The COVID-19 pandemic fuelled the current infodemic of misinformation, is my firm – though unproven – belief. I was happy to read that at least The Lancet agrees. In this recent editorial, they state that the consequences of misinformation (misleading data shared unintentionally) and disinformation (deliberately deceptive information) reach far beyond COVID-19 or vaccine scepticism and will only increase with decisions of major platforms to reduce fact-checking.
Two weeks ago, Facebook announced reducing the amount of fact-checking and transferring fact-checkers from California to Texas “where fact-checkers are expected to be less pre-opinionated”.
Opinions matter
On March 21st 2020, Donald Trump tweeted “HYDROXYCHLOROQUINE & AZITHROMYCIN, taken together, have a real chance to be one of the biggest game changers in the history of medicine.” That message showed up 78,800,580 times in people’s feeds. HCQ for acute COVID-19 was adopted globally, and some randomised trials came to an early end as patients refused randomisation, insisting on getting HCQ instead. We now know that in hospitalised patients with acute COVID-19, HCQ increased mortality rates with 10%.
Quantifying misinformation
In my search for any quantification of these processes, I came across this Science study from last year. For misinformation to have impact, the message must (a) be seen and (b) influence someone’s opinion. Focusing on misinformation about COVID-19 vaccines on Facebook in 2021, they quantified the content that was reported (‘flagged’) as misinformation to Facebook fact-checkers, as well as content that was categorised as expressing vaccine scepticism but was not flagged, as it frequently came from trustworthy sources. The non-flagged content dwarfed the flagged content, which was 0,3% (8,7 million) of all vaccine sceptical content (2,7 billion!). An example of a non-flagged post: “A ‘healthy’ doctor died two weeks after getting a COVID-19 vaccine; CDC is investigating why” from the Chicago Tribune (April 8, 2021, viewed by 55 million people on Facebook).
The “persuasiveness” of content (treatment effect) was experimentally quantified through surveys. Bottom-line: flagged content had a higher treatment effect, but the overall impact of all non-flagged content was much higher. Good news: reducing fact-checking will hardly affect the impact of vaccine scepticism. Bad news: there is a much larger impact of vaccine scepticism coming from trustworthy sources.
Infection control
According to Lancet, “combating misinformation takes a systematic approach akin to curbing the spread of infectious agents: finding and containing the source; proactively identifying the most vulnerable to its effects; and immunising the population against false claims by providing clear educational resources.”
We need vaccines against misinformation.