Estimation of survival following the onset of dementia is an extremely important public health issue as America ages. A recent study reported median survival after dementia to be 3.3 years, much shorter than previously estimated. Likewise, the amount of illness that occurs as cognition declines presents a huge burden on family and health care resources alike. Accurate estimation of both of these measures is needed to predict burden for caregivers and on health resources. Recently, the Cardiovascular Health Study (CHS), a multi-site observational cohort established to investigate risk factors for heart disease and stroke in the elderly, completed an ancillary study (funded by NIA) to evaluate dementia in a subset of 3602 of its participants. A total of 480 cases of incident dementia resulted, 330 (68.8 percent) Alzheimer's disease (AD), 52 (10.8 percent) vascular dementia, 76 (15.8 percent) both AD and vascular dementia and 22 (4.6 percent) other types. We propose to utilize these data along with the extensive clinical and surveillance data of the CHS cohort to answer the following questions: (1) What is the duration of survival for individuals classified with incident dementia participating in the Cardiovascular Health Study (CHS)? (2) What are the rates of hospitalization and nursing home admission following onset of dementia? (3) Do these rates differ by type of dementia, age, gender, race or ApoE status? (4) Do these rates differ between individuals evaluated with incident dementia and others of similar age and gender enrolled in CHS? (5) What are the predictors of survival for individuals with incident dementia? We will merge together data from the following sources to achieve our goals: (1) dementia classification, type of dementia, prevalence/incidence status, and date of onset from the CHS Memory Study database; (2) all demographics, risk factor data (including MRI and neurologic symptoms) and cognition scores from the CHS clinic database; (3) date and cause of death, status at end of follow-up, and all hospitalization data from the CHS Events database; and (4) nursing home information from HCFA MEDPAR files. Cause of death and hospitalization diagnoses will be coded to meet analytic needs. Hospital and death records already collected, will be accessed and abstracted for additional information as needed. Analyses will include calculation of length bias (if present), estimation of median survival and diagnosis-specific rates of hospitalization and nursing home admission by selected demographics and type of dementia. Cox proportional hazards regression will be used to identify predictors of survival after onset of dementia. We will also continue to compile datasets and support use of the date for study collaborators. The strengths of this application include the breadth of data already available for the CHS cohort, the inclusion of well-documented incident cases of dementia, and up-to-date surveillance. This application will support continued analyses of an important database designed to provide valuable research for the prevention and impact of dementia in this country.
| Fitzpatrick, Annette L; Kuller, Lewis H; Lopez, Oscar L et al. (2005) Survival following dementia onset: Alzheimer's disease and vascular dementia. J Neurol Sci 229-230:43-9 |
