Using a multi-state model to enhance understanding of geriatric patient care.

Objectives: To use multi-state Markov chain modelling to analyse data on geriatric patient care, and to make comparisons between male and female patients.

Methods: Estimation, from observed data, of covariate (age of patient and date of admission to hospital or community care) dependent parameters of statistical models for time in care and subsequent events.

Results: Differential effects of these covariates shown on the parameters of the models for female and male patients, where these parameters can be interpreted as affecting different features of the distributions of time in care.

Conclusions: Multi-state modelling is an appropriate means of analysing data on geriatric patient care and can reveal underlying patterns of differential effects, some of which may not be apparent from more routine data processing.