Skip Navigation
Skip to contents

Ann Occup Environ Med : Annals of Occupational and Environmental Medicine

OPEN ACCESS
SEARCH
Search
Advanced Search
mobile menu button

Articles

Page Path
HOME > Ann Occup Environ Med > Volume 38; 2026 > Article
Original Article Occupational differences in benzene-related biomarker levels beyond traditional industrial settings: findings from the Korean National Environmental Health Survey, 2015–2023
Chul Gab Lee1,2,*orcid
Annals of Occupational and Environmental Medicine 2026;38:e19.
DOI: https://doi.org/10.35371/aoem.2026.38.e19
Published online: June 15, 2026

1Department of Occupational and Environmental Medicine, Chosun University Hospital, Gwangju, Korea

2Gwangju Branch of Korea Occupational Disease Surveillance Center, Gwangju, Korea

*Corresponding author Chul Gab Lee Department of Occupational and Environmental Medicine, Chosun University Hospital, 365 Pilmun-daero, Dong-gu, Gwangju 61453, Korea E-mail: cglee@chosun.ac.kr, eecg@daum.net
• Received: March 26, 2026   • Revised: June 1, 2026   • Accepted: June 8, 2026

© 2026 Korean Society of Occupational & Environmental Medicine

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

  • 378 Views
  • 12 Download
prev next
Full Article

Download PDF Download PDF

  • Background
    Benzene is a group 1 carcinogen, and urinary trans,trans-muconic acid is a key biomarker of benzene exposure. Whether specific occupational groups in Korea have higher benzene body burdens beyond environmental levels has not been systematically evaluated using national biomonitoring data with smoking-adjusted biomarkers.
  • Methods
    We pooled adult data from three cycles (3rd–5th) of the Korean National Environmental Health Survey (KoNEHS, 2015–2023; n = 10,786). Creatinine-adjusted urinary trans,trans-muconic acid (UttMACr) was analyzed across nine occupation groups using complex-samples general linear models with nested covariates. Sampling weights were rescaled within each cycle to harmonize weight scales across waves. Smoking was controlled using questionnaire-based variables and five-level creatinine-adjusted urinary cotinine (COTCr). Variance decomposition quantified predictors' contributions, and sensitivity analyses included COTCr-stratified models and analyses limited to the economically active population.
  • Results
    Although population geometric mean UttMACr declined by 51% from the 3rd to 5th cycle, occupation remained a significant independent predictor after adjusting for temporal trends, smoking, demographic, and lifestyle covariates (ΔR² = 0.0018, p = 0.003). Cleaning/guard/elementary workers showed the highest elevation in the geometric mean ratio (GMR: 1.13; p < 0.001), followed by food/textile/other manufacturing (GMR: 1.12; p < 0.05) and chemical/metal/machinery manufacturing workers (GMR: 1.08; p < 0.05). COTCr-stratified analyses revealed occupation-specific patterns attenuated in pooled models: among participants with COTCr ≤ 1.0 μg/g creatinine, construction/mining (GMR: 1.62; p < 0.001) and agriculture/fishery (GMR: 1.41; p < 0.01) showed marked elevations.
  • Conclusions
    Population UttMACr levels in Korea declined over time, but occupation-specific differences persisted after adjustment for trends, smoking, and sociodemographic factors. These findings suggest occupational benzene exposures may extend beyond traditionally recognized industrial settings and support more targeted biomonitoring, particularly for worker groups not routinely prioritized in current surveillance programs.
  • Keywords: Benzene; Occupational exposure; Biological monitoring; Trans,trans-muconic acid; Cotinine
Benzene is classified by the International Agency for Research on Cancer as a group 1 human carcinogen, and chronic exposure is associated with hematopoietic malignancies.1,2 In Korea, the occupational exposure limit for benzene has been progressively tightened over time. The 8-hour time-weighted average, historically influenced by American Conference of Governmental Industrial Hygienists threshold limit value–based practice, was set at 10 ppm in 1986, reduced to 1 ppm (3,190 μg/m³) in 2003, and further tightened to 0.5 ppm in 2016. A short-term exposure limit was introduced at 5 ppm in 2007 and later tightened to 2.5 ppm in 2016.3-6 In the environmental sector, an air quality standard of 5 μg/m³ (0.00157 ppm) for benzene was established in accordance with European standards at the end of 2006 and was implemented beginning in 2010 as monitoring infrastructure expanded.7 Despite these regulatory changes, benzene exposure remains a significant concern in both industrial and community settings. In cities hosting petrochemical complexes, ambient benzene concentrations exceeding the environmental standard have been reported at some sites.8,9 During maintenance or turnaround operations in petrochemical facilities, workers may also experience exposures exceeding occupational benchmarks.10,11
Another challenge is unrecognized benzene contamination in products that are not labeled as benzene-containing. Material safety data sheets (MSDS) for petroleum-derived paint thinners and cleaning agents formulated with varying proportions of toluene, xylene, ethylbenzene, and styrene often state that benzene is absent; however, measurable benzene contamination has been documented in some such products.12,13 Therefore, personnel involved in the use of petroleum-based cleaning agents or similar solvents may be at risk of benzene exposure, which is not accounted for in the conventional hazardous substances communication system.
Benzene exposure arises through both occupational and non-occupational pathways, including tobacco smoke, motor vehicle emissions, indoor combustion, industrial emissions, and vapors from consumer products.14,15 Urinary trans,trans-muconic acid (UttMA) is a widely used biomarker of benzene exposure, with an elimination half-life of approximately 5 hours.16 At low exposure levels, dietary sorbic acid can substantially increase UttMA,17,18 so its specificity is limited compared with S-phenylmercapturic acid (SPMA).19 Despite these limitations, UttMA was selected as a biomarker of benzene exposure in this study because of its historical use in national biomonitoring programs, including previous Korean National Environmental Health Survey (KoNEHS) cycles. Accordingly, analyses based on UttMA require careful attention to smoking and other potential sources of background exposure.
The KoNEHS is a legally mandated nationwide biomonitoring program conducted every 3 years since 2009 under Article 14 of the Environmental Health Act.20 KoNEHS data have been used to describe urinary benzene metabolite levels in the general Korean population and to examine associations with selected health outcomes.21,22 However, published KoNEHS analyses have not specifically evaluated whether benzene-related biomarker levels differ across occupational groups after rigorous control for smoking and temporal trends. Therefore, this study aimed to compare UttMA levels across occupational groups and to estimate the extent to which occupation contributes to internal benzene exposure beyond background environmental exposure. This exploratory analysis aimed to determine whether systematic differences in urinary benzene biomarker levels exist among occupational groups, beyond what can be attributed to smoking and temporal trends. A Korean translation of this article is available in Supplementary Data 1.
Study design and participants
This study used pooled data from the 3rd (2015–2017; n = 3,787), 4th (2018–2020; n = 4,239), and 5th (2021–2023; n = 4,279) cycles, yielding a total of 12,305 adults aged 19 years or older. Each cycle used a stratified two-stage probability-proportional-to-size cluster sampling design based on census enumeration districts, with stratification by administrative region and area type (urban, rural, and coastal), thereby preserving national representativeness. Participants provided spot urine and venous blood samples and completed interviewer-administered questionnaires on demographics, lifestyle, residential environment, and dietary habits. To focus on occupational comparisons among working-age adults, students and military personnel were excluded. Participants with urinary creatinine concentrations outside the World Health Organization validity range of 0.3–3.0 g/L were also excluded to reduce distortion caused by extreme urine dilution or concentration.23 The final analytic sample comprised 10,786 participants.
Occupation classification
Occupation was classified according to the two-digit sub-major groups of the Korean Standard Classification of Occupations (KSCO). To harmonize codes across survey waves, KSCO 6th-revision codes used in cycle 3 were mapped to KSCO 7th-revision codes used in cycles 4 and 5.24 In total, 44 sub-major codes were harmonized. For analysis, occupations were regrouped into nine broader categories defined a priori with consideration of plausible benzene-related exposure potential and sample size (Table 1, Supplementary Tables 1 and 2): (A) homemaker/unemployed (reference group), (B) agriculture/fishery, (C) service/sales, (D) manager/professional/clerical, (E) driver/transport, (F) chemical/metal/machinery manufacturing, (G) food/textile/other manufacturing, (H) construction and mining-related trade occupations (construction/mining), and (I) elementary occupations (including cleaning and security).
The regrouping was intended to improve statistical power while retaining occupational interpretability. The homemaker/unemployed group was designated as the reference category because these individuals were less likely to experience workplace chemical exposures and therefore approximate general environmental background exposure levels. Although this group is not entirely free from benzene exposure (e.g., through household cleaning products), it represents the lowest exposure baseline available in a general population survey. To verify the robustness of the findings, a secondary analysis using agriculture/fishery as the reference was conducted within the economically active population.
Outcome variable and covariates
The primary outcome was the natural logarithm of creatinine-adjusted urinary trans,trans-muconic acid (LogUttMACr; μg/g creatinine). Because spot urine concentrations are influenced by urine dilution, UttMA was normalized to urinary creatinine according to standard biomonitoring practices and KoNEHS reporting conventions. Covariates are detailed in the footnotes to Table 2. In brief, covariates included residential region, a traffic exposure score, sex, age, education, monthly household income, alcohol consumption, regular exercise, and the frequency of grilled meat and grilled fish consumption.
