Bridging the gap between pragmatic intervention design and theory: using behavioural science tools to modify an existing quality improvement programme to implement "Sepsis Six".

BACKGROUND: Sepsis has a mortality rate of 40 %, which can be halved if the evidence-based "Sepsis Six" care bundle is implemented within 1 h. UK audit shows low implementation rates. Interventions to improve this have had minimal effects. Quality improvement programmes could be further developed by using theoretical frameworks (Theoretical Domains Framework (TDF)) to modify existing interventions by identifying influences on clinical behaviour and selecting appropriate content. The aim of this study was to illustrate using this process to modify an intervention designed using plan-do-study-act (P-D-S-A) cycles that had achieved partial success in improving Sepsis Six implementation in one hospital.

METHODS: Factors influencing implementation were investigated using the TDF to analyse interviews with 34 health professionals. The nursing team who developed and facilitated the intervention used the data to select modifications using the Behaviour Change Technique (BCT) Taxonomy (v1) and the APEASE criteria: affordability, practicability, effectiveness, acceptability, safety and equity.

RESULTS: Five themes were identified as influencing implementation and guided intervention modification. These were:(1) "knowing what to do and why" (TDF domains knowledge, social/professional role and identity); (2) "risks and benefits" (beliefs about consequences), e.g. fear of harming patients through fluid overload acting as a barrier to implementation versus belief in the bundle's effectiveness acting as a lever to implementation; (3) "working together" (social influences, social/professional role and identity), e.g. team collaboration acting as a lever versus doctor/nurse conflict acting as a barrier; (4) "empowerment and support" (beliefs about capabilities, social/professional role and identity, behavioural regulation, social influences), e.g. involving staff in intervention development acting as a lever versus lack of confidence to challenge colleagues' decisions not to implement acting as a barrier; (5) "staffing levels" (environmental context and resources), e.g. shortages of doctors at night preventing implementation. The modified intervention included six new BCTs and consisted of two additional components (Sepsis Six training for the Hospital at Night Co-ordinator; a partnership agreement endorsing engagement of all clinical staff and permitting collegial challenge) and modifications to two existing components (staff education sessions; documents and materials).

CONCLUSIONS: This work demonstrates the feasibility of the TDF and BCT Taxonomy (v1) for developing an existing quality improvement intervention. The tools are compatible with the pragmatic P-D-S-A cycle approach generally used in quality improvement work.