Dr. Lorna Thorpe, a professor of population health at the New York University School of Medicine, and Dr. Richard Moggitt, a professor of biomedical informatics at the Emory University School of Medicine, recently led a study to measure the incidence of Long Covid among adults and children by using electronic health records.
Their research team found that for every 20 adults with COVID, between 2 and 5 developed Long Covid. And for every 25 children with COVID, 1 developed Long Covid. The incidence of Long Covid was higher for women than for men; for people hospitalized for COVID than for people who were not; and for people with more pre-existing conditions. Incidence was also higher for adults older than 65 years (than for younger adults); and for children aged 12 though 17 years (than for younger children). Long Covid incidence spiked in the months after each surge of a new SARS-CoV-2 variant. And Long Covid incidences for recent (Omicron) variants were higher than those for earlier ones.
Estimating the Long Covid incidence is challenging because symptoms of Long Covid often overlap with symptoms of other diseases. Indeed, estimates by earlier studies have varied from 6% to over 50%. Given this ambiguity, the research team also calculated a “minimum incidence” of Long Covid by measuring the “excess incidence” of Long Covid symptoms. To do this, the team first measured the incidence of symptoms among people who could not have had Long Covid. (In other words, some other diseases must have caused these symptoms in these people.) For example, in early 2019 some people’s symptoms resembled those of Long Covid, even though Long Covid could not have caused these symptoms (since the pandemic had just begun).
Similarly, in early 2021, some people had symptoms of Long Covid despite testing negative for SARS-CoV-2 infection. The team subtracted this background incidence from the incidence of Long Covid symptoms among people with COVID in later months and years (when Long Covid could have caused such symptoms). In this way, the team found the lower-bound estimates of excess incidence—for every 25 adults with COVID, 1 developed Long Covid; and for every 100 children with COVID, 1 or 2 developed Long Covid.
Overall, these results are important because they show that despite new treatments and vaccines and greater immunity among the population over these last few years, Long Covid incidence has not gone down. Thus, Long Covid is still an important public health matter needing serious attention and resources.
This study is also noteworthy by addressing variations in earlier studies—variations in how they defined disease, the periods they considered, and the populations they examined for estimating incidence. Here, the researchers considered four definitions of the disease, including the definition recently set by the National Academies of Sciences, Engineering, and Medicine. They considered different periods: throughout the entire study period (March 2020 through July 2023); by month; and by “wave” (of SARS-CoV-2 variants). Finally, they used three databases of electronic health records, comprising records from well over six million people.
Also noteworthy was how the study addressed critical questions like whether Long Covid incidence varies after different infection waves by different strains of SARS-CoV-2 and whether Long Covid incidence has gone down in recent years.
The research team noted several reasons Long Covid incidence may actually be higher than what they estimate. First, the team used a strict definition of Long Covid in children, which may have left out many bona fide cases of Long Covid. Also, the team may have used an overly inflated value for baseline incidence of symptoms (used to estimate excess incidence). This baseline value may have been too high because people in these control groups may have falsely tested negative for infection, or may have been more likely to have other health issues with Long Covid–like symptoms. A high baseline value would result in underestimating the excess incidence of Long Covid.
For future electronic health record studies to estimate Long Covid incidence, the team recommended considering the specific SARS-CoV-2 variants infecting individual participants and whether individual participants were vaccinated. Many electronic health records do not have this information.
This work was part of the NIH RECOVER initiative. Also part of the team was Solve’s own Emily Taylor.
