As school mask mandates return, critics claim pro-mask study plagued by methodological problems

"We can't make a scientific question a moral one just to shut down debate," epidemiologist says, referring to claim that masks mitigate "structural racism."

Updated: November 15, 2022 - 11:32pm

As an early flu season intersects with RSV and COVID-19 infections, schools that have reimposed mask mandates or said they might do so are getting reinforcement from a study of Boston-area school districts published in the New England Journal of Medicine.

Researchers at the Harvard and Boston University schools of public health and the Boston Public Health Commission claim that COVID cases rose by about 45 per 1,000 students and staff within 15 weeks of the Massachusetts mask mandate ending.

But the research is drawing fire for failing to account for variables that could have affected the results — and speculating that not wearing masks is racist. 

This conclusion lets mask proponents claim their opponents are "bad people" rather than scientists questioning their methods, University of California San Francisco epidemiologist Vinay Prasad wrote in an introduction for a lengthy analysis of the study in the Sensible Medicine newsletter. "We can't make a scientific question a moral one just to shut down debate," he said.

Healthcare professionals themselves may be wary of school mask mandates returning. Medical publisher MedPage Today inexplicably shut down a Twitter poll referring to the NEJM study several days early, showing that 97% of respondents opposed school mask mandates even "[w]hen respiratory cases are high."

The NEJM paper was coauthored by New York Health Commissioner Mary Bassett, who claimed in February that masks in schools had cut Omicron wave infections by more than 90% in a month in that state.

It received glowing coverage last week in the Washington Post and New York Times, but also as a preprint this summer before undergoing peer review, despite similar criticisms of its methodology. 

One pediatric infectious disease specialist wondered if the preprint was an academic hoax, while Prasad said then its failure to note Massachusetts exempts masked individuals from testing as "close contacts" made it unfit for publishing.

The near-ubiquity of natural immunity in American children, especially amid the Omicron variant wave, also raises the question of what meaningful benefit mask mandates would provide even if effective.

"Severe clinical forms" of pediatric COVID cases declined from 1.4% in "initial episodes" to 0.3% in reinfections, according to a new preprint tracking pediatric infections and reinfections in a Serbian province from March 2020 through July 2022.

The research by the Stanford Meta-Research Innovation Center, run by medical professor John Ioannidis, found reinfected children were 4.7 times more likely to have severe disease during their initial infection. Their reinfection hospitalization rate was less than half the initial, "and none resulted in death."

The Boston mask study contrasted Boston and Chelsea, the only school districts to end the 2021-2022 academic year with mask mandates, with districts that had similar COVID trends before ending their mandates within weeks of the state mandate ending in February. 

The researchers emphasize mask mandates stayed longer in districts with worse buildings, more students per classroom and higher percentages of low-income, disabled, black, Latino and English language-learning students. 

Universal masking not only can reduce the "loss of in-person school days" but also mitigate "effects of structural racism in schools, including potential deepening of educational inequities," they said.

Perhaps responding to criticism from Prasad or peer reviewers, the authors claim that the "additional testing of unmasked close contacts could explain less than 7% of the estimated excess cases" under the "most extreme assumptions."

In her review for Sensible Medicine, epidemiologist Tracy Beth Hoeg said the study data show case rates started going up in unmasking districts "independent of when they dropped the mask mandate." She broadly questioned the researchers' choice to use "difference in difference analysis," a technique that assumes differences between treatment and control groups are "constant over time" and that "the treatment itself" is the only thing that changes between them.

The problem for this assumption is that "the difference between the mask lifters and controls is not stable," and testing rates "were declining at different rates" before the state mandate ended, said Hoeg, who advises the Florida Department of Health and ran an earlier CDC-promoted study comparing COVID spread in schools versus their surrounding communities.

While a Yale medical professor emphasized that "masked schools got tested *more*" in-house, which suggests masks were more effective than observed, Hoeg noted the study doesn't include district-level information on testing rates including at-home rapid tests.

This is "a hugely important limitation and potential confounder," she said, noting the CDC kept recommending close-contact testing for unmasked students through spring. The richer districts that dropped mask mandates "may have been more likely to test."

Several districts also had lifted their mask mandates early due to waivers based on reaching 80% vaccination rates, she noted. CBS News Boston reported three weeks before the state mandate ended that 42 districts had received waivers since the September 2021 policy change.

While the authors said their analysis accounted for lower vaccination rates in the districts that remained masked, Hoeg said the "more highly vaccinated students" in the unmasking group might actually have been prone to higher risk of reinfection as their immunity waned, further complicating any mask causality.

The less vaccinated masked group may have gained "more durable protection going into spring" because of Omicron wave infections, she said, noting "seroprevalence rates" across the Boston area weren't included in the study.

Other critics weighed in. The difference-in-difference analysis is "preposterous and inappropriate: pandemic growth is nonlinear and depends on multiple unreported factors," said University of California Berkeley machine-learning professor Ben Recht, who scrutinized the methodology of an earlier mask study.

NEJM didn't respond to queries to explain why it published the paper despite the foundational criticisms of the preprint.

Just the News Spotlight