Contemporary Issues in Medical Informatics: Good Health IT, Bad Health IT, and Common Examples of Healthcare IT Difficulties
On the ecosystem of HIT

Background On The “Ecosystem” of Commercial Healthcare IT


Clinical IT is a world characterized by serious impediments to success, resulting in issues such as these (thanks to Health IT discussion site EMRUpdate for some of these links):

  • "Half of all current EMRs fail!", from 1/2007 Technology for Doctors (link to PDF)
  • "Avoiding EMR meltdown.  About a third of practices that buy electronic medical records systems stop using them within a year", from 12/2006 AMNews (link)
  • "The failure rates of EMR implementations are also consistently high at close to 50%", from Proceedings of the 11th International Symposium on Health Information Management Research – iSHIMR 2006 (link to PDF)
  • "Industry experts estimate that failure rates of Electronic Medical Record (EMR) implementations range from 50–80%.", from 7/2006 A Commonsense Approach to EMRs (link to PDF)
  • Kaiser Permanante HIT Meltdown (link)
  • Milton Keyne's (UK) Care Records System caused 'near meltdown' (link)


and many other mishaps a similar nature.  The common factor is waste of time, money and resources, and potential harm to patients and patient care. 


It should be considered that results like these regarding a drug or therapy would be the cause of serious questions if not mass lawsuits, if such a drug or therapy were being promoted as essential and miraculous (think VIOXX).   That health IT receives such uncritical support suggests a very complex “ecosystem” with many constituents, with agendas that for various reasons benefit from health IT despite possible downsides.


In reading about HIT difficulties it is important to understand the “ecosystem” of commercial health IT, that is, the identity and nature of the principal constituents and stakeholders, and their interrelationships.  Familiarity with this environment is useful in order to place the social and organizational issues affecting HIT diffusion in the proper context.  I should note that the following applies primarily to healthcare IT in the private and academic sectors.   The federal sector and military have their own health IT ecosystem, with many parallel issues but with distinct issues as well.


I believe the ecosystem of HIT is unique in the annals of information technology.  Just beneath the apparently smooth-running operations of medicine is a complex, poorly bounded, conflicted, highly variable, uncertain, and high-tempo work domain [Nemeth & Cook, JBI 2005;38:262-263].  Never before has such a complex, costly, interdependent and culturally sensitive science and profession come under so much pressure by outsiders to adopt a technology based on major assumptions and perhaps blind faith regarding cost/benefit ratio, advancements in medical practice, and other “silver bullet” factors that are largely unproven and may or may not be true. 


It should be noted that while the difficulty of placing basic medical histories, labs and other data online is not overwhelming, moving towards interoperability, adding the far more complex datasets of the medical subspecialties (for example, invasive cardiology), and adding clinical decision support and other high level cognitive or “artificial intelligence” functions – where many of the touted benefits of HIT are claimed – becomes exponentially more difficult.  This is due to the daunting scientific, conceptual, terminological, statistical, workflow and others complexities of biomedicine. 


It should also be noted that today’s fastest 3 GHz quad-core, 64-bit computer with 8 gigabytes of RAM and a terabyte hard drive, or even today’s fastest teraflop supercomputer, is no more “intelligent” than this author’s 1970’s-era, 8-bit, 64K memory, 2 MHz Intel 8080-based Heathkit H8, or the 1960’s 4K DEC PDP-8/S minicomputer via which this author first gained hands-on computer experience.  It should also be remembered that overzealous diffusion of IT creates unexpected problems; a common example is stress and loss of productivity due to email overload.  HIT is not about technology, it’s about information science, medical wisdom and common sense.


That said, the ecosystem of HIT is more complex than just clinicians, hospitals and HIT vendors.  While not a comprehensive list, the HIT ecosystem is characterized by the following role types and subtypes:


  • Optimists or Idealists.  They believe HIT will "revolutionize" medicine without seeming to exhibit much concern about potential political or societal downsides or potential unwelcome effects on the medical professions and their practitioners, and relentlessly promote only HIT virtues, real and imagined. 


