Hui Pheng Teoh

MSc in Data Science for Health and Social Care, The Usher Institute

Supervised with K.Banas

Identifying and explaining key trends in Scottish accident and emergency attendances and waiting times.

Accident and emergency (A&E) attendances and waiting times have been the focus of media attention in recent years. Prolonged A&E waiting times can have detrimental effects on patient outcomes. This dissertation explores key trends in Scottish A&E activity and investigates the influence of patient demographics, time and location on A&E attendances and waiting times. The insights gained may contribute to A&E service improvements by highlighting areas of high demand and poor performance. Line graphs, heat maps and choropleth maps were created to illustrate data trends, and a generalised linear model (GLM) based on Poisson regression was built for forecasting purposes to guide planning and service development. Differences were observed in A&E attendance volumes depending on department type (emergency department or minor injuries unit) and health board. After accounting for population size, Scottish health boards were found to have similar A&E attendance rates. Comparisons between 2012 and 2022 data revealed that all Scottish health boards had recorded a decrease in the percentage of patients seen within 4 hours, despite reductions in attendance rates in most health boards during this period. A&E attendances were highest in March through to August, with variations in periods of high service demand (time of the day and day of the week) depending on department type. The GLM results aligned with previous findings regarding the direct relationship between deprivation and the number of A&E attendances, and observations of higher attendances in patients over 75 years of age. The dissertation successfully created a model to predict future A&E attendances and lays the foundation upon which future models can be refined and built upon. This research has provided some key insights into A&E activity. It has also revealed that the relationship between A&E attendance numbers, waiting times, arrival hour, patient demographics and location is complex and is influenced by additional factors not encapsulated by this dissertation.

Resulting publications:

PRE-PRINT COMING SOON.