ABSTRACT
Much geographical work about mental illness has been aggregative in nature and, consequently, has ignored variation that might be attributed to individuals. The research reported here is intended to reverse this trend by applying the methods of categorical data analysis to the detailed patient histories contained in the Salford Psychiatric Case Register. The paper begins by reviewing the changing prevalence rates for various diagnoses and types of care in Salford between 1968 and 1986. In the light of this background log-linear data models are fitted to evaluate relationships between service usage and patient age, diagnosis and marital status and residential location at two survey points within the study period. Significant relationships are examined in terms of their temporal stability (such as the diagnosis of schizophrenia and a marital status of single) and instability (the changing role of the community psychiatric nurse) to isolate the main combinations of categories to influence the uptake of psychiatric services.