• End of life savings will likely disappoint

    Andrew Sullivan picks up Chris Conover’s post on the other 1%–health care costs–where 1% of the population accounts for 20% of the spending.

    Andrew asks:

    I wonder how much of the cost for the top 1 percent in costs is spent in the last few days or weeks of life?

    Focusing on Medicare, which insured around 8 in 10 persons who died last year, the proportion of total Medicare expenditures in the last year of life has been fairly stable at around 1 in 4 dollars for 3-4 decades. From a recent Annals of Internal Medicine paper by Amy Kelley and others, a focus on the last 6 months of life shows:

    Median total Medicare expenditures in the last 6 months of life, adjusted to 2006 U.S. dollars, were $22 407 (range, $0 to $391 773; interquartile
    range, $9087 to $40 802).

    While difficult to directly compare all these numbers, the 1% identified by Conover could be thought of as the sickest subset of a very sick subset that is heavily focused in the Medicare population. Seeking to reduce end of life spending in the Medicare population seems like an obvious place to start in order to address our health care cost problems, but I have termed such savings the Fools Gold of health reform:

    It is doubtful that a focus on reducing EOL spending per se will result in as much savings as is often assumed, for one simple reason:  The concept of the last year of life is inherently retrospective.  You do not know when the last year of someone’s life started until it ends.  The stylized fact that leads to the assumption of wasteful EOL spending., i.e., 1 in 4 dollars spent on care in the last year of life, is based on an inherently retrospective concept that does not translate easily into the prospective decision-making that would be needed to reduce wasteful, futile or harmful spending in the last year of life.

    This doesn’t mean there is nothing that can be done. In fact, I have shown with colleagues that hospice reduces Medicare costs by around $2,300 from the point of hospice initiation as compared to what costs would have been without hospice (a 6-7% reduction in last year of life costs in our sample). The biggest daily savings due to hospice come in the last few days of life (see figure below) when hospice users avoid hospitalizations that are typically common near death, and maximum savings were realized in our study with a hospice stay of around 8 weeks (median hospice stays are today around 16 days). Thus, around 7 in 10 persons using hospice would reduce Medicare expenditures by using hospice for longer before death, and a little over 4 in 10 Medicare decedents now use hospice prior to death (see page 266 of this report).

    There are things we can do to reduce costs at the end of life like seeking to expand hospice use, but we likely need a more systematic way of asking whether care is “worth it” throughout the system in order to develop a sustainable health care system. Only focusing on end of life spending will likely disappoint because doing so will require us to determine when people are near death AND to then change how they are treated. The technical problems are hard enough, but the cultural and political ones inherent in this strategy are enormous. I expand in great detail on these issues and my view that our nation has a cultural fear of death that enables the politics of health care cost control and rationing to be so potent in my book (ch. 3-6). We need to try everything, but I think that focusing on end of life spending as a cost control strategy is unlikely to be a panacea for our cost problem.


    Amy S. Kelley, MD, MSHS; Susan L. Ettner, PhD; R. Sean Morrison, MD; Qingling Du, MS; Neil S. Wenger, MD, MPH; and Catherine A. Sarkisian, MD, MSHS. Determinants of Medical Expenditures in the Last 6 Months of Life. Ann Intern Med. 2011;154:235-242.

    Donald H. Taylor, Jr. Jan Ostermann, Courtney Harold Van Houtwen, Karen A. Steinhauser, James Tulsky.  What Length of Hospice Use Maximizes Expenditure Savings: Evidence from the U.S. Medicare Program.  Social Science and Medicine 2007;65:1466-78.

    Kathleen T. Unroe et al. Resource Use in the Last 6 Months of Life Among Medicare Beneficiaries with Heart Failure, 2000-07. Arch Intern Med 2011:171(3):196-203.

    • “Only focusing on end of life spending will likely disappoint because doing so will require us to determine when people are near death AND to then change how they are treated. The technical problems are hard enough, but the cultural and political ones inherent in this strategy are enormous.”


      I’ve always found that the observation that most healthcare spending takes place near the end of life about as helpful as the observation that most food consumption takes place at meal-times.

