Barely a week after the fall 2020 semester kicked off on the University of Wisconsin-Madison campus with in-person instruction, new confirmed cases of COVID-19 on Sept. 10 surged well past previous single-day records in Dane County, home to the university, the state capital and the second largest metro area in Wisconsin.

"This is a dramatic increase," warned Public Health Madison & Dane County in a tweet that morning. The local health department announced there were 456 more confirmed cases of COVID-19 in Dane County than the previous day. A few hours later, Dane County Executive Joe Parisi issued a public letter to UW-Madison Chancellor Rebecca Blank and UW System Interim President Tommy Thompson urging them to send undergraduate students home, citing the same figure.

But in its own daily update on the status of COVID-19 in the state, the Wisconsin Department of Health Services announced an even higher number of newly confirmed cases in Dane County for Sept. 10: 482.

Nearly a week later, an online data dashboard where the local health department updates information about the spread and impact of COVID-19 in Dane County showed 294 new cases on Sept. 10 while attributing 487 new cases to the day prior.

So, which of these figures is correct?

It's a straightforward question that could be expected to have a simple answer. But it isn't quite so simple because, as it turns out, all three figures are accurate — even if on the surface they seem contradictory.

How is this possible?

There are a couple key reasons for why daily state and local health department updates about COVID-19 reports often appear different: timing and definitions.

Health officials across Wisconsin report new coronavirus cases based on varying 24-hour periods, and they use several definitions for "new cases" when assigning them to specific dates. As a result, discrepancies between daily state and local COVID-19 data abound in Wisconsin. But that doesn't mean the information is inaccurate or unreliable — it simply means a little explanation is needed to make sense of it.

Let's break it down.

Just like other reportable diseases, when someone in Wisconsin tests positive for the coronavirus that causes COVID-19, the diagnosis must be reported to local public health authorities. This reporting usually happens through the statewide Wisconsin Electronic Disease Surveillance System. The secure online system allows healthcare practitioners and laboratories around the state to quickly upload basic details about new COVID-19 cases, including the place and time they're identified.

In the case of COVID-19, local health departments use this information for a number of purposes like contact tracing and deciding where to target testing and informational campaigns. The state health department also keeps close tabs on local information, which it uses to inform its broader response to the pandemic in Wisconsin.

Wisconsin's reporting requirements for COVID-19 are strict: As a Category I communicable disease, confirmed and suspected cases must be reported to local health officials "immediately" by phone, and a full case report must be filed via fax, mail or, most commonly, the electronic surveillance system within 24 hours.

These case reports form the basis of the state's and many local health department's daily public COVID-19 updates, including Public Health Madison & Dane County's Sept. 10 update highlighting the surge in local cases largely affiliated with UW-Madison.

Many, though not all, local health departments in Wisconsin regularly share COVID-19 statistics. Among the ones that do, most update information about new confirmed cases, deaths and hospitalizations in their jurisdictions, while some provide additional demographic and testing information.

Even though both state and local health departments use the same data to inform their daily updates, they do not necessarily pull new case reports from the electronic system at the same time. As a result, the number of new cases they report on a day-to-day basis can vary between days. The discrepancies are liable to be especially pronounced during the course of a surging local outbreak, when new case reports flood the system, sometimes on an hourly basis.

Public Health Madison & Dane County and other local health departments tend to pull new case reports first thing in the morning so contact tracers can start their work as quickly as possible. Meanwhile, the Wisconsin Department of Health Services, which isn't directly responsible for contact tracing, usually begins downloading new case reports a bit later.

Jennifer Miller, an agency spokesperson, said compiling the data is a two-hour endeavor every day and usually begins by 9 a.m., though it sometimes occurs as late as noon.

The state health department updates its public statistics every day at 2 p.m., while local health departments that provide daily updates refresh their statistics at various times. For example, Public Health Madison & Dane County does so at 9:30 a.m., the Milwaukee County Department of Health & Human Services releases its data at 1:30 p.m., and Waukesha County Public Health updates its data at various times of day.

This timing helps explain why the state initially reported dozens of more new cases in Dane County for Sept. 10 than the local health department did. There were simply more new case reports in the system by the time the state pulled data than when Dane County health officials did so earlier in the morning.

But that still doesn't explain why Dane County's COVID-19 data dashboard attributes a far higher number of new cases to Sept. 9 than Sept. 10. This difference has to do with how health authorities define new cases. For local epidemiological purposes, it's more informative to match new cases with the date of a positive test result — not the date the test result was publicly reported.

"If we receive a test result that is three days old, we update the number of case counts from three days ago," said Christy Vogt, a spokesperson for Public Health Madison & Dane County. It's one reason why health departments often label recent data — usually from the latest two-week period — as provisional.

"Numbers for recent dates are subject to change as tests continue to be processed," Vogt added.

Most of the 456 new cases Dane County announced on the morning of Sept. 10 had test result dates from Sept. 9 or earlier, and as even more test results from Sept. 9 trickled in over subsequent days, the record-breaking number grew to 487.

At the same time, UW-Madison maintains its own online COVID-19 data dashboard that it updates daily with new cases, though it assigns the new cases to the date the test results are reported to the university — not the date of the test result. This approach can help explain why UW-Madison's count of 280 new cases for Sept. 10 is only 14 cases fewer than new cases for that date reported in all of Dane County.

(WisContext maintains a daily COVID-19 tracker based on data updated daily by the state health department, which is why it shows 482 new confirmed cases for Dane County on Sept. 10.)

Vogt noted one more difference between the COVID-19 statistics shared by Dane County and the state health department: The agencies calculate test-positivity rates differently. This metric tracks the percentage of COVID-19 tests that come back positive and can be an important indicator of whether there is enough testing in a community.

Vogt said part of the difference in test-positivity rates reported by Dane County and Wisconsin as a whole lies in the different ways they attribute cases to dates when they update daily counts.

"But part of the reason is because our denominator includes all tests … and DHS's denominator includes only people tested," Vogt said.

This difference doesn't affect the calculations too much since most people aren't tested multiple times in a short period of time. However, Vogt said the county decided to calculate the rate this way to "best capture testing capacity" in its jurisdiction.

These data discrepancies, while due to legitimate disease reporting practices, can cause confusion, or worse, they could lead watchful members of the public to become suspicious of health authorities — a common theme of the highly-politicized COVID-19 pandemic.

These differences are all the more reason to focus on longer-term trends rather than daily data updates, according to Amanda Simanek, an epidemiologist at the University of Wisconsin-Milwaukee who helps advise the pandemic response in the state's largest metro area.

"It's better to look at a seven-day trend," Simanek said. "Because some of those anomalies get ironed out."

Health authorities in Wisconsin's largest city have moved away from providing daily data updates in part for this very reason, according to Jeanette Kowalik, the outgoing commissioner of the Milwaukee Health Department. She said Milwaukee shifted from daily updates to two per week.

Kowalik noted that the less frequent updates have provided the city's department with some breathing room to address occasional problems with the data like incomplete case reports. She also said biweekly updates provide a clearer picture of local trends for the disease's spread and lamented a lack of clear data reporting guidance from federal health officials.

Kowalik described the varying data reporting practices as "so unorthodox for public health because we typically follow the lead of the feds."

"That chain has been broken, and everyone has had to somewhat fend for themselves," she said. "So there's so much confusion across jurisdictions and across states about what [data] really matters [and] how we present it."

And while the Wisconsin Department of Health Services has some authority to set reporting standards, state law limits the agency's ability to dictate how and when local departments share data publicly.