Predicting dementia from chronic conditions among clderly

Objective: Examine whether chronic conditions can predict onset of dementia over a four-year period among the elderly in five age cohorts. Methods: Medicare claims were tracked (1996-2000) for 398,724 beneficiaries comprising five cohorts (G1: 65-69 years, n=130,714; G2: 70-74 years, n=119,554; G3:75-79 years, n=82,873; G4: 80-84 years, n = 44,262; and G5: 85+ years, n=21,721). Elderly who had both stroke and dementia (ICD-9 codes) in 1996 were excluded from the analyses. Using each chronic condition that existed in 1996, a path model was used in predicting dementia during 1997-2000 from each chronic condition directly and indirectly through stroke. Results: Several significant trends emerged. First, Dementia occurred among 1.6% of G1, 2.6% G2, 6.7% G3, 9.2% G4, and 14.3% G5. Stroke and transient ischemic attack (TIA) were the strongest and most consistent positive, non-demographic predictors in all age cohorts. Diabetes (G-1 to G3) and congestive heart failure (CHF) (in G3 and G5) were also significantly and positively associated while hyperlipidemia was significantly protective (in G1-G5). Blacks and women were at significantly greater risk in each cohort (G1-G5). Conclusions: Results support the hypothesis that early treatment of stroke and related chronic conditions aggressively pursued will contribute to reducing the occurrence of dementia among the elderly.