By Anne Blythe
North Carolina’s public health team recently noticed some anomalies in the daily data dumps that LabCorp has been providing the state to help track COVID-19.
That prompted data analysts to probe further into the differences they were seeing between electronic and manual reports, and as a result, the state Department of Health and Human Services announced corrections on Wednesday to daily and cumulative COVID-19 test counts.
North Carolina has revised downward the total number of tests performed during the pandemic to 1,823,283 from the 2,044,727 previously reported. That 221,444 overcount in tests performed, public health leaders contend, does not have an impact on the key metrics and trends North Carolina uses to guide pandemic decisions.
Brian Caveney, chief medical officer and president of LabCorp Diagnostics, said in a statement that the DHHS team had informed the North Carolina-based company last week of the inconsistencies.
“We determined that from late April until last week, Pixel by LabCorp at-home test collection kits that originated out-of-state but were processed in North Carolina were inadvertently included in the manual data submission to the state,” Caveney said. “We quickly corrected the issue and provided the updated manual reports to NCDHHS. LabCorp’s daily electronically reported data was accurate and unaffected by the error, and this issue does not affect other states or any results reported to patients or their providers.”
The glitch and larger impact
While it might seem that the number of cumulative tests reported on the DHHS dashboard would have an impact on the percentage of tests that are positive because the baseline number would shift, Health and Human Secretary Mandy Cohen explained how that’s not the case.
It turns out COVID-19 data comes to the state in two forms: one electronically filed from the direct source and the other submitted by phoning or faxing it in and then entered manually.
The public health data entry team only manually enters test numbers and positive results for labs that do not submit their data electronically. Since LabCorps also submits electronic data, those correct numbers populated the DHHS dashboard automatically, according to Cohen.
About 80 percent of the labs submit their data electronically, she said, and it is only that information that DHHS uses to figure out the percentage of cumulative tests that are positive.
“Because we wanted to have that apples to apples numerator, denominator for only those labs that submit to us electronically,” Cohen explained.
For the 20 percent of the labs that report their data through surveys that require manual entry, the DHHS team only collects positive lab case numbers but not negative test results and they do not add those labs into the total when calculating the percent positive rate.
“So when we go to capture the percent positive, we only use a subset of our total lab data,” Cohen explained. “For those of you who have been deep in the weeds with us on our data have known, and we have it posted on our website, that our percent positive has always been calculated from the labs that submit to us electronically.”
How the error happened
Nonetheless, the inconsistencies come in the middle of a political year when mixed messaging about the coronavirus pandemic has bred skepticism and mistrust over varied state response plans in the absence of a national strategy.
People with little to no formal education in science and health care have become armchair epidemiologists and data crunchers, poring over every last number to support or criticize state response plans.
North Carolina initially took an aggressive stance against the novel coronavirus and closed down public schools, businesses and ordered people to stay at home except for essential visits to grocery stores, pharmacies and health care facilities.
Currently, the state is in an extended Safer At Home Phase 2 plan, which keeps bars, gyms, playgrounds, bowling alleys and large entertainment venues closed because they are deemed to be high risk for the spread of COVID-19.
When Gov. Roy Cooper extended the order last week into September, he said that while the public health team had noticed glimmers of hope in North Carolina’s trends and metrics, they wanted to monitor the same data as public schools and universities bring students back.
In a briefing with reporters on Wednesday, Cohen said she and her team are committed to data transparency and will work to smooth out wrinkles in a system that allowed for the error.
“We are in a pandemic and had to stand up a lot of data systems very quickly and we need to improve them over time,” Cohen said. “You have my commitment to continue to be transparent about what we are doing and always answering your questions.”
Cohen said two teams in her department have been involved with gathering and reviewing data from hospitals, labs and emergency departments to help make sense of the overall impact of COVID-19 during the pandemic.
But it wasn’t until this past week, when they came together as part of a DHHS effort to shift the process to even more electronic reporting, that they found the discrepancies in overall LabCorp tests.
