Few Trust Most Health Content on Social Media, Autism Claims Follow Thimerosal Policy Shift, and Misleading Narratives About SSRIs in Pregnancy

Featured: KFF’s Latest Poll Finds That Over Half of Adults Use Social Media For Health Information, But Few Trust Most of the Content They See

With public trust in government health agencies as reliable messengers declining, the latest KFF Tracking Poll on Health Information and Trust finds that over half (55%) of adults, including larger shares of young adults and Black and Hispanic adults, say they use social media to find health information and advice at least occasionally. Large shares of adults report seeing information on social media in the past month on about weight loss, diet, or nutrition (72%) and mental health (58%), and about four in ten (38%) report seeing vaccine-related content. Smaller shares say they saw information on social media about abortion (30%) and birth control (22%) in the past 30 days.

The poll also finds that most adults are skeptical of the health information and advice they see across social media platforms. Fewer than half of different social media platform users say they find “most” or “some” of the health information they see on the platforms they use trustworthy. Less than one in ten say they think “most” of this content is trustworthy. On some of the most widely used social media apps or sites including TikTok and YouTube, larger shares of younger adults compared to older adults trust the health information and advice they see.

The latest poll also asked about social media influencers and found that so far, their influence on health is relatively small. As KFF President and CEO Drew Altman wrote in a recent column, just 14% of the public say they regularly get health information and advice from influencers online, and an even smaller share of social media users (5%) can name a particular health influencer they trust.

Thimerosal Vaccine Claims Spread Online After Federal Policy Shift

False claims that a mercury-based vaccine preservative, thimerosal, is harmful to children gained traction in late July, coinciding with a federal decision to no longer recommend flu vaccines containing the compound. The narrative, which incorrectly links thimerosal to autism and other developmental issues, appeared in online conversation after Health and Human Services (HHS) Secretary Robert F. Kennedy Jr. rescinded the CDC’s recommendations of flu vaccines that contain the ingredient. Kennedy later accepted recommendations by the CDC’s Advisory Committee on Immunization (ACIP) to endorse routine influenza vaccination this year and to only recommend thimerosal-free formulations. The decision followed a June meeting and vote of the CDC’s Advisory Committee on Immunization Practices (ACIP), where a former leader of a group known for its opposition to vaccines presented unsupported claims that thimerosal is toxic to children. On July 23, the day the decision was announced, some X users with large followings cited the policy shift in posts that made false claims about thimerosal. One account, with more than 500,000 followers, repeated the claim that it causes autism and wrote, “You wouldn’t catch me dead taking a flu vaccine… never.. but this is good for those who religiously get it.”

Thimerosal, a mercury-based compound used to prevent contamination in multi-dose vaccine vials, was removed from routine childhood vaccinations in 2001, not because of any evidence of harm but in response to public concern about mercury exposure. Thimerosal contains a form of mercury called ethylmercury, which is excreted by the body much more quickly than methylmercury, the form of mercury found in fish which can accumulate in the body and have hazardous health effects. Confusion between the two compounds contributed to public unease, and the FDA and medical groups recommended removing the compound as a precautionary measure and to preserve confidence in vaccines. Some virologists have argued that the 2001 removal, while intended to increase confidence in vaccines, may have inadvertently reinforced false claims linking them to autism. Dozens of studies since have found no evidence of harm, and rates of autism diagnoses have steadily increased since the compound was largely removed from vaccinations.

In the United States, thimerosal is still used in a small portion of multi-dose influenza vaccine vials to prevent fungal or bacterial contamination. Because these vials are generally less expensive per dose and require less storage space, they are primarily used in high-volume or resource-limited settings, such as mass vaccination clinics. They also remain an important option in many low- and middle-income countries where single-dose vials are harder to distribute and store. The decision to rescind recommendations for thimerosal-containing vaccines may have global effects, leading to reduced access and increased cost, as international health authorities have historically closely watched recommendations by ACIP. In the U.S., insurance coverage for these vaccines may be affected, as coverage requirements are linked to CDC and ACIP recommendations in almost every case. Medical organizations, including the American Academy of Pediatrics (AAP), have further cautioned that the change could further the spread of false claims about vaccines and contribute to rising vaccine hesitancy.

Polling Insights: KFF’s April 2025 Tracking Poll on Health Information and Trust found that  three in four (74%) adults say they are either “very confident” or “somewhat confident” that vaccines for the flu are safe. Confidence in flu vaccines, however, differs across partisans. While majorities across groups are at least “somewhat” confident these vaccines are safe, Democrats (57%) are much more likely than independents (36%) or Republicans (24%) to say they are “very” confident. Confidence is also lower among parents of children under age 18 compared to non-parents (24% vs. 44% are “very” confident these vaccines are safe).

Online Narratives Misrepresent Antidepressant Safety in Pregnancy

About one in 10 people experience depression during pregnancy, and around 6% of pregnant people in the U.S. are treated with SSRIs (selective serotonin reuptake inhibitors). Online conversation about these antidepressants in July reflected public confusion about the safety of their use during pregnancy. In late July, social media users repeated exaggerated claims of risk following a July 21 meeting of an FDA advisory panel that presented unbalanced information by emphasizing potential harms of SSRIs during pregnancy beyond what current scientific evidence supports, while giving little attention to documented benefits. Some of the panelists claimed that perinatal usage of the medications can lead to autism, miscarriage, and other harms, or that they offer no therapeutic value, despite strong evidence demonstrating the importance of these medications in addressing maternal mental health. Some social media users shared and amplified these claims, inaccurately describing SSRIs as especially harmful for pregnant and breastfeeding people. One X user, with more than 500,000 followers, wrote, “Evil… so evil. Doctors are actively prescribing SSRI’s to Pregnant Women that Cause Birth Defects in Babies.”  