Smoking-related adjustment included current smoking status, secondhand smoke exposure frequency, and creatinine-adjusted urinary cotinine (COTCr). COTCr was modeled as a five-level ordinal variable. To address potential residual confounding by tobacco smoke, a COTCr value of ≤1.0 μg/g creatinine was used as the reference category, and higher values were categorized into percentile-based groups (1st–50th, 51st–75th, 76th–95th, and >95th percentiles). Using both questionnaire-based and biomarker-based smoking indicators was intended to reduce residual confounding from active smoking, secondhand smoke, and low-level nicotine exposure not fully captured by self-report.25-27
KoNEHS uses a residence-based stratified sampling design, oversampling rural and coastal areas for environmental health monitoring. Because KoNEHS sampling is residence-based rather than occupation-based, the weighted occupational distribution may differ from the national labor force structure. While sampling weights adjust for residential-area selection, they do not include occupation-specific post-stratification. Accordingly, weighted occupational proportions should not be interpreted as representing the national labor force structure.
Pooling strategy: weight rescaling and unique sampling identifiers
All analyses accounted for the complex sampling design by specifying strata, primary sampling units, and sampling weights in the complex-samples module of SPSS version 29 (IBM Corp., Armonk, NY, USA). A major challenge in pooling KoNEHS cycles was that cycle 3 used standardized relative weights with a mean near 1.0, whereas cycles 4 and 5 used population-projection weights on a much larger numeric scale. If combined directly, cycle 3 would have had minimal influence on pooled estimates. To harmonize the contribution of each cycle, weights were rescaled within each cycle so that the mean weight equalled 1.0, allowing pooled analyses to reflect cycle sample sizes rather than incompatible target-population scales.
Because each KoNEHS cycle represents an independent sampling frame and may reuse similar stratum or cluster codes, unique cycle-specific identifiers were generated for strata and primary sampling units before pooling. This step prevented the artificial fusion of clusters across cycles, which could otherwise underestimate standard errors and overstate statistical significance.
Statistical analysis
Weighted arithmetic means, standard deviations, geometric means (GM), and geometric standard deviations of UttMACr were calculated by occupation and participant characteristics. Weighted percentile distributions of UttMACr were also generated by survey cycle, sex, worker status, and smoking status (Supplementary Table 3). These distributions—stratified by cotinine-based smoking category, including a non-smoker/passive-smoker/active-smoker classification based on COTCr—are intended to provide population-level reference values against which occupational benzene biomarker levels can be compared after accounting for background tobacco smoke contributions. We fitted a sequence of complex-samples general linear models with LogUttMACr as the dependent variable (Table 2). Exponentiated regression coefficients are presented as geometric mean ratios (GMRs), interpreted as the ratio of the geometric mean UttMACr for each occupation group relative to the reference group. The model sequence was as follows: model 1 included occupation only; model 2 added survey cycle; model 3 added region; model 4 included occupation, survey cycle, and region; model 5 added demographic, lifestyle, and traffic-exposure covariates; model 6 added smoking variables to model 4; and model 7 included all covariates. The incremental contribution of occupation to explained variance was estimated by comparing the full model with a reduced model of the same specification but excluding all occupation indicators. To quantify the relative contribution of predictor blocks, we calculated changes in weighted R² across the sequential model hierarchy. Because these estimates are order-dependent, they were interpreted descriptively. The three smoking-related covariates were entered simultaneously to maximize tobacco-related confounding control. Although self-reported smoking status and COTCr are correlated by design (Cramér’s V: 0.86), all variance inflation factors remained below 5, and the condition number was 14.8, indicating no problematic multicollinearity (Table 2).
Two sensitivity analyses were performed. To assess whether the inclusion of nonworkers influenced the occupational contrasts, we repeated the full model among economically active participants using agriculture/fishery as the reference group (Table 3). To further address potential residual confounding from tobacco smoke, we stratified the full model by COTCr level (≤1.0 vs. >1.0 μg/g creatinine), thereby separating participants by objective smoking intensity and allowing occupational contrasts to be examined within strata of objectively measured tobacco exposure (Table 4).
Ethics statement
This secondary analysis used de-identified KoNEHS data provided by the National Institute of Environmental Research (NIER). Cycle 5 data access was approved under NIER data-use approval No. NIER-2021-01-01-013. Because the analysis used de-identified survey data, no additional participant contact or intervention was involved.
Participant characteristics
Table 1 summarizes the characteristics of the 10,786 participants across nine occupation groups. The highest GMs for UttMACr were observed in construction/mining (79.6 μg/g creatinine), chemical/metal/machinery manufacturing (78.5), and food/textile/other manufacturing (77.9), and lowest in agriculture/fishery (64.5). A marked secular decline was observed across KoNEHS cycles. The population GM decreased from 108.0 μg/g creatinine in cycle 3 to 65.1 μg/g creatinine in cycle 4 and 51.3 μg/g creatinine in cycle 5, and this pattern was broadly consistent across occupation groups. Current smoking prevalence ranged from 11.4% in homemakers/unemployed participants to 42.5% in construction/mining workers.
Stepwise regression and variance decomposition
Table 2 presents the stepwise modeling results. In the unadjusted model (M1), six occupation groups had significantly elevated GMRs relative to the homemaker/unemployed reference group. Adding survey cycle (M2) produced the largest single gain in explained variance (ΔR² = 0.1163), indicating that the secular trend was the dominant predictor of UttMACr variation. Adjustment for smoking-related variables alone (M6) substantially attenuated several crude occupational contrasts. For example, the GMR for driver/transport decreased from 1.10 to 0.97, that for construction/mining from 1.15 to 1.00, and that for chemical/metal/machinery manufacturing from 1.11 to 0.99, suggesting that part of the crude elevation in these groups was attributable to differential smoking prevalence rather than occupation alone.
In the fully adjusted model (M7), three occupation groups remained significantly elevated: cleaning/guard/elementary workers (GMR: 1.13; p < 0.001), food/textile/other manufacturing workers (GMR: 1.12, p < 0.05), and chemical/metal/machinery manufacturing workers (GMR: 1.08; p < 0.05). The occupation-specific increment in explained variance was small but statistically significant (ΔR² = 0.0018; F(8,10745) = 2.976, p = 0.0025).
Full model covariates and economically active population analysis
Table 3 presents the full-model covariate estimates. Relative to cycle 3, the GM was 39.5% lower in cycle 4 (GMR: 0.605; p < 0.001) and 51.0% lower in cycle 5 (GMR: 0.490; p < 0.001). Current smoking was associated with a 28.6% elevation in UttMACr (GMR: 1.286), and the highest COTCr category (>95th percentile) was associated with a 70.1% elevation (GMR: 1.701), indicating a strong exposure-response relationship. Women showed higher levels than men (GMR: 1.243; p < 0.001), and rural residence was associated with slightly lower levels than urban residence (GMR: 0.931; p < 0.01). In the economically active population analysis, with agriculture/fishery as the reference group, only cleaning/guard/elementary workers remained significantly elevated (GMR: 1.124; p < 0.05); the broadly similar findings between the full-population and economically active models suggest that the main inference was not driven solely by the inclusion of nonworkers.
Cotinine-stratified analysis
Table 4 shows that cotinine stratification revealed occupational contrasts that were less apparent in the pooled models. Among participants with COTCr ≤ 1.0 μg/g creatinine, construction/mining showed the largest elevation observed in the study (GMR: 1.62; p < 0.001), followed by agriculture/fishery (GMR: 1.41; p < 0.01) and food/textile/other manufacturing (GMR: 1.36; p < 0.05). Among participants with COTCr > 1.0 μg/g creatinine, only cleaning/guard/elementary workers remained significantly elevated (GMR: 1.12; p < 0.01). These findings indicate that residual tobacco-related background can obscure modest occupation-related biomarker differences, even after multivariable adjustment.
Hierarchical variance decomposition
When predictor contributions were partitioned sequentially according to the prespecified sequential entry order shown in Table 5, among the total model R² (0.178), survey cycle accounted for the largest share (ΔR² = 0.1163, 65.3% of total R²), followed by smoking-related variables (25.5%), demographics/lifestyle/traffic exposure (6.1%), region (0.6%), and the unique contribution of occupation (1.0%) (Table 6). The unadjusted occupation R² (M1: 0.0044, 2.5%) was reduced to 1.0% after full adjustment, indicating that approximately 60% of the crude occupational effect was confounded by smoking and other correlated factors. Nevertheless, the adjusted occupation ΔR² remained significant (p = 0.003), confirming that occupation is independently associated with UttMACr variation even after accounting for all measured confounders. The total unexplained variance was 82.2%, reflecting unmeasured sources including individual metabolic variability and dietary sorbic acid intake. The relatively small unique contribution of occupation is consistent with the use of broad occupational categories and with the dilution of workplace signals in a general-population survey that includes many nonworkers and environmentally exposed individuals.
Overview of the main findings
Using pooled KoNEHS data from 2015 to 2023, this study found that UttMACr declined substantially over time in the Korean adult population, yet occupation-specific differences persisted after adjustment for temporal trends, smoking, and sociodemographic covariates. The most consistent elevation was observed among cleaning/guard/elementary workers, whereas cotinine-stratified analyses additionally revealed notable elevations among non-smoking construction/mining and agriculture/fishery participants.
