Patient Triage and Prioritization Under Austere Conditions
Below is a blog I write for INFORMS. See the original post at https://www.informs.org/Blogs/ManSci-Blogs/Management-Science-Review/Patient-Triage-and-Prioritization-Under-Austere-Conditions
Triage and prioritization help allocate limited resources to patients who need it most in an effort to save as many lives as possible. Thus, it has become a common practice in both daily emergency department operations and mass-casualty events. However, the information that is required to prioritize patients, such as urgency and resource requirements among others, may not be immediately available. Thus, decision makers have to spend some time on triage and collect such information (perfect or imperfect), which results in delay in the actual treatment of patients. This is especially the case in war zones and economically deprived regions, due to extreme resource restrictions, a single provider may be the sole person in charge of providing emergency care to a group of patients. Therefore, it is important to balance the time spent on triage with the time spent on the actual service. This decision problem shares some similarity with the classical exploration-exploitation dilemma. In fact, the problem at its core, i.e., how to strike a balance between the time spent on acquiring more information and that spent on acting on the available information, is not even unique to services. In our daily lives, we constantly prioritize our tasks by assessing the relative value of prioritizing one task over the other given the available information.
The paper “Patient Triage and Prioritization Under Austere Conditions,” by Zhankun Sun, Nilay Tanık Argon and Serhan Ziya sheds some light on this problem. The authors develop a mathematical framework to capture the essential features of the problem and analyze this formulation to provide insights that can be helpful in making decisions in practice. The authors are able to completely characterize the optimal policy on if and when to use triage, which depends on patients’ clinical severity, service requirements, as well as the current patient mixture. One of the key insights from their finding is that it is better to triage when there are many patients waiting for service. When there are only few patients, service of all patients will not take too much time. Thus, the value of information that will be obtained through triage does not justify the additional waiting that all patients will have to endure. The authors also identify conditions under which it is optimal to use static policies, such as triage all patients and triage no patient. These static policies are simple and easy-to-implement thus are appealing in practice.
The authors then demonstrate that it is possible to develop simpler heuristic policies (compared to the optimal policy) based on their analytical results. They test the performance of their policies under more realistic and complex settings, relaxing some assumptions enforced on the mathematical model for the sake of tractability. The numerical study shows that the best heuristic policy performs close to the optimal policy, and outperforms the best static policy in most test cases. One insight from the numerical results is that the information learnt from triage is most beneficial when the cost of learning (delay caused by triage) is small and when the resource is extremely limited. Those results are also shown to be robust to the mixture of important and less important patients, which is not easy to estimate in practice. In the end, the authors discuss other possible scenarios that their model and results can be useful, such as the search and rescue operations, intelligence collection management, among others.
Read the full article at: https://doi.org/10.1287/mnsc.2017.2855.