Forecasting demand for long-term care services.

ABSTRACT

This article analyzes three methods used to forecast the transition of long-term care clients through a variety of possible home and facility placements and levels of care. The test population (N = 1,653) is derived from the larger population of clients admitted in 1978 to British Columbia's newly established Long-Term Care program. The investigators have accumulated 5 years of service-generated data on moves, discharges, and deaths of these clients. Results show that the first-order Markov chain with stationary transition probabilities yields a superior forecast to state-by-state moving average growth and state-by-state regression analyses. The results of these analyses indicate that the Markov method should receive serious consideration as a tool for resource planning and allocation in long-term care.