Cleaning/guard/elementary workers as a consistently elevated group
The most consistent signal in this study was the elevation among cleaning/guard/elementary workers. This group remained elevated in the fully adjusted model and in the economically active population analysis, and it was also elevated among participants with COTCr > 1.0 μg/g creatinine in the stratified analysis. Although the absolute magnitude of elevation was modest, its consistency across analytical approaches strengthens the conclusion that this group may experience sustained benzene-related exposure above background levels. These workers may encounter benzene-related exposures from petroleum naphtha-derived cleaning agents, solvent-containing products, and building maintenance materials. Although the 13% relative elevation was modest, its consistency suggests recurrent exposure rather than a chance finding. Because trans,trans-muconic acid (ttMA) reflects recent exposure, the present data cannot establish cumulative dose or long-term persistence, and individual-level longitudinal data would be required to confirm chronic exposure patterns. Notably, this finding challenges the prevailing assumption that meaningful occupational benzene exposure in Korea is limited to heavy industrial or petrochemical environments.12,13 Given that these workers are not subject to mandatory working environment measurements (WEM) for benzene under current Korean occupational health regulations, this population may be overlooked by current surveillance systems.
The elevation observed among cleaning/guard/elementary workers may be interpreted through several potential mechanisms. First, direct chemical exposure could occur through petroleum-derived cleaning agents, idling vehicles in parking facilities, and residual indoor air contamination, as discussed above. Second, the finding may reflect indirect exposure pathways related to occupational precariousness—such as environmental tobacco smoke in poorly ventilated workplaces, residential proximity to traffic and industrial emissions, and informal multiple job holding—which are commonly associated with this workforce. Both interpretations remain speculative at present, and further targeted research, including task-level exposure assessment, would be needed to clarify which mechanisms, if any, contribute to the observed elevation.
Temporal decline and its interpretation
The crude population GM decreased by approximately 52.5%, from 108.0 μg/g creatinine in cycle 3 to 51.3 μg/g creatinine in cycle 5. This decline likely reflects multiple concurrent changes, including tighter fuel and vehicle emission controls, broader ambient air quality management, and stricter occupational benzene limits.4,28,29 Despite this overall decline, the persistence of relative occupational differences suggests that some workplace-related exposure sources remain even amid broad environmental improvements. Analytically, because survey cycle explained the largest share of variance, pooling KoNEHS waves without cycle adjustment could yield misleading occupational estimates.
Smoking confounding can mask occupational patterns
The multi-component smoking adjustment (smoking status, secondhand smoke exposure, and COTCr level) accounted for 25.5% of explained variance, confirming tobacco smoke as the second most important determinant after temporal trends (Table 6). A notable feature of the analysis was the extent to which smoking altered occupational contrasts (Table 4). Importantly, cotinine stratification revealed marked elevations in selected occupation groups that were not obvious in the pooled models. Construction/mining workers showed a 62% elevation (GMR: 1.62) among non-smokers but no effect among participants with higher cotinine levels (GMR: 1.01); agriculture/fishery showed a 41% elevation only in the low-cotinine stratum. Tobacco-smoke-related background exposure may overwhelm modest occupational signals, analogous to the difficulty of detecting a modest light source against a bright background.28,29 These results suggest that conventional regression adjustment may be insufficient when smoking prevalence differs sharply across occupations and when the biomarker itself is sensitive to tobacco-related exposure.30 In future biomonitoring studies of low-level benzene exposure, cotinine-based stratification or similarly robust smoking control may be a useful complement to standard covariate adjustment.
Agriculture/fishery and construction and mining-related trade occupations findings
The elevations observed among participants with low COTCr in the agriculture/fishery and construction/mining groups were unexpected and should be interpreted cautiously. Potential exposure sources include fuel handling, diesel exhaust, biomass combustion, pesticide-related solvents, and other mixed exposures encountered during field or outdoor work.31 Because the present occupation groups were broad and no task-level exposure measurements were available, these findings should be considered hypothesis-generating rather than definitive evidence of specific sources.
In particular, the agriculture/fishery elevation among non-smokers (GMR: 1.41) warrants cautious consideration. Although urinary cotinine is widely used to reflect recent tobacco-derived nicotine exposure, it may be less sensitive to non-tobacco combustion sources. Candidate exposure pathways that have been suggested in the literature include biomass burning for residential heating and cooking—which may still be encountered in some rural areas of Korea and has been described as a possible source of benzene—combustion exhaust from agricultural machinery using diesel or gasoline fuel, and benzene-containing solvent carriers that may be present in certain pesticide formulations. Such exposures would not necessarily be fully captured by cotinine alone. Accordingly, the residual elevation among non-smoking agriculture/fishery workers may be partly attributable to non-tobacco environmental and occupational benzene sources, although this is only one of several possible explanations and should not be solely attributed to residual smoking confounding. Definitive attribution would require task-level exposure measurements that were not available in the present study.
Interpreting the small but significant occupational ΔR²
The unique occupation-related increment in explained variance was ΔR² = 0.0018, equivalent to 0.18 percentage points of total variance. This ΔR² is small and may appear negligible, but this estimate should be interpreted in light of the study design. KoNEHS is a general-population biomonitoring survey rather than an occupational cohort, 36.5% of participants were homemakers/unemployed, and occupations were aggregated into broad groups. Under these conditions, any true workplace-related signal is likely to be diluted. Thus, the statistical significance of the occupation term should not be interpreted as evidence of a large effect size, but neither should the small ΔR² be dismissed as meaningless. Nevertheless, occupation maintained a statistically discernible association even after controlling for major confounding variables. Under the Korean Occupational Safety and Health Act,32 WEM are generally triggered by the presence of regulated hazardous substances identified in workplace chemical inventories and safety documents, and special health examinations are conducted for designated exposures. However, MSDSs for many petroleum-derived products used in workplaces may not specify benzene when it is present only as a trace contaminant; consequently, WEM and biological monitoring for benzene may not be conducted. Consequently, administrative records may classify such workers as unexposed. Given the carcinogenicity of benzene, even modest and recurrent differences in exposure may warrant preventive attention.33-36
Strengths and limitations
This study has several strengths. It used a large nationally representative biomonitoring dataset, pooled three survey cycles while explicitly addressing incompatible weight scales, and incorporated both questionnaire-based and biomarker-based smoking control. The analysis also examined model building sequentially and presented the incremental contribution of occupation transparently.
Several limitations should be acknowledged. First, the cross-sectional design precludes causal inference. Second, dietary sorbic acid intake was not measured; because ttMA has limited specificity at low benzene exposure levels, unmeasured dietary sorbic acid intake across occupational groups may have contributed to residual confounding.17,18 Third, spot urine specimens reflect recent exposure and are subject to within-person variability. Fourth, the broad occupational categories may have introduced substantial exposure misclassification. Fifth, no workplace air-monitoring or task-level data were available to validate putative exposure sources. Sixth, a healthy worker effect could bias estimates toward the null in physically demanding occupations such as construction/mining. Seventh, the higher UttMACr values observed in women may partly reflect sex-specific creatinine excretion differences.37 Future studies would benefit from concurrent measurement of more specific biomarkers such as SPMA and from linkage to job-exposure matrices.
In pooled KoNEHS data from 2015 to 2023, UttMACr decreased markedly over time in the Korean adult population, but occupation-specific differences remained after adjustment for smoking, temporal trend, and sociodemographic factors. Cleaning/guard/elementary workers showed the most consistent elevation, and cotinine-stratified analyses suggested additional signals among non-smoking construction/mining and agriculture/fishery participants. The population-level UttMACr percentile distributions stratified by cotinine-based smoking status (Supplementary Table 3) may serve as background reference values for comparing benzene biomarker levels in occupational monitoring programs. These findings do not establish causality, but they indicate that benzene-related exposure may not be confined to traditionally recognized industrial settings.
Because ttMA is strongly influenced by smoking and has limited specificity at low exposure levels, more specific biomarkers such as SPMA may be preferable for targeted surveillance of occupational benzene exposure. More targeted biomonitoring and exposure assessment, particularly for worker groups not routinely captured by existing surveillance programs, may help identify overlooked occupational exposure scenarios.

