Modelling stroke patient pathways using survival analysis and simulation modelling

Abstract

Stroke disease is the third leading cause of death in the UK, placing a heavy burden on society

at a cost of 7 billion pounds per year. Prolonged length of stay in hospital is considered to be an inefficient use of hospital resources. In this paper we present results of survival analysis that utilise length of

stay and destination as outcome measures, based on data from the Belfast City Hospital. Survival probabilities were determined using Kaplan-Meier survival curves and log rank tests. Multivariate Cox proportional hazards models were also fitted to identify independent predictors of length of stay including

age, gender and diagnosis. Elderly patients showed a decreased hazard ratio of discharge. However, gender was not a significant hazard risk for length of stay in hospital. Those patients with a diagnosis of

cerebral haemorrhage showed an increased hazard ratio and hence were most likely to have a shorter length of stay and to die in hospital. Those who were eventually discharged to a Private Nursing Home had the lowest probability of early discharge. On the basis of these results we have created several groups, stratified by age, gender diagnosis and destination. These groups are then used to form the basis of a simulation model where each group is a patient

pathway within the simulation. Various scenarios are explored with a particular focus on the potential efficiency gains if length of stay in hospital, prior to discharge to a Private Nursing Home, can be reduced.