Dear Mr. President, a Data Model for my Electronic Health Records Nearly Killed Me

Blogger: Joe Bugajski

Mr. President, your historic economic stimulus package (The American Recovery and Reinvestment Act of 2009), appropriated $19 billion for health information technology ("Technology Gets a Piece of Stimulus", New York Times, January 25, 2009). This week, your Director of the Office of Management and Budget (OMB), Peter Orszag, shockingly held that half of the US operating deficit can disappear with lower healthcare costs and these will obtain through electronic healthcare records (Daily Show, 6 April 2009). Today, the Wall Street Journal wrote that you proudly proclaimed that electronic healthcare records for the members of the US military, like my youngest son, and continuing through Veterans Affairs "will provide a 'seamless system' to facilitate information sharing and cut red tape, ending the need for veterans to transfer military records to receive benefits". Whereas Star Wars and Star Gate movie fantasies provide great fun, witnessing you, a world leader, spew delusional visions of a nation-covering, interoperable, secure, private, reliable, accurate, and instantaneous electronic healthcare data network is at best terrifying and at worst pernicious.

Two months ago and for 100 hours I battled for my life with a networked, state-of-the-art, secure, electronic healthcare record system. It connected two of arguably the most advanced medical facilities in the world. One was a modern clinic built by generous benefactors from the computer industry. The other was a world renown hospital at a top ten university. Both facilities bristled with brain-power and hi-tech gadgets. All records were "computerized" - that was what was so very wrong with the care that I did not receive, albeit competent and attentive nursing staff managed my care as they shuffled me around the emergency room (ER) and intensive care unit (ICU). (For more about what went wrong and how it went wrong, read my personal blog.)

The reason things went so wrong during my clinic, then hospital, stay, and the reason, Mr. President, your grand vision of a universal IT health data network is so screwy, is that healthcare data cannot be reliably modeled. An unreliable data model for health records; like those at the two world-class, completely electronic facilities, that "cared for me"; accelerate prescription errors, prevent staff from efficiently delivering services, heighten life-threatening risks, and dramatically increase costs. Data models are the technical instructions for software to make "computerized health records" possible. Indeed, data models make all computer records possible. According to my friend, colleague, and data modeling guru, Joe Maguire, "data modeling is a family of techniques used to describe the kinds of information that are important to an enterprise". Healthcare data is clearly important but that fact alone does make for reliable data models. Good models require stable data and good data modelers. Unfortunately, healthcare data is unstable. Sadly, good data modelers are scarce.

The first problem with modeling healthcare data is that models must represent certain concepts (and not others) that will remain stable and true long enough to be built into computer software then used by healthcare providers and patients. Mr. Obama, have you noticed just how much knowledge has, is, and will be accumulating in the medical sciences? Knowledge is codified using words - medical knowledge uses copious quantities of difficult words taken from several languages. Words that recur frequently in a particular context become imbued with a meaning that includes the context (e.g., the White House). Such words then come to symbolize a bigger idea than originally intended (i.e., a "house" that happens to be "white", versus your administration and not the house). In medicine, how many stable words exist? These words - nay, well-formed concepts, repeatable, agreed by the medical community - can be modeled and added to computers to store records. Go one step further. Specialization in ER, ICU, cardiac care, pulmonology, oncology, radiology, and other medical subjects exists because the cumulative knowledge defines a large ontology. The ontology, taxonomy, skills, and knowledge of an medical subject area then can be referenced with one word - the name of the specialty. Unfortunately, words that refer to a concept in one specialty often mean something different in another specialty.

The second problem is the lack of good modelers. These people, specialists in data engineering, a subfield of software engineering, transform concepts into graphical and lexical patterns that are used to create computer records. The concepts they model are words (nouns and verbs) plus concepts used by practitioners to describe a patient's medical condition, or a critical care pathway, medications, instructions to patients and nursing staff, tests, and diagnoses. Who among us has the modeling skills to encode this data? As information varies across specialties, how should it be encoded? Empirical evidence suggests that engineers who built the electronic health records network at the two facilities that "cared for me" tried to do this, but they failed. Their data model had irreconcilable silos of information spread across specialties and expressed as incomplete taxonomies (entities), inadequate ontologies (attributes), and poor associativity (relationships). Hence, when programmers added those bad data models to the health information systems, those systems later lost critical information about patients' condition, listed wrong medications,isolated prior diagnoses from current observations, in short, made very bad medicine.

