August 29, 2010
Anesthesiologists Demand To Outrun Supply?

Will a 3% per year growth in demand for anesthesiologists in the United States outrun supply?

A recent study from the RAND Corporation, one of the country’s most trusted analytic organizations, finds a current shortage of 3,800 anesthesiologists and 1,282 nurse anesthetists. However, if current trends continue, a dramatic shortage of anesthesiologists and a significant surplus of nurse anesthetists are projected by 2020.

This brings up some questions that I do not have answers for. But the questions are pretty interesting and worth investigating:

  • Is it necessary today to have 1 anesthesiologist per operation? Or could a small number of anesthesiologists monitor many operations with the help of software and assistants? How hard would it be to develop software and equipment that would enable anesthesiologists to handle many operations at once?
  • How long until software can do as good of a job as anesthesiologists?
  • What are the prospects for reducing by anesthesiologist demand by speeding up operations? For example, is duration of surgery shortened by current generation robotic assist surgical equipment such as the da Vinci surgical system? Will future robotic surgical devices operate autonomously much faster than human surgeons can?
  • Will rejuvenation therapies result in an increase or decrease in demand for surgeons? For example, will stem cell therapies decrease the demand for surgeries more than installation of grown replacement organs will increase demand for surgeries?
  • Will effective cancer cures via, for example, nanotech-based precise cancer cell killers reduce the overall demand for surgery? Or will avoidance of death from cancer (and elimination of need for cancer surgeries) just make people live longer to need many other kinds of surgeries?

To put it another way: Should a 20 year old today choose surgery or anesthesiology as a career path? Are these safe bets? Or are they at high risk of getting heavily automated in the 2020s and 2030s?

The RAND researchers expect a 3% yearly increase in demand for anesthesiologists. That results in a doubling in 24 years. My guess is that at some point in those 24 years demand will peak and then start contracting due to automation.

In its baseline projection, assuming demand for services grow at the rate of 1.6 percent annually for both anesthesiologists and nurse anesthetists, the RAND study projects a shortage of close to 4,500 anesthesiologists and a surplus of close to 8,000 nurse anesthetists by 2020. However, if the growth in demand is assumed to be 3 percent to account for the aging population, the RAND study projects a shortage of physician anesthesiologists as high as 12,500 by the end of the decade. In this likely scenario, the surplus of nurse anesthetists by 2020 could be as high as 15,000.

The really highly paid and high skilled occupations are not immune to automation. Given that medical costs in the United States now exceed 17.35 of GDP and are on course to surpass 20% of GDP it makes sense to heavily automate the most expensive tasks in medicine. We can not afford the current trend in medical expenditures. Thing's got to give. What policies could increase the rate of automation of medicine?

Share |      Randall Parker, 2010 August 29 01:42 PM  Policy Medical

Basil Ransom said at August 29, 2010 2:22 PM:

I was well acquainted with an anesthesiologist about 5 years ago. I asked him if anesthesiology could be automated or at least de-skilled, i.e. simplify it so that a nurse could do the job. I know next to nothing about the profession, but at a fundamental level, the job consists of considering a number of factors to determine how much anesthetic to administer. Find a way to digitize all the relevant factors, and drawing on a database with millions of cases, a computer dictates how much anesthetic to administer. If possible, backwards validate it against past cases to see if it matches with the optimal dosage according to experienced doctors. My friend the doctor said automated administration of anesthesia was impossible, but couldn't explain why.

Randall Parker said at August 29, 2010 2:37 PM:

Basil Ransom,

He probably is repelled by the idea.

One doesn't have to achieve full automation to achieve huge cost reductions. If 2 anesthesiologists could oversee 5 operations with software that warns them of problems then that'd cut 60% of the cost of anesthesiology. It'd probably cut costs even more because with less demand for anesthesiologists their fees would go down.

Radiology strikes me as another specialty amenable to automation. Imagine what image processing algorithms will be able to do 20 years from now. Near instantaneous analysis with lower error rates.

macj said at August 29, 2010 3:06 PM:

A medical student, not an anesthesiologist, but maybe I can answer some.

At our hospital, one anesthesiologist could supervise a couple nurse anesthetists.

Anesthesia is more than just turning dials. They are responsible for the airway (kind of a big deal). They are also often the doctors that you see running the ICU.

My impression of robotic surgery is that as of today, it takes much longer to perform than open or laparoscopic operations because the surgeons aren't as experienced. I've seen hysterectomies with the DaVinci that take 2 hours with a traditional approach take 7 hours with the robot. Is that better for the patient. Hospitals have to buy the DaVinci because its a marketing thing, and in the future, you might be able to do operations with the DaVinci that could not be done any other way, but laparoscopic surgery is improving too. Robots still need a human operator (a surgeon) and does not change the anesthesia requirements for a case.

