Effects of Cognitive Impairment on Hospitalizations in Heart Failure Patients




Lykens, Kristine A.
Bhatti, Megan
Miller, Thaddeus
Nandy, Karabi


Journal Title

Journal ISSN

Volume Title



Abstract Purpose: Heart failure (HF) is a major health problem in the United States (US) linked to poor survival rates, high rehospitalization rates and high healthcare cost. HF is positively associated with aging and its impact on US health and healthcare systems is expected to grow as the baby boomer generation enters their retirement years. The same is true for another chronic health risk, cognitive impairment. There is a clear, negative impact on prognosis and healthcare outcomes associated with cognitive impairment in HF patients, but less is known about how these affect systems outcomes such as overall hospitalization. We compared hospitalization patterns among aged HF patients with and without comorbid cognitive impairment to identify associated risks and outcomes. Methods: The data for this analysis will be obtained from the Medical Expenditure Panel Survey. MEPS is a survey that is conducted each year through the Agency for Healthcare Research and Quality (AHRQ) including both a household and an insurance component. MEPS provides relevant data including number of hospital discharges, diagnosis codes, etc. The sample is a stratified, cluster sample to ensure racial representation. Statistical analyses will be performed using SAS. Estimates of risk of hospitalization for patients with and without cognitive disorders will be calculated using self-response weights. Associated standard errors will be calculated using replicate weights obtained from MEPS. For the first question, differences in the distribution of risk for patients with and without cognitive disorders will be evaluated using weighted Chi-squared tests. A weighted logistic regression will be used to find factors associated with hospitalization risk by examining associations among demographic and other characteristics with the outcome. Goodness-of-fit of the model will be assessed using the Hosmer-Lemeshow test and a deviance test. Results and Conclusions: pending