September 17, 2018

An Aspirin a Day...

The headlines this week—aside from the hurricane, the typhoon, and the charge of sexual misconduct in the Supreme Court nominee—are all about aspirin. For older people, unless you live in the Carolinas or Hong Kong, this is definitely the story. A new study (reported as 3 separate studies but really just one study with three different endpoints) threatens to unseat aspirin from its coveted spot as the little-pill-that-could.
A single aspirin a day, many people believed, could stave off heart disease, stroke, cancer, and perhaps dementia. If taken as a “baby aspirin,” a dose of 81 mg a day instead of the 325 mg in a regular aspirin tablet, and with a special “enteric” coating to protect the lining of the stomach, it was touted as effective with virtually no side effects. The truth, unfortunately, seems to be that it is neither effective nor devoid of side effects when taken by healthy older people.
The study, published online in the New England Journal of Medicine, examines three plausible possible benefits of low dose, enteric-coated aspirin. First, they ask whether aspirin has a desirable effect on cardiovascular events such as heart failure requiring hospitalization, stroke, or heart attack. They found no difference in benefit between healthy older people in the US or Australia (where older was defined as over 70 except in blacks and Hispanics, where it was defined as over 65) who took 100 mg of aspirin and those who did not.
Next, they looked at whether aspirin has an effect on how long healthy older people live without developing a disability. Again, they found no statistically significant difference between those who took aspirin and those who didn’t.
Finally, they examined overall mortality in the aspirin-takers and the non-aspirin takers. Once again, the two groups were indistinguishable.
There was, however, one striking difference in outcomes between the 9525 people who were randomized to take aspirin and the 9589 people who were randomized to placebo: the risk of bleeding was significantly higher. And by bleeding, the investigators meant major bleeding such as a gastrointestinal bleed or an intracranial hemorrhage. 
Not only did this randomized controlled study fail to show any benefit from taking aspirin, and not only did it show an increased risk of harm, but even when the results were subjected to subgroup analysis, no group emerged as potential beneficiaries. The authors looked at the composite endpoint (dementia, death, or persistent disability) in several pre-specified subgroups. One was gender: in the past, aspirin has been touted as preventive for healthy men but not women; in this study, neither men nor women benefited. Another was frailty (though I’m not quite sure how 421 of the “healthy” elderly subjects could have met the definition of frailty): in this study, neither the frail nor the non-frail benefited. If anything, there was a trend towards worse outcomes in the frail group, though the numbers were so small that the difference was not statistically significant and might well be due to chance.
No study is perfect and this one is no exception. The median period of observation was 4.7 years, a relatively short period with respect to the time needed to develop dementia or heart disease. The analysis was done on an “intention to treat” basis, which is the way such studies are supposed to be analyzed, but in fact only 2/3 of the people assigned to take aspirin were actually taking it by the end of the study period. The benefit of aspirin might therefore have been under-estimated. The risk of bleeding, however, which was already substantial in the aspirin users, may have also been under-estimated. For some reason, the study used a 100 mg dose even though a standard baby aspirin contains only 81 mg: maybe the results would have been different with an even smaller dose. But the strengths of the study are impressive. It was randomized; follow up was almost complete; data collection seems to have been thorough and careful.
I have a confession to make: for several years, I took a baby aspirin every day. I’m under 70 and I’m female, so my physician did not recommend that I take aspirin. I took it nonetheless because I really don’t want to have a stroke and thought that just maybe taking aspirin was something I could do to help. I took it because years ago, before I went to medical school, I worked in a hematology research lab and spent my days studying platelet aggregation. It turned out that people who had taken a single aspirin tablet within two weeks of my testing their blood showed markedly decreased clumping of platelets, blood cells that are critically involved in the clotting process. About a year ago, I had several episodes of subconjunctival hemorrhage, a benign form of bleeding involving the blood vessels of the eye. I worried the bleeding might be related to aspirin, so I stopped taking it. 
Today, the evidence is compelling that for people without heart disease or dementia or stroke, an aspirin is more likely to cause harm than good. As of now, aspirin has joined the ranks of other failed panaceas such as estrogen and calcium supplements. 

September 09, 2018

Looking Ahead

British researchers recently projected care needs of the very old.  We would do well to pay attention to their analysis—or, if we think Americans are substantially different from their British counterparts, then we should replicate the analysis with our own data. Their study concludes that the number of very old people (over age 85) who will be highly dependent will double by 2035. This despite a marked decline in the anticipated rate of dementia and the growth of the population who remain fully independent. The seeming paradox arises from an increase in the number of comorbidities that the very old will have and the interaction between multimorbidity and dementia. In short, if you do survive to extreme old age and if you are one of the still substantial number of people who develops dementia, you probably will also have several chronic diseases. The combination spells dependence, where the study authors define high dependency as requiring 24-hour care, medium dependency as needing help at regular times daily, low dependency as needing help less than daily, and independence as free from care needs.

A few charts make the points best.

If we look at the proportion of all 85+ year-old men who were highly dependent in 2015, we can see that it was low then and it will have fallen further by 2035. Very old women are somewhat worse off than men today and the discrepancy will increase by 2035. Since the total number of octogenarians will rise considerably over the next 20 years, the total number of dependent, very old men and women will be larger than it is today--in the UK, the numbers will double.

 If, instead, we look at the number of additional years that people who turn 65 can expect to live, we have a more nuanced view. Men who turned 65 in 2015 can anticipate another 18.7 years of life, of which 11.1 years will be spent entirely independent and 1.4 years will be in a state of high dependency. A man who turns 65 in 2035, projections suggest, can expect another 22.2 years of life, during which he will spend 15.2 years independent and only 1.1 years dependent.

Women who turned 65 in 2015 can expect to live another 21 years, during which they will be independent for 10.7 years and very dependent for 2 years. The next generation, which turns 65 in 2035, will have a life expectancy of 24.1 years and will independent for 11.6 of them. Unfortunately, the length of severe dependency will increase to 2.7 years.

The implication of all is that we will need a huge increase in the number of caregivers to accommodate the needs of the very old. Unless we are far more successful in rolling back the rate of chronic disease than we have been to date (cardiovascular disease, diabetes, and stroke are the leading offenders) and can also dramatically cut the risk of dementia, we need to start planning now. We should encourage more people to become nursing aides. This will involve raising the pay and enhancing the status of the job. We will also want to seriously consider increasing immigration to fill the needs of the oldest old. And we should not assume we are better off than our British counterparts: on the contrary, older Americans today experience more sickness and disability than our European counterparts and our health care system devotes less attention to social than medical problems.

