SIOUX FALLS, S.D. — South Dakota is facing a Thanksgiving test in its battle with COVID-19.

The state, home to one of the worst COVID-19 outbreaks in the nation, is at a crucial "juncture," according to modeling from Sanford Health, the South Dakota-based health system.

"As we approach the Thanksgiving holiday, our team is looking at two scenarios — one in which community spread remains consistent, and another scenario in which people choose to gather with multiple people and multiple family members outside of their household," said Andy Munce, Sanford vice president of operations.

"We are at a juncture where we can slow the spread, but we need everyone in our community to do their part to take care of themselves so we can take care of the patients who need us," he said.

November has been particularly grim for South Dakota, even with 10 days left in the month. A surging COVID-19 has killed 280 people in November, raising the state's virus death toll to 705, as state officials reported another 31 fatalities on Thursday, Nov. 19.

Newsletter signup for email alerts

The state Department of Health reported 1,068 people who have newly tested positive for COVID-19, raising the state's total to 69,742 although 51,153 have recovered. Active cases, those who have tested positive and are still considered able to spread the virus, slid to 17,884 on Wednesday, down significantly from a peak of 19,360 cases on Sunday.

There are 578 people hospitalized in South Dakota due to COVID-19, down from a peak of 607 cases on Nov. 10 but nearly double the hospitalizations seen a month ago.

South Dakota is second in the nation for the number of new COVID-19 cases per capita in the last 14 days, behind only North Dakota, according to an analysis of state data from The New York Times. The state has also marked new daily records in positive cases, active cases, and hospitalizations, taxing hospitals and exhausting healthcare workers.

Thanksgiving could prove a unique, and unwelcome, challenge for South Dakota, whose approach -- relying on individuals to take steps to avoid spreading COVID-19 -- has clearly proven unsuccessful. Nearly two in five Americans are likely to attend large holiday gatherings, according to a national survey conducted by the Ohio State University Wexner Medical Center.

And with about one in 46 South Dakotans current known to be positive with COVID-19 and able to spread the virus, the state is a dangerous hotspot for the risk of spreading the virus via holiday get-togethers, according to the COVID-19 Event Risk Assessment Planning Tool developed by the Georgia Institute of Technology.

Sanford Health is a Sioux Falls-based health system with major medical facilities in South Dakota, North Dakota and Minnesota. Since early March, its data and analytics team has used algorithms to model the spread of COVID-19 in the state, providing projections crucial to hospital administrators and state health officials alike.

Munce said hospitalizations lag about two weeks behind increases in positivity rates in testing, and the months of analysis have strengthened the health system's modeling.

"Over time, the algorithms get smarter as we see how this disease reacts and we know how many hospitalizations admissions we have," Munce said. "Every week our team predicts what the next seven days will look like with good reliability."

South Dakota has no statewide COVID-19 restrictions, and Gov. Kristi Noem has repeatedly said she doesn't believe she has the authority to issue a masking mandate, a move recommended by many public health experts since research shows wide-scale masking can help reduce the spread of the virus.

In the last week, several South Dakota cities have approved local mask mandates, joining Brookings, which has had a mask mandate for several months.

On Tuesday, the city council in Sioux Falls, the state's largest city, approved a mask mandate for those in retail businesses and city owned facilities. The council members whose votes proved decisive in the vote cited support from healthcare leaders as crucial to their decision.