Using survival models to analyze the effects of social attributes on length of stay of stroke patients

Abstract

Hospital length of stay (LOS) is a widely accepted indicator of hospital activity and performance of clinical care. We have collected and analyzed data on LOS of 4,086 stroke patients discharged from the Singapore General Hospital (SGH) for the duration 2004–2007. We chose to study stroke patients' data because stroke is an important chronic disease which requires significant social support and healthcare resources. We used survival analysis to study the effects of social attributes on LOS, using discharge from hospital as the time to event. Gender, age, ethnicity and subsidy status were studied as covariates. We discovered that old age increases the probability of long stay. Indian race decreases the probability of stay while Other makes the patient stay longer as compared to Chinese. Patients who were not fee-subsidized have lower chances of stay than those who were. Gender and Malay race did not have a significant effect on the stay probability. The effects of ethnicity and paying status on LOS reflect the influences of cultural environment and socioeconomic status. This demonstrates the importance of social determinants on healthcare and their consequent effects on the utilization of healthcare resources.