Identifying and Reducing Insulin Errors in the Simulated Military Critical Care Air Transport Environment: A Human Factors Approach.

Abstract

Introduction: During high-fidelity simulations in the Critical Care Air Transport (CCAT) Advanced course, we identified a high frequency of insulin medication errors and sought strategies to reduce them using a human factors approach.

Materials and methods: Of 169 eligible CCAT simulations, 22 were randomly selected for retrospective audio-video review to establish a baseline frequency of insulin medication errors. Using the Human Factors Analysis Classification System, dosing errors, defined as a physician ordering an inappropriate dose, were categorized as decision-based; administration errors, defined as a clinician preparing and administering a dose different than ordered, were categorized as skill-based. Next, 3 a priori interventions were developed to decrease the frequency of insulin medication errors, and these were grouped into 2 study arms. Arm 1 included a didactic session reviewing a sliding-scale insulin (SSI) dosing protocol and a hands-on exercise requiring all CCAT teams to practice preparing 10 units of insulin including a 2-person check. Arm 2 contained arm 1 interventions and added an SSI cognitive aid available to students during simulation. Frequency and type of insulin medication errors were collected for both arms with 93 simulations for arm 1 (January-August 2021) and 139 for arm 2 (August 2021-July 2022). The frequency of decision-based and skill-based errors was compared across control and intervention arms.

Results: Baseline insulin medication error rates were as follows: decision-based error occurred in 6/22 (27.3%) simulations and skill-based error occurred in 6/22 (27.3%). Five of the 6 skill-based errors resulted in administration of a 10-fold higher dose than ordered. The post-intervention decision-based error rates were 9/93 (9.7%) and 23/139 (2.2%), respectively, for arms 1 and 2. Compared to baseline error rates, both arm 1 (P = .04) and arm 2 (P < .001) had a significantly lower rate of decision-based errors. Additionally, arm 2 had a significantly lower decision-based error rate compared to arm 1 (P = .015). For skill-based preparation errors, 1/93 (1.1%) occurred in arm 1 and 4/139 (2.9%) occurred in arm 2. Compared to baseline, this represents a significant decrease in skill-based error in both arm 1 (P < .001) and arm 2 (P < .001). There were no significant differences in skill-based error between arms 1 and 2.

Conclusions: This study demonstrates the value of descriptive error analysis during high-fidelity simulation using audio-video review and effective risk mitigation using training and cognitive aids to reduce medication errors in CCAT. As demonstrated by post-intervention observations, a human factors approach successfully reduced decision-based error by using didactic training and cognitive aids and reduced skill-based error using hands-on training. We recommend the development of a Clinical Practice Guideline including an SSI protocol, guidelines for a 2-person check, and a cognitive aid for implementation with deployed CCAT teams. Furthermore, hands-on training for insulin preparation and administration should be incorporated into home station sustainment training to reduced medication errors in the operational environment.