AM

arithmetic mean

CI

confidence interval

COTCr

creatinine-adjusted urinary cotinine

CSGLM

complex-samples general linear models

EAP

economically active population

GM

geometric mean

GMR

geometric mean ratio

GSD

geometric standard deviation

ISCO

International Standard Classification of Occupations

KoNEHS

Korean National Environmental Health Survey

KSCO

Korean Standard Classification of Occupations

LogUttMACr

natural logarithm of creatinine-adjusted urinary trans,trans-muconic acid

MSDS

material safety data sheet

NIER

National Institute of Environmental Research

SD

standard deviation

SHS

secondhand smoke

SPMA

S-phenylmercapturic acid

ttMA

trans,trans-muconic acid

UttMA

urinary trans,trans-muconic acid

UttMACr

creatinine-adjusted urinary trans,trans-muconic acid

WEM

working environment measurement

WLS

weighted least squares

Competing interests

The author declares that they have no competing interests.

Acknowledgments

The author gratefully acknowledges the National Institute of Environmental Research for granting permission to use the raw data from the Korean National Environmental Health Survey.

Supplementary Data 1.
Korean translation of "Occupational differences in benzene-related biomarker levels beyond traditional industrial settings: findings from the Korean National Environmental Health Survey, 2015–2023."
aoem-2026-38-e19_Supplementary-Data-1.pdf
Supplementary Table 1.
UttMACr by major KSCO.
aoem-2026-38-e19_Supplementary-Table-1.pdf
Supplementary Table 2.
UttMACr by major ISCO.
aoem-2026-38-e19_Supplementary-Table-2.pdf
Supplementary Table 3.
Percentile of UttMACr.
aoem-2026-38-e19_Supplementary-Table-3.pdf
Table 1.
Characteristics of study participants by occupation group: UttMACr GM and weighted percentages
Characteristic Occupationa Total
A B C D E F G H I
No. 3,937 919 1,305 2,389 456 566 280 285 649 10,786
UttMACr AM (SD) 108.6 (240.5) 97.3 (141.5) 112.5 (161.5) 105.4 (146.2) 108.3 (114.8) 110.0 (104.5) 119.7 (146.3) 114.1 (120.1) 117.4 (146.8) 108.6 (181.6)
UttMACr GM (GSD) 67.2 (2.45) 64.5 (2.29) 71.1 (2.44) 67.1 (2.47) 73.8 (2.39) 78.5 (2.27) 77.9 (2.42) 79.6 (2.28) 76.3 (2.44) 69.5 (2.43)
Survey cycle
 3rd 104.3 (30.6) 93.7 (24.4) 115.9 (27.1) 105.4 (29.1) 116.0 (30.8) 106.6 (36.2) 109.0 (31.2) 134.4 (34.6) 125.6 (29.2) 108.0 (29.8)
 4th 64.7 (32.0) 62.9 (44.3) 61.1 (37.3) 63.9 (34.0) 64.8 (30.2) 71.5 (35.6) 74.3 (33.1) 63.9 (30.3) 75.9 (33.7) 65.1 (34.0)
 5th 48.4 (37.4) 50.1 (31.4) 57.5 (35.6) 49.2 (37.0) 57.1 (39.0) 59.6 (28.2) 60.8 (35.7) 57.4 (35.1) 51.7 (37.1) 51.3 (36.1)
Sex
 Male 59.5 (30.2) 63.5 (65.8) 70.1 (40.3) 65.8 (60.0) 73.6 (97.5) 78.4 (92.5) 73.5 (57.0) 78.8 (95.4) 74.7 (39.3) 67.9 (51.3)
 Female 70.8 (69.8) 66.6 (34.2) 71.8 (59.7) 69.0 (40.0) 78.6 (2.5) 79.6 (7.5) 84.2 (43.0) 96.6 (4.6) 77.3 (60.7) 71.3 (48.7)
Age (years)
 19–29 51.1 (12.1) 102.8 (1.6) 64.3 (16.1) 61.9 (18.1) 66.0 (7.3) 73.8 (15.1) 78.6 (12.5) 62.1 (9.9) 142.6 (5.2) 61.1 (13.3)
 30–39 75.3 (13.5) 82.1 (4.8) 84.6 (16.8) 65.5 (27.9) 93.2 (11.1) 95.5 (21.1) 88.9 (17.2) 72.5 (13.3) 100.1 (6.2) 74.2 (17.6)
 40–49 70.9 (13.0) 64.8 (7.4) 77.9 (19.7) 73.6 (30.6) 93.9 (20.3) 70.5 (28.1) 75.3 (19.2) 110.3 (18.1) 104.1 (9.3) 75.2 (19.8)
 50–59 75.3 (15.9) 75.2 (24.6) 67.3 (28.6) 66.4 (17.5) 83.0 (33.0) 91.2 (23.6) 77.3 (32.7) 90.7 (33.9) 88.3 (23.3) 74.7 (20.9)
 60–69 68.2 (21.1) 59.8 (29.1) 65.5 (13.6) 63.0 (4.8) 52.2 (21.4) 54.4 (9.1) 65.7 (14.8) 63.5 (20.3) 62.8 (26.6) 64.3 (15.4)
 ≥70 64.4 (24.3) 57.9 (32.4) 65.9 (5.1) 54.2 (1.2) 46.8 (6.9) 69.2 (3.0) 108.1 (3.5) 49.8 (4.5) 61.9 (29.3) 62.8 (13.1)
Education
 ≤Elementary 56.5 (36.1) 56.8 (48.8) 57.2 (26.5) 54.9 (9.1) 65.0 (33.2) 65.2 (19.3) 72.6 (31.2) 56.9 (35.3) 63.8 (47.9) 58.3 (27.3)
 Middle school 73.3 (12.5) 57.8 (18.5) 76.5 (12.2) 42.8 (6.4) 56.9 (14.4) 68.2 (10.1) 69.3 (16.9) 104.4 (15.2) 88.0 (13.5) 66.7 (11.2)
 High school 67.3 (26.7) 82.5 (23.6) 73.2 (30.9) 62.5 (27.8) 86.3 (39.8) 87.5 (39.4) 87.2 (35.6) 94.8 (33.4) 82.1 (28.6) 71.9 (29.2)
 ≥College 82.7 (24.7) 84.4 (9.1) 81.0 (30.4) 75.4 (56.7) 84.5 (12.6) 80.4 (31.3) 78.9 (16.3) 89.5 (16.1) 119.9 (10.0) 79.3 (32.2)
Monthly household income (KRW)
 <2 million 64.4 (39.4) 63.8 (56.8) 68.1 (21.2) 64.3 (6.4) 85.7 (22.5) 98.8 (9.4) 80.5 (30.0) 109.3 (25.2) 69.5 (47.8) 67.9 (26.1)
 2–3 million 73.3 (19.3) 59.9 (19.9) 82.2 (20.9) 74.0 (14.8) 70.9 (27.0) 86.8 (29.4) 85.3 (22.8) 83.3 (24.6) 79.1 (19.3) 75.9 (19.5)
 3–5 million 66.2 (22.9) 61.8 (14.4) 67.0 (28.2) 70.2 (34.1) 71.8 (31.5) 68.5 (37.9) 72.7 (32.6) 61.7 (30.9) 94.6 (24.1) 69.2 (28.1)
 ≥5 million 68.3 (18.4) 87.4 (8.9) 70.2 (29.6) 63.0 (44.7) 68.3 (19.1) 78.8 (23.3) 74.2 (14.6) 74.5 (19.3) 64.7 (8.8) 66.9 (26.3)
Region
 Coastal 63.7 (4.6) 59.3 (36.6) 67.2 (6.3) 62.5 (3.2) 85.8 (6.6) 112.8 (4.6) 100.1 (6.2) 51.1 (2.9) 146.5 (6.6) 69.2 (6.2)
 Rural 59.6 (11.4) 65.7 (46.5) 58.6 (11.2) 66.4 (8.6) 73.8 (14.4) 69.7 (13.5) 66.9 (12.8) 95.2 (11.5) 73.5 (14.1) 64.8 (12.8)
 Urban 68.5 (84.0) 73.6 (17.0) 73.3 (82.4) 67.3 (88.2) 72.8 (79.0) 78.4 (81.9) 78.3 (81.0) 78.9 (85.5) 72.7 (79.3) 70.3 (81.0)
Traffic exposure
 Low 65.2 (30.9) 64.0 (56.1) 72.6 (28.1) 65.4 (23.1) 76.0 (28.8) 88.5 (28.8) 83.2 (31.0) 85.5 (27.9) 81.2 (32.5) 69.5 (29.4)
 Medium 68.3 (50.5) 65.4 (41.3) 72.5 (50.2) 67.4 (55.5) 74.0 (49.2) 77.7 (49.9) 71.9 (52.4) 79.9 (52.5) 75.8 (53.3) 69.9 (51.6)
 High 67.7 (18.6) 60.5 (2.6) 66.1 (21.7) 68.2 (21.4) 70.3 (21.9) 68.4 (21.3) 89.0 (16.6) 71.1 (19.6) 67.6 (14.3) 68.3 (19.0)
Smoking
 Never/former 65.4 (88.6) 58.3 (78.8) 67.3 (81.7) 61.8 (79.4) 56.7 (58.2) 64.1 (58.7) 71.7 (73.9) 60.5 (57.5) 68.8 (84.2) 64.1 (80.0)
 Current 83.4 (11.4) 94.3 (21.2) 90.7 (18.3) 91.6 (20.6) 106.5 (41.8) 104.9 (41.3) 98.7 (26.1) 115.2 (42.5) 131.7 (15.8) 95.8 (20.0)
SHS exposure
 None 67.0 (92.1) 63.2 (86.7) 70.0 (84.2) 64.6 (80.5) 69.1 (74.9) 75.2 (67.6) 81.9 (79.7) 75.5 (68.7) 74.0 (88.6) 67.9 (84.2)
 1–4 times/week 62.9 (5.1) 72.2 (9.8) 75.4 (9.6) 70.3 (13.9) 73.7 (15.6) 94.4 (18.4) 55.2 (11.6) 65.4 (11.4) 84.2 (6.9) 72.2 (10.1)
 ≥5 times/week 83.3 (2.8) 78.4 (3.5) 80.2 (6.2) 101.7 (5.6) 123.5 (9.4) 75.8 (14.0) 78.4 (8.7) 106.8 (19.8) 120.5 (4.4) 91.7 (5.7)
COTCr level
 ≤1.0 63.7 (21.7) 67.5 (15.1) 71.5 (17.4) 62.2 (22.0) 60.7 (11.4) 64.4 (15.4) 76.4 (15.3) 89.4 (16.1) 68.0 (21.9) 65.1 (19.8)
 1.0–p50 61.6 (41.0) 58.5 (39.9) 63.2 (40.0) 61.1 (38.9) 53.3 (32.8) 65.7 (33.2) 73.9 (37.4) 53.4 (29.8) 65.7 (37.5) 61.7 (38.8)
 p50–p75 70.9 (22.4) 47.8 (20.6) 70.8 (20.9) 61.4 (15.9) 66.9 (13.9) 61.4 (8.9) 59.2 (20.0) 54.2 (10.9) 70.1 (20.8) 65.9 (18.6)
 p75–p95 79.3 (12.6) 80.0 (16.7) 84.0 (18.6) 84.1 (19.1) 91.1 (30.3) 93.7 (35.2) 96.0 (19.9) 96.1 (29.8) 120.5 (15.3) 86.6 (18.4)
 >p95 125.9 (2.3) 137.0 (7.7) 120.3 (3.1) 118.6 (4.1) 145.3 (11.5) 153.9 (7.3) 127.8 (7.3) 151.9 (13.3) 143.7 (4.5) 132.2 (4.4)
Alcohol drinking
 Non-drinker 65.7 (38.8) 63.1 (38.3) 69.6 (24.7) 64.8 (15.9) 57.7 (16.6) 72.1 (14.0) 83.2 (24.6) 92.8 (19.3) 73.5 (37.0) 67.1 (27.2)
 Occasional 64.3 (36.7) 65.0 (24.5) 71.8 (34.7) 67.5 (37.7) 81.3 (25.0) 72.5 (32.8) 80.2 (33.1) 66.1 (23.3) 77.2 (35.7) 68.2 (34.9)
 Frequent 74.4 (24.5) 65.7 (37.1) 71.4 (40.6) 67.5 (46.3) 75.9 (58.3) 84.4 (53.2) 73.4 (42.3) 81.5 (57.4) 78.9 (27.3) 72.5 (38.0)
Regular exercise
 No 68.1 (56.7) 66.6 (75.7) 74.2 (67.8) 67.4 (61.3) 78.0 (54.7) 78.9 (67.1) 93.4 (67.0) 84.8 (68.8) 77.9 (60.1) 71.2 (61.6)
 Yes 66.0 (43.3) 58.5 (24.3) 65.1 (32.2) 66.6 (38.7) 69.0 (45.3) 77.8 (32.9) 53.9 (33.0) 69.2 (31.2) 73.8 (39.9) 66.8 (38.4)
Grilled meat consumption
 Never/rarely 64.1 (37.6) 59.1 (40.5) 66.7 (27.9) 64.6 (18.1) 64.9 (23.9) 77.0 (21.6) 69.6 (34.5) 88.1 (25.0) 74.0 (45.8) 66.1 (29.5)
 Monthly 70.0 (49.3) 67.9 (50.0) 71.6 (53.5) 68.1 (59.2) 76.6 (59.0) 73.4 (58.7) 81.8 (53.3) 75.9 (54.8) 71.9 (41.4) 70.6 (53.4)
 Weekly+ 65.9 (13.2) 71.1 (9.5) 76.5 (18.6) 66.4 (22.7) 77.5 (17.1) 98.0 (19.7) 86.7 (12.2) 79.6 (20.2) 102.9 (12.8) 72.1 (17.0)
Grilled fish Consumption
 Never/rarely 65.4 (71.4) 64.3 (70.9) 68.9 (67.5) 65.1 (65.2) 69.5 (70.9) 79.1 (71.6) 73.9 (72.0) 82.7 (70.9) 72.2 (71.0) 67.7 (69.2)
 Monthly 72.5 (23.1) 65.9 (23.1) 73.9 (26.3) 71.0 (30.5) 85.4 (22.1) 70.1 (23.9) 91.5 (22.8) 69.8 (21.8) 70.5 (20.3) 72.4 (25.4)
 Weekly+ 68.7 (5.5) 61.6 (6.0) 85.4 (6.2) 70.1 (4.3) 84.8 (7.0) 128.4 (4.4) 81.5 (5.2) 80.9 (7.3) 144.7 (8.6) 79.5 (5.5)

UttMACr: creatinine-adjusted urinary trans,trans-muconic acid; GM: geometric mean; ttMACr: creatinine-adjusted trans,trans-muconic acid; AM: arithmetic mean; SD: standard deviation; GSD: geometric standard deviation; KRW: Korean Won; SHS: secondhand smoke; COTCr: creatinine-adjusted urinary cotinine; p50, p75, and p95: the 50th, 75th, and 95th percentiles, respectively.

aOccupation: the two-digit sub-major groups of the Korean Standard Classification of Occupations (KSCO) were reorganized into nine groups according to potential benzene exposure. A = homemaker/unemployed; B = agriculture/fishery; C = service/sales; D = manager/professional/clerical; E = driver/transport; F = chemical/metal/machinery manufacturing; G = food/textile/other manufacturing; H = construction/mining; I = elementary occupations (including cleaning and security).

