Abstract :
Specialized patient care facilities such as coronary care units, intensive care units, renal dialysis units, and cancer radiation therapy units usually constitute scarce resources in a health care delivery system. When patients "compete" for these facilities, decisions on patient selection are made. It is not infrequent to hear the motto of "treating high-risk patients first." A quantitative model implied by the motto is presented. It is assumed that patient classes could be established in terms of the relative effectiveness to be expected from treatments in the facility. For each patient class, arrivals are assumed to be Poisson. The state of the system at a decision point corresponds to the number of patients of each class already in the facility as well as the class of the arriving patient. If the objective is to optimize the overall system effectiveness, and exponentiality assumptions are made about the length-of-stay distributions, then the problem can be formulated in a Markov renewal program. To illustrate the approach, alternative admission policies for a four-bed coronary care unit are studied. There it is shown that the average overall mortality rate can be reduced when a relatively discriminating admission policy is used at a high occupancy level. Simulation runs are also made to evaluate the effects of the exponential approximations.
