In just two weeks, a computerized double-check of intravenous drug doses programmed by nurses discovered a potentially harmful miscalculation 28 times-alerting caregivers before they could send overdoses coursing into patients' veins through automated IV pumps.
The test of a new technology for safer administration of IV medications was limited to an eight-bed trauma and intensive-care unit at a North Carolina hospital, but it provided some of the first field results illustrating the impact of a new class of computer intelligence for detecting serious dosing errors at the bedside.
Although the incidence of dangerous dosing represented only a fraction of the more than 2,000 drug calculations programmed during the trial period, it was equivalent to two "catches" per day for a handful of beds, the study's researchers say.
Findings of the clinical trial, conducted at the University of North Carolina Hospitals, were presented earlier this month at the midyear meeting of the American Society of Health-System Pharmacists in New Orleans.
A vulnerability in the system
The combination of IV-drug delivery machines and an alert system for preventing doses outside a defined minimum and maximum range tackles a vulnerability in the medication process that can quickly lead to harm, says James McAllister, director of pharmacy for the 672-bed Chapel Hill, N.C., hospital complex.
"When a nurse gets to the bedside to do an IV, there's no one looking over her shoulder," McAllister says. "If an error is made, it's done-it's in the patient already."
When a mistake is made in entering instructions, it often involves punching in wrong numbers that can increase by a factor of 10 some of the most potent medications in healthcare therapy. Among the catches was a dose of a powerful sedative set at 255 micrograms per kilogram of body weight per minute, which a nurse changed to 25 micrograms after getting a warning. The incident occurred at 3 a.m.
The clinicians used a software system called MedNet from Abbott Laboratories, which operated in conjunction with an electronic infusion therapy machine also marketed by the drug and medical device manufacturer.
Nurses, pharmacists and physicians at the medical center reached consensus on a list of 42 drugs to monitor, and ranges were set for a safe and effective dose. The results were compiled into a software "library" that operates in conjunction with machines that deliver manually programmed doses of medication.
When a pump was programmed with dosage instructions outside those predetermined ranges, the nurse loading the information was alerted to take a closer look and verify that the medication strength was appropriate for that particular patient's medical situation, McAllister says.
Most of the time the dosage was correct, reflecting the critical conditions of patients coming to the ICU from the emergency room or after major surgery. Nurses went ahead with the challenged dosages 89% of the time, according to the study.
By setting "soft" limits for powerful but highly therapeutic heart drugs, painkillers, sedatives and other IV solutions, a busy ICU operation can contend with situations where usual limits aren't broad enough, McAllister says. Up to a certain strength, the software permitted nurses to go ahead with the order, but the system also recognized hard limits that could not be exceeded. In those cases, it prevented a caregiver from programming a dosage that clearly endangered a patient, McAllister says.
Despite the high percentage of overrides, the clinicians overseeing the loading of drug ranges opted to go with a narrow minimum-maximum range where there was a high potential for toxicity and harm.
The alert system provided "a built-in redundancy without having to have a second nurse there," he says.