Our business intelligence (BI) team was engaged by BaNES, Swindon and Wiltshire Clinical Commissioning Group (CCG) to develop a system-wide demand and capacity model to support winter planning. Also, to highlight capacity risks across acute and community providers.
There was a need to forecast demand for services based on recent demand. A modelling tool that could embed a consistent approach across the system, and via the modelling of multiple scenarios help identify risks, capacity gaps, and pinch points for particular times of the year, types of bed, and the provider was needed. Plus, a model which supported winter planning.
We carried out the work in a phased approach:
- Pre-initialisation - discussions with stakeholders on likely data requirements, data availability, and access to data
- Phase 1 - agreement of scope and approach with key stakeholders
- Phase 2 - segmentation of population and length of stay (LoS) projection - i.e. for each patient grouping, based on patient/admission characteristics, what would be the expected LoS based on historical data.
- Phase 3 - development of an Excel-based model to forecast bed usage, discharges, and community demand, including testing, presentation to stakeholders, and documentation/handover
An innovative aspect of our work was using Decision Trees in the R statistical software package to segment baseline admissions data into homogenous cohorts, using LoS as the output variable to define the groups.
Our work was carried out in the summer of 2021 and presented at the Urgent Care & Flow (UC&F) Board in September 2021.
- The model achieved the system's objective of providing a snapshot of projected demand and capacity within the Urgent Care system over a 6–9-month period for several scenarios, highlighting risks and mitigations.
- The common modelling approach adopted across the system provided a consistent and comparable picture, facilitating system-wide discussion and actions.
- The underpinning methodology enabled the model to be easily updated. This answers the other stated aim for the project of enabling regular updates i.e. using the model with updated data to allow a comparison of the latest demand and capacity data against the original projections.
- The work took a partnership approach – steered by critical members of the UC&F Board and localities but delivered between the core modelling team and BI representatives from across partner organisations. We played a crucial role in this approach, leading discussions across BI teams and developing templates to help ensure consistency of source data across organisations.
- The model identified a range of bed gaps across different scenarios for each provider, including statistical tolerance levels to reflect the associated levels of uncertainty. The gaps helped identify the potential levels of mitigation needed, against which projected winter scheme impacts could be compared.
The support from SCW was integral in providing our system with a robust approach to planning and managing through a challenging inter period. Importantly, the model gave partners within the system a trusted evidence base around which to collaborate and agree - Sam Wheeler, Assistant Director Business Intelligence - System Architecture and Transformation, NHS BSW CCG