The Ex-Post Dead Are Not Ex-Ante Hopeless

It’s well known that a large faction of medical spending occurs in the last 12 months of life but does this mean that the money spent was fruitless? Be careful as there is a big selection effect–we don’t see the people we spent money on who didn’t die. A new paper in Science by Einav, Finkelstein, Mullainathan and Obermeyer finds that most spending is not on people who are predicted to die within the next 12 months.

That one-quarter of Medicare spending in the United States occurs in the last year of life is commonly interpreted as waste. But this interpretation presumes knowledge of who will die and when. Here we analyze how spending is distributed by predicted mortality, based on a machine-learning model of annual mortality risk built using Medicare claims. Death is highly unpredictable. Less than 5% of spending is accounted for by individuals with predicted mortality above 50%. The simple fact that we spend more on the sick—both on those who recover and those who die—accounts for 30 to 50% of the concentration of spending on the dead. Our results suggest that spending on the ex post dead does not necessarily mean that we spend on the ex ante “hopeless.

…”Even if we zoom in further on the subsample of individuals who enter the hospital with metastatic cancer…we find that only 12% of decedents have an annual predicted mortality of more than 80%.

Thus, we aren’t spending on people for whom there is no hope but it doesn’t follow that it’s the spending that creates the hope. What we really want to know is who will live or die conditional on the spending. And to that issue this paper does not speak.

Comments

Similarly, studios and publishers would presumably benefit from concentrating their spending only on the movies and books that will be hits. The trick is identifying ex ante which those will be.

The Ex-Post Dead would be a good name for a band, but only highbrows would get it.

Why no citation, or even mentioning of who the authors of this paper are?

+1 to no citation. It's hard to critique a paper one cannot read.

A lot of people who get an original diagnosis of Stage 4 cancer of some type are given aggressive chemo despite the fact that the outcomes are seldom positive (survival past six months which is the general clinical marker for a positive outcome). Claims data alone will not have such information.

There has been virtually no improvement in long term morality from cancer, despite the $billions spent since the “War on Cancer” was declared. What’s worse, is that the system is incentivizing “miracle” cures, when the real cure is at the end of a fork. Eat a whole food plant-based diet and your risk of cancer plummets. Not much money in Big Broccoli. Yet, our entire food and medical system is designed to have people eat crap and “cure” the result with a magic bullet. It’s totally insane and tragic. My advice is to save yourself and family with lifestyle choices and let Rome burn. Too many people make money from this cycle of death and disease to stop it soon.

“Eat a whole food plant-based diet and your risk of cancer plummets.“

Lol, no.

Lol, yes

http://science.sciencemag.org/content/360/6396/1462
0
0

Report
Predictive modeling of U.S. health care spending in late life

Liran Einav1,2, Amy Finkelstein1,3,*, Sendhil Mullainathan1,4, Ziad Obermeyer5

'Predictive modeling of U.S. health care spending in late life'

So, this actually says very little about health care costs of people not predicted to die in 12 months who are not included in Medicare.

In other words, we are excluding the medical costs of all under 65 accident victims who live for several days, weeks, or months after the accident. And of heart attack or stroke victims who are younger than 65.

Yep, sounds completely definitive in the way that a carefully defined set of data always is.

Given that it is mostly old people who die, I wouldn't categorize this analysis as narrow (as you seem to imply).

"Death is highly unpredictable." So much for medical science.

Come to think of it, why do medics consider themselves equipped to harangue us on eschewing dietary fats - or whatever the cause célèbre of the day may be - when they are useless at predicting death?

A truly idiotic statement, surpassing your baseline level.

Because 'statistics'.

Yes, please cite the paper. It is an interesting idea, but I have doubts about a paper that depends on the nominal probabilities from a machine learning model. It is irrelevant whether the probability of death is more than 50% (or 90% or 10%) - what matters is the ordering and whether the higher probabilities are, in fact, more likely to die. So, the paper should not be based on what fraction of the people had probabilities > 50%.

"what matters is the ordering and whether the higher probabilities are, in fact, more likely to die."

