Assorted links


Adjacent Possible Medicine's predictions sound good, but studies show that patients aren't that interested in optimizing their health care. Outcomes are secondary to signaling, touching, being fussed over, being noticed, being listened to, and having one's individuality affirmed by the provider. If the medical center of the future wants to "treat patients like numbers", it will have to develop some kind of complementary "humanistic" structure to disguise its real workings. I would suggest hiring actors to wear white coats and hover of the patient throughout the patient visit, nodding and wrinkling their foreheads concernedly, touching the patient reassuringly, and cooing sympathetically, while the real doctors -- pretending to be some kind of powerless assistants -- sit at the keyboards taking data and solving the patient's problems.

How do I get the #%@#$ away from those people? PPACA?


Really interesting point. I've wondered along these same lines- are patients better off with care they don't want? Are they better off not liking their doctor as much as they used to?

I wish I had something compelling to say, but instead I'm just going to sit here and think about your comment....

Thanks for reading!


Very good points. Hopefully, in twenty years we'll have some freedom of choice, so that those who want to go for "touching" can do so, while those who aim for outcome can take a different route (Watson, MD).

For non-surgical health care, we are already at the point where doctors can be largely replaced by expert systems. Like all things, we make it affordable through mass-production and automation. Our goal in healthcare should be to eliminate human labor as much as possible.

Veridical Driver,

If I may respectfully disagree- assessing and treating a sick person is still vastly more complex than our non-human systems are currently capable of. Further, what would you prefer, care from machines that are smarter than today's doctors, or care from machines that are smarter than yesterday's doctors that are utilized by tomorrow's doctors?

Personally, I'd take machine + doctor of machine any day of the week.

The point is that, just because we upgrade the tools doesn't mean the tool users will become obsolete. It means that the tool users can do even more marvelous things now.

Thanks for your comments!


Your choice isn't "Doctor + Machine" vs "Machine". Your choice is "Machine" vs "No Doctor".

Sure, I rather have all my clothes custom tailored by experts who studied and apprenticed for 12 years - but that isn't an option if you want to clothes everyone. Mass produced clothing is not as good as custom tailored clothing, but it is better than half the population going naked.

Likewise, if you are wealthy enough, you would be able to afford the luxury of a human doctor... but if we want highly effective care to the maximum number of people, we need to make health care drastically cheaper. We do this with mass-production and automation,.

If I may respectfully disagree- assessing and treating a sick person is still vastly more complex than our non-human systems are currently capable of.

That's true. But does it get better results?

What if it turned out that the complications and human intuition were actually getting worse results on average?

The vast majority of diagnoses (about 90% at the time of my relatively recent medical education) are made by a competent physician talking to a patient. A competent physical exam will give you another 5-6%. Of the remaining diagnoses, a goodly percentage are untreatable. Bottom line is, replacing competent physicians with machines will lead to more expensive and less competent medicine.

I don't understand your comment. Isn't the accuracy of those diagnoses the issue? Do physicians have a clear advantage on all diagnoses by that measure?

The linked article wasn't describing modern technology being used to clarify the accuracy of a physician's diagnosis and help guide the treatment; it described a bunch of linked technology administering care and effectively replacing physicians. This is nonsensical. Most laypeople have the mistaken idea that, when they have their blood drawn or x-rays taken or CT scans done, the doctor is doing that to figure out what's wrong with them. In fact, the vast majority of those tests are done to generate income (I'm being a little extreme here, but not as much as you might think) or to do some CYA work (though malpractice is much less of a driver of medical costs than most people realize) or (and this is the correct use) do determine the extent of the diagnosis that the physician has already made or at least already suspects, and to give some measurable parameters for managing that pathology (e.g., the doctor knows the patient has hepatitis, and wants to confirm what type, and get a baseline of the degree of disease).