      Once you start parsing the end-of-life spending it becomes very difficult to disaggregate useful spending from futile spending, and even that normally requires making a series of thorny value judgments that are unlikely to be universally shared and thus impossible to compile into an algorithmic set of national guidelines.

      As a general rule of thumb, if something is impossible, it’s not wise or prudent to attempt it. Formulating universal guidelines for end-of-life care that actually reduce spending is not practically, politically, or ethically achievable in a system in which third parties pay the entire bill.

      The only factor that I can see having an impact upon the choices that individuals and families make under the current regime is setting a hard ceiling upon lifetime payouts per-individual under Medicare/Medicaid, beyond which the estate is billed for every dollar until the estate is exhausted.

    • Isn’t the simple truth that there are no panaceas for cost control? One of our problems is that too many people keep looking for the silver bullet. It doesn’t exist. Cost control is going to take end of life and hospice care AND reducing wasteful errors AND better coordinated accountable care AND getting rid of administrative waste and fraud AND reduced malpractice costs and defensive medicine AND…It’s time to quit looking for the easy answers and start doing ALL the hard stuff.

    • @DShea
      Yep. Barrier is the lack of a political deal that makes all the hard stuff the task of both political parties

    • If we can only marginally control end of life costs and the other 4 percent of the population that costs another 25 percent is most likely severely ill, what hope is there to control costs at all? Either we raise taxes dramatically or institute cost limits on users and providers. There is no other way absent going to a universal health plan; which, once we all come to accept the staggering costs in front of us, may finally come to pass. Looking at the figures realistically, is there really any way to reduce the cost fo the other 95 percent of us; meaning what can be squeezed?

    • Don
      Well stated.

      You mentioned that 7 out of 10 patients using hospice longer would reduce the Mcare spend.

      Embedded within that notion are the same problems you speak of so eloquently in your post, mainly, acceptance–both cognitively and clinically–of EOL care.

      To promote faster transfer to hospice would require clinician engagement in the “conversation” much earlier than present. Intervening in hospital before the death spiral begins means ambivalent prognostication (read: avoidance), as opposed to the retrospective certainty of savings that invariably accompanies our Monday morning quarterbacking…after the death. The quarterbacking is the imaginary savings we all want, but will never achieve in full. The tools are simply not there. JayB above makes the point well.


      • @Brad F
        Medicare hospice policy a bit chaotic now as well. Lots of worry about very long lengths of use (~5% use for more than 180 days), while 25% use hospice for 5 days or less. MEDPAC and congress more focused on very long stays out of worry of fraud and/or development of back door LTC benefit. Need to both increase short stays (all the problems you note; 25percentile of hospice has been 5 days for around 15 years) while slowing very long ones. Even moving the 25 percentile from 5 to 10 days would have some effect and almost certainly improve QOL as well…it is just hard to figure out how to do it.

    • Don–

      Do you know if we have end-of-life-care data that incorporates the age of death. In other words, are we spending more money keeping the handful of 40-year-olds alive who have a terminal illness, or is the money instead going toward extending the life of thousands of 85-year-olds who have already lived longer than the average person?

    • Here is a way to solve end of life spending problems that avoids the retrospective issue.

      Health care rationing through self-selection

      Background facts:
      Patents expire after 17 years.
      The time between death and retirement is a little bit more than 17 years
      Health care spending is concentrated in the last few years of life.

      At age 67, each citizen fills out a simple form, checking one of two boxes:

      Option 1: For the rest of my life, I would like access to all healthcare technology, procedures and drugs that are available today, my 67th birthday. My out of pocket expenses will be limited to a $20 copay for each procedure and prescription. I will have no coverage for any technologies, procedures or drugs that are developed and FDA approved after today’s date. However, I can pay for them out of pocket if I chose.