Table 2.
Stepwise adjusted GMRs (95% CIs) for UttMACr by occupation group
Occupation No. GMR (95% CI)
M1 M2 M3 M4 M5 M6 M7
A 3,937 1 1 1 1 1 1 1
B 919 0.96 (0.89–1.04) 0.98 (0.91–1.05) 0.99 (0.91–1.08) 0.99 (0.91–1.07) 1.01 (0.93–1.10) 0.95 (0.88–1.03) 1.02 (0.94–1.11)
C 1,305 1.06 (1.00–1.12) 1.07 (1.02–1.13) 1.06 (1.00–1.12) 1.07 (1.02–1.13) 1.04 (0.98–1.10) 1.04 (0.99–1.10) 1.04 (0.99–1.10)
D 2,389 1.00 (0.96–1.04) 1.00 (0.96–1.05) 1.00 (0.95–1.04) 1.00 (0.96–1.04) 0.98 (0.94–1.03) 0.97 (0.93–1.01) 1.00 (0.95–1.04)
E 456 1.10 (1.01–1.20) 1.10 (1.02–1.19) 1.10 (1.01–1.20) 1.10 (1.02–1.19) 1.12 (1.02–1.21) 0.97 (0.89–1.05) 1.07 (0.99–1.16)
F 566 1.17 (1.09–1.26) 1.11 (1.04–1.19) 1.17 (1.09–1.26) 1.11 (1.04–1.19) 1.11 (1.03–1.20) 0.99 (0.93–1.06) 1.08 (1.01–1.16)
G 280 1.16 (1.04–1.29) 1.15 (1.04–1.27) 1.16 (1.04–1.29) 1.15 (1.04–1.28) 1.13 (1.02–1.25) 1.08 (0.98–1.19) 1.12 (1.01–1.24)
H 285 1.18 (1.06–1.32) 1.15 (1.04–1.28) 1.18 (1.06–1.32) 1.15 (1.04–1.28) 1.16 (1.05–1.29) 1.00 (0.90–1.10) 1.10 (0.99–1.22)
I 649 1.13 (1.05–1.23) 1.14 (1.06–1.23) 1.14 (1.05–1.23) 1.14 (1.06–1.23) 1.15 (1.07–1.23) 1.12 (1.04–1.20) 1.13 (1.06–1.22)
R² (ΔR² = 0.0018)a 0.0044 0.1207 0.0053 0.1217 0.1326 0.1622 0.1780

Models were CSGLM with LogUttMACr as the dependent variable; exponentiated coefficients are reported as GMRs. M1: occupation group only; M2: M1 + survey cycle (3rd/4th/5th); M3: M1 + residential region (urban/rural/coastal); M4: M1 + survey cycle + region; M5: M4 + sex, age, education, monthly household income, alcohol drinking, grilled meat/fish consumption, regular exercise, and traffic exposure; M6: M4 + smoking status, secondhand smoke exposure, and COTCr category; M7: M4 + all covariates from M5 and M6.

GMR: geometric mean ratio; CI: confidence interval; UttMACr: creatinine-adjusted urinary trans,trans-muconic acid; CSGLM: complex-samples general linear models; LogUttMACr: log-transformed creatinine-adjusted urinary ttMA; COTCr: creatinine-adjusted urinary cotinine; Occupation: A = homemaker/unemployed [Ref]; B = agriculture/fishery; C = service/sales; D = manager/professional/clerical; E = driver/transport; F = chemical/metal/machinery manufacturing; G = food/textile/other manufacturing; H = construction and mining-related trade occupations (construction/mining); I = elementary occupations (including cleaning and security).

aIncremental R² test (ΔR²): an additional reduced model with the same specification as M7 but excluding all occupation indicators (k = 8) was estimated. The incremental R² = R²(full) − R²(reduced) = 0.1780 − 0.1762 = 0.0018. F(8, 10745) = 2.976, p = 0.0025.

Sampling weights were rescaled within each cycle (mean = 1.0) to balance cycle contributions.

Table 3.
GMRs from the fully adjusted model (M7) for UttMACr by occupational group
Variable GMR (95% CI)
Full population Economically active population
Occupation
 Homemaker/unemployed 1 (Excluded)
 Agriculture/fishery 1.020 (0.940–1.106) 1
 Service/sales 1.044 (0.990–1.101) 1.019 (0.926–1.121)
 Manager/professional/clerical 0.995 (0.951–1.042) 0.969 (0.881–1.067)
 Driver/transport 1.072 (0.987–1.165) 1.046 (0.936–1.169)
 Chemical/metal/machinery manufacturing 1.079 (1.005–1.159) 1.037 (0.933–1.151)
 Food/textile/other manufacturing 1.120 (1.013–1.238) 1.083 (0.955–1.230)
 Construction/mining 1.102 (0.995–1.221) 1.072 (0.943–1.217)
 Elementary occupations (including cleaning and security) 1.134 (1.056–1.218) 1.124 (1.014–1.246)
Survey cycle
 3rd 1 1
 4th 0.605 (0.581–0.629) 0.601 (0.572–0.632)
 5th 0.490 (0.468–0.514) 0.494 (0.465–0.525)
Sex
 Male 1 1
 Female 1.243 (1.197–1.290) 1.204 (1.149–1.262)
Age (years)
 19–29 1 1
 30–39 1.149 (1.086–1.215) 1.071 (1.001–1.146)
 40–49 1.166 (1.104–1.232) 1.095 (1.025–1.169)
 50–59 1.187 (1.121–1.257) 1.072 (1.000–1.148)
 60–69 1.135 (1.064–1.211) 0.959 (0.882–1.043)
 ≥70 1.157 (1.075–1.244) 0.964 (0.867–1.071)
Education
 ≤Elementary 1 1
 Middle school 1.000 (0.945–1.059) 0.921 (0.856–0.992)
 High school 1.009 (0.958–1.062) 0.987 (0.923–1.055)
 ≥College 1.051 (0.986–1.121) 0.988 (0.909–1.073)
Monthly household income (KRW)
 <2 million 1 1
 2–3 million 1.089 (1.037–1.144) 1.071 (1.003–1.144)
 3–5 million 1.026 (0.977–1.076) 1.009 (0.947–1.075)
 ≥5 million 1.053 (1.001–1.108) 1.026 (0.960–1.097)
Region
 Urban 1 1
 Rural 0.931 (0.887–0.977) 0.938 (0.883–0.996)
 Coastal 1.001 (0.933–1.075) 1.037 (0.952–1.130)
Traffic exposure
 Low 1 1
 Medium 1.015 (0.979–1.052) 0.986 (0.942–1.032)
 High 1.006 (0.960–1.054) 0.988 (0.932–1.047)
Smoking
 Never/former 1 1
 Current smoker 1.286 (1.193–1.386) 1.296 (1.185–1.417)
SHS exposure
 None 1 1
 1–4 times/week 0.985 (0.934–1.038) 0.997 (0.939–1.058)
 ≥5 times/week 1.075 (1.004–1.151) 1.061 (0.982–1.146)
COTCr level
 ≤1.0 1 1
 1.0–p50 1.018 (0.976–1.062) 1.019 (0.966–1.076)
 p50–p75 1.058 (1.006–1.112) 1.003 (0.940–1.070)
 p75–p95 1.224 (1.133–1.322) 1.206 (1.097–1.325)
 >p95 1.701 (1.536–1.885) 1.643 (1.457–1.854)
Alcohol drinking
 Non-drinker 1 1
 Occasional (≤2/month) 1.032 (0.991–1.075) 1.018 (0.964–1.075)
 Frequent (≥1/week) 1.013 (0.971–1.057) 0.959 (0.908–1.012)
Exercise
 No 1 1
 Yes 0.984 (0.953–1.017) 0.974 (0.936–1.014)
Grilled meat consumption
 Rarely 1 1
 Monthly 0.994 (0.957–1.032) 0.961 (0.916–1.010)
 Weekly or more 1.003 (0.954–1.055) 0.975 (0.915–1.038)
Grilled fish consumption
 Rarely 1 1
 Monthly 1.011 (0.975–1.049) 1.017 (0.972–1.064)
 Weekly or more 1.009 (0.941–1.082) 1.061 (0.972–1.158)

Exp[β] from log-linear regression. Weights were rescaled to mean 1 within each survey cycle. Agriculture/fishery is the reference group in the economically active population model.

GMR: geometric mean ratio; UttMACr: creatinine-adjusted urinary trans,trans-muconic acid; CI: confidence interval; KRW: Korean Won; SHS: secondhand smoke; COTCr: creatinine-adjusted urinary cotinine; p50, p75, and p95: the 50th, 75th, and 95th percentiles, respectively.

Table 4.
Fully adjusted (M7) GMRs stratified by COTCr (μg/g creatinine) level
Occupation group No. Full population Economically active population
COTCr ≤1.0 COTCr > 1.0 COTCr ≤1.0 COTCr > 1.0
Homemaker/unemployed 3,937 1 1
Agriculture/fishery 919 1.412 (1.118–1.784) 0.945 (0.868–1.029) 1 1
Service/sales 1,305 1.073 (0.931–1.236) 1.031 (0.975–1.091) 0.802 (0.605–1.063) 1.075 (0.972–1.189)
Manager/professional/clerical 2,389 1.031 (0.921–1.153) 0.987 (0.940–1.037) 0.786 (0.590–1.048) 1.022 (0.924–1.130)
Driver/transport 456 1.019 (0.778–1.333) 1.059 (0.972–1.154) 0.795 (0.552–1.144) 1.095 (0.976–1.228)
Chemical/metal/machinery manufacturing 566 1.129 (0.925–1.378) 1.069 (0.992–1.152) 0.827 (0.602–1.137) 1.094 (0.981–1.221)
Food/textile/other manufacturing 280 1.361 (1.018–1.819) 1.072 (0.965–1.191) 0.966 (0.654–1.426) 1.117 (0.979–1.275)
Construction/mining 285 1.622 (1.216–2.164) 1.012 (0.908–1.127) 1.168 (0.804–1.696) 1.050 (0.918–1.201)
Elementary occupations (including cleaning and security) 649 1.177 (0.989–1.401) 1.123 (1.039–1.213) 0.912 (0.682–1.219) 1.185 (1.050–1.322)

Model 7 (M7, full model) adjusted for survey cycle, region, sex, age, education, monthly household income, alcohol drinking, grilled food consumption, regular exercise, traffic exposure, smoking status, secondhand smoke exposure, and urinary cotinine-to-creatinine ratio category where applicable. Full population: COTCr ≤ 1.0 μg/g creatinine, R2 = 0.1724; COTCr > 1.0 μg/g creatinine, R2 = 0.1891. Economically active population: COTCr ≤ 1.0 μg/g creatinine, R2 = 0.1718, n = 1,270; COTCr > 1.0 μg/g creatinine, R2 = 0.1992, n = 5,579.