There are at least two subgroups, the “True Believers” or Pollyannas cheerleading for health care information technology, and the Pundits, who make a living out of promotion of HIT and of themselves, such as at the increasing number of HIT-related conferences and seminars in recent years.  In fact the latter may be able to make a decent living entirely based upon giving talks and seminars at regional and national HIT conferences and meetings.  The Pundits may be supported by the Opportunists, below, creating a type of circular conflict of interest.


  • Opportunists.  They come in two subgroups, the Industrialists, who will leverage the enthusiasm generated by the optimists to make money, e.g., HIT vendors and their trade organizations (such as HIMSS), healthcare insurers, and other payers, with little focus on downside issues; and the Ideologues, who will use the enthusiasm to advance their ideological goals such as increasing control over clinicians and/or ushering in nationalized healthcare.


  • Technicians.  These are the IT personnel who design and implement HIT, who probably act as a negative feedback or inhibitory force through not understanding medical culture and the "hiding in plain sight" complexities of healthcare and HIT.


The major pathologies of the Technicians include a “control mentality” and the belief that the methodologies of business computing or management information systems that may be appropriate in, say, accounting and finance (and perhaps not even there considering the failure rates), are appropriate and indeed sacrosanct – sacred and inviolable - for any domain.  They believe such rigid methodologies for IT development, implementation and lifecycle are appropriate even for a domain as complex, poorly bounded and unpredictable as clinical medicine where more agile methodologies are essential (this issue is discussed at length on this site). 


Technicians also believe inappropriately that the leadership of health IT can be generic, i.e., that health IT leaders need no experience in biomedicine.  Finally, Technicians generally are uncomfortable with uncertainty, which in medicine is a given.  “Metrics”, often time consuming to gather, ill conceived and harmful, or meaningless are used as an emotional crutch to “treat the leadership” instead of to improve actual patient care.  Examples are found in several case studies on this site. 


Such beliefs are in fact harmful in fields requiring creativity and flexibility in the approach to computing, such as clinical medicine and scientific discovery, especially when such individuals are in leadership roles with real power over others. 


  • Consultants.  These consist of individuals, small boutique management consultant organizations, and large, powerhouse consulting companies such as McKinsey and Deloitte with divisions involved in health care and HIT.  Their members are drawn from many of the other groups, including pundits, academics, technicians and informatics.   Their actual professional education and experience, especially in clinical and medical informatics domains, varies widely. 


They might be considered a species of opportunist in the HIT Ecosystem.   They are usually quite costly.  A major goal of HIT consultants, of course, is to produce billable hours and repeat engagements.  While the advice of HIT consultants can be quite valuable, it should be recognized that, in effect, consultants have somewhat of a built in conflict of interest towards actually solving problems and making their client organizations self-sufficient.


I have seen half a million dollars spent by one hospital on HIT consulting engagements where the consultants were young, very smart, but relatively inexperienced individuals, and the product was a thick, fancy book filled with “so what’s” and other useless information that simply gathered dust due to its impracticality.


When engaging consultants, an organization should as a due diligence take an inventory of the specific backgrounds of both the consultant managers and the “worker bees”, to determine where they fit in the ecosystem and how this will affect their “given wisdom.” 


  • Medical Informatics.  These come in two subgroups, academic and applied (practical).   Medical informatics personnel ideally have training in both clinical medicine and biomedical information science (informatics).  This cross training provides bona fide informaticists with a unique perspective on the acquisition, synthesis and application of information to problem solving and program development in clinical and biomedical areas (more on the “bona fide” issue below).


Academic informatics.  These personnel include the pioneers who leveraged IT, biomedical information science, and experimental work to make clinical IT possible.  They also used to be the key personnel in evaluation studies.  However, they are much smaller in number than the other stakeholders. 


The investment in formal academic training is considerable and does not offer good employment advantages in HIT, except in academia itself.  This is in contrast to easy to obtain “certifications” such as here and here which do offer industry advantage in hiring and promotion, although the apparent lack of rigor in these pseudo-certifications tend to make them a good fit in the “opportunist” cloud.  Academics are also distracted by the necessity of grant writing and fund seeking from limited sources to make a significant impact on HIT, especially in the areas where it is most needed in the early 21st century, post marketing surveillance of HIT systems. 


