Suicide variations between English neighbourhoods over 2017-21: The role of spatial scale

Abstract

Geographic studies of suicide variation typically focus on predictors at the same level as the event rates, and the possible interplay between different spatial scales does not generally figure. In this paper we focus on suicide variations between 6856 small area census units in England, but against a background provided by nine regions, broad urban-rural categories, and 155 local labour markets. Suicide death totals vary considerably between the small areas, with more areas than expected having no deaths, so we apply zero inflated regression. With this framework, we consider the relative contribution of factors at higher and lower spatial scales in explaining small area suicide contrasts, and why some areas have unduly elevated or unduly low suicide rates. We find significantly lower suicide levels in English metropolitan regions, after allowing for neighbourhood influences, but considerable heterogeneity in risks within broader spatial units. Varying incidence in general is associated significantly with all observed neighbourhood risk factors (social fragmentation, socioeconomic status, mental ill-health, ethnic mix), but low fragmentation and low psychiatric morbidity are the only significant influences on unduly low incidence.