Sunday, December 07, 2008

Open Letter to President Barack Obama on Healthcare Information Technology: The "We're Right" Wing vs. The "Left Out" Wing

Mr. President, it is obvious that healthcare requires change. Change is needed in order to assure equal access to all, to end disparities in quality and service, to create sanity in costs, and to accomplish other reforms that will promote healthcare “justice” for all.

You have just called for a significant expansion of use of healthcare information technology (HIT) by hospitals and physicians as part of your economic rescue plan. I strongly applaud such an incentive.

As in other industries using information technology, HIT such as electronic medical records, physician order entry, and clinical decision support systems will play a leading role in healthcare transformation. Without such technology, the healthcare system is far too complex to be managed well.

And managed well it must become. However, there is a major caveat regarding such a national initiative.

Mr. Obama, I offer the following suggestions from the perspective of my medical specialty area, Biomedical Informatics, the formal study of computer applications in biomedicine [1]:

HIT is an experiment. It is, as yet, unproven on a large scale. There have been many warning signs that it is an experiment that could go awry. HIT can be used to assist in healthcare transformation, but not if the HIT itself is defective or mismanaged.

Unfortunately, the problems with HIT are indeed considerable.

Use of HIT after 30-plus years of effort and billions of dollars spent remains limited. As per the 2008 statistics in the New England Journal of Medicine, just four percent of physicians in the U.S. reported having an extensive, fully functional electronic-records system, and just thirteen percent reported having a basic system. Most hospitals are also lacking the technology to any meaningful extent [2].

Clinicians (physicians, nurses and others) are struggling to use awkwardly designed HIT, designed as if for quiet, solitary business offices yet costing millions of dollars per hospital. Hospital boards, executives and department heads are often frustrated by the pushback and resentment they get from their clinicians regarding such HIT.

Too few in the HIT industry, however, ask why this situation exists and what might be done to eradicate these problems.

The healthcare industry and the HIT sector have been reliably tone deaf on these issues, which results in the very low diffusion of HIT. Platitudes, excuses, and blame placed solely on endusers (i.e., the clinicians) are the norm.

Why is this so?

Mr. Obama, I believe that there are two major ideological camps in what I call the “HIT Ecosystem” [3].

On one side are the “HIT industrialists" - the "We're Right" Wing.

The HIT industrialists, including developers, vendors, consultants, and IT personnel in hospitals view HIT simply as another variety of business computing technology, such as finance and accounting systems meant for use in calm, solitary office environments.

They cling to their “religion” of management information systems approaches and methods for design, implementation and evolution of business IT, believing these approaches as valid for the back office as for the pediatric clinic, emergency department or intensive care unit.

Nothing could be further from the truth. This is a leap of logic of absurdist proportions that would be comical if its implications were not so serious.

It is as if the industrialists believe one should be able to travel to the moon in a hot air balloon. The moon is "up" and balloons go "up", therefore, why not? All that's required are the right "processes" -- with which the Acme Hot Air Balloon Co. executives can accomplish anything -- and ignoring those pessimistic scientists, engineers and other experts who speak of vacuum of space and radiation and all those esoteric "gotchas" that are bad for business! (See my Powerpoint presentation to the IEEE Medical Technology Policy Committee on this issue at this link.)

To the Moon in a Hot Air Balloon!

The industrialists forget -- or never knew -- that a thousand generic workers following the finest of process will always be outperformed by a few people who actually know what they're doing.

The HIT industrialists usually refuse to accept more progressive, “activist” approaches to these issues that recognize that clinical computing and business computing are not different sides of the same coin, but are in fact distinct subspecialty areas of computing requiring distinct and specialized approaches.

The industrialists would also have us believe that the HIT “economy” is fundamentally sound and that we should proceed with business as usual.

I shall refer to the industrialists as the “We’re Right” Wing of the HIT Ecosystem.

I use this title based on their often reactionary and procrustean views towards HIT. The adjective “procrustean”, derived from certain practices of Procrustes, a figure from Greek mythology, is defined as “designed to produce strict conformity by ruthless or arbitrary means” [4]. These views, that the “old” ways of designing and implementing IT are the “right” and in fact the only way to proceed, are strongly held. I believe they are, however, being made obsolete in an age of unparalleled IT innovation.

