Length of stay of patients dying in central region hospitals in Thailand

Abstract

        The statistical model was constructed to describe the variation in length of stay (LOS) of patients who died in hospital. The total of 257,076 records of mortality cases were included to determine the effect of factors on LOS for the period of October 1999 to September 2007, reported from provincial hospitals in Central Thailand to the Ministry of Public Health.

        We considered principal diagnosis (ICD-IO), age, gender, hospital size, and the geographic region as predictors for LOS. Chi-squared test was to investigate associations between the factors and LOS. All predictors were found to have a statistically significant association with LOS (p-value < 0.05). Multiple logistic regressions with weighted sum contrasts were used to compare proportions for each factor after adjusting for categorical covariates. The results showed that the percentage of longer LOS (7 days or more) was 35.2 among all patients. LOS increased with age, and hospital size. Injury was the principal diagnosis most likely to have short LOS, whereas cancer patients had the highest LOS. In term of geographic region, Bangkok had the highest percentage of patients with longer LOS, whereas the Southeast of Bangkok had the lowest percentage.