One DHHS team was involved with manual reporting and the other dealt mostly with electronic data fed directly into older DHHS systems ill-prepared for the amount of information flow necessary to keep the public well-informed during a pandemic.
“It wasn’t until we brought those teams together last week as we tried to do some process improvements to take away that manual process that we saw the discrepancy in the numbers,” Cohen said. “When we did, we brought that to LabCorp. They quickly identified what the issue was and then we’ve been working to sort of fully scope out that data issue, and you can see on our web site an Excel sheet that takes you day over day the corrections we made.”
Finding good news in numbers
While the numbers do not change most of the key metrics that the state has used to guide decision-making during the phases of reopening, the discrepancy does show that North Carolina is not conducting as many coronavirus tests as the public health team would like to see.
“What we are actually seeing is there has been a slowdown in testing throughout our region,” Cohen said. “I reached out personally to a number of other state colleagues to understand what they’re seeing in their testing. They’ve also seen a slowdown.”Cohen acknowledged that North Carolina needed to ramp up testing even more.
“We know that testing turnaround times had started to balloon back in July,” Cohen added. “We are hypothesizing that folks heard about those longer testing times, and maybe if they are going to be more inclined to get a test, maybe they became less inclined to get a test. So we want to make sure that folks realize testing turnaround times have improved.”
The time that people now wait for test results is closer to two to three days, a steep decline from the seven- to 12-day windows that some labs were experiencing.
As part of its COVID-19 attack plan, the state has helped organize or sponsor free testing in communities of color and in other places where virus spread has been nettlesome.
More charts and graphs on horizon
Cohen gave a peek at a presentation she plans to make on Thursday, outlining key trends and metrics.
Several weeks ago, North Carolina’s case numbers were going up rapidly and the percentage of tests that were positive hovered at 10 and 11 percent.
During the early part of this week, the number of new cases was not going up as quickly as they were in July. Lab-confirmed cases dropped into the hundreds for the early part of this week, but Cohen said Wednesday that they had topped 1,000 again by midweek.
“The drop in the number of new positives is absolutely not impacted by this data discrepancy and this data change that we are making,” Cohen stressed. “The lower number of cases is the fact that we’ve been working hard on our trends. We’ve been working hard to slow the spread of this virus, and I think this continued progress that we’re seeing.”
Hospitalizations have been declining slightly, though there still were 1,062 people in hospital beds on Wednesday with severe illness related to the coronavirus.
Cohen stressed again that her department will be as transparent as it can about the data collected from LabCorp and other players in the state’s coronavirus response.
“At the end of the day, the accuracy and quality needs to be both at the state level but also at our partners that are submitting that data,” Cohen said. “The responsibility for quality is on all of us, whether it’s hospitals or labs or others who submit the data to us. We’ll continue to be focused on data quality and data transparency as we go forward.”
Coronavirus by the numbers
According to NCDHHS data, as of Wednesday afternoon:
- 2,204 people total in North Carolina have died of coronavirus.
- 137,895 have been diagnosed with the disease. Of those, 1,122 are in the hospital. The hospitalization figure is a snapshot of people hospitalized with COVID-19 infections on a given day and does not represent all of the North Carolinians who may have been in the hospital throughout the course of the epidemic.
- 116,969 people who had COVID-19 are presumed to have recovered. This weekly estimate does not denote how many of the diagnosed cases in the state are still infectious.
- To date,1,823,283 tests have been completed. As of July 7, all labs in the state are required to report both their positive and negative test results to the lab, so that figure includes all of the COVID-19 tests performed in the state.
- People ages 25-49 make up the largest group of cases (44 percent). While 13 percent of the positive diagnoses were in people ages 65 and older, seniors make up 80 percent of coronavirus deaths in the state.
- 331 outbreaks are ongoing in group facilities across the state, including nursing homes, correctional and residential care facilities.
- There are 3,244 ventilators in hospitals across the state and 897 ventilators in use, not just for coronavirus cases but also for patients with other reasons for being in the hospital.
Republish our articles for free, online or in print, under a Creative Commons license.