Largely as a result of the “We’re Right” wing’s views, costly, morale damaging (especially to clinicians) HIT difficulties and failures are commonplace, and the benefits of which this technology is capable are not being realized.

Clinicians must toil with ill conceived or badly designed IT in their already harried ten to fifteen minute patient visits, and the industry is largely deaf to their misery. It’s bad for the IT business.

On the other hand, knowledgeable clinician-informaticists in the aforementioned field of Biomedical Informatics, true computer scientists (as distinguished from management information systems or MIS personnel), sociologists studying IT, and other related specialists, some working for more than thirty years, have found better ways to develop and manage HIT than has the traditional IT industry, which itself has a dismal record of project cost overruns, failures, and system abandonment in fields far less complex than clinical medicine.

Sadly, the specialists are often pushed to the margins of the HIT Ecosystem. I will call this “we actually know what we’re doing” group of specialists the “Left Out” Wing.

It is particularly ironic that the Biomedical Informatics group is marginalized, in that it was that group of pioneers who experimented in the early days of computing and essentially invented the foundational elements for today's health IT. Even then, the pioneers noted issues in HIT that are unchanged many decades later. For example, as I note at my academic website on HIT difficulties:

As far back as 1969, EMR and Medical Informatics pioneer Donald A. B. Lindberg, M.D., now Director of the U.S. National Library of Medicine at NIH, made the following observation.

He wrote that "computer engineering experts per se have virtually no idea of the real problems of medical or even hospital practice, and furthermore have consistently underestimated the complexity of the problems…in no cases can [building appropriate clinical information systems] be done, simply because they have not been defined with the physician as the continuing major contributor and user of the information" (Lindberg DAB: Computer Failures and Successes, Southern Medical Bulletin 1969;57:18-21).

With regard to Biomedical Informatics specifically, the U.S. National Institutes of Health (NIH) has spent hundreds of millions of dollars over the past twenty years to expand Biomedical Informatics through funding of rigorous training at many of this nation’s major universities, as have many other educational organizations on their own funds.

Yet, the lessons of Biomedical Informatics and the other specialist areas have yet to diffuse into the HIT industry and effect needed “change” in that sector. When informatics experts bring their activism into the equation, they are often not taken seriously or placed into “advisory roles” (that is, away from true leadership roles) in most healthcare organizations.

Experts in the Left Out wing offer sage advice. “It is unwise spending millions on Electronic Medical Records without investing thousands in Biomedical Informatics expertise”, these specialized personnel say [5].

Yet there are many “unwise” organizations within healthcare. Examples abound of healthcare organizations not just passively but actively and almost absurdly rejecting the presence or counsel of the specialists as if the organizations are run by intellectually challenged leadership (does this sound familiar regarding other industries in the news recently? See example here), and I believe there is great value in studying the results of this lack of insight [6]. However, others do not.

The following is meant not to be humorous but to reflect a dangerous reality:

The Industrialist "We're Right" Wing's Simplistic and Overoptimistic View of Healthcare Environments

The Informaticist "Left Out" Wing's Reality-Based View of Healthcare Environments

If you think the above is exaggerated, Mr. Obama, read my account "Serious Clinical Computing Problems in the Worst of Places: An ICU" at this link regarding an actual situation some years ago in an ICU at the largest healthcare provider in Mr. Biden's home state, and the cavalier attitudes taken by its executive administration over endangerment of ICU patients via computer.

Imagine for a moment the horror of being unable to intervene due to administration's priorities of "everyone getting along" rather than the absolute protection of patients. Yet this is not an uncommon scenario in the IT backwater of hospitals.

This state of affairs has significant real world implications.

Even now I hear of ill conceived and poorly implemented ICU systems that endanger premature infants, EMR's that interfere with adult patient care and clinician mental focus in offices and clinics, order entry systems that encourage rather than prevent medication errors, and systems that impair hospital operations to the point of putting seriouly ill patients at risk of being misdiagnosed and intermittently losing patient information.

