Implementing after action review in an Irish tertiary hospital: a mixed-methods evaluation of safety culture and second victim experience and support

BACKGROUND: Evidence for After Action Reviews (AAR), a non-hierarchical debriefing approach, is limited. We explored AAR implementation and measured safety culture and second victim experience and support, before and after introduction of AAR at an Irish hospital. METHODS: A mixed-methods study which coincided with the COVID-19 pandemic and a health system cyber-attack was conducted. Hospital selected staff were trained as AAR facilitators using a simulation-based programme (July-Sept 2021). The intervention was the hospital introduction of AAR over 12-months (Sept 2021-Aug 2022). Before training, the Hospital Survey on Patient Safety 2.0 (HSOPS 2.0) and Second Victim Experience and Support Tool (SVEST) were distributed to all staff (N = 586), and repeated at 12-months with additional questions on AAR awareness, frequency and duration. At 6-months, focus groups with trained facilitators explored implementation enablers and barriers. Costs were estimated. Quantitative analysis used chi-squared, Fisher's exact and t-tests to compare pre-/post- responses. Qualitative data was analysed using the Capability, Opportunity, Motivation - Behaviour (COM-B) model. Findings were integrated via triangulation, and the level of implementation was categorised using Roger's theory of the diffusion of innovations. RESULTS: Nine per cent (n = 50) of staff were trained as AAR facilitators. Survey response rates were 33% pre- and 16% post- AAR implementation. After 12-months, 37% of staff were aware of AAR and 10% had participated in one, at €48 average cost. Approximately, 39 AARs were conducted. The ratio of informal to formal AARs was 2.6:1. No change in overall safety culture occurred despite a pandemic and cyber-attack. Response to error improved. Triangulation of results confirmed improvement in response to error was associated with staff awareness and participation in AAR. Second victim experience and support showed no change in psychological and physical distress, yet professional self-efficacy, perceived organisational support and turnover intentions worsened, with no observed association with AAR. CONCLUSIONS: Simulation-based training and national-level policy supported early adoption of AAR. Implementation of AAR was associated with improved response to error at the hospital, but not second victim experiences.