To the extent that medical therapy improves, this could be a substitute for surgeons and surgery. The breast surgeon I was with thought that in 20 years, breast cancer would have medical therapy only an that surgery as a treatment for it would be obsolete. Who knows if that is true or what it means for demand for surgery and anesthesiology.

Underachiever said at August 29, 2010 4:19 PM:

Basil and Randall,

I recently argued that a computer program should soon be able to replace many general practitioners. I argued that a computer program could diagnose general conditions as well as prescribe medicines, tests, and treatments. I used many of the same arguments that Basil just did with the anesthesiologist example. My pre-med friend was extremely skeptical, but she couldn't give any good reasons for her skepticism. I imagine that people just don't want to imagine that they could be replaced.

foobar said at August 29, 2010 5:24 PM:

Everyone's problem here is that they are thinking rationally, but there are a number of irrational forces in the healthcare system that will prevent most of these automation suggestions from being implemented.

It is important to distinguish between the regulatory framework for a board-certified physician and a medical device. There are different regulations in place for the "manufacture" of each, even if they otherwise are serving the same purpose.

For that reason, a device that does the job of an anesthesiologists -- like the one described by Basil Random -- is unlikely to reach the market, even if the device is on net better. It could cost tens or hundreds of millions of dollars to get approval for such a device.

Innovation in healthcare is heavily burdened by regulation. That includes possible rejuvenation therapies, but at least there's probably a large enough market to get big companies interested in those.

SteveSC said at August 29, 2010 7:53 PM:

Re: "I recently argued that a computer program should soon be able to replace many general practitioners."

Medicine is a 'high-touch' occupation. Devices and automation can't replace all physicians for the same reason print, TV, and Web ads haven't replaced all sales people. You can develop all the prediction algorithms you want, but sometimes it takes an experienced person to listen, observe, ask the right questions, put their findings in the context of an internal, often non-verbal 'database', and come up with the right action.

Of course, the low end of practitioners can be replaced by more consistent devices/software, but that will likely make the high end even more valuable.

Basil Ransom said at August 29, 2010 10:44 PM:

Foobar, I agree about the obstacles to such technologies (as I wrote below, before I saw your comment). Our best hope may lie with nationalized health care systems like the British NHS, which are always looking to save money. On the other hand, their costs and thus their incentives to cut costs, are not as great.

I tried to post this earlier, but had to step away from the computer for awhile:

Randall, I found a recent paper and press article discussing a device that reflects our ideas exactly, abstract and press article below:

I recently did a few minutes of Google searching on radiology and outsourcing. It turns out there is little to no offshoring occurring, as our laws pose a greater obstacle to automation and offshoring than do our current technical limitations. Currently, one must have the credentials to practice medicine in America to do radiology analysis anywhere. There is, however, common use of nighthawks, doctors who do the analysis miles away by computer. Apparently, there is little to no barrier to distance scan analysis (which is outsourcing). The messy web of legal liabilities, doctors' groups, medicare and government laws governing the practice of medicine will greatly slow the rollout of automation in medicine. It would happen after a combination of a rise in medical tourism and demonstrated success, most likely in other countries, to spur automated medical technologies.

Also, what's important is the reduction in workload due to automation, not the absolute elimination of need for human practitioners. The frequency with which the latter occurs is nothing, as compared with the former.

PacRim Jim said at August 30, 2010 12:34 AM:

By the time Democrats get through with the medical profession, anesthesiologists will be making minimum wage and/or the standards will be lowered to bring in thousands of marginally qualified hacks.

Biobob said at August 30, 2010 7:05 AM:

"By the time Democrats get through with the medical profession, anesthesiologists will be making minimum wage and/or the standards will be lowered to bring in thousands of marginally qualified hacks."

Yes, but they will be making millions from all those unrequired tonsilectimies and foot amputations, so I expect big things to come!!!

Eric London said at August 30, 2010 7:41 AM:

A design handles the exceptions as well as the typical. In my case, during a recent operation, my heart stopped, probably due to a reaction to the drugs used for the anesthesia. The surgeon performed compressions and the anesthesiologist pumped epinephrine, and obviously the two of them got things started again.

BBM said at August 30, 2010 8:49 AM:

As an actual surgeon, I think that probably the last people to get replaced will be surgeons and anesthesia.

The med student is spot on on every point. The anesthesiologist is the person that puts the person to sleep, puts in the breathing tube and takes it out, and manages the airway. Airway problems are the worst... you have only a few short minutes before the pt expires. Not computer program or someone that stayed in a Holiday Inn Express last night can manage that. This is true even of routine cases.