Please do not misunderstand, Mr. President, the medical personnel at the clinic and the hospital were professional, competent, and knowledgeable. It is just that when they interacted with me and other patients, then translated that interaction into electronic health information systems, there was always a fight. Indeed, healthcare professionals wasted between 40% and 60% of the time they had allotted to patient care with making electronic health records work very poorly.

Since the time of my illness, I met and spoke with a dozen medical professionals and healthcare IT experts. They unanimously confirmed my sickbed analysis of the faults with electronic health records. Most longed for handwritten charts hanging at the foot of every patient's bed (see, Professor Dr. Armstrong-Coben’s New York Times Op-Ed ) - now, so do I.

Mr. President, before your administration pours billions of our grandchildren's yet to be hard earned dollars into the biggest, scariest, and most wasteful boondoggle of an IT project the world has ever seen, please instruct your health IT experts to carefully analyze the strengths, weaknesses, opportunities, and threats (SWOT) associated with building a national heath information network using today's technology. Tell them to take the simplest steps first. Make them prove results in small projects. Insist that your experts read my paper, "Data Integration: Fantasies and Facts". It explains how to start and manage a large scale data integration project.

If our nation simply accepts your vision while healthcare IT vendors collect a lion's share of the $20 billion stimulus bounty, individuals and businesses will pay higher medical costs, patients will receive inferior care, medical professionals will loose more of their precious time fighting IT systems instead of delivering better care, and you will be a one term president.

With profound admiration and respect,

Joe Bugajski

Comments

[Loraine Lawson]

Excellent piece. I posted about it at IT Business Edge, so hopefully you'll get some traffic off our site.
http://www.itbusinessedge.com/cm/blogs/lawson/is-it-capable-of-building-a-national-electronic-health-records-system/?cs=31858

[Ognian Pishev]

The following paragraph is a condensed description of the requirements for a proper data model in healthcare:

"The first problem with modeling healthcare data is that models must represent certain concepts (and not others) that will remain stable and true long enough to be built into computer software then used by healthcare providers and patients. Mr. Obama, have you noticed just how much knowledge has, is, and will be accumulating in the medical sciences? Knowledge is codified using words..."

So let's separate what is stable from what is changing or is likely to change. In order to do so openEHR (www.openEHR) defines a two-level modelling methodology and two sets of specifications - an information part and a knowledge part. The first one defines a stable reference model that describes the EHR as a container that holds compositions (documents) which in turn may have entries (observations, evaluations, instructions, etc.). Although the selection of entries is based on a certain model of decision making in healthcare, the model itself does not attempt to represent clinical knowledge. It comes from a variety of sources - terminologies, ontologies, classifications, measurement systems, etc. - and the application of computable knowledge representation to the information model is how data acquires its context-specific meaning. The full computable data models of clinical concepts are called archetypes: blood pressure is an archetype, so is a synoptic report, pathology lab result, etc.
The archetype is meant to represent all the clinical knowledge re. a certain concept. However, in order to make a practical use of archetype libraries we combine them in semantic templates, which, for example, may specify that we need the systolic and diastolic blood pressure, and the arterial, or the pulse, neither are we interested in discussing whether the patient was sitting, standing, etc.
In different circumstances, e.g. in sports medicine we may want to have a baseline reading and then repeat it every 5,10mins. Archetypes allow for different constraints in different templates, so long as we use the same basic clinical models. Once we've selected the template we can further enhance the semantics by querying available terminologies (SNOMED, ICD) and producing bindings to the terminology subsets that should be used with particular archetypes/templates. If our understanding of medicine is enriched, we can produce a new version of the archetypes, or in order to dig deeper, we can specialise them and so on.
The beauty of the openEHR approach which formed the basis of an international standard (CEN/ISO 13606) is that it produces fully computable EHRs that serve as the platform for application development. Semantic interoperability, which simply means that data captured by one application should be reusable by any other application without transformation, is based on archetype libraries freely available in the public domain. Archetypes are produced by clinicians who are the data modelers using a freely available Archetype Editor. ADL, the archetype definition language, is part of the international standard and every single bit of the openEHR specifications is available from the openEHR foundation.

