Abstract

We used data from four bi-yearly interviews, from 1984 to 1990, in a longitudinal study of aging (LSOA) to build a model where the dependent variable, difficulty in managing money, was used as a proxy for cognitive decline. We analyzed its multiple correlation with a set of independent risk factors: age, presence of coronary heart disease (CHD), history of myocardial infarction (MI), history of stroke, diabetes status, history of Alzheimer’s disease, and body mass index. The data was analyzed with the statistical software R, and empty fields were imputed with the last observation carried forward (LOCF). We also removed the people at baseline who had difficulty managing their money, leaving a population of N = 4994 individuals older than 70 years of age. The only significant variables found in our stepwise regression were age and diabetes status. We did not attempt a cross validation of this model, but instead generated the receiver operation characteristic (ROC) curve and its area under the curve (AROC). The value found for AROC was 0.63. A practical application of the model is provided. The author and his collaborators are currently applying the methodology presented in this paper to data bases more suited for cognitive-decline.