Data were obtained from the 2016–2018 Korea National Health and Nutrition Examination Survey (KNHANES), which was conducted by the Korea Centers for Disease Control and Prevention. The KNHANES is a self-report survey conducted in Koreans of all age and is designed to gather annual national data on sociodemographic, economic, and health-related conditions and behaviors. The survey is consisted of three components (health interview, health examination and nutrition survey), all of which are conducted by trained staff members including physicians and medical technicians 16.
Of the 24,269 survey participants, we excluded who tested positive for serologic markers for liver disease (hepatitis B, hepatitis C, or liver cirrhosis) (n=735), were aged <20 years who did not undergo blood testing conducted by the KNHANES (n=6,868), and were not representative of covariates considered in the study (failed to answer the survey questionnaires) (n=3,148). Accordingly, the final sample size consisted of 13,518 participants (Fig. 1). This study was an analysis of existing data; thus it did not require approval by ethics review board. The data that was used in this study is the KNHANES and it has been getting an annual review and approval by Korea Centers for Disease Control (KCDC) Research Ethics Review Committee since 2007.
NAFLD is the main dependent variable in this study. NAFLD in this study was diagnosed according to the hepatic steatosis index (HSI), which was developed by the Department of Internal Medicine and Liver Research Institute in Seoul National University College of Medicine to efficiently select individuals for liver ultrasonography 17. The HSI formula was derived via logistic regression model using serum alanine aminotransferase (ALT) to serum aspartate aminotransferase (AST) ratio, body mass index (BMI), and diabetes mellitus status: HSI= 8 × (ALT/AST ratio) + BMI (+ 2, if female; + 2, if with diabetes mellitus)17. Participants were considered to have NAFLD if their HSI value was above 36.
The main independent variable is the participants’ sitting time. Sitting time was measured by asking participants to report the following question adopted from the International Physical Activity Questionnaire (IPAQ) 18. “How many hours did you spend time sitting per day during the last week?” This questionnaire indicates the time spent in academic and/or leisurely activities in a reclined or seated position. Participants’ responses to sitting time were divided into 4 categories using quartiles: < 5, 5 -7, 8 – 10, and > 10 hours.
Sociodemographic, economic, and health-related factors were also considered in the study. Sociodemographic factors included age, educational attainment, and marital status. Economic factors included household income and occupation. Health-related factors include sleeping time (hours), total energy intake ((carbohydrate(g) x 4 kcal/g) + (protein(g) x 4 kcal/g) + (fat(g) x 9 kcal/g)), physical activeness (active: ≥150 min of moderate activity, ≥ 75 min of vigorous activity, or a mixture of both for ≥ 150 min; inactive: <150 min of moderate activity, <75 min of vigorous activity, or a mixture of both for <150 min), pack years of smoking, current drinking status, comorbidity of hypertension, and comorbidity of diabetes mellitus.
The frequencies and percentages of participants were calculated for each of the categorized variables included in the study. The variables included in the analysis were all categorical, those that were not initially categorical were converted into categories (age, BMI, total energy intake, etc.). The chi-square (χ2) test was performed to assess the chi-square differences between the groups within each categorized variable (Table 1). Multiple logistic regression analysis was used to calculate the odds ratios (with 95% confidence intervals) for NAFLD according to the participants’ report on sitting time (Table 2). The sub-group analysis for NAFLD stratified by the participants’ sex, physical activeness, and obesity status defined by BMI was also performed using multiple logistic regression (Fig. 2).
The reported odds ratios were adjusted for all covariates considered in the study. The sampling weight variables were applied in the analysis to improve the representativeness of the sample. KNHANES has constructed sample weights to take into account survey non-response, over-sampling, post-stratification, and sampling error. The use of sample weights in the analysis is recommended to produce an unbiased national estimate. For all data analysis, we used SAS version 9.4 (SAS Institute, Inc, Cary, NC, USA) and the significance level was set at p value < 0.05.