Examining Patient Safety Events Using the Behaviour Change Wheel: A Cross-Sectional Analysis

Precursor-level safety events (PSEs) pose greater patient risk than no-harm events but are not as severe as serious safety events. Despite their potential for harm, the underlying determinants associated with PSEs are poorly understood. This study aimed to use a behavior change framework to understand the underlying determinants of PSEs and whether associated action items aligned with the behavior.

This cross-sectional study took place in a maternal/pediatric hospital. A total of 58 prerecorded PSEs were analyzed using the Behaviour Change Wheel (BCW); a behavioral framework that identifies sources of behavior and proposes intervention types that address said behavior. Researchers and clinicians independently coded each PSE's underlying determinant and action items using the relevant components of the BCW. The types and frequency of underlying behavioral determinants and intervention types for each PSE were documented. A matrix, based on the BCW, reflected how often the underlying behavior aligned with the corresponding action item.

Of the 58 PSEs, six behavioral determinants and seven intervention types were identified. Environmental context/resources was the behavioral determinant coded most often (25.4%); education was the most common intervention type (45.8%). Several underlying determinants (24.6%) and action items (8.3%) received no code due to limited information. Based on the BCW matrix, 34.2% of behavioral determinants were addressed with interventions that would target the underlying behavior, while 37.8% did not align, and 28.1% could not be coded due to missing behavioral information.

This study identified poor alignment between types of interventions and underlying determinants in more than one third of analyzed PSEs. This included using educational interventions in about 50% of events, despite this type of intervention being ineffective for most of the coded behaviors. Further, alignment of many safety events could not be determined due to limited reported information. This highlights a need to design more systematic, behavior-informed approaches to reporting PSEs and identifying interventions to effectively change behavior.