Abstract

STUDY OBJECTIVES--To assess the value of death certification for the epidemiological study of dementia, the frequency with which the condition was recorded on death certificates of patients diagnosed with some form of dementia before the age of 65 years was studied. A further objective was to identify variables associated with failure to record dementia on the certificate. DESIGN--A cohort of patients with presenile dementia, differentiated by a clinical algorithm applied to hospital case records, was traced through the National Health Service Central Registry and details of certified causes of death were obtained. SETTING--The Northern Regional Health Authority in England. SUBJECTS--Prevalent cases of presenile dementia resident in the northern health region during 1986 traced up to April 1992. MEASUREMENTS AND MAIN RESULTS--The underlying cause of death was recorded as dementia or as Alzheimer's disease in 53% of cases of clinically diagnosed presenile Alzheimer's disease, 33% of cases of presenile vascular dementia, and 10% of cases of presenile dementia secondary to another neurological condition. Dementia or Alzheimer's disease was recorded in any part of the certificate in 75% of cases of Alzheimer's disease, 52% of vascular dementia, 33% of other dementias, and in 65% of cases overall. Dementia or a cerebral condition of a kind that can result in dementia was recorded in 80% of all cases. Failure to mention dementia was related to the clinical type of dementia, shorter duration of illness, and earlier period of study. CONCLUSIONS--The underlying cause of death seriously understates the frequency of dementia, but when the recording of other brain disease is taken into account the presence of potentially dementing brain disease is recorded much more frequently. It is suggested that coding chronic conditions present at death, such as dementia, in addition to those causing or contributing to death would improve the value of death certificates for epidemiological purposes.