On the use of multi-state multi-census techniques for modelling the survival of elderly people in institutional long-term care

A good estimate of the length of stay (LOS) of elderly people in institutional long-term care (ILTC) is essential to both care providers (care home owners) and purchasers (local authorities in the UK) in order to forecast their resource availability and future requirements. Existing methods usually rely on cohort data, which need to be collected on a continuous basis for several years and are often not available electronically. In this paper we develop an approach based on multi-census data to fit a Markov model in continuous-time for the LOS of elderly people in ILTC (residential and nursing care). We use 3 year placement data provided by London Borough of Merton Social Services to compare the estimates of the model parameters obtained using multi-census data with those derived from cohort data, hence to validate the multi-census approach in this context. The implications of the proposed multi-census approach and its advantages in the context of long-term care planning are discussed.