Mental disorders and ecological structure in Nottingham

This paper presents the findings of a specifically geographical investigation of the incidence, distribution and social/environmental correlates of mental disorder in Nottingham. Three diagnostic groups were examined, namely schizophrenia and paranoia, the effective psychoses (or manic depression) and the non-psychotic mental disorders. Analysis of two distinct cohorts of first contact patients for both the schizophrenia and affective psychoses diagnoses yielded markedly dissimilar incidence rates. It is therefore suggested that observed variations in the incidence of specific mental disorders between cities probably reflect such nosocomial (i.e. service related) factors as diagnostic accuracy and the levels of local psychiatric services. It is therefore dangerous to ascribe the variations solely to the contrasting social/environmental milieux found in these cities. The spatial distributions of the selected mental disorders within Nottingham were analysed in the context of a set of 12 distinct ecological areas. These areas were derived from a matrix comprising 105 sub-areas and 40 variables via principal components analysis and a non-hierarchical clustering algorithm. Marked spatial variation in the incidence of all the mental disorders was demonstrated via both traditional mapping and probability mapping techniques. Pearson product-moment analysis revealed that the distributions of the two diagnostically distinct schizophrenia and affective psychoses cohorts were virtually identical. The relationships between the inception rates for the mental disorders and social/environmental attributes were analysed. Stepwise multiple regression models for both the principal components and the individual census variables revealed strong and systematic relationships with all the mental disorders. Although such ecological analysis has limited explanatory power it nevertheless provides useful insights into mental disorder-environmental relationships. It is argued that these can be subsequently best examined at a disaggregated (i.e. behavioural) level.