Assessing life expectancies of institutionalized older persons in Hong Kong

Background: In Hong Kong, it has been projected that, by the year 2010, there will be 0.86 million persons aged 65 or older. The demand for residential care places continues to rise. Information on the survival of older persons will allow better management of care in institutional settings. Objective: The Weibull model was developed to predict the 18-month survival of a group of Chinese nursing home residents. Methods: In this prospective cohort study, 590 older residents from four nursing homes with different levels of care were enrolled. The main outcome measure was survival at 18 months. Information on subject demographics, clinical data and disability levels as measured by the Patient Assessment Instrument (PAI) was collected at baseline. Life expectancies were derived from accelerated failure time models. Results: Independent predictors of decreased survival derived from the Weibull model included increased age, male gender, diagnosis of cancer, presence of malnutrition and functional dependence. Significant interaction was observed between male gender and PAI score. The goodness of fit of the model was satisfactory. Conclusions: From this study, the model for predicting life expectancy in nursing home residents can assist policy makers in planning long-term care. It can also guide clinicians to make more appropriate management decisions for their older patients. Copyright (C) 2004 S. Karger AG, Basel.