Alternative statistical approaches to identifying dementia in a community-dwelling sample

Area of Science:

• Epidemiology
• Biostatistics
• Gerontology

Background:

• Dementia diagnosis relies on various statistical models.
• Limited research compares alternative models for dementia identification.
• Understanding group-specific risk factors is crucial for targeted interventions.

Purpose of the Study:

• To compare logistic regression and recursive partitioning (tree-based models) for identifying individuals with dementia.
• To determine the most effective statistical approach for characterizing dementia risk groups.
• To identify key predictors of dementia in an elderly population.

Main Methods:

• Analysis of a sub-sample from the Duke Established Populations for Epidemiologic Studies of the Elderly.
• Comparison of stepwise multiple logistic regression and classification tree analysis.
• Inclusion of predictors: gender, age, chronic health conditions, basic and instrumental activities of daily living (IADL), and cognitive status.

Main Results:

• Cognitive status and IADL were significant predictors in logistic regression, with cognitive status being most important.
• Classification tree analysis also identified cognitive status as the primary dementia criterion.
• Among the cognitively unimpaired, older age was a risk factor; among the impaired, IADL problems were significant.

Conclusions:

• Both logistic regression and classification trees can identify dementia predictors.
• Classification trees excel at identifying specific high-risk groups within the population.
• Logistic regression is effective for targeting specific predictive characteristics for dementia.