Evidence and Insights (Assumptions)
Note Collation of insights exercise available internally as some of these leverage internal data, modelling or stakeholder insights. a An overview of the considerations to gather such insights is given below.
Insights supported the project both in terms of model context and scope; model formulation; parametrisation; calibration; validation and relevant use cases.
In the previous section the used data assets and data publications were mentioned. Other forms of insight, whether based off of such data assets or alongside them, include academic and gray literature, internal analysis and modelling pieces, operational procedures and subject-matter-expert knowledge and judgement.
On one hand, these existing insights beyond core data assets can help direct the exploratory analysis and the model specification more efficiently. For instance, with regards to already known patterns or sources of variation in demand, flow, supply, job cycle time or ED characteristics.
Further, while the model focusses on ambulance operational performance, such insights help to contextualise the work in terms of other key outcomes like clinical outcomes and staff experience, other healthcare settings, expected inter-trust variation or the macroenvironment (e.g. pandemic).
Crucially also, some mechanisms such as dynamic escalations and behavioural changes in response to pressure may be hard to define and parametrise from existing data. These would need evidence-based assumptions.
This includes:
- ED queue prioritisation
- Patient behavioural - balking, reneging, redirection to walk-ins
- Double Services Ambulance (DCA) fleet available for step-up
- Use of interventions (e.g. cohorting)
- Impact of pressure on allocation (conveyance, dispatch, no-send)
- ED queue prioritisation
- ED escalations (capacity, early discharge) when under pressure
- Other known standard operating procedures and/or behavioural changes when under pressure
Where addressed, this could be either based on:
- Empirical emerging evidence, data-driven historic empirical relations (e.g. conveyance % as a function of Cat2-response time or another pressure-indicative KPI).
- Written procedures under normal operations and different levels of escalation.
- SME feedback, namely capturing anectodal evidence or information from local data sources.
- Modeller baseline assumption, which could be improved namely via calibration.
Some useful reads
A small subset of useful background reads is given below:
NHS England, Delivery plan for recovering urgent and emergency care services (January 2023)
AACE, Delayed hospital handovers: impact assessment of patient harm (November 2021)