This is not the point of the paper (though they do answer that question). The authors write, "That one-quarter of Medicare spending in the United States occurs in the last year of life is commonly interpreted as waste." The point is if you can predict reasonably well who will die, you might be able to reduce spending. But what they find is that even if you lower the prediction threshold to 50%, the people above that threshold only account for 5% of the spending. Moreover, using their methods, they can't even predict death that well. "[W]e find that only 12% of decedents have an annual predicted mortality of more than 80%."

I haven't yet been able to access the paper, but my point is that the probability values are meaningless. If only a small fraction of Medicare recipients die each year, few of the observations will have predicted probabilities above 50%, yet alone much above it. It wouldn't shock me if almost everyone had a probability less than 50%. So, if few are above 50%, then obviously they will account for only a small % of Medicare spending. What we want to know is if we look at the highest x% of the predicted probabilities (not using any arbitrary threshold), what % of the spending is accounted for.

I gotcha. Basically, even if we couldn't get an accurate measure of the predicted probability of death, as long as the rank order was fine, then we could use that to potentially understand whether money could accurately be withheld in the last year of life.

Put another way: if the highest 10% of predicted probability of death (whatever the range) covered 90% of those who died, the algorithm could still be quite useful.

I would make the distinction between seniors with an acute (or episodic) illness and seniors with a chronic illness (the latter including heart disease, cancer, arthritis, etc.), but that's seldom done. Of course, old age is a chronic illness (there's no cure for it). Almost everything about our health care system is based on a model designed for acute (or episodic) illness not chronic illness. Indeed, ACA, for all of its improvements, did not actually address chronic illness - recurring annual deductibles and co-pays over an extended period can break even those with significant savings. But what about the improvements in survival rates among cancer patients, isn't that progress for that chronic illness, especially when survival rates in America are compared with those in other countries? It's true that survival rates in America are higher, but it depends on the meaning of "survival". Survival is measured from the date the cancer is detected to the date of death, not the from the date the cancer first developed to the date of death. With all the diagnostics and so-called preventive care in America, detection is much earlier than in other countries. Cancer patients don't necessarily live longer, but they live longer knowing they have cancer.

This article only touches the tip of the iceberg. The more significant issue is the reason the aged become chronically ill in the first place. The standard American diet (SAD) is, without any question, one of if not the worst in the world. Chronic illness takes years to develop (cancer, atherosclerosis, Type II diabetes, etc.) and diet is the primary culprit. Worse than the final year of life are the decades spent treating these entirely preventable diseases. Until the US population adopts a more whole food plant-based lifestyle (minimize refined and processed foods and animal products), there will be no relief. Big Pharma, the processed food and animal industries, along with the medical industries and politicians they have in their pockets will see that the current system of creating and treating sick people will never change until then.

+1.
Not much chance of that happening.
SAD: Standard American Diet.
Nice acronym!

I recall the lack of sympathy for people with AIDS, who were blamed for their illness and, hence, weren't deserving of either sympathy or the funds for treatment or finding a cure. I'm not sure if that's unique to America, but the attitude extends well beyond AIDS to heart disease, cancer, diabetes: if they had gotten exercise or had a better diet, they wouldn't have the chronic illness. Even the chronically ill who lead a puritanical life and die are described as having "lost" the "fight". Death doesn't become America. And neither does illness. Of course, one can make the analogy to "losers" in the marketplace, who don't deserve public assistance. In life and in death the American attitude is you are responsible for and deserve your fate. Have a nice day.

Resources are finite, and socialized anything becomes impersonal quick. If it's a choice between an otherwise healthy 40 year old with lymphoma and a 65 year old obese diabetic who's eaten french fries his whole life, society should choose the former to keep the common weal solvent. I wish we lived in the Garden of Eden too. Recognition that we don't is part of adulthood.

Socialized health care is rationed health care. Therein the state (death panels) not the market decides on who lives.

It's supply and demand. Money in and money out. All are finite. But, demand (relatively) tends to outstrip supply especially when the government takes over, interferes, intervenes; and/or does not do its duty, e.g., de facto amnesty and open borders for 12 million.

