursing home staff, and any other medical personnel who came into contact with the patient. The earliest form of such an order was the “out of hospital DNR” order, a form signed by the patient and by his attending physician that stated that cardiopulmonary resuscitation should not be attempted. A more nuanced and complex approach is the Physician Orders for Life-Sustaining Treatment (POLST), piloted in Oregon. The POLST is supposed to build on a discussion with the patient about his goals of care. Based on that discussion, the patient and physician fill out a document that indicates which of various treatments (CPR, a ventilator for breathing, artificial nutrition) may be used. It’s similar to the Medical Directive, but does not ask the patient to imagine being in any of several conditions (the newer versions of the Medical Directive, incidentally, are expanded to include 6 situations), and it is a doctor’s order, not a statement of preferences.

POLST forms have caught on in many states. According to the National POLST organization, 14 of the 50 states have endorsed POLST (though only 2 of these have ‘mature’ programs), 27 are developing POLST programs, and 10 (including Washington, DC) have no programs. In Massachusetts, it’s the Massachusetts Medical Orders for Life-Sustaining Treatment (MOLST) and in West Virginia, it’s Physician Orders for Scope of Treatment (POST). But the big question is whether these forms are successful in assuring that the patient’s voice is heard. A new study in the Journal of the American Geriatrics Society—the largest analysis of the outcomes of the POLST program to date—says they are. 

The authors examined all 58,000 adult deaths in the state of Oregon in 2010-2011 and found that 17,902 had submitted a POLST form to the “POLST Registry.” When they studied the preferences those patients expressed on the POLST, they found that 66% wanted a focus exclusively on comfort, 6.4% wanted maximal medical treatment, and 26.7% wanted something in between. They also found that patients who had completed a POLST form were more likely than those who had not be white, very elderly, and female, to live in an urban area, to have more than a high school education, and to have a diagnosis of cancer. When they looked at subpopulations with particular diagnoses, they found that 41% of people dying of cancer had filled out a POLST compared to 35% of people with Alzheimer’s disease and only 25% of people with heart disease.

To answer the question of the impact of POLST completion on the care the patients received, the authors looked at whether the patient died in the hospital—not necessarily an accurate marker of whether the medical order was carried out. It turns out that 1/3 of people who did not complete a POLST died in the hospital, compared to only 6% of people who had. Looked at differently, 6.4% of patients who specified a comfort-only approach (and specifically said they would not want to be hospitalized unless essential for comfort) on the POLST were in fact hospitalized, compared to 44.2% of those whose POLST forms said they were to have no restrictions on treatment and 34.2% of people with no POLST.

There’s much we still don’t know about the POLST. We don’t know how often patients got interventions they had specifically stated they didn’t want. We don’t know how often physicians recommended treatments or diagnostic tests and then proceeded with them—or changed course—because of what was specified on the POLST. Most crucially, we don’t know whether patients in fact discussed their general state of health, their prognosis, and their goals of care with their physician before completing a POLST form. We don’t know whether the 71% of patients who died without a POLST on record had been offered such a form and declined to fill it out or whether the subject hadn’t been broached with them.

What we do know from many studies is that when patients discuss their condition and their wishes with their family members and their doctors, they have a much better chance of having a comfortable final phase of life than if they do not. Translating those wishes into an advance directive or a medical order may turn out to be less important than having the discussion in the first place.

Hey Ho, ACO!

Health care costs in America continue to be outrageously high and rising, although there is some evidence that the rate of rise has been slowing, at least within the Medicare program. The proposed fixes to the problem have either focused on patients (claiming that if only patients had more “skin in the game,” they would stop buying care that costs so much), on government (which in principle could control prices or, in the case of Medicare, limit the procedures and treatments it will pay for), or on physicians (encouraging them to provide “high value care” and to eschew “low value care” through a variety of incentives).

The focus on patients hasn’t worked out too well since decision-making rests largely with doctors and often results in choices deleterious to health when patients do actually choose. Health insurance companies continue to promote this option with “high deductible health plans” and tiered reimbursement (more limited reimbursement for more expensive drugs or higher cost hospitals) and free market economists, along with free market politicians continue to advocate “patient choice” and competition as the answer, despite the limitations of this approach to date. 

The focus on government is a problem, at least at the federal level, since government cannot seem to agree on much of anything and is unlikely to pass legislation that could be construed as rationing. And limiting choice, even when the choices that are eliminated are unambiguously bad choices, is seen as rationing. Even having the government (in the form of Medicare) pay for conversations about whether patients want expensive, minimally effective treatment near the end of life elicits the “death panel” canard.

The remaining option is for doctors to solve the problem. One of the leading strategies for doing this is through managed care, which puts the responsibility for rationalizing care, for avoiding needless duplication of services, and for preventing cost shifting from one part of the health care system to another by putting all the components into one integrated system of care. Now this is in principle possible to do with traditional managed care organizations that really do control outpatient services, hospital services, physician services, and post-acute services for their patients. But managed Medicare (the Medicare Advantage Plans) remain a small part of the Medicare program. To allow patients to continue to avail themselves of fee-for-service Medicare but at the same time assure the cost/quality benefits of managed care, the Affordable Care Act came up with the idea of Accountable Care Organizations (ACO). In this arrangement, patients do not actually join an integrated system. They simply see whatever primary care and specialty physicians they wish and go to whatever hospitals they want, but Medicare assigns them to a particular ACO. The ACO is then expected to lower its rate of spending while providing a high level of care. The ACO is rewarded if it succeeds and penalized if it fails. How can this possibly work, if the ACO cannot exercise control over the services “its” patients receive?

