A Scottish record linkage study of risk factors in medical history and dementia outcome in hospital patients.

The relative risk (RR) of acquiring a hospital diagnosis of dementia was estimated in 101,104 patients (1 in 25 fraction) randomly selected from total admissions to Scottish general hospitals between 1968 and 1977. The patients were allocated to putative at-risk groups according to main diagnosis at time of index admission, and RR of dementia (ICD 9,290) was contrasted between ten risk groups and a reference group. Record linkage was used to reduce admission episodes to individual cases, to link general and psychiatric Scottish Morbidity Records (SMR 1 and SMR 4), to identify subsequent admission diagnoses of dementia, and to establish person-years-at-risk for each case by linking to the Registrar General's mortality file. For males, RR was significantly increased in the 'hypertensive' risk group (RR 3.88:95% CI 2.18-9.89: p < 0.05) and was significantly reduced in the 'arterial disease' risk group (RR 0.57:95% CI 0.34-0.95: p < 0.05). There was a trend towards increased risk in the cerebrovascular category. For females, RR was significantly reduced in the 'cancer' risk group (RR 0.46:95% CI 0.34-0.63: p < 0.05). There was no evidence of significant alteration in RR in a number of other possible risk groups (endocrine, head injury, CNS/NS disease, ischaemic heart disease, peptic ulcer). Morbidity registers provide access to large data sets of low reliability. No attempt was made to distinguish between Alzheimer's disease and multi-infarct dementia cases, which limited scope for comparison with studies focused on Alzheimer's disease. Previous reports positive associations between head injury, thyroid disorder, and subsequent Alzheimer's disease were not replicated.