First and foremost among the “We’re Right” wing is an aversion to detailed investigation of IT failure, as should be performed, NTSB style. However, doing so might be bad for business. In the industrialist culture of IT, it’s bad to air the dirty laundry.

As a result, our government has been seduced by the promise, the potential, the Siren Song if you will of HIT, and shielded from information on its true challenges, difficulties, downsides and failures. An "irrational exuberance", a Syndrome of Inappropriate and Uninformed Overconfidence in Computers prevails in healthcare (my webcast on irrational exuberance in HIT can be heard at online journal "Government Health IT" here).

As a result of this seduction and censorship of the unpleasant realities, seasoned legislators such as Sen. Max Baucus are led to make claims such as:

“Everybody talks about health IT. We all know we need it. It hasn't happened. Why? Partly because we're America. We're not a single-payer system like the UK, which can say, you hospitals, you have to put this in because we're paying your bills” [7].

This implies that Sen. Baucus believes the UK should have an easy time of it in their multi-billion dollar national HIT program, NHS Connecting for Health.

Yet this has proven to be profoundly untrue:

The future of the NHS's £12.7bn computer programme was in doubt last night after its managers acknowledged further delays in introducing a system for the electronic storage and transmission of patients' records.

Connecting for Health, the NHS agency responsible for the world's biggest civil IT project, said it was no longer possible to give a date when hospitals in England will start using the sophisticated software that is required to keep track of patients' medical files.

Christine Connelly, the Department of Health's recently appointed head of informatics, is understood to be reviewing whether the programme is a cost-effective way of improving the quality and safety of patient care [8].

I should add that the “appointed head of informatics” has no apparent formal Biomedical Informatics background, but had been CIO of candy/beverage company Cadbury Schweppes, a puzzling background to lead a national HIT initiative.

If health IT is easier in the UK due to their single payer system, then God help us here in the United States.

Lessons from HIT difficulties and failures are rarely spoken of and often hidden. Detailed investigations down to fine grained causes that are routinely performed in other areas of technology failure – airlines, trains, etc. – are not done when health IT fails. This is due to the industrialist biases.

Yet, as per the European Federation for Medical Informatics (EFMI) Working Group for Assessment of Health Information Systems, you should be aware that bad HIT can kill [9].

Further, the repeated advice of the “We’re Right” wing of the HIT Ecosystem is that the cause of failure, besides those darn clinicians, is neglecting to adhere to obvious and simplistic “best practices” typical for business IT. These practices largely involve common sense, such as involving end users at all stages of development, focusing on the value added rather than the technology, and other “sterile technique is needed in surgery” advice.

Unfortunately, even in non-healthcare IT, technology projects have a bad reputation for going over budget and schedule, not realizing expectations, for providing poor return on investment, or in failing outright and being de-installed. Surveys and reports on the acceptability of new IT systems seem to highlight constantly the same problems and probable causes of failure yet businesses, large and small, continue to make mistakes when attempting to improve information systems and often invest in inappropriate or unworkable changes without proper consideration of the likely risks [10].

In HIT, these problems are perhaps worse, with the added complexity of the unpredictable, poorly bounded environment of healthcare.

Enter Biomedical Informatics, computer science, social science and the other true experts. Advice often heard from this “Left Out” wing, but largely dismissed by the industrialists, include:

  • Simplistic “best practices” of business IT are often inadequate for the complexities of IT in clinical settings, and even these are often not followed in healthcare IT.
  • CIO’s are not unaware of, nor do they go out of their way to pass up, best practices. The reasons that commonsense best practices, e.g., strong clinician involvement and empowerment in health IT projects are not universally employed is a sociological issue, and requires investigation.
  • The most important "best practice" of all is often not followed: the practice of knowing what you do not know, identifying those who do know, becoming familiar with their work, and utilizing them and their knowledge maximally. Most hospital CIO's are unaware or have a very narrow and faulty view of Biomedical Informatics. Considering the impact of the CIO role in healthcare, and the ease of obtaining such information via the web, this amounts to fundamental negligence.
  • There are highly specialized and nuanced “best practices” arising from decades of Biomedical Informatics, IT-sociology and other research of which traditional IT personnel are generally unaware, and therefore do not use. This is harmful to the healthcare enterprise.
  • The reasons for “worst practices” seen of many stakeholders – such as active and passive aggressive behaviors that occurs for many reasons and negate any “best practices” - need to also become deeply understood, not treated in a cavalier manner or ignored. These behaviors reflect hidden, critical human issues of which nobody speaks, and it is these unspoken issues that can cause HIT to fail.
  • Continued inattention to HIT ease of use to maximize vendor profits, and issues such as pay for performance formulas based on electronic medical records data, but prepared in a typical corporatized manner unfair to clinicians and harmful to medicine’s core values, will be recognized as such by clinicians. This will lead to increased active and passive aggression against HIT that will cause HIT (even otherwise well designed HIT) to fail.

