Simulation-based occupancy recommendations for adult medical/surgical units using admissions scheduling systems.

TLDR: The contention in the remainder of the paper is that almost any hospital medical/surgical unit should operate well in excess of that model's widely used occupancy recommen?

Abstract: Minimum cost operation of a hospital requires the correct number of beds to meet the de? mand placed on the facility. An excess of beds results in inflated operating and construction costs, while a bed shortage is unacceptable for a variety of reasons, such as the lack of quality care to the community. Several models have been developed and used to assist planners in finding the correct bed size to meet a given demand. These models are inadequate in an environment where stringent cost containment is second only in importance to quality of care. The pre? vious models, such as the Hill-Burton formu? las and the Poisson approximation, are inad? equate because they are incompatible with contemporary admissions scheduling systems. These systems alter the behavior of the census so that analytical models based upon the Pois? son assumption do not fit the results and, along with normative models such as the Hill Burton program, allow too many beds. The implications of the Poisson assumption will be discussed at length here because the contention in the remainder of the paper is that almost any hospital medical/surgical (M/ S) unit should operate well in excess of that model's widely used occupancy recommen? dations. In most cases this also leads to op? eration in excess of the state Hill-Burton rec? ommendations for medical/surgical units, thus Hill-Burton is rejected as a consequence.1 The Poisson assumption for hospital census was first used in the 1940s,2 and has been ap? plied worldwide to determine the size of hos? pital facilities.3 An appealing feature of the Poisson assumption is the simplicity of the parameters; the census mean is equal to the census variance. The factor which has varied from application to application has been the amount of that variance to allow in sizing a particular facility. The first models used from three to four standard deviations which cor? responded to at least the 99.9 percentage point of the Poisson distribution. Later model build? ers reduced this coefficient. One low value used is 2.06 standard deviations, correspond? ing to the 98 percentage point of the normal approximation to the Poisson distribution: