Prevalence of Health Misinformation on Social Media: Systematic Review

Authors: Victor Suarez-Lledo, Javier Alvarez-Galvez
Venue: Journal of Medical Internet Research, 2021 — DOI

TL;DR

This systematic review of 69 studies characterizes health misinformation topics and their prevalence across social media platforms. Health misinformation was most prevalent on Twitter regarding smoking and drugs; vaccines, noncommunicable diseases, and pandemics were also common topics. The paper provides comparable measures of health misinformation prevalence and identifies key analytical techniques used to study this phenomenon.

Contributions

• First systematic review providing a comprehensive characterization of health misinformation on social media with standardized measures of prevalence
• Identified six principal health misinformation categories: vaccines (32% of studies), drugs/smoking (22%), noncommunicable diseases (19%), pandemics (10%), eating disorders (9%), and medical treatments (7%)
• Classified studies by methodological quality and analytical approaches, including content analysis, sentiment analysis, and social network analysis
• Mapped platform-specific patterns of health misinformation dissemination and characterization

Methods

Systematic review following PRISMA guidelines. Searched MEDLINE and PREMEDLINE for articles published before March 2019 focusing on health misinformation in social media. Extracted 69 eligible studies through full-text review. Classified studies across four dimensions: (1) descriptive information, (2) search strategy evaluation, (3) information evaluation, and (4) quality of methodology and reporting.

Results

Main findings by health topic:

• Vaccines (32% of studies): 96% of studies on Twitter/general; 14% specifically on HPV vaccine hesitancy; misinformation prevalence ranged widely from 1% to 87%, with 22% averaging across all vaccine-related studies
• Drugs and smoking (22%): 75% of studies examined opioids, marijuana, e-cigarettes; consumption of misinformation on Twitter reached 87% in some studies; drug-related claims often lacked scientific evidence
• Noncommunicable diseases (19%): Included cancer, diabetes, hypertension; 13% of studies on YouTube; prevalence of misinformation 33%-77%; often promoted unproven cures or fallacies
• Pandemics (10%): COVID-19, Zika, Ebola, H1N1; limited quantity and quality of information available; studies using content and sentiment analysis predominant
• Eating disorders (9%): Pro-eating disorder content prevalent and widespread; communities on social media normalized harmful behaviors
• Medical treatments (7%): Lowest misinformation prevalence (30%); content primarily from official sources and reliable accounts

Methodological findings:

• Twitter and Tumblr most common for social network analysis (19% of studies)
• YouTube and Instagram focused more on content/quality evaluation
• 24% of studies used evaluating content approaches; 16% used content/text analysis; 6% sentiment analysis
• Only 10% used content evaluation methods to assess quality of health information

Platform-specific patterns:

• Health misinformation most prevalent on Twitter and related to smoking products and drugs
• Different topics concentrate on different platforms: vaccines prominent on Twitter/YouTube, eating disorder content on pro-eating disorder communities
• Less common sources: Instagram, MySpace, Pinterest, Tumblr, WhatsApp, VK

Connections

• Related to Vaccine hesitancy and vaccine safety concerns via empirical measurement of misinformation prevalence on social platforms
• Complements Infodemic — characterizes domain-specific infodemic patterns in health
• Cross-referenced with Social Media Analysis — methodological review of analytical techniques
• Related to COVID-19 misinformation and the infodemic — pandemics one of six identified health topic categories
• Informs Misinformation and fake news detection — evidence base for health-specific detection approaches

Notes

This is the first systematic effort to provide comparable metrics across health misinformation studies. The heterogeneity in definitions, methodologies, and measurement approaches limits direct comparability, but the paper advances the field toward standardized measurement. A key gap is the lack of causal studies on behavioral downstream effects (e.g., vaccine refusal, drug consumption). The paper's focus on English-language, peer-reviewed articles published before 2019 means it may not capture rapid evolving platform dynamics or non-English misinformation ecosystems. The finding that vaccine and drug-related misinformation are the most prevalent suggests these domains warrant targeted interventions and further research.