Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Industrial Engineering


Taaffe, Kevin M

Committee Member

Craig , Janet B

Committee Member

Garrett , Sandra K

Committee Member

Mayorga , Maria E


The success of a healthcare facility evacuation depends on communication and decision-making at all levels of the organization, from the coordinators at incident command to the clinical staff who actually carry out the evacuation. One key decision is the order in which each patient is chosen for evacuation. While the typical planning assumption is that all patients are to be evacuated, there may not always be adequate time or resources available to move all patients. In these cases, prioritizing or ordering patients for evacuation becomes an extremely difficult decision to make. These decisions should be based on the current state of the facility, but without knowledge of the current patient roster or available resources, these decisions may not be as beneficial as possible.
Healthcare facilities usually consider evacuation a last-resort measure, and there are often system redundancies in place to protect against having to completely evacuate all patients from a facility. Perhaps this is why there is not a great deal of research dedicated to improving patient transfers. In addition, the question of patient prioritization is a highly ethical one.
Based on a literature review of 1) suggested patient prioritization strategies for evacuation planning as well as 2) the actual priorities given in actual facility evacuations indicates there is a lack of consensus as to whether critical or non-critical care patients should be moved away from a facility first in the event of a complete emergency evacuation. In addition, these policies are 'all-or-nothing' policies, implying that once a patient group is given priority, this entire group will be completely evacuated before any patients from the other group are transferred. That is, if critical care patients are given priority, all critical care patients will be transferred away from the facility before any non-critical care patient.
The goal of this research is to develop a decision framework for prioritizing patient evacuations, where unique classifications of patient health, rates of evacuation, and survivability all impact the choice. First, I provide several scenarios (both in terms of physical processing estimates as well as competing, ethically-motivated objectives) and offer insights and observations into the creation of a prioritization policy via dynamic programming. Dynamic programming is a problem-solving technique to recursively optimize a series of decisions. The results of the dynamic programming provide optimal prioritization policies, and these are tested with simulation analysis to observe system performance under many of the same scenarios. Because the dynamic programming decisions are based on the state of the system, simulation also allows the testing of time-based decisions. The results from the dynamic programming and simulation, as well as the structural properties of the simulation are used to create assumptions about how evacuations could be improved.
The question is not whether patient priorities should be assigned - but how patient priorities should be assigned. Associated with assigning value to patients are a variety of ethical dilemmas. In this research, I attempt to address patient prioritization from an ethical perspective by discussing the basic principles and the potential dilemmas associated with such decisions.
The results indicate that an all-or-nothing, or a 'greedy' policy as discussed in the literature may not always be optimal for patient evacuations. In some cases, a switching policy may occur. Switching policies begin by evacuating patients from one classification and then switch to begin evacuations from the second patient class. A switch can only be made once; after a switch is made, all remaining patients from the new group should be evacuated. When there are no more patients of that group remaining in the system, the remaining patients from the class that was initially given priority should be evacuated. In the case of critical and non-critical care patients, switching policies first give priority to non-critical care patients. When the costs of holding patients in the system are not included in the models - and the decisions are just based on maximizing the number of saved lives - the switching policies may perform as good or better than the greedy policies suggested in the literature. In addition, when holding costs are not included, it is easier to predict whether the optimal policy is a greedy policy or a switching policy.
Prioritization policies can change based on the utility achieved from evacuating individual patients from each class, as well as for other competing objective functions. This research examines a variety of scenarios - maximizing saved lives, minimizing costs, etc. - and provides insights on how the selection of an objective impacts the choice. Another insight of this research is how multiple evacuation teams should be allocated to patients. In the event that there is more than one evacuation team dedicated to moving a group of patients, the two teams should be allocated to the same patient group instead of being split between the multiple patient groups.