REGIONAL— As the state makes moves to re-open the economy after Gov. Tim Walz lifted his stay-at-home order last Sunday night, a revised Minnesota pandemic model predicts the COVID-19 peak will come sooner and that more people could ultimately die than its projections suggested in April.
Revealed in a press conference last week, the adjusted model now projects 29,000 COVID-19-related deaths in Minnesota over a 12-month period, a jump of 7,200 over its previous estimate. By the end of May, total deaths in Minnesota could be between 1,400 and 2,000, more than double the deaths reported as of last Friday. And the peak for maximum intensive care usage shifted from mid-July to June 29.
“I share the hope with you that we’re wrong and this is a pessimistic outcome of the model,” said Stefan Gildemeister, a health economist with the Minnesota Department of Health. “Nevertheless, I think it is a plausible outcome based on what we understand about the rate of transmission of the disease, the rate of contacts, the rate of the mortality in the population as is, and sort of generally the disease dynamics.”
According to Gildemeister, the higher estimate of deaths reflects the fact that researchers had originally overestimated the extent to which Minnesotans would follow social distancing and stay-at-home orders. The modeling team initially projected that 50 percent of people would comply with social distancing guidelines, but that number was revised in the latest model to only 38 percent. They assumed 80 percent would follow stay-at-home guidelines, but that, too, was revised downward, to 59.5 percent. In both cases that means social contacts were greater than expected and therefore the virus had more opportunity to spread.
“As we find that there’s more mixing going on in the population than we had assumed, the disease has spread farther and the potential for individuals becoming infected and requiring hospital and ICU care increases, and therefore mortality increases,” Gildemeister said.
One possible indication that recommended guidelines on social distancing and travel weren’t having the expected effect was illustrated in a chart Gildemeister presented comparing total daily deaths projected by the model against actual MDH death reports. The two virtually mirrored each other until April 19, when actual deaths started increasing faster than the model predicted. As of April 26, the last date provided by Gildemeister, the actual death count, 272, was nine percent higher than the 250 projected by the model.
Gildemeister did not suggest that decreased social distancing and excessive travel by Minnesotans were solely responsible for the new model predictions. In fact, he emphasized that numerous aspects of the model were updated based on Minnesota’s experience and new data about the novel coronavirus from the U.S. and abroad. The model was adjusted to account for those who have contracted the virus and can spread it but show no symptoms, and for deaths occurring outside of hospitals, such as in long-term care facilities. Where researchers in some cases had to depend on foreign data in past iterations of the model, it now relies primarily on U.S. and Minnesota data, including state death and hospitalization data through April 26.
“The COVID-19 pandemic remains in its early stages,” Gildemeister said. “It’s been three months in Minnesota. It’s been four or five months across the globe. The epidemic and the evidence still are very new. There is still a great amount of uncertainty about the impact of key metrics.”
A common thread among the scenarios examined in the model by researchers is that the mitigation strategies imposed by Gov. Walz pushed the peak farther into the future, while having only a marginal effect on the projected number of deaths. Walz’s orders, as the governor had explained when he issued them last March, were intended to buy the state time to ramp up its hospital capacity to deal with the expected influx of COVID-19 patients.
“Time doesn’t just buy us the ability to do things differently, but it also buys us time to get a more accurate understanding of what the disease turns out to be like and inform state leaders and the public about that. Time is our friend in many ways,” Gildemeister said.
DOH Commissioner Jan Malcolm reiterated that models aren’t designed to provide point-in-time predictions.
“This is intended to illustrate broad directional trends and changes,” she said. “It’s very tempting when it looks this tidy to draw conclusions about specific points in time or specific degrees of change but that’s not how the model is intended to be used. It’s intended to illustrate relative effect.”
Illustrating that is the variance of uncertainty reported for mortality over 12 months. While 29,000 is the primary projection for deaths, the model allows for as few as 16,000 to as many as 44,000.
“They reflect a mix between capabilities of the model and responses to questions that we’ve had in the past,” Gildemeister said, noting that models are only one piece of information used by decision-makers. “The purpose of this is to educate us collectively and say here’s what could happen under a set of assumptions and a range of interventions by policy makers. These are not recommendations. Modeling serves really an educational purpose for state leaders and the general public to tell us about the likely outcome of the disease given a set of assumptions.”
Gildemeister said the model would continue to evolve as more data about COVID-19 becomes available.