The DaVinci system is really cool and makes some ooperations that are ~impossible to do laparoscopic possible (only a small number of case types, like prostatectomy). But it currently takes substantially longer (setup time) than laparoscopic surgery for cases that are possible to do laparoscopically, and the pain/results will be no better than if the case were done laparoscopically (just more expensive). And it is several million dollars for the system. There are things coming down the line that will probably improve the time issue, though.

The US system is unique in its ability to stimulate development of new technologies and drugs. Yes, there are inefficiencies in the system, but eventually the entire world benefits. The US market alone pays the developmental and entrepreneurial costs of these developments. The rest of the world free rides off these developments paying ~cost for drugs etc. We have to be careful that health care reform does not destroy what is unique and valuable in the US system... because if it does, who will we sponge off of?

Black Death said at August 30, 2010 9:03 AM:

Some thoughts....

Most lay people and even many physicians don't really understand what hospital-based physicians (radiologists, anesthesiologists and pathologists) do and how their practices work.

Automation is a tool that physicians use to obtain better results - it doesn't replace them. Anesthesiologists use computer programs to optimize drug doses, but the computer doesn't take charge when a patient suffers cardiac arrest in the middle of a surgical procedure. Giving anesthesia is risky business, not to be trusted to amateurs. The computer can help, but it's the doctor who ultimately decides.

Similarly, the image-analysis specialties, radiology and pathology, use computer tools, but it's not the computer that makes the call about whether or not your lung comes out. That decision is made by a board-certified physician.

Robotic surgical devices are used because they generally produce better results. Robotic prostatectomies have a lower incidence of impotence and incontinence, and the patients generally leave the hospital the next day. But the robots are expensive, and the procedure takes longer.

Nighthawking is very commonly used in radiology ( Radiology studies obtained in the middle of the night in US hospitals are read in real time by radiologists mostly located in Australia, where the time difference makes it daytime. The nighthawk radiologists are mostly American grads, with US boards and licenses. They obtain privileges at the hospitals they serve and bill for their services just as other physicians do. Their Australian visas preclude them from other types of medical practice. Nighthawk is the biggest user of broadband in Australia.

cancer_man said at August 30, 2010 6:32 PM:

Doctors are going to be very surprised how their profession is turned upside down by 2020 (far fewer doctors and nurses, lower salaries) due to both robotics/screening innovations and pills that dramatically decrease the risk of getting cancer, heart disease and stroke.

There are regulations that may slow change down a bit, but once it is obvious how advanced computers and robots become later in the decade, watch Congress overturn these 435 to 0. The AMA can't stop this.

Dr. Leroy said at August 31, 2010 11:45 AM:

Don't know about anesthesia, but I'll give you the benefit of my 25 year experience in internal medicine.

Back when I started my IM practice in the 1980s the govt decided medical care was too expensive because us greedy doctors, gasp, charged for what we did. So by the early 1990s the feds socialized medical billing. If I bill Medicare more than the government says I can, I don't get paid and I may go to jail. The insurance companies quickly followed. Socializing doctors' fees saved money. It also drove doctors out of internal medicine. People smart enough to be doctors can make more money doing jobs whose fees have not been socialized. Now MORE THAN HALF the new doctors in IM training in the US are foreign medical graduates.

The feds have turned Doctor into another job Americans just won't do.

It must be the case that there are too few anesthesiologists because there used to be too much freedom -- and the federal government put a stop to that.

Anesthesia MD said at August 31, 2010 12:11 PM:

There are simply too many variables for the practice of anesthesia to be automated, and this is unlikely to change in the next 20 to 30 years. What you will more likely see is the increased utilization of nurse anesthetists (CRNAs) supervised by MD anesthesiologists. However, if the government takes over medicine entirely and decides to re-imburse all anesthesia at Medicare/Medicaid rates, you won't see any MD anesthesiologists or CRNAs. No one would be willing to take the the medical responsibility and legal lability for the $50k a year such a setup would provide after expenses (Not to mention the 60 hour work week). On the positive side, with no anesthesia, the government will save a bundle since there won't be much surgery.

Ron from IL said at August 31, 2010 12:24 PM:

From a recent study: (

CONCLUSION: Data from the American Society of Anesthesiologists, Closed Claims database suggest that anesthesia at remote locations poses a significant risk for the patient, particularly related to oversedation and inadequate oxygenation/ventilation during monitored anesthesia care. Similar anesthesia and monitoring standards and guidelines should be used in all anesthesia care areas.

