New Tech Signals

New tech is usually adopted not for direct productivity gains, but to signal one is in fashion, one is technically capable, etc.  From a Post Oped Tuesday:

President Obama's proposed health-care reforms include investing $50 billion over five years to promote health information technology. Most notably, paper medical records would be replaced with linked electronic records to try to improve quality of care and lower medical costs. The recently enacted stimulus package included $20 billion for health IT. …

Yet while this sort of reform has popular support, there is little evidence that currently available computerized systems will improve care. … Large, randomized controlled studies — the "gold standard" of evidence — in this country and Britain have found that electronic records with computerized decision support did not result in a single improvement in any measure of quality of care for patients with chronic conditions including heart disease and asthma. …

They do little to prompt greater and more widespread health-care practices that are known to be effective. Health IT has not been proven to save money. Moreover, personal financial ties have been found between some researchers and the companies that produce these systems, and as far back as 2005 studies have shown that health IT developers are about three times more likely to report "success" than evaluators who had no part in system development.

What's more, evidence suggests that adoption of some computerized systems has not helped but harmed patients. After the Children's Hospital of Pittsburgh added automated prescribing recommendations to a commercial electronic records system, the institution documented a more than threefold increase in the death rate among child patients. Another leading system contributed to more than 20 different types of medical errors.

I emailed one of the authors for cites, and they check out; this is solid.  Economists are familiar with this scenario:

In 1987 Robert Solow, a Nobel Prize-winning economist, famously said: “You can see the computer age everywhere but in the productivity statistics.” It was only in 2003 that The Economist felt comfortable boldly proclaiming: “The 'productivity paradox' has been solved.”

As Dr Solow observed, most countries saw productivity growth slow in the 1980s and early 1990s, just as computers were becoming widely used. Techies grumbled, economists sharpened their pencils — and businessmen ignored the argument and went on buying the kit. But the conclusion was clear: new technologies on their own do not raise productivity. Rather, companies and individuals must figure out how to make best use of them in order to reap their rewards.

There are important lessons in this for the ongoing debate about broadband.  … The OECD released its latest report on May 19th. It surveys the broadband landscape to December 2007, and tells a warm tale. …. But the excellent report … examines why broadband is actually useful. And here the authors face a problem: there simply is not good data to show that broadband matters. … The most innovative country in the OECD as measured by number of start-up internet companies that rise to global prominence is America, which has a mediocre standing on the OECD broadband rankings. …

Paul David, an economist at Oxford University, has shown that electric power, introduced in the 1880s, did not immediately raise productivity. Not until the late 1920s—when around half of America's industrial machinery were finally powered by electricity—did efficiency finally climb. … Although policymakers and the public might feel that super-fast broadband is essential, that view is based more on faith than fact.

Of course one could argue that adopting tech early for signaling reasons benefits everyone else by helping the tech to evolve.  Could be, but sounds an awful lot like wishful thinking.

GD Star Rating
Tagged as: , , ,
Trackback URL:
  • Bill Harshaw

    As some comments on the op-ed noted, the article conflates electronic medical records and computerized medical recommendations–they should be considered separately. And the key weasel words are “currently available”. We never would have eliminated all those secretaries if we had stopped office automation with the first IBM MT/ST back in 1968. Early adopters can be early learners, getting a start up the learning curve.

  • R, Keene

    I work on a large IT project to automate all aspects of hospital information.

    1.) The end users, e.g. the doctors and nurses, definitely like our system once they get accustomed to using it. Recently I was in the emergency room with one of my kids, and asked the nurses what they thought of the system. They said they love it and can’t live without it.

    2.) the $/line-of-code is awful. I have also worked in start up companies where the $/LOC is very good. The root cause of the inefficiency is first, the size of the project, and second, the overhead of government enforced and outdated project management methodologies.

    3.) In general, tasks that are computer like, such as math, accounting, programming, email, and customer record filing, have huge gains when done on computers. Those were also the earliest uses of computers.

    4.) Tasks that are not naturally computer like, such as digital white boards, ‘mind mapping’, and such are very bad on computers.

  • R, Keene

    Some followup: Overall productivity for a given profession has increased or decreased due to computers according to how applicable computers are to the job. For example, programmer productivity has directly increased to a huge extent as IDEs have gotten better at helping the programmer.

    Some jobs such as sales have increased due to customer management solutions, cell phones, and the internet.

    There have been some productivity losses due to things like internet browsing at work, instant messaging at work, etc.

    There have also been some fiascoes where productivity projects have failed badly, such as the aborted multi-billion dollar US traffic control upgrade project.

    General Rule:
    – Overall business productivity is mostly increased by better business practices, coupled with conservative computer system upgrades.