      Option 2: For the rest of my life, I would like access to all heathcare technology procedures and drugs that are available today, plus any that are developed in the future. My out of pocket expenses for all procedures and prescriptions (both existing and new) will be 50% of the actual price

      – If the relative out-of-pocket percentages are set correctly, a large majority of people would likely choose Option 1
      – This would reverse the current trend of allocating the most expensive healthcare technologies (i.e. those still on patent) to the least able to pay (those in the last years of life)
      – Rationing via this self selection process may be more politically palatable than market/price signals, HMO restrictions, “death panels” or general supply constraints caused by price controls
      – It is a bit difficult to argue that it is cruel to limit somebody to the best healthcare available today (or whatever “today” is when they turn 67).
      – If some magic pill were invented to cure all disease, or if there is a new procedure that is significantly less costly than a prior procedure, an exception could be made and it could be included for Option 1 participants (although this could create a slippery slope)
      – There is a nice symmetry in that people get free access to all the technology developed by their age cohort, but have no right to technologies developed by subsequent generations.
      – Every citizen would have chosen their path, and thus would have less basis to complain about injustices (of course, complaints would still occur, but perhaps be less impactful)

    • I have never seen a person get surgery over the age of 80 and not regret it. Of course that is only 2 out of 2. Not so many people reach that point. And having worked for a bit as a CNA, there are many folks kept alive as old zombies, some it seems against their will. As individuals we need to control our end more.

    • A comparison with another country with advanced health care would be enormously instructive.

    • I once had a conversation with a doctor about spending in the last year of life. His response was to invite me to walk with him down the hallway of his hospital.

      “Which patients are going to die in the next six months,” he asked. “You tell me which ones and I’ll be glad to focus on reducing the cost of their care.”

    • -Marginally off-topic but the institutional attitude towards who can give DNR orders, who needs to be notified, and who needs to consent seems significantly different in the NHS.

      -Spouse was working as a physician in NZ and noted much the same attitude. One could argue that this occurs in a cultural setting that grants implied consent by default and no one expects to get the full court press if they enter the hospital with a potentially terminal illness over a certain age – but I can’t imagine folks in the US granting physicians that kind of lattitude, or all physicians wanting to exercise it. Interesting question for me is whether the difference practice with regards to DNR orders predates or is a consequence of single payer regimes.

      • @JayB
        “Interesting question for me is whether the difference practice with regards to DNR orders predates or is a consequence of single payer regimes. &lt”;—that is interesting question, which came first, etc.

    • @Don

      This is a very helpful conversation, and I thank you for it. I work as an internist, and I would like to share a few observations:

      1) Folks are right to point out we’re bad at prognosis (i.e. predicting exactly how long someone has)
      2) That said, there is tremendous waste involved in end-of-life care (not just expense, WASTE). This is particularly true if you look at the intensive care unit. My understanding is that we are heavier utilizers than other industrial countries, and that patients with terminal illnesses are more likely to receive ICU care in the States. I’d welcome any statistics you have on this. It seems like an obvious target for savings.
      3) The culture of medicine – and oncology in particular – is bad at addressing prognosis and goals of care. If you want to get the hospice use numbers up than you need the doctors caring for these patients to view that as an appropriate and timely part of their job.

      • @Jonathan
        I am going to be blogging some more on this topic next week and will keep these questions in mind. We are doing a study in which patients with cancer and their caregivers discuss preferences for care if there are resources constraints. One interesting aspect of the study has been the differential response of particular oncologists and/or different clinics in terms of their willingness to refer patients to the study; some have a great deal of reticence because of the topic. Thanks for writing with your observations.

    • -Useful observations Jonathan, but there’s a fine line to walk at the center of an ethical landscape that’s bounded by absolute medical futility on one side and unsentimenal euthenasia on the other. The comments below aren’t directed at you, just generaly observations regarding doctors and end-of-life care.

      Unless the life being saved and the money being spent are the physician’s personal property, the subjective value judgments about cost and quality of life are solely the province of the patient, and of the family if the patient is no longer in a condition to do so.

      I have an immense amount of respect for the medical profession, but neither the life of the patient nor the money being spent are properties that they’re entitled to exercise personal discretion over. Even in the case of a single-party system, where the funding and thus the de-facto control of the patient’s body reside in the state – the locus of control belongs elsewhere.

