CHI+MED logo banner

Patient Safety with Interactive Medical Devices:
a multi-disciplinary symposium

Reading, UK; 24th–25th June 2015

Abstract of a presentation

View presentation slides

Reducing intravenous therapy errors: Testing the usability of a system for central control of multiple infusion pumps

Fokie Cnossen, University of Groningen, Institute of Artificial Intelligence and Cognitive Engineering

Frank Doesburg & Maarten Nijsten, University of Groningen, University Medical Centre

Most errors in the intensive care unit (ICU) are related to intravenous (IV) therapy. Infusion pumps are notoriously hard to operate, and as most ICU patients receive several IV-therapies simultaneously, each with their own administration regime, scheduling medication imposes a high cognitive workload on ICU nurses. A second contributing factor is the considerable number of lumens that run from the IV-pumps to the patient, complicating connecting infusion syringes to the correct IV-connection sites on the patient side.

Proposed system
To tackle the problem of “IV spaghetti” and to optimize operation and scheduling of multiple infusion pumps, we propose a new control system that for intermittent medication determines when which medicine has to be started and ended. The system takes medication schedules and dosages, other fluids, and potential incompatibilities into account. The system will include a display of information from the IV pumps, such as time left until syringe is empty, dosage of fluids, etcetera. The system can also manage a “relay” mode: gradual emptying of a syringe while a new syringe takes over, ensuring constant blood levels.

Another aim of the proposed system is to reduce the number of lumens by using the same lumen for different medications by including a pharmacologically neutral buffer fluid between incompatible drugs. We call this novel administration technique Multiplex infusion. In addition to reducing infection risks, the proposed integrated system should reduce nurse’s cognitive workload and reduce human errors.

Study 1: Medication administration assistant
As a proof of concept, we first developed a user interface for controlling multiple infusion pumps that actively assists nurses during error-prone tasks such as modifying flow rates and responding to alarms. We performed a usability study on a computer simulation of this user interface, and compared it with a simulation of conventional manual control. 16 ICU nurses participated in the within-subject experiment. Participants were asked to perform a number of everyday tasks. These included replacing an almost empty syringe in response to an alarm, changing the administration rate and dealing with interruptions. Results showed that with similar task completion times for both interfaces, with integrated control participants responded faster to alarm events, while their overall error rate was lower.

Study 2: Tubing assistant
At present we are designing an interactive instructional module to assist nurses in connecting tubes from the infusion pumps to the patients. This is an essential part of our system, as the system will have to be fully aware of how the pumps are connected to the patient. We will assess the effectiveness and usability of the module soon.

The results from our first study are quite promising: the number of medication errors was significantly lower with our system compared to the conventional system, while nurses were also faster in dealing with alarms. We therefore believe that this system will increase patient safety in the ICU.