« Sweden fact of the day | Main | Should we diversify our charitable giving? »
Why is Medicine so Primitive?
The practice of modern medicine is surprisingly primitive. My doctor only recently started to provide printed prescriptions instead of the usual scrawl. Incorrectly filled prescriptions can be serious and computer printed prescriptions are an obvious response yet even today only one in four physicians use some form of electronic health records and only one in ten really use electronic records to follow a patient's entire history. My credit card company knows far more about my shopping history than my physician knows about my medical history.
Medicine is primitive in another way. The number of treatment regimes supported only by tradition and authority is very high. Here's a recent example:
For the past 30 years or so, doctors have routinely given pregnant women intravenous infusions of magnesium sulfate to halt contractions that can lead to premature labor.
...[a] team reviewed 23 clinical trials worldwide involving 2,000 women who had received the drug to quell contractions. They found that it did not reduce preterm labor and that more babies died when their mothers took the drug than in a control group where the mothers had not been given it.
...Grimes and Nanda estimate that about 120,000 American women receive mag sulfate each year for premature contractions, and they say some evidence suggests it may be associated with 1,900 to 4,800 fetal deaths annually in the United States.
This would be a shocker except for the fact that stories like this are common - by some accounts a majority of medical procedures are not supported by serious scientific evidence. Indeed, what are we to make of a profession where evidence-based medicine is only a recent and still far from accepted movement?
Why is medicine so primitive? One reason is that medicine is the largest area of the economy still dominated by artisanal production. I will be blunt: We need assembly line medicine, medicine that is routinized, marked and measured. As I have argued before I would much prefer to be diagnosed by a computerized expert system than by a physician. The HMOs, Kaiser in particular, have done good work on measuring the effectiveness of different procedures but much more needs to be done to bring medicine into the twentieth century let alone the twenty first.
Posted by Alex Tabarrok on October 16, 2006 at 07:20 AM in Medicine | Permalink
Comments
How dare you question practitioners of a field without yourself being in that field? /sarcasm (reference to Caplan's paper on mental illness and the reactions thereto)
Posted by: Constant at Oct 16, 2006 7:40:56 AM
How cleancut is the relationship between artisanal practice, and the technical and organizational obstacles?
My primary care physician has my history on paper: the efficiency and quality benefits of electronic records really show up when the specialist I see doesn't have to get all the data by hand. The portability of electronic records makes switching physicians/hospitals/systems cheaper, so organizations have less incentive to invest and, more importantly, standardize.
This standardization woud make research to develop evidence-based medicine much cheaper.
Further question: if you wanted to make medicine more routinized, should you create policies that favor the insurers/payers or the hospital systems (e.g. Partners in Boston, UPMC in Pittsburgh)?
Posted by: Allan Friedman at Oct 16, 2006 7:50:32 AM
The problem is that the physicians (or any other single group) just do not own or control enough of the process to make end-to-end computerization worthwhile.
If you really want to use an industrial process, maybe someone can point to a company that has successfully implemented end-to-end enterprise resource planning successfully. There are many more stories of failures in ERP than success. Why would medicine be any different?
Posted by: superdestroyer at Oct 16, 2006 8:19:27 AM
One key issue that prevents the 'continuous improvement' culture that has benefitted manufacturing assembly lines is medical malpractice litigation. When I was in medical school, every specialty had a monthly (sometimes weekly) meeting usually called 'Grand Rounds', where errors in care were discussed and the sources of those errors aggressively probed. Some years later we were told to 'be careful' in discussing specific errors because the discussions could open the door to malpractice litigation. Soon after that, Grand Rounds devolved into the presentation of academic papers and consumption of food provided by pharmaceutical companies.
Concurrently, the number of autopsies declined to nearly zero. Without these quality feedback systems many doctors are unaware of how their colleagues are practicing and what works. Even in many group practices, a culture has grown up that ignores 'practice variations'. With a dearth of evidence, and to prevent internal conflict, doctors will 'sign off' on medical orders that they disagree with, simply because the patient 'belongs' to the other physician and they are simply 'covering' while on call.
Medical malpractice attorneys like to claim that they are the only check on poor doctors and bad medical practices, yet the industry also has a hand in preventing effective use of other quality enhancing methods. Cultures of continuous improvement in manufacturing require a 'no-fault' environment to discover and analyze unintentional errors. Imagine how ineffective this method would be if any time a worker 'stopped the line' because of a problem, one or more of his/her coworkers would be hauled into a multi-year adversarial procedure that could result in heavy fines or the loss of jobs.