GMR: geometric mean ratio; COTCr (μg/g creatinine): creatinine-adjusted urinary cotinine.

Table 5.
Hierarchical sequential model comparison of weighted R2 for LogUttMACr
Model Covariates added R² ΔR² % R² ΔR² formula Remarks
Sequential model comparison
 M1 Occupation only 0.0044 Unadjusted baseline
 M2 M1 + Survey cycle 0.1207 0.1163 65.3 R²(M2) − R²(M1) Dominant: secular temporal trend
 M3 M1 + Region 0.0053 0.0009 0.5 R²(M3) − R²(M1) Marginal (cycle absent)
 M4 M1 + Cycle + Region 0.1217 0.1173 65.9 R²(M4) − R²(M1) Time + space together
 M5 M4 + Demographics, lifestyle, traffic exposure 0.1326 0.0109 6.1 R²(M5) − R²(M4) Sociodemographic/behavioral (entered before smoking)
 M6 M4 + Smoking variables 0.1622 0.0405 22.7 R²(M6) − R²(M4) Tobacco exposure (unadjusted for demographics)
 M7 Full model (all covariates) 0.1780 0.0454 25.5 R²(M7) − R²(M5) Final adjusted model
 M8a M7 excluding occupation dummies 0.1762 0.0018 1.0 R²(M7) − R²(M8) Occupation unique (fully adjusted)

n = 10,786. Outcome: LogUttMACr; weighted least squares (WLS) with cycle-rescaled weights. R2 = weighted R2; ΔR2 = incremental R2 [ΔR2(Mj←Mi) = R2(Mj) − R2(Mi)]; %R2 = ΔR2/R2(M7) × 100. Covariates were added sequentially, so each ΔR2 is the variance gained given the variables already in the model; demographics/lifestyle were entered before smoking because socioeconomic factors are distal antecedents of tobacco use. M6 (M4 + smoking only) is shown for reference and lies off the main sequential path used in the variance decomposition (Table 6).

LogUttMACr: natural logarithm of creatinine-adjusted urinary trans,trans-muconic acid.

aM8 = the full model (M7) with all eight occupation indicators removed; occupation as a block was jointly significant, F(8, 10745) = 2.976, p = 0.0025.

Table 6.
Variance decomposition for LogUttMACr
Variance decomposition; sequential partition of M7 R² (= 0.1780) ΔR² % of M7 R² % ΔR² formula Remarks
Occupation 0.0044 2.5 R²(M1) M1 alone
 + Survey cycle 0.1163 65.3 R²(M2) − R²(M1) M2 vs. M1
 + Region 0.0010 0.6 R²(M4) − R²(M2) M4 vs. M2, cycle controlled
 + Demographics, lifestyle, traffic exposure 0.0109 6.1 R²(M5) − R²(M4) M5 vs. M4, before smoking
 + Smoking variables (after demo/lifestyle) 0.0454 25.5 R²(M7) − R²(M5) M7 vs. M5, after demo/lifestyle
 Total explained variance R²(M7) 0.1780 100.0
Occupation (adjusted) 0.0018 1.0 R²(M7) − R²(M8) M7 vs. M8 (unique)
Unexplained variance (1 − R²(M7)) 0.8220

The covariates in M7 explain only 17.8% of the variance in LogUttMACr (R2(M7) = 0.1780); the remaining 82.2% is attributable to factors not captured by this model. The M7 R2 (0.1780) is partitioned cumulatively to sum to 100%. The top "Occupation" row is occupation's unadjusted contribution (R2(M1) = 0.0044, 2.5%), whereas "Occupation (adjusted)" is its unique contribution after all other covariates, computed as R2(M7) − R2(M8) = 0.0018 (1.0%); the drop from 2.5% to 1.0% reflects variance occupation shares with survey cycle and smoking. M7 was adjusted for survey cycle, region, sex, age, education, monthly household income, alcohol drinking, grilled-food consumption, regular exercise, traffic exposure, smoking status, secondhand-smoke exposure, and urinary cotinine-to-creatinine ratio category. For model and variable definitions, the ΔR2 formula, and the M8 joint-significance test, see the footnote to Table 5.

LogUttMACr: natural logarithm of creatinine-adjusted urinary trans,trans-muconic acid.

  • 1. Benzene: IARC monographs on the evaluation of carcinogenic risks to humans Vol. 120. Updated 2018. https://publications.iarc.who.int/576. Accessed March 1, 2026.
  • 2. Yoon JH, Kwak WS, Ahn YS. A brief review of relationship between occupational benzene exposure and hematopoietic cancer. Ann Occup Environ Med 2018;30:33.ArticlePubMedPMCPDF
  • 3. Kang SK, Lee MY, Kim TK, Lee JO, Ahn YS. Occupational exposure to benzene in South Korea. Chem Biol Interact 2005;153-154:65–74.ArticlePubMed
  • 4. Park D, Choi S, Ha K, Jung H, Yoon C, Koh DH, et al. Estimating benzene exposure level over time and by industry type through a review of literature on Korea. Saf Health Work 2015;6(3):174–83.ArticlePubMedPMC
  • 5. Ministry of Employment and Labor of Korea. Permissible exposure limits (PELs) for chemical substances and physical agents (Labor Ministry Notification No. 2020-48, Dec 22, 1986; later revisions). https://msds.kosha.or.kr/msds/BB00200M01.do?bbsTySeCd=002&cmmnDetailCd=CA05&nttId=14#. Updated 2013. Accessed March 1, 2026.
  • 6. Ministry of Employment and Labor of Korea. Permissible exposure limits (PELs) for hazardous agents in annexed Table 11-3 under Article 81-4 of Enforcement Rule of the Occupational Safety and Health Act. https://www.law.go.kr/lsInfoP.do?lsiSeq=187393&ancYd=20161028&ancNo=00169&efYd=20161028&nwJoYnInfo=N&efGubun=Y&chrClsCd=010202&ancYnChk=0#J8953305. Updated 2016. Accessed March 1, 2026.
  • 7. Ministry of Climate Energy and Environment of Korea. Environmental standards: Annex 1, Article 2 of the Enforcement Decree of the Framework Act on Environmental Policy. https://www.law.go.kr/LSW/lsInfoP.do?lsiSeq=76291&ancYd=20061204&ancNo=19745&efYd=20070101&nwJoYnInfo=N&efGubun=Y&chrClsCd=010202&ancYnChk=0#0000. Updated 2007. Accessed March 1, 2026.
  • 8. Kim AR, Bang JH, Lee S, Sim CS, Kim Y, Lee J. Distribution of volatile organic compounds by distance from industrial complexes and potential health impact on the residents. Int J Environ Health Res 2024;34(12):4202–13.ArticlePubMed
  • 9. Lee DK, Park JM, Jang JH, Jung JS, Kim MK, Heo J, et al. Spatiotemporal assessment of benzene exposure characteristics in a petrochemical industrial area using mobile-extraction differential optical absorption spectroscopy (Me-DOAS). Toxics 2025;13(8):655.ArticlePubMedPMC
  • 10. Chung EK, Shin JA, Lee BK, Kwon J, Lee N, Chung KJ, et al. Characteristics of occupational exposure to benzene during turnaround in the petrochemical industries. Saf Health Work 2010;1(1):51–60.ArticlePubMedPMC
  • 11. Chung EK, Jang JK, Koh DH. A comparison of benzene exposures in maintenance and regular works at Korean petrochemical plants. J Chem Health Saf 2017;24(3):21–6.Article
  • 12. Choi S, Kwak S, Park D, Jeong JY. Potential risk of benzene in petroleum-derived products used from 1974 to 2012 in Korea. Aerosol Air Qual Res 2019;19(3):548–58.Article
  • 13. Kim DY, Kim HS, Lim DS, Kim KY. Benzene exposure assessment of printing workers treating petroleum-based cleaner in South Korea. Ind Health 2023;61(4):283–90.ArticlePubMedPMC
  • 14. Yu R, Weisel CP. Measurement of the urinary benzene metabolite trans,trans-muconic acid from benzene exposure in humans. J Toxicol Environ Health 1996;48(5):453–77.ArticlePubMed
  • 15. Weisel CP. Benzene exposure: an overview of monitoring methods and their findings. Chem Biol Interact 2010;184(1-2):58–66.ArticlePubMedPMC
  • 16. Boogaard PJ, van Sittert NJ. Biological monitoring of exposure to benzene: a comparison between S-phenylmercapturic acid, trans,trans-muconic acid, and phenol. Occup Environ Med 1995;52(9):611–20.ArticlePubMedPMC
  • 17. Ruppert T, Scherer G, Tricker AR, Adlkofer F. Trans,trans-muconic acid as a biomarker of non-occupational environmental exposure to benzene. Int Arch Occup Environ Health 1997;69(4):247–51.ArticlePubMedPDF
  • 18. Scherer G, Renner T, Meger M. Analysis and evaluation of trans,trans-muconic acid as a biomarker for benzene exposure. J Chromatogr B Biomed Sci Appl 1998;717(1-2):179–99.ArticlePubMed
  • 19. Rahimpoor R, Jalilian H, Mohammadi H, Rahmani A. Biological exposure indices of occupational exposure to benzene: a systematic review. Heliyon 2023;9(11):e21576.ArticlePubMedPMC
  • 20. Ministry of Climate Energy and Environment of Korea. Environmental Health Act Article 14 (Basic Investigations of National Environmental Health). https://elaw.klri.re.kr/eng_service/lawView.do?hseq=48628&lang=ENG. Updated 2018. Accessed March 1, 2026.
  • 21. Lee S, Kim JH, Choi YH, Kim S, Lee KM, Park JB. A review of the literature using the Korean National Environmental Health Survey (cycle 1-3). J Environ Health Sci 2021;47(3):227–44.
  • 22. Shin SS, Yang EH, Lee HC, Moon SH, Ryoo JH. Association of metabolites of benzene and toluene with lipid profiles in Korean adults: Korean National Environmental Health Survey (2015–2017). BMC Public Health 2022;22(1):1917.ArticlePubMedPMCPDF
  • 23. World Health Organization. Biological monitoring of chemical exposure in the workplace: guidelines. https://iris.who.int/bitstream/handle/10665/41856/WHO_HPR_OCH_96.1.pdf. Updated 1996. Accessed March 1, 2026.
  • 24. Ministry of Data and Statistics. 7th Revision and Notification of the Korean Standard Classification of Occupations (Ministry of Data and Statistics Korea Notification No. 2017-191). https://mods.go.kr/board.es?mid=a10403040000&bid=107&act=view&list_no=361380. Updated 2017. Accessed March 1, 2026.
  • 25. Siegmund B, Leyden DE, Zikulnig E, Leitner E, Murkovic M, Pfannhauser W, et al. The contribution of dietary nicotine and dietary cotinine to salivary cotinine levels as a nicotine biomarker. Food Chem 2001;74(3):259–65.Article
  • 26. Benowitz NL, Jain S, Dempsey DA, Nardone N, Helen GS, Jacob P. Urine cotinine screening detects nearly ubiquitous tobacco smoke exposure in urban adolescents. Nicotine Tob Res 2017;19(9):1048–54.ArticlePubMedPMC
  • 27. Pratt K, Hilty A, Jacob P, Schick SF. Respiratory exposure to thirdhand cigarette smoke increases concentrations of urinary metabolites of nicotine. Nicotine Tob Res 2023;25(8):1424–30.ArticlePubMedPDF
  • 28. Fustinoni S, Consonni D, Campo L, Buratti M, Colombi A, Pesatori AC, et al. Monitoring low benzene exposure: comparative evaluation of urinary biomarkers, influence of cigarette smoking, and genetic polymorphisms. Cancer Epidemiol Biomarkers Prev 2005;14(9):2237–44.ArticlePubMedPDF
  • 29. Kim S, Vermeulen R, Waidyanatha S, Johnson BA, Lan Q, Rothman N, et al. Using urinary biomarkers to elucidate dose-related patterns of human benzene metabolism. Carcinogenesis 2006;27(4):772–81.ArticlePubMed
  • 30. Carrieri M, Tranfo G, Pigini D, Paci E, Salamon F, Scapellato ML, et al. Correlation between environmental and biological monitoring of exposure to benzene in petrochemical industry operators. Toxicol Lett 2010;192(1):17–21.ArticlePubMed
  • 31. Sabarwal A, Kumar K, Singh RP. Hazardous effects of chemical pesticides on human health–cancer and other associated disorders. Environ Toxicol Pharmacol 2018;63:103–14.ArticlePubMed
  • 32. Ministry of Employment and Labor of Korea. Chapter VIII Management of Employees’ Health Occupational Safety and Health Act. Url. Updated 2025. Accessed March 1, 2026.
  • 33. Koh DH, Park JH, Lee SG, Kim HC, Jung H, Kim I, et al. Development of Korean carcinogen exposure: assessment of the exposure intensity of carcinogens by industry. Saf Health Work 2022;13(3):308–14.ArticlePubMedPMC
  • 34. Koh DH, Choi S, Park JH, Lee SG, Kim HC, Kim I, et al. Development of Korean CARcinogen EXposure: occupation-based exposure matrix. Ann Work Expo Health 2025;69(2):191–200.ArticlePubMedPDF
  • 35. Guo H, Ahn S, Zhang L. Benzene-associated immunosuppression and chronic inflammation in humans: a systematic review. Occup Environ Med 2020;78(5):377–84.Article
  • 36. Wang T, Cao Y, Xia Z, Christiani DC, Au WW. Review on novel toxicological effects and personalized health hazard in workers exposed to low doses of benzene. Arch Toxicol 2024;98(2):365–74.ArticlePubMedPDF
  • 37. Suwazono Y, Akesson A, Alfven T, Jarup L, Vahter M. Creatinine versus specific gravity-adjusted urinary cadmium concentrations. Biomarkers 2005;10(2-3):117–26.ArticlePubMed

