Study Data Showed Vaccinated Kids Shed COVID Up to 3 Times Longer Than Unvaxed

The authors of a study published Oct. 23 in JAMA Pediatrics concluded that vaccinated and unvaccinated children who got COVID-19 were equally contagious. That’s how the media reported on the study, but the data showed that vaccinated kids shed the virus for up to 3 times longer than unvaccinated children.

Children who received the COVID-19 vaccine and those who didn’t both shed the virus for a median time of three days, according to a research letter published Oct. 23 in JAMA Pediatrics.

The authors’ findings led them to suggest that school policies requiring students with COVID-19 to stay at home for five days are appropriate and that schools need not consider vaccination or booster status in back-to-school policies.

However, although this advice may be sound, it was based on an unsound statistical trick that hid the significant disparities in viral shedding between vaxed and unvaxed kids.

While all unvaccinated children in the study were clear of the virus by day six, 10 of 52 vaccinated children (19%) took double the time to be virus-free. Three subjects (about 6%) were still infectious on day 10 — three times longer than the reported median.

How long children remained contagious after a COVID-19 diagnosis remains a discussion topic, despite global data showing children overwhelmingly survive infection.

A recent study found kids rarely transmit the virus to adults, which is unsurprising given a 2021 study showing that nose/throat swabs from infected children were only half as likely — compared with swabs from adults — to contain virus capable of causing illness.

Here’s how researchers conducted the study

Between April and September 2022, researchers led by Neeraj Sood, Ph.D., a health policy expert at the University of Southern California, recruited 76 children ages 7-18 with a positive PCR test for COVID-19. Half were male and 68.4% (52 subjects) were vaccinated for COVID-19.

Of the vaccinated kids, 35 had received two doses, 15 received a third shot (booster), and two did not provide details.

Investigators took throat swabs from the subjects on the day they tested positive (day 0), and every two days thereafter up to day 10.

Samples were tested for their ability to infect cultured cells through a “cytopathic” or cell-changing effect. Cytopathic assays use microscopes that are not powerful enough to see viruses or their entry into cells but instead, look for visually obvious changes over time in test cells exposed to the viral swab. Microscopic examinations occurred on days six, eight and 10.

Investigators reported the “median duration of infectivity” — how long a child was infectious — as three days for both vaccinated and unvaccinated subjects.

The median of a dataset is the middle value, with an equal number of values above and below. The average, or mean, equals all the values divided by the number of entries.

The mean and median of the same data can be very different. For example, the median value of the dataset 2, 3, 5, 8, 42 is 5 (the number in the middle), but the average is 12. Using median values is legitimate in this case because 5 is closer to a “typical” value and there is only one outlier.

But Sood’s tactic of designating the 20% of his subjects who continued to shed virus for up to 10 days as “outliers” is dubious. All, it turns out, were vaccinated.

As Figure 1 shows, the 10 vaccinated subjects (19%) who remained contagious at the end of day six were not outliers and their infectious status was not a rare, unexplained event but typical for this dataset.

Using median instead of mean infectivity duration downplays the significance of much of the vaxed subject data in Sood’s study. The approach was wrong for the unvaccinated group as well, as those data are tightly distributed around the three-day mark with no outliers.

Equating median values for vaxed versus unvaxed allowed researchers to combine data from the two groups and report that 14 participants (18.4%) were infectious on day five and three (3.9%) were still shedding on day 10.

Again, all subjects still positive after day six were vaccinated — but hiding this inconvenient fact allowed the authors to conclude, without considering the more informative mean values, that “there was no association between duration of infectivity and vaccination or booster status.”

This contradiction is evident in Figure 1B.

Figure 1B. Percentage of subjects infectious over time, by vaccination status.

Percentage (vertical axis) of vaccinated (orange solid line) and unvaccinated (black solid line) subjects contagious since their initial COVID-19 diagnosis in days (horizontal axis). The numbers below the graph represent the number of patients still shedding virus, with “unvaccinated” and “vaccinated” groups shown separately. By day 6 no unvaccinated subjects were contagious but 10 vaccinated patients were still shedding virus. On day 10, when the last cell-infection test was run, three vaccinated subjects were still shedding virus.

How the media reported it

Several media outlets picked up the story.

MedPage Today issued a straight-up rehash of the authors’ conclusions: “There was no difference in duration of infectivity by vaccination status.”

The University of Minnesota’s Center for Infectious Disease Research and Policy news reported that some children remained infectious for longer: “Fourteen children (18.4%) were still infectious on day 5 and 3 (3.9%) remained so on day 10.”

However, the center failed to mention that all children who remained infectious after day five were vaccinated.

Medical Dialogues repeated both vaccine-favorable narratives: “The median time of infectivity was 3 days, with 18.4% of children still infectious on day five and 3.9% infectious on day 10. The study also found no association between how long children were infectious and whether they were vaccinated.”

Why combine vaxed and unvaxed data?

Deliberately excluding relevant data but analyzing them correctly is one way to cherry-pick which results to present and which to obscure. Sood did not do this.

Since he provided all his data he resorted to a more common and accepted approach based on statistical legerdemain or “precise falsehoods.” An analogy from everyday life is an attorney who knows his client is guilty but works hard to acquit him nonetheless.

Cherry-picking and selective statistics are part of a larger group of questionable research practices as old as research itself.

One may guess at the motives behind such practices — for example expecting one result (vaccinated children recover from COVID-19 faster than unvaxed) but learning halfway through a study that another outcome (no they don’t) is inevitable.

But it usually comes down to the simple fact that no one likes to be wrong.

HTML Hyperlinks


RSS Feed