August 20, 2018

Hospital at Home?

For years, I’ve been arguing that frail older patients are often better off being treated for acute medical problems at home than in the hospital. I began making this claim during my residency when I wrote about The Ethics of Cardiopulmonary Resuscitation,” claiming that it created cognitive dissonance for physicians in hospitals to go all out for some patients and to withhold selected treatments for others. Better, I suggested, to keep those individuals who didn’t want everything under the sun, and particularly those who didn’t want attempted cardiopulmonary resuscitation, out of the hospital altogether. 

I picked up the theme during my geriatrics fellowship in a paper entitled “Adverse Consequences of Hospitalization in the Elderly," which quantified the frequency with which older patients developed iatrogenic complications, unrelated to the medical problem for which they had been admitted and found it was 40.5 percent in those aged 70 or older, compared to 8.8 percent in those under 70. And instead of suggesting that we build higher siderails—or institute the kinds of interventions that eventually were adopted by Acute Care for the Elderly (ACE) units—we try to avoid hospitalization altogether. 

Over time, increasing evidence has accumulated that many people can be cared for at home and that this approach not only decreases the chance of adverse events (including unwanted treatment), but also results in comparable outcomes. A new study in JAMA Internal Medicine  suggests a way to pay for “hospital at home” model--by paying a single fee for acute care plus 30 days of post-acute care.
The Hospital at Home model (which, unlike the study authors, I won’t abbreviate as HaH since that makes it seem like a joke) used in this study offers hospital-style care for any of 19 common medical conditions (including pneumonia, urinary tract infection, flare of heart failure, and cellulitis) in clinically stable, interested adults who have adequate home supports. Their outcomes were compared to those of contemporaneously hospitalized adults who were eligible but either declined to participate in the study or couldn’t participate because services could not be set up for them at the particular time (night time) or day (weekends) when they showed up in the emergency department.
The results were overwhelmingly positive. Patients cared for at home, when compared to controls, were much less likely to be (re)hospitalized for any reason in the month after discharge (8.6 percent vs 15.6 percent); they were less likely to go to a skilled nursing facility after treatment of the acute episode (1.7 percent vs 10.4 percent); and they were far more likely to rate the quality of their care as excellent (68 percent vs 45 percent). While 12.2 percent of the Hospital at Home patients did end up in the hospital for treatment of the acute illness, there were no differences in mortality. There were also no differences in complications such as falls or infections, but the number of either of these adverse events was too small to draw any definitive conclusions.
On the other hand, as the authors of the accompanying commentary indicate, the study was a case control study, not a randomized trial, and the “controls” were significantly different from the cases: they were younger (a lot younger—an average of 71.5 compared to 76.0), less well educated, and better functioning (55.5 percent had at least one area of difficulty in daily activities, compared to 71.5 percent in the Hospital at Home patients). However, finding that the home care patients fared at least as well and, in many instances, better than their hospitalized counterparts, even though they were older and frailer suggests that they may have more to gain by staying out of the institution.
It's heartwarming to see evidence that treating many medical problems at home is achievable without sacrificing any important benefits and to discover there are plausible ways of paying for such care. My concern is that the “hospital at home” model, while an improvement over the hospital environment precisely because it takes place in a familiar location, may bring too much hospital technology with it. I would like to have a better understanding of just what hospital at home includes. I’m reminded of the final days of Pope John Paul II, dying of Parkinson’s disease and its complications. The Pope was cared for at home—but with a feeding tube, a ventilator, and physician care. He was, for all intents and purposes, receiving ICU care at home. So, by all means, let’s make medical care available to older people at home; let’s keep them out of the hospital if possible. But that should not mean bringing the hospital in entirety to them.

August 12, 2018


Think “A Man Called Ove,” Fredrik Backman’s bittersweet novel about a lonely man who tries repeatedly and unsuccessfully to commit suicide but ultimately finds meaning in life by helping others. Water down Ove-- take away his rage against injustice and dilute his frustration with modern technology. Make him 85 instead of 59-going-on-80. Preserve the pet cat, the needy neighbor, and the enduring love for his deceased wife. What you get is the protagonist of Elizabeth Berg’s new novel, “The Story of Arthur Truluv.” The book is a sentimental portrait of another octogenarian, Arthur Moses, nicknamed “Truluv” by the lost teenager he befriends because of his devotion to his wife, whose grave he visits daily. 

It’s high time we had more fictional treatments of older people, so I’m happy we have another contribution to the genre. And Truluv is an endearing old fellow with a big heart—albeit one apparently affected by heart failure. He, unlike Ove, is a glass-half-full sort of a fellow, whereas Ove, at least until the end of his story, is a glass-with-a-hole-in-it sort of a guy. While I appreciate that Berg does not completely sugar coat old age, giving both Arthur and his neighbor Lucille an assortment of maladies and expeditiously killing off Lucille’s elderly boyfriend with a massive heart attack, Arthur is so nice that he’s a bit hard to take. His teenaged friend Maddy, pregnant and estranged from her father, needs both a job and a place to stay? No problem, Arthur lets her move in and serve as his housekeeper. The irksome woman across the street, whose main virtue is her baking skill, decides she’d like to move in, too? Fine, she can do the cooking. Maddy hasn’t seen her father in months and secretly would like to reconcile with him before she gives birth? Arthur intuits what’s going on and invites him for Thanksgiving dinner. 

I’m a pushover for over-the-top amazing octogenarians, so I confess I enjoyed reading this book. It’s short—222 small pages, peppered with dialogue—so I only had to devote a few hours of my life to this endeavor. Maybe I would have liked it more if I hadn’t recently read “A Man Called Ove,” a tragicomic masterpiece with a protagonist who seems old even if he isn’t, a NY Times bestseller that has been translated into 43 languages. But maybe Elizabeth Berg will next turn her talents to writing about an octogenarian who is frail but who nonetheless finds meaning in life. Now that would be a major contribution to the literature on aging.