It is exceedingly rare that these tests reveal unsuspected pathology without something in the patient's history and current symptoms already indicating that pathology. In everyday medical practice, when lab tests are ordered as a matter of 'routine,' abnormal lab tests only cloud the picture, and usually result in further testing, which further clouds the picture, and can sometimes lead to unnecessary treatment and a bad outcome. One reason for this is that the 'normal range' for most lab tests is the range that includes 95% of a group of seemingly healthy people. By definition, 5% of a healthy cohort will have lab values at any given moment that fall outside the normal range. One axiom of good medicine is to never order a test or a lab unless you have a good reason to, and have a plan for what you'll do if it comes back abnormal. The flip side is that when doctors orders tests that aren't really called for, and they get abnormal results, more often than not they simply ignore those results. That's why ordering a standard 'SMAC20' panel of metabolic tests and a chest x-ray on people who aren't symptomatic is usually bad medicine.

There are very few diseases that have a specific definitive lab test. Most lab tests are like vital signs -- deviations from the 'normal range' require context to be understood, and further lab tests cannot and do not provide a fraction of the context one gets from simply spending 10 minutes talking to a patient and observing them during that time. The solution is not more data streaming into a machine. All that data is a small part of a big picture, and 'expert systems' will never put that picture together as well as an experienced physician who spends some time with a patient.

If we take things a step further and consider issues of false positives from this blanket metabolic testing (which is inevitable), and the real risks of even relatively benign testing, and the real and probably huge risks of unnecessary treatment triggered automatically by such expert systems, and the scenario described in the article is beyond idiotic, and certainly not in the direction of either better medicine or more cost-effective medicine. Of course, when your premise is that medicine is like the recorded music industry, I guess we should expect a certain amount of idiocy.

Thanks for an extremely interesting comment,


I think you misunderstand the author. What he is describing is basically statistical process control. It doesn't mean we will do it well, but if the product becomes things you buy at Wal-Mart rather than having to carve 3 hours out of our lives to go see the doctor once every couple years we have a good shot.

[Gonna sound off-topic disclaimer]

Wife has sore throat. 50/50 viral or bacterial. Won't get the slow-growing strep test back until Thurs. That's a long time to wait. Doctor offered us a prescription. Noone nagged for it. I understand antibiotic overuse. Aren't most people just going to get the prescription filled? Wouldn't a bunch of faster weak signals be better than a 5 day definitive test?

@Andrew: The article talks about collecting diverse medical/physiological data on a remote and ongoing basis to drive medical diagnosis and treatment. The author misinterprets what goes on in the doctor's office, and takes it to the extreme, and comes up with an answer that doesn't make the slightest bit of sense. The problems with western medicine is NOT that there is too little data, and it is NOT managing data, and it is NOT that doctors are overwhelmed by he data. As I said, the vast majority of diagnoses are made by spending some time with patient, and sometimes laying hands on the patient. Ironically, not only does this approach provide the best medicine, but even when there is no useful treatment per se provided, that interaction itself is known to promote better health.

We've been evolving two systems of medicine: (1) the fee-for-service model, where doctors who see the most patients (and who necessarily minimize the time spent with those patients), and who do the most procedures, make the most money, and (2), the HMO model, where doctors who see the fewest patients, and who spend the least time with them and do the fewest procedures make the most money. Both systems are biased away from optimal patient care, which depends upon time spent with patients, and which also depends upon coordination of care among caregivers. These two crucial aspects of good medicine are not only not rewarded for physicians, but they are actively punished in the economic sense.

Re: your wife's sore throat: a sore throat is going to be viral over 90% of the time. Step is usually an easy diagnosis to make -- the tonsils are covered in pus, among other things. In questionable cases, the rapid strep test is very specific and takes a few minutes in the doctor's office. Since there is about a 20% false negative with that test, this can't always rule out strep throat. A standard throat culture takes 1 to 2 days. A sore throat caused by strep despite the absence of pus, and with a negative rapid test, is going to be about 1% or fewer of the cases of sore throats. Many physicians will still treat with antibiotics, but it's a judgment call, and it's the kind of judgment call that remotes wireless sensors and statistical aggregators are never going to be able to accomplish. By the way, what 'weak indicators' are you suggesting the doctor is ignoring but that the remote sensors would have picked up?