Figure & Data

REFERENCES

    Citations

    Citations to this article as recorded by  

      • Cite
        CITE
        export Copy Download
        Close
        Download Citation
        Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.
        Format:
        • RIS — For EndNote, ProCite, RefWorks, and most other reference management software
        • BibTeX — For JabRef, BibDesk, and other BibTeX-specific software
        Include:
        • Citation for the content below
        Occupational differences in benzene-related biomarker levels beyond traditional industrial settings: findings from the Korean National Environmental Health Survey, 2015–2023
        Ann Occup Environ Med. 2026;38:e19  Published online June 15, 2026
        DOI: https://doi.org/10.35371/aoem.2026.38.e19
        Close
      • XML DownloadXML Download
      Related articles
      Occupational differences in benzene-related biomarker levels beyond traditional industrial settings: findings from the Korean National Environmental Health Survey, 2015–2023
      Occupational differences in benzene-related biomarker levels beyond traditional industrial settings: findings from the Korean National Environmental Health Survey, 2015–2023
      Characteristic Occupationa Total
      A B C D E F G H I
      No. 3,937 919 1,305 2,389 456 566 280 285 649 10,786
      UttMACr AM (SD) 108.6 (240.5) 97.3 (141.5) 112.5 (161.5) 105.4 (146.2) 108.3 (114.8) 110.0 (104.5) 119.7 (146.3) 114.1 (120.1) 117.4 (146.8) 108.6 (181.6)
      UttMACr GM (GSD) 67.2 (2.45) 64.5 (2.29) 71.1 (2.44) 67.1 (2.47) 73.8 (2.39) 78.5 (2.27) 77.9 (2.42) 79.6 (2.28) 76.3 (2.44) 69.5 (2.43)
      Survey cycle
       3rd 104.3 (30.6) 93.7 (24.4) 115.9 (27.1) 105.4 (29.1) 116.0 (30.8) 106.6 (36.2) 109.0 (31.2) 134.4 (34.6) 125.6 (29.2) 108.0 (29.8)
       4th 64.7 (32.0) 62.9 (44.3) 61.1 (37.3) 63.9 (34.0) 64.8 (30.2) 71.5 (35.6) 74.3 (33.1) 63.9 (30.3) 75.9 (33.7) 65.1 (34.0)
       5th 48.4 (37.4) 50.1 (31.4) 57.5 (35.6) 49.2 (37.0) 57.1 (39.0) 59.6 (28.2) 60.8 (35.7) 57.4 (35.1) 51.7 (37.1) 51.3 (36.1)
      Sex
       Male 59.5 (30.2) 63.5 (65.8) 70.1 (40.3) 65.8 (60.0) 73.6 (97.5) 78.4 (92.5) 73.5 (57.0) 78.8 (95.4) 74.7 (39.3) 67.9 (51.3)
       Female 70.8 (69.8) 66.6 (34.2) 71.8 (59.7) 69.0 (40.0) 78.6 (2.5) 79.6 (7.5) 84.2 (43.0) 96.6 (4.6) 77.3 (60.7) 71.3 (48.7)
      Age (years)
       19–29 51.1 (12.1) 102.8 (1.6) 64.3 (16.1) 61.9 (18.1) 66.0 (7.3) 73.8 (15.1) 78.6 (12.5) 62.1 (9.9) 142.6 (5.2) 61.1 (13.3)
       30–39 75.3 (13.5) 82.1 (4.8) 84.6 (16.8) 65.5 (27.9) 93.2 (11.1) 95.5 (21.1) 88.9 (17.2) 72.5 (13.3) 100.1 (6.2) 74.2 (17.6)
       40–49 70.9 (13.0) 64.8 (7.4) 77.9 (19.7) 73.6 (30.6) 93.9 (20.3) 70.5 (28.1) 75.3 (19.2) 110.3 (18.1) 104.1 (9.3) 75.2 (19.8)
       50–59 75.3 (15.9) 75.2 (24.6) 67.3 (28.6) 66.4 (17.5) 83.0 (33.0) 91.2 (23.6) 77.3 (32.7) 90.7 (33.9) 88.3 (23.3) 74.7 (20.9)
       60–69 68.2 (21.1) 59.8 (29.1) 65.5 (13.6) 63.0 (4.8) 52.2 (21.4) 54.4 (9.1) 65.7 (14.8) 63.5 (20.3) 62.8 (26.6) 64.3 (15.4)
       ≥70 64.4 (24.3) 57.9 (32.4) 65.9 (5.1) 54.2 (1.2) 46.8 (6.9) 69.2 (3.0) 108.1 (3.5) 49.8 (4.5) 61.9 (29.3) 62.8 (13.1)
      Education
       ≤Elementary 56.5 (36.1) 56.8 (48.8) 57.2 (26.5) 54.9 (9.1) 65.0 (33.2) 65.2 (19.3) 72.6 (31.2) 56.9 (35.3) 63.8 (47.9) 58.3 (27.3)
       Middle school 73.3 (12.5) 57.8 (18.5) 76.5 (12.2) 42.8 (6.4) 56.9 (14.4) 68.2 (10.1) 69.3 (16.9) 104.4 (15.2) 88.0 (13.5) 66.7 (11.2)
       High school 67.3 (26.7) 82.5 (23.6) 73.2 (30.9) 62.5 (27.8) 86.3 (39.8) 87.5 (39.4) 87.2 (35.6) 94.8 (33.4) 82.1 (28.6) 71.9 (29.2)
       ≥College 82.7 (24.7) 84.4 (9.1) 81.0 (30.4) 75.4 (56.7) 84.5 (12.6) 80.4 (31.3) 78.9 (16.3) 89.5 (16.1) 119.9 (10.0) 79.3 (32.2)
      Monthly household income (KRW)
       <2 million 64.4 (39.4) 63.8 (56.8) 68.1 (21.2) 64.3 (6.4) 85.7 (22.5) 98.8 (9.4) 80.5 (30.0) 109.3 (25.2) 69.5 (47.8) 67.9 (26.1)
       2–3 million 73.3 (19.3) 59.9 (19.9) 82.2 (20.9) 74.0 (14.8) 70.9 (27.0) 86.8 (29.4) 85.3 (22.8) 83.3 (24.6) 79.1 (19.3) 75.9 (19.5)
       3–5 million 66.2 (22.9) 61.8 (14.4) 67.0 (28.2) 70.2 (34.1) 71.8 (31.5) 68.5 (37.9) 72.7 (32.6) 61.7 (30.9) 94.6 (24.1) 69.2 (28.1)
       ≥5 million 68.3 (18.4) 87.4 (8.9) 70.2 (29.6) 63.0 (44.7) 68.3 (19.1) 78.8 (23.3) 74.2 (14.6) 74.5 (19.3) 64.7 (8.8) 66.9 (26.3)
      Region
       Coastal 63.7 (4.6) 59.3 (36.6) 67.2 (6.3) 62.5 (3.2) 85.8 (6.6) 112.8 (4.6) 100.1 (6.2) 51.1 (2.9) 146.5 (6.6) 69.2 (6.2)
       Rural 59.6 (11.4) 65.7 (46.5) 58.6 (11.2) 66.4 (8.6) 73.8 (14.4) 69.7 (13.5) 66.9 (12.8) 95.2 (11.5) 73.5 (14.1) 64.8 (12.8)
       Urban 68.5 (84.0) 73.6 (17.0) 73.3 (82.4) 67.3 (88.2) 72.8 (79.0) 78.4 (81.9) 78.3 (81.0) 78.9 (85.5) 72.7 (79.3) 70.3 (81.0)
      Traffic exposure
       Low 65.2 (30.9) 64.0 (56.1) 72.6 (28.1) 65.4 (23.1) 76.0 (28.8) 88.5 (28.8) 83.2 (31.0) 85.5 (27.9) 81.2 (32.5) 69.5 (29.4)
       Medium 68.3 (50.5) 65.4 (41.3) 72.5 (50.2) 67.4 (55.5) 74.0 (49.2) 77.7 (49.9) 71.9 (52.4) 79.9 (52.5) 75.8 (53.3) 69.9 (51.6)