"If the government took over the Sahara Desert, in five years they'd run out of sand." That is why I am a conservative independent.

For some on the left, it's wrong to spend money (to add one-year, say) on an old American who "by accident of birth" (saw that too many times on the web) worked for 50 or so years, paid taxes and health/Medicare premiums for 50 years; maybe fought for her/his country. Why should he/she get anything?

They need the relatively-limited health-care funds for illegal crooks/invaders (addicted/alcoholic, dishonest, uneducated, violent) when they shoot or stab each other; or, like I witnessed in an ER, when Pedro breaks over Jose's head a length of iron pipe.

Free market health care is rationed just as much as socialized health care.
Actually, since it is rationed by who can afford it, free market health care may be rationed even more than socialized health care.

Not really.

Price allocation means that excess demand generates income and typically investment. Even if the rich have better care, they move resources into the system which ultimately increases the ability of the system to care for all. For example if we both have hernias and I pay more I go first, but if I pay enough more the surgeon may work an extra shift or hire another surg tech to be able to maximize his income.

Government or your typical insurance system allocates based on waiting and paperwork. These are simple deadweight losses. For example we both need hernia repairs. So we both weight. The surgeon does his quota each day and we wait for when our turn comes. This means we live with pain and the risk of incarceration.

In theory, government or insurance should be able to anticipate the number of surgeons needed, how to staff their ORs, and ensure that we have enough resources dedicated to avoid long wait times. Having spent time in systems like Tricare, I just do not see this. Centralized planning is difficult. I have, however, seen more than one hospital get built by surgeons because the money was there to support it (up until that got banned by the ACA).

From what I have seen overseas and follow in trade publications, foreign governments tend to set their medical budgets through political processes more than through anything approaching price signaling. This means we end up seeing waitlists proliferate, lost man-hours of labor, and proliferation of paperwork.

Unfortunately, these days insurance companies are so tightly regulated and are such behemoths that they function much more closely as analogues of governments.

The market only works if people know what it is. Consumers can't shop around or make informed decisions beforehand based on pricing opaqueness. When people can't shop around it becomes in the businesses best interest to lower quantity provided and increase prices (standard monopoly behavior). This is the exact opposite of "increases the ability of the system to care for all."

Sure, can you provide actual historic examples of a modern healthcare system working the way your beautiful theory describes.

There are well over 100 independent countries in the world now, if you theory had any basis in fact I would think one of those countries would have at least tried it.

This is exactly how veterinarian care works. Cosmetic surgery also functions in a similar fashion. Eye surgery has also tended to follow this model.

Internationally, Germany allows for two tiered health insurance precisely to capture this sort of affect. Switzerland likewise tries to use supplemental insurance to get price signals to maintain quality and distribution (they also have a fun deductible system that captures a good amount of price information). Singapore, Taiwan, and many other places use a public-private hybrid system that seems to both capture a lot of price signaling and allow the wealthy to buy better access with higher prices.

In general, a full Beveridge system seems to be less functional to me than systems that capture price data or those which allow the wealthy to dump more money into the system.

I am confused. You are saying that healthcare functions like a monopoly which results in lower care and more rent seeking, but the solution is to turn it into a larger monopoly?

And consumers most certainly can shop around. The Surgery Center of Oklahoma makes a good bit of bank by posting prices online and doing surgeries for a literal fraction of what my local colleagues charge. The Free Market Medical Association is a small growing offshoot that is moving things around throughout the country.

What really straight jackets consumers are insurance plans. You go where your plan is accepted or you go to the nearest emergency facility. The former means that we are back to a centralized bureaucracy making allocation decisions with diminished price data - less helpful. The latter represents an extremely small percentage of overall healthcare spend.

Ultimately, I hope we eventually transition to some sort of event type insurance. You get sick, insurance sends you to a doc of their choice to evaluate it. You are diagnosed, you get a flat reimbursement fee per the diagnoses and then you choose where/how to treat it.

