Computerized provider order entry (CPOE) systems can introduce unintended patient harms and it will take more than better design to eliminate them, according to Robert L. Wears, a professor in the Department of Emergency Medicine at the University of Florida.

Systems are designed based on “work-as-imagined”--an order is placed electronically and the appropriate department carries it out--yet “work-as-done” reflects what actually happens in the real world, Wears writes in an article posted recently to the Agency for Healthcare Research and Quality's Patient Safety Network.

He tells of an intern inexperienced on a CPOE system who ordered "CT Abdomen and Pelvis with contrast" from a long drop-down list when she wanted oral contrast, which she ordered separately, but not intravenous contrast. Unfortunately, this included both oral and intravenous contrast bundled with the scan, and the patient developed signs of contrast nephropathy.

Interface flaws can introduce picklist errors, such as having entries too close together, lists that are too long, entries that are worded too similarly or even misalignment of entries for a person standing rather than sitting.

Good picklists, Wears says, use clean typography with good contrast, well-spaced entries, carefully worded choices and require minimal scrolling.

He urges system administrators to monitor order/cancel/reorder triples--how often a provider places an order, realizes it’s incorrect, cancels it and places a different order--which he says are a sign of picklist problems.

In the real world, when an order is placed, generally a nurse “fiddles” with it, he says, clarifying facts with the physician (Did you want oral contrast AND intravenous contrast? Just oral? Just intravenous?)

More empirical studies and observations of real clinical work will be required to move closer to truly represent “work-as-done” in clinical systems, he says.

Still, research published online last month in the Journal of the American Medical Informatics Association seems to support the claims made by Wears. For the study--with contributors from Atrius Health, Brigham and Women’s Hospital, Columbia University Medical Center, Kaiser Permanente Northwest, Partners HealthCare, the University of Illinois at Chicago and the University of Pennsylvania--researchers reviewed all patient medication error reports from six sites participating in a Food and Drug Administration-sponsored project examining CPOE safety.

Of 2,522 reports of errors in the medication ordering phase, more than half (51.9 percent) were related to CPOE. Of these, CPOE facilitated the error in 13.1 percent and potentially could have prevented the error in 86.9 percent.