While academics do perform some rigorous evaluation work, they do not usually consider observational or anecdotal evidence worthy of much consideration, thus their work is limited in quantity and exposure beyond obscure academic journals.  The HIT industry largely ignores it.  Informatics academics who deviate substantially from faith in the claims made about HIT are even fewer in number, belonging to the “realists” group below.  A metaphor that describes the academic informatics ideology comes from Scott Adams:  “I eat strawberries every day and every time I do, I get hives an hour later.  But since this is not a scientifically controlled experiment and therefore causality is not proven, I will continue to eat strawberries.”


Academic informatics is subject to the pressures of “publish or perish” and the pursuit of tenure.  This causes a focus on “popular” (i.e., to a relatively closed and similar-thinking circle of academic peer reviewers) and/or arcane subjects.  These subject are often of little or no interest to those managing HIT and dealing with the typical human-issue difficulty scenarios.  Articles on HIT dysfunction are not popular.  The American Medical Informatics Association, for example, did not wish to publish a book of case studies, modeled after this website and authored by several members of the Clinical Information Systems Working Group, myself included, on “lessons learned” from HIT difficulties.  The group had to seek out other publishers.


Academic informatics is also subject to the distractions of academic vanity and petty internecine conflicts prevalent in many institutions of higher learning. 


A rather disappointing example of this phenomenon was just recently experienced by this author, where the academic proprietor of a site that aggregates selected academic articles on HIT, the “Informatics Review”, attacked my work through recommending to one of my own former students -- a relationship that was presumably unknown to the Informatics Review proprietor -- that the former student not use me as a professional reference.  This author never worked in any capacity with the person making the negative recommendation.  It is probably not coincidental that this person’s article aggregation site is supported by HIT industry funding.


In effect, academics are largely unable to keep up with the unbridled enthusiasm for mass dissemination of HIT and its social implications, and have become a relatively powerless group in terms of critical thought on HIT.


Applied Informatics.  These personnel include Informal and Formal subtypes, the latter ranging from certificate holders (from legitimate, accredited academic institutions, as opposed to “certification” from opportunist groups) to those with BS or MS degrees, and at the top tier, those with formal postdoctoral informatics education. 


Subgroup members promote themselves as having knowledge and experience beyond traditional business computing or “management information systems” (MIS) personnel.  Ideally, this would include clinical education and experience (such as MD or nursing), as well as formal training in computer and biomedical information science, controlled vocabularies, human computer interaction, medical decision making strategies, and other complex informatics topics. 


In reality, however, the label “medical informatics” or “healthcare informatics” has been misappropriated by a wide variety of stakeholders who “do something” with computers in a biomedical environment of some kind.  The actual education and expertise of these stakeholders in both medicine and informatics ranges from basically none, to the novice (little true informatics skills and experience), to those with certificate, bachelors’ or masters’ level informatics training, to true experts with extensive doctoral and post-doctoral academic training in informatics as well as applied experience in real-world non-academic settings. 


The latter group should perhaps be called “Applied Formal Informaticists” or otherwise differentiated from the “medical insta-maticists” who’ve simply adopted a desirable term to describe their amateur level skills and experience.  The field of informatics is akin to medicine in the early 1900’s, pre-Flexner Report, before rigorous standards were set for medical education and certification.  Caveat emptor applies when utilizing personnel lacking in formal medical and informatics education and experience who use the title “informatics.” 


The article Training the Next Generation of Informaticians”, a 2004 report from the American College of Medical Informatics, is useful in understanding formal medical informatics training, while the essay “On Medical Informatics, MIS, and Leadership of Clinical Computing” is useful in understanding the misuse of the “informatics” label.  Another useful article in understanding the low-skilled subgroup is “Unskilled and unaware of it: how difficulties in recognizing one's own incompetence lead to inflated self-assessments”, Kruger & Dunning, J Pers Soc Psychol. 1999 Dec; 77(6):1121-34), as members of that subgroup, especially those in the traditional IT fields, often overestimate their true abilities in complex biomedical settings.


