Modeling fluctuations in a hospital's census

Reducing variability in a hospital's census improves the efficiency of the hospital and reduces the overall operating cost. A deterministic and stochastic four stage model is created for the purpose of modeling the hospital census and its variability. Stochastic admissions are handled as a mixed Poisson process with admission rates which depend on day of the week. The stochastic discharge process is accomplished through use of a trinomial random variable. The results are used to determine whether the most variability comes from surgical or medical admissions or surgical or medical elapsed stay. It is found through calculation of correlation that surgical admissions make a strong contribution to the variability of the overall census. The partial correlation coefficients are calculated to reveal that medical average elapsed stay explains a good portion of the census not explained by medical admissions, surgical admissions, and surgical elapsed stay. Through experimentation with the admissions process, it is determined that a reduction in the variation of surgical admissions results in a significant reduction in the variation of the census. A reduction in the variance of medical admissions experiments which reduced the medical and surgical elapsed stay also result in a reduction of the variance.