Fact check: Study projects coronavirus peak, then moves the dates

Eric Litke, USA TODAY
·5 min read

Specifics have been hard to come by in this pandemic.

We don’t know how many people are infected, how fast the coronavirus is spreading or how long we may need to quarantine.

So the first study to project a date when each state’s cases would peak drew a lot of interest — especially since Wisconsin jumped off the virtual page.

The study, released March 26, said cases wouldn’t peak in Wisconsin until May 22 — a month later than most states, and nearly two weeks later than any other state in the nation.

The date went viral. News outlets across the state picked it up, and one write-up of the story from Madison365.com received nearly 40,000 comments, reactions and shares on Facebook in a day.

And then the study changed.

An updated analysis posted four days after the original study moved Wisconsin’s projected peak by almost a month to April 26.

So what’s up with the shift? Which date is right?

Here’s what we found.

The claim: A study projects Wisconsin's coronavirus cases will peak April 26

The analysis from the Institute for Health Metrics and Evaluation — an independent research center based at the University of Washington — developed a model projecting coronavirus growth across the United States for the next four months.

It projects deaths as well as a variety of measures of hospital utilization — beds used, intensive care unit days and ventilator days used. That estimate was based on COVID-19 deaths so far, data on hospital capacity and use and more detailed coronavirus hospitalization data available in some localities.

The institute estimates the pandemic would peak nationally around mid-April and result in more than 80,000 deaths, a summary of the study said.

The state-by-state projections were based on a statistical model — essentially a detailed formula that projects the future virus spread based on death rates and other data available to this point.

And those projections moved a lot after the initial release.

The update

Four days after the original publication, on March 30, researchers added 1,730 nationwide deaths to the model and tweaked the methodology. They said the updates created more precise estimates.

The updated projections changed the estimated peak for all but four states. Wisconsin moved the most, to 26 days earlier, but several others saw dramatic shifts as well. Maryland’s estimated peak moved to 25 days later, Missouri to 21 days later and Idaho to 13 days sooner.

Ali Mokdad, a senior faculty member at the institute, told the Milwaukee Journal Sentinel the original model — started around the beginning of March — assumed Wisconsin would have a Safer at Home-style order in place about two weeks sooner than it did. Gov. Tony Evers issued the order banning all but essential services and travel on March 24.

The update factors in that actual date along with additional death data, which dramatically changed the model’s analysis of the state.

The new projected peak for Wisconsin — April 26 — is still later than what state health officials are saying. Wisconsin Department of Health Services Secretary Andrea Palm said March 30, the state expected the curve to begin flattening (i.e. peaking) in about 10 days.

"It’ll be a couple of weeks before we start to really understand how well we are doing, social distancing, safer at home, to really judge ourselves against the worst-case models," Palm said in a media briefing with Evers and other officials. "We really do believe it’s another 10-plus days before we are going to be able to see evidence of a flattening off of the new daily cases."

The viral story from Madison365.com quickly updated to reflect the new numbers, so Facebook posts of that story shared before or after that update now show a headline that accurately references the new April 26 peak.

Instead of being the latest peak in the country, Wisconsin is now right in the middle of the pack — tied for 20th-latest with three other states. The peak is two weeks after the projected national peak of April 15.

Our ruling: True

This is an unusual fact check, since at the time most of those stories were shared, the post referenced a May 22 date that is no longer the best estimate from this study. But Facebook’s design updates a linked story when the story is updated. We aren’t getting into whether this is an accurate model, since it’s a prediction of the future and there’s no way to say for sure. But we can examine the accuracy of the Facebook post itself. What is currently online in this viral post is an accurate reflection of the study, albeit one that requires a bit of extra context to understand the genesis of the current figures. We rate this TRUE.

Our fact-check sources

Contact Eric Litke at (414) 225-5061 or elitke@jrn.com. Follow him on Twitter at @ericlitke.

Thank you for supporting our journalism. You can subscribe to our print edition, ad-free app or electronic newspaper replica here.

This article originally appeared on USA TODAY: Fact check: Changing coronavirus peak projections