    Some Anti-Rules:
    – Being leading-edge in business computing is usually a waste of money.
    – If the computer becomes the center of what you are doing, then your customer is no longer the focus.

  • Paul Crowley

    Do you tend to make more posts critical of the Obama administration than you did of the Bush administration because you are more sympathetic to Obama and take care to post against the grain of your sympathies, or because you think your readership are more sympathetic to Obama and so such posts will be more novel, or because this reflects your feelings about these administrations? Or am I mistaken in this perception?

  • Carl Shulman

    My main interest in electronic medical records has been the potential to produce big datasets to study treatment effectiveness, risk factors, etc.

  • frelkins

    Of course where I used to work we had a large division in electronic medical records. Let’s be honest. Many people are killed or seriously harmed by medical errors – these appear largely caused by errors with drugs or not having a patient’s medical history easily available. This has been documented by many, many good studies. The economic cost is huge.

    The idea is that electronic medical records will reduce the errors in prescriptions, thus injury, thus malpractice, etc. Also, by reducing paperwork, and its inherent errors, society can save more costs, and processing thru insurers and the government will be more efficient.

    The problem is that the health care/insurance process is still far from uniform, transcribing records from paper to machine is a burden for small practices and is also error-prone, the government systems suck, and most software is poorly designed for the medical task. But these issues can be tackled and solved with time.

    I can’t judge whether the Obama plan, which is very incremental, is good or not. So I can’t comment on that.

  • Julian Morrison

    The irony is that medical records are where computerization might help a lot – of the sort that absolutely would not be considered. Suppose you made everyone’s medical records public, and let Google index them…

  • spencer

    How do you reconcile these results with the extremely good results the Veterans Administration reports with its use of electronic records.

  • HH

    Could this be a case where wishful thinking actually brings about a positive result?

    I have to note, all the technologies mentioned above are subject to network effects AND economies of scale, which might well explain the delay between the appearance of a technology and its reflection in productivity. I have a hard time imagining a delay for technologies that don’t depend on interaction: I imagine that if someone were to develop a better shoe/shirt/TV/chair/pen, that technology spreads much faster because it doesn’t depend on other people’s actions.

    Phone lines, high speed internet, and medical records become more useful as more people use them. No point in having a phone if no one else has one, nor in having electronic medical records if no other doctor can receive them or send them to you. Could it be just a coordination problem? Perhaps an individual practice would receive no net benefit or might even incur a net cost if it adopted alone, but because it interacts with other providers, it would be a net benefit to all. Perhaps a mandate will simply bring about the necessary critical mass of users faster than it would have happened otherwise.

  • diogenes

    I don’t think electronic medical records (EMRs) will really provide an outcome benefit in the short term. A lot of people have strong incentives for (EMRs) to be adopted though. For health economists, it could provide another mechanism for cost control. For IT people it creates jobs.
    From a medical perspective — its nice to have access to all a patients records — I’m not sure if it will wind up reducing mistakes overall, but it does occasionally speed things up. And of course, having access to all that data provides research oppurtunities.

    I agree computers don’t magically improve productivity — on the other hand, are you really going to argue a typewriter is more efficient than a word processor? Email is not a productive for of communication? The Net isn’t an efficient mechanism of information retrieval? Speadsheets aren’t efficient for accounting? When used for clear cut purposes — the productivity of computers is clear cut. In medicine, I don’t think the EMR has a clear cut benefit — however, I don’t think its delusional to have some wishful thinking and think some useful benefit will emerge (maybe via a network effect).

    I do think the argument for EMRs is a bit of wishful thinking, but its not a terrible idea either, its worth a shot.

    If you saw an archaic and fragmented paper record at multiple health facilities, your first reaction would be this is absurdly inefficient. We should at least see if computers can help. Maybe it will turn out like education where it seemed to be more hype than benefit — but only one way to find out.

  • frelkins

    A counter study? Why would EMRs work in Europe but not the UK and the US? Or why would they work better for dialysis than asthma and heart disease?

    Why wouldn’t there be an obvious benefit in being chipped with your own medical information or tattooed with a database reference in case of emergency? Perhaps the better market for this is in fact for private companies to have me pay to send in my own records and have myself chipped, rather than depend on an overwhelmed solo practitioner or small group practice to deal with it.

  • TGGP

    This blog began in late 2006, so a better comparison might be what Robin writes about Obama starting in 2014 (assuming he wins the next presidential election).

  • TGGP

    Casey Mulligan has a recent NYTimes post where he says studies show that highly specialized doctors “make better diagnoses and fewer mistakes”. That would have been precisely the sort of factor I would have assumed was more about status-affiliation than effectiveness, rather than technology. My own company primarily serves businesses just now getting into the technology of the 80s and ditching pen & paper for computer systems. I’m told we drastically increase their productivity and reduce their costs, but I’d expect my employers to say that regardless.

