Bayesian spatial and ecological models for small-area accident and injury analysis

Abstract

In this article, recently developed Bayesian spatial and ecological regression models are applied to analyse small-area variation in accident and injury. This study serves to demonstrate how Bayesian modelling techniques can be implemented to assess potential risk factors measured at group (e.g. area) level. Presented here is a unified modelling framework that enables thorough investigations into associations between injury rates and regional characteristics, residual variation and spatial autocorrelation. Using hospital separation data for 83 local health areas in British Columbia (BC), Canada, in 1990-1999, we explore and examine ecological/contextual determinants of motor vehicle accident injury (MVAI) among male children and youth aged 0-24 and for those of six age groups (<1, 1-4, 5-9, 10-14, 15-19 and 20-24). Eighteen local health area characteristics are studied. They include a broad spectrum of socio-economic indicators, residential environment indicators (roads and parks), medical services availability and utilisation, population health, proportion of recent immigrants, crime rates, rates of speeding charge and rates of seatbelt violation. Our study indicates a large regional variation in MVAI in males aged 0-24 in British Columbia, Canada, in 1990, and that adjusting for appropriate risk factors eliminates nearly all the variation observed. Socio-economic influence on MVAI was profoundly apparent in young males of all ages with the injury being more common in communities of lower socio-economic status. High adult male crime rates were significantly associated with high injury rates of boys aged 1-14. Seatbelt violations and excess speeding charges were found to be positively associated with the injury rates of young men aged 20-24. This and similar ecological studies shed light on reasons for regional variations in accident occurrence as well as in the resulting injuries and hospital utilisation. Thereby they are potentially useful in identifying priority areas for injury/accident prevention and in informing regional health planning and policy development.

Epidemiology of child pedestrian casualty rates: Can we assume spatial independence?

“…However, this model did not consider spatial dependencies within the casualty rates and will be discussed later. In contrast to the collision location based analysis popular in the UK, MacNab (2003MacNab ( , 2004 has recently presented a spatial analysis from Canada which used hospital derived data aggregated by the area in which the casualty is resident.…”

Exploring Spatial Structure behind the Road Mortality of Regions in Europe

“…Moreover, it enables data sharing (i.e. risk smoothing) in space, which often results in more reliable risk prediction (MacNab 2004).…”

The Influence of On-Premise and Off-Premise Alcohol Outlets on Reported Violent Crime in the Region of Waterloo, Ontario: Applying Bayesian Spatial Modeling to Inform Land Use Planning and Policy

“…Methodologically, the Bayesian spatial regression modeling approach applied in this research overcomes limitations of modeling count data for small-areas (MacNab 2004;Li et al 2013). In small-area spatial analysis, crime counts and population are often small and crime rate estimates may be misrepresentative of crime risk (i.e., a small change in count or population results in a large change in crime rate) (Law and Chan 2011).…”