Aug 29, 2022 - Health

Twitter's medical information problem

Illustration of a Twitter verified sticker with the "no" symbol replacing the checkmark.

Illustration: Maura Losch/Axios

Twitter played an outsized role in disseminating public health information during the COVID pandemic, but it's coming under fire for the gatekeeping role it's played in elevating — or quieting — qualified voices.

Why it matters: For better or worse, a lot of people get important information from social media, particularly since COVID-19 arrived.

"You never knew who was going to be the next leading health care voice," said Fumiko Chino, a radiation oncologist at Memorial Sloan Kettering Cancer Center.

  • Twitter "certainly elevated some voices. Some good, who were trying to promote good public health messaging, and others who were not promoting good evidence-based medicine," she said.

The big picture: Last week, the social media giant's acknowledged its efforts to squelch misinformation about COVID have occasionally gone awry.

  • The accounts of multiple doctors, scientists, and patient advocates — even those with a blue verification badge — were suspended and factual health tweets flagged as false under the platform's misinformation policy — a problem experts say hurts the credibility of the field, the Washington Post reported.
  • Many of the tweets have since had the misinformation labels removed, and the suspended accounts have been restored, per the Post.
  • "We acknowledge the mistakes made in these cases and are reviewing our team's protocol to safeguard against such mistakes in the future," Twitter spokeswoman Celeste Carswell said in an email to Axios.

Be smart: It's an example of the unwieldy challenge of conveying the most accurate information from authoritative voices during a global health crisis.

  • In the first months of the pandemic, Twitter said it worked with public health authorities to identify and give the blue verification badge to more than 1,600 accounts belonging to experts that were used to offer facts and guidance.

Yes, but: A study led by Memorial Sloan Kettering researchers published in JAMA Network Open found that those verified physician sources in the first year of the pandemic were far more likely to be male (71%) rather than female (29%).

  • It also found an over-representation of experts from the U.S. compared to the rest of the world.
  • At the time the data was gathered in November 2020, there was not a transparent delineation about why certain accounts were verified and others weren't, said Chino, a co-author of the study who received a blue verification badge on her own Twitter account early on in the pandemic.

What they're saying: "The concern we have is that this may promote what we already know in medicine, which is male voices are promoted more, they're elevated more and they are given a bit more weight," Chino said.

The other side: After a public feedback process, Twitter rolled out a more transparent verification system in May 2021, a spokeswoman said. Those changes included adjusting language to be clear about what each notability category refers to and adding criteria to various categories to gain the badge, Carswell said.

  • Chino said she appreciates the clarification. But she still sees room for bias in how Twitter ultimately determines who is considered "authentic, notable, and active," especially for physicians who must now apply under the "activist, organizers, and other influential individuals" category.
  • She pointed specifically to the example of Kim Manning, a frequent public health speaker and medical professor at Emory with more than 100,000 followers, who has been unable to get verified.

The bottom line: It's not entirely surprising we're seeing inequities in society reflected on social media platforms. But it's critical to hold these platforms accountable for whose voice gets aired, Chino said.

  • "For me, it charges me to put a fine eye in terms of who I'm elevating, who I'm retweeting," Chino said. "Because on this platform of Twitter, it matters."
Go deeper