That is a million years away unfortunately as we are locked in ever larger rounds of monopsony begetting monopoly with the regulators on all sides being captured by their industries. So I suspect we will muddle along with places like Vermont trying and failing to implement universal healthcare and ever larger hospital systems/insurances behaving less like marketplaces and more like government bureaucracies. None of this will lower costs, but it will reward paper pushers so it a lock until there is a revolt against the status quo.

Rationing healthcare in the US is an oxymoron. The US has increased healthcare spending from about 5% of GNP in the ‘60s to more than 17% today. This is a function of virtually no controls on the medical industry, particularly Big Pharma, and an incredibly unhealthy lifestyle by most Americans. Over 2/3 of Americans are overweight or obese which directly correlates to epidemics of chronic disease that didn’t exist before our food industries became so efficient at producing unhealthy and addictive foods. Add to that a government that not only enables this system, but actively promotes these unhealthy foods with taxpayer dollars through the USDA. These costs will continue to skyrocket until America wakes up to these facts and takes control through healthier eating and lifestyle. The system is incentivized to make people sick and then treat them.

All healthcare is rationed healthcare. However I would sooner trust medical professionals to perform triage (they do it regularly in disaster settings, and in a milder way in hospital ERs) than trust the "market" whose judgments on these sorts of things are random and irrational.

Do you trust the market to deliver affordable, life-preserving food? Because it's worked wonderfully there.

Medical professionals are market actors like everyone else. I've personally seen doctors over-treat and under-treat out of financial incentive whether it's private insurance or Medicare/Medicaid. The "market" is ubiquitous. Subsidized/socialized care doesn't actually displace a health care market; it distorts it.

That market for food is radically different than the market for healthcare. Healthcare is much more akin to the market for justice or the market for sanitation: leave those to the randomness of the market and you will not get an equitable system, and most likely a woefully inadequate one. Every nation on Earth that is remotely civilized and with the means has chosen top nationalize healthcare to some degree (whether through single payer or whom other structure). That the US is still debating this in 2018 (mainly because some trick ninnies and failed ideologues don't like paying taxes) is as howlingly absurd as if we still debating indoor plumbing and public sewers.

So are you saying that the U.S. has already nationalized healthcare? Or that it's not remotely civilized?

Has it? Surely you realize there are massive government agencies that are responsible for making sure your food is the safe. Occasionally some things like the E-coli outbreak still get through and make national news. A couple hundred years ago before greater restrictions were put on the market a couple of people dying from a food outbreak would not be news

Do you trust the market to deliver affordable, life-preserving food?

A popular analogy, but a poor one.

Nobody goes years eating hardly anything, or maybe nothing at all, and then faces a situation where they need a mountain of food, or they will die or have their life severely constrained.

Even in optimistic cost scenarios medical care will be expensive when you need a lot of it, so most will want insurance. Food is actually pretty cheap. We spend a lot more on it than we need to in order to survive, even to eat a basic American diet. Not so with medical care.

As one of those medical professionals I would say you are woefully taxing my abilities.

Suppose my hospital wants to consider opening a new urgent care. I do not know what our current borrowing rate is. I have only the foggiest idea of what land prices for commercial real estate are. I can make a guess, likely not good, at what compliance costs will be for construction. All of these are information that can be tricky to gauge but are essential to know before considering opening a new location. Even more tricky is figuring out how this need enterprise would affect current resource utilization. Will other urgent cares get fewer visits and should we cut back staff? Should ER be higher staffed due to a larger number of patients coming?

My skillset is in keeping you alive. I know more than I want about finances and business, but there is a reason why we have admin to take care of things.

For all of this, price is a very honest signal in a world with terribly dishonest signals. Every doctor can think of ways to spend money that are helpful, few can see the system enough to know where to budget the next marginal dollar.

The real question is not who gets in to see me first - the MVC with head trauma or the GSW with profuse peripheral bleeding - that's too easy. The real question is how many of me do you hire, where do you place them, and how do you use them efficiently. Physicians can do well, particularly in aggregate, but even when we own the hospital we typically tend to buy expertise or rely on price signals.

