Since December 2019, when we first heard about cases of a novel coronavirus in Wuhan, China, the Arizona public health system has been closely monitoring COVID-19 on an international, national, and local level. Mitigating the spread of COVID-19, as well as responding to the impacts of the virus, remains our highest priority. Since our first case was reported in Arizona back in January, we have been working to protect our populations that are most severely impacted by the disease and preparing our healthcare system for a surge in cases.  While many of the current models show that Arizona’s capacity is sufficient to meet the projected need for hospital beds and ventilators, in order to protect Arizonans, we have been preparing for a worst-case scenario while working to facilitate much better outcomes.

Back in February and March, when we didn’t know as much about COVID-19 as we know now, predictive models were based on the very limited experience and data from our Chinese counterparts in Wuhan and Guangzhou, China. Using Arizona-specific population data and modeling formulas from Harvard, the team at the Arizona Department of Health Services (ADHS), developed our initial Arizona projection looking at the anticipated need for inpatient and intensive care unit (ICU) beds required to treat COVID-19 patients. Based on that initial modeling, an estimated 13,000 additional inpatient beds and an additional 1500 ICU beds would be needed to care for Arizonans with COVID-19. When we calculated the potential spread, we estimated our peak resource needs would fall between the middle to end of April. Using those numbers, ADHS started to develop plans and work with healthcare partners to ensure we would have enough access to care to meet the demand.

As more information about the virus started coming out, additional models became available online. It’s important to note: these models all vary dramatically and are updated as new data is available. The two most prominent are found at healthdata.org and COVIDActNow.org. The models at healthdata.org are developed by the Institute for Health Metrics and Evaluation (IHME), an independent global health research center at the University of Washington. Early in the response, these models were predicting peak resource use around April 20, with 5,342 inpatient beds, 787 ICU beds, and 436 ventilators needed. The IHMEmodels have always been more optimistic than the original ADHS projections, and are updated every couple of days based on the data and mitigation strategies put into place. As of today, April 22,  this model forecasts that Arizona has already passed our peak of resource utilization and only requires 424 inpatient beds, 103 ICU beds, and 92 ventilators*. This is well under our available resources and current hospital capacity.

The COVIDActNow.org model, otherwise known as the U.S. Interventions Model, is a data platform that projects COVID infections, hospitalizations, and deaths across the United States. It was built with input from experts at Google, Stanford University, Georgetown University, and other public health and analytic experts. In its early stages, this model was less optimistic than our initial Arizona projection, predicting tens of thousands of hospitalizations and deaths with our healthcare system becoming overloaded at the end of May. The COVIDActNow model is updated on a regular basis and incorporates current data and the state’s mitigation strategies to come up with newer predictions. Currently, it is showing that Arizona is predicted to be able to meet any COVID-19 healthcare requirements with our current available capacity and our current mitigation strategies in place*.

Over the past several weeks, ADHS has partnered with experts from Arizona State University and the University of Arizona to develop a more targeted, Arizona-specific model, with the most recent update received on Tuesday, April 21. This group of experts has worked on the COVID-19 response with the Centers for Disease Control and Prevention (CDC) and the National Institutes of Health (NIH). This model was extensive, taking into account our current data, mitigation strategies, and potential summer effects on viral transmission. It produced various scenarios that gave us a baseline estimate, with high and low ranges of potential hospitalization and ICU needs of Arizonans. The initial data we received 2 weeks ago from our Arizona team showed an estimated need for hospitalization of 15,000 individuals and 7,000 ICU beds needed. The recently updated version included newer data, and the model shifted. The most recent baseline estimates a peak need for 600 hospital beds and 300 ICU beds around May 22.

Last week, our federal partners from the Centers for Disease Control and Prevention (CDC), the Assistant Secretary for Preparedness and Response (ASPR), and Federal Emergency Management Agency (FEMA) shared the modeling they had built for Arizona based on projection formulas developed by the Johns Hopkins University. This federal model takes into account the current data from the state as well as the mitigation strategies Arizona has put into place. While we are still pending approval from our federal partners to publicly share this data, this model is updated regularly and is the tool the federal government is using when determining resource allocations on a national level. This model predicts our peak resource utilization to occur around June 11, assuming our mitigation strategies are lifted at the end of the current Stay Home, Stay Healthy, Stay Connected order on April 30th. Given that our goal was to reduce transmission of COVID-19, if this model holds true, this later peak would reflect the success of those mitigation strategies. Its projections, even with the mitigation strategies lifted, predict that our current resources, including inpatient beds, ICU beds, and ventilators, will meet a healthcare surge due to COVID-19. This model appears the most realistic and the predictions are reassuring.

As you can see, the model projections vary widely and are highly sensitive to our mitigation strategies. All of our efforts to date have decreased the transmission of COVID-19 in our community and helped our healthcare system increase resource capacity to meet Arizona’s healthcare needs. While most of the models show that we currently have the capacity to meet the healthcare demands for Arizona, it is the responsibility of public health to plan for the worst-case scenario. We want to make sure every Arizonan can access the level of care they need at the time they need it. This is why we are still working on developing plans for alternate care sites, such as the one at St. Luke’s, and facilitating the Arizona Surge Line to help coordinate transfers of patients to prevent surge at any one hospital. As the data evolves, so will our plans.  While the models may try to predict what lies ahead, they are simply predictions. Moving forward, the best course of action is to continue using all of our real-time, Arizona specific data to assess the health of our healthcare system and evaluate the trend of our cases to make decisions that are best for Arizona.

* These electronic models may change by day, so the data presented on the website may not match the numbers posted in this blog.