Presentation
PS3 - Clean Hands, Clear Outcomes: A Formal Methods Analysis of Anesthesiology Induction Hygiene
SessionPoster Session 1
DescriptionAnesthesia induction is the process of transitioning a patient from a conscious to an unconsciousness state and placing of an endotracheal tube that supplies the patient with oxygen and anesthetic. This is a complex, fast-paced process that involves numerous steps and interactions between practitioners. Infection control during anesthesia induction remains a concern due to the significant interaction between the practitioners' hands, patients' mouths, and multiple pieces of operating room equipment. The richness and complexity of these interactions can also facilitate human error (where humans skip critical hygiene steps; Biddle et al., 2018) and make it difficult to isolate where an HAIs originated. This is a serious problem, as HAIs are associated with significant morbidity and mortality and can lead to sepsis, which accounts for over 250,000 deaths annually in the United States (Revelas, 2012; Rhee et al., 2017).
To address this problem, we have developed a novel formal methods approach that uses robust mathematical techniques to model and prove properties about anesthesia induction procedures (Rose et al., 2023). In particular, it allows analysts to model the items and pairs of hands used in an induction procedure. Steps in the process are also modeled, inclusive of hand hygiene activities such as cleaning hands and/or adding or removing layers of gloves. Procedural steps account for what objects practitioners are touching and/or what objects get touched together. Nondeterministic options in the model can allow for the skipping of hygiene steps, practitioners using different hands (left of right) in single handed procedures, and inadvertently touching their hands together. Analysts can also control which objects start dirty or clean to account for lapses in between-procedure housekeeping. When run, this method can prove whether things ever become dirty when they should not. If such a violation is possible, the method produces a counterexample that shows exactly how it occurred.
In this study, we observed and modeled the anesthesia induction process based on a video published by the Auckland Academic Health Alliance (see https://www.youtube.com/watch?v=VtUN1a6zbDM). This involved two practitioners collaborating to perform the induction with a set procedure. We modeled each sequential interaction between practitioners, patients, equipment, and other environmental objects in an MS Excel spreadsheet. This was automatically translated into the input language of the Prism model checker (Kwiatkowska, et al., 2002) using custom Python software created for the method. Using Prism's symbolic model-checking and path-generating capabilities, we verified the cleanliness of critical equipment and generated traces to visualize contamination. This was done for multiple versions of the model to determine the effect of skipping hygiene steps, inadvertent hand touching, and potentially lapses of housekeeping between operations.
Our analyses of the induction procedure revealed that it resisted the spread of infection around the operating room under normative conditions, even when allowing for nondeterminism in which hand was used, and allowing for the touching of a given practitioner’s hands. However, the procedure was sensitive to the skipping of singular cleansing steps and between step cleaning/housekeeping. To illustrate this last finding, if the tray (a location where other equipment is placed during the procedure) or the pyxis plate (a plate used to hold syringes during the procedure) are dirty at the beginning of the procedure, this could result in the port (where induction agents are used to intravenously administer induction agents) becoming dirty: a potential HAI. This scenario also resulted in the tray- or plate-originating infection spreading to medication containers used in the procedure. This is a less critical outcome, but shows how infections agents could spread to unexpected places due to a lapse in housekeeping.
To address this problem, we have developed a novel formal methods approach that uses robust mathematical techniques to model and prove properties about anesthesia induction procedures (Rose et al., 2023). In particular, it allows analysts to model the items and pairs of hands used in an induction procedure. Steps in the process are also modeled, inclusive of hand hygiene activities such as cleaning hands and/or adding or removing layers of gloves. Procedural steps account for what objects practitioners are touching and/or what objects get touched together. Nondeterministic options in the model can allow for the skipping of hygiene steps, practitioners using different hands (left of right) in single handed procedures, and inadvertently touching their hands together. Analysts can also control which objects start dirty or clean to account for lapses in between-procedure housekeeping. When run, this method can prove whether things ever become dirty when they should not. If such a violation is possible, the method produces a counterexample that shows exactly how it occurred.
In this study, we observed and modeled the anesthesia induction process based on a video published by the Auckland Academic Health Alliance (see https://www.youtube.com/watch?v=VtUN1a6zbDM). This involved two practitioners collaborating to perform the induction with a set procedure. We modeled each sequential interaction between practitioners, patients, equipment, and other environmental objects in an MS Excel spreadsheet. This was automatically translated into the input language of the Prism model checker (Kwiatkowska, et al., 2002) using custom Python software created for the method. Using Prism's symbolic model-checking and path-generating capabilities, we verified the cleanliness of critical equipment and generated traces to visualize contamination. This was done for multiple versions of the model to determine the effect of skipping hygiene steps, inadvertent hand touching, and potentially lapses of housekeeping between operations.
Our analyses of the induction procedure revealed that it resisted the spread of infection around the operating room under normative conditions, even when allowing for nondeterminism in which hand was used, and allowing for the touching of a given practitioner’s hands. However, the procedure was sensitive to the skipping of singular cleansing steps and between step cleaning/housekeeping. To illustrate this last finding, if the tray (a location where other equipment is placed during the procedure) or the pyxis plate (a plate used to hold syringes during the procedure) are dirty at the beginning of the procedure, this could result in the port (where induction agents are used to intravenously administer induction agents) becoming dirty: a potential HAI. This scenario also resulted in the tray- or plate-originating infection spreading to medication containers used in the procedure. This is a less critical outcome, but shows how infections agents could spread to unexpected places due to a lapse in housekeeping.
Authors
Event Type
Poster Presentation
TimeMonday, March 314:45pm - 6:15pm EDT
LocationFrontenac Foyer
Digital Health (DH)
Simulation and Education (SE)
Hospital Environments (HE)
Medical and Drug Delivery Devices (MDD)
Patient Safety and Research Initiatives (PS)