EMR Failures and a Recipe for Their Avoidance in the Future

Everyone involved in healthcare information technology (HIT) is familiar with the history of frequent failure or inadequate performance of EMRs. Just to get a rough handle on the breadth of this problem, I searched Google using the following search terms: failure OR problem EMR OR EHR. Here is the resulting search that shows 2,670,000 hits. Not a perfect measure of the depth of the problem but a good way to launch this discussion.

With this background information in mind, it would seem reasonable to assume that some HIT professionals would be wringing their hands. After all, hospital CIOs and CMIOs are changed with the purchase and deployment of EMRs. Mr. HIStalk opines below about how to avoid EMR failures in response to a query from a blog reader:

From Greg Tourniquet: "Re: CIS failures. AMIA [American Medical Informatics Association] keeps talking about the value of publishing CIS [clinical information system] failures and lessons learned. There is a formal initiative that we can look forward to: A group of battle-scarred CMIOs is writing a book; they recently put out a request on their listserv for ‘tales from the trenches.’ ...

[Response from Mr. HIStalk] That’s what the industry needs. That plus an assessment tool that I’ve advocated previously: a readiness checklist that would tell a hospital how high it should set its sights, i.e. if the culture and change management capability is primitive, don’t run off and buy a $50 million clinical system - stick with ancillary department task automation, data analysis, and integration and call it a job well done. The money wasted by the hospital industry on ineffective IT implementations is embarrassing. It’s not the vendors’ fault - nobody made them buy - but they consistently underestimate the challenge despite ample available evidence. I’d buy that book.

Although a book by "battle-scarred CMIOs" with "tales from the trenches" might prove to be an amusing read on a rainy day, my instincts tell me that it would offer few real solutions to the EMR conundrum. My rationale for such a statement is hinted at in Mr. HIStalk's proposed solution above: stick with ancillary department task automation. I will take his definition of ancillary information systems to primarily involve LISs, RISs, PACs, and pharmacy information systems. Unfortunately, few CIOs and CMIOs have in-depth knowledge about the lab, radiology, or pharmacy systems. They are managed by the knowledge and process domain experts in the specific hospital departments, frequently with hosting by the central IT departments. However, this hosting could be provided more efficiently and cheaply from remote server farms.

I respectfully disagree with Mr. HIStalk's assertion that this sorry state of affairs is "not the vendors' fault." For their part, EMR vendors continue to pursue their short-term strategy of selling inadequate and poorly designed systems that that are difficult to install, don't scale up, or perform inadequately. CIOs and CMIOs keep buying inadequate EMRs, despite the long  history of failure of these systems, because they are paid to deploy "modern" systems and there are few if any in the market that respond to their wish lists. You have two sets of players that are pursuing a flawed business model in concert.

For me, the answer is the wide adoption by EMR vendors of the federated system architecture whereby the ancillary content experts are allocated "white space" in the EMR to organize and populate with the information generated in their own ancillary departments. The "integration problem" raised by Mr. HIStalk is solved and the EMR becomes an ordering and reporting conduit for the various specialized hospital systems including nursing and physician applications. The bad news for the CMIOs under this approach is that most of their jobs will disappear. The good news is that simplicity and rationality will take hold and that the history of failure for EMRs will begin to decline.

Drs. Ul Balis and Mark Routbort will lecture on both order entry and results reporting using a federated architecture at the Lab InfoTech Summit on April 11 in Las Vegas.

Comments

[bev MD]

Dr. Friedman;

Could you elaborate on how the federated system architecture differs from the "best of breed" approach, wherein a separate LIS, RIS, ED system, etc. are all interfaced to a central HIS - which contains the patient demographics and order entry system? At our hospital, the "party line" was that maintaining those interfaces was too difficult and thus we had to migrate to an integrated system like Cerner Millenium - with the predicted loss of functionality for the individual departments....