CPOE System Can Create Errors, Study Finds

Kate Traynor

Computerized prescriber order-entry (CPOE) systems are widely touted as a way to improve patient safety, but the systems have the potential to introduce new errors into the order-entry process, according to researchers at the University of Pennsylvania School of Medicine.

In a report published in the March 9 Journal of the American Medical Association, the researchers examined how hospital staff interacted with a CPOE system that was in place at the university hospital from 1997 to 2004.

After conducting surveys and interviews and shadowing hospital staff as they worked, the researchers uncovered nearly two dozen situations in which use of the CPOE system could lead to medication errors.

The researchers found that one source of mistakes arose when information generated by the CPOE system was fragmented or was not integrated with other hospital data. For example, prescribers sometimes interpreted the smallest available dosage size of a medication displayed in the CPOE system as the lowest recommended dosage for the product. In such cases, the prescriber is mistaking pharmacy inventory information for clinical information, and can choose an inappropriate dosage for the patient.

Nearly 75 percent of the 261 prescribers who participated in this portion of the study mistakenly used the CPOE system to identify low-dosage information for infrequently prescribed medications. Similarly, about 80 percent of prescribers used the CPOE system to determine dosage ranges for infrequently ordered drugs.

Another type of error identified by the authors was caused by a poor "human-machine interface." Such errors included selecting the wrong patient or drug from the CPOE display. According to the report, for the CPOE system studied, "names and drugs are close together, the font is small, and...patients' names do not appear on all screens."

Nearly half of prescribers reported that the poor-quality CPOE display made it difficult to correctly identify patients. About 70 percent said the need to work through many screens to display a patient's entire medication summary created uncertainty about which drugs had been administered.

According to the report, the CPOE system regularly crashed two or three times a week for 15 minutes at a time. About 85 percent of study participants said that medication administration delays occurred during such system shutdowns. Another problem related to system shutdowns was the sending of medications to a patient's room after the patient was moved elsewhere, because the move was not documented in the pharmacy computer while the CPOE system was down.

The report noted that the study involved potential medication errors and not actual adverse drug events resulting from faulty use of CPOE. According to the report, pharmacists act as a check on the CPOE system by reviewing all of the hospital's medication orders. About 4 percent of the hospital's orders are rejected by a pharmacist, the report stated.

The authors recommended that CPOE implementation plans should focus not on technology alone but also on the usual work flow and habits of the people who will use the system.

"CPOE systems need to be very carefully designed and implemented, as well as constantly evaluated and improved," lead author Ross Koppel explained in a statement. "As these systems continue to be improved, designers should understand that their programs must seamlessly integrate into an institutional context of infinite complexity."

Koppel acknowledged that the study involved "only one older CPOE system in a single setting," but he said the study's findings "reflect what is happening in health care facilities across America that have adopted CPOE systems as a key patient-safety initiative."

The University of Pennsylvania Health System currently uses an upgraded CPOE and clinical information system from the same manufacturer as the one Koppel and his colleagues evaluated for their study.