Tuesday, October 13, 2009

Another Major HIT Project Setback at UCSF: Vendor, Client or Both at Fault?

In yet another example of a major health IT project setback, in August I wrote about UCSF's apparent problems with health IT implementation that I learned about through anonymous comments at the HisTALK blog. At "Lessons Unlearned: Health IT Failure, Act 2" I wrote:

I find it remarkable that this resource-wasting scenario (with possible adverse patient care repercussions) can occur:

  • In a state that's in a severe economic crisis,
  • At an organization that failed severely in a HIT and administrative IT merger ten years ago (in the failed, late 1990's attempted merger between UCSF and Stanford's medical centers, see the 2000 stories "UCSF/Stanford: Marriage was rough; divorce is expensive" here and "A thousand MIS personnel cannot merge two healthcare systems" here),
  • With an EHR product, Centricity, that is the descendant of Logician that others have implemented successfully (including myself, speaking from experience),
  • With GE, a major global high technology vendor, presiding over this new failure at a major academic medical center,
  • With ample preventive material available in books, journals on the web about such failures (e.g., at the many pages and links here and here, as just a few examples).

I asked if vendor and hospital executives bother to read such materials.

Here is an update in this poorly-covered mystery:

Friday, October 9, 2009
UCSF halts clinical IT installation
San Francisco Business Times - by Chris Rauber

Dr. Sam Hawgood, newly appointed dean of the University of California, San Francisco's School of Medicine, confirmed that UCSF has put the brakes on installation of a $50 million clinical IT system from General Electric [which had acquired Medicalogic's "Logician" EHR some years ago - ed.]

In late August/early September, the Business Times was unable to reach high-ranking UCSF officials to comment on anonymous reports on the respected HISTalk blog and by sources that UCSF was unhappy with early results of the electronic medical records system installation. An informed insider says GE was “way behind schedule” in writing code, and “UCSF got fed up with the endless GE delays,” and is looking to identify other vendors for a drug order entry system.

“We are taking a pause to evaluate our best options moving forward, and we will be making a decision in the next two to three months, and then moving forward aggressively,” Hawgood said. The delay will not put the IT project behind schedule, he said, because “once we make a decision regarding the vendor, we’ll be back on track for an aggressive installation.” [Unfortunately, that sounds like wishful thinking or spin to me - ed.]

UCSF has brought in consultant Kurt Salmon Associates to help it evaluate the IT project’s woes, which were said to be creating considerable frustration within UCSF Medical Center, and obviously the School of Medicine as well.


This setback is of great concern to me. I believe such scenarios could become commonplace in coming years as healthcare organizations bow to the ARRA-created pressure to computerize "or else" by 2014 - that is, suffer reimbursement penalties for not being "meaningful users" of HIT. (Whatever that somewhat presumptuous term describing a largely experimental technology will ultimately come to mean.)

I have frequently written about the HIT vendors being dominated by those with an MIS (management information systems or "business computing") background, and unshakably and arrogantly deficient in talent management where Medical Informatics expertise is concerned. GE may also suffer from domination by engineers whose primary experience is at the level of capital equipment, PACS etc., much as I wrote about competitor Philips Medical here and here.

Having once worked for a GE competitor myself, Comdisco Healthcare Group, and having asked GE representatives about what a phrase on a banner at an RSNA show stating "GE: Leader in Radiology Informatics" meant -- and getting blank stares and comments that "it has something to do with the computers connected to our xray devices" back in return-- my concern was that the problems are not just UCSF centric but vendor centric as well.

UCSF is a big, complex organization, with a lot of very smart clinicians and lots of politics, and I conjecture that GE bit off more than it could chew regarding development, customization and deployment of major health IT at such an organization. It requires far more than technical excellence.

My concern is that GE, along with many if not most of the other major health IT vendors, lack the Medical Informatics and Social Informatics talent and depth to make our ambitious national EHR plans a reality. The overselling of vaporware and "yes, we can do that, no problem" promises by sales and marketing are also a concern, as I find common in HIT where salespeople promise the world to close a deal. Then, the technical people need to play catch up to the grandiose promises made by their creative sales colleagues.

I fear in a few years we will be in the situation that the UK's National Programme for IT (NPfIT) in the NHS is in now.

If we want to avoid that fate, we as a country must:

  • Increase transparency and information diffusion about HIT difficulties and failures greatly. That my website on HIT failures is still nearly unique on the Web after ten years is symptomatic of a true lack of information sharing on real world HIT problems. My monitoring of access patterns to the site as reported in this 2006 AMIA poster (PDF) strongly suggests the demand for such material far exceeds the supply. The AMIA/AHIMA book of which I am an associate editor entitled "H.I.T. or Miss: Lessons Learned from Health Information Technology Implementations" and the new journal "Applied Clinical Informatics" whose Editor-in-Chief is Dr. Chris Lehmann, informaticist at Johns Hopkins, are a start -- but just a start on candid information diffusion about applied HIT realities.
  • Health IT vendors need to understand that those in MIS and engineering are, in the context of complex clinical settings where clinician-supportive HIT is to be developed and deployed, often dyscompetent (they fail to maintain acceptable standards in one or more areas of professional practice) or even incompetent (lacking the requisite cognitive and non-cognitive abilities and qualities to perform effectively in the scope of professional practice). Lacking an understanding of medical culture and the nature of medical settings is a highly compromising deficit. It leads to mission hostile HIT devices such as shown here. There needs to be much better talent management in that regard.
  • Suboptimal HIT vendor performance, and defective HIT devices, should not be tolerated. Repeat purveyors of such technology should be materially sanctioned for wasting precious healthcare resources. Whether this happens primarily in the courtroom, or in the court of "consumer opinion" by HIT buyers -- based on transparent consumer reports on HIT - remains to be seen.

-- SS

Addendum: a reader familiar with the UCSF situation largely confirms my suspicions as above regarding vendor capabilities and the vendor's biting off more than it could chew, greatly delaying deliverables. They did say, however, that I was not correct (in this case) about the issue of salesperson "promise the moon" behavior, and that this was not a factor in the project stoppage.