Background It is estimated between 15% and 40% of patients attending emergency departments could be treated in General Practice (GP). Hospitals throughout England have introduced GP services in or alongside emergency departments (GPED) to try to reduce demand. Our aim was to explore the impact of GPED and the extent to which it has become a part of routine practice.

Methods Qualitative study consisting of: non-participant observation of 142 individual clinical encounters and 413 semi-structured interviews with key stakeholders (policymakers, service leaders, ED staff, GPs, patients and carers). This was distributed across 64 NHS emergency departments in England, including 10 case sites. A coding framework was formed with patient collaborators and used with the pen portrait method for case sites. Findings were mapped onto the four main constructs of NPT: coherence, cognitive participation, collective action and reflexive monitoring.

Results There was widespread disagreement at individual, stakeholder and organisational levels regarding the purpose and potential impact of GPED (Coherence). Participants criticised policy development and implementation and viewed it as a rushed, top-down generalised solution to local problems. The amount of ‘work’ staff were willing to invest into GPED was hindered by tensions between ED and GP staff; often stemming from different attitudes to risk, perceptions of the GPED GP role and of what should be considered a GPED appropriate patient (Cognitive participation). Streaming and implementation issues (e.g. inter-professional relationships and structural support) and staffing and resource constraints affected how staff used GPED and the extent that it was embedded into routine practice. Concerns that GPED may encourage patients to attend ED and strong views around ‘appropriate’ ED attendance also influenced how staff and patients viewed GPED (Collective action). There was a lack of consensus as to whether GPED could be considered a success, due to variations in GPED model, site-specific patient mix and governance arrangements (Reflexive monitoring).

Conclusion Translating policy into practice is complex. Our findings highlight challenges of applying a national policy locally. We identified a series of success factors (e.g. inter-professional working, leadership, staffing) for introducing GPED, a number of which are commonly cited as barriers/enablers for introducing health policy in the wider literature.