AnesthesiaMD said at August 31, 2010 12:27 PM:

I am a board certified anesthesiologist, practicing for 25 years. The problem I see with automating my job is twofold. A large part of what I do is "Hands On", i.e. actual physical things I do to the patient, including but not limited to airway management, placing invasive vascular lines from a simple IV to an arterial line, a central venous catheter, or a pulmonary artery catheter. Placing nerve blocks using anatomic, nerve stimulator or ultrasound guidance, placing neuraxial anesthetics and catheters, i.e. spinal and epiduaral anesethesia, post operative and labor analgesia.
I am often the only and last physician to actually examine the patient the day of surgery to determine if there has been any change in his medical status that would preclude safe surgery at that time (and I have, catching a developing heart attack, decompensating asthma, new arrythmias, etc).
The actual administration of an anesthetic requires a synthesis of multiple factors including the patient's initial medical status, medications, type and passed as well as remaining duration of surgery, current vital signs, depth of anesthesia, and fluid balance, in conjunction with considering at least 5 different drugs which each perform a different function for a routine case, as well as over 100 other drugs used to intervene if things go bad.
I don't think I could write down what goes into my decision making process, much less come up with a program to do what I do. Remember, I spent 4 years getting a biochemistry BS, 4 years in medical school, and 4 years as a resident before I was allowed to have the primary resposibility for a patient as an attending physician. I earn between 80 and 100 credit hours of continuing medical education a year to keep my skills current.
Do you really want to replace me with a computer program and a tech with 6 months training when you are on the table?

Ken Royall said at August 31, 2010 12:31 PM:

As a person who has had operations and been a frequent consumer of medical services, AND who is also a software engineer, I have no desire to see any automation of the kind being discussed here. Automation is great but it is not the answer to every question. There are root causes as to why their are fewer anesthesiologists, those should be addressed. I would imagine that if you go back far enough government meddling is at the root of it. The market will fill the void if it is allowed to work properly.

looking closely said at August 31, 2010 1:00 PM:

Show me the computer algorithm that can intubate someone with laryngeal edema, and I'll show you the day when anesthesiologists need to be concerned about having their jobs computerized.
Show me the computer algorithm that can tell if someone is properly oxygenated *without ANY external monitors*, etc.
Show me the computer algorithm that that spot a drug addict on sight and compensate drug dosages accordingly (including denial of meds), etc.
Show me the computer algorithm that can put in epidural anesthesia into a nervous 300lb primagravid woman, etc.

Now show me the patient who wants that computer and a tech doing their anesthesia (which is generally more risky than the actual surgery) instead of a trained and experienced anesthesiologist.

Ain't going to happen.

The more you try to dilute anesthesiologists by spreading them thinner with computer and techs, the more likely you are to run into problems that only the anesthesiologist can fix.

Felix Kasza said at August 31, 2010 1:08 PM:

> Given that medical costs in the United States now exceed 17.35 of GDP and
> are on course to surpass 20% of GDP it makes sense to heavily automate the
> most expensive tasks in medicine.

I disagree with this rather fundamental premise. The cost driver in medicine is the patient, not the doctor. A hundred, no, fifty years ago, if you had cancer, surgery might be attempted in some cases, but mainly your GP held your hand while you expired at home, in your own bed, palliated by cheap morphine. Hand-holding, at $100 a pop, is far less expensive than MRIs, CTs, PETs, -mab and -mib drugs, and lengthy ICU stays.

Give up the desire to eke out those last few days of life, be quit with maintaining unaware and completely demented Alzheimer's patients on three pressors in the ICU while they are actively trying to die, and the amount we spend on medical drops by half or more. (It would, of course, be a disaster for doctors who mostly bill procedures, but if such a radical change ever comes, it'll take us decades to make the transition.)


David said at August 31, 2010 1:10 PM:

Lets put it this way, the anesthesologist is the person trying to keep the jerk surgeon from killing you. Are you sure that you want this to be a computer? The number of things that can happen in surgery can not be listed. It requires a human and one that can think, not an idiot nurse. Having been a doc for 40 years, I am all for people taking care of themselves (like most doctors). I do not believe in our current way of delivery of service and feel you should get any med you want anytime, any where, no problems, no identification, no prescription. Then if you need the doctor, go to him.

looking closely said at August 31, 2010 1:11 PM:

Sure, with actual films being obsolete now, and most radiologic images being generated purely electronically, there is no longer a need for a radiologist to be physically present at the facility where the image was generated any more. So long as you have a good tech taking the images, they can be transmitted nearly anywhere instantly electronically and the radiologist could be interpreting the image from the computer monitor across the hall, in their bedroom, in a different hospital, or even in a different country.

But to the ones who think that you're going to replace radiologists with fancy image-reading computers, that ain't going to happen anytime soon.

Proper interpretation of medical images is incredibly complex. The radiologist has to draw on all sorts of shape and pattern recognition, color recognition, evaluate radiologic and other artefacts, aberrant anatomy, make judgements including specific patient information and history, etc. This requires tremendous amounts of parallel processing and is a lot more sophisticated than any computer can do right now.