  • Grant

    Aren’t outcomes going to be extremely varied? Why not simply let some practice and hospitals switch when they want to (i.e., do nothing)? It seems to me that some doctors, nurses and aids have the computer skills necissary to take good advantage of new technology, while it may only befuddle others. I know some doctors who have to have every latest gadget, and others who can scarcely use a computer.

    Julian, couldn’t people make their medical records public (or semi-public) if they wanted to? If the benefits were there I could see some people using a system like that.

  • davidc

    Is there much evidence about the use of technology in education?
    What i can find seems to point to it having negative consequences

  • MZ

    Carl Shulman said: My main interest in electronic medical records has been the potential to produce big datasets to study treatment effectiveness, risk factors, etc.

    It’s funny that you mention that. I published a paper in 2000 on how risk factors affect the longterm outcomes of endarterectomy procedures. Well, I didn’t really publish it, I got my name on it, but I didn’t most of the grunt work. I was a undergraduate at the time and a group of doctors needed someone to cull through their patient records and collect data on what turned out to be 41 “clinical and angiographic” factors (things like lifestyle, patient history, family history, etc). In all, 254 records were included, although I had to go through at least twice as many more, so maybe >700 to get that data. I spent ~15 hrs a week for 4 months in the medical records room, and half that time was spent pilfering through the stacks of records, trying to find the right files. I probably could have cut the workload in half if the records were computerized.

  • Richard Pointer

    Dr. Hanson,

    Would Self-digitizing my medical records help me? Having some sort of implanted, readable chip be of use?

  • Unsupervised

    It’s so easy to say we need a nationwide electronic medical record system. Sounds so easy and quick. And cheap, too, I’m sure.
    And whose version of the Electronic Medical Record (EMR) will you be using?
    The VA’s? City hospital of wherever’s? University-of-whatchamacallit’s? What’s to ensure my EMR from my ER visit in Maine when I was on vacation will match with my EMR at my local hospital or doctor’s office? Who will decide on the encryption for transmission via Internet circuits? The doctor’s office uses a MAC-based EMR. The university a locally-developed LINUX one. My hospital uses a Windows-based system.
    I doubt if there’s a common database structure.
    How about the Military’s AHLTA system of outpatient electronic medical records? As of this writing, AHLTA isn’t compatible with MAC, LINUX or VISTA, only XP Service Pack 2 (not even Service Pack 3). And you can’t use any Internet Explorer version newer than 6 (not IE7, not IE8, and I’m not even going towards the non-Microsoft ones). It’s a dog. (but the government developed it, so it’s gotta be good, right?)

  • Adam Nelson

    On the technology/productivity angle, I always thought it was interesting that the most efficient firm I’ve ever seen was Wal-Mart and their in store inventory system was a single terminal that still had a green screen all the way up to 2001, but unlike any other store, everyone knew our could quickly find out how our sales were doing compared with the prior week/month/year last year for each product in their department. I think their overall IT budget was about 2% of 10 year average profits, but one study attributed a huge % of the nation’s productivity to their gains alone.

  • Dan

    There are two stages to the adoption to any new technology: the assisting phase and then the integrating phase.

    In the first phase, the new technology is used to improve existing designs and processes. For example, the Romans used iron instead of bronze for swords and spearheads, but the basic conduct of war remained the same. IN the digital age, early on improvements were made because of things like not having to send a clerk down to the file room to get records, they could be pulled up nearly instantaneously on a terminal.

    IN the second phase, new processes and designs are created that could not have been achieved using the old technology. For instance, iron swords became longer, narrower, and for their size lighter, making them more easily used on horseback, and thus the medieval knight was born. IN the example of electricity, it wasn’t until new tools were developed that took advantage of the ability to place the power source within the tool itself (replacing the system of belts and pulleys powered by a steam engine or water wheel) that the productivity gains were achieved.

    With computers and information now medicine is just beginning to move into the second phase. In order to truly achieve the gains from EMRs and other new technologies, the very process of medicine is going to have to change. What form that will take, I don’t know, and anybody who says they do is either a fool or about to get very rich (possibly both).

  • Chase Johnson

    Unsupervised: The failure modes you discuss have an easy workaround: print out the records. Any reasonably well-designed electronic system can easily fall back on non-electronic methods. At that point, you have no advantage from the electronic system, but in time it would be limited to special cases. As the sophistication of EMR systems grows, compatibility between different hospitals will become less of a problem. Individual EMR products will gain increased ability to translate between different document formats. Check out a piece of modern word processing software. The format reading and writing options are staggeringly numerous, because compatibility is an important component of the word processor’s job.