In any event, actually putting physicians in charge of where to build hospitals and the like is actually prohibited by law. Section 6001 of the ACA limits the ability of physician owned hospitals to expand to meet needs that physicians see. There are complicated reasons why this is the case, but when physicians have actually wanted to stake their own money and serve patients where they see a need, they cannot. I can legally decide if you will live or die in triage, but I am prohibited from building a hospital if I think it makes sense to build a hospital next door to you.

"Socialized health care is rationed health care"

A truly idiotic statement.

All health care is rationed.

But enjoy your 'co-pays', hyper expensive generic drugs, $200 Band Aids and rejected insurance claims!

If everything is rationed, then nothing is rationed. We might as well strike "rationed" from the dictionary.

None of the problems you mentioned (which might or might not be actual) would be ameliorated in any way by nationalizing healthcare. The U.S. government already spends more- as a % of GDP!- on healthcare than any other country.

I didn't mention nationalising health care, you nong.

Shove your straw man up your backside.

Re: If everything is rationed, then nothing is rationed.

Logically that statement is nonsense. You can see that easily if I switch the terms to say "If all people are human no people are human".
And , it shouldn't take a degree in economics to understand that all goods in limited supply are rationed by one method or another.

Re: social healthcare, for all the talk of the amazing successes of healthcare funded out of general taxation, it is the case that:

The American system spends large fractions of money are spent on very old people, in very large amounts, on long shots for very marginal extension of life in the last months of life. The insurance systems targets resources towards old people who have been healthy and "done the right thing" their whole lives, paid into the system and are at stages where it is very hard to keep them alive.

Socialized healthcare does not, and redistributes money to those who are in ill health who provide QALYs ("Quality-adjusted life years"), e.g. younger people who are easier to treat and will have a good health life expectancy with cheap interventions.

It's no accident that with the US system you got the "War of Cancer" (with its successes and failures) and in the UK, with the NHS, we got the birth of palliative care. It's why the UK has awful cancer survival rates, but is very cost per reduction or mortality, while the US has great cancer survival rates, and high costs.

Thus the root cause for the "inefficiency" of the market system, and the "efficiency" of social health care. But there is not an efficiency difference of them doing the same thing, and still less anything to do with the "corporate interests" that so capture the imagination of leftists (even the late, great Hawking). There's a difference because the systems systematically "choose" different people to help. Once again, as so often in life, selection explains it all.

I'm not sure if that's unique to America, but the attitude extends well beyond AIDS to heart disease, cancer, diabetes: if they had gotten exercise or had a better diet, they wouldn't have the chronic illness.

Or if they hadn't smoked for half a century. Forgot about that one, huh?

No, It doesn't extend beyond the US. In Europe we provide help to anyone even to those whose options in life may not have been the best.

That's how it works in truly charitable and Christian societies, you know...

The US spends a far greater percentage of GNP (17%) on healthcare than any other country in the world. By that standard, we are the most “Christian” country in the world. The real problems are a population with the unhealhiest lifestyle in the world combined with a food and health system that is incentivized to make people sick and then treat the consequences.

The comparison of poor diet to smoking is amusing. Starting in the 70s the government promoted a diet based on bad science and special interests that harmed the health of my generation.

How should they be held responsible?

They used to subsidize tobacco production as well.

The government is in bed with a food industry that is incentivized to produce addictive and unhealthy food. The USDA subsidizes $Bs to the animal and sugar industries that continue to create the unhealthiest population on earth. With over 2/3 of us overweight or obese, we could significantly reduce healthcare costs by simply switching to a whole food plant-based lifestyle. All the money is working against that, so don’t count on your doctor, government or food companies to make a change soon.

Maybe you can go back in time and find the people responsible and punish them in some way.

By the way I doubt anyone would have been harmed by actually following the government's advice on nutrition, even if it wasn't perfect. Sweets and soda were never recommended.

I also would like to see the paper. That being said, my comment is taken from nearly 2 decades of experience as a surgical critical care physician. We are, indeed, fairly poor at predicting individual mortality, but somewhat better at predicting return to independent activities. None of those predictions however seems to be important for a significant number of mostly family members who are thrust into the position of making withdrawl of support decisions on critically ill patients.

