Factors associated with cardiovascular disease risk perception bias among adults undergoing health examinations in Guangzhou, China: A Bayesian network analysis

Objective To identify factors associated with cardiovascular disease risk perception bias and examine their probabilistic dependencies. Methods A cross-sectional study was conducted among 424 adults undergoing health examinations at a tertiary hospital in Guangzhou, China, from March to May 2025. Perceived 10-year cardiovascular disease risk categories were compared with those estimated using the Prediction for Atherosclerotic Cardiovascular Disease Risk in China model. Risk perception bias was defined as disagreement between these categories: overestimation when the perceived category was higher, and underestimation when it was lower. The final logistic regression model retained sex, smoking status, stress, number of chronic diseases, and health literacy. These factors and risk perception bias formed the Bayesian network. Results Risk perception bias occurred in 191 participants (45.05%): 176 (41.51%) overestimated and 15 (3.54%) underestimated risk. Female sex and high health literacy were associated with lower odds of bias; passive smoking, moderate or severe stress, and one chronic disease were associated with higher odds. Smoking status, health literacy, and stress had direct probabilistic dependencies with risk perception bias. Conclusions Risk perception bias predominantly involved overestimation. Behavioral, psychological, and health-literacy factors may inform tailored cardiovascular risk communication.