RT Journal
A1 Reitz C, Tang M, Schupf N, Manly JJ, Mayeux R, Luchsinger JA
T1 A summary risk score for the prediction of alzheimer disease in elderly persons
JF Archives of Neurology
JO Archives of Neurology
YR 2010
FD July 1
VO 67
IS 7
SP 835
OP 841
DO 10.1001/archneurol.2010.136
UL http://dx.doi.org/10.1001/archneurol.2010.136
AB Objective 
    To develop a simple summary risk score for the prediction of Alzheimer disease in elderly persons based on their vascular risk profiles.Design 
    A longitudinal, community-based study.Setting 
    New York, New York.Patients 
    One thousand fifty-one Medicare recipients aged 65 years or older and residing in New York who were free of dementia or cognitive impairment at baseline.Main Outcome Measures 
    We separately explored the associations of several vascular risk factors with late-onset Alzheimer disease (LOAD) using Cox proportional hazards models to identify factors that would contribute to the risk score. Then we estimated the score values of each factor based on their β coefficients and created the LOAD vascular risk score by summing these individual scores.Results 
    Risk factors contributing to the risk score were age, sex, education, ethnicity, APOE ε4 genotype, history of diabetes, hypertension or smoking, high-density lipoprotein levels, and waist to hip ratio. The resulting risk score predicted dementia well. According to the vascular risk score quintiles, the risk to develop probable LOAD was 1.0 for persons with a score of 0 to 14 and increased 3.7-fold for persons with a score of 15 to 18, 3.6-fold for persons with a score of 19 to 22, 12.6-fold for persons with a score of 23 to 28, and 20.5-fold for persons with a score higher than 28.Conclusions 
    While additional studies in other populations are needed to validate and further develop the score, our study suggests that this vascular risk score could be a valuable tool to identify elderly individuals who might be at risk of LOAD. This risk score could be used to identify persons at risk of LOAD, but can also be used to adjust for confounders in epidemiologic studies.