I began to realize this early in my career, but have started to come to a understanding of the motivations behind it after reading “ I began to realize this early in my career, but have started to come to a understanding of the motivations behind it after reading “The Elephant in the Brain: the Hidden Motives in Everyday Life” by Simler and Hanson. In the part of their book that talks about healthcare decisions, they posit the hypothesis that much of our healthcare spending is to demonstrate love and its actual result has little to do with the decision to implement it. When having end-of-life discussions with patients families I definitely have noticed that for a relatively large plurality of people the act of making sure that “everything is done“ is to demonstrate to them selves, to family, and to a lesser but important extent, society as a whole, that they didn’t “give up” on their loved one.

This admirable demonstration of love becomes a problem because the resources that are being expended for marginal results in some cases or not the patients or families own but societies as a whole. I am starting to feel that the entire field of end-of-life care in this country is a modern version of the Potlatch where extravagant gifts goods are exchanged and often destroyed to demonstrate social standing and community loyalty. In the case of the US it is to demonstrate the importance of “life” and filial love.

This is why the debates around limiting end-of-life care in the face of “futility” tends to become very acrimonious because one side define futile as “no reasonable chance of meaningful recovery” and the other side finds the act of providing such care meaningful.

I’m trying to put my thoughts on this subject together for a paper so if any interested person wishes to collaborate please feel free to contact me.

And if you use the Potlatch idea, quote me.

+1.
Atul Gawande's book " Being Mortal" addresses some of these concerns in an excellent manner.

On the other hand, I’ve been struggling with writing an advance directive tightly enough to preclude life extension care after dementia diagnosis. After having watched a couple of family members go through that decline, I think it’s worth avoiding. The best I’ve been able to come up with is “apply hospice protocols” even if you are not officially in hospice.

The best “insurance” against dementia and Alzheimer’s is a whole food plant-based lifestyle. Both diseases are caused by vascular inflammation, the same as heart disease. If you are cursed with the APO4 gene, which could be the case with a family history, your risk is much higher. The only way to avoid it is by avoiding the foods that produce the inflammation in the first place. Animal products. You may want to check out nutritionfacts.org for more details.

Nutritionfacts.org is a purveyor of junk science, don't trust them

The site summarizes and critques current peer reviewed medical literature and research on various topics related to nutrition. All its articles provide the source literature and research for your own review. It doesn't create any "junk science" on its own, but highlights a lot of the "junk science" out there. So, I guess you could say they expose a lot of the "junk science" that exists, much of which is funded by none other than the food, drug and medical industries with profit motives to maintain the status quo. The site is non-profit. My friends, family and I have followed its advice and have eliminated several chronic diseases or risk factors ( high blood pressure and cholesterol, IBS, excess weight, rheumatoid arthritis, Type II diabetes, etc.), so if that's "junk science", I'm buying into it.

I've had both parents and three grandparents go through end-of-life care and agree familial attitudes can make a big difference. The treatment would tend to migrate towards the more interventional baseline.

Probably just as problematic is the fact that the medical industry has every financial incentive to provide these interventional treatments. There is no limiting standard the medical profession is required to meet, so family choice is the only limitation to provide these procedures. Cost is often a non-factor and suffering of the patient is often secondary to familial emotions.

I believe I was lucky that in all my cases, there was mutual agreement that hospice was better than more radical interventions. I could see through the decision process, though, that this would have been a major issue if there was any disagreement among my siblings or other family members.

@ James McNeill: After seeing the results of ventilation on a traumatically injured and eventually deceased 50-yo family member, my esteemed mother now refers to ventilation as "billing."

I saw a tragic documentary about a TBI patient who was saved via heroic efforts only to be comatose for a decade. He came out of the coma around age 30 with limited faculties and greatly reduced mobility. One tragic consequence was he no longer recognized his daughter, and constantly made sexually suggestive comments to her. The scene where the family members asked the doctors if there was any hope of improvement was painful. The man had been saved, but the one the family knew and loved was gone forever. And in his place was a chairbound, creepy child-man who was going to be around for the next 20 years.

