Admissions optimisation and premature discharge decisions in intensive care units

This paper studies the admission and premature discharge decisions in the intensive care unit (ICU). While the previous admission policy first come, first served treated all patients equally, this paper divides patients into two classes. One class (class 1) is the more critical patients who cannot be prematurely discharged to other hospital units. The other class (class 2) is patients who can be prematurely discharged to other hospital units in order to accommodate new class 1 patients. We formulate a dynamic programming model to determine the best policy for allocating available beds to different classes of patients and reducing premature discharging costs. By analysing the model, we conclude that there is a threshold (i.e. a fixed number of available beds) for class 2 patients in each time period. If the number of available beds is lower than the threshold, the request of a class 2 patient will be rejected. Otherwise, he/she will be accepted. We find that the survival benefits follow a marginal diminishing effect. The management team can decide how many beds should be allocated to ICU by considering the balance of budget and survival benefits. We also establish the lower and upper bounds of probability that a patient is admitted and prematurely discharged on the same day. The bounds can be used to evaluate our policy and adjust the parameter to improve the policy. The computational experiments illustrate that the proposed policy is better than the traditional policies and the obtained threshold is lower than the threshold that premature discharging is not permitted. The average survival benefits are computed with all initial states. The proposed method is effective and can help ICUs to obtain a relative high survival benefits per day.