Posted by: SteveSC at Oct 16, 2006 8:35:15 AM
Alex,
You wrote: "I would much prefer to be diagnosed by a computerized expert system than by a physician."
It may well be the case that YOU would be better diagonosed by a machine, but I would argue that this is because you are such an admirably rational chap. The problem is that there are lots of people who are so introverted and/or warped that their best hope is to develop a trusting relationship with their physician that facilitates their being transparent about their problems, history, and life style choices. Don't forget Alex that we are social animals; a good physician is like Sherlock Holmes unraveling a mystery in these cases; a machine could have solve few, if any, of the cases the Holmes (and Watson) unraveled.
So announce your view that "I would much prefer to be diagnosed by a computerized expert system than by a physician" again, and again, and I will agree with you every time, for YOU that is.
Posted by: jim at Oct 16, 2006 9:13:08 AM
One problem preventing improvement here is the doctor's self-conception as a scientist. But, the vast majority of doctors are not scientists in any plausible sense. They are closer to auto mechanics or, at best, enginers than to scientists. But, by thinking they are scientists they become very reluctent to take outside instruction from actual scientists.
Posted by: Matt at Oct 16, 2006 9:56:34 AM
"Medical malpractice attorneys like to claim that they are the only check on poor doctors and bad medical practices, yet the industry also has a hand in preventing effective use of other quality enhancing methods."
Sounds very similar to the effect of Sarbanes-Oxley on the financial industry. You'd be amazed at how many technologies and quality improving things are eschewed just because they enhance the ability of future litigants and/or SEC investigators to misconstrue something or expose the firm to undesirable and pointless litigation. SOX protects investors by decreasing shareholder wealth.
Posted by: Johnny Debacle at Oct 16, 2006 9:59:08 AM
There are more failures than success in ERP? I guess you should contact the CIO at Walmart and tell her to immediately stop spending on IT.
Posted by: bjak at Oct 16, 2006 10:17:51 AM
Medicine in many cases is best seen as a social ritual that brings comfort, rather than a set of biological interventions that bring about therapeutic results.
Indeed, the degree to which the social ritual brings comfort is responsible for a large component, perhaps the major component, of its therapeutic effect -- the placebo effect. This makes me question whether treating medicine as an "industrial process" is likely to succeed, or is even desireable. There is some debate as to whether the placebo effect improves objective outcomes (ie - can cure cancer) but most medical treatment is used for chronic conditions as a sympomatic relief -- and there is no doubt of its effectiveness there.
Of course there are interventions that are necessary and effective, but tough to know exactly which ones.
BTW, I'm somewhat upset to hear the specifics about magnesium sulfate therapy. They put my wife on it for our first child. Her pressure crashed after delivery and she almost died, wonder if it was from the mag sulfate. . .
Posted by: Matthew Cromer at Oct 16, 2006 10:19:42 AM
I agree that medicine needs to become more standardized and data driven, and less artisanal. Kaiser and Group Health have accomplished a lot here, but another example of progress (libertarians take note) is the Veteran's Administration.
Posted by: Bill Gardner at Oct 16, 2006 10:43:53 AM
You want to read "The Score" by Atul Gawande in last week's New Yorker (http://www.newyorker.com/fact/content/articles/061009fa_fact). It's a fascinating study of the change in medicine (Caesarians) due to the introduction of metrics. (The Apgar score quantifies the status of the newborn, giving doctors something to shoot to improve. Use of the forceps can't be taught, it's a craft, while the Caesarian can be taught.)
Posted by: Bill Harshaw at Oct 16, 2006 10:45:29 AM
Atul Gawande recently addressed this distinction AND touched on evidence-based medicine in his article for the New Yorker, THE SCORE. As always, it is well-written.
www.newyorker.com/printables/fact/061009fa_fact
You should read Don Berwick. He's started an Organization called the Institute for Healthcare Improvement (IHI). I think you'd find the stuff of his 100,000 lives campaign thrilling.
http://www.ihi.org/IHI/Programs/Campaign
Finally, as for your assembly-line comment, I think that it is one thing to say "this is broken" and quite another to say "this will fix it." The problem of care that is "routinized, marked and measured" is much more difficult than you may have imagined.
This article will get you started, and the second link is for a treasure-trove of related articles.
http://content.nejm.org/cgi/content/full/350/23/2409
http://www.healthcareproblems.org/Bibliography/BIBMeasuringQualityOfCare.htm
I enjoy your blog.
