‘Coding’ errors prompt retraction of paper on long COVID in kids
JAMA Pediatrics has retracted a controversial 2023 paper on the incidence of long COVID in children after the authors discovered a raft of “coding” errors in their analysis that greatly underestimated the risk of the condition.
The article – a research letter titled “Post–COVID-19 Condition in Children” – was written by a group of researchers in Canada led by Lyndsey Hahn, of the University of Alberta, in Edmonton. It has been cited eight times, according to Clarivate Analytics’ Web of Science, and garnered significant attention on mainstream and social media sites, including by critics who said the authors fatally botched the definition of long COVID.
According to the authors, the incidence of long covid in kids was “strikingly low”, occurring in just 0.4% of young patients. Symptoms of infection in kids typically resolve within two weeks, they added.
But those reassuring findings hinged on several errors in the analysis that made the incidence of long COVID in children look less than a third of what the researchers should have reported.
Per the retraction notice, signed by senior author Piush Mandhane:
"Since publication, we have identified methodological (analytical) errors in our original report that impacted our estimate of the prevalence of post–COVID-19 condition (PCC) in children and adolescents and our reported associations between pre- and post–COVID-19 symptoms.
One coding error led to the miscategorization of children for whom data were missing to be considered asymptomatic. Another error led to the exclusion of 15 children because of a mistaken date cut-off.
In addition, we originally reported a study sample of children between 8 and 13 years of age. We subsequently identified participants with COVID-19 (cases) who were recruited between 1 and 7.49 years and 14.5 and 19 years of age. These participants were included in the original PCC analysis. We had concluded that the incidence of PCC was 0.4% (1/271). In correcting these errors, we found that the incidence of PCC was 1.4% (4/286).
The researchers also identified coding errors affecting their analysis of symptoms before and after infection, and in the duration of fevers.
In correcting these errors, we found that pre–COVID-19 headache, fever, and fatigue were associated with post–COVID-19 symptoms of the same. However, pre–COVID-19 cough, rhinitis, and sore throat were not associated with post–COVID-19 symptoms of the same."
The notice concludes with an apology to the readers and editors of the journal, and states all seven authors agreed with the decision to retract the work.
Blake Murdoch, of the University of Alberta Health Law Institute – which you may recognize as the same university as the authors of the article – was co-author of a January 2024 letter to JAMA Pediatrics pointing out several flaws in the Hahn paper. Murdoch told us he and his colleagues were glad to see the retraction. With a caveat:
No, this does not address the concerns we raised, these are in fact additional errors! Dr. Mandhane’s retraction note chalked it up to coding errors, which certainly seem to exist, but there is no mention of the basic errors in the manuscript we laid out in our letter, such as changing the definition of PCC between the Introduction and Methods as I discuss in an old tweet thread. Also, the study authors’ response letter to us seemingly attempted to claim they used components of a definition of long covid that was not published until after the study was completed. …
Perhaps the authors are attempting to save face by focusing on the coding errors rather than the egregious and obvious methodological problems, such as claiming to use the WHO definition of PCC but then inventing their own.
Mandhane did not respond to a request for comment.
The retraction is the latest of more than 400 involving papers about COVID-19, by our count.
https://retractionwatch.com/2024/08/20/coding-errors-prompt-retraction-of-paper-on-long-covid-in-kids/