We are the UK's leading organisation dedicated to realising the potential of scientists and engineers returning to research following a career break

Risk Assessment of Delayed Discharges from NHS Hospitals Using Computer Modelling

The number of people, and particularly elderly people, in Scotland admitted to hospital as emergency inpatient admissions has risen steadily over the last twenty years. Emergency admission often occurs when the patient or primary healthcare providers wait until crisis point before seeking help and, then, the only option is emergency admission. This creates two problems for NHS Scotland. First, it is often the same patients being admitted via emergency facilities on a repeated basis and this creates a high demand for these facilities. In response to this, NHS Scotland have developed SPARRA, a computer model that predicts the risk of a given patient being readmitted, or admitted in the first place, based on their clinical history. While still in development, this model is able to make reasonable predictions of likelihood of readmission based on simple assumptions of how factors in the patient’s history combine to increase the risk of readmission.

The second problem with the increasing rate of emergency admissions is that these patients do not typically have provision for follow-on care after the hospital is ready to discharge them. These patients then become delayed discharges with no route out of the acute services into which they were admitted. The key problem is that the presence of delayed discharges constrains the admission of new emergency patients, and indeed patients generally, as delayed discharges tie up valuable resources, and in particular beds.

This proposal seeks to provide a predictive model for risk assessment of delayed discharges. The programme of research will combine complex data visualisation, computational modelling and statistical modelling to create a new kind of predictive model. The model will be based on extended analysis of the factors that contribute to the risk of becoming a delayed discharge once admitted via emergency facilities. Importantly, it will be developed in partnership with NHS Tayside who will serve both as data providers and experts. The predictive power of this new modelling approach will be compared with a model of the same type as SPARRA. We hope to both create a new model for predicting the risk of delayed discharges and contribute to the ongoing development of SPARRA by indicating any performance increases associated with our approach, and will ensure that the development methodology is disseminated throughout NHS Scotland.