August 07, 2018

The Truth About Hospice

Hospice is widely touted as the best form of care near the end of life. As a result, the proportion of people who are enrolled in hospice at the time of death has been steadily increasing. Nearly half of all Medicare patients are receiving hospice care at the time of death. But a new report suggests that the quality of care they receive may not be what it should be. Moreover, a chunk of the $16.7 billion that Medicare spends on hospice (that was in 2016; it's more today) goes towards care that is unnecessary or not provided at all. What exactly is the problem? How widespread is it and how can patients and families identify hospices that provide high quality care?
While substandard care and fraudulent billing are distributed throughout the country, these practices occur disproportionately in the 64 percent of hospices that are for-profit. The bad behavior predominantly involves two specific variants of hospice care: general inpatient hospice care and hospice services that are provided within institutions, whether skilled nursing facilities or assisted living facilities. General inpatient care is an intensive form of care offered when symptom control cannot be achieved elsewhere (ie at home], so the patient is transferred to a hospital or skilled nursing facility. For this type of care, Medicare pays a per diem rate of $720, compared to $187 for home hospice (actually this is the rate for days 1-60, after that, the rate drops to $147 per day). The kinds of abuses that have been reported are indeed disturbing: the report gives several examples.
However,  General Inpatient Care accounts for only 1.5 percent of all the days that patients were enrolled in hospice. 
Other cases of allegedly inappropriate billing take place in Skilled Nursing Facilities (SNF) or in Assisted Living Facilities (ALF). Some instances were unequivocally fraudulent, as when hospices paid facilities to enroll patients in hospice, sometimes without the knowledge of patients or their families.  But again, it’s important to keep this in perspective.  A minority of hospice patients are cared for in these environments: 25 percent of hospice beneficiaries during the period the study examined (2006-2016) lived in a SNF and only 13 percent in an ALF.
Several of the criticisms leveled by the report are of questionable significance. The study says that many hospices fail to develop the mandatory “comprehensive care plan,” a document that spells out what services the patient needs and how they will be provided. This failing was especially egregious in the inpatient setting, where in 85 percent of cases, no such plan was documented. But the absence of a written plan doesn’t mean there was no plan or that inadequate care was provided. It means just what it says—the hospice did not create the requisite piece of paper. Maybe, just maybe, they thought it was a waste of time, a meaningless bureaucratic requirement.
Hospice care is also criticized for failing to provide physician visits. Actually, hospice does provide one physician visit. A face to face visit with a physician at the time of admission to hospice has been mandatory since 2011, so hospice patients do see a physician at least once during their hospice stay, which in many cases is short: over half of all hospice patients are enrolled for less than 30 days, and over one-fourth are enrolled for a week or less. Many hospice patients and their families would like more involvement of the medical profession, particularly their primary care physician or the relevant specialist, but that’s different from suggesting that the quality of care was poor because there were no medical visits at all.
Skilled nursing facilities are accused of double dipping, of billing Medicare for services that they already routinely provide as part of their usual care. This has been a worry since Medicare agreed to cover hospice care in a SNF. Care by nurses and nurses’ aides are a normal component of nursing home care. But that doesn’t mean that the services offered by hospice are redundant; on the contrary, the reason that hospice can be beneficial in the SNF environment is precisely because the usual nurses may not be skilled in pain management and the amount of aide time may be totally inadequate for a dying patient.
Lastly, the report argues that the payment system incentivizes hospices to minimize services and to cherry pick patients, selecting low-need, long-stay patients, such as those with advanced dementia, rather than high-need, short-stay patients, such as those imminently dying of cancer. We’ve known about this phenomenon for some time. In fact, CMS responded by modifying its payment system to pay more for the first few days of care and for the last few days of care and less for the in-between days. The practice of complementing high-cost patients with low-cost patients is only a problem if the low-cost patients don’t deserve to be enrolled in hospice at all. If they do, and there’s every reason to think they do, then it’s simply a good business strategy. Hospitals do the same thing when they add services for which they are generally reimbursed well to compensate for the services for which they are poorly reimbursed. The alternative would be to abandon the per diem payment system and go to fee-for-service. But that’s the model that is gradually fading away from the rest of the medical system because it promotes the use of volume rather than quality.
What should we conclude about hospice care in America? How should prospective patients respond to this report and how should Medicare respond? 
Patients and their families would do well to consult Hospice Compare, the CMS website, for basic information about quality. This tool relies principally on patient and family satisfaction surveys rather than objective measures, but it’s very revealing. Rather than add more measures, which is what the authors of the report propose,  I think the tool should be publicized: I was familiar with Hospital Compare and Nursing Home Compare, but I confess I did not know about Hospice Compare until recently--and the site was launched in August, 2017. 
If families have a choice between a not-for-profit hospice and a for-profit hospice, they may find it prudent to select the not-for-profit option.  On average, not-for-profit facilities are associated with higher quality care. 
As to steps that Medicare should take, increased across the board regulation, which is what the report advocates, seems unduly burdensome. Far wiser would be to target enhanced supervision and regulation to those areas that need it most, namely skilled nursing facilities and assisted living facilities. Stiffer penalties for malfeasance would also be wise. As with banking fraud, it’s not enough to slap the owner or executive director on the wrist. Jail time is a much more powerful disincentive to bad behavior. And the bad behavior we’re talking about here is neglecting people who are dying and in pain.

July 29, 2018

Treatment for Alzheimer's or False Alarm?

The most promising treatment of Alzheimer’s disease to date uses monoclonal antibodies to rid the brain of amyloid plaques. These plaques are widely thought to be crucial in the development of Alzheimer’s disease, the form of dementia that afflicts over 5 million Americans—and, unless we find a way to cure or prevent it, will afflict millions more in the coming years. Researchers hope that these drugs will prove to be “disease-modifying,” that they will disrupt the cascade of events that produces the symptoms of the disease. 

So far, the studies of monoclonal antibodies in this setting have been very disappointing: Bapinezumab: 3 failed trials. Solanezumab: 3 failed trials. Crenezumab: 1 failed trial. In the past year, three major pharmaceutical companies, Eli Lilly, AstraZeneca, and Merck, all stopped development of the monoclonal antibodies they were testing against Alzheimer’s disease. Another company, Pfizer, closed its neurodegenerative disease research section entirely. Now we learn that a Phase II trial of the as yet unnamed agent “BAN2401” actually “looks promising.” How promising?