By the way, I don't minimize that doctors participate in the bias towards unnecessary treatment. You're correct that it's not always the patient nagging for antibiotics. While I constantly had friends (I refused so often they gave up) who would ask me to call in prescriptions for them because they were certain they had this or that illness, at the same time many physicians ignore the real risks of antibiotic treatment and err on the side of just giving the prescription. Then they don't have to worry about following up with the strep culture results, and they know the patient leaves certain that the doctor did 'everything they could.' It's a silly game doctors play. Bronchitis is one of the worst diagnoses for this. I've learned to just keep my mouth shut when someone tells me they're on antibiotics for their 'bacterial bronchitis.'

It doesn't matter. Doctors get input from the patient, analyze that input, and produce an output. That can be automated.

One of my comp-sci professors (who was also an M.D.) developed expert systems in the 1970s that where more effective at diagnosing illness (and referring a patient to an appropriate specialist) than most family doctors at the time. What you do is a lot less complicated than you imagine, and outside some very difficult specialties, can essentially be simplified into something like a flow chart.

It isn't difficult to get a computer to do decision-making for known problems. Computers still would have a difficult time doing automated surgery, and can't solve new problems... but for diagnosing 90+% of the problems that most doctors see, we could do that stuff back in the 70s.

I second Veridical Driver's comment.

As for "‘expert systems’ will never put that picture together as well as an experienced physician who spends some time with a patient." - this is like saying that computers will never be able to play chess as well as humans do.

@ Vertical Driver -- computer scientists have been saying that they can replace doctors for decades, you're right. And under highly artificial conditions, a computer algorithm might compete fairly successfully with a GP in referring patients to specialists. But then the computer can't do much of anything that a specialist is doing (not just surgery), so little is gained, and the computer isn't going to save much money on the treatment of those patients who get treatment from the GP. The cost of GP's screening patients for serious disease, and making appropriate referrals, is not anywhere near the major driver of our runaway medical costs.

On another note, we could turn this around and point out that the vast majority of patient visits to doctors are ultimately unnecessary with respect to patient outcome. Either the illness is completely non-serious, or it is self-limiting and will get better even without treatment, or has no significantly useful treatment. Of those illnesses that do have useful treatment, that treatment often doesn't involve on-going physician input. And we have to figure in iatrogenic illness, and well as treatment side effects that are often worse than the original problem. So we could use those statistics to show that spending time playing Tetris is generally about as useful as a visit to the doctor when one feels ill. And then what have we accomplished?

@ Jamie -- to a layperson, chess may seem as complicated as practicing medicine. It ain't. Hey, guess what -- a computer can beat a smart human at Jeopardy! I guess that means computers should be setting monetary policy.


Are you telling me that the total amount of patient data relating to a typical condition, say congestive heart failure, isn't overwhelming? The real-time, dietary, exercise, fluid status, myocardial status, blood pressure and oxygenation, combined with genomic, social, and environmental features is not staggeringly complex?

Just because doctors lack to tools to gather, organize, and interpret this information doesn't mean it doesn't exist. It is knowable, so long as we have the tools. We don't yet, so we diagnose and treat with clinical exam skills, and that's great for now, and I have deep respect and admiration for those skills.

But I sure hope those skills are eventually seen as primitive.

Saying this will never happen... ooohhh.... I'm definitely putting my money on it happening.

I am interested in the potential downsides to Romers charter city proposal.
Specifically, that a foreign government could sponsor a charter city within another country to assure that it adopts and adheres to rules that the city has in its charter. While we may think this is a great way for the US to sponsor development in an African country, what we do, China can do as well. It could charter or sponsor a new African city. US gunboat diplomacy of the early 20th century involved intervention to protect financial interest, and it is easy to see that if the sponsor claims the charter city is breaching the charter and reneging on financial commitments, that too could be grounds for invasion.

