Although numerous studies have been conducted to discover the spatial patterns of road crashes, relatively few have focused on the patterns of road crashes suffered by socially disadvantaged groups, while simultaneously accounting for urban environmental features. This study used advanced econometric (negative binomial regression) and spatial (geographically weighted Poisson regression) approaches to capture latent geographical diversity in crash patterns. The police-reported crash data for the over-65 population in metropolitan Adelaide, Australia, were investigated for two periods: before and after COVID-19. Using both spatial and nonspatial models, the effects of land use mix, population density, road network design, distance to the central business district, and accessibility of public transit on crash frequency, and location at the neighborhood level were investigated. The findings revealed that, in addition to sociodemographic factors, the aforementioned components had nonlinear effects in varied geographical contexts. Although the number of crashes fell by 20% during the periods studied, the fundamental reasons for such incidents did not change. The results of the study could assist academics and policy makers in Australia to better understand the multidimensional implications of the built environment on the road safety of the elderly—a vulnerable group in society who were disproportionately affected by the global pandemic. The hybrid technique presented in this research has the potential to be useful in other scenarios experiencing varying crash patterns.