Applying computer modelling, simulation and socio-technical systems analysis to improve NHS dementia care outcomes

Description

Around 40% of people over 75 in hospital are likely to have dementia (Sampson el al., 2010). People with dementia have complex needs which are often poorly met by hospital systems, process and environment (Porock et al., 2015; Crowther et al, 2017). They also have worse outcomes than people without dementia including; longer admissions, and increased falls, fractures and mortality (Dewing & Dijk, 2016) representing a significant cost to the NHS and society. It is poorly understood how hospital systems, processes and environment interact to affect outcomes.

Improving hospital dementia care outcomes is a national research priority (DoH 2015). This is not just a ‘medical’ problem it is a quintessentially management problem too - requiring understanding of the interrelated consequences that concurrent, systemic factors (e.g., care pathways, training programmes, hospital ward conditions) have, as well as the medical conditions themselves. Research has demonstrated that socio-technical systems (STS) approaches can be applied to understand causal chains leading to medical care outcomes (e.g., Clegg et al, 2014; Hughes et al., 2017) and to understand the complex interplay between human, technical and management factors involved. The complexity involved is a challenge to traditional research methods and linear thinking (Davis et al, 2019; Hughes et al, 2012) and one that is perpetuated by the ethical challenges involved in researching such vulnerable patients.

Computerised modelling, specifically agent-based modelling and simulation (ABMS), offers a way of exploring how such STS variables interact to affect outcomes, in a safe and controllable environment. Computer simulations operate within an infrastructure, representing the features and parameters of the simulated environment (e.g., a ward containing a specified range and number of facilities, as well as patients and staff). Individual ‘agents’ are included to represent different types of people (e.g., patients, doctors, relatives), or artefacts (e.g., beds, chairs, medical stations). Agents can be asked to work through specified tasks (e.g., administering medications). In each case, the tasks could involve movement, happen at particular time intervals, and could take a given amount of time. The simulations then run over a period of time, with the behaviour of these agents generated in alignment with simple ‘rules’ (e.g., ‘if X happens, then do Y’). Additionally, agents can be given ‘attributes’ (e.g., dementia severity), and rules can be developed on this basis. For instance, if we know that patients with a particular type of dementia display a particular behaviour pattern 70% of the time, we could incorporate this level of probability into the rules (e.g., it might take longer to care for them). Simulations then generate actions in real time, in accordance with probabilistic instructions (i.e., on 70% of occasions X will happen, and on 30% of occasions Y will happen).

In healthcare research computerised simulations are novel, yet offer a range of potential benefits for understanding and improving dementia care.

In summary, this project aims to:

1. Identify socio-technical systems (STS) factors influencing hospital based dementia care outcomes;
2. Develop a simulation prototype that incorporates these factors, to establish the utility of computer modelling and simulation for examining hospital based dementia care;
3. Use this simulation model to test (in silico) the impact of changes to physical design, management approaches or clinical practices on dementia care outcomes, thus demonstrating proof of concept.

Impact

The potential impact of this research is far reaching:

1. Data collection will use STS Analysis and Scenarios Planning Approaches which have already demonstrated benefit to the NHS via the ESRC funded MALT project (see Hughes et al., 2017), and the research underpinning the model development will draw together data sources on a range of STS aspects. This alone will have the potential to impact care outcomes, management and processes within dementia wards.
2. There are a lack of tools available that enable researchers to study complex, wicked problems in ways that demonstrate useable, practical outputs (see Davis et al, 2019). Simulation can enable researchers to start, stop, pause and rewind to see how events emerge, which can help explore tipping points and emergence, improving understanding of phenomena. The simulation model will demonstrate proof of concept in this domain.
3. Simulating complex healthcare challenges has the potential to open up a whole new way of researching for researchers, Trusts and patients. By enabling the trialling of interventions ‘off-line’, and then the exploration of related cost, safety, and time-savings, such approaches will have the potential to contribute meaningfully to scenario planning activities within Trusts.
4. Once an initial proof of concept has been developed, the sophistication of the models can become more complex. The aspiration is that over time, initial investment in this approach will enable such models to become more commonplace.