[Reporters/editors: If you are looking for a sidebar or feature, below is some information that may be useful for you.]
Rockville, Md. -- Waning COVID-19 vaccine effectiveness. Real-world effectiveness of vaccines against early variants of the virus. Reduced symptoms and viral load among early breakthrough cases. A common thread links these major issues: Abt Associates gathered the critical data for the Centers for Disease Control and Prevention (CDC) to make its scientific, evidence-based conclusions and decisions. We’re doing so with increasing speed to provide data to inform urgent CDC policy decisions, such as the recent recommendations on booster shots. And we provided updated data on vaccine effectiveness for the August 24 Morbidity and Mortality Weekly Report (MMWR).
The task of gathering the data as the pandemic swiftly evolves has posed enormous challenges. Our journey started in May 2020 when Abt proposed to CDC that we use our existing network of clinical healthcare organizations, established seven years ago for surveillance of novel flu viruses, for COVID-19 work.
Abt pivoted the focus to monitor COVID-19 transmission for five groups: first responders, healthcare workers, and essential workers; pregnant women and infants; middle-aged and older adults; older adults in continuing care retirement communities; and households.
Then when vaccines became available earlier than anticipated, the task expanded to monitoring vaccine effectiveness. When the Delta variant spread, the focus shifted yet again to include those cases.
Getting data isn’t always easy. First responders, healthcare workers, and essential frontline workers in the Research on the Epidemiology of SARS-CoV-2 in Essential Response Personnel (RECOVER) study had high exposure risk and thus were among the first to get vaccines. But they have been stretched thin: physically from long work hours, mentally from dealing with people who believe disinformation about COVID-19 vaccines, and emotionally from dealing first-hand with mass deaths in high-risk environments.
Essential response personnel have been stretched thin, but they have found time to provide key data for the RECOVER study each week for over a year.
But they have found time to provide key data for the RECOVER study each week via text messages, surveys, and weekly nasal swabs for COVID tests, sticking with the study for over a year now.
“This strikes me as heroic and brings about an upwelling of gratitude and respect for our participants whenever I think about it,” says Abt programmer Tyler Morrill.
Of course the pandemic continued to evolve rapidly. A month ago, the RECOVER team had settled into a routine pace of biweekly reports. Our published data in the July 22 issue of the New England Journal of Medicine about real-world effectiveness against infections and serious symptoms had only three breakthrough infections for fully vaccinated RECOVER participants from mid-December through April 10. Then on July 26, a spike: suddenly we detected six breakthrough infections in a single week.
We alerted the CDC but didn’t know if it was a blip or worrisome trend. The next week turned out to be similar, as was the week after that. The result was what the CDC announced in the August 18 White House press conference: a decline in estimated vaccine effectiveness against SARS-CoV-2 infection from 92 percent to 64 percent after the Delta variant started to spread. That was one reason for the recommendation for widespread use of boosters.
Additional data led to slightly different numbers in the August 24 MMWR: a decline from 91 percent to 66 percent when Delta was predominant and an overall effectiveness rate of 80 percent since the vaccines first rolled out December.
To cope with the rapid pace of change, Abt invested in data visualization dashboards in Tableau that display new data soon after the initial processing. These data show vaccination coverage, trends in positive cases, total positive counts by vaccination status, and the crude vaccine effectiveness rate, which we calculate by comparing the rates of infection for vaccinated people to those of unvaccinated people. The dashboards quickly show new trends such as reduced effectiveness. We also developed programs that can incorporate new client requests rapidly.
As alarming Delta data came in from around the world, the CDC asked us to speed up our reporting. Wednesday deadlines accelerated to Tuesday, and on Friday, August 13, the CDC asked for the data by the following Monday night, a sign of the urgency for CDC to receive the data for decision-making.
Trouble is, some of the data don’t arrive until late Sunday night. The importance of the issue meant it was imperative to be fast, but we also had to be accurate. That put pressure on our programmers and data quality assurance team. We reorganized our processes, but could not eliminate the data quality checks. So Abt staffers worked early mornings and late nights to meet the changing deadlines and were on call over weekends to ensure responsiveness. We were able to deliver the data at 11:57 a.m. on Monday so that CDC analysts could begin their preliminary analysis in advance of the Wednesday press conference.
How could we be both fast and right? Fortunately, we had been fine tuning our data management procedures for five months. All changes to the code we use to process data are completed and fully vetted in advance so that we can download the data, run the processing scripts, and not hold up delivery of the datasets to CDC.
Many data quality checks are built into the code. Issues that arise are flagged and addressed in a parallel process without affecting our ability to deliver the data. We send data quality issues to our five participating study sites around the country to resolve discrepancies or questions at the source. We can deliver data quickly without fear we are cutting corners and sacrificing quality.
It’s not just a matter of being accurate, though. Providing context and appropriate cautions also is important so that policymakers and the public interpret the findings correctly. For the August 24 MMWR, for example, it was important to note that we didn’t know if the evidence of declining effectiveness was due to the powerful Delta variant, weakening of the vaccine over time, or both.
Uncertainty also stems from the limited length of observation (less than two months) and the small number of infections (43). We need more data to understand this dynamic—and we are collecting it.
For those in the data trenches week in and week out, the work has become both professional and personal.
Brian Sokol, the data lead on the project, had to travel recently after being privy to the new data about reduced vaccine effectiveness. The issue for him was not whether to wear a mask in the airport but how many to wear (he settled on two). And his sister asked if their 79-year-old mother should get a third shot.
“It’s extraordinary and a little eerie to be working on something that is so immediate in terms of both its public health impact and direct personal impact,” he says. “You can practically see our numbers getting turned into both public health guidance and personal behavior in real time.”
About Abt Associates
Abt Associates is a global consulting and research firm that uses data and bold thinking to improve the quality of people’s lives. From combatting infectious disease and conducting rigorous program evaluations, to ensuring safe drinking water and promoting access to affordable housing—and more—we partner with clients and communities to tackle their most complex challenges. https://www.abtassociates.com