2.1. Data
Data from the 6-year (2014–2019) Korean National Health and Nutrition Examination Survey (KNHANES), conducted by the Korean Disease Control and Prevention Agency (KDCA), were analyzed in this study. The KNHANES employed a stratified, multistage cluster-sampling design on a representative nationwide South Korean population [19]. The World Health Organization's (WHO) Global Physical Activity Questionnaire (GPAQ) guideline was reflected in the KNHANES starting in 2014, and the COVID-19 pandemic triggered to suspend the pulmonary function test starting in year 2020. Hence, this study employed KNHANES data from 2014 to 2019. As the use of raw data of KNHANES is publicly available, ethics approval was exempt. All responses from participants in KNHANES are collected anonymously from the KDCA.
2.2. Study population
In the 2014–2019 KNHANES, all participants were recorded for the study population (n=47,309). Firstly, to acquire the results of the pulmonary function test, we excluded those who were not subjected to the test, those below the age of 40, and those who did not have the test results (n=28,795). Of the remaining 18,514 participants, excluding the case of lung-related diseases on-going prevalence (tuberculosis [n=9], lung cancer [n=12], asthma [n=333], chronic obstructive pulmonary disease [n=50]; total n=377 with 27 overlapped) and the case with missing variables used in this study (n=2,123), the final study population comprised 16,014 participants. Due to the restricted number of pulmonary function test participants, individuals with pulmonary function test results in the overall dataset are assigned a distinct weighted value in the relevant analysis.
2.3. Variables
2.3.1. Physical activity
Physical activity, the variable of interest, was measured using GPAQ [20]. The WHO established the GPAQ, a standardized questionnaire that assesses physical activity by three distinct domains (work, recreation, and transport) and is currently used in 100 countries. Responded Moderate- and vigorous-intensity of physical activity in each domain was calculated as Metabolic Equivalents for Task (MET)-minute per week as follows:
intensity of physical activity × minute per day × times per week
The intensity of physical activity score was 8.0 at vigorous physical activity and 4.0 at moderate or transport-related physical activity. Given that the respondents were above 40 years old, we determined a sufficiently active categorization as performing physical activity consistent with the GPAQ guidelines: at least 30 minutes of moderate-intensity activity or walking per day on at least five days of a typical week; 20 minutes of vigorous-intensity activity per day on at least three days of a typical week; or 5 days of any combination of walking and moderate- or vigorous-intensity activities totaling at least 600 MET-minutes per week [21]. In addition, since the guidelines recommend that this age group perform muscle-strengthening activities 2 or more days a week, this study finally defined it as participating in regular physical activities, including this activity [1]. For subgroups, questionnaire domain of physical activity was divided into four groups based on a large proportion of the individual's MET score: (1) work, (2) recreation, (3) transport, and (4) inactive. Type of physical activity was divided into four groups based on aerobic or muscle-strengthening physical activity: (1) aerobic, (2) muscle-strengthening, (3) both types, and (4) inactive. Furthermore, intensity of physical activity was divided into three groups based on individual's MET score: (1) high (≥3,000 MET), (2) moderate (<3,000 MET), and (3) inactive (low).
2.3.2. Pulmonary function
Pulmonary function, the dependent variable, was measured using spirometry. Four technicians conducted spirometry using a dry rolling seal spirometer (Model 2130; SensorMedics, Yorba Linda, CA, USA) [22]. Survey data on non-smokers without a history of respiratory diseases or symptoms, as well as normal chest X-ray results, were used to create spirometric prediction equations. The results were interpreted using the proportion of forced vital capacity (FVC) and forced expiratory volume at 1 second (FEV1) compared to normal value of prediction equations. Four kinds of results were obtained: normal (FEV1/FVC ≥70% and FVC ≥80%), restrictive pattern (FEV1/FVC ≥70% and FVC <80%), obstructive pattern (FEV1/FVC <70% and FVC ≥80%), and mixed pattern (FEV1/FVC <70% and FVC <80%) [23]. Pulmonary functional defect was defined as restrictive, obstructive, and mixed pattern in this study. For subgroups, each four categories of variables were examined.
2.3.3. Related covariates
The following variables were entered in the model as potential covariates based on previous studies [12, 13, 24]. Sociodemographic factors: sex, age, scale of residence, household income, education level, marital status, and three-groups of the longest occupation according to the Korean Standard Classification of Occupation [25], which is originally categorized into 10 (white collar: manager, professionals and related worker and office worker; pink collar: service worker and sales worker; and blue collar: agriculture/forestry/fishery, craft and related trades workers, machine operators and assemblers, labor workers, and soldier). Health behavioral or other related factors: body mass index, smoking status, pack-year of smoking, drinking (more than once in a month: yes or no), hypertension, diabetes, and the year variables.
2.4. Statistical analysis
Sex-specific factors can influence physical activity and pulmonary outcomes [26]. Accordingly, all analyses were sex-stratified, and the results were proposed in the same way. To identify the general characteristics of the study population, number and percentage were presented with the results of a chi-square test. Multiple logistic regression analysis was performed to identify the association between physical activity and pulmonary function. For subgroup analyses, stratification of other covariates, pattern of pulmonary function defect, and physical activity were examined to investigate detail relationship. In subgroup analysis for pattern of pulmonary function, multinomial logistic regression was utilized owing to the multinomial category of dependent variable. The association between all variables in this study was estimated using odds ratios (ORs) and 95% confidence intervals (CIs). SAS version 9.4M6 (SAS Institute, Cary, NC, USA) was used for all statistical analyses.