Like its cousins, BAN2401 is a “humanized monoclonal antibody.” That means an antibody that was developed in mice and modified so it wouldn’t be rejected as alien by the human immune system. Now mice don’t normally get Alzheimer’s disease but certain strains have been genetically engineered so that they do. Using the genetically engineered mouse model, a monoclonal antibody was created that binds to amyloid. Today, this chemical is called BAN2401; when it enters the next stage of assessment, it will presumably acquire a name that ends in “zumab” to indicate it is just such a humanized monoclonal antibody. So, what do we know about this nameless entity?

We know that BAN2401 made it through a Phase I trial that tested its safety, tolerability, and pharmacokinetics in 80 people with mild to moderate dementia. We know it then entered Phase II of testing in people aged 50-90 with evidence of amyloid on their PET scans and either mild Alzheimer’s disease or mild cognitive impairment (a condition that is not dementia but often evolves into dementia). Five different dose regimens were administered intravenously and compared to placebo in terms of their effects on cognition and on brain amyloid deposits on PET scan. We know that when the preliminary results were announced last December, no cognitive benefit was reported. But now, after 18 months of study, the final results are in. They were presented at the Alzheimer’s Association International Conference in Chicago this past week.

What we don’t know is how this study will fare when subjected to peer review. We don’t know what the final paper will look like. But here’s what I can glean from the press release by Eisai, the company that collaborated with Biogen to develop and test the drug, and the reports by the news media:

--The amount of amyloid in the brain (as measured by PET scan) decreased with all the doses tested. The more drug, the greater the change. The findings were statistically significant. 

--There was no discernible effect on cognitive function except in the 161 people who were treated with the highest dose of BAN2401. In those 161 people, cognitive function deteriorated over time, just as it did in all the other groups, but it deteriorated less.

--Cognitive function was measured in a way that differs from the way it has been measured in other studies of treatments for Alzheimer’s disease. The researchers used a composite measure made up of several scales, each of which has been individually validated as a way to assess mental status, but only one of which is regularly used to report the outcome of clinical trials of Alzheimer’s drugs. 

--The patients tested included a mix of people with mild dementia and people with mild cognitive impairment (who don’t actually have dementia but aren’t entirely normal either) and the study had no way to determine whether the effectiveness of BAN2401 was any different in the two subgroups.

What does all this mean? The PET scan changes indicate that the monoclonal antibodies were successful in destroying amyloid deposits in the brain. That doesn’t prove that whatever effect the drug had was due to its amyloid-busting, but it’s suggestive. It also indicates that to the extent that the drug was a failure, it wasn’t a failure because it didn’t destroy amyloid. 

How should we interpret the cognitive changes? It’s hard to know, given that the researchers used a non-standard means of measuring clinical decline (a test they called ADCOMS). We cannot, for example, say that BAN2401 is better than one of the other monoclonal antibodies that was deemed a failure since nobody looked at the effect of the other drug on the “ADCOMS.” What we can say is that a 30 percent change in the rate of decline may be statistically significant but clinically, it’s a lot less impressive. Everyone  deteriorated, including those given the highest dose regimen; they just deteriorated a little less. How much of a difference actually occurred depends on the absolute decline: suppose the score went from 100 at baseline to 80 at the end of the study in the controls (I don’t know what the actual numbers are, so this is hypothetical). A 30 percent difference in the rate of decline means that the score in the treated group dropped from 100 to 86.  Is that clinically meaningful? Probably not.

So, no, BAN2401 is not likely to likely to dramatically change the course of Alzheimer’s disease. I hope the FDA does not allow the drug’s manufacturers catapult it into a clinical trial prematurely, before it goes through Phase 3 testing. But it has taught us a few valuable lessons. 

We now know that monoclonal antibodies can be designed that destroy amyloid plaques in the brain. We know that only the highest tolerated dose of the drug has any chance of being clinically useful. We should also realize the importance of studying patients with early dementia and those with MCI separately—it’s possible that using monoclonal antibodies once dementia has set in is too late, and I suppose that it’s also possible that using them before dementia has developed is too early. Finally, the study reminds us of the importance of a uniform methodology in conducting this kind of work. If the test that was previously regarded as the gold standard, the ADAS-Cog, is not the right test to use, then researchers need to agree on that and decide collectively what test to use instead. 

ABC News reported the drug as “giving patients hope” and showing “big promise.” Fortune magazine asserted that “world leaders want to end Alzheimer’s by 2025; a new drug breakthrough means we just might.” Investors are closer to the mark: Eisai’s stock price fell 10 percent after the data were reported.

July 23, 2018

The Right Way to Treat Cancer

Kudos to Judith Graham of Kaiser Health News for yet again identifying a new development that has the potential to enormously affect the medical care of older people. Her article, Geriatric Assessments Could Fine Tune Cancer Care for Older Adults,” pointed me to the American Society of Clinical Oncology's (ASCO) new guidelines for the management of patients over age 65 with cancer. The headline (and I know that journalists don’t generally write the headlines for their articles) is the only part of the article that’s misleading: the changes suggested would not merely “fine tune” oncologic care for older people, they would radically transform it.

Here’s the situation: 70 percent of cancer patients are over age 65 and that number will rise in the next 20 years. Looked at differently, the risk of developing cancer at some point if you are a man over age 70 is one in three, and the risk for women is one in four. Despite the predominance of older cancer patients, most clinical trials of chemotherapy drugs include few if any older people: they have too many other problems to meet the eligibility criteria for participating in the study. But those same people, with all their other medical problems and functional impairments, are regularly given chemotherapy, based on the results of studies that didn’t involve anyone like them. That’s a problem. 

What we do know is that 50 percent of older patients with advanced cancer experience severe toxicity from chemotherapy in the first three months of treatment. We also know that the Institute of Medicine’s recent study, “Delivering High Quality Cancer Care,found that the current delivery system is “ill-equipped to address the needs of older patients.” Oncologists agree with this assessment, stating that “the care of older adults with cancer needs to be improved” and putting the responsibility on themselves. Specifically, only one quarter or less of oncologists felt confident in their ability to recognize dementia, identify a fall risk, or assess functional status. And we know from a recent study that patients shift their priorities from sheer survival to quality of life as functional status declines. In light of these observations, what does ASCO recommend?

