Misperception of body weight and associated socioeconomic and health-related factors among Korean female adults: A nationwide population-based study

BACKGROUND: Misperception of body weight is associated with various psychological and health problems, including obesity, eating disorders, and mental problems. To date, female-specific risk factors, including socioeconomic or health-related lifestyle features, or their indicative performance for the misperception in Asian women according to age groups remain unknown. OBJECTIVES: To investigate the prevalence and associated risk factors for the mismatch in self-perceived body weight and evaluated the classification performance of the identified risk factors across age groups in female adults. METHODS: We analyzed data of 22,121 women (age 19–97 years) from the 7-year Korea National Health and Nutrition Examination Survey dataset (2010-2016). We evaluated self-perceived body weight of the participants with their actual weight using the body mass index cut-off and grouped them by age: early adulthood (19–45), middle adulthood (46–59), and late adulthood (≥60). Logistic regression was conducted in each age group based on their weight misperception. The classification performance of the identified risk factors was evaluated with a bagging tree ensemble model with 5-fold cross-validation. RESULTS: 22.2% (n=4,916) of the study participants incorrectly perceived their body weight, of which 14.1% (n=3,110) and 8.2% (n=1,806) were in the underestimated and overestimated groups. Among the age groups, the proportion of participants who misperceived their body weight was highest in late adulthood (31.8%) and the rate of overestimation was highest in early adulthood (14.1%). We found that a lower education level, absence of menopause, perception of themselves as unhealthy, and efforts for weight management were significantly associated with the overall misperception (overestimation or underestimation) of body weight across age groups. Based on the identified risk factors, the highest area under the receiver operating curve (AUROC) and accuracy of the best classification model (weight overestimation in all participants) were 0.758 and 0.761, respectively. Adding various associated lifestyle factors to the baseline model resulted in an average increase of 0.159 and 0.135 in AUROC for classifying weight underestimation and overestimation, respectively. CONCLUSIONS: Age, education level, marital status, absence of menopause, amount of exercise, efforts for weight management (gain, loss, and maintenance), and self-perceived health status were significantly associated with the mismatch of body weight.