Here is a link to some of Romers work describing the concept:

What would be so bad about China sponsoring a city? I could definitely see it being win-win. I just don't know if China is particularly interested in exporting governance, as it seems content to deal with existing governments and just extract resources. After all, it has a large enough supply of fairly cheap labor at home and doesn't need to do much to encourage labor productivity.

In the data shown in (1), most of the faster-than-GDP growth of infotech is confined to the 1990s tech boom, and even after the irrational exuberance of the stock market is factored out, that was still the period with the strongest real productivity growth in recent history. The thesis is hard to sustain on this data.

Sadly, I was born with an awful lot of bad medical conditions. About a year ago I was in the hospital with a very strange set of conditions. It took a team of doctors a few days to finally diagnose me and give me the right treatment. As they tell it, my condition was extremely dangerous and could have been very bad if things had been done differently. I'm rather glad they didn't do some of test and treatments they had talked to me about at various points (which I, rather dazed by my ill health, could barely process).

One of the reasons my condition got so bad is because I self diagnosed myself with the flu in the beginning. My initial symptoms were flu like. As I had done many times before, I went to patient first (that nurse place and the CVS you don't need an appointment for) figuring they could just write me a prescription for my usual flu anti biotic. Not just because it was cheaper, but because it was a lot quicker and less hassle. Not only did she miss my real condition (who can blame a nurse), but the prescription actually worsened my condition significantly.

Now there is no way to know if a doctor would have done things differently. It was only after they got blood test results back that anyone started asking questions. But I sure wish I had gone to a doctor originally. And I definitely am glad I didn't go to webMD or something like that.

You thought you had the flu, so you sought out a medical professional who would give you an antibiotic. This is an excellent example of why the consumer driven medical model that the Republicans are pushing won't work. If you'd have been correct, and you did have the flu, you would have been wasting your time and your money and doing something that was dangerous (taking antibiotics unnecessarily), while also running around spreading the flu. That was the best case. The worst case is what almost happened to you.

Unfortunately, many of the attempts to fix American medicine in recent years have made things worse, so I don't want to imply that I think our current system works, but 'more choice' and 'malpractice reform' are examples of trying to treat influenza with antibiotics. They won't save money and they won't give us better care. But at least they'll generate more profits for the non-medical part of the medical industry. Yippie!

"the consumer driven medical model that the Republicans are pushing won’t work"

What technocratic nonsense. This anecdotal example is nice but a counter anecdote would be that in a socialized health care system, someone with a bacterial pneumonia ois not treated, because on average, most pneumonias are viral and self limiting.

Demand for health care, irrespective of cost, is infinite, and does not saturate - there is always more health, but there is always only a finite amount of health care resources available - even if 100% of a country's resources are devoted to health care.
To distribute the available health care in a population, a metric is required. An optimal health care system then distributes the health care according to this metric. The two metrics are:
- patient satisfaction - what you call 'consumer driven'
- 'objective outcome' based measures - what you would probably call 'expert driven'

Patient satisfaction metrics are based on individual preferences, allowing individuals to optimize their personal balance between cost and satisfaction. Objective outcome based measures are based on collective preferences across the population, allowing 'czars' or 'experts' or 'health care authorities' or 'NHS' or 'CBG' to optimize the balance of cost and satisfaction across the population.

Unfortunately, these usually do not overlap, as in this example, where 'dave' preferred one purchase based on his expected satisfaction, and 'objective outcome based measures' might have lead (with hindsight) to another, different, health care purchase, i.e. going to a doctor.

If objective, outcome based measures existed this might work. The problem is that individuals are different, and have different preferences for health.
For example, one objectively measurable outcome based measure for eye surgery is visual acuity. But some individuals who do outdoor sports prefer the highest visual acuity, let's say 20/15, while other patients, who spend most of their time reading and watching television, are okay with 20/30.
To deal with this, 'quality of life' based measures were developed (you can look up the extensive literature on this), which are obtained by asking large numbers of people to atrribute preferences to different aspects of their life, and determine a value - like would you rather live 2 more years with poor eyesight, or only more year with good eyesight. These measures then supposedly capture the aggregate health care preferences of a population, so that the 'experts' can decide what to spend resources on.

