Personalising high-risk perioperative care with dynamic models of clinical trajectory
Major elective surgery carries a mortality of 3.6% in the UK [Pearse 2012] with non-fatal post-operative complications increasing length of stay and leading to long-term reductions quality of life as well as reducing long-term survival in affected patients.
This burden falls disproportionately on the elderly: consequent socioeconomic costs and ageing demographics make improving post-operative outcomes a significant public health consideration. Stratification of patients for post-operative intensive care has received much attention, but many perioperative complications occur later.
Whilst early warning scores help to identify patients who are deteriorating in the resource-limited ward setting, there is a need for a system that can identify patients at risk of complications so that clinical contact and investigations can be prioritised prospectively. My project will use deidentified perioperative episodes (CUH has >20,000 historical high-risk patient episodes) to build dynamic prediction models for perioperative complications by day.
In this project, I will establish models to better describe and predict from typical patient profiles classes and attempt to better understand the impact of complications on length of stay. These will be used to create better dynamic risk predictions to guide resource allocation and inform shared decision making.
What will help to make digital transformation a reality in the NHS?
The key to digital transformation is implementation. The key to implementation is broad buy-in. The key to buy-in is education and understanding.