Using realist evaluation to test a programme theory of ‘embeddedness’ in a palliative care Integrated Care Pathway in primary care

Background: A primary care locality developed and implemented an Integrated Care Pathway (ICP) for those with life limiting illnesses in fourteen GP practices. The ICP focuses on identifying patients early, regardless of disease type, and used proactive and patient centred interventions to plan for a good death. To date, few studies have examined how and when ICPs work best in the context of palliative care. This presentation will focus on one particular initial programme theory on embeddedness, and discuss how it was refined throughout the evaluation. Design: A realist evaluation of the ICP was conducted, which aimed to highlight what in the ICP worked, for whom and in which circumstances. Data collection included focus groups with ICP implementers, GP practice data and palliative care quality visit observations. GP practice data was analysed using cluster analyses which can be used to discover relatively homogeneous groups of GP practices based on selected characteristics. Results: The initial programme theory stated that the number of people who die in their chosen location (outcome, O), will depend on the GP practice (context, C) they are registered with and how embedded the ICP is (mechanism, M) within that practice, as indicated by the number of interventions used per patient. Three GP practices in particular demonstrated high performance in terms of intervention use and preferred place of death. The CMO configurations in these practices were compared with those in a lower performing practice in order to refine the programme theory. This highlighted that desirable outcomes including preferred place of death and intervention use (O), resulted from strong peer support and leadership delivered by a champion or opinion lead (M) taking place in a context where there was a shared nursing team on site. Discussion: Overall, the ICP was thought of as 'successful' as it statistically increased intervention use across all practices (p