Friday, March 5, 2010

Computers and Prostate Problems in Pennsylvania, East and West

At "Bungled Brachytherapy, Computer Interfaces and Other Mysteries At The Philadelphia Veterans Administration Hospital" at this link I reported on serious problems involving brachytherapy treatment of prostate cancer at the VA Medical Center in Philadelphia.

One of the issues involved computer problems, in the form of failure to network a key computer involved in treatment evaluation.

Now at the other end of the state, Pittsburgh, more prostate-related computer problems have occurred:

Prostate cancer test interpretation flawed
By Walter F. Roche Jr.
PITTSBURGH TRIBUNE-REVIEW
Friday, March 5, 2010

A computer programming error caused West Penn Allegheny Health System's laboratory to send physicians incorrect interpretations of prostate cancer tests for 288 patients over 15 months.

Hospital officials say physicians who ordered the tests were advised about the errors in recent weeks. They were sent revised, corrected interpretations, said Dr. Jan F. Silverman, chairman of the Department of Pathology and Laboratory Medicine.


One wonders how such a "programming error" can occur.

Silverman said actual test results were correct, and most physicians would rely on those and not interpretations. He said hospital officials found no evidence that incorrect test interpretations resulted in delayed or improper care ... The erroneous interpretations were provided on a test physicians use to assess whether patients need biopsies of their prostates. The test provides a comparison of total prostate specific antigen, or PSA, versus free or non-attached PSA.

That there was no apparent delayed or improper care was by happenstance. The purpose of health IT, however, is not to give physicians the opportunity to have a lucky day, or to have placed upon them the additional cognitive burden of deciding which is correct: the test results, or its interpretation.

This episode raises another question: in addition to whatever "programming error" caused this problem, was there no QC of the actual reports to ensure the "interpretation" matched the pathological, serological and other results and findings?

Dr. Ralph Miller, head of the Allegheny Prostate Cancer Center, said it was "theoretically possible, but very, very unlikely" that erroneous interpretations resulted in delayed or improper care.

[That may be true, but is it due to luck and/or the inconvenient fact that most physicians 'did not rely' (i.e., ignored) the computer-generated interpretations? Also, will luck run out the next time a "programming error" occurs, resulting in dead patients? - ed.]

Silverman said those interpretations were sent between Oct. 1, 2008 and January. Of 818 PSA tests the West Penn Allegheny Core Laboratory performed during that period, 412 included comparisons of the two PSA figures. Of those, 288 included incorrect interpretations of that ratio, Silverman said.


That's a very high percentage of error. Should that have occurred in a drug trial, the FDA would likely have been all over the responsible parties. However, health IT is unregulated, therefore all that's required is an "please excuse us, we'll do better the next time" from the involved parties.

The programming error was discovered recently when a physician questioned an interpretation, Silverman said.

I have written before on these electronic pages that physicians and clinical settings should not be the testing labs for IT personnel, with the clinicians using their clinical skills in locating programming bugs.

-- SS