The new guideline, “Practical Assessment and Management of Vulnerability in Older Patients Receiving Chemotherapy,” makes a series of evidence-based suggestions. From the perspective of patients and their families, what’s important to know is that oncologists who are contemplating administering chemotherapy to people over age 65 should provide a geriatric assessment to identify problems they might not otherwise recognize. Specifically, they should take simple steps to look for falls, to determine how well patients can function day to day, to identify all other medical diagnoses besides the cancer that might impact treatment, to screen for depression, to evaluate cognition and nutrition, and to assess the social support system. Because the guideline is meant to be practical, it suggests specific tools that oncologists can use to achieve these ends. The guideline also advocates determining the approximate life expectancy, both the cancer life expectancy and the non-cancer life expectancy. 

What is the point of all this? How might it help older patients with cancer? The main objective is to avoid both over-treatment and under-treatment, where “too much” and “too little” must be assessed in light of the best available data about what treatment could achieve and knowledge of the patient’s preferences. How the oncologist presents the options should reflect the facts about what the consequences of treatment are likely to be. How much misery a person is willing to endure might well be affected by how much he or she has to gain in the best-case scenario and how much there is to lose in the worst-case, given the nature of the cancer and the patient’s other medical conditions. In short, the information from a geriatric assessment should guide the process of shared decision-making involving the physician, the patient, and the patient’s family, leading to an “integrated, individualized plan of care.”

But there’s another potential outcome, one that perhaps paradoxically undermines the accuracy of the predictive tools on which the recommendations rely. There is the possibility that performing a geriatric assessment will lead to implementing the recommendations of that assessment with resultant improved capacity to withstand the rigors of chemotherapy. The authors of the guideline recognize this possibility. They say, “While there are not yet completed RCTs [randomized controlled trials] that demonstrate that GA-guided care, or ‘GA with management,’ definitely improves outcomes of older patients with cancer, this care model has been shown to improve outcomes in older noncancer populations.”  Randomized controlled trials are reportedly ongoing to evaluate the ability of “GA-guided care” to actually improve outcomes in older cancer patients—not merely to modify their willingness to undergo chemotherapy in the first place, but also their longevity and quality of life if they do accept chemotherapy.

The first step is for patients and caregivers to request and oncologists to perform geriatric assessment. The second step is to act on the findings of the assessment. That doesn’t just mean using the information to guide decision-making about cancer treatment; it means, first and foremost, making some changes that will optimize the patient’s overall condition. These changes might include physical therapy, modification of the non-cancer medical regimen, or providing more social support. 
It might be necessary to delay instituting chemotherapy, and delay talking about whether to start chemo until certain basic interventions have taken place. 

One analogy that may be helpful is rehab. When a patient and family are struggling to decide if it’s going to be possible to return home after a debilitating hospitalization, it’s often helpful to begin with a stay at a rehab facility. The goal is to optimize the patient’s condition before making important decisions about where to live. A second analogy is depression. Depression may color a patient’s decisions about preferences for care. In general, whenever feasible, it’s desirable to treat depression before accepting a patient’s views about, say, limiting care for some other medical condition, as representing his or her true wishes. 

So, by all means, let’s do geriatric assessments on older patients with cancer. How, exactly, that will shape care remains to be determined.

July 16, 2018

Us or Them?

Nearly 30 years have passed since Congress passed the Patient Self-Determination Act, enthroning advance care planning as an important part of care in the last phase of life. While the frequency with which patients designate someone to serve as their proxy in the event of incapacity has increased dramatically and the use of various planning instruments such as living wills and instructional directives has also risen, in most settings no more than one-third of patients with serious illness actually have conversations with their physicians about their preferences. 

Assuming that it’s a good idea for patients with serious illnesses to speak with their physicians about their goals of care, whose responsibility is it to raise the question? Does it make more sense to concentrate on educating physicians to communicate well about goals or to focus on empowering patients? 

Over the years, different initiatives have targeted one population or the other. The Robert Wood Johnson’s “Project on Death in America,” a multi-hundred-million-dollar project launched in 1994, focused to a large extent on prospective patients. The AMA’s program, “Education on Palliative and End-of-Life Care” (EPEC), also introduced in the 1990s, is targeted entirely to physicians (in fact, the “P” in the title used to refer to physicians: when first launched, the program was called “Educating Physicians on End-of-Life Care”.) More recently, a series of videos by Angelo Volandes of ACP Decisions aims to show patients and their families what various medical treatments entail, while the “Serious Illness Conversation Guide” from Susan Block, Atul Gawande and others at Ariadne Labs offers a checklist of questions to help physicians structure their discussions.

So, who needs more attention, the doctors or the patients? A new study in Health Affairs may offer some clues. I should point out at the outset that the Health Affairs article does not seek to weigh in on the question of whether to concentrate on physicians or on patients when designing interventions to promote advance care planning. The aim of the study, as indicated by its ponderous title, “Factors Contributing to Geographic Variation in End-of-Life Expenditures for Cancer Patients,” is to understand why some regions of the US spend so much more on care near the end of life than other reasons, with no discernible difference in outcomes. 

There are methodological problems with this study, as with all studies that start with death and work backwards—it is possible that although the people who died had similar outcomes (perhaps not surprising, as they all died), other people who weren’t considered in the study because they didn’t die were more likely to do well if more money was spent on them. Nonetheless, we have two groups of people with advanced lung or colorectal cancer on whom very different amounts of money were spent—in the lowest quintile, the average outlay was $10,131 and in the highest quintile, the average expenditure was $19,318). Was the decision to spend more coming from patient pressure or was it something that physicians were pushing for?

What the authors found was that in the high spending areas, physicians were less knowledgeable about treating dying patients (by their self-report), less comfortable providing care to patients near the end of life, and had less favorable attitudes toward hospice than their counterparts in the low spending areas. The high spending areas also tended to have more specialty physicians but fewer primary care physicians, as well as fewer hospices (per capita) than elsewhere. But—and here’s the key—patients’ beliefs about what they wanted were no different in high and low-spending areas.

All the statistically significant findings were in the domain of physicians, not patients:

Now, this finding does not directly translate into the question of who drives decision-making near the end of life in general and advance care planning discussions in particular. But it strongly suggests that physicians play an outsized role in shaping what happens to patients. Many patients don’t have pre-determined preferences; their values do not unambiguously determine what kind of medical treatment they should get. Or, if they do have some idea of what would be best for them, they are nonetheless strongly influenced by the views of their physicians. If “shared decision-making” is to work, both partners need to be informed and on board—and the physicians are particularly in need of some attention.

