Exposure to health misinformation, self-care confidence and delayed care-seeking among adults in the UK: a cross-sectional survey

Objectives Generate population-level evidence on exposure to misleading health information and its behavioural implications within contemporary self-care. Design Cross-sectional, self-administered online survey. Setting United Kingdom. Participants 1,414 adults (mean age 46.8 years) recruited via a closed online panel using quota sampling by age, gender and ethnicity. Main outcome measures Self-reported exposure to false or misleading health information; perceived ability to recognise misinformation; delays in professional care-seeking following exposure. Secondary measures included self-care practices, trust in information sources and recognition of common misinformation tactics. Multivariable ordinal logistic regression examined predictors of recognition ability and care-seeking delays; Spearman correlations assessed associations between numeracy and recognition of misinformation tactics. Results Exposure to misleading health information was widespread: 81.8% of respondents reported encountering false or misleading content in the previous six months. Confidence in recognising misinformation was high overall, with strong educational gradients; university and postgraduate education were independently associated with higher perceived recognition ability. However, educational attainment did not protect against behavioural consequences. Exposure to misleading content showed a clear dose-response association to professional care-seeking, even after adjustment for age, sex, education, ethnicity and urbanicity. Compared with no exposure, odds of delay increased among those exposed once or twice (aOR 1.53), monthly (aOR 1.83) and weekly (aOR 1.89). Asian and Black participants had significantly higher odds of delay compared with White participants and urban residents had higher risk than peri-urban residents. Numeracy skills were positively associated with recognising sensational headlines and cherry-picked evidence, but not with detection of AI-generated or impersonated content. Conclusions Pharmaceutical and health misinformation was widespread and linked to inappropriate care-seeking. Although higher education increased perceived recognition, it did not eliminate behavioural risk. Traditional health literacy skills helped identify some forms of misinformation but offered limited protection against AI-enabled deception. These findings position pharmaceutical misinformation as a patient safety and access issue requiring coordinated, equity-sensitive, system-level responses alongside individual literacy efforts.