Posted by: topher at Oct 16, 2006 10:49:29 AM
Matt: your comment seems awful harsh considering this is an economics blog. Economists are certainly closer to auto mechanics (diagnosing what went wrong or right after the fact) than scientists (using experiments designed to avoid bias). They also often seem willing to denigrate the non-licensed class of "armchair economists." And just like your doctor, they probably are right that their training and experience gives them a leg up in diagnosing problems. Sometimes this is correct, but it should not be used as a cudgel to discount or dismiss opponents to the conventional wisdom.
Also for Alex: since medicine is lagging as primitive, where would you rate the field of public health? They certainly have been taking a lot of flack from the right this week (with the Lancet article), but public health interventions have far more proven results than most medical treatments...
Posted by: Dan K at Oct 16, 2006 10:52:02 AM
Perhaps primitive is not the right description to describe medicine as
practiced today, but certainly it is out of step with the times. As a
practicing internist I see two problems that make it difficult to make
changes. First, doctors are paid on a fee for service basis, and payment
for services is falling and costs are rising. Technologies such as e-mail
and electronic medical records require a sizable investment of capital
with only a marginal increase in productivity and no increase in
reimbursement. There are many federal laws such as HIPPA which drive up
the cost to implement technologies, reduce efficiency in a practice,
and do not increase payment.
Second, our legal system makes it difficult for doctors to enter the
21st century in the way we practice.As it is, evidence based
medicine is not the standard of care in most states.Under our current
litigation system, doctors are judged by the standard of care as
practiced by expert testimony, not on the basis of sound evidence based
medicine. Many doctors have lost lawsuits because their care was
appropriate by the standards of evidence based medicine, it may not have
been the standard of care according to an expert witness.I don't see
this issue changing until our medical litgation system is overhauled.
Posted by: DrTom at Oct 16, 2006 11:13:58 AM
I second what Dr. Tom has said above.
Medical practice is very much a "cottage industry" or artisan business if you like. This makes implementation of EHR or other uniform systems very difficult. Add to that the daily time and money pressures and modernization is very difficult.
As far as diagnosis, I prefer a real human physician, who can spot physicial, social and emotional aspects of my condition that computerized lab tests never could. Internist and family docs are absolutely amazing much of the time for their ability to pull together a great deal of data in a big hurry and be right.
And of course human behavior drives much of of health system (quit smoking, eat less, exercise more, etc.).
Posted by: save_the_rustbelt at Oct 16, 2006 11:38:33 AM
BTW, the 1998 study referenced seems to have involved 93 pregnancies in a study originally designed to test whether magnesium sulfate prevents cerebral palsy. They got 7 fetal deaths from 46 patients on magnesium sulfate, and 0 from 47 patients taking similar drugs. I can only access their abstract, but I would really like to read their Methodology section. The problem is this: they claim 15% excess deaths, w/ the interval going from 11%-19%. So if 120,000 mothers took magnesium sulfate last year, that would have led to a minimum of 13,200 fetal deaths.
So Grimes and Nanda are basing their estimate of 2k-5k fetal deaths on a study that suggests 13k-24k fetal deaths. Can you really have it both ways? If the number of fetal deaths couldn't possibly be in the range suggested by the study, that implies that the study was a mistake, and so it can't be used to support an estimate of 2k-5k deaths. Can someone wiser in the ways of stats-fu explain this to me?
Posted by: anonymous at Oct 16, 2006 12:30:45 PM
Oh, in relation to the explanation of the disparity between Grimes and Nanda's estimates and the study; I don't want you to tell me that its only a 95% confidence interval, so there is a 2.5% chance that it is below that. I know the results are conveivably consistent. It would also be conceivably consisent if G&N were claiming that magnesium sulfate *saved* lives. But if the real number of fetal deaths from mgso2 were 5k, the implication would be that there was a serious error in the methodology of the 1998 study, so that study could not be used to *support* the 5k estimate. I think that explains why I am puzzled
Posted by: anonymous at Oct 16, 2006 12:35:34 PM
Part of it could be doctor training. I'm a phd student doing research on computer aided detection and diagnosis for radiologists. Radiologists are trained by viewing thousands of images. Any new tool or change in presentation can really throw them for a loop, since much of their training no longer applies. To introduce an expert system would require some training to figure out how to incorporate it.
Part of it is also workflow. In many cases doctors would rather scribble their prescriptions and move on than take a minute to find a computer and print a prescription. There are lots of cases in radiology where the most modern visualization methods (3D, color maps, etc) just take too long to prepare and are thus never used by the doctors. Even retrieving old images can take too much time.