July 08, 2018

Where We Die

Honoring patient preferences, which is shorthand for providing an approach to medical treatment consistent with what patients say they want, has become a fundamental tenet of American medicine. And one preference that the vast majority of Americans share, according to multiple studies, is the wish to die at home. 

Whether dying at home is actually what patients want when they are faced with impending death, rather than an abstract preference expressed when they are healthy, is another matter—and I’ve previously argued that as hospital-based palliative care improves and home-based palliative care places an ever-growing burden on families, hospitals have become more attractive as a site for dying. But what patients tend to agree on is that they don’t want to suffer as they are dying.

To the extent that hospital care entails interventions such as ventilators or ICU treatment or chemotherapy, patients are reluctant to subject themselves to this type of care, especially if its likelihood of prolonging life is remote. To the extent that fewer hospital deaths and more home deaths is a marker for less suffering at the end of life, change in this direction is very desirable. A new study by Teno and colleagues shows we are continuing to make progress in this direction.

A few years ago, Teno et al performed a similar study comparing the experience of patients in 2000 to the comparable experience in 2009. What they found then was a marked decrease in the percentage of elderly Medicare fee- for-service decedents (health-policy-speak for people over age 65 with conventional Medicare who died) who expired in the hospital (32.6 percent vs 24.6 percent). Over the same period, however, they found ICU use increased among decedents in the last month of life (from 24.3 percent to 29.2 percent) as did the percentage of dying patients who underwent a transition of care (nursing home to hospital, for example) in the last 3 days of life (10.3 percent to 14.2 percent). The current study updates these findings by extending the period of analysis to 2015 and by adding data from older patients enrolled in Medicare Advantage programs, who now account for 30 percent of the Medicare population.

What they discovered today was that the proportion of hospital deaths among the fee-for-service group has continued to fall, going from 32.6 percent in 2000 to 24.6 percent in 2009 to 19.8 percent in 2015. ICU use in the last 30 days of life, which had risen between 2000 and 2009, remained stable at the 2009 level in 2015. Transitions to another site of care in the last 3 days of life, which had also risen between 2000 and 2009, went back down in 2015 to the same level as in 2000. And the chance of being enrolled in hospice at the time of death rose from 21.6 percent in 2000 to 50.4 percent in 2015. When the investigators looked at a sample of Medicare Advantage patients, they found these individuals had the same experience in 2015 as their fee-for-service counterparts.

What does all this mean? I suspect what it means is that when we know with a high degree of certainty that someone is going to die in the very near future, we tend to focus on comfort care. If physicians, patients, and families recognize that death is imminent, hospitalization is relatively unlikely, as is ICU care or transfers from home or nursing home to another site of care. However, physicians often cannot be so certain that death is likely to occur in the next few weeks or months. As long as the usual strategy is to pursue maximally aggressive care until death is virtually sure to occur in the immediate future, and then to abruptly transition to care focused exclusively on comfort, the picture we see today is likely to continue.

There is another approach. That approach involves opting for a goal that is neither comfort only nor life-prolongation at any cost. Instead, maximization of function is paramount; treatment aimed at prolonging life is also acceptable, provided it will not affect quality of life in a major way. So, too, is comfort a goal, but only to the extent that it does not conflict with maintaining function. For people who are frail, extremely old, or both this alternative strategy translates into fewer hospitalizations, fewer ICU stays, and fewer transitions of care in the final stage of life, whether that period is measured in weeks, months, or even years. 

Medical treatment does not have to be all or none; there is something in between. It’s quite possible that many people would opt for this type of care—if only they knew it existed.