Ditto for interpretation of pathology slides. The best currently available machines are OK to SCREEN (not interpret) pap-smear slides, which are about the simplest interpretation in all of pathology. . .you're basically just comparing nuclear size. But even THAT technology, which has been in evolution for over a decade, still can't detect OTHER abnormalities in pap smears (like infections) and isn't reliable enough to certify results without human pathologist confirmation.

John T. Gregg, MD said at August 31, 2010 1:19 PM:

I find it amusing and not a bit sad, that in this decade – or two or three – of whining about the impersonal nature of medical care, one would hear proposals for the computerization of medical procedures – “plug them into the hose and let the robot cut/pierce/lase…” – are you kidding!!!
On the one hand, everyone picks on the cognitive specialist trying to TALK with you and EXAMINE you for 4 problems in15 minute exam, the span dictated by the blunt economics of insurance/Medicare/Medicaid reimbursement. On the other hand, failure-to-diagnose lawsuits – diseases, drug interactions, complications, lifestyles, etc – lead the pack.
The specialist must quickly, ably and cheaply diagnose, fix, and follow whatever simple/complex problem you may (or may not) have – without failure and with little work on your part, despite whatever instructions you are told are important. Any risks are on him/her, and your medical conditions/lifestyle habits/whatever are no excuse for failure of perfection and beauty in the outcome.
Three things to know:
1. Medical care – cheap, quick, quality – pick two of the three;
2. Quality is provided by personal interaction – your needs, situation, condition – not by referral to “everyone’s” needs – population dynamics and statistical studies are necessary but not sufficient guides to diagnosis and treatment;
3. Thus far, the US medical care system is where “everyone else” goes when they cannot get it at home – OB care for the illegal population, heart surgery for Canadian bigwigs, transplants for shady oriental characters – where will you go after we are in fact just like the rest?

fiftyville said at August 31, 2010 1:39 PM:

I would no more trust an automated anesthesia device than I would fly as a passenger on a completely automated aircraft. Why?

Experience: Unless the programmer has had extensive experience in anesthesiology, no one has the time or patience to sit and make sure the information transfer between expert and programmer is complete, or to look over the programmer's shoulder and make sure that the information was understood and correctly implemented.

Quality: Is the software developer experienced in medical software, or was his firm the low bidder and the last project he worked on a video game involving zombies?

Support: If something goes wrong, is expert help immediately available 24/7, or will there be a wait queue for someone named "Bob" in Bangalore whose English is just a little hard to follow?

Security: Is the device subject to infection from computer viruses via an infected USB drive or inadequate network security? Is the operating system reliable and not in need of frequent patching? Is it vulnerable to power spikes? Battery backup?

It is impossible to adequately plan and program for all possible medical contingencies, and the middle of an operation is no place for finding that out. The perfect device is a well-trained and experienced human.

Paul Manner, MD said at August 31, 2010 2:31 PM:

Fair questions. As an orthopaedic surgeon, here are my answers.
1. Is it necessary today to have 1 anesthesiologist per operation? Or could a small number of anesthesiologists monitor many operations with the help of software and assistants? How hard would it be to develop software and equipment that would enable anesthesiologists to handle many operations at once?

Let's rephrase that. "Is it necessary today to have 1 pilot per airplane? Or could a small number of pilots monitor many planes with the help of software and assistants? How hard would it be to develop software and equipment that would enable pilots to handle many airplanes at once?"

If something goes wrong with the patient's airway, you've got about 4 minutes before the patients is braindead. That's not a lot of time, particularly if you have to hit CTL-ALT-DEL to get the computer to reboot. I'll pass.

2. How long until software can do as good of a job as anesthesiologists?

We have software to analyze billing and reimbursement coding. It's wrong about 10% of the time. And billing and coding aren't changing minute by minute (just feels that way).

3. What are the prospects for reducing by anesthesiologist demand by speeding up operations. For example, is duration of surgery shortened by current generation robotic assist surgical equipment such as the da Vinci surgical system? Will future robotic surgical devices operation autonomously much faster than human surgeons can?

Nil. Robots are a pain to set up, they take up space, time, and effort, and they slow procedures significantly. Robodocs were tried for total hip replacement for years; on average, 20% would have to be switched to manual because the robot failed, the revision rate was higher, and the robots shredded the muscles around the hip. They're no longer used.

4. Will rejuvenation therapies result in an increase or decrease in demand for surgeons? For example, will stem cell therapies decrease the demand for surgeries more than installation of grown replacement organs will increase demand for surgeries?
Name one rejuvenation therapy that actually works. You still have to get the stem cells in there. It's not as if you can drink them and they magically go where needed.

5. Will effective cancer cures via, for example, nanotech-based precise cancer cell killers reduce the overall demand for surgery? Or will avoidance of death from cancer (and elimination of need for cancer surgeries) just make people live longer to need many other kinds of surgeries?