      Physicians can ethically serve as highly trained medical technicians, consultants, and confidants but aren’t entitled to determine who lives and dies, and when. I’d feel less strongly about this if I hadn’t been privy to conversations where physicians let their workload or their personal sentiments shape their conduct in ways that crossed this line. Those lines get especially blurry in a single payer environment where physicians are forced to be stewards of the public purse.

      Overworked young interns slapping DNR orders on critically ill elderly patients without the knowledge or consent of either the patients or their families crosses dozens of bright red ethical lines but my observation is that the institutional incentives that physicians operate under in that environment make such conduct much more likely.

    • @JayB

      I appreciate your comment and share completely your concern about doctors playing God. That said, an intensive care unit intervention is a series of aggressive medical procedures. I know it because I’ve been the one doing it. In fact the intervention – not the step back – is the ultimate loss of autonomy for many patients, and most don’t realize this when they enter into it. For some patients there’s good evidence that this works (i.e. it prolongs life or reduces sufferring). In many cases there’s no evidence that it does anything.

      I would argue that we need to determine in what circumstances these interventions are justifiable based on evidence. In situations in which there’s no chance that an intervention can prolong life or improve quality, it shouldn’t be offerred. To put it differently, we don’t offer people appendectomies when they don’t have appendicitis, because even if they want one it’s not going to help them. It’s not medically indicated. So why are we offering them CPR, intubation, and ventilator support when they have metastatic non-small cell lung cancer?

      I believe in patient autonomy to a point, but I think we’ve crossed a line and are now inflicting a lot of suffering without enhancing quality or longevity. We shouldn’t feel obligated as a country to do things that don’t work and that are diverting resources from where they would be more usefully deployed. We certainly shouldn’t demand that tax payers finance such a historically unprecedented and perverse set of priorities.

      Given the impossibility of reaching universal agreement on where to draw these lines, there ought to be a system that allows a certain amount of flexibility. For example, if you are a strongly religious person, and you believe that medical intervention should be carried out until the last possible breath, there ought to be an insurance plan that caters to you. Certainly there are many doctors who would be happy to provide the care. But I see no reason whatsoever why the federal government should be expected to pay for this.

    • One helpful thing would be for Medicare to give greater EOL access to Hospice care for degenerative illnesses of old age such as the various dementias (e.g. Alzheimer’s and Lewy-Body Disease). Hospice has been a godsend for my mother (who has LBD, which includes the Parkinson’s disease which many people are familiar with).

      When some of the typical LBD things happen, the staff at her facility can call hospice instead of sending her to the ER. She does not go to the hospital, and staff combined with hospice ensure that she gets the care which she really needs which is generally bed-rest, pain relief, love, and attentive care.

      What she does NOT need is being shipped to the hospital, having her medications disrupted (which provide much needed comfort and symptom relief), and lots of scans and tests to look for things that nobody is going to treat anyway even if they are found just to satisfy the academic curiosity of some hospitalist.

      And goodness knows the taxpayers really shouldn’t be paying for all that intervention. As pointed out above, hospice is a bunch cheaper. So – what’s the problem? Simple. With diseases such as LBD the decline isn’t always smooth – it’s usually described as a “roller-coaster”. So you could get easily get a 2-3 month period when someone is “stable” so they get kicked out of hospice. (And since in hospice no one is sending them to the hospital to be tortured, then they might well stablize for a while).

      Once they are no longer in hospice, the next thing you know that person ends up in the ER – which A) costs everyone bunches of money and B) is very bad for the patient and causes them to then have a sudden and significant decline – with significantly decreased comfort and dignity on top of it.

      Thanks for letting me rant a bit! I live in constant fear that my mother will get kicked out of hospice.

      • @Janet
        I am sorry about your mom. There are some pilot projects in the Affordable Care Act to allow experiments with opening the rules of hospice (not having to certify < 6 months to live)....what are called concurrent care demonstrations,with concurrent meaning getting expanded hospice/pall care without unelecting something else. There are some tests done in younger than 65 commercial insurance by Aetna that had good results (improved quality, satisfaction and some cost savings and not increase which is the worry). One interesting tidbit is that expansions of hospice/pall care tend to be expected to reduce costs in addition to other benefits. There is nothing else I can think of that has this standard (usually we are left wondering if increased cost is worth it).