June 28, 2018

MedPAC Speaks

I once called MedPAC the most influential organization in the medical arena that most Americans have never heard of. That was in a blog post from April, 2009 called “Follow the Money.” I continue to find MedPAC—the non-partisan, independent congressional agency that advises on issues related to the Medicare program—an excellent source of information and analysis. Its 17 members, representing academia, public health, medicine, and industry, make recommendations to Congress for revisions to the Medicare program twice a year. The March report is called “Medicare Payment Policy” and the June report is “Medicare and the Health Care Delivery System.” This month’s report contains more than the average share of interesting material. 
There’s a section specifically mandated by Congress that analyzes the Hospital Readmissions Reduction Program (it found, on average, that the program led to a 3.1 percent decrease in readmissions for five major medical problems, although it couldn’t be sure that the program caused the drop). Another section addresses how to improve Medicare’s approach to paying for medical devices (among other tactics, limit the ability of physician-owned distributors through greater regulation and accountability).  There’s a chapter on Accountable Care Organizations (possibly effective in controlling costs and improving quality, but with various modifications proposed). But what intrigued me most was the final chapter on “Medicare coverage policy and use of low-value care.”
MedPAC defines “low-value care” as services (including treatments, medications, and tests) that are of “little or no benefit” and where the risk of harms exceed the potential benefits. That, of course, is a value-laden definition since it assumes “benefit” can be clearly established and quantified, ditto for “harms,” and that we know the probability of either benefit or harm. Nonetheless, a remarkable number of treatments, medications, and tests are widely agreed to fall into this category. They comprise the list of no-no’s compiled by specialty professional societies for the Choosing Wisely campaign which, in its first five years of operation, came up with 525 recommendations. 
For frail elders, incidentally, the relevant professional organization inventoried by Choosing Wisely inveighed against gastrostomy tubes in people with dementia, obtaining urine cultures for people without urinary symptoms, use of lipid lowering agents in people with a limited life expectancy, screening for breast, colorectal, or prostate cancer in individuals with a life expectancy of less than ten years, hospitalization without taking into account goals of care, and aggressive blood pressure regimens.
The first observation MedPAC makes about low-value care is just how widespread it is among Medicare fee-for-service beneficiaries. Using two measures developed by Aaron Schwartz, Bruce Landon, and colleagues, the commission estimated that between 25 and 42 percent of patients over age 67 received at least one value service per year, costing Medicare between $1.9 and $8.5 billion. Imaging tests, cancer screens, and other diagnostic and preventive tests accounted for the lion’s share by volume, with cardiovascular tests and imaging the major offenders in terms of cost.
MedPAC concludes by considering six “tools” that could improve the situation. The commissioners raise the possibility of expanding the use of pre-authorization, a much-despised approach used by prescription drug plans and hospitals to limit the use of expensive medications. This tactic would create administrative obstacles to the use of various tests and procedures—and would further the bureaucratization of medicine deplored by many physicians. 
They also propose implementing “clinician decision support” programs and provider education, claiming that inappropriate antibiotic prescribing has been reduced in this way. No one could disapprove of trying to make clinicians wiser and more knowledgeable—the question is how to achieve this. Many of the usual “decision support” and “education” programs have been of dubious benefit. 
Another option is increased cost sharing. This is a favorite approach of commercial insurance programs but mercifully has rarely been used by Medicare—it places the burden for deciding whether a test is worthwhile squarely on the shoulders of patients, often without providing them with the information necessary to decide. It encourages wealthy people to squander their resources on low-value tests since the cost-sharing involved is trivial for them, and it may distort the decision-making of the less affluent by leading to their accepting cheap but useless tests and rejecting expensive but useful ones. 
Then there’s the strategy of revisiting Medicare’s “national coverage decisions” periodically—for the small number of tests in which the choice about coverage is made at the national level. The problems with this approach are that it fails to address the vast majority of coverage decisions that are made at the local level and that the reality is that revisiting coverage decisions has historically led to Medicare’s paying for more procedures, not fewer.
That leaves only two options to MedPAC. The first is introducing new payment models in which providers are held accountable for both cost and quality. This is the rationale behind Accountable Care Organizations and it may be working, according to MedPAC. It’s not entirely clear since the organizations that were able to lower the use of low-value care by and large choseto belong to an ACO so there may be something about the organization, not the ACO structure. The second option is to introduce cost effectiveness into determinations of Medicare payment.
MedPAC doesn’t explicitly endorse balancing cost and benefit in determining whether Medicare will pay for a service and if so, how much it will pay. It is at pains to review the long and sad history of attempts to introduce cost into Medicare’s decision-making. In fact, MedPAC includes a 14-page “primer” on Medicare’s decision-making process, with several highlighted sections on the failed efforts in 1989 and again in 2000 to use some form of cost-effectiveness. Both attempts were effectively derailed by the device industry that lobbied heavily against a strategy that they knew would result in restrictions on their ability to sell any device that is “safe and effective” (the FDA standard) at whatever price they wish.
But why, if including cost in coverage decisions is out of the question, does MedPAC go on and on explaining what cost effectiveness analysis is, reviewing the ethical objections to the use of QALYs (Quality-Adjusted Life Years), and detailing the history of efforts to define “reasonable and necessary,” the phrase used in the original 1965 Medicare legislation to describe what Medicare must cover? Why all the verbiage devoted to cost-effectiveness analysis in the commercial sector and by the non-profit Institute for Clinical and Economic Review? 
I have to assume that the reason is that the MedPAC commissioners recognize that while better physician education and new improved ACOs may be of some use, the most effective strategy is for Medicare to refuse to pay for “low-value care” when that is a euphemism for “useless intervention.” It is true, as the report indicates, that there is no explicit statutory provision including cost as a component of Medicare’s coverage decisions. But neither is there any statutory prohibition. It is time to try once again to exploit the loophole, perhaps using the language of “value” that is currently in vogue—after all, who could argue against choosing treatment that is valuable?

June 25, 2018

Reforming Medicare: Enhancement or Evisceration?

Recent reports indicate that Congress will try to slash Medicare in order to balance the budget—making older people and disabled people shoulder the cost of its enormous tax cuts. The proposed plan, according to the Washington Post, would extract $537 billion dollars from the Medicare program over the next decade. At the same time, the budget passed by Congress and signed into law by the president in February created CHRONIC (Creating High Quality Results and Outcomes Necessary to Improve Chronic Care Act) which, the NY Times suggests, is a hidden jewel buried in the voluminous budget bill. Which is it? Is Medicare headed for enhancement or for evisceration?

Thus far, the cuts are theoretical (it’s not clear that the House Budget Committee will get very far with its recommendations) whereas the reforms are real—or will be when they go into effect in 2020. CHRONIC is to be lauded for accomplishing several important goals. 

First, the act recognizes that good outcomes among people with chronic conditions are contingent on what are not strictly medical services—as well as access to physicians, hospitals, pharmaceuticals, and medical devices. Wheelchair ramps and grab bars, as well as other products that lie outside the traditional definition of “durable medical equipment,” can in the future be paid for by Medicare—at least by Medicare Advantage plans that opt to offer them. This strategy does not go as far as the National Health Service did in the UK with its “personal health budgets” that allowed patients or their representatives to decide in concert with their physicians how to spend their share of the health care pie. The result—and the program, despite some vocal protests, has been so successful that it was recently expanded—is that patients with early dementia can choose, for example, to spend NHS money on creating a garden that will keep them engaged, potentially obviating the need either for medication to control symptoms of agitation or for institutionalization in a nursing home. Nor does the US strategy apply to traditional Medicare: in an effort to make Medicare Advantage plans, which currently have 19 million members, even more attractive (furthering the Republican goal of privatizing Medicare), only MA plans will be allowed to reimburse for these new supportive services. 

Second, CHRONIC permanently authorizes Medicare Special Needs Plans (SNPs) that cater to the highest risk Medicare beneficiaries including those living in institutions. These are special types of Medicare Advantage plans that offer enhanced integration and coordination of care, a critical feature for this complex population. 

Finally, CHRONIC extends its support of non-traditional forms of care, of which the allowance for grab bars was one example, to telemedicine (particularly relevant for homebound patients and in rural communities) and to home care (expanding the Independence at Home Program 50 percent from 10,000 enrollees to 15,000). 

So, what’s not to like? Two cautionary notes. First cautionary note: CHRONIC focuses overwhelmingly on Medicare Advantage plans (which currently cover 32 percent of Medicare beneficiaries), not on traditional Medicare (which covers the other 68 percent). This is no surprise, as the Republican Congress, which is interested in privatizing Medicare, sees shifting to the MA model as a route to achieving this goal. In principle, I don’t have any problem with expanding the number of MA plans (currently there are 3300, according to MedPAC, the Medicare advisory council) as they offer great potential for the coordination of care so essential to frail elders, but it will be essential to maintain the regulatory oversight of CMS if these plans are to be guaranteed to provide quality care. Moreover, we need to begin collecting detailed data on the utilization and outcomes of MA members. Right now, almost all of the voluminous data gathering by the federal government exclusively deals with fee-for-service enrollees so no granular analysis of the performance of MA plans is possible. 