Possibly; antibiotics and proton pump inhibitors have essentially eliminated surgery for peptic ulcers. I doubt that's done much for overall costs of health care, although the patients are better off. More likely, people will simply live longer, which drives up costs.

dr kill said at August 31, 2010 2:47 PM:

Yeah, well, my idea is that you losers use a computer for anesthesia, and you losers use a robotic surgeon. Give me the American medical school graduate with 5 or 10 years of experience every time. What is wrong with you people? Gas-passing is an art, not a science. I can't describe it to you either, but I know competence when I see it. Stupid idea, stupid, uninformed and ignorant comments.

Narniaman said at August 31, 2010 2:49 PM:

I have an interesting background, inasmuch as I have a degree in electrical engineering, worked for years as an engineer, went back to medical school, did a general surgery residency, practiced general surgery for many years, and have also been involved in the development of a fairly wide variety of medical software.

To suggest that a computer with a sophisticated software system could take the place of an anesthesiologist is absurd. A computer robot is not going to be able to intubate the patient -- especially when it's a crash intubation and the patient is vomiting. Is a computer going to be able to get a central line in on a patient who has a ruptured aortic aneuyrism and their BP is 60/0? Is a computer even going to be able to place a ng tube? How long is it going to take the computer to place a epidural catheter?

How good is the computer going to be at even listening to lung sounds? Is it going to recognize that the reason the patient is hypoxemic is that the endotracheal tube has gone down the trachea too far?

Computers are useful and powerful and very very fast -- but also very, very limited in what they can do.

dr kill said at August 31, 2010 2:57 PM:

Except for the comments from the MDs. I salute you, especially the surgeons. Imagine spending 12 years learning to do bypass surgery and having most of it replaced by plastic stints. I would never recommend anyone go to med school. Human med will end up like vet med, all women and all part -time. Or all Nigerian. You are welcome to them. I'll pay cash and see the real doctor. There will always be a market for competent physicians with discerning Americans.
To further my rant, is there anything the Feds have co-opted that hasn't led to a burgeoning cash industry for the alternative? Education, agriculture, if the Fed is in it, someone is offering an alternative that knowing Americans find attractive.

Spocougar said at August 31, 2010 3:30 PM:

I agree with the comments by the physicians above, for the most part. I'm an anesthesiologist. We are just now (as a profession) integrating computerized charting. The system we use now is very good, but regularly crashes. When one crashes, they all go down across multiple operating rooms. What are you going to do when your computer doc crashes? Who's going to respond? A computer geek? In this profession you can go from a totally healthy person to a vegetable in a few minutes. You have to have someone on site. Beyond those technical issues, there is a reason why we go to 8 years of medical school and residency training, and then wait a year before we can be board certified - and only after passing a written and oral exam. There are more variables for any given patient then could reasonably be programmed and accounted for. Creating a computer to provide anesthesia (disregarding all of the actual hands on stuff we do) would be like trying to exactly predict weather a week away with NO TOLERANCE for error. The thought that anesthesia is just measuring the right amount of anesthetic and delivering it is a misinformed perception.

A couple of years ago I took care of a teenager who was severely mentally handicapped, had physical contractures in all of her extremities, extremely developmentally delayed, and had no real quality of life. She required intensive care in every aspect of her life. Turns out that she, at the age of 3 or 4, was a totally normal little girl that went in to get a tonsillectomy. During the surgery the oxygen level in her blood was very low for just a few minutes (not sure on how that happened, but that is the responsibility of the anesthesiologist in the room) and the result was a massive brain injury that led to her current condition. I have a 4 year old daughter. Would you want your little girl in the hands of a computer, or a highly trained doctor? Things can still happen badly with the doctor, but I'll take him/her any day of the week - and that's not going to change this decade or any time soon.

not anon or anonymous said at August 31, 2010 4:23 PM:

"Would you want your little girl in the hands of a computer, or a highly trained doctor?"

If you have ever boarded an airplane, then you have in fact trusted your life to extremely complex computer systems, which account for dozens of variables and have very limited tolerance for errors.

Of course, real life-critical computer systems are built to exacting standards: the hardware is highly redundant and fault-tolerant, the software is often formally proven to be bug-free, and obviously they don't "crash regularly and bring down multiple operating rooms". I would readily trust my anesthesia dosages to a computer system if it was built to similar standards.

