MANAGING A HOSPITAL GERIATRIC WARD: IDENTIFYING COST-EFFECTIVE STRATEGIES

This paper presents a method for modelling the expected cost for a group of patients' accumulated time in hospital. By expanding an existing method for modelling the survival of patients in care, which incorporates a special type of Markov model in a Bayesian Network (BN), the methodology presented in this paper attaches a cost variable to the final node of the BN, enabling the estimation of departmental costs and the ability to investigate potential policy changes. An application of the model is illustrated using data relating to the admissions of 6994 patients into the geriatric department of a major U.K. hospital between 1969 and 1985. The model's practical use is demonstrated by considering the potential benefits to hospital managers investigating the implementations of making policy changes and the influence of future decisions on the hospital expenditure.