Exposure to health misinformation on social media across key health domains: a systematic review and meta-analysis of survey-based studies

Exposure to health misinformation on social media has emerged as a significant public health concern; however, survey-based evidence remains conceptually heterogeneous across health domains and outcome definitions. This systematic review and meta-analysis synthesized individual-level observational studies examining direct exposure to health misinformation across key domains, including COVID-19, vaccination, cancer, and oral health. Following PRISMA 2020 and MOOSE guidelines, PubMed, Web of Science, and Scopus were searched (2010–2025). Eligible studies were observational and survey-based, reporting prevalence or determinants of individual-level exposure to health misinformation encountered on social media. Exposure was operationalized as (i) perceived exposure (self-reported encountering of misinformation) or (ii) item-level encounter/recognition of predefined misinformation claims framed as prior exposure. Constructs reflecting belief, attitudinal agreement, susceptibility, or behavioral responses were excluded from prevalence pooling to prevent conceptual conflation. Interventional or experimental correction studies were excluded to preserve comparability with naturalistic observational exposure measures. Random-effects meta-analyses were conducted in R (meta/metafor), with heterogeneity quantified using I². Subgroup analyses were conducted by health domain, geographic region, and platform context. Eight studies (N = 22,780) met inclusion criteria. Reported prevalence estimates varied substantially (10%–87%) across domains and exposure operationalizations. The pooled prevalence was 59.0% (95% CI: 44.0–73.0); however, heterogeneity was extreme (I² = 99.8%). Accordingly, the pooled estimate is interpreted as a descriptive contextual summary rather than a generalizable population parameter. Subgroup analyses suggested domain-dependent patterns, with comparatively higher reported exposure in COVID-19 and oral health contexts and lower levels in cancer-related contexts. Across studies, younger age, lower health or digital literacy, and minority ethnicity were recurrently associated with higher reported exposure. Platform-related associations were context-dependent and varied by outcome construct and health domain, indicating that platform effects operate as environmental modifiers rather than intrinsic determinants. Exposure to health misinformation on social media appears common but highly variable across health domains, operational definitions, and platform environments. Given extreme between-study heterogeneity, reliance on cross-sectional self-reported measures, and variability in exposure operationalization, findings should be interpreted as contextual rather than universally generalizable. The most informative insights derive from domain- and context-specific patterns, which may inform targeted and evidence-sensitive public health strategies.