But these quality of life based measures are elusive as well, because people are different and change over time, and they are ultimately still subjective. There is simply no way 'experts' can better attribute health care resources than well informed individuals.

We have had a 100 year run of excellent results in health care based on the assumption that people are essentially the same and disease causes and cures follow Gaussian distributions. All that low hanging fruit is gone now.


We don't really have a consumer driving medical model and when I worked for consumer product company X, we gave the consumer what we knew he really wanted, not what he may have said he wanted that he didn't really want. That's all I want from medicine.

I am also not 100% sure dave really means "antibiotic" when he says "antibiotic."

I think you might be interested to to go back a few posts where Alex refers to a video by Megan McArdle and look at the table. What the democrats are pushing is a simple false choice between "consumer driven" or "expert driven" when the situation is a lot more complicated, and Megan McArdle outlines the various actors and their incentives.

I skipped watching the McArdle video when I first saw it, so thanks for the reminder. I think what she has to say about antibiotics fits exactly with some of the points I'm trying to get across. The American medical system is full of perverse incentives. Applying market principles to medicine generates many of those perverse incentives, and has exactly the kind of horrible long-term outcomes that McArdle details.

I think both Democrats and Republicans are promoting false choices. I personally think the Republican choices are the easiest to discredit, but that might be because their choices are simple to understand and have never been shown to work. But ultimately it's a sham debate, and voices of logic and reason, like MrArdle's in that video, are drowned out by the political gibberish and sloganeering (I mean, she is proposing Death Panels by restricting antibiotic use, isn't she?).

How is it, by the way, that patients can't be customers (a la Krugman), but at the same time doctors cannot withstand the withering onslaught that results in them cowering in the corner and throwing bottles of antibiotics at patient zombies?

I kinda doubt it's both ways unless the popular meme is flawed.

@ ad*m: You wrote "This anecdotal example is nice but a counter anecdote would be that in a socialized health care system, someone with a bacterial pneumonia ois not treated, because on average, most pneumonias are viral and self limiting."

I kind of stopped reading there, because this is simply false. The standards of good medical care is fundamentally the same in America, Canada, France, Germany, etc. I know quite a few people who have received care under the American system, and European systems. None have ever complained about being denied care in Europe, and studies of those systems back this up. The idea that in 'socialized medicine' an illness that has a significant risk of being rapidly fatal will be ignored because it's statistically unlikely is a political meme, not a reality.

And by the way, a minority of pneumonias are viral (around 25%), and many of those viral pneumonias also need medical treatment. The non-viral pneumonias absolutely need rapid treatment.

Also, the one medical system where I *have* seen potentially serious diagnoses ignored or minimized is in American prepaid systems (either HMOs, or more distributed prepaid groups). In these systems, if the gatekeeper (usually not a physician) correctly guesses that a patient is going to recover with treatment, they're financially ahead to thwart that patient's care. On the other hand, there is little incentive for a physician in a socialized system to either provide unnecessary care, or to deny necessary care. Whatever the flaws of those systems, THAT is among their strengths.

@Andrew: I don't think either party has a clue about what we need to do to make American medicine better and more accessible and cheaper. And I think the AMA, the hospital groups, and the insurance companies are also part of the problem. I feel like I'm on the Titanic while one group is rearranging the deck chairs, and another group is arguing about what tunes the band should play, and there are a bunch of 'experts' in a safe, warm ship alongside debating the best way to load up the inadequate number of lifeboats.

In the above, I meant to write "if the gatekeeper (usually not a physician) correctly guesses that a patient is going to recover withOUT treatment, they’re financially ahead to thwart that patient’s care." And don't think for a minute that this doesn't happen frequently.

Oh good! That sentence made a horrifying sort of sense the other way.

