This The rule of thumb has been to plan a hospital for 80 percent occupancy. The figure is said to provide for fluctuations in patient census, and historically, many hospitals do operate at about that occupancy rate. rule of thumb, however, may imply construction of beds which lie unused a great part of the time. A more rational approach to hospital sizing considers the fraction of the time that patients must be scheduled or turned away, with occupancy rate a derived figure rather than a design figure. This section discusses a method of determining hospital size on the The basic idea is to specify a confidence level (or probability) that the hospital can accommodate requests for admission, and to derive the number of required beds from census figures. The confidence level is prescribed on the basis of judgments as to acceptable levels of scheduling and the implicit social cost of scheduling or refusing admissions. basis of service rather than occupancy. To derive the number of beds from census figures, the statistics of patient occupancy are needed. These statistics can be gathered from the past operation of the hospital, or in the case of new construction, inferred from projected average census figures. To infer the statistics of patient census, the following assumptions are made: 1. Patients (or their physicians) request admission independently of one another and at a constant rate over time 2. Patients' stay in the hospital is not influenced by the occupancy rate 3. Patients can be admitted to any empty bed in the hospital of the time. ཨཱ། Under these assumptions, the statistics of patient census are adequately described by a Poisson distribution uniquely determined by the average census. Using this distribution, the number of beds required is derived from the average census and the chosen confidence level. Some discussion of these assumptions is necessary to determine whether this method can be applied to the reference hospital. Though admission may be requested for a patient, his condition does not necessarily require immediate hospitalization; for example, much surgery can be scheduled. The observed fluctuation in admissions, therefore, will be less than that predicted by this model. Though there is a seasonal variation in admission rates, over the peak season the admission rate is nearly constant. From these two factors we conclude that Assumption 1 leads to a conservative estimate of the number of beds required. The second assumption would have the same effect, since it has been observed that length of stay does tend to decrease with increasing occupancy in the same hospital. Assumption 3 requires further consideration. Increased utilization will result from making each bed serve as wide a range of patients as possible. However, some compartmentalization is necessary. Extremely ill patients go to the Intensive Care Unit; patients with infectious diseases must be isolated; obstetrical wards must be segregated by law in some states. From the viewpoint of patient management, segregation by sex may be required on the wards. In the reference hospital, there are enough private rooms to be used for isolation when desired. Even if the ICU, CCU, pediatric beds, and private rooms are excluded, enough beds will remain so that a patient can be admitted to any empty bed in almost every case. Using this distribution of census statistics, the planning assumptions of the reference hospital's size will be tested. Initially, the original hospital had 250 beds operating at 70 percent occupancy; the new plans |