Finally, doesn't this behavior happen in other fields? I read the book "My Life as a Quant" and remember some of the options pricing discussion. When the traders(?) needed a tool for bond options they adapted Black-Scholes very quickly because there was nothing else, but when the quants came up with a more valid model the traders took a lot of convincing.
Posted by: Richard at Oct 16, 2006 12:36:18 PM
The New Yorker article by Gawande is highly recommended. It illuminates 'process' and why medicine changes to X and not Y. Or doesn't change.
Topher provided a link - see above.
I suspect the best way to improve medicine is great training and better conditions at the second level - nurses, paramedics, techs, and even the billing and insurance staff.
And availability is important. If a surgeon moves to Beverly Hills and will only do breast implants for thirty years it doesn't really matter if he scribbles the presciptions, he is not 'available' as a physician in any meaningful way.
Posted by: K at Oct 16, 2006 1:35:53 PM
The first answer that leaps to my mind is that the supply of doctors is artificially limited by the AMA (mostly at the state level). And there are many other gov't interventions that prevent doctors from getting the market pressure that would lead to the rapid progress in medicine that we have seen in less regulated industries like computer technology.
Posted by: Stephen W. Carson at Oct 16, 2006 1:42:40 PM
I agree with "Matt" - that physicians are less scientists than they choose to believe.
Since physicians co-opted the term "doctor" from academics such as (dare I say) economists,
I say we respectfully ignore the point made by "Dan K" and take back the title.
-Drew
Posted by: Drew Battista at Oct 16, 2006 1:57:30 PM
All good comments. I want to echo Stephen Carson though. Like lawyering, doctoring seems like a medieval guild to me, that jealously protects it's turf. For example, I don't think I really believe in the genius GP's powers of diagnosis. I don't know if there are any studies on what triggers diagnosis of major illness, but from what I can tell, diagnosis of illness is pretty straight forward in the vast majority of cases. If I am correct, then we don't need to have full MDs, with their overly-rigorous (almost like a hazing) training as the first responders. We can have slightly better-trained nurses. But the AMA is going to fight that like crazy.
And in response to the posters above who said that the doctor/patient interaction was an important part of the diagnosis experience, most doctors I have ever had were just plain socially inept. Just weird. It's the nurses that have the human contact abilities. I think there is a lot of research that supports this idea. From what I have read, the medical training process selects a certain type of individual, and these are not people people.
Posted by: cw at Oct 16, 2006 2:07:54 PM
One reason for the primitive state of medicine is that patients are in a poor position to evaluate the true quality of the care they received. This sounds stupid, of course; after all, you can tell whether you are sick or well, and you can tell whether a particular ailment has been cured or not. But generally there is no way for an individual patient to know whether a particular procedure was necessary or was performed merely from custom or to pad the bill. Suppose a doctor gives antibiotics for a viral infection, and the infection clears up; we are likely to give the doctor credit. Or a fetal monitor shows an elevated heartrate and a C-section is performed; we tend to believe that the procedure was necessary, and most people will believe the doctor performed a life-saving operation.
Consider that a vast number of alternative practitioners, from aromatherapists to foot reflexologists, are able to stay in business by convincing their patients of the merits of treatments which must by and large be useless. Some of the processes of modern medicine are in the same class, and they have the additional cachet of being part of the approved edifice of Medical Science as Approved By The AMA. Confirmation bias will work to maintain the illusion of effectiveness.
Unfortunately data gathering by itself won't solve all problems, since in many cases only short-term outcomes are considered and subtle long-term downsides would be overlooked. For example, Vitamin K supplementation for newborns has been shown to prevent a number of deaths from VKDB, but aside from some leukemia studies no one has attempted to see whether there are long-term ill effects of this procedure.
Many of these problems could be addressed by datamining a sufficiently large database of complete medical histories. Some large drug companies, such as Pfizer, have terabytes of patient data that could be used. Given the potential for liability in cases where existing procedures were shown to be detrimental, however, I expect that the data will remain private.
Posted by: bbartlog at Oct 16, 2006 2:25:06 PM
My wife's doctor just received an electronic mechanism that automatically and electronically files prescriptions with the local pharmacy.
I'm so happy that we have finally entered the 1980's IT world in medicine!
Posted by: Mr. Econotarian at Oct 16, 2006 4:33:28 PM
VA hospitals have done good work on computerization.
Posted by: joeo at Oct 16, 2006 4:35:23 PM