There are many other points of wisdom contained in a huge body of literature that, in fact, could have helped avoid the major problems within the UK’s national HIT initiative, and that can prevent the same from happening here.

Mr. President, I implore you and your advisers and staff to become personally familiar with these issues. Seek out the advice of Biomedical Informatics experts, computer scientists, those who study social issues in computing, and others. Have your advisers read books such as “Medical Informatics 20/20: Quality and Electronic Health Records through Collaboration, Open Solutions, and Innovation” by the VHA pioneers [11], “Managing Technological Change: Organizational Aspects of Health Informatics” by informatics researchers Lorenzi and Riley [12], and “Understanding And Communicating Social Informatics” by IT social issues innovators Kling, Rosenbaum and Sawyer [13].

Push as strongly for HIT reform as for healthcare reform itself, lest our HIT initiatives suffer the same delays – and the same costly failures – as the UK’s national electronic medical records program.

If this is not done, Mr. President, billions of precious healthcare dollars that might be spent on “IT misadventure” in a time of unprecedented national financial challenges and hardships might simply be better spent on delivery of needed medical services, health insurance and other "safety net" interventions.

HIT in 2009 and beyond, as healthcare itself, is at a crossroads. If we do HIT right, we get rewarded. If we do HIT wrong, then the results shall be deserved.

My takeaway lesson is simply stated: healthcare’s defects cannot be effectively changed or reformed via healthcare IT, if that healthcare IT itself is defective.

Both require change.


Scot Silverstein, M.D.


[1] “Training the Next Generation of Informaticians - A Report from the American College of Medical Informatics.” J Am Med Inform Assoc. 2004 May–Jun; 11(3):167–172. PDF at this link.

[2] “Electronic Health Records in Ambulatory Care - A National Survey of Physicians.” NEJM 359:50-60.

[3] “Introduction to the Complex Ecosystem of Healthcare IT.” Essay at website “Sociotechnologic Issues in Clinical Computing: Common Examples of Healthcare IT Difficulties.” Scot Silverstein MD, Drexel University, Philadelphia, PA, (accessed Nov. 26, 2008).

[4] The American Heritage® Dictionary of the English Language: Fourth Edition, 2000.

[5] Paraphrasing William Hersh, M.D., Professor and Chair, Department of Medical Informatics & Clinical Epidemiology, Oregon Health & Science University,

[6] “Medical Informatics, Information Technology Leadership, and Clinical IT Success”, Scot Silverstein MD, Drexel University, Philadelphia, PA, (accessed Nov. 26, 2008).

[7] “Health” supplement, Wall Street Journal, Nov. 24, 2008, p. R10

[8] “Bank bailout puts £12.7bn NHS computer project in jeopardy”, John Carvel, The Guardian, Oct. 29, 2008, (accessed Nov. 26, 2008).

[9] “Bad Health Informatics Can Kill.” European Federation for Medical Informatics (EFMI) Working Group for Assessment of Health Information Systems, (accessed Nov. 26, 2008).

[10] “Developing a risk estimation model from IT project failure research.” M Bronte-Stewart, Computing and Information Systems, University of Paisley, 2005, 9(3): 8-31. (accessed Nov. 26, 2008).

