6 states might be manipulating COVID-19 data as they reopen, reports suggest — here's what can skew the numbers

tsonnemaker@businessinsider.com (Tyler Sonnemaker)
·6 min read
coronavirus sample test technician france
A technician scans test tubes containing live coronavirus samples in the laboratory for the Robert Ballanger hospital in Aulnay-sous-Bois near Paris, France, April 30, 2020.

Gonzalo Fuentes/Reuters

As new coronavirus case counts appear to slow and decline in parts of the US, all 50 states have started easing lockdown restrictions.

Public health experts have repeatedly warned that states risk triggering new waves of cases if they reopen prematurely, without robust information about who's infected as well as data on how and where the virus is still spreading.

However, news reports suggest that six states — Texas, Virginia, Vermont, Florida, Georgia, and Arizona — might be fudging their numbers in various ways that drum up support for decisions to lift stay-at-home orders.

Here are the decisions each state has made that might skew coronavirus numbers.

Texas, Virginia, and Vermont have reportedly combined numbers for two different types of coronavirus tests, thereby making their testing capacities look artificially inflated.

greg abbott coronavirus
Texas Gov. Greg Abbott.

Tom Pennington/Getty Images

Officials in Virginia, Texas, and Vermont reportedly combined the numbers of completed tests for two separate types: diagnostic tests that identify people currently infected, and antibody tests that can tell whether people have recovered from the virus. 

Mixing the two counts can make it seem like these states have been testing more people than other states that are only counting diagnostic tests. The move to combine the totals into one also makes it difficult to identify the trends in daily case counts relative to diagnostic tests completed.

The Associated Press reported on Tuesday that officials in Vermont and Virginia have since stopped combining the figures in this way.

Virginia Gov. Ralph Northam told the AP the correction caused "no difference in overall trends."

Texas Gov. Greg Abbott said Monday that the state wasn't mixing together its figures, despite state health officials saying last week that they had in fact been treating the data this way, the AP reported.

Combining the results of both types of tests can shed light on the total number of infections (assuming people are not double-counted), but doing so can also hinder officials' abilities to make data-based decisions about easing restrictions based on the latest number of active infections.

Texas and Virginia are still seeing a rise in their new cases each day, while Vermont's appear to be plateauing, according to data from the COVID Tracking Project.

A top researcher in Florida says she was fired after she refused to 'manually change data to drum up support for the plan to reopen.'

MIAMI GARDENS, FLORIDA - MAY 06: Florida Gov. Ron DeSantis takes his mask off as he prepares to speak during a press conference at the Hard Rock Stadium testing site on May 06, 2020 in Miami Gardens, Florida. Gov. DeSantis announced during the press conference that a COVID-19 antibodies test will be available. The test can show if a person has had the virus in the past without showing symptoms, and therefore may be immune to it. The test will be available to first responders and health care workers first, with the goal of being able to expand testing to the general public.
Florida Gov. Ron DeSantis.

Joe Raedle/Getty Images

Rebekah Jones, who developed Florida's coronavirus-tracking dashboard, told WPES, a local CBS affiliate, that she was fired because she refused to "manually change data to drum up support for the plan to reopen."

As Florida has started reopening, some data has reportedly gone missing, and access to the raw data used to generate dashboards has become a challenge, Florida Today reported. Jones said in an email to Florida Today that the new team in charge of the dashboard may not be as committed to "the same level of accessibility and transparency."

A Florida health department spokesperson told Business Insider, however, that Jones was fired because she made "unilateral decisions to modify the Department's COVID-19 dashboard without input or approval from the epidemiological team or her supervisors."

The White House's guidelines for reopening say states should wait until they've seen 14 consecutive days of declining total cases — or declining positive tests as a percent of total tests. Florida officials say that aside from two spikes in April, case numbers are declining. Jones' firing, however, has led some to wonder whether the state has been publishing accurate numbers.

A recent chart from Georgia's Department of Public Health falsely depicted a decline in cases in the state's hardest-hit counties because it mixed up times and locations.

georgia governor brian kemp coronavirus mask
Georgia Gov. Brian Kemp (middle).

Ron Harris/AP Photo

A corrected version of the graph showed declines weren't actually being observed. The Atlanta Journal-Constitution also pointed out other data mishaps in Georgia, including incorrectly reporting case and death counts as well as changing the color scale used in a county-by-county case map such that fewer appeared to have severe outbreaks.

The reporting also suggested that Georgia's entire reopening strategy may be based on a fundamental misinterpretation of its data.

Starting in April, Georgia began assigning new positive cases to the date a patient first reported symptoms rather than the day their test came back positive — and due to a lag in its data collection efforts, the new method appears to underreport recent cases at first, according to the Journal-Constitution. 

Two researchers from the Georgia Institute of Technology wrote in an article for Slate that Gov. Brian Kemp appeared to take that data at face value when he decided to let businesses reopen.

Georgia hasn't seen a spike in cases since reopening.

Arizona told its experts to stop making projections days before the state reopened since they had suggested the outbreak there had not yet peaked.

WASHINGTON, DC - JUNE 13: U.S. President Donald Trump (C) speaks as Iowa Gov. Kim Reynolds (L) and Arizona Gov. Doug Ducey (R) listen during a working lunch with governors on “workforce freedom and mobility” at the Cabinet Room of the White House June 13, 2019 in Washington, DC. President Trump spoke about the administration’s work to overhaul occupational licensing laws, job training programs and child care policies at companies around the country.
Arizona Gov. Doug Ducey.

Alex Wong/Getty Images

Arizona's health department told a team of experts from two of the state's universities to stop their work after they predicted COVID-19 cases had not yet peaked in the state.

The modeling team, made up of at least 23 experts from Arizona State University and the University of Arizona, had estimated in late April that the peak of the outbreak might not come until May 22 or later.

The order came hours after Gov. Doug Ducey announced he would lift some restrictions on businesses in the coming days, and a day before President Donald Trump visited the state.

Ducey said he had "confidence that we are going in the right direction," and The Washington Post reported that the state plans to rely on federal projections that are unavailable to the public.

Models can vary widely depending on how they incorporate different sources of information and are constantly revised as new data become available, but they can still be a useful tool for policymakers.

Arizona is still seeing a rise in new cases, according to the COVID Tracking Project.

Sinéad Baker, Haven Orecchio-Egresitz, and Morgan McFall-Johnsen contributed reporting to this story.

Read the original article on Business Insider