COVID-19 data from Georgia’s Department of Health could be ‘misleading,’ experts say

Tribune Content Agency

Public health and data experts have raised concerns about the clarity of some COVID-19 data published by the Georgia Department of Public Health less than a week after Gov. Brian Kemp decided not to extend the shelter in place order for most state residents.

Following the update of its website last week, some experts say the changes represent a better attempt by officials to show coronavirus trends a little more than two months after the state confirmed its first cases.

However, some of the new material has caused confusion.

The state health department clarified the way it is counting newly confirmed cases. The measure is tied more closely to when a person begins feeling sick or when a test sample is first collected from a person.

That data, displayed on a chart on the website, may be missing cases from the past two weeks, according to the state health department.

The potentially missing data results in graphs that show Georgia is seeing a sharp decline in new cases over time, although that likely isn’t happening, experts said.

Experts said the state is not clearly explaining its method of reporting new cases, and the chart showing cases over time could be misleading or confusing the public.

“They shouldn’t even put (the last two weeks) on the graph or publish them,” said TJ Muehleman, co-founder of the Standard Co., a data company that assists groups like the World Health Organization with collecting and analyzing various health data. “Any time you publish data that is subject to recasting at a later day, you are going to confuse people. That is a certainty. I’ve seen that with the state of Georgia. … People are confused by this data.”

The new dashboard also left off data the state previously reported to the public, including the number of non-Georgia residents who tested positive for the disease and a table which listed the age and county of residence for those who’ve died.

The Georgia Department of Public Health did not respond to any of McClatchy’s questions regarding the state’s new coronavirus webpage.

But late last week, it did add missing data McClatchy asked about and said researchers were working to make sure the data reported to the public is as accurate as possible.

The concerns come as Gov. Kemp allowed various businesses to reopen despite warnings from some public health experts who aren’t convinced that Georgia has done enough to ease its coronavirus restrictions.

“I continue to believe, in spite of progress to date, that reopening businesses as the governor has begun to do is both reckless and premature,” said Dr. Harry J. Heiman, a clinical associate professor at Georgia State University’s School of Public Health.

WHAT IS NEW THAT MIGHT CONFUSE GEORGIANS?

In its latest update, the state health department defined how it reports newly confirmed coronavirus cases.

The department can use three different measures to help identify the date of COVID-19 cases. They are:

The date when symptoms first appeared.

The date when the first positive test sample was collected, if the date when symptoms first appeared can’t be used.

The date the case is reported, if the dates for symptom onset and first positive sample can’t be used.

The state health department charts these cases under its graph titled “COVID-19 Cases Over Time.”

“This chart is meant to aid understanding whether the outbreak is growing, leveling off, or declining and can help to guide the COVID-19 response,” the graph reads.

However, the state health department acknowledges some data from the past two weeks may be missing from their graph. When asked, representatives for the department did not say how much data might be missing.

The state health department also did not respond to questions regarding how often the state knew when a person’s coronavirus symptoms first appeared or when a test sample was taken, meaning it’s unclear how consistently Georgia is able to backtrack someone’s illness.

Using their new method, the state reported 158 new cases for May 1.

On the same day, McClatchy recorded 1,236 new cases in Georgia. McClatchy has tracked Georgia’s new cases by taking the number of total coronavirus cases confirmed as of 7 p.m. and subtracting that number from the total coronavirus cases reported the previous day at 7 p.m.

Attributing cases to when symptoms first appear is the most accurate way for Georgia officials to assess coronavirus trends, Heiman said.

But, the potentially missing data means the number of new cases over the past two weeks is likely higher than what the state is currently reporting, he added.

“Without a clearer ‘disclaimer,’ I do think this can be misleading to the general public as it will almost always appear that the cases and deaths from the most recent period are declining,” Heiman said. “I would ignore projected trend lines for the past 7-10 days because, as noted, reporting in the most recent 14-day period won’t account for all cases.”

Dr. Thomas Tsai, an assistant professor in the Department of Health Policy and Management at Harvard T.H. Chan School of Public Health, said the state’s new cases graph is “clearly an underestimate.”

“It’s important to base the data on the part of the curve you know is reliable. I wouldn’t draw confidence that there’s (been) a true decline in cases,” he said.

WHAT IS GONE, AND WHAT ISN’T CONSISTENT

In some instances, the Georgia Department of Public Health has contradicted itself when reporting coronavirus data. For a time last week, the total number of coronavirus cases and deaths reported by the state differed on a single web page.

Near the top of the state health department’s web page late on April 28, the state reported 24,865 COVID-19 cases and 1,036 deaths.

However, copying and pasting the county-by-county list a little further down the page resulted in only 23,727 cases and 1,017 deaths.

That left 1,138 cases and 19 deaths unaccounted for. The state was still reporting cases where the county of residence was unknown but was not publishing data on non-Georgia residents.

When asked about the discrepancy, state health officials did not explain why the two data points did not match.

When the state health department published its coronavirus data Thursday at noon, the case numbers for non-Georgia residents reappeared with no explanation. The number of cases on the county-by-county table and on the top portion of the web page matched.

A death list that includes the ages, county of residence, sex and information about underlying health conditions was re-added to the website late last week after disappearing. However, users must download the state’s data reports to receive this information.

“It damages trust,” Muehleman said of the discrepancies. “There have been all kinds of squirrelly data inconsistencies in the last few days.”

“Real-time (data) makes things challenging,” he added. “They need to say ‘whoops, our bad.’ Just admitting when they screw something up goes a long way towards building that trust back up.”

WHAT IS THE STATE DOING TO ADDRESS THE PROBLEM?

While the Georgia Department of Public Health did not answer McClatchy’s questions about the updated website, a representative said aspects of the data on the new web page are being updated continuously.

“The data team is working daily to update aspects of the dashboard to provide an accurate and complete picture of COVID-19 in Georgia, and the progress being made to stop the spread of the virus in the state,” said spokesperson Nancy Nydam.

Muehleman, who is also part of the COVID Mapping Project which tracks coronavirus stats in all U.S. states and a few territories, said the data for Georgia is not 100% correct, but it’s close.

Still, Georgia has some issues with data consistency not seen elsewhere in the nation, he said.

“Every state, you have to look at this data through a bit of a skeptical lens. We haven’t tested adequately enough to know the scope of the (disease,)” he said. “Unlike other states, who’ve glided into a place where the data may not be 100% accurate but it’s at least consistent, Georgia is not necessarily accurate and not necessarily consistent.”

———

©2020 Columbus Ledger-Enquirer (Columbus, Ga.)

Visit the Columbus Ledger-Enquirer (Columbus, Ga.) at www.ledger-enquirer.com

Distributed by Tribune Content Agency, LLC.