[11] Medical Informatics 20/20: Quality and Electronic Health Records through Collaboration, Open Solutions, and Innovation. First edition. Goldstein, Groen, Ponkshe & Wine, Jones & Bartlett Publishers, 2007, ISBN-13: 9780763739256

[12] Managing Technological Change: Organizational Aspects of Health Informatics. Second edition. Lorenzi & Riley, Springer-Verlag Publishers, 2004, ISBN: 0387985484

[13] Understanding And Communicating Social Informatics: A Framework For Studying And Teaching The Human Contexts Of Information And Communication Technologies. Information Today Publishers, 2005, ISBN 1573872288.


maggiemahar said...

Everything you say here sounds very logical. Clearly, there is a huge difference between using IT in an office and using IT in an ER.

Healthcare is not like other industires is so very many ways . . . and, clearly, clinicians should be at the center of designing it.

I know only a little about Health IT so please excuse a few naive questions:

I am under the impression that the Mayo Clnic's Health IT system works very well (it even includes voice recognition, so docotors don't have to type)within Mayo.

And the VA's system, VISTA, works well--within the VA.

Is there anything to learn from these systems that could be applied to nationwide interoperable Health IT?

On HealthBEat, we have written about using VisTA as a starting point. Is this simply a naive idea?

Finally, from talking to people in Sweden, I'm under the impression that Health IT works well there. Is this mistaken? IT is, of course,a much, much smaller country. . . but can we learn anything from Sweden or other European countires?

Thanks- Maggie Mahar (

MedInformaticsMD said...

clearly, clinicians should be at the center of designing it.

Yet many if not most IT personnel would not see this as clear at all.

Certainly, the issue of clinician leadership of health informatics initiatives is undecided. The chief of staff of the hospital where I trained recently opined that it is "not a sine qua non" that a physician should lead clinical informatics projects. I was taken aback by this "learned helplessness" opinion, which in essence states that intrusions into clinical medicine by non-clinicians is simply OK. I cannot imagine any other profession that would take such a milquetoast approach to increasingly critical tools of its trade.

I know only a little about Health IT so please excuse a few naive questions:

Naive is not the correct word. I'd say "pre-informed questions."

I am under the impression that the Mayo Clnic's Health IT system works very well

It's hard to know what "very well" means in the context of IT. What are the metrics? What is an acceptable capital and operating cost per annum per bed? What is an acceptable IT-induced error or operations difficulty rate?

That said, the literature points to usefulness and widespread availability of clinical IT at the Mayo. Several factors contribute, but other questions must be asked.

First, Mayo has been at the forefront of biomedical informatics research for a long time, with a well developed presence in the domain. They have worked on clinical IT for many years, and the investment has been considerable. The creation of Mayo's EMR began in 1993 (link). Much of their systems, I believe, are proprietary and/or not easily ported to other organizations with different cultures, structures, etc. This is not atypical for organizations advanced in HIT usage.

However, there are also several questions raised. The development process was likely rich with "lessons learned" in how not to do things. In fact I recall some early stories in the 1990's about this at conferences.

Yet, try to find information on Mayo setbacks, failures, false starts, revised plans, costs overruns, etc. and you will likely have a hard time.

As at my website on HIT difficulty and on HC Renewal as well, I opine that there is far too much hype about HIT "success" (whatever that means) and far too little on lessons learned from other organizations. There is far too little organizational information sharing on "worst practices", let alone best practice -- which design metodologies, business IT practices, management fads, etc. simply don't work in HIT, which vendors and vendor products are unreliable or untrustworthy, etc.

The takeaway is that HIT is a cottage industry. It evolved as a facilitative tool to HC professionals. It seems somewhat in parallel with the dot-com mania, it was transformed from facilitative tool to godlike savior of mankind.

The syndrome of inappropriate overconfidence in computing, an "irrational exuberance" over HIT now prevails. Listen to my Government Health IT webcast on that issue at this link.

I fear our government has been spoon fed a steady diet of irrational exuberance on this issue and will proceed naively - the appropriate term - down a path that might lead directly to HIT hell.

