Study Design
The present study is an observational study with a cross-sectional study based on data from the Korea National Health and Nutrition Examination Survey (KNHANES) that was conducted to compare the differences of demographics, health behaviors, and health literacy between individuals with IFG and non-IFG. Furthermore, considering the cultural characteristics of Korea, IFG related factors according to sex were identified.
Data Source and Participants
This investigation utilizes the Korean Health and Nutrition Examination Survey (KNHANES) data, which is a survey conducted by the Korean Ministry of Health and Welfare and the Korea Centers for Diseases Control and Prevention (KCDC). The KNHANES evaluates the health and nutrition status of Koreans, monitors trends in health risk factors and major chronic diseases, and provides data for the development and evaluation of Koreans health policies and programs. This survey used a stratified, multistage probability sampling design to select housing units. This included stratifying by region in the first stage and layering by sex and age in the second stage. To represent the entire Korean adult population and account for the complex sampling procedure, sampling weights were used. The weighted sample for the KNHANES reflected the sampling fraction and nonresponse bias adjustments.
In total, data from 24,269 people who participated in the KNHANES VII (2016–2018) were analyzed to determine whether the inclusion criteria were met. Data that met the exclusion criteria, such as those under 29 or over 65 years and diagnosed with type I or II diabetes, were then removed (excluded data n=14,350). The KNHANES survey comprised the following: The Health Interview Survey, Health Behavior Survey, Nutrition Survey, and Health Examination Survey, and this study’s variables were collected from these four surveys.
Assessment of Measurements
The KNHANES VII (2016–2018) was used to obtain information on demographic characteristics, hematologic examinations, health behaviors, and health literacy (i.e., recognition of nutrition fact labels and utilization of the nutrition fact labels).
Demographic characteristics
Sex, age, economic status (low, middle, high), education level (≤ middle school, ≥high school), employed status (yes or no), duty type (day, shift), perceived health status (good, bad), body mass index (< 23 kg/m2 or ≥ 23 kg/m2), and perceived stress (never feel, feel a little, feel much) were used to assess socioeconomic status.
Economic status was based on the average monthly household income. Household income was partially adjusted by sex and age based on monthly equivalent income (= monthly household income/ number of families). Education level was categorized based on graduation status, such that completion, withdrawal, enrollment, and leave of absence were categorized with the preceding academic background.
Citizens in the Republic of Korea receive mandatory education through middle school in accordance with Article 31 of the Constitution. Therefore, the education status was categorized based on whether each participant was a middle school graduate or had received more than compulsory education.
For employed status, participants were asked “Have you worked for more than an hour in the last week for income? Or have you worked for your family for more than 18 hours unpaid? If you were originally working but are on a temporary leave of absence, this is considered work.” For duty type, people who answered “yes” in response to “Do you work from 6 am to 6 pm?” were classified as day type, and people who answered “yes” in response to “Or do you work during a different time period?” were classified as shift type.
The question regarding perceived health status asked, “What do you usually think about your health?” and the result was categorized into good or bad. Body mass index (BMI) was calculated using measured height and obesity. In this study, participants with a BMI of less than 18.5 kg/m2 were categorized as underweight, those with a BMI of 18.5–22.9 kg/m2 were categorized as normal weight, and those with a BMI of at least 23.0 kg/m2 were categorized as overweight [16]. Participants were considered as not overweight or overweight based on whether their BMI was < 23 kg/m2 or ≥ 23 kg/m2, respectively.
Health behaviors
Dining out was classified as less than once a month, less than twice a week, 3 to 4 times a week, or every day. Participants’ alcohol consumption during the recent year was categorized as not at all, less than once a week, and more than two times a week. For smoking, “Smoking every day” and “Smoking occasionally” were classified as “yes,” and “I smoked in the past, but not now” was classified as “no.”
The American Diabetes Association (ADA) [17] recommends that people with prediabetes and IFG walk for at least 150 minutes a week. Weekly walking time was categorized as less than 150 minutes or more than 150 minutes a week. Weekly aerobic activity equivalent to medium-intensity physical activity for over 2 hours and 30 minutes or high-intensity physical activity for over 1 hour and 15 minutes or a mixture of medium- and high-intensity physical activity (high intensity for one minute or medium intensity for two minutes) was classified as “yes” or “no.”
Total sitting time per day was calculated based on the usual sitting time during the week, and was categorized into sitting less than four hours a day or sitting for more than four hours a day.
Health literacy. Health literacy was measured by two questions. Participants responded to questions regarding the recognition of nutrition fact labels and utilization of nutrition fact labels with “yes” or “no.” The outcome variable, “the presence of IFG” was assessed through responses to questions regarding whether participants have fasted for 8 hours without diabetes and have a fasting glucose level greater than 100 mg/dl but less than 126 mg/dl [1].
Statistical Analysis
To increase the sensitivity of this study and reduce hidden biases such as sampling bias, complex samples analysis procedures were implemented in consideration of sampling weights, stratification variables, and cluster including two or more survey districts by city/state/house type variables using the IBM SPSS 25.0 program. Through the process of applying these weights, the sampling bias was reduced. Missing data were statistically excluded. First, data from the selected sample were analyzed for individuals with non-IFG and IFG. Chi-square tests were conducted to compare the percentage of all variables for individuals with non-IFG and IFG. For the logistic regression analysis of the main outcomes, Shapiro-Wilk test was performed to test for normality, and the normality assumption was satisfied with p-value > .05. To get the evidence for collinearity among the independent variables, whether Variance Inflation Factor (VIF) was is greater than 10. As results, there were not found s evidence for collinearity. That is, there were no variables greater than 10. The chi-square statistic was computed comparing the observed frequencies with those expected under the linear model using the Hosmer-Lemeshow tests. A non-significant chi-square found in this study (x2= 10.40, p > .05), indicating that the data fit the model well.
In the next step, the logistic regression model was conducted using variables with significant results in the univariate analysis to investigate the association between variables that were statistically significant at less than 0.05. Moreover, multivariate logistic regression was stratified into female and male.