Minnesota sharpens modeling estimates of COVID-19

Social distancing isn't working as well as originally predicted in the COVID-19 pandemic, but it will still help protect Minnesota from the highly infectious coronavirus, new state modeling research has shown.

A statewide stay-at-home order has reduced the predicted number of COVID-19 deaths, researchers from the University of Minnesota and the Minnesota Department of Health reported Wednesday. However, the order reduced face-to-face contact and disease transmission by 55%, compared with the 80% projected in earlier modeling.

The latest modeling also predicted 29,000 COVID-19 deaths over the next 12 months under current social distancing conditions, compared with 57,000 if the state had done nothing. The toll is higher than what an earlier state model predicted.

The latest model also predicted 1,700 deaths in Minnesota by the end of May — though with 638 COVID-19 deaths as of Wednesday the state is behind that pace.

"We hope these are pessimistic estimates, but it is certainly a plausible outcome," said Stefan Gildemeister, state health economist.

Gov. Tim Walz received the modeling data earlier this week, and weighed it as part of his decision Wednesday night to end the current stay-at-home order. Retailers with social distancing plans can open Monday, and bars and salons can start planning to open June 1.

Group gatherings must be smaller than 10 people, though, and Walz urged the public — especially the elderly and those with health problems — to remain cautious.

The latest modeling predicted 29,000 deaths over the next 12 months, but with a range from 16,000 to 44,000. State health officials stressed that the modeling was never intended to forecast specific death counts, but rather to assess whether changes in social policies and restrictions would cause those numbers to decline.

One scenario estimated a lower total — 22,589 deaths over 12 months — if the state maintained a stay-at-home order through May, and relied on 20,000 diagnostic tests per day to identify cases and prevent widespread outbreaks in communities and businesses.

That scenario was based on an assumption that tests are 90% accurate in identifying infected people. A separate model estimated results with only 70% testing accuracy.

Modeling for the first time also analyzed the impact of Minnesota following guidance by the U.S. Centers for Disease Control and Prevention, and only pulling back on social distancing when its COVID-19 hospitalizations had declined for 14 days.

The state would see 26,294 COVID-19 deaths in the next year under this scenario, but would spread out cases and dramatically reduce the pressure on hospitals and their ICU bed capacity.

Walz has viewed the CDC recommendation as conservative for Minnesota, where he hopes a wealth of testing and medical resources will allow for a faster rollback of stay-at-home restrictions when businesses and organizations can offer adequate social distancing and safety for workers and customers.

"There is a social-psychological factor of how long people can do this," he said.

Modeling was instrumental early on for Walz, who issued school closures and statewide stay-at-home orders based in part on state-specific forecasts showing that COVID-19 cases would outstrip the available supply of critical care beds and ventilators in hospitals.

Minnesota hospitals have worked to increase ventilator capacity, and now have 3,702 available — though 858 remain on back order. Statewide, 562 ventilators were in use Wednesday by patients with COVID-19 or other unrelated medical problems. Walz said in viewing the latest modeling that he had to estimate the continued economic harms against whether hospitals could add many more beds or ventilators.

The latest modeling predicted that social distancing measures in general — such as staying 6 feet apart and refraining from large groups — would only reduce disease transmission by 38% rather than earlier projections of 50%. Waning public compliance factored into that reduction.

"Mitigation (in China) was very different from what we are experiencing here," he said.

Gildemeister said the latest model was recalibrated with estimates of social distancing effectiveness until its predicted number of cases matched the actual number of cases of COVID-19 spreading today.

"It is possible for these mitigation effects to change further depending on how the population interacts in response to good weather or sort of how compliance varies over time," he said.

Death estimates under the Minnesota modeling continued to be higher than those elsewhere. Modeling by the Los Alamos National Laboratory in New Mexico estimated 1,200 deaths in Minnesota by June 21 and the Massachusetts Institute of Technology predicts 1,527 deaths by June 15.

Columbia University modeling as of May 4 estimated a peak of 47,500 cases in Minnesota by June 15, but a more rapid spread of the virus once normal business and activity levels resume.

Minnesota's model has changed other underlying assumptions beyond the effect of social distancing.

The latest model also was based on feedback from doctors that 100% of people would die from COVID-19 if they needed ventilator care in hospitals and didn't receive it.

On the other hand, the model assumed that patients only needed eight days in intensive care, compared with prior models that predicted 10 or 22 days. The lower number means Minnesota's existing ICU capacity can stretch further than originally estimated.

Gildemeister said that modeling continues to show that social distancing delays the peak of cases and buys hospitals time to prepare. However, the restrictions wouldn't spread out, or flatten, the peak of cases over a longer period of time unless they stayed in place until at least September.

"Maybe that is entirely undesirable and not worth the trade-off," Gildemeister said, "but if flattening is what we are looking for, that is one of the tools that is available to us."