OP52 Dementia progression in the ageing population: a computer simulation accounting for individual cognitive and functional decline variability. (3rd September 2019)

• Record Type: Journal Article
• Authors: Evenden, DC; Walsh, BM; Brailsford, SC; Roderick, PJ
• Is Part Of: Journal of epidemiology and community health. Volume 73(2019)Supplement 1
• Page Start: A25
• Page End: A26
• Publication Date: 2019-09-03
• Subjects: Dementia -- service use -- simulation; Public health -- Periodicals; Epidemiology -- Periodicals; 614.4
• DOI: 10.1136/jech-2019-SSMabstracts.53
• Languages: English
• ISSNs: 0143-005X

Abstract

Abstract : Aim: To improve planning for the growing older population and the complex care needs of people with dementia, it is important to recognise the variability of individual cognitive and functional decline and associated care costs. However, many studies exploring care demand simplify varied patient trajectories by partitioning dementia patients into severity categories, therefore failing to capture heterogeneity. Our aim was to develop a computer simulation that models dementia progression longitudinally, driven by population-level dementia onset, mortality, and ageing and including individual variability. Methods: Each modelled age group contains two stocks and three flows using the System Dynamics methodology. A cognitively normal (CN) stock is connected via an incident flow to a stock representing people with dementia (PWD). The 65 to 105 year age range is modelled using eight contiguous 5-year age groups. Ageing is implemented by transferring CN and PWD survivors from each stock to the next oldest age group at 5-year intervals. Mortality flows complete the structure for each age group. Published sources provide flow rates. Agent-based methods are used to model individual attributes and outcomes for dementia cases. Progression is modelled by defining a progression rate type for each agent and deriving individual severity progression coefficients. These are based on fixed and random effects regression models, sampled from probability distributions. Results: Mean progression rates are consistent with published studies. By including individualised random effects, the model demonstrates a complex relationship between decline, severity and service use. Rapid decline leads to higher annual care costs with higher mortality rates. Slower decline leads to lower annual care costs over a longer time period. By incorporating these different trajectories within the model, dementia, care costs, and QALYs can be partitioned by fast, intermediate, and slow progression types to more fully support targeted recommendations for planning service delivery. Discussion: Our computer simulation model shows that accounting for the considerable variability in dementia progression rates as well as severity categories provides more accurate representation of the variation in patient trajectories and outcomes. This modelling method hybridises population-level epidemiology and individual-level pathology, allowing future simulation scenarios to explore the relative impact of different intervention approaches across the dementia population.