Safety-Critical Healthcare Technology Design

Preventable medical errors are a severe problem in healthcare, causing over 400,000 deaths per year in US hospitals alone. In acute care, the branch of medicine encompassing the emergency department (ED) and intensive care units (ICU), error rates may be higher to due low situational awareness among clinicians engaged in safety-critical procedures. Here, clinicians often work together in high-stress scenarios that require strong team communication and coordination. The inability to work together effectively can lead to adverse patient harm, or worse, patient death.

Clinical teams may rely on both paper and computer-based tools to help them coordinate better. Paper-based cognitive aids are one such tool that teams may use to reference information and tasks for a situation to prevent possible errors. Likewise, tools like electronic health records might help the team document information during safety-critical scenarios or access critical patient information. However, as health information technology (HIT) becomes more commonplace in healthcare, it is becoming clear that the design does not always meet the needs of clinicians in safety-critical contexts.

This is a critical problem, because poorly designed HIT can lead to patient harm or death. Attending to the ways new HIT impacts the clinical workflow will help designers make improvements to their software before compromising patient safety. Most HIT takes a 'shoot first, ask questions later' approach to design, where problems are only corrected after they manifest. They are not built with clinicians' workflow and patient safety at the forefront. Patient lives are worth the time to design effective HIT that supports clinicians' individual needs.

In order to design more effective HIT, designers must look beyond a single hospital or department. Healthcare is not a ``one size fits all'' domain, and each care setting differs in its culture, environment, and practices. Technology must be adaptable to meet these differences, or it is unlikely to be effective in practice.

This dissertation explores this concept of design using a multi-institutional approach in a collaborative HIT tool we call Visual TASK (Team Awareness and Shared Knowledge). The goal of Visual TASK is to improve clinicians' situational awareness and cognitive workload in safety-critical scenarios, such as resuscitation.

In this dissertation, I outline the ethnographic fieldwork, design, and evaluation of Visual TASK studied in the context of three hospitals. Our ethnographic study focuses on understanding some of the environmental and resource constraints facing clinicians across three hospitals. I then outline how we used these findings in the design of Visual TASK, and walk through an evaluation of the tool at two US hospitals.

My work has multiple contributions in the fields of human-computer interaction, health informatics, and safety-critical technology design. First, I identified human-factor problems clinicians faced across hospitals, including different spacing constraints and team management. Understanding these differences are important for designers, so that they can create technology which can accommodate constraints in the work environment.

Second, I found a number common problems with electronic health record systems (EHRs) and cognitive aid use in safety-critical environments. This includes problems with technology usability and workarounds clinicians employ to manage team size and roles. For example, we found at several institutions that EHRs were so poorly designed that clinicians used paper towels as a workaround. Third, I identified problems with clinicians fixating and becoming distracted by cognitive aids in safety-critical scenarios. Clinicians often over-rely on cognitive aids, to the point where it directly interferes with how they deliver care and their situational awareness. This can impose challenges for the rest of the team.

Fourth, I designed and developed Visual TASK to support team situational awareness with the capability to adapt to end-user needs. Visual TASK provides a shared resource display for the entire team to guide them through a procedure. It provides detail of each individual task for each role in a safety-critical scenario, so that all participants have the information they need to do their jobs. It also provides a history chart of completed, and a list of upcoming events, so participants can reference the team's status.

Fifth, I found that gesture-based technology is unusable in safety-critical environments, and may cause patient harm. Gesture-based tools like the Kinect are unfit for safety-critical environments, such as the intensive care unit, due to possibilities of occlusion and space constraints. Using gesture-based input modalities may only create new problems for clinicians, such as usability problems and distracting them from the situation.

Finally, I evaluated Visual TASK with clinical teams at two US hospitals. I found that Visual TASK can significantly improve situational awareness and reduce cognitive workload for team members. This is an important finding, because it shows that collaborative technology may be able to help reduce preventable medical errors that are caused due to a lack of situational awareness and high-cognitive workload.

My research informs the design of future safety-critical technology. Technology must be adaptable to clinicians' needs, because poorly designed technology will disrupt clinicians' workflow and can lead to patient death. Technology designers must take into consideration how clinicians cognitively process tasks, coordinate as a team, and understand their workflow to create effective solutions. This can help prevent some of the severe problems that compromise patient safety and help reduce the number of preventable medical errors that occur today. Patient lives are worth the time investment in fully understanding the magnitude new technology has on the clinical workflow.