Objective: The objective of this study was to compare the prevalence of dementia using different diagnostic systems, and to investigate the influence of the different diagnostic components (memory impairment, personality changes, definition of other intellectual functions) on the prevalence.
Methods: A general population sample of 1,019 elderly living in Gothenburg, Sweden was investigated by using the Comprehensive Psychopathological Rating Scale as well as specific assessments relevant for dementia diagnoses. Diagnoses were given according to the 9th and 10th version of the International Classification of Diseases (ICD-9, ICD-10) as well as the 3rd revised and 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R, DSM-IV). Further, "historical" criteria for dementia were applied as had been used in older studies.
Results: DSM-IV dementia occurred most frequently (9.6%), followed by dementia according to "historical" criteria (7.4%), DSM-III-R (6.3%), ICD-10 (3.1%), and ICD-9 (1.2%). The kappa values for the agreement between the diagnostic systems were between 0.166 and 0.810. The requirement of both long-term and short-term memory impairment in DSM-III-R and personality changes in ICD-10 explained most of the differences. When these requirements were held constant, DSM-III-R, DSM-IV, ICD-10 and "historical" criteria identified predominantly the same persons as demented (kappa: 0.810-1.000).
Conclusion: Prevalence of dementia varied widely depending on diagnostic classification system used. For DSM-III-R, DSM-IV, ICD-10, and "historical" criteria, the definitions of personality changes and combinations of memory impairment lead to differing prevalence rates, whereas the definitions of other intellectual functions have little impact.