Second cautionary note: while some of the provisions of CHRONIC appear to address programs, in fact the legislation is often grounded in how the programs are to be reimbursed. And the underlying philosophy is that the way forward lies with “value-based” care. I’ve blogged about this before, most recently in my post “V is for Value.” My concern about this approach is that it assumes that better and less costly medical care can be obtained simply through tweaking reimbursement. It’s the triumph of the economists’ view of health care as an industry subject to manipulation like other industries. The trouble with this insistence that VBP is key to all our problems, aside from the fact that so far value-based reimbursement systems such as pay-for-performance have not succeeded, is that it discounts the role of culture, advertising, and popular expectations. It is these factors, and not just payments to physicians, for example, that shape the enthusiasm for technology manifested by patients, corporations, and physicians alike.

CHRONIC is an admirable piece of legislation—for what it includes. What should concern us, however, is what it leaves out.

June 20, 2018

Much Ado About CPR

Two short essays in this week’s New England Journal of Medicine present differing perspectives on how to treat a desperately ill woman w in the ICU of an American hospital, a woman who is unable to make her own decisions and has no advance directive. One author responded that the right course of action was to “institute a DNR order” and the other that it was to “continue full resuscitative measures,” setting up a quasi-debate. I say a quasi-debate because the two authors, as per the editor’s instructions, simply laid out their own arguments without responding to the other’s point of view. I know, because I am one of the two.

I asserted that the physician should declare the patient DNR—a woman with no written advance directive and no designated health care surrogate who was dying of multiple organ failure triggered by sepsis and unresponsive to maximal medical therapy. Attempted CPR in this setting, I argued, was as close to futile as medical treatment ever is; moreover, the patient had clearly stated to a friend and neighbor that she would not want aggressive measures when facing overwhelming odds. I stand by my arguments, which I won’t repeat here, but I regret that the discussion focused on the wrong question.

The wrong question is whether the patient should have a DNR status; the right question asks about the general treatment strategy rather than a specific medical intervention and about the process for deciding, not just the outcome of the process. It moves the discussion beyond considerations of futility—a debated that raged in the 1980s and ultimately led to the recognition that futility is meaningless without reference to what treatment is intended to accomplish. Maintaining a patient who is in a persistent vegetative state on artificial nutrition is futile if the goal is to restore her to full functioning as a thinking human being. But it’s entirely appropriate if the goal is to sustain life in the sense of a heart that beats and lungs that respire. Discussing how to make medical decisions for patients who have lost capacity and how to think about treatment for those who are dying allows us to discuss much more important decisions than whether or not to attempt CPR.

Ever since 1984, when a New York hospital was found guilty of putting purple dots on patients’ charts signaling that they were not to be resuscitated—without their knowledge or input or that of their next of kin—we’ve been obsessed with DNR orders. I suspect that more ink has been spilled on whether, when and why to write a DNR order than on any other medical decision. The reason, presumably, is that it seems irreversible: if a patient sustains a cardiac arrest and CPR isn’t performed, the patient will surely die. Other seemingly momentous decisions often prove not quite so momentous—even Karen Ann Quinlan, the first patient to bring the possibility of not intervening medically to public attention, lingered for nine years after her “life-sustaining” ventilator had been disconnected. A choice not to resuscitate is far more unambiguous—although CPR, it should be pointed out, is hardly a guarantee of life.

Cardiac arrest may have special cachet because it seems to divide the living from the dead. But in fact, for older patients who are critically ill, it is often merely the last step in what is often a relentless progression of markers on the way to death. From the perspective of human suffering and of doing good (and avoiding harm), other steps along the way are often far more significant. After all, a person whose heart has stopped and who is not breathing is no longer able to experience anything, neither the existential angst from awareness of imminent death nor the physical discomfort of aggressive treatment. The nurses and doctors who attempt CPR in a dying patient may feel guilty of assault; they often regard intubation, chest compressions, and electrical shocks as an undignified way to end life, but they are the ones experiencing malaise, not the patient.

Subjecting a patient who is in the final phase of life to dialysis for four hours a day, three times a week, or to major surgery to repair a damaged heart valve, or to ventilator care for advanced lung disease, by contrast, may cause pain and suffering to the person receiving treatment. Both the potential benefit of treatment (longer life) and its potential burdens (ranging from delirium and functional decline to a shorter life) are difficult to assess quantitatively. These decisions, which are difficult to make with a competent patient and correspondingly more fraught with an incapacitated patient, are the ones we should be discussing, not CPR.

Ironically, despite the now 42-year history of DNR orders, there is still confusion about whether DNR indicates that CPR will not be performed in the event of a cardiopulmonary arrest (as it is defined by the American College of Cardiology and other authorities) or something more. Indeed, the opposing piece in the New England Journal of Medicine talks about a time-limited trial of “continued resuscitative measures,” which presumably refers to ongoing ventilator treatment for respiratory failure, dialysis for kidney failure, and fluids and antibiotics for sepsis, rather than for a few minutes of attempted CPR. This uncertainty about what DNR means reflects another problem with the obsession with “getting the DNR order,” and that is the widespread belief that treatment decisions must be all-or-nothing: if you don’t want CPR, then surely you don’t want any potentially life-prolonging medical interventions. The reality is that many people don’t want burdensome treatments that are extremely unlikely to be beneficial, but that doesn’t mean they want to focus exclusively on comfort. 

There’s much more to discuss—I haven’t even touched on the process for making medical decisions. That includes not only who the surrogate should be in the event of decisional incapacity and what standard that surrogate should use for making decisions (substituted judgment is the usual standard, with best interests the back-up if no information about the patient’s preferences is available), but also how the discussion should unfold. Typically, physicians are so eager to reach a decision about what to do that they jump to this step in the process without first clarifying the patient’s underlying health state (in the NEJM vignette, the crucial bit of information is that the patient was dying), as though preferences are independent of any broader context. But for now, let me leave you with the thought that both medical decision-making and advance care planning involve much more than checking off items on a menu, and that if we do engage in intervention-specific planning, CPR should be low on the list of what we discuss.