Like, if the patient dies now, they won't have to pay increasing costs as he ages....

2. Health care in twenty years’ time? Forty?

If health care in twenty years’ time will be that different what virtue is to the current medical licensing gauntlet?

I love your engagement with this topic.

Please allow me to address a fundamental point of my post that I don't think came through terribly well:

Disruptive technology is a special kind of development. It is not about bigger, fancier tools that require exceedingly expert handlers. Rather, it is about smaller, cheaper tools, yet smarter tools that a novice can operate.

In the post I used Clayton Christensen's example of a smartphone as disrupting the music industry by putting a small, cheap, smart tool in the hands of novice music consumers.

Healthcare has lagged in implementing such disruptive technologies, for various important and understandable reasons. In my post, I'm simply envisioning what the world will look like when these disruptions start to occur.

Forgive the wanton optimism, but everything will get better. Patients will be empowered to be more proactive in their health, better data will be available to carry out better prevention, and these two features will combine to drastically reduce cost and bring better healthcare to broader swaths of the population.

It will REDUCE the unnecessary testing you're talking about, because disruptive devices won't be testing, they'll be monitoring. We won't have mere snapshots like today, but trendlines integrated with other trendlines, medication history, diet and exercise habits, and genomic data. Such cheap, smart devices will reduce testing, reduce the need for fancy scans, and reduce the acute exacerbations the prompt emergency room visits.

To be clear: These ideas are largely Clayton Christensen's in "The Innovator's Prescription," THE most intriguing book on healthcare I know of, combined with Eric Topol and Daniel Kraft's respective TEDTalks.


Aaron, I appreciate you clarifying your thoughts. I still think the music industry is a terrible industry to compare to medicine (though I'm hard pressed to think of what industry is comparable, since I think health and medicine defy many conventions of both logic and market economics). In the music industry over just the last century or so we see a story of constant technological disruption. Mass production allowed ordinary people to acquire quality musical instruments, so sheet music was all the rage. Then music recording devices (Victrola and so on) disrupted that. Then inexpensive radios and the development of the radio industry damaged the recording industry. A series of improvements in recorded music (LPs, 45s, various forms of tapes, CDs) came along, each to some extent disruptive of the status quo, while radio gradually was eclipsed by television. Then the digital music revolution happened, and that's caused further disruption. I'm sure this is a half-assed description of the music industry, but my point is that it's an industry that has always been in flux. The smartphone is but one more jagged bump in a path that has been all over the map, and the evolution of music consumption has been unpredictable from one decade to the next for a long time. Can you guess where the music industry will be in 20 years? I bet you'll be wrong.

Medicine, on the other hand, is practiced pretty much the same way it's been practiced for centuries. We have better understandings of physiology and pathology, and have better tools and tests and treatments, public health is vastly improved, but the practice of medicine is not so different from what Hippocrates did almost two and a half millenia ago. All these centuries of disruptive technology, and somehow doctors still do much the same thing. I think a better question than 'where will medicine be in 20 years' is, why has medicine changed so little over such vasts period of time, and why is it so similar in day-to-day practice across such a wide range of countries?

I would also suggest that patients already have the ability to be proactive in their health, and yet somehow choose to ignore the overwhelmingly clear data they already have. Who needs trendlines of trendlines? What is some marvelous smartphone device going to tell someone who is morbidly obese, who drinks too much, who enjoys occasional unprotected sex, who goes ahead and eats that chicken that was sitting on the counter and might not have been properly cooked? How is a disruptive ap going to change the behavior of a 35 year old guy who is in such pathetic health that he wheezes after going up a flight of stairs? What will it do for the mother who really believes that vaccines are dangerous? Obviously, I'm skeptical, but I could be wrong. Even if I am wrong, I think this focus on disruptive medical technology is a diversion.

