Technology Hasn't Eliminated Medication Errors Yet, USP Reports

Cheryl A. Thompson

Nearly 20% of the medication errors reported to a national database in 2003 pertained to problems with computerization or automation, the United States Pharmacopeial Convention (USP) announced in December.

The statistic is based on the 235,159 records submitted anonymously by 570 health care facilities, including hospitals and outpatient units, to USP's Web-based MedMARx database.

"Computer entry has been an increasing cause of error in MedMARx, in both rate and frequency, among all causes reported for medication errors," Diane D. Cousins, vice president of the USP Center for the Advancement of Patient Safety (CAPS), told reporters during a December 20 teleconference.

In 2003, MedMARx received 27,711 reports of computer-entry errors.

John P. Santell, CAPS director of educational program initiatives, said hospitals and health systems have been turning to computerization as a way to eliminate the errors, such as those caused by illegible handwriting, inherent to the manual processing of medication orders.

"Although computers have made improvements in the provision of care, USP's MedMARx data findings suggest that there is still much room for improvement," he said.

Computer-entry errors have steadily increased over the years, Santell said, starting as the seventh leading cause of errors in 1999 and reaching fourth place in 2003. But this increase could in part be a consequence of health care facilities' greater use of computer technology, he noted.

USP defines a computer-entry error, he said, as the "entry of incorrect or incomplete information into a computer system associated with the medication-use process."

Of the computer-entry errors reported in 2003, 71% were "intercepted" before reaching the patient, Santell reported, and 0.74% of computer-entry errors resulted in actual harm to the patient.

Some 82% of the computer-entry errors, he said, originated after the prescriber made the medication order but before a health care professional administered the dose to the patient, possibly because information did not pass seamlessly from the start of the medication-use process to the end. For example, errors occurred when personnel had to enter data into the pharmacy computer system from a handwritten medication order.

More attention must be paid to how a computer system is used in an actual health care environment, Santell said. "Both health care practitioners and information technology programmers need to work together to construct computer-entry display screens and answer some key questions like, Do you want to spell out the dosage, or should it be represented numerically?" Spelling out the dosage, rather than using numerals with a comma or decimal point, can help to avoid some errors, he explained.

As with handwritten medication orders, health care professionals can misunderstand the intent of abbreviations in computerized orders, Santell said. "There are limitations on the character lengths in the data-entry screens for many computer-entry systems, and this forces information technology programmers to truncate, or abbreviate" drug names, dose frequencies, routes of administration, and dosage forms.

More than half of the computer-entry error reports citing a person's inadequate performance as the cause—the foremost reason for this type of error—also mentioned distraction or interruption as a contributing factor, Santell said.

"The implication here is that computer-entry tasks should be viewed as serious cognitive actions requiring clinical judgment," he said, and that "health care professionals should be sufficiently insulated from distractions and interruptions and adequately trained on the system."

Distractions contributed to nearly 57% of the computer-entry errors reported to USP. MedMARx introduced computerized prescriber order entry (CPOE) as a distinct type of error in May 2003 and analyzed the data in collaboration with the federal Agency for Healthcare Research and Quality. More than 7000 of the reports submitted in 2003 cited CPOE as the cause of the error; 33% of these reports mentioned transcribing, documenting, dispensing, administering, or monitoring—not prescribing—as the phase in the medication-use process tied to the error. The errors attributed to CPOE resulted in patient harm 0.1% of the time.

As for errors linked to the use of automated dispensing devices, 1.3% of these errors resulted in patient harm, Santell said.

Nearly 70% of the 8000-plus errors reported with automated dispensing devices in 2003 involved an incorrect dose or drug, he said. Many of these errors, he explained, occurred because someone filled the device with the wrong quantity or strength of drug product, placed the correct product in the wrong drawer or bin, or retrieved the wrong product from the device.

USP suspects that problems with the use of automated dispensing devices "may actually be more widespread than previously recognized," Santell said.

"We believe these findings should stimulate a national review by professional organizations and manufacturers of these devices to develop best practices with the goal of reducing medication errors," he said.

MedMARx, which launched in August 1998, now receives about 20,000 reports each month on actual and potential medication errors, Cousins said.

About 1.5% of the reports received in 2003, she said, described instances in which a medication error harmed the patient or the patient died. This percentage is less than in previous years, she added.

Cousins said the latest statistics are part of MEDMARX 5th Anniversary Data Report: A Chartbook of 2003 Findings and Trends 1999–2003, available from USP.