Thursday, December 25, 2008

A Biomedical Informatics Manifesto

"We need to recognize that in our society's efforts to improve healthcare, we are attempting to create virtual clinical tools for complex and unforgiving medical settings that happen to involve computers, and that facilitate healthcare, not information systems that happen to involve doctors and clinicians that (in line with the bellicose grandiosity characteristic of IT marketing) will "revolutionize" healthcare. - S. Silverstein (me - ed.)

The American Medical Informatics Association is moving towards having the field of Biomedical Informatics formally declared a medical subspecialty with the help of a $300,000 grant from the Robert Wood Johnson Foundation. While this is good, it is a long time in coming.

I have been thinking a great deal about the recent Joint Commission Sentinel Events Alert on the potential dangers of poorly conceived and implemented healthcare IT. I wrote on this issue at the post "Joint Commission Sentinel Events Alert On Healthcare IT".

Noteworthy is the fact that the research of current Biomedical Informatics specialists and allies figure heavily in the Alert’s references.

I can only wonder what motivated the Joint Commission to issue such a potentially industry-adverse advisory (adverse to common and profitable health IT industry practices that create these issues, that is) at this time. President-elect Obama, after all, has made HIT a significant part of his agenda.

Perhaps as cases of HIT difficulty and patient harm become more commonly known, the Commission does not wish to appear to have been caught unaware as have other regulatory agencies in the face of recent turmoil in the housing and financial sectors. I have heard rumors from my readers, in fact, of major stories about HIT malfunction that might in fact reach a wide public audience in the coming years. Considering the turmoil in the housing, economic and investment sectors, perhaps the Joint Commission does not want to be seen in the same light as a Fannie Mae or an SEC.

In any case, the alert's appearance raises a meta-issue:

Why was such an alert even necessary?

None of the issues the alert raises, ranging from safety risks, unintended and preventable adverse events that these implementations can create or perpetuate via fundamental errors of either commission or omission, ill conceived human-machine interfaces and organization/system design, etc. would be very commonplace if the teachings of 40+ years of Biomedical Informatics research was heeded.

Some of those researchers could probably have written the Joint Commission alert on HIT in the 1960's. In fact, they largely did.

Biomedical informatics research goes back to the pioneers such as the writings in 1968-9 of Dr. Donald Lindberg, (now director of the U.S. National Library of Medicine at NIH and a key figure in funding for Bioemedical Informatics postdoctoral training), the "Ten Commandments" of Octo Barnett regarding HIT in 1970, the five major reasons for lack of HIS success by Morris Collen in 1972, the four reasons for lack of HIT diffusion by Friedman and Gustafson in 1977, and many others [1].

Why has it taken forty years for this research to be reinvented and regurgitated by a major healthcare regulatory agency? How many billions of dollars have been wastefully spent, and how many patients have been adversely affected by bad informatics, in all those years?

Biomedical informatics research currently includes the work of the formal departments and centers in the field such as these funded by the NIH in the U.S. and others funded by their own universities, departments of Biomedical Informatics in other countries, sociologists (e.g., Ross Koppel, Jos Aarts to name just a few), social informatics experts, the iSchool consortium, and many others working in a cross-disciplinary manner.

In my own case, I have been writing on these issues since my eye-opening experiences as a Chief Medical Informatics Officer (CMIO) in the mid-1990's at Delaware's largest healthcare organization. I am astonished that the issues I wrote of then are current today; I often hear the exact same stories from current CMIO's over a decade later.

I can say that none of the Joint Commission warnings were necessary on my watch as CMIO; I was cognizant of all these problems as a result of several years of studying the issues in the time I spent in the deep cave known as the Yale Cushing/Whitney Medical Library during my Biomedical Informatics fellowship and faculty time, as well as by direct observation using the simple skills of a medical internist but applied to IT.

My writings have been widely viewed (see my 2006 AMIA poster "Access Patterns to a Website on HIT Failure", ppt file at this link). Yet, as a Google or other search engine search on "healthcare IT failure", "healthcare IT difficulty" or similar concepts demonstrates, explicit information on this topic remains curiously limited. On the contrary, a search on "medical malpractice" brings up literally millions of "hits."

Why do the healthcare IT dangers that are the subject of the JC alert occur at all? A clue can be found in the following observations.

The problem can be summed up like this:

Biomedical Informatics as a specialty might as well be invisible. Amateurs rule HIT.

(Note: I use the term "amateur" not in a pejorative sense, but in the sense that I am a radio amateur or “ham” with formal licensure after governmental examinations and significant technical and applied telecommunications experience. I am not a telecommunications professional, however, who should be, say, leading a critical telecommunications initiative for a hospital or large business.)

Here is my most recent and indeed most personal evidence Biomedical Informatics is still largely invisible. My former chair of medicine from my residency, a consummate clinician trained at Yale and now Chief of Staff of an 800+ bed hospital complex, opined regarding an opening for a Director of Clinical IT at his organization (in reference to my application, which was summarily ignored by the IT department other than the typical form letter reply) that "I do not believe it is a Sine qua non that a physician be at the helm of the clinical informatics enterprise". He knew of my career path since my residency in the mid 1980's.

Now, this same person wrote a letter of recommendation in 1992 to Yale's Center for Medical Informatics about my qualifications for a Biomedical Informatics postdoctoral fellowship at his alma mater. The letter indeed mentioned the evening in 1986 when I went "way outside the box" as a resident at his hospital and used my IT expertise to repair a CT scanner's computer that was malfunctioning and that had been abandoned for the evening by the vendor.

My repair brought the CT scanner online, and changed the intervention on a mid 20's unknown male brought to the ED with profound mental status changes from a potentially catastrophic lumbar puncture to the appropriate neurosurgical intervention - to manage a huge hydocephalus caused by a benign obstruction within his brain. The patient survived.

