Reducing Infusion Pump-Related Medication Errors: The Use of a “Mental Model” to Guide Drug Library Development and Implementation
Event Type
Poster Presentation
TimeThursday, April 152:00pm - 3:00pm EDT
LocationMedical and Drug Delivery Devices
Adverse drug events are the single leading cause of medical injuries1, and medication infusions present a great potential for harm, particularly in the field of pediatric anesthesia. While standard infusion pumps allow manual programming of rate (or dose) over time, “smart” infusion pumps incorporate an electronic library of medications along with medication-specific “soft” and “hard” limits in an effort to decrease medication errors due to programming errors. The use of this medication library or Drug Error Reduction System (DERS) may be bypassed by the user, negating its potential benefit, if it is perceived as difficult or time-consuming to use. Pump programming errors commonly include decimal point errors/power of 10 misprogramming (slip), mental math error, and memory lapse (mcg/kg/min v mcg/kg/h). Literature suggests that smart pumps do not eliminate all medication errors, and compliance with DERS utilization is essential2. We used principles of user-centered design to develop and implement a novel medication library for use in the pediatric operating room (OR) in order to increase the likelihood of achieving high levels of compliance with DERS utilization and reduction in pump- programming related medication errors.
Infusion pumps without a drug library are designed for maximum flexibility so that they may be adapted for use in a wide variety of clinical settings1. Even at a children’s hospital, the same pump that delivers vasopressors to a 600-gram preterm infant in the neonatal intensive care unit may be used to care for a 150-kilogram teenager in the OR, and therefore must be designed without rate limitations in place. Unfortunately, this level of adaptability results in increased opportunities for error, as pumps must be manually programmed with drug concentration, patient weight, and delivery units - often in dynamic and time-pressured environments. In such circumstances, one wrong button push or decimal point error may be devastating. Inadvertent substitution of micrograms for milligrams, for example, may result in a 1,000-fold increase in drug delivery. Therefore, dose error reduction software has the potential to minimize the opportunities for error by utilizing weight-based rate and dose limits.
In order to reduce the likelihood of pump-related medication errors, our institution developed a drug library customized for care of pediatric patients in September 2019. Initial training sessions with the pediatric anesthesiology team revealed high levels of frustration with the complexity and difficulty of standard alphabetical categories in which the drug library was to be accessed. As a result of conversations with team members who had undergone the initial training session, it was clear that the most frustrating aspect of drug selection was determining where in the alphabetical categories the first letter of the medicine would fall. A second frustration was the number of button pushes required to scroll through potential medications prior to selection.
We used this feedback to re-develop the initial menu from one of alphabetical categories to one of medication effect. One anesthesiologist described frustration with having to sing the “alphabet song” to determine where a medication would be located within the pump’s drug library. This adaptation to the “mental model” of the anesthesiologist considers the fact that anesthesia practitioners select and administer medications based on their desired clinical effect, and not their position in the alphabet. We obtained data regarding the frequency of use of infusion medications during the preceding 12 months to prioritize inclusion in the profile. The medications that were used most were included in the “Frequently Used” tab. One or two button pushes within the anesthesia profile displays the eight medications used most frequently by our anesthesia team, accounting for more than 75% of all infusion medications used in the operating room. If they are not included in the “Frequently Used” category, medications are categorized by their desired clinical effect. Categories include “Vasopressors”, “Anti-Hypertensives”, “Analgesics” and “Sedatives”, among others. “Soft” limits were designed for situations in which a clinical indication outside of the limits may exist and there was limited risk to the patient. “Hard” limits were designed for drugs in which incorrect pump programming would result in a serious safety event; the user is unable to override the limit. For some medications, these hard limits were specifically set in order to alarm in response to probable decimal point errors, a common source of pump programming error within pediatric medicine. The pump was also designed to require double entry of the patient's weight in order to reduce the likelihood of errors related to incorrect entry of patient weight.
Usability testing of the medication library organized by clinical effect revealed a significant improvement in user satisfaction and willingness to use the DERS system. Users were given a list of six different cases, each of which required multiple infusions to be started. Anesthesia providers worked independently to program the medications with a trained pharmacist and anesthesiologist available to answer questions or address any concerns. Feedback was obtained from anesthesia providers which confirmed our new approach to drug library organization and to refine specific settings for “soft” and “hard” infusion limits.
Results and Discussion
Interestingly, in reviewing reported data from our institution’s Quality and Safety Reporting System, there was not a significant difference in overall reporting anesthesia-related medication errors that reached the patient and required intervention or prolonged monitoring. It is important to note that this reporting system is voluntary and likely represents a small proportion of actual medication errors. While our institution has a strong and stable reporting rate, these data are limited in their ability to detect a meaningful impact on medication errors.
Querying the infusion pumps themselves tell a different story, however. For Q4 of 2020, for example, out of a total of 7850 infusions, 121 times (1.53%) a soft limit caused the user to change their pump programming, and 137 times (1.66%) a hard limit prevented the user from continuing. In reviewing these events, several 10-fold to 1000-fold dosing errors were found and prevented, including an attempted 1 mg/kg bolus of remifentanil instead of 1 µg/kg (1000-fold error), an attempted 1mg/kg bolus of phenylephrine instead of 1 µg/kg (1000-fold error), a milrinone infusion inadvertently programmed to deliver 50 mcg/kg/min instead of the desired 0.5 mcg/kg/min, among numerous others.
Anesthesiology’s average monthly compliance (83.5%, median 88.8%) with the drug library is significantly higher than the hospital average of 67%, thought to be secondary to the improved usability of the library. We are in the process of quantifying user satisfaction with the structure and content of the anesthesia medication library and the perceived impact on the likelihood of making an inadvertent medication error.
Drug library design and implementation that is user centered and based on the mental model of medication selection of the anesthesiology provider is an effective means of averting significant medication errors. Thoughtful, iterative, user-centered design has the ability to result in high levels of compliance with DERS utilization and a low incidence of alarm limit override.

1. Reves JG. “Smart Pump” Technology Reduces Errors. APSF Newsletter. Spring 2003: 18(1).
2. Ohashi K, et al. Benefits and risks of using smart pumps to reduce medication error rates: a systematic review. Drug Safety. 2014: 37(12): 1011-20.