@ blade doc

Just like bank regulators ignored the you would suppose so useful question of: What is ex post more dangerous to the bank system, that which ex-ante perceived as risky, or that which is ex ante perceived as save?

Opportunity Cost & Choice

Who pays these elderly medical costs -- and who is actually making the choice to spend such large amounts for that purpose, versus other possible spending purposes ??

Ultimately the government pays for almost two-thirds (or 68%) of health care spending by the elderly (age 65+), with Medicare accounting for 55% of the coverage, Medicaid covering 10% and other government programs covering the remaining 3%. Private insurance covered approximately 13% of the elderly’s medical expenses.

These government spending "decisions" are a vague mass of individually small choices made by countless multitudes of government politicians and bureaucrats... with no clear understanding of the net effects of that spending 'system'.

Of course the government is hugely in debt and cannot afford these huge medical costs. Holders of U.S. Dollars will ultimately pay these costs via continuing Dollar devaluation.

The people making these elderly medical care spending "choices" ... do so indirectly with no consideration of opportunity costs -- and these "deciders" are not themselves paying the costs of their vague decisions.

In most cases it is the doc in consultation with the patient and family that decides on the treatment plan and thus the spending. So most spending is doc directed.

And because in most cases doctors are paid fees for each service, the incentives are all screwed up. No incentive to reduce treatment, ever.

This always seemed like the obvious reply whenever someone made that point, but it's good that someone really looked at it carefully.

This paper misses the point. Almost none of those receiving medicare are working and contributing outside of making their families a little happier that they are living longer. It makes much more sense to spend this money on younger workers who are raising the next generation. Let the old folks spend on their own healthcare.

One of my favorite short stories I read last year was by Margaret Atwood. In it, there is a massive uprising because young people blame the older generation for all their problems. Nursing and retirement homes are destroyed, old people are assaulted, etc. The title of is "Torch the Dusties", with dusties referring to the old. I partly liked just because of the title.

Medicare the great equalizer. Most all of you will get old enough to 'enjoy' Medicare benefits.

What do you suggest? Mandatory euthanasia upon retirement? Moreover the current system does help younger people since the burden of supporting their elders is spread over the whole population-- which includes people like me whose elder relatives are all deceased. Otherwise non-wealthy people would find supporting their elders an insupportable burden.

Looks like Google hopes to help with that. Instead of LMGTFY, we might get "let me google your death for you."

Scalable and accurate deep learning with electronic health records
https://www.nature.com/articles/s41746-018-0029-1

They predict in-hospital mortality well. I wonder what the percentage of deaths in Einav, et al's sample are in-hospital. If it is a large portion, they might want to adjust their predictive model. If not, then the paper you reference doesn't have much relevance to their analysis.

Yes but what's the utility for spending on someone older than 60? 70? 80? 90? We must draw lines and limit expenses, otherwise soon enough they will be drawn for us.

This very dependent on a bunch of individual patient factors. Physical and mental age more important than numerical. Prognosis, home and local medical/social supports, expectations of the patient and family, to name a few. Most older patients decline on very aggressive interventions. When I was in medical training in the '70's the natural transition of patients toward less aggressive treatment tended to occur at about age 70. Now it's about 85. And much of that has to do with new medical technologies, where folks are living longer. And technologies that though expensive are easier to administer to the elderly. For instance many chemos are much less toxic, many surgeries less invasive.

The high cost of end of life care is not necessarily due to aggressive treatment. Supportive care also costs a lot. Dying people generally cannot care for themselves and often require assistance. Even if we did nothing but provide supportive care for anyone over 70 with a chronic illness the bill is still going be a large one.