The statement directed towards me that "a physician is not a Sine qua non of a healthcare informatics initiative" thus clearly identifies me as "a physician", not a "formally trained and experienced biomedical informatics professional who is also a physician." (One would hopefully never hear that "a surgeon is not a Sine qua non of major abdominal surgery.")

That the field of biomedical informatics could have been so invisible to such a person suggests to me the field is simply a Black Hole to others even less well informed about its practitioners:


The environment around an invisible Black Hole. Is this how the
healthcare industry sees the field of Biomedical Informatics?



Other occupations (I've personally seen IT personnel, MBA's, business consultants, "process re-engineers", 'techie' docs and nurses with just enough IT knowledge to be dangerous, social workers, and a cadre of others) are thus free to commit what I term "cross-occupational piracy" upon HIT. They do not do a stellar job of it.

Thus, the Joint Commission alert becomes necessary.

This phenomenon was also apparent in pharma. I note the following essay at the excellent blog "In the Pipeline" by medicinal chemist Derek Lowe, PhD, in a Dec. 19, 2008 post entitled "My Compound Goes Where the Wild Goose Goes" that:

A colleague and I got to talking yesterday about something that I'm sure many chemists have noticed. Have you ever chased down some reaction or compound in the literature, only to find yourself wild-goosing back to some obscure journal that no one has ever read - just because no one can be bothered to publish a modern procedure?

Here's how that typically works. You run a SciFinder search on Molecular Structure X. A list with a dozen references comes up. There's a Tet. Lett. from 2002, but what are the chances it'll have any spectral data (or anything useful at all?) Ah, there's one from Tetrahedron in 1995, that should do. So you look over the PDF, search for your compound. . .there it is, number 17. Now to the experimental. . .and you find in the first paragraph that "Compound 17 was prepared according to a published procedure", footnote thirty-eight. And the footnote is to. . .ay, it's to a Chem. Ber. paper from 1932. Ausgezeichnet!

Oh-kay. Back to that SciFinder reference list. How about that Tet. Lett. paper? Nope, on inspection, it turns out to reference the 1995 paper you just looked at. What else? There's a JOC from 1984, let's try that. Good ol' JOC, solid stuff. Well, digging up that PDF, you find that it refers to a 1980 paper from the same group from Synthesis. Hrm. So you chase that one down, there it is, compound 9, and the experimental for it is. . . footnoted to the 1932 paper. Again.


What he is referring to is the "wild goose chase" that often plagues medicinal and other chemists in performing syntheses, essential to the drug discovery process. He is pointing out the use of the cheminformatics resource known as "SciFinder" by the American Chemical Society. An ACS division provides the largest databases of publicly disclosed chemistry-related information and makes the information accessible through search and retrieval software such as SciFinder, along with links to the original literature and patents.

SciFinder is an essential tool to the modern research chemist. Some pharmas (e.g., Pfizer) made this tool widely available to its scientists. Which makes my experience in Big Pharma astonishing, in that I as a science library director in a company with a declining pipeline of new drugs had to fight a long battle with a non medical, non science-background IT VP over ending rationing of said tool, among others, to save a few million $ per annum. I was dumbfounded. This in a drug giant boasting $60 billion in market clout and 60,000 employees.

In the end, my department received only a fraction of what I asked for, and that was soon taken away (not from me personally, but from the scientists, ultimately) as budgets declined. My appeals to senior management as a Biomedical Informatics professional, a.k.a. information scientist, might as well have been invisible as well.

I was ultimately laid off by the VP for my efforts, leaving laggards who had tolerated this mind-boggling, truly absurd situation that impaired drug discovery behind.

Black hole OJ287, mass of 18 billion Suns.

Unfortunately, I doubt FDA will issue a similar alert about the dangers of research IT leadership by non scientists.

In effect, the Joint Commission Sentinel Alert affords an opportunity for a return to some sanity in healthcare IT endeavors.

It is my view that the Biomedical Informatics community and its allies need to leverage this perhaps first-ever acknowledgment from an organization with clout that HIT is not a harmless panacea and needs to be subject to informed discipline and rigor (as medicine itself). They need to leverage it to end the "black hole of invisibility", irrespective of the bureaucratic and prolonged process of having the field declared a medical subspecialty, the outcome of which is uncertain.

The alert needs to be leveraged each and every time a Biomedical Informatics professional finds themselves being marginalized. It needs to be leveraged as a hammer. "Political correctness" (aptly abbreviated "PC", pun intended) on these matters ultimately gets patients killed. You know you're afflicted with the terminal PC bug when you feel reluctance to tell your CIO and IT personnel who lack training, knowledge and experience in clinical medicine, and who've just concluded contracts for the world's worst clinical IT without sufficient end user involvement, that they, in fact ... lack training, knowledge and experience in clinical medicine.

When told that the CIO or IT people or the COO or the CFO or the MBA process re-engineers "know better" than informatics experts what clinicians should be doing and what they need to care for patients; that informatics experts don't have enough project management or other mundane experience; that they are too academic; that doctors don't do things with computers; that a biomedical informatics expert is not the "Sine qua non" of a major healthcare informatics initiative; that the vendor whose products impaired a national HIT initiative overseas has assured us their products are just dandy, and so forth ...

... the reply should itself be a well-justified bit of bellicosity.

Biomedical informatics professionals should stand up, speak out, and make it count:

"What part of 40+ years of Biomedical Informatics research and experience don't you understand?"

-- SS

Reference:

[1] A History of Medical Informatics in the United States 1950-1990 (pp. 167-175), Morris F. Collen B.E.E., M.D., 1995, American Medical Informatics Association, ISBN 0-9647743-0-5.