One major pathology of the HIT ecosystem is its “natural selection” of leaders who lack clinical and informatics experience.  In continuation of a pattern I noted a decade ago (link), I regularly receive solicitations for hospital "Director of Clinical Informatics" positions that require neither clinical nor informatics training or experience, rather traditional business/MIS backgrounds.  I consider such positions possibly harmful, and certainly not helpful, to hospitals and clinicians due to its putting incumbents in such positions outside their core competencies.   I raise the possibility that organizations and journals that provide advice and management consulting to hospital I.S. departments are in part responsible.


In the example here just recently received (July 2008), the qualifications for "the Director of Clinical Informatics" are a "BS with major in IT or provider related field such as accounting/finance, related MBA, IT program level management experience."  I doubt those with formal clinical and informatics credentials would even be considered.


  • Realists (a minority.)  These are people who see HIT as a facilitative tool to clinicians if done right, done well, with consideration to downsides and unexpected consequences, and not overdone and oversold.  Some are informaticists and academics in other fields such as sociology, while others are wise healthcare and IT industry professionals.


Examples are the WorldVistA and Open Source HIT communities, who have inherited what this author considers a “rational practice” approach to development and diffusion of HIT, and a number of commercial vendors usually catering to the smaller private practice physicians. 


WorldVistA was formed to extend and collaboratively improve the VistA electronic health record and health information system for use outside of its original setting. The system was originally developed by the U.S. Department of Veterans Affairs (VA) for use in its veterans hospitals, outpatient clinics, and nursing homes. WorldVistA has a number of development efforts aimed at adding new software modules such as pediatrics, obstetrics, and other functions not used in the veterans' healthcare setting. 


The book “Medical Informatics 20/20: Quality And Electronic Health Records Through Collaboration, Open Solutions, And Innovation” by Goldstein, Groen, Ponkshe and Wine (Amazon link here) illustrates this approach.  I use the book in my own graduate informatics courses, and am cited in it for my views on HIT as expressed in this website.


  • Data Merchants.   Facilitated by information technology, a lucrative industry that collects and sells market research, sales and clinical data to others for profit has arisen.  Customers include the insurers and payers, government, pharmaceutical companies, clinical research organizations that perform drug studies under contract from pharmas, and others.  Examples include Verispan and IMS Health.  Such organizations have an inherent conflict of interest in pushing for widespread electronic health records, regardless of effects on patient care and clinicians, as this technology can enhance their ability to engage in their merchandising of health data and improve their margins.


  • Headhunters.  These people make money by assisting organizations to find HIT leadership candidates.  Some are “retained” (paid to act as exclusive agents in a position search), and others are freelance.  Some are innovative, hard working and unfortunately affected by the fads affecting hiring in most of the IT fields, as in this example from a recruiter commenting on this website: 


What is happening to MDs trying to change careers is providing a window into broader issues about professionals in society today - narrow training, pigeonholing in the marketplace, difficulty making lateral and cross-industry transition, what a handicap it is to be creative, entrepreneurial, or cross-disciplinary in the current marketplace, and the wasted intellectual capital represented by the high caliber of individuals who can't find ways to fruitfully plug themselves into the marketplace.  I continue to be amazed at this general phenomenon...the remarkable quality of a number of candidates I've met, and the lack of recruiters' ability to get them in the door of good companies. The interesting part of the story is that when I am able to get access to high level execs in some of these companies (not just IT, but devices, pharmaceuticals, etc. also) they are dismayed at the quality of those that they hire. They know that something is wrong in how the recruitment process is working (e.g., one of the major device cos. just devoted the time of 1 FTE in Human Resources to 'finding innovative ways of identifying and recruiting good talent into the company'). 