A new article in JAMA Internal Medicine suggests that it can’t.

The authors looked at the behavior of over half a million Medicare beneficiaries in 2010-2011 (before the ACO system was put into effect) and found that one-third of them would have been assigned to a different ACO in 2011 from the one they would have been assigned to in 2010. This means that the incentive for an ACO to invest in chronic disease management programs, care coordination, and patient education would be small as they cannot expect to reap the benefits of their investment until much later—by which point the patients might no longer be part of their ACO. They also found that while patients tended to see a primary care physician within the ACO—only 8.7% of primary care visits were to doctors outside the ACO—they saw specialists outside the ACO two-thirds of the time.  If ACO’s can exercise only limited control over the utilization of resources by their patients, how can they possibly control costs?

They can’t. Managed care “lite” is an oxymoron. Either the care is managed or it isn’t. ACOs represent a valiant attempt to preserve patient choice in all things medical but at the same time to accrue the benefits of integrated, coordinated, managed care. The two aims are contradictory. Time to rethink the ACO.

A Tale of Two Medicines

Imagine two versions of a medication that are essentially identical but one costs 100 times as much as the other. Using the more expensive variety enriches the drug company that manufactures it and the doctors that administer it—with patients and the American people paying the price. Yet that’s exactly what’s happening with two drugs that are used to treat a form of macular degeneration and diabetic macular edema, 2 common conditions that cause blindness, principally in older people. And the reason this is happening is the steadfast refusal of Congress to allow Medicare to consider cost in making reimbursement decisions.

The story came to light when Medicare released data in April about how it spends its money. It turns out that eye doctors are among the physicians who get more than a million dollars a year from Medicare. The reason: they inject the drug Lucentis (ranibizumab) into the eyes of their patients, an effective treatment for macular degeneration or macular edema, and are paid the average price of the drug plus 6% overhead by Medicare. They could, however, inject a far cheaper version of the drug, using the chemotherapy agent Avastin (bevacizumab) off-label. In absolute dollars, Lucentis costs $2023 per dose and Avastin costs $55 per dose. And a typical patient is treated up to 12 times per year, sometimes for years.

It turns out that that National Institutes of Health (NIH) actually funded a study of over 1000 patients comparing treatments for the more common of the 2 eye conditions for which the drug is used, neovascular macular degeneration. The study found no difference in effectiveness between the 2 drug variants, Lucentis and Avastin. Not enough patients were enrolled to determine if there are statistically significant differences in safety. That is, it’s possible that one version of the drug is safer—and since a compounding pharmacy has to convert the chemotherapy drug Avastin into the drug used by eye doctors, this is a theoretical possibility—but we don’t know for sure. Subsequent studies have found minor differences in the safety profiles of Avastin and Lucentis, which are both made by the same drug company.

Carefully performed modeling of what we can expect over the next 10 years—making reasonable assumptions about how common the eye conditions will be, how beneficial treatment is likely to be, the extent of side effects—shows that if current practice patterns continue, in which 2/3 of people are treated with the cheaper drug and 1/3 with the costlier drug, then Medicare will spend $20 billion on these 2 treatments and patients will spend $5 billion (on co-pays). If, however, everyone started using the cheaper drug, then Medicare would spend only $2 billion and patients only $420 million. And if everyone switched to the more expensive drug, Medicare would spend $57 billion and patients would spend $14 billion. So what should we do?

A recent opinion piece in JAMA suggests that Medicare, which is currently prohibited from negotiating over drug prices, could get around this problem by using a technical loophole. It can exercise flexibility in how it figures out how much to pay for a product or service if the current payment amounts are “inherently unreasonable because they are either grossly excessive or deficient.” To date, this process has never been invoked in paying for medications.

The article in Health Affairs describing the predictive model suggests that Medicare could selectively increase reimbursement to doctors who use the cheaper drug or that Congress could modify the regulations for “biosimilars” (as the generic versions of biologically complex drugs such as Lucentis are called). All these approaches take as a given that for Medicare to spend 1/6 of its entire Medicare Part B budget on one drug is absurd. Just how absurd becomes apparent if we look at the cost-effectiveness of Lucentis: medical costs would go up $2 million/QALY (quality-adjusted life year) if everyone took Lucentis instead of Avastin, though most health economists consider $50,000-$100,000/QALY a reasonable amount to pay.

What all this tells me is that it’s time we stop pretending that cost-effectiveness analysis is unethical. Spending $25 billion in 10 years when we could achieve the same end by spending $2 billion is what’s unethical. Comparative effectiveness research, which is supported by the Affordable Care Act and which forms the foundation of the new independent research agency, the Patient Centered Outcomes Research Institute (PCORI), is legally prohibited from considering cost. But if we don’t start considering cost explicitly, we will pay dearly in the not so distant future.