Computerized physician order entry has been hailed as a breakthrough for reducing medication errors, but the systems can introduce new types of mistakes that can themselves lead to adverse drug reactions and other errors, according to research conducted by Dr. Kathleen E. Walsh of the University of Massachusetts, Boston, and her colleagues.
The investigators sought to determine the rates and incidence of computerized physician order entry (CPOE) in pediatrics, noting that computer-related errors are the fourth leading cause of medication errors, according to the U.S. Pharmacopeia's MedMARx system (Pediatrics. 2006;118:1872–9).
Their review looked at use of CPOE over a 9-month period in a single urban teaching hospital with 40 general and surgical pediatric inpatient beds, 4 pediatric intensive care unit beds, and 15 neonatal intensive care unit beds. (The emergency department, operating room, and postanesthesia care unit were not included in the study because they did not have the system.) Only residents could write and enter orders.
Forty patients per month were randomly selected. Reviews of the charts, hospital incident reports, and order-entry logs were done by two pediatric nurses and then two physicians. Overall, they reviewed 352 of 975 PICU, NICU, and inpatient ward admissions and 6,916 medication orders.
The researchers detected 104 errors, of which 20 were computer-related. Seven were serious medication errors, and 13 were errors with little potential for harm. The rate of computer-related errors was 10 per 1,000 patient-days; the rate for serious computer-related errors was 3.6 per 1,000 patient-days.
Two of the serious errors were duplicate medication orders, four were menu selection errors, and one was a serious keypad entry error. There were eight order set errors, none serious.
Three of the 7 serious errors were orders for ceftriaxone overdoses that generated alerts, but physicians overrode the alerts. None of the computer-related errors reached a patient.
They concluded that the rate of serious medication errors in their study was slightly higher than that found in similar studies. The researchers found fewer types of errors than had been previously reported, but, they said, “our findings do support the assertion that problems with the human-machine interface can pose a risk to patients.”
The study was limited by the fact that it was done in a single hospital that relied on a single order-entry system. But some design features are similar across different software programs, the authors said.