A multistate model and its standalone tool to predict hospital and ICU occupancy by patients with COVID-19. Issue 2 (February 2023)

• Record Type: Journal Article
• Authors: Lafuente, Miguel; López, Francisco Javier; Mateo, Pedro Mariano; Cebrián, Ana Carmen; Asín, Jesús; Moler, José Antonio; Borque-Fernando, Ángel; Esteban, Luis Mariano; Pérez-Palomares, Ana; Sanz, Gerardo
• Is Part Of: Heliyon. Volume 9:Issue 2(2023)
• Publication Date: 2023-02
• Subjects: COVID-19 -- Health resources -- Multistate models -- Hospital and ICU occupancy -- Predictive tool; Research -- Periodicals; Medical sciences -- Periodicals; Natural history -- Periodicals; Social sciences -- Periodicals; Earth sciences -- Periodicals; Physical sciences -- Periodicals; 507.2
• DOI: 10.1016/j.heliyon.2023.e13545
• Languages: English
• ISSNs: 2405-8440

Abstract

Abstract: Objective: This study aims to build a multistate model and describe a predictive tool for estimating the daily number of intensive care unit (ICU) and hospital beds occupied by patients with coronavirus 2019 disease (COVID-19). Material and methods: The estimation is based on the simulation of patient trajectories using a multistate model where the transition probabilities between states are estimated via competing risks and cure models. The input to the tool includes the dates of COVID-19 diagnosis, admission to hospital, admission to ICU, discharge from ICU and discharge from hospital or death of positive cases from a selected initial date to the current moment. Our tool is validated using 98, 496 cases positive for severe acute respiratory coronavirus 2 extracted from the Aragón Healthcare Records Database from July 1, 2020 to February 28, 2021. Results: The tool demonstrates good performance for the 7- and 14-days forecasts using the actual positive cases, and shows good accuracy among three scenarios corresponding to different stages of the pandemic: 1) up-scenario, 2) peak-scenario and 3) down-scenario. Long term predictions (two months) also show good accuracy, while those using Holt-Winters positive case estimates revealed acceptable accuracy to day 14 onwards, with relative errors of 8.8%. Discussion: In the era of the COVID-19 pandemic, hospitals must evolve in a dynamic way. Our prediction tool is designed to predict hospital occupancy to improve healthcare resource management without information about clinical history of patients. Conclusions: Our easy-to-use and freely accessible tool (https://github.com/peterman65 ) shows good performance and accuracy for forecasting the daily number of hospital and ICU beds required for patients with COVID-19.