Modeling Framework

The ADRD-MM simulates individual incident ADRD cases from diagnosis to death. A simulated incident case is randomly generated and enters the model. At the point of entry (diagnosis) the person with ADRD’s personal characteristics (e.g., race/ethnicity) and the characteristics of their caregiver (e.g., relationship to the person with ADRD) are generated using published incident statistics and observational data. Thus, the simulated individuals mimic an actual population as observed in data. The model uses a combination of parameters estimated from racially/ethnically diverse data.

Data from the National Alzheimer’s Coordinating Center (NACC; 13% of the sample was non-Caucasian), were used to estimate linear mixed effects models to predict clinical trajectories and these models included quadratic terms for time that account for nonlinear change. NACC data were also used to estimate a Weibull model to predict transitions from the community to nursing home. Data from the Aging, Demographics, and Memory Study (19% of the sample was non-Caucasian) linked to CMS Medicare data were used to estimate generalized linear models to predict time spent caregiving and health care expenditures.

ADRD-MM uses the estimated prediction models to simulate monthly declines in the clinical features (cognition measured using the Mini-Mental State Examination, function measured as the number of limitations based on the Functional Activities Questionnaire, and behavior measured as the number of behavioral symptoms present on the Neuropsychiatric Inventory), transitions (from the community to nursing home [a distinction is made for short/long stay admissions], from Medicare to Medicare-Medicaid, and mortality), and costs to Medicare, Medicaid (home and community based care and nursing homes), and families (value of family care and out-of-pocket medical and nursing home expenditures) in 2015 dollars.

The model simulates multiple incident cases, each with their own personal characteristics, to estimate the average lifetime cost of ADRD. No single source provides longitudinal data on declines in ADRD clinical features and policy outcomes. ADRD-MM is innovative and synthesizes data from racially diverse sources to quantify the lifetime effect of ADRD on family and policy relevant outcomes associated with usual ADRD care.

Baseline Characteristics

Model Inputs