Study design
The present study was a cross-sectional descriptive study that was conducted nationwide to identify differences in distorted body weight perception and its related factors according to gender among middle-aged Koreans.
Data source and participants
Data for this study were obtained from the Korea National Health and Nutrition Examination Survey (KNHANES). The KNHANES has been administered annually by the Korea Centers for Disease Control and Prevention (KCDC) since 1998 using a multi-stage clustered probability sampling method. The participants in the KNHANES are a nationally representative sample of civilians in Korea, and about 10,000 individuals aged one year and over are included each year. Therefore, the potential for sampling bias in the data can be limited.
The KNHANES consists of three components (health interview, health examination, and nutrition survey) and is performed by trained medical staff at a mobile examination center and dieticians’ visits to the homes of the participants.
In this study, raw data were extracted from the 6th KNHANES (2014–2015) and 7th KNHANES (2016–2017) and collected by the stratified colonies system extraction method. All statistics of this survey have been calculated using sample weights assigned to sample participants. Among all data (N = 31,207), we extracted data on the target population of this study, which was middle-aged adults (45–64 years old) (N = 9172). We excluded participants who had missing data on body weight perception and BMI (N = 809), resulting in a final weighted sample of 8363 (91.17% of the total middle-aged sample).
Assessment of measurements
Among the three components of the KNHANES, socioeconomic status, anthropometric measures, body weight perception, health behaviors, food intake, and psychological status (perceived stress, depression, and suicidal ideation) were measured.
Socioeconomic status
Age, education level (≤ middle school versus > middle school), having a spouse (yes or no), having a job (yes or no), type of job (blue-collar worker or white-collar worker), household income quartile (the lowest, low, moderate, high), number of family members living together, and type of health insurance (National Health Insurance, National Workplace Health Insurance, Medical Aid) were used to assess socioeconomic status.
Anthropometric measures
Body mass index (BMI) was calculated using height (cm) and weight (kg) and classified into three groups: less than 18.5 kg/m2 = underweight, 18.5–22.9 kg/m2 = normal weight, more than 23.0 kg/m2 = overweight [12]. One item asked about “weight changes” in the last one year, and responses included “no change,” “weight loss,” and “weight gain.”
Daily energy consumption was classified as insufficient or excessive energy intake according to the Korean Foods and Nutrients Guideline (KFNG) [15]. Participants who consumed less than the recommended amount of daily energy intake were categorized as the insufficient group, and those who consumed more than the recommended amount of daily energy intake were categorized as the excessive group. In the KFNG, the recommended amount of daily energy consumption varies by age: 1900 kcal for women and 2400 kcal for men aged 30–49 years, and 1800 kcal for women and 2200 kcal for men aged 50–64 years [14].
Body weight perception
To assess body weight perception, participants were asked “what do you think about your body weight?” Response choices included “perceived as thinner,” “perceived as normal,” and “perceived as obese.” Distorted body image regarding body weight was measured by comparing the difference between perceived body image regarding body weight and actual body size based on BMI categories [1]. Consequently, participants were categorized into two groups: the non-distorted group, in which perceived body weight corresponded with the BMI category; and the distorted group, in which perceived body weight was below their BMI category (underestimation) or was above their BMI category (overestimation).
Health behaviors
Health behaviors consisted of regular participation in the national health screening program (yes or no), perceived health status (good or not bad/bad), having daily activity limitations (yes or no), weight control effort (yes or no), drinking (yes or no), heavy drinker (yes or no), smoking (yes or no), and regular daily walking exercise (yes or no). Heavy drinker was considered drinking on average 40 g/day (pure alcohol) or more in men and 20 g/day or more in women. Because there is no official guideline on a single standard drink size in Korea, we defined the amount of alcohol per 1 standard drink as 8 g in accordance with the guidelines on alcohol consumption [16]. For women we also gathered data regarding menopause (with or without).
Psychological status
Participants were asked about feelings of depression lasting more than two weeks, any suicidal ideation in the past year, and the presence of stress in daily life. Each question was responded to with “Yes” or “No.”
Statistical analysis and complex samples analysis
To increase the sensitivity and normality of this study, and reduce sampling bias, the complex samples analysis procedures were implemented by considering sampling weights, stratification variables, and cluster variables using the IBM SPSS 25.0 program.
Most statistical analyses assume that the data collected are from a simple random sample of the population of interest; however, if sampling units are widely spread out geographically, complex sampling designs are more efficient than simple random samples both in the choice of participants and in the feasibility of collecting data.
The Korea Centers for Disease Control and Prevention recommends complex sample analysis for analysis using raw data from the KNHANES. Complex sample data analysis programs such as SPSS Complex Samples analysis are designed to address the sampling design elements such as weights, stratification, and clusters. Prior to conducting the data analysis, researchers should merge multiple data files to create a complex sample plan file. In this case, weights are used to improve representativeness and accuracy. Weight were given from the KNHANES. This is because there were limitations in extracting data evenly from the population, and non-response errors can also differ due to differences in the number of households and populations between design and survey time. According to KNHANES’ raw data analysis guide, missing data is treated as a “system missing value” in the analysis. Missing data was statistically excluded. First, data from the selected sample (adults aged 45–64) were analyzed separately by gender. Cross tabulation analysis was conducted to identify the distribution of distorted and non-distorted participants in each gender. Next, descriptive analysis was performed to examine the distribution of all variables by gender. Complex samples chi-square tests and t-test were then conducted to compare the percentage or mean of all variables by gender. Finally, variables that had a significant association in the bivariate analyses were entered in a logistic regression model. Logistic regression was performed to examine the association between distorted body weight perception and the variables according to gender. The significance level was set at .05.