RESUMO
This paper argues that the use of socio-demographic indicators to represent actual or potential demand for mental health services needs to be informed by knowledge of local settlement conditions and trends. Following a selective review of the literature on the use of socio-demographic indicators in mental health care planning and on the spatial ecology of mental illness, a case study of a non-metropolitan jurisdiction in southern Ontario, Canada is presented. The results of a regression-based analysis reveal a strong association between the socio-demographic composition of populations and hospitalization rates for mental illness. The pattern of correlates appears to be underlain by residential location processes that 'filter' populations on the basis of age and socio-economic status. It is concluded that, in the absence of this knowledge of local social geography, applications of indicators methodologies run the risk of being 'black boxes'.