Then there are the “boutique” HIT headhunters.  These sometimes espouse beliefs at odds with fundamentals in healthcare.  From an article in the journal “Healthcare Informatics”:


I don't think a degree gets you anything," says healthcare recruiter Lion Goodman, president of the Goodman Group in San Rafael, California about CIO's and other healthcare MIS staffers. Healthcare MIS recruiter Betsy Hersher of Hersher Associates, Northbrook, Illinois, agreed, stating "There's nothing like the school of hard knocks."   In seeking out CIO talent, recruiter Lion Goodman "doesn't think clinical experience yields [hospital] IT people who have broad enough perspective. Physicians in particular make poor choices for CIOs. They don't think of the business issues at hand because they're consumed with patient care issues," according to Goodman.


These are unhelpful beliefs about education and expertise that most clinicians would reject out of hand.  Even more unhelpfully, many HIT headhunters lack formal backgrounds in either healthcare or IT.   Yet hospitals pay headhunters handsomely for IT talent management, itself a poor use of scarce healthcare resources.  (It is as if the NFL paid me, a medical informaticist who knows little about football and does not watch it, to find the best football athletes.)


The best way in the opinion of this author to “align IT to the business” is to hire IT and executive leaders aligned to the business.   If the business is biomedicine, that does not mean hiring leaders from McDonald’s or International Paper.


  • Policy makers and regulators.  These people need to make sense of the market and the interactions of the stakeholders and actors above. 


An excellent illustration of another of the pathologies of the “HIT Ecosystem” – a self-imposed bias on revealing its “warts” - can be gleaned through a simple information science experiment.  Search engines such as Google serve as a surrogate for popularity or lack thereof of particular subjects.   An example of this regarding the biases of the news media is here.  In medicine, a Google search on “medical malpractice” produces thousands or even tens of thousands of relevant hits.  A search on “healthcare IT failure” or “healthcare computing difficulties” or related concepts, however, produces almost no relevant hits, other than this website and related links.   This is a stunning finding. 


Searching repositories of specialized eJournals (e.g., PubMed) does somewhat better, but most public officials and regulators are probably unaware of such resources and journals.  In effect, the HIT Ecosystem is self-preserving through an apparent self-imposed censorship of negative information.  Policy makers thus largely hear only the positives about HIT.  This phenomenon can be construed as a type of self serving, deceptive business practice in this author’s opinion.


That policy makers and regulators are becoming more familiar with the issue of overoptimism on HIT is evidenced by a new provision in proposed HIT legislation recently released by Energy & Commerce: H.R. 6357, the “Protecting Records, Optimizing Treatment, and Easing Communication through Healthcare Technology Act of 2008", a.k.a. the "PRO(TECH)T Act of 2008."


I note the E&C HIT bill calls for the National Coordinator (ONC) at HHS to prepare a report:


"IMPLEMENTATION REPORT. - The National Coordinator shall prepare a report that identifies lessons learned from major public and private health care systems in their implementations of HIT systems, including information on whether the systems and practices developed by such systems may be applicable to and usable in whole or in part by other health care providers" (Item 5, page 12).


The "lessons learned" provision in the proposed legislation suggests Congress is aware that there are lessons to be learned, which implies they are also aware of difficulties, failures etc. that waste precious healthcare resources and time in the interactions of the components of the HIT ecosystem.


As an aside, it is possible that the numerous web “hits” I noted from domain "" on this website of collected HIT difficulties over the past few months played a role in this proposed language.  I track this site's viewers by IP for research purposes (e.g., see this 2006 AMIA poster "Access Patterns to a Website on Healthcare IT Failure": Abstract [pdf], Poster [ppt]. Evidence for this is the language at this site's introductory page to lessons learned:


"Organizational and human factors issues associated with healthcare IT have led to project difficulties and failures. Detailed case accounts might improve knowledge sharing between healthcare organizations on lessons learned and best implementation practices. Web-based, detailed information on healthcare and other IT project difficulty that can be used as “lessons learnedby others in their own projects is uncommon ...We believe filling the information gap on healthcare IT difficulties is an essential goal to which medical informatics specialists can contribute, and that doing so would be helpful to patients and the healthcare community."


The E&C language bears striking similarities to that wording, which I crafted years ago and which is, like this website itself, uncommon.