Analysis Of Hospital Admissions For Extractions Using Geographically Weighted Regression

Objectives: The number of paediatric hospital admissions for dental related extractions remains a cause for concern, despite dental diseases being largely preventable. While local investigations have taken place, little is known about national trends, and how the relationship between hospital admissions and key predictors vary across England. The aim of this study was to analyse the spatial variation in paediatric hospital admissions for extractions in relation to caries and deprivation data.
Methods: Hospital admissions data (for all dental related reasons) were taken from the Hospital Episode Statistics (HES - 2017/18), for children aged up to 19 years. Additionally, mean dmft data for 5-year-olds were taken from the Dental Public Health Epidemiological Programme for England (2015, 2017), while deprivation data were taken from the Indices of Multiple Deprivation (IMD - 2015). All data were collected at local authority level. Geographically weighted regression was used to analyse the relationships between hospital admissions, mean dmft and deprivation across England, as well as the strength of these relationships, and how these varied spatially.
Results: Mean dmft and IMD scores were found to be significant predictors of paediatric hospital admissions. The analysis demonstrated considerable variation in trends nationally, with some areas exhibiting positive associations between the predictor variables and HES data (as mean dmft scores and deprivation increased, so did the number of hospital admissions), while in other areas the opposite was found (as mean dmft and deprivation scores increased, hospital admissions decreased).
Conclusions: This research is the first to use geographically weighted regression to study national trends in paediatric hospital admissions for dental extractions. The analysis demonstrates considerable variation in the relationship between admissions, caries and deprivation. Due to the use of aggregate data, inferences about causal mechanism are limited. Local case studies may provide additional contextual information that helps to explain why these patterns may be occurring.