Physical activity assessment
All types of physical activity (work physical activity, recreational physical activity, commuting physical activity, and sedentary behavior) were assessed using the GPAQ. Whether they engage in more than moderate intensity work physical activity, recreational physical activity, and commuting physical activity during the week. Sedentary duration <600min or ≥600min in a 24-hour period. The codes "1" and "2" indicate whether this type of physical activity or sedentary duration <600min and ≥600min are in the final database, respectively.
Smoking behavior assessment
Smoking behavior data is extracted from the SMQ dateset, which provides survey participants' cigarette use history, age of start, use in the last 30 days, cigarette brand, sub-brand, and other relevant details. For adults 18 years of age or older, trained interviewers ask questions at home using a computer-assisted Personal Interview (CAPI) system. The codes "1" and "2" represent whether or not you smoke at the current stage.
Covariate
Covariates included gender, age, race, education, marital status, and income-poverty ratio. A total of 2,015 participants were divided into three age groups: 20-39 years, 40-59 years, and >60 years. Race is divided into Hispanic, non-Hispanic white, non-Hispanic black, non-Hispanic Asian, and other races. The education level is divided into below high school, high school and above high school. Marital status was divided into cohabitation, married living alone (widowed, divorced, separated) and never married. The poverty ratio is a measure of poverty measured by dividing household income by the survey year. In this study, the poverty ratio was used to create two income conditions, poor (<1.3) and middle income (≥1.3) [
20].
Statistical analysis
We used Microsoft Excel 2010 to extract and merge the raw data and exclude missing and useless (rejected, don't know) items. The database includes adults 20 years of age and older with complete information. For the purpose of this study, we tested the significance of the differences in covariates between the "smoking" and "non-smoking" groups. Rank sum test was used for quantitative variables and chi-square test for categorical variables. We used a binary logistic regression model to analyze the relationship between different types of physical activity and smoking behavior. All data were analyzed using the Statistical Product and Service Solutions (SPSS) version 26.0, and a P-value less than 0.05 was considered statistically significant (bilateral test). Variables that were statistically significant in the univariate analysis were included in the stepwise binary logistic regression analysis. In univariate analysis, all covariables (P < 0.05) except gender (P=0.077) were statistically significant. In the significance test of measurement data, age P < 0.05 (variance was not homogeneous), P(bilateral) < 0.001, the difference was not statistically significant. Therefore, age and sex were not excluded as confounding factors in logistic regression analysis. A-entry=0.05 and a-exit=0.10 were used to select and exclude confounding variables.
When analyzing the relationship between physical activity at work and smoking behavior. We took work physical activity as the independent variable (1=yes, 2=no) and smoking (1=smoking, 2=no smoking behavior) as the dependent variable. To exclude the effect of confounding variables, we built the following models: Model I: Only the independent variable work physical activity was adjusted. Model II: Adjusted for independent variables in model I plus demographic variables (race, education, marital status, income-poverty ratio). Model III: Adjusted for model II plus variables for recreational physical activity, commuting physical activity, and sedentary behavior.
When analyzing the relationship between recreational physical activity and smoking behavior. We took recreational physical activity as the independent variable (1=yes, 2=no) and smoking (1=smoking, 2=no smoking behavior) as the dependent variable. To exclude the effect of confounding variables, we built the following models: Model IV: Only the independent variable recreational physical activity was adjusted. Model V: Adjusted for independent variables in model IV plus demographic variables (race, education, marital status, income-poverty ratio). Model VI: Adjusted for model V plus variables for work physical activity, commuting physical activity, and sedentary behavior.
When analyzing the relationship between commuting physical activity and smoking behavior. We took commuting physical activity as the independent variable (1=yes, 2=no) and smoking (1=smoking, 2=no smoking behavior) as the dependent variable. To exclude the effect of confounding variables, we built the following models: Model VII: Only the independent variable commuting physical activity was adjusted. Model VIII: Adjusted for independent variables in model VII plus demographic variables (race, education, marital status, income-poverty ratio). Model IX: Adjusted for model VIII plus variables for work physical activity, recreational physical activity, and sedentary behavior.
On the relationship between sedentary behavior and smoking behavior. We took sedentary behavior as the independent variable (1=yes, 2=no) and smoking (1=smoking, 2=no smoking) as the dependent variable. To exclude the effect of confounding variables, we built the following model: Model X: Only the sedentary behavior of the independent variable was adjusted. Model XI: Adjusted for independent variables in model X plus demographic variables (race, education, marital status, income-poverty ratio). Model XII: Adjusted for model XI plus variables for work physical activity, recreational physical activity, and commuting physical activity.