Wednesday, March 23, 2005

Wash. Post on Cedars-Sinai CPOE failure

For anyone who doubts clinical IT is an area where angels fear to tread, there's this story in the Washington Post:

Cedars-Sinai Doctors Cling to Pen and Paper
By Ceci Connolly, Washington Post Staff Writer
Monday, March 21, 2005

Excerpts, with notable passages in bold and my interspersed comments in italics:

... For every doctor, nurse and executive here, there is a different explanation of what went wrong. The technology, created in-house, was clunky and slow. Only a fraction of the 2,000 doctors with privileges at the hospital were involved in developing the system, even though they faced a dramatic change in the way they practiced medicine, from jotting notes on a clipboard to logging onto a computer to type in their treatment and medication orders. Training was insufficient, and administrators opted for what Hackmeyer called a "big bang" implementation rather than switching one ward at a time.

Now, two years later, the hospital often viewed as an industry leader is being held up as a cautionary tale in the drive toward bringing medicine into the computer age, and officials here say they have no intention of trying again for at least a year.
(note: my opinion on their initial remediation plans were published in Health-IT World in Sept. 2004 -- SS)
The marriage of information technology and medicine is all the rage in health policy circles. Five years after the Institute of Medicine issued a landmark report cataloguing the life-and-death consequences of medical errors, corporate leaders, politicians and physicians are embracing computer-assisted health care.

Yet the spectacular failure at Cedars-Sinai -- described by Bush's technology guru as "the worst case" he has seen -- demonstrates how difficult it can be to make the transition. Even well-financed, sophisticated hospitals face enormous hurdles moving from the Marcus Welby era of pen and paper to one in which doctors spend precious minutes entering data into a machine that never went to medical school and does not have the flexibility to make nuanced judgment calls.
(Neither did most of the IT people developing and leading implementation of such machines, but that seems a problem that is rarely addressed -- SS.)
... "The important lesson of the Cedars-Sinai case is that electronic health record implementation is risky," David J. Brailer, national coordinator for health information technology, said in an interview. "Up to 30 percent fail."
(We're talking billions of dollars of capital and expense dollars that hospitals simply do not have an abundance of -- SS)

... Each time a patient arrived, pulmonary specialist Andrew S. Wachtel would have to find a computer (preferably one of the newer, faster ones), log in and begin checking boxes in at least a half-dozen categories to indicate the patient's symptoms, allergies, diagnosis, tests and medications. A task that once took three minutes to scribble shorthand at the patient's bedside suddenly devoured 30 to 40 minutes, he said.
(Reminds me of the passage in the book "House of God" about finding a medical student who only triples an intern's work - SS)
... even techies found flaws. The system refused to recognize even slight misspellings, so Hackmeyer's efforts to order the laxative Dulcolax -- easily understood by nurses even if he was off by a letter or two -- were thwarted by the computer. It was also impossible to use it to order "clear liquids and advance diet as tolerated," another routine instruction when easing a patient back to solid foods, he said.
(Even Microsoft Word does a better job on typos than that -- SS)
... But the biggest complaint -- with potentially dangerous implications -- involved the automatic alerts that flashed on the screen every time a doctor made an out-of-the-ordinary request. Designed to catch errors before they occur, the alerts became an unending series of questions, reminders and requests on fairly basic decisions.
Infectious disease specialist Stephen Uman said he went around in circles trying to give patients the antibiotic Vancomycin. Although the recommended dosage is 928 milligrams, Uman knows to round up to 1 gram because pharmacies dispense the medication in multiples of 250 milligrams. But when he typed 1 gram into the computer, the machine rejected the request.

Cedars-Sinai was unable to strike a balance between useful computer warnings and a machine that seemed to constantly cry wolf, acknowledged Harold, the former chief of staff. "Buried in those annoying alerts is probably one life-saving alert," he said.
("Unable" to strike a balance? That sounds like an interesting story in itself. -- SS)
... Even with that data, Cedars-Sinai is in no rush to try again. The hospital is waiting for the technology to improve and perhaps for more young, tech-savvy doctors to arrive. In the meantime, Neil Romanoff, the physician who oversees safety procedures here, said the hospital relies on extra layers of staff to double- and triple-check its procedures.
The technology will improve when those behind it have a better grasp of the complexities of medicine. Pioneers in EMR such as Morris Collen, Donald Lindberg, Octo Barnett, and others published guidelines on how to best implement clinical IT that warned against the cornucopia of problems mentioned in this article. Sadly, they did so starting in -- and perhaps even before -- the 1960's.
A good source on references to the wisdom of the pioneers is "A History of Medical Informatics in the United States 1950-1990, by Morris Collen MD, section 3.4. For example, Octo Barnett's clinical IT "10 Commandments" (written in 1970) as reproduced in Collen's book on page 169, as well as 1960's and 70's material by Lindberg, Lamson, Collen, Davis, Baker, and numerous others address many of the substantial problems encountered by Cedars-Sinai in 2003 as documented above.

Medical informaticists are specifically educated in the necessity of minimizing the problems mentioned above, I should add, yet often are not sufficiently empowered to effect change. In my case, efforts to avoid such issues in clinical IT projects were often resisted by IT personnel who "knew better" than the "informatics guy" whose ideas were "way out there" (the actual words of an executive in charge of IT, as reported to me by people who disliked his micromanagerial and bullying style. Of note, that executive is now a CEO at a major healthcare facility).

-- SS

1 comment:

singularityblog said...

I know this post is now 4 years old, but I just ran across it and wanted to thank you for your analysis following the article, and for providing a list of useful resources. I wish more bloggers would use the platform to inform and connect to the best resources rather than just try to keep all the traffic on their site. Thanks!