Inconsistencies and problems with Covid-19 data collection in Texas have clouded the picture of the pandemic’s trajectory in the state, to the point that some residents and officials say they cannot rely on the numbers to tell them what is really going on.
The state has overlooked thousands of cases, only to report them weeks after infection. It has made major adjustments to its case and death counts, defining them one way and then another, suddenly reporting figures for some counties that were vastly different from those posted by the local health department.
“The changing of gears and data reporting at the state has a lot of public health departments feeling a significant case of whiplash,” Mayor Ron Nirenberg of San Antonio said.
The daunting task of reporting coronavirus cases and deaths in real time has strained public health departments across the country. But none have shown more repeated cracks than Texas.
The state has faced these data problems as infections surged in the summer and schools and colleges began to reopen for the fall. Changes in the state’s figures have been large enough to affect national trends, and have sown confusion and distrust at a time when the state says it needs public support to avoid another surge.
“If everyone was counted in a timely way, then maybe people would be more careful, and maybe people would understand that this is real,” said Debra Zukonik of Rockwall County, where the deaths of her brother and a friend did not appear to be reflected in state and local figures for weeks afterward.
Cases and deaths peaked over the summer in Texas and have been trending downward since then, according to a New York Times database that uses both state and county figures. But the virus is still spreading rapidly in many areas of the state, with an average of more than 3,500 new cases a day in September.
Public health officials and researchers place the blame for the state’s data problems on Texas’ antiquated data systems and a reliance on faxed test results, which limit the state’s ability to track every infection and death in many of its 254 counties. They also say that the vast state’s decentralized structure — with many local governments, some of them tiny, running their own public health operations — is ill-suited to coping with the crush of Covid-19.
“It’s a colossal undertaking, and because it’s happening in real time, there will inevitably be situations where we have to update or correct something,” said Chris Van Deusen, spokesman for the Texas Department of State Health Services.
“Texas has these 57 strong, independent organizations that will do what they do,” he said, referring to the state’s local health departments. “That’s just the situation we’re in.”
Other states have also grappled with some of these data challenges, including glitches and backlogs in California and a debate over transparency in Florida. But Texas, even as it has tried to make improvements, has been troubled by multiple issues.
Cases lost and later found
In August, after upgrading one of its reporting systems, Texas discovered backlogs of thousands of positive test results that the state and its laboratories had overlooked or miscoded from late spring and summer. That meant case counts in some counties had been artificially low for months — and would be artificially high in the final weeks of August and into September as the omitted cases were belatedly added.
Dallas County reported 5,361 confirmed cases on a day in mid-August when the county had just 166 actual new cases; the rest were from the state’s backlog. The state has added a few thousand more backlogged cases since then, many dating from June or July.
“It’s definitely very frustrating for all of us,” said Dr. Philip Huang, director of Dallas County Health and Human Services. “If there’s a big backlog or delay, it doesn’t reflect the actual number of people who test positive today, or even two days ago.”
Public health experts and local officials say the backlogs represent thousands of missed chances to perform timely contact tracing, and have given the public reason to be skeptical of the data.
But state officials said that knowing about the backlogged cases earlier would not have changed the overall trend in the state or the guidance given to the public. And they say the reporting system upgrade, despite its flaws, shows that Texas is moving in the right direction.
“The data that we have now is far more accurate than what we had last month, and maybe even a couple of weeks ago,” Gov. Greg Abbott said on Aug. 23, after the backlog issue came to light. “That data should be very close to being perfectly accurate in the coming weeks and coming months.”
Even as the backlogs subside, confusion about cases persists because state and local health departments do not report them uniformly. While the state includes confirmed cases based on a molecular test, at least 17 counties also report probable cases based on antigen testing, symptoms or exposure to an infected person, as recommended by the Centers for Disease Control and Prevention.
The case count differences raise doubts among the public, said Annette Rodriguez, health director for the Corpus Christi Nueces County Public Health District, which reported 4,000 more cumulative cases than the state on Sept. 11.
“It makes them question: Who’s right? Who isn’t? Who’s credible?”
Counting deaths one way, then another
In late July, Texas changed the way it tallies coronavirus deaths. It started basing its counts on death certificates filed with the state, instead of gathering figures from local health department websites, which often rely on reports from hospitals and physicians. The change was intended to make the tallies for all 254 counties consistent and comparable, but it has distorted data for several locations, confusing residents and local officials alike.
The change led the state to add 675 deaths to its tally in a single day — a jump plainly visible on the national death curve.
Then, the next day, the state lowered its new figure by more than 200, citing an “automation error.”
The state’s tally for Harris County, which includes Houston, took one of the largest leaps: from 636 deaths on July 26 to 1,100 on July 27. In the same 24-hour period the county noted only 10 deaths; the spike was due mainly to the state’s change of method.
In other counties, the state’s new method yielded lower figures than local health departments’ websites were reporting. The day after the change, Hidalgo County in the hard-hit Rio Grande Valley said its cumulative death toll was 531; the state’s figure was 254.
Hidalgo officials said the discrepancy was probably caused by delays of up to three weeks in issuing death certificates in the county. In a detailed letter on Facebook, they said the state had made the change without consulting them.
Mr. Van Deusen, the state public health spokesman, said it “could have done a better job communicating about that to the local jurisdictions.”
In one of more than 100 comments posted on Facebook in response to the county’s letter, a Facebook user wrote, “Do you understand that these discrepancies lead us all to think that you are hyping the numbers?”
Dave Montgomery contributed reporting.