Facebook says my recent column about face masks is “missing context” and “could mislead people,” based on an assessment by “independent fact checkers.” That judgment and the analysis underlying it show how reflexive deference to government agencies distorts supposedly “independent” summaries of scientific evidence on controversial issues, especially issues related to COVID-19 control measures. When one of those agencies gets something wrong, criticism of its position is apt to be labeled “misleading” on social media platforms that strive to police COVID-19 “misinformation” at the government’s behest, regardless of what the evidence actually shows.

My column summarized the results of January 30 Cochrane Library review that considered 18 randomized controlled trials (RCTs) aimed at measuring the effectiveness of surgical masks or N95 respirators at reducing the spread of respiratory viruses. Judging from those studies, the Cochrane review found, wearing a mask in public places “probably makes little or no difference” in the number of infections. The authors said that conclusion was based on “moderate-certainty evidence.”

Does the Cochrane review prove that masks are worthless in protecting people from COVID-19? No. But it does show that the Centers for Disease Control and Prevention (CDC) misled the public about the strength of the evidence supporting mask mandates, which was the point I made in my column.

The authors of the Cochrane review suggest several possible explanations for their results, including “poor study design,” weak statistical power “arising from low viral circulation in some studies,” “lack of protection from eye exposure,” inconsistent or improper mask use, “self-contamination of the mask by hands,” “saturation of masks with saliva,” and increased risk taking based on “an exaggerated sense of security.” It is possible that some subjects in these studies did derive a benefit from wearing masks, but that effect was washed out by the behavior of other subjects who did not follow protocol, especially if those subjects took more risks than they otherwise would have because masks gave them “an exaggerated sense of security.”

It is nevertheless fair to say that the Cochrane review is inconsistent with the CDC’s statements about masks. After casting doubt on the value of general mask wearing early in the pandemic, the CDC decided the practice was so demonstrably effective that it should be legally mandated even for 2-year-olds. The CDC’s mask advice initially did not distinguish between surgical masks or N95s and the commonly used cloth masks it eventually conceded were far less effective. Its message was that wearing a mask—any mask, apparently—”reduc[es] your chance of infection by more than 80 percent,” a claim it said was supported by the best available evidence.

CDC Director Robert Redfield averred that masks were more effective than vaccines at protecting people from COVID-19. His successor, Rochelle Walensky, insisted that “the evidence is clear.” But the evidence on which the CDC relied came from two sources with widely recognized drawbacks.

Laboratory experiments provide good reason to believe that masks, especially N95s, can reduce the risk that someone will be infected or infect other people. But those experiments are conducted in idealized conditions that may not resemble the real world, where people often choose low-quality cloth masks and do not necessarily wear masks properly or consistently.

Observational studies, which look at infection rates among voluntary mask wearers or people subject to mask mandates, can provide additional evidence that general mask wearing reduces infection. But such studies do not fully account for confounding variables.

If people who voluntarily wear masks or live in jurisdictions that require them to do so differ from the comparison groups in ways that independently affect disease transmission, the estimates derived from observational studies will be misleading. Those studies can also be subject to other pitfalls, such as skewed sampling and recall bias, that make it difficult to reach firm conclusions.

Despite those uncertainties, the CDC touted an observational study that supposedly proved “wearing a mask lowered the odds of testing positive” by as much as 83 percent. It said even cloth masks reduced infection risk by 56 percent, although that result was not statistically significant and the study’s basic design, combined with grave methodological weaknesses, made it impossible to draw causal inferences.

RCTs aim to avoid these problems by comparing disease rates among subjects randomly assigned to wear masks in real-life situations with disease rates in a control group. That design makes the evidence produced by RCTs stronger than the evidence produced by laboratory experiments or observational studies. When they are conducted properly, RCTs support the inference that a difference in outcomes can be attributed to the intervention they test, because the treatment group and the control group are otherwise similar.

If wearing a mask had the dramatic impact that the CDC claimed, you would expect to see some evidence of that in RCTs. Yet the Cochrane review found essentially no relationship between mask wearing and disease rates, whether measured by reported symptoms or by laboratory tests. Nor did it confirm the expectation that N95s would prove superior to surgical masks in the field. The existing RCT evidence, the authors said, “demonstrates no differences in clinical effectiveness.”

According to Facebook, making these points “could mislead people.” But in fact, it was the CDC that misled people by insisting that the case was closed on masks and mask mandates while citing impressive but empirically shaky estimates of their effectiveness.

The “independent fact checkers” on whom Facebook relies, who work for an organization called Health Feedback, give the game away by contradicting themselves. “Multiple studies show that face masks reduce the spread of COVID-19,” their headline claims, echoing the CDC. Health Feedback’s “key take away” modifies that claim, saying “evidence suggests that widespread mask usage can reduce community transmission of SARS-CoV-2, especially when combined with other interventions like frequent handwashing and physical distancing” (emphasis added).

Already we have moved from a confident assertion about what “multiple studies show” regarding the effectiveness of masks in particular to a qualified statement about what “evidence suggests” regarding the effectiveness of multiple precautions taken in conjunction with each other. But the whole point of RCTs is to isolate the impact of a specific intervention—in this case, face masks.

According to Health Feedback’s conclusion, “a growing body of evidence from RCTs and observational studies suggests that consistent mask-wearing can effectively reduce the spread of respiratory viruses like SARS-CoV-2 in both healthcare and community settings” (emphasis added). Health Feedback’s fact checkers concede that “the extent to which community mask-wearing contributes to limiting the spread of different respiratory viruses and in different circumstances is still unclear” (emphasis added). They note potential weaknesses in the RCTs covered by the Cochrane review and say more research is needed to definitively settle the question of how effective masks are.

Contrast that gloss with the position taken by the CDC, which says “the evidence is clear” that wearing a mask “reduc[es] your chance of infection by more than 80 percent.” Walensky said that remarkable reduction applies to “the flu,” “the coronavirus,” and “even just the common cold.” The CDC also thinks the evidence is clear that mask mandates work in schools and other settings, despite the methodological problems with the observational studies it cites.

Health Feedback’s analysis ostensibly addresses the claim that the Cochrane review “demonstrates” face masks “are ineffective at reducing the spread of COVID-19 and other respiratory diseases.” I did not make that claim. But the fact check, which cites two publications of my column (on this website and in the Chicago Sun-Times), also objects to my statement that “the CDC grossly exaggerated the evidence supporting mask mandates.” Health Feedback not only fails to show that assessment is wrong; it reinforces the point that the CDC distorted the science to support a predetermined conclusion.