I haven't yet read the paper, which of course never stops anyone here from commenting. But my best guess is that it more or less confirms what we already know - that there are several different "trajectories of illness" at the end of life, not all of which are predictable or expensive. A claims-based algorithm is like viewing patients through a stained-glass window anyway, and in the absence of clinical data I doubt you could identify a large population of dying patients in the first place. Claims data combined with clinical judgment and a patient's baseline level of function is much more accurate. But if you did train an algorithm on the single variable of mortality, you're going to bring in a bunch of old people going gently into that good night in an assisted living facility or who are well supported and comfortable at home with home hospice. Those are the kind of dying people who are less inclined to get chemo, go to the ICU, etc and thus aren't all that expensive.

Anyway, as other commenters have said already, there is plenty of waste in life's last chapter, whether you want to define that as the last year or the last five years. Getting sick patients care in lower acuity settings - whether they are dying or not - is not only cost-effective but very likely helps chronically sick people live longer.

The facts mentioned in the article just highlight the absurdity of delivering medical care as a consumer good in the first place.

It is a strange marketplace where one can't choose to leave and has no knowledge of what is being sold.

The doc in so many cases has to be a/the major advocate in the patients best interest. This is the problem of asymmetrical information, where the patient doesn't know all the details, implications and potential outcomes of many acceptable treatment plans.

Arrow's seminal paper and why we have Medicare in the first place:

http://www.who.int/bulletin/volumes/82/2/PHCBP.pdf

That medicine is a consumer good as opposed to a public good is unavoidable. There may be perfectly sound, ethical arguments for why everybody should pay for everybody's medical care. Regardless, resources are finite, demand is infinite, and everything has a cost. Rationing will occur which is why Canadians who can afford to supplement their single-payor coverage with private insurance and Europe limits its military budgets despite being next door to Russia.

It's a dismal topic. Nobody wants to believe that their life or well-being depends on their finances or on bureaucratic decision-making.

"Saving their lives so they'll cost even more next year."

The more precise conclusion: medicare administrative data is not very good at identifying people who are likely to die in the next year. Part of this is real; even people who look very bad off in terms of just a list of diagnoses (eg class 4 heart failure, severe copd, esld, advanced hiv) often have median survival more than a year. The other part has to be the paucity of what they know; only sometimes do ICD-9 codes convey the severity of diagnoses (ICD-10 is much more granular, but is new) and rarely do they capture essential elements like a patient's performance status and response to therapy. I'm surprised how few sure bets on mortality they find (eg many stage 4 solid cancers), perhaps that's a function of not using the admission data for terminal stays. Surely many in-hospital deaths have admission diagnoses which would identify very high mortality, such as a combination of severe malnutrition and severe chronic disease or infection superimposed on severe chronic disease.

For all the talk of the amazing successes of healthcare funded out of general taxation, it is the case that:

The American system spends large fractions of money are spent on very old people, in very large amounts, on long shots for very marginal extension of life in the last months of life. The insurance systems targets resources towards old people who have been healthy and "done the right thing" their whole lives, paid into the system and are at stages where it is very hard to keep them alive.

Socialized healthcare does not, and redistributes money to those who are in ill health who provide QALYs ("Quality-adjusted life years"), e.g. younger people who are easier to treat and will have a good health life expectancy with cheap interventions.

It's no accident that with the US system you got the "War of Cancer" (with its successes and failures) and in the UK, with the NHS, we got the birth of palliative care. It's why the UK has awful cancer survival rates, but is very cost per reduction or mortality, while the US has great cancer survival rates, and high costs.

Thus the root cause for the "inefficiency" of the market system, and the "efficiency" of social health care. But there is not an efficiency difference of them doing the same thing, and still less anything to do with the "corporate interests" that so capture the imagination of leftists (even the late, great Hawking). There's a difference because the systems systematically "choose" different people to help. Once again, as so often in life, selection explains it all.

The relevant statistic (25% in the final year) indicates some low-hanging cost-savings fruit. If you could better predict mortality, you could save a lot of money. It’s enticing because it doesn’t require a scientific breakthrough or a massive cultural change. Just better machine learning, which is already improving exponentially.

Just as long as the doc is there at the bedside, along with the family making the (final) decisions. The AI is just another opinion, not the 'death panel'.

Comments for this post are closed