(P27-009-23) Impact of Cardiovascular Biomarkers on Dementia

Objectives: Worldwide prevalence of dementia is expected to reach 82 million by the year 2030, and to nearly double that by 2050. The study was conceived to provide predictive tools to aid HCPs in identifying at-risk patients. A longitudinal (20 years) retrospective observational study was conducted using IBM Explorys dataset of patient electronic health records from U.S. healthcare systems to explore associations between cardiovascular biomarkers and dementia and to develop a predictive model for dementia.

Methods: Dementia and No Dementia cohorts were formed using patient diagnosis records and compared across selected cardiovascular biomarkers. Data were analyzed by Wilcoxon rank sum test, logistic regression, or classification trees. Selection bias was mitigated using propensity score matching.

Results: Odds of diagnosis with dementia were calculated. In male and female patients, respectively, with high triglyceride, odds were 29% and 19% higher; in those with low high-density lipoprotein (HDL) cholesterol, odds were 26% and 27% higher; and in those with high hemoglobin A1C (HbA1c), odds were 47% and 28% higher. In female patients with high total cholesterol to HDL ratio, odds were 44% higher; in those with high systolic BP, odds were 16% higher; and in those with high pulse pressure, odds were 21% higher. The predictive model predicted dementia if triglyceride (mg/dL) >175 or triglyceride < 175 and HbA1c ≥ 8.5%; no dementia if triglyceride < 175 and HbA1c < 8.5%. Accuracy of the model was 56% (baseline accuracy = 50%). All P values < 0.05.

Conclusions: This study showed association between lipids, blood pressure, HbA1c and dementia. Our results allow identification of vulnerable populations for early intervention and provide direction for nutritional product innovations to meet their needs.

Funding Sources: Abbott Laboratories