Anesthesiologists will still be needed to handle the airway, deal with non-typical patients etc. But having a computer system deal with the routine cases would reduce the effort expended by humans, thus enabling better care in unforeseen contingencies as well.

kmh said at August 31, 2010 4:27 PM:

All scientific progress, including medicine and economics is not linear. Any of the above might be reasonably accurate out for maybe as long as a decade. After that, anticipate a paradigm shift of some sort, either supply or demand or both. Two potential examples: 1) Some breakthrough in rendering people unconscious other than medications using a technique that does not require human monitoring. 2) A sudden drop in demand for surgery--general surgeons laughed when cimetidine came out and a few pundits said this could be the end of the need for emergency gastrectomies due to bleeding ulcers. 3) Changes in the reimbursement mechanism that puts increased value on cogntive rather than procedural services.

married to ICU RN said at August 31, 2010 6:23 PM:

Can you hold your breath long enough to re-boot the Robot? What about your child, or your Mother? Of course you will need Anesthesia Doctors and CRNA's always one breath away. When it comes to things that can't be replaced by robots, real-time anesthesia is at the top of the list.

cancer_man said at August 31, 2010 7:46 PM:

Not surprisngly the doctors and nurses are missing the point. Robots and diagnostic software of 2020 would astonish them if they could go into the future ten years where computer power will be more than a thousand times as powerful as today. Rejuvination technology is still at an early stage (although you cah see a finger tip being regenerated using pig stem cells on youtube)but it will be routine by 2020.

Getting back to Randall's question about a 20 year old med student chosing surgery or anesthesiology as a career path.

He or she can anticipate that when they are 30 (in 2020) or maybe 35 (in 2025), they will not be doing that job. Along the way, it is likely that the salaries will drop as well.

For the doctors showing their decade of training, I'd like to see them take _current_ final exams in chemisrty, microbiology as well as pass their boards. If the AMA is so concerned about quality, then all doctors should be required to pass the tests every 7 or 8 years to make sure they are current.

Randall Parker said at August 31, 2010 8:23 PM:


You can rail against cost-cutting measures. You can view various automation schemes as impossible or absurd or foolhardy. But the US economy can not sustain the rate of medical cost increases that it has experienced for the last few decades. When medicine was 3% of GDP it could double. When it was 6% of GDP it could double. Once it hit 12% of GDP one last doubling probably became a bridge too far. We are now at 17.3% of GDP for medicine and rising. That rise can't go too much further.

Cost cutting pressures are going to intensify. Either some tasks have got to become automated or less care must be delivered. If you tell me that automation isn't practical then where must the care cuts come from?

Randall Parker said at August 31, 2010 8:58 PM:

Regards the references to pilots flying multiple airplanes: I suspect some people here are not aware of how automated airplane flying has become. Airbus especially has reduced the number of decisions that pilots make:

But planes built by Airbus, a European consortium based in Toulouse, France, give computers, not humans, the final authority on flight decisions.

Airbus' American rival, Boeing, also uses fly-by-wire systems on its newest planes, but their pilots have the ability to override the computers in an emergency.

As Airbus puts it, the "deflections of the flying control surfaces on the wing and tail are no longer driven directly by the pilots' controls, but by a computer which calculates exactly which control surfaces are needed to make the aircraft respond as the pilot wishes."

Inevitably computers will make mistakes that crash an airplane. But pilots make killer mistakes too. I expect computer mistake rates to continue to drop while pilot mistake rates do not drop as much.

Some fighter jets are so aerodynamically unstable and require so many rapid adjustments that computers must operate the control surfaces. Human decision-making loops are too long to control them.

Automated Carrier Landing Systems demonstrate the possibility of pilot-less flight. Predator drones are also getting automated landing systems.

Then there are cars. First came Anti-Lock Braking Systems or ABS. But car computers have gone further. Heard of Electronic Stability Control? Perhaps you drive a car with ESC? Did you know it adjusts both throttle and brakes? Required on all cars in the US by 2012. Car driving automation will continue its incremental advance. Automated parking systems take over parallel parking in some luxury models.

DARPA is taking vehicle automation much further with its Grand Challenge for driverless vehicles. The vehicles in this contest keep getting better and so DARPA has upped the challenge into the Urban Challenge with a more complex environment.

darwinian roadkill said at August 31, 2010 10:46 PM:

A link to a story on McGill's University "McSleepy" robotic anesthesiologist system.

Delfin J Beltran MD said at August 31, 2010 10:47 PM:

The fundamental problem with the concept of replacing a human being with "automation" in any task is the belief that mankind is capable of all things great and small combined with the belief that if you wish or pray hard enough it will exist in reality. I had the privelege of growing up with open heart surgery as an active anesthesiologist from 1956 to 2004. My experiences ranged from using the first Bird ventilator (serial number 00001) to working with two of the three inventors of computers, Doc Hurd and Dean Brown. When I went to work at Stanford in January 1968 I found it necessary to develop the clinical criterea to define brain death and support that diagnosis with a valid clinical laboratory test, as I was the medical director of the four beds of ICU that existed at the time of Dr.Shumway's first heart transplant and those patients who were donating their heart were my medical-legal responsibility. The concept of " Quality " in medical care was and is a political farce designed to impower ignorant humans with the power to legislate away their lives. The 1990 Five year Health Plan had one paragraph that stated that " Medical Quality " was indefineable, the only true statement in that volume, followed by about fifty pages of gibberish fleshing out an invalid concept. Similarly the concept of technology being created to solve a problem, has been well demonstrated to be a new tool that is capable of solving an increased range of situations previously defined as untreatable with existing tools. In the process of implementing the new innovation, previously unknown problems are also revealed only to await the inventive genius of another innovator to define the new solution. These ideas of solving humanities conflicts with life have exposed more human fraility, fraud and human greed and dishonesty that are without soulution because of the imperfection of man that has occured as an undesireable consequence of that imperfection.

