Bed occupancy rate modeling in Shiraz hospitals

Abstract Background: The Bed Occupancy Rate (BOR) among Iranian hospitals was reported to be 57.4% in 2001 indicative of a significance difference with standard index of 75%. Objective: To evaluate the bed occupancy rate modeling in Shiraz hospitals. Methods: This was a cross-sectional study carried out on random samples from different hospitals in Shiraz during the first half of 2004. The Poisson models including the negative binomials and mixture Poisson were used for BOR modeling. Hierarchical mixture Poisson, negative binomial and hierarchical two-component normal distribution were applied for the role of each hospital ward. Findings: Our findings showed that the average BOR was 66.4% in Shiraz hospitals. Regarding the relative hemogenity in observation of BOR in different hospitals, the hierarchical mixture Poisson and negative binomial models failed to accurately explain this variable and it was the hierarchical mixture two-component normal model which gave an accurate explanation for BOR found in our study. In this model, one of the components indicates the BOR observation close to mean and the other represents the outlier observations. Conclusion: The hierarchical modeling in which a hospital ward acts as a random component was shown to lead to the highest value for BOR.