Hawaiʻi’s Covid models, responses, lessons learned

University of Hawaiʻi

buildings and skyline

In one of the first comprehensive studies on the state’s COVID-19 pandemic response, a group from the Hawaiʻi Pandemic Applied Modeling Work Group (HiPAM), that is chaired by University of Hawaiʻi at Mānoa faculty member Victoria Fan and broadly included UH Mānoa applied researchers towards a specific policy challenge, identified key takeaways to inform how to better respond to the next pandemic. The key takeaways include:

  • COVID-19 modeling in Hawaiʻi benefited from the incorporation of state-specific data.
  • The models are dynamic and should be interpreted as such.
  • After considering model outputs and issuing appropriate warnings, actions should be taken immediately.
  • These models require a firm grasp of epidemiologic concepts, therefore, policymakers should involve an interdisciplinary scientific advisory group as early as possible to translate models into actionable items.
  • There is a need for communicators who help to translate and communicate complex ideas into simple concepts for policymakers and the public.

“This case study is intended to and may help future policymakers seeking to navigate this complex landscape of models and draw upon practical lessons learned on how to make appropriate evidence-based decisions using models,” said the study titled “Mitigation Planning and Policies Informed by COVID-19 Modeling: A Framework and Case Study of the State of Hawaii,” which was published in the International Journal of Environmental Research and Public Health on May 18. Authors included experts from the UH Mānoa Thompson School of Social Work and Public Health, Hawaiʻi Data Collaborative, UH Mānoa Department of Mathematics, UH Mānoa John A. Burns School of Medicine, Hawaiʻi Department of Health, Hawaiʻi Department of Defense, and the Center for Global Development in Washington, D.C.

The study examined different COVID-19 prediction models, followed by the subsequent policy decisions that were made and then the lessons learned from the process. The methodology and data were based on a review of the historical facts and real-world events through experiences of the authors of the paper.

Prediction models

swab and tube

During the early pandemic crisis, two models widely circulated were the University of Washington Institute for Health Metrics and Evaluation model and the Imperial College London model. After the technical team realized that these two models were not suitable for local customization, the team focused on two other models—Epidemic Calculator and the University of Basel model. The key characteristics in selecting the models included, model objective (for example, COVID-19 hospital impacts vs. community spread), local customizability (ability to insert state-specific parameters) and age distribution (accounting for age to project the case, hospitalization and fatality numbers more accurately). By 2021, the team relied on the locally tailored model developed by UH Mānoa Professor Monique Chyba.

Key policy decisions

The authors explained how and which models were used to inform key policy decisions in Hawaiʻi, both in terms of managing capacity during a surge and reopening amidst a decline. During a surge, the study analyzes which models were used to inform decision making in three selected use cases or scenarios: determining whether there was adequate hospital bed capacity in the state and adequate PPE (personal protective equipment) in the state, assessing the need for isolation and quarantine facilities from the surge of the second wave in fall 2020, and the role public communication played during the Delta surge in summer 2021, and the Omicron surge in fall 2021.

Models were also used to inform policy decisions on when to loosen restrictions during a decline, particularly reopening to domestic and international travelers. While some models can predict COVID-19 spread under certain mitigation measures, there is still much uncertainty about the basic scientific facts and assumptions of COVID-19, for example the extent of screening for asymptomatic transmission and the infection fatality rate, making policy decisions difficult.

The authors said that COVID-19 will continue to affect the community for decades, therefore, “policymakers will need to shift from the use of models that focus on hospital capacity and reopening, to models identifying long-term health and economic impacts of COVID-19, such as mental health, access to non-COVID-19 health care services, education, and other dimensions of the social determinants of health.”

Shutdown of HiPAM

Fan, HiPAM founder and associate professor of health policy, also announced that HiPAM intends to close down from June 30, 2022 due to lack of available funds.

“HiPAM has been an amazing collaboration in which applied researchers and health professionals came together during a time of great community need,” Fan said.

“We are grateful for the opportunity to partner with others in the state to serve our community,” HiPAM co-chair Thomas Lee commented. “HiPAM gave to the community our knowledge, expertise and skills to provide timely and unbiased information to the public using best available evidence and ever-evolving science.”

Chyba, HiPAM mathematician and professor of mathematics, added, “HiPAM demonstrated the best of how university partnerships can be relevant to the real world and innovative during a time of crisis.”

HiPAM was supported by several sources including the Hawaiʻi Data Collaborative, UH Mānoa Provost’s Office, National Science Foundation, Hawaiʻi Department of Health, and most recently, the Hawaiʻi Department of Defense.

This work is an example of UH Mānoa’s goal of Excellence in Research: Advancing the Research and Creative Work Enterprise (PDF), one of four goals identified in the 2015–25 Strategic Plan (PDF), updated in December 2020.

/Public Release. This material from the originating organization/author(s) may be of a point-in-time nature, edited for clarity, style and length. The views and opinions expressed are those of the author(s).View in full here.