cancer_man said at September 1, 2010 1:03 AM:


Can you name one airline crash that was caused by computer failure in the past 25 years? The plane that "landed" on the Hudson River last year was on auto pilot when the "hero" pushed the button.

I talked to a couple of pilots recently and asked when they thought they would be out of a job. Both said "never" because even though it could essentially be done today, nobody would want to fly.

But we will see companies outside the U.S. offer cheap flights with a cheap pilot with little experience , or no pilot, and when people understand there are no crashes, the entire industry will move to pilotless flights. Sometime around 2020.

Just Wonderin' said at September 1, 2010 6:44 AM:

Can someone explain why higher growth in demand (3% vs. 1.6%) would result in a greater oversupply of nurse anesthetists (15,000 vs. 8,000)?

CRNA said at September 1, 2010 7:51 AM:

I have administered anesthesia for over 30 years. People who suggest that it can (or will) be somehow administered by "computer program" understand neither anesthesia nor computers.

JoelP said at September 1, 2010 10:55 AM:

It's true that autopilots are useful in flying specific models of plane through specified weather patterns. Show me the autopilot that can fly any plane (from an Airbus to a Zero) through rain, shine, or hurricane. Then we can talk about replacing anesthesiologists with computers. Computers can chart (though today they fail at that too often). They can calculate drug dosages. They'll probably be able to intubate in a decade. But anesthesia is all about figuring out the weird unspecified situations and how to manage them, and that's what computers are worst at.

Nick G said at September 1, 2010 12:25 PM:

Back in 1980 I saw research that demonstrated that automated patient interviews produced patient histories that were well above the average quality produced by physicians. Similarly, automated diagnostic systems in the ER were found to produce better diagnoses than those of all but the very best physicians.

Where are the new and improved descendants of those systems today?

I'm afraid the pace of change is a lot slower than we'd like.

Nick G said at September 1, 2010 12:31 PM:

anesthesia is all about figuring out the weird unspecified situations and how to manage them, and that's what computers are worst at.

I'm skeptical. I think you may be paying too much attention to zebras*. I'd be more concerned about the incompetent, hung-over or inattentive anesthetists that compromise surgery outcomes.

If you want to hear about physician mistakes, talk to support staff, like nurse, instead of the doctors.

*old medical aphorism: when you hear hoofbeats, think horses, not zebras. IOW, don't look for interesting, exotic diseases in preference to common problems.

Fred said at September 1, 2010 5:40 PM:

"Where are the new and improved descendants of those systems today?
I'm afraid the pace of change is a lot slower than we'd like."

A doctor who is now in his 80s showed software was better at making correct diagnosis than doctors and lobboed Congress to allow their use, but the AMA easily blocked him and the general idea. A couple of doctors interviewed said the same as some here have written: "I've had 8 years of schooling, then residency, etc." The interviewer forgot to point out that the doctors remember only a small fraction of that schooling they had in the 1970s.

The pace is likely to be slow a few years longer, but eventually a critical mass forms and rapid change will occur. This is bad news for most of the doctors and nurses in terms of work since most will have to retrain for something, but it is great news for everyone including dotors and nurses who will benefit from all of the coming advances.

Burt said at October 22, 2010 2:32 PM:

Wow. A lot of thoroughly uninformed conversation here.
The practice of anesthesiology requires broad technical KNOWLEDGE, expert MANUAL SKILLS which are honed through years of training in med school and residency, and most importantly JUDGEMENT accumulated over years to avoid trouble in the first place and to resolve difficult medical problems when they do inevitably arise.

To think this is all going to be automated or farmed out to marginally trained technicians with good results is naive and reflects a very poor understanding of what goes on in the operating room. It is this kind rather uninformed thinking that is resulting in the current legislative process of dumbing down medicine. There is a strong and dangerous trend toward the utilization of less well trained personnel thoughout medicine. The driving force is of course economic. It is a very disappointing development and the quality of medical care will indeed suffer if it continues.

Dr. Burt

Post a comment
Name (not anon or anonymous):
Email Address:
Remember info?

Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright ©