Developing a logic model and dark logic model for hospital at home (virtual wards): a qualitative study

Abstract Introduction Hospital at home (HaH), also known as virtual wards in the UK, are innovative models allowing patients to get hospital-level care at home, supported with the use of technology and wider healthcare team. There has been rapid growth of these models in the UK, with an ambition to scale-up and standardise models of care. However, evaluations varied lacking evidence on their implications in practice and patient safety. Logic models are often used to synthesise diverse evidence or evaluate complex interventions exploring intended outcomes and causal mechanisms. Aim This study aimed to develop a logic model (intended outcomes) and dark logic model (unintended outcomes) to systematically review and illustrate existing evidence, with input from key stakeholders. This would support priority setting for further research, implementation of services or quality improvement. Methods This study involved triangulation of three workstreams. Firstly, a systematic search for published literature was undertaken involving document analysis using a pre-piloted data collection form, subsequently used to form initial draft logic models. Secondly, key informants, participated in online semi-structured interviews to review the draft logic models. Key informants were professionals involved in HaH, recruited through known contacts, (purposive sampling) and snowballing. Interviews were transcribed and analysed by content analysis, using hybrid inductive and deductive coding, to refine the logic models. Thirdly, an online focus group with consenting participants from the previous phase was undertaken, audio-recorded and analysed, to gain consensus on the final logic models. Results Document analysis using a range of documents (n = 30) up to March 2024, included national guidance, secondary literature, primary studies and virtual ward case-studies. Draft logic models were presented to key informants (n = 12) who had varying levels of involvement in HaH or virtual ward services, including pharmacists (n = 7), consultant geriatrician (n = 1), advanced nurse practitioners (n = 2), clinical services manager (n = 1), and an advanced physiotherapist practitioner (n = 1), from diverse geographical locations in England. Four participants took part in the focus group. Valuable stakeholder insight helped refine the logic models, highlighting key areas for sustainability of services such as clinical and organisational ‘buy-in,’ clear communication and building effective cross-sector relationships within networks. This includes effective resourcing; like technology, workforce, integration of systems and digital infrastructure. Funding to sustain these services and recognising the importance of medicines provisions and governance were key themes highlighted. Conclusion This study provides clear visual representation of intended and unintended outcomes of HaH (virtual wards) in England, and key contributing factors, strengthened by a review of the literature and stakeholder involvement. Despite the modest sample size, this work can support prioritising future research or improving current services before scaling-up. This study highlights clear implementation, workforce and operational challenges that need to be explored to be able to deliver acute care outside of usual hospital settings. Consideration of how to maximise the potential of current resources both from a workforce and operational perspective needs to be prioritised to enable the foundations for a future of seamless personalised and holistic care.