Study population
This study used data from KoNEHS Cycle 3 (2015–2017), a representative cross-sectional sample of the population of Korea. KoNEHS Cycle 3 was conducted between August 2015 to June 2017 to ensure homogeneity of the sample composition for each year considering regional and seasonal distribution. The study population of KoNEHS Cycle 3 comprised 6167 individuals, including 2380 children (≥3 years old) and 3787 adults (≥19 years old). Different multi-stage stratified cluster sampling methods were used between children and adults. In the present study, data from adults were used. For adults, the first stratification was centered on local administrative districts and coastal areas based on the Population and Household Census 2015 provided by Statistics Korea. The second stratification was based on the proportion of residential-complex districts, as well as location within 5 km of the east, south, and west coast, which related to socio-economic status. Ultimately, KoNEHS Cycle 3 included 233 districts nationwide, including 20 areas that had national air quality monitoring stations.
The KoNEHS Cycle 3 collected questionnaires and urine and blood samples for 16 clinical tests and performed analyses for 26 harmful environmental substances (e.g., phthalates and VOC metabolites). In the present study, the urinary cotinine concentration, the primary metabolite of nicotine, was used as a biomarker for SHS exposure. Cotinine is a specific and sensitive biomarker for SHS exposure with an average half-life of 17 h [15]. This study was approved by the institutional review board of the National Institute of Environmental Research in Korea (NIER-2016-BR-003-01, NIER-2016-BR-003-03).
Based on questionnaires, the sample of 3787 adult participants was limited to never and former smokers (n = 3183). Next, our sample was limited to those who lived in apartments, attached housing, or detached housing (n = 3168) because participants who answered “others” could not be distinguished. We then limited our samples to participants whose proportion of daily time spent indoors at home, at the workplace, indoors somewhere other than home or workplace, on transportation, and outdoors as recorded by the questionnaires was more than 80% (i.e., 1152 of 1440 min) (n = 3094). Finally, our samples were limited to participants whose urinary creatinine concentrations were between 0.3 and 3.0 g/L [16] (n = 2701). Among the 2701 qualified participants, 126 were excluded because their urinary cotinine concentration was higher than the cut-off point of 53 μg/L [17], and they were suspected of being smokers. Ultimately, a total of 2575 non-smoking adults were included in the final analysis.
Smoking home status
Non-smokers were classified as participants who answered “I have never smoked” and “I used to smoke in the past but not anymore.” Smokers were defined as participants who answered “I smoke now” and were excluded from this study. Among the non-smokers, those who lived in smoking homes were defined as those who responded “yes” and those lived in smoke-free homes were defined as those who responded “no” to the question “Do you live with any smokers at home?”
Socio- demographic characteristics
Socio-demographic characteristics such as sex (male or female), age (19–39, 40–59, or ≥60 years), type of housing (apartment, attached housing, or detached housing), household income (<1000, 1000–1999, 2000–2999, or ≥3000 USD/month), ventilation duration at home (<30, 30–59, 60–599, ≥600 min/day), self-reported weekly SHS exposure (no or yes), time spent in residential indoors (<780, 780–1079, ≥1080 min/day) and outdoors (<10, 10–69, ≥70 min/day), and job classification (unemployed, manual occupation, non-manual occupation, or hospitality venue worker). Levels of household income and ventilation duration at home was divided into quartiles and time spent at residential indoors and outdoors was divided into tertiles.
For this study, jobs were classified into unemployed (unemployed, students, or stay-at-home parents), non-manual occupations (manager, office/service/sales workers, or expert/related workers), manual occupations (skilled/functional workers, machine operators, assembly/simple labor workers, or agricultural/forestry/fishery workers), and hospitality venues (restaurant, bar, cafe, fast-food franchise, or bakery workers) based on the job-classification code or written job title in the raw data from KoNEHS Cycle 3. Similar classifications have been used in a previous study [18]. Jobs in hospitality venues in this study were classified separately from non-manual occupation to examine differences in urinary cotinine concentrations by job type after the implementation of smoke-free regulations in these places started from 1st January, 2015 [5].
Urinary cotinine
Mid-stream urine samples were collected and frozen at -20°C until laboratory analysis. For urinary cotinine analysis, we added 250 µL of internal standard, 50 μL of 0.1 M sodium hydroxide, and 0.5 mL of chloroform to 3-mL urine samples. The solution was centrifuged, and the upper layer was removed. Next, 0.2 g of sodium sulfate was added to remove residual water and 3 μL of solution was injected into a gas chromatograph/mass spectrometer (Clarus 600 T, PerkinElmer, USA) to estimate urinary cotinine concentrations. The detail analytical methods for urinary cotinine have been described in elsewhere [14, 19]. The method detection limit (MDL) for urinary cotinine was 0.3 μg/L. Urinary cotinine concentrations below the MDL were assigned a value of 0.2 μg/L (MDL/ √2).
Statistical analysis
Statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc., Cary, NC, USA). In the analysis, the domain option in SAS was used to control for subgroups with full clusters in reducing the dataset of interest. The PROC SURVEYFREQ function was used to calculate percentages of socio-demographic variables and determine the differences in these percentages between those living in smoking and smoke-free homes.
Natural log (ln)-transformed creatinine-unadjusted and creatinine-adjusted urinary cotinine concentrations were used in statistical analyses due to the skewness of the distribution of the untransformed data. Creatinine-adjusted urinary cotinine was calculated by using the urinary cotinine concentration to urinary creatinine concentration ratio (μg/g creatinine [Cr]). SAS PRO SURVEYMEAN was used to calculate the overall geometric means (GMs) and 95% confidence intervals (CIs) of the Cr-unadjusted and Cr-adjusted urinary cotinine concentrations. The same procedure was used to determine the GM and 95% CI of the Cr-adjusted urinary cotinine concentrations of non-smoking adults living in smoking and smoke-free homes. Using SAS PROC SURVEYREG, covariate-adjusted least-square geometric means (LSGMs) and 95% CIs of Cr-adjusted urinary cotinine concentrations of non-smoking adults living in smoking and smoke-free homes by socio-demographic variable were estimated in multivariable regression models. Socio-demographic variables, including sex, age, type of housing, household income, ventilation duration at home, weekly SHS exposure, time spent at residential indoors and outdoors, and job classification, were included as covariates. Among the socio-demographic characteristics, we did not use education level, former smoker status, or time spent at the workplace because of potential collinearity between age and education level (Spearman’s rho = -0.62), sex and former smoker status (Spearman’s rho = -0.69), and time spent at residential indoors and that at the workplace (Spearman’s rho = -0.72). Similar criteria have been used in previous studies [18]. The same SAS procedure was used to conduct multivariable linear regression analysis to examine relationships between urinary cotinine concentrations and socio-demographic characteristics in non-smoking adults who lived in smoking and smoke-free homes. A p-value of 0.05 was considered significant in all analyses.