We are attempting to go from cottage industry to mature one in one giant leap. I'm afraid Neil Armstrong's "giant leap" for mankind was an easier task.

the VA's system, VISTA, works well--within the VA

The VA's clinical IT works well within that environment, indeed. Again, what they have now took decades. It is not without problems. There is a huge controversy right now about bringing Cerner in to update the ultra-complex lab modules. The question is, can the VA accomplishments be ported to other environments?

Note that one of the books I recommend in the HC Renewal posting is by VA experts - Medical Informatics 20/20. The book came out last year as an attempt to share the wisdom the VA has accumulated over decades on HIT. I do not believe it has been read by most of the nation's hospital executive teams, however. I am using it in teaching graduate courses in health informatics. That is my contribution to sharing this material.

Another controversy is over the planned merge of ViSTA with the active military EHR system AHLTA. That merger alone will likely take many years and require hundreds of millions of dollars or more to accomplish.

On HealthBEat, we have written about using VisTA as a starting point. Is this simply a naive idea?

No. It is being done, in Linux-like fashion by entrepreneurs. See the websites of the WorldVista project, for example. However, understand the dataset and applications themselves require massive rework, customizations, and iterative refinements as they become adapted to different medical specialties, environments and countries. It's certainly easier than moving from scratch, but not "easy."

Finally, from talking to people in Sweden, I'm under the impression that Health IT works well there.

Yes, a small country with an HIT system that may not be easily duplicated here. If you've read my posts on the UK's problems, you'd wonder why the UK doesn't just do what Sweden did. The answer is one of: they could not, or would not. Lack of comprehensive understanding of how debacles occur exists at present. It is what you don't know that will hurt you, where complex technology and social systems (e.g., medicine) are concerned.

Also, your question raises a definitional issue. What does "works well" mean? What are the metrics? What actual nitty-gritty capabilities are in place? Basic capabilities such as notetaking? Intermediate, e.g., context sensitive data entry and well controlled terminology sets that are easy to apply in a consistent fashion? Advanced? (e.g., robust clinical decision support).

It's hard to know a comprehensive answer. It's worse in the U.S., but current diffusion rates here after 30+ years of effort do not look encouraging.

My point is that before we assume that a cottage industry can be turned into a successful megalith overnight, we should do our collective homework.

While HIT problems may be good for the IT and management consulting businesses, they are not good for the healthcare business, already struggling under great financial duress.

blog said...

I'm very sympathetic to your critique of the "industrialists". The real world is uncertain, and the scientific method requires us to always suspect our own beliefs and evidence-supported assertions. Anyone who treats empirical matters as religion is bound to falter.

In that same spirit, I would ask you to re-evaluate the allusions you appeared to make early in the letter to government-run healthcare. Having lived in other countries, and done some consulting for the UK's NHS, I'm terrified of their low quality, bureaucrat-rationed care. Did you know that after 2006, the UK's NHS has allowed choice, that the british are leaving government care by the thousands ?

MedInformaticsMD said...

blog wrote:

In that same spirit, I would ask you to re-evaluate the allusions you appeared to make early in the letter to government-run healthcare.

Thanks for the comment. Actually, I'm close to centrist (see "Political Map of Pharmaceutical Bloggers" here) and I make no systemic suggestions for a healthcare model. I only make suggestions for how healthcare IT can be made a productive part of whatever model is selected, by powers much above my pay grade.

I can only try to keep a tidy house in my own neighborhood, which is HIT!

sydney said...

That's a very good analysis. Did you know you can post this sort of thing at Obama's website Click on the "tell us your story" button and submit it. Maybe they'll listen. Maybe.

Larry said...

We need IT advancements in medicine if only to help streamline EMRs within the medicinal community. The hitech act ensures the privacy and security of those involved in the process, a major concern of those who were on the fence about Eperscribing. DrFirst is an excellent example of a company using technology to streamline medicine. Check out their website at

Scot M Silverstein MD said...

We need IT advancements in medicine if only to help streamline EMRs within the medicinal community.

While I agree we need IT advancements in medicine (most notably centering on human factors improvement, computational linguistics, rational thinking about what really needs to be digitized, and other mission-friendly adaptations, not on technology), that is a weak excuse indeed for spending tens of billions of precious healthcare dollars on experimental technology in the next decade.