Preprint servers are being used to censor scholarly papers critical of the Centers for Disease Control and Prevention (CDC) and policy errors made by the Biden administration, according to Vinay Prasad, M.D., MPH.
Prasad, hematologist-oncologist, professor and director of a medical policy lab at the University of California San Francisco said his lab has submitted hundreds of papers to preprint servers — but only those that deviate from the official COVID-19 narrative have been rejected.
The basis for rejecting those papers is inconsistent with standards applied for other topics or with the fundamental rules of the servers, Prasad explained in a recent Substack post and YouTube video on the topic.
“All [rejected papers] are consistent with scientific practices and norms, and similar papers not critical of the CDC or Biden administration have been accepted,” Prasad wrote.
“If only papers that praise the CDC are acceptable by preprint servers,” he added, “then the role of science as a check and balance on incorrect policy is subverted.”
Preprint servers were established to address inefficiencies in academic publishing. The peer-review process typically takes months or more, delaying the real-time sharing of scientific findings with the public.
Also, many journals are proprietary and can only be accessed through expensive personal or institutional subscriptions.
Preprint servers offer a location for scientific reports and papers to be available to the public while the paper goes through peer review — making scientific findings available immediately and for free and opening them up to broader public debate.
There is no peer-review process for preprints, although there is a vetting process.
Prasad said preprint servers are supposed to be neutral and supposed to post all research conducted with a clearly explained and reproducible methodology. The goal, Prasad said, is to be inclusive and make scientific debate possible.
But instead, Prasad said, several of the preprint servers have become “gatekeepers” for what science gets published.
“When the people who work there are rabid, politicized and biased,” he said, “I think that’s a problem.”
‘The preprint servers are really a disgrace’
To test the bias of preprint servers, Prasad’s lab did a comprehensive analysis of its own preprints. Lab staff analyzed outcomes for all preprints they submitted to SSRN, medRxiv (pronounced “med archive”) and Zenodo servers — which he noted is a “reproducible systematic methodology.”
They found “a startling pattern of censorship and inconsistent standards from preprint servers. Preprint servers appear to be playing politics,” Prasad wrote.
“Preprint servers are not supposed to be journals — they are not supposed to reject articles merely because the people running them disagree with the arguments within.”
But the analysis suggests they are doing just that. Even the preprint of Prasad’s analysis was rejected by both medRxiv and SSRN.
“They don’t want to post a preprint that makes their own preprint server look bad,” he said, adding, “That’s pathetic. You have to post it because it’s factually true.”
Examples of censored articles demonstrate the bias behind the servers’ decisions.
One paper re-analyzed a study published in the New England Journal of Medicine that found cloth masks reduced the rates of COVID-19 transmission in some Massachusetts schools.
Prasad’s lab found that by simply lengthening the timeframe of the data analyzed in the observational analysis, the paper’s conclusion was invalidated. The data showed instead that masks did not slow the spread.
The servers didn’t post their paper, he said, citing “the need to be cautious about posting medical content.”
Prasad said the article was censored, “because it calls into question something that, frankly speaking, is pretty stupid, which is masking children with cloth masks — a stupid intervention derived by someone who cannot read randomized controlled trials and then pushed with the full force of the federal government — with no credible evidence and no randomized data.”
The servers censored Prasad’s lab’s comprehensive analysis of preprint bias using the same justification — “the need to be cautious about medical content.”
He said, “No one could possibly believe that this paper would require the need to be cautious — it merely documents your [preprint server’s] prior screw-ups.”
Another censored preprint reported on the lab’s systematic review and meta-analysis of Pfizer-BioNTech vaccination in 5 to 11-year-olds. The paper critiqued the U.S. Food and Drug Administration’s (FDA) approval of the drug for that age group.
The SSRN also removed that preprint citing, again, “the need to be cautious about medical content.”
SSRN used the same “boilerplate language” to remove the lab’s review of methodologies and conclusions in Cochrane reports. Those findings supported the controversial conclusion of the Cochrane review that community masking had no impact on slowing the spread of COVID-19.
Another article, rejected by medRxiv, which documented statistical and numerical errors made by the CDC during the pandemic, was already accepted by a journal and is one of the 10 most downloaded articles of all time on SSRN.
“Here’s the point,” Prasad said, “You don’t have to agree with me, but this preprint server is not even letting the audience of scientists decide. Who gave them this authority? They don’t get to peer review the article. That’s not their purview.”
Overall the team found that 38% of their submissions to preprint servers were rejected or removed. Yet, these rejected articles eventually were published and extensively downloaded.
“The median number of downloads for a rejected/removed article that was later accepted by a different server was 4,142 vs. 300 for articles submitted and accepted without rejection or removal,” he said.
Their analysis found that overall, Zenodo does not censor articles, but SSRN and medRxiv do.
So, he said, these organizations, established to make science transparent and uncensored have become gatekeepers “for their friends, for the views that they like.”
He also said their policies were inconsistent, with no clear scientific principles guiding rejection.
“They’re rejecting 38[%] of our articles because these are critical of establishment COVID-19 policies,” he said, adding:
“The COVID-19 pandemic is in fact a great example of … how people in power suppress minority views even when those views are meritorious — like toddlers shouldn’t mask, school should be open, you don’t need to mandate boosters, you shouldn’t mandate boosters for young men, nobody who had COVID should have to get the vaccine — those are sensible medical policies that are correct.
“History will vindicate them. They were all censored at one time or the other… The preprint servers are really a disgrace.”