Some studies cited by panelists suggested potential risks, but psychiatrists and obstetricians criticized these studies for poor design and inadequate control for confounding factors, including the presence or severity of maternal depression. When accounting for these factors, research has largely supported the safety of SSRI use during pregnancy. A large cohort study controlling for confounding factors, for example, found no evidence of association with autism or other neurodevelopmental disorders. Other members of the panel also claimed, without evidence, that SSRIs are wholly ineffective, conflicting with numerous studies, including meta-analyses of randomized controlled trials that have shown SSRIs to be more effective than placebo in treating major depressive disorder.  

Misleading claims that overemphasize the risks of antidepressants in pregnancy may prompt patients to discontinue or delay treatment, putting them at risk for poor outcomes. Untreated perinatal depression carries well-documented and significant consequences for both maternal and fetal health, including preterm birth, low birth weight, and developmental delays. It may also lead to reduced prenatal care, higher rates of substance use, and increased risk of suicidality. According to the CDC, mental health conditions are a leading cause of pregnancy-related deaths, accounting for 22.5% of such deaths in 2020.  

While most evidence supports the safety of SSRIs in pregnancy, some studies have found small increased risks for adverse outcomes, including bleeding complications at delivery. Because the potential harms of medication must be weighed against the risks of leaving depression untreated, guidelines adopted by the American College of Obstetricians and Gynecologists (ACOG) emphasize that treatment for depression, including the use of SSRIs or selective norepinephrine reuptake inhibitors (SNRIs), should be highly individualized. The Substance Abuse and Mental Health Services Administration (SAMHSA) has called for more education on the use of medications during pregnancy, noting that people who had experienced mental health conditions during pregnancy reported that having clearer information would have improved their experiences. Medical organizations, including ACOG, the American Psychiatric Association (APA), and the Society for Maternal-Fetal Medicine (SMFM) stress that the harms of untreated depression in pregnancy may, for many patients, outweigh the risks of medication and therefore recommend a collaborative, shared clinical decision-making process. Research has shown that this practice, in which patients and healthcare providers work together to make informed healthcare decisions, can help improve trust in the physician-patient relationship. 

Influencers Promote Unproven Health Benefits of Nicotine

Mentions of the supposed health benefits of nicotine have appeared widely on social media, with some health influencers and accounts with large followings sharing posts that claim that it is not addictive and that it improves focus, boosts brain function, and may prevent or cure neurodegenerative diseases. Some X accounts, including one with more than 100,000 followers, posted a video clip of a podcast interview in which a chiropractor and influencer claimed nicotine was not addictive and was a cure for Parkinson’s disease, Alzheimer’s disease, multiple sclerosis, COVID-19, and glioblastoma brain tumors. Another account, with more than 800,000 followers, posted a podcast clip of a prominent political commentator attributing his health to his use of nicotine. KFF’s latest Tracking Poll on Health Information and Trust found that 15% of adults who use social media, including 23% of those ages 18-29, say they regularly get health information and advice from influeners on social media.

Claims of nicotine’s health benefits are largely made without evidence. Researchers are investigating therapeutic applications of nicotine, specifically for Parkinson’s disease and Alzheimer’s-related cognitive decline, but results from observational and animal studies have not yet been replicated in large-scale clinical trials. Curative properties have not been demonstrated, and nicotine is not approved by the FDA for any of these therapeutic uses. A 2008 correlational review of nicotine and Parkinson’s disease, for example, suggested that nicotine could help reduce symptoms and drug-induced side effects, but a later clinical trial found that nicotine did not slow progression of the disease. For Alzheimer’s-related cognitive decline, a small trial investigating nicotine and mild cognitive impairment found some improvement in attention and memory, but the authors cautioned that further studies were needed, which are still underway.

As these speculative claims about nicotine’s health benefits continue to circulate, they may obscure established health risks, encourage use by non-users, and deter quitting among current users. Despite claims by some social media users, nicotine is known to be highly addictive, and a 2015 review found that it increases risk of cardiovascular, respiratory, and gastrointestinal disorders and has adverse effects on the immune system and reproductive health. It is particularly harmful in some populations, including pregnant people and adolescents, where nicotine exposure can be toxic to a developing fetus and interfere with brain maturation.

AI-Generated Public Health Campaigns Most Effective With Human Oversight

Findings from a series of studies published in PNAS Nexus suggest that artificial intelligence (AI) could streamline public health media campaigns by selecting real-time, community-generated messages from social media to create persuasive health messaging. Researchers from the University of Pennsylvania, University of Illinois, Emory University, and government and community agencies developed an AI system to automatically generate an HIV prevention and testing campaign for counties in the U.S. The system collected messages from social media posts and evaluated whether they were actionable, relevant, and appropriate for the target population, men who have sex with men (MSM).

Researchers recruited 260 participants to review 36 selected messages and rate them on actionability, appropriateness, accuracy, relevance, and effectiveness. 12 messages were selected by the AI and reviewed by a human researcher, 12 were AI-selected without human review, and 12 were control messages taken from a simple keyword search. Participants rated messages selected by the AI higher than messages from the control group, with those vetted by a human scoring highest. In a separate test, researchers provided access to the tool to public health agencies and community-based organizations, finding that the AI selection process made them more likely to post HIV prevention messages on social media.

Researchers noted, though, that using machine learning tools in this way may introduce algorithmic bias, unintentional errors caused by existing prejudices in the data used to train AI systems, particularly when automated processes are used without a structured human review. They emphasized the continued importance of human oversight to mitigate this bias, as well as the risk of amplifying false information from community-generated messages. Among study participants, messages reviewed by human researchers were consistently rated more persuasive and accurate.