Current discussions of EHRs in the US are describing the HOW and not the WHAT part. If we start by defining the semantics of medical data, then it will be fairly easy to specify a logical record architecture and provide health IT vendors with content specifications for shareable EHRs. It is up to them to design the implementation of a semantically interoperable EHR, which is the only meaningful way of providing for the "meaningful use" of IT in healthcare.

[Christopher Fahey]

It sounds to me like the problem is primarily one of user interface design. It sounds like almost no thought or effort or money was devoted to making the user interfaces actually useful to medical professionals. Every problem you've described sounds like one that is the direct result of letting systems engineers design the user interface.

By blaming the data model, I fear you are focusing attention on a symptom and not on the real underlying problem. The result will be that the data model will be improved endlessly, but the user interface for entering and retrieving the data will still be based on the technology and not on the humans.

Even a perfect data model is useless if the medical professionals can't use it.

To those of us in the user experience and interface design world, the failure of EHS is a profound tragedy because so much that you complain about can be solved if your industry would simply focus on the correct problem.

[cheap laptops canada]

I still prefer the old way when charts were hung at the bed of each patients bed containing detail of his health problems.

[Glen]

The archetype is meant to represent all the clinical knowledge re. a certain concept. However, in order to make a practical use of archetype libraries we combine them in semantic templates, which, for example, may specify that we need the systolic and diastolic blood pressure, and the arterial, or the pulse, neither are we interested in discussing whether the patient was sitting, standing, etc. http://www.hotfilemediafire.com

[Increase GPRS Speed]

The beauty of the openEHR approach which formed the basis of an international standard (CEN/ISO 13606) is that it produces fully computable EHRs that serve as the platform for application development. Semantic interoperability, which simply means that data captured by one application should be reusable by any other application without transformation, is based on archetype libraries freely available in the public domain.

[Electronic Gadgets]

"a nation-covering, interoperable, secure, private, reliable, accurate, and instantaneous electronic healthcare data network is at best terrifying and at worst pernicious."i am in your side.

[LV Cosmic Blossom]

Many cultural forces set the stage for the new art, too.

[how to win at blackjack]

The first one defines a stable reference model that describes the EHR as a container that holds compositions (documents) which in turn may have entries (observations, evaluations, instructions, etc.). Although the selection of entries is based on a certain model of decision making in healthcare, the model itself does not attempt to represent clinical knowledge.

[acid reflux diet]

By blaming the data model, I fear you are focusing attention on a symptom and not on the real underlying problem. The result will be that the data model will be improved endlessly, but the user interface for entering and retrieving the data will still be based on the technology and not on the humans.

[HarrietHOWELL25]

Make your own life more easy get the business loans and all you want.

[computers from warehouse]

It's always amazing reading or commenting on a blog from which we get a full knowledge. Same as here I have found some really interesting information which is simply a great boost to my knowledge.

[computers from warehouse]

It's always amazing reading or commenting on a blog from which we get a full knowledge. Same as here I have found some really interesting information which is simply a great boost to my knowledge.

[oakley outlet]

Dear John et al.,

Hello, Kimia et al.

A reader whom I will call Doug wrote to ask that question. What is your answer for Doug?

[קניית דומיין]

Accusing the data model Im afraid that we are focusing on the symptom and not the real underlying problems. The result is that the data model getting better, but the user interface for entering and searching the data is still based on technology rather than people.

[Acid Reflux]

Wow thats very informative and inspirational. I appreciate your views.

[laser pointer]

Interesting title and interesting article. It offers a good suggestion on data modeling.

[men glasses]

will call Doug wrote to ask that question

[casino online]

the USA government is trying to manage with all country's holes in the budget...

[Chiropractor sydney]

If our nation simply accepts your vision while healthcare IT vendors collect a lion's share of the $20 billion stimulus bounty, individuals and businesses will pay higher medical costs, patients will receive inferior care, medical professionals will lose more of their precious time fighting IT systems instead of delivering better care, and you will be a one term president.

This is so true. I hope his message reach the president.

[Arbeej12]

http://www.stlouisbridal.com

[EMR]

An electronic medical record is not just a typed record of the patient encounter, but an extremely useful decision support tool. The data can be entered into the EMR via any of the two general mechanisms: direct entry by the physician using point and click templates or transcription of dictated notes.

[Poker gratuit]

Thanks !

[beebond@narod.ru]

On the day resembling it it's always a vastly fantastic pleasure to understand your personal phrases
http://kamagra.name