I think the big questions for reshaping medicine are two fold. One, how do we make good care accessible for everyone? That's NOT quicker and cheaper diagnoses, which disruptive technology might help with (though I would say the history of medicine shows that this will mostly be available and utilized by people who already have access to health care, and are already taking care of themselves). I was in grad school while CT scanners, PET scans, and MRIs were being developed . It was exciting stuff, and I even got paid good money to be an experimental subject in some of the early development (stupid of me, but I was young). I love new medical technology, and can see the gains we've made from it in diagnosis and treatment. But, frankly, even the staggering advances of CT/MRI/PET haven't dramatically impacted the general quality of life and health of Americans.

The expensive part and the difficult part is the treatment. Right now there are multiple websites that allow a reasonably intelligent person to diagnose themselves. I've looked at them, and they're not bad. So now it's virtually free for you to figure out that you likely have ischemic heart disease. But you don't have insurance. That ugly mole might well be melanoma, but you don't have medical coverage. How does the disruptive technology make this situation better? How does it help more people get good treatment without bankrupting themselves?

The other question, the elephant in the room, is how do we make the hard decisions to allocate health care in a way that society can manage? Or, as the Republicans would put it, who's going to run the Death Panels? So your disruptive technology helps us all stay healthier longer, and it catches some diseases sooner, and everyone suddenly listens to their medical smartphone aps in a way they never listened to their doctors. It's a heavenly, healthy age. Until we all get into those last few years of life, when it all starts breaking down. Our bodies, that is.

We don't need more tests and monitoring to tell us this. People in the last few years of their life are generally disabled, and have multiple illnesses and chronic conditions. Your disruptive technology, if it's as good as you hope, has extended the average life span a few years, and kicked the ball down the field a little further. But were still faced with the fact that we've evolved a system where the majority of health care costs are run up in the last couple of years of a person's life, with a huge amount of that happening in the last month or so. On a societal level it's a huge waste, a staggering misallocation of a valuable resource.

What you hope to happen with monitoring might even make things worse. New technology that helps more people live longer means more people will enter those final years and months in a demented and/or extraordinarily frail state. Very few people have the foresight to create living wills or specify their wishes years before they get to that decrepit state, and it's impossible to have that discussion with a demented patient, so they get all possible treatment. If you've never had the pleasure to doing chest compressions on a 95 year old woman in multi-organ failure who is suddenly in cardiac arrest, then I'm not sure you can appreciate how much expensive, tragic futility goes on in those final weeks.

I'm sorry I'm so cynical about this. I've pretty much lost what little faith I had in our political process, and I've felt out of step with much of the medical industry for a while. I'll check out Christensen's book, maybe it'll change my mind, but I don't think we can find technology that will pull our medical system out of the death spiral it's in.

But were still faced with the fact that we’ve evolved a system where the majority of health care costs are run up in the last couple of years of a person’s life, with a huge amount of that happening in the last month or so. On a societal level it’s a huge waste, a staggering misallocation of a valuable resource.

"Doctor, it hurts when I do this."
"Then don't do that."

This is a social issue, not primarily a medical issue.

If we didn't have medical insurance, then people would pay for the medical care they wanted and could afford, and this would mostly not come up. When a person became so debilitated that somebody else executed their estate, then somebody rational would decide when to stop paying medical bills. Problem solved.

If private insurance companies handled old people, they could raise rates until old people couldn't afford it. Problem solved.

If government could choose how much money to spend, that would settle it. Problem solved.

It would be acceptable for physicians to say "We've done just about all we can.".

We already have a growing hospice movement. Lots of people don't want to spend their last month in an agony of resuscitation.

When my grandfather died in the early 1950's, his MD signed for no treatment and then went on a fishing trip. Nobody at the hospital was ready to do treatment against his orders, so my grandfather died in a few days. When my mother died a few years ago, her MD said no treatment. She couldn't drink water by herself, so when they removed the IV she died of dehydration in a few days. This isn't exactly new.

Aaron, I searched your blog for references to Robin Hanson. You should check out his writings on healthcare and why it doesn't seem to improve with Moore's Law like other industries.

Comments for this post are closed