       High 67.7 (18.6) 60.5 (2.6) 66.1 (21.7) 68.2 (21.4) 70.3 (21.9) 68.4 (21.3) 89.0 (16.6) 71.1 (19.6) 67.6 (14.3) 68.3 (19.0)
      Smoking
       Never/former 65.4 (88.6) 58.3 (78.8) 67.3 (81.7) 61.8 (79.4) 56.7 (58.2) 64.1 (58.7) 71.7 (73.9) 60.5 (57.5) 68.8 (84.2) 64.1 (80.0)
       Current 83.4 (11.4) 94.3 (21.2) 90.7 (18.3) 91.6 (20.6) 106.5 (41.8) 104.9 (41.3) 98.7 (26.1) 115.2 (42.5) 131.7 (15.8) 95.8 (20.0)
      SHS exposure
       None 67.0 (92.1) 63.2 (86.7) 70.0 (84.2) 64.6 (80.5) 69.1 (74.9) 75.2 (67.6) 81.9 (79.7) 75.5 (68.7) 74.0 (88.6) 67.9 (84.2)
       1–4 times/week 62.9 (5.1) 72.2 (9.8) 75.4 (9.6) 70.3 (13.9) 73.7 (15.6) 94.4 (18.4) 55.2 (11.6) 65.4 (11.4) 84.2 (6.9) 72.2 (10.1)
       ≥5 times/week 83.3 (2.8) 78.4 (3.5) 80.2 (6.2) 101.7 (5.6) 123.5 (9.4) 75.8 (14.0) 78.4 (8.7) 106.8 (19.8) 120.5 (4.4) 91.7 (5.7)
      COTCr level
       ≤1.0 63.7 (21.7) 67.5 (15.1) 71.5 (17.4) 62.2 (22.0) 60.7 (11.4) 64.4 (15.4) 76.4 (15.3) 89.4 (16.1) 68.0 (21.9) 65.1 (19.8)
       1.0–p50 61.6 (41.0) 58.5 (39.9) 63.2 (40.0) 61.1 (38.9) 53.3 (32.8) 65.7 (33.2) 73.9 (37.4) 53.4 (29.8) 65.7 (37.5) 61.7 (38.8)
       p50–p75 70.9 (22.4) 47.8 (20.6) 70.8 (20.9) 61.4 (15.9) 66.9 (13.9) 61.4 (8.9) 59.2 (20.0) 54.2 (10.9) 70.1 (20.8) 65.9 (18.6)
       p75–p95 79.3 (12.6) 80.0 (16.7) 84.0 (18.6) 84.1 (19.1) 91.1 (30.3) 93.7 (35.2) 96.0 (19.9) 96.1 (29.8) 120.5 (15.3) 86.6 (18.4)
       >p95 125.9 (2.3) 137.0 (7.7) 120.3 (3.1) 118.6 (4.1) 145.3 (11.5) 153.9 (7.3) 127.8 (7.3) 151.9 (13.3) 143.7 (4.5) 132.2 (4.4)
      Alcohol drinking
       Non-drinker 65.7 (38.8) 63.1 (38.3) 69.6 (24.7) 64.8 (15.9) 57.7 (16.6) 72.1 (14.0) 83.2 (24.6) 92.8 (19.3) 73.5 (37.0) 67.1 (27.2)
       Occasional 64.3 (36.7) 65.0 (24.5) 71.8 (34.7) 67.5 (37.7) 81.3 (25.0) 72.5 (32.8) 80.2 (33.1) 66.1 (23.3) 77.2 (35.7) 68.2 (34.9)
       Frequent 74.4 (24.5) 65.7 (37.1) 71.4 (40.6) 67.5 (46.3) 75.9 (58.3) 84.4 (53.2) 73.4 (42.3) 81.5 (57.4) 78.9 (27.3) 72.5 (38.0)
      Regular exercise
       No 68.1 (56.7) 66.6 (75.7) 74.2 (67.8) 67.4 (61.3) 78.0 (54.7) 78.9 (67.1) 93.4 (67.0) 84.8 (68.8) 77.9 (60.1) 71.2 (61.6)
       Yes 66.0 (43.3) 58.5 (24.3) 65.1 (32.2) 66.6 (38.7) 69.0 (45.3) 77.8 (32.9) 53.9 (33.0) 69.2 (31.2) 73.8 (39.9) 66.8 (38.4)
      Grilled meat consumption
       Never/rarely 64.1 (37.6) 59.1 (40.5) 66.7 (27.9) 64.6 (18.1) 64.9 (23.9) 77.0 (21.6) 69.6 (34.5) 88.1 (25.0) 74.0 (45.8) 66.1 (29.5)
       Monthly 70.0 (49.3) 67.9 (50.0) 71.6 (53.5) 68.1 (59.2) 76.6 (59.0) 73.4 (58.7) 81.8 (53.3) 75.9 (54.8) 71.9 (41.4) 70.6 (53.4)
       Weekly+ 65.9 (13.2) 71.1 (9.5) 76.5 (18.6) 66.4 (22.7) 77.5 (17.1) 98.0 (19.7) 86.7 (12.2) 79.6 (20.2) 102.9 (12.8) 72.1 (17.0)
      Grilled fish Consumption
       Never/rarely 65.4 (71.4) 64.3 (70.9) 68.9 (67.5) 65.1 (65.2) 69.5 (70.9) 79.1 (71.6) 73.9 (72.0) 82.7 (70.9) 72.2 (71.0) 67.7 (69.2)
       Monthly 72.5 (23.1) 65.9 (23.1) 73.9 (26.3) 71.0 (30.5) 85.4 (22.1) 70.1 (23.9) 91.5 (22.8) 69.8 (21.8) 70.5 (20.3) 72.4 (25.4)
       Weekly+ 68.7 (5.5) 61.6 (6.0) 85.4 (6.2) 70.1 (4.3) 84.8 (7.0) 128.4 (4.4) 81.5 (5.2) 80.9 (7.3) 144.7 (8.6) 79.5 (5.5)
      Occupation No. GMR (95% CI)
      M1 M2 M3 M4 M5 M6 M7
      A 3,937 1 1 1 1 1 1 1
      B 919 0.96 (0.89–1.04) 0.98 (0.91–1.05) 0.99 (0.91–1.08) 0.99 (0.91–1.07) 1.01 (0.93–1.10) 0.95 (0.88–1.03) 1.02 (0.94–1.11)
      C 1,305 1.06 (1.00–1.12) 1.07 (1.02–1.13) 1.06 (1.00–1.12) 1.07 (1.02–1.13) 1.04 (0.98–1.10) 1.04 (0.99–1.10) 1.04 (0.99–1.10)
      D 2,389 1.00 (0.96–1.04) 1.00 (0.96–1.05) 1.00 (0.95–1.04) 1.00 (0.96–1.04) 0.98 (0.94–1.03) 0.97 (0.93–1.01) 1.00 (0.95–1.04)
      E 456 1.10 (1.01–1.20) 1.10 (1.02–1.19) 1.10 (1.01–1.20) 1.10 (1.02–1.19) 1.12 (1.02–1.21) 0.97 (0.89–1.05) 1.07 (0.99–1.16)
      F 566 1.17 (1.09–1.26) 1.11 (1.04–1.19) 1.17 (1.09–1.26) 1.11 (1.04–1.19) 1.11 (1.03–1.20) 0.99 (0.93–1.06) 1.08 (1.01–1.16)
      G 280 1.16 (1.04–1.29) 1.15 (1.04–1.27) 1.16 (1.04–1.29) 1.15 (1.04–1.28) 1.13 (1.02–1.25) 1.08 (0.98–1.19) 1.12 (1.01–1.24)
      H 285 1.18 (1.06–1.32) 1.15 (1.04–1.28) 1.18 (1.06–1.32) 1.15 (1.04–1.28) 1.16 (1.05–1.29) 1.00 (0.90–1.10) 1.10 (0.99–1.22)
      I 649 1.13 (1.05–1.23) 1.14 (1.06–1.23) 1.14 (1.05–1.23) 1.14 (1.06–1.23) 1.15 (1.07–1.23) 1.12 (1.04–1.20) 1.13 (1.06–1.22)
      R² (ΔR² = 0.0018)a 0.0044 0.1207 0.0053 0.1217 0.1326 0.1622 0.1780
      Variable GMR (95% CI)
      Full population Economically active population
      Occupation
       Homemaker/unemployed 1 (Excluded)
       Agriculture/fishery 1.020 (0.940–1.106) 1
       Service/sales 1.044 (0.990–1.101) 1.019 (0.926–1.121)
       Manager/professional/clerical 0.995 (0.951–1.042) 0.969 (0.881–1.067)
       Driver/transport 1.072 (0.987–1.165) 1.046 (0.936–1.169)
       Chemical/metal/machinery manufacturing 1.079 (1.005–1.159) 1.037 (0.933–1.151)
       Food/textile/other manufacturing 1.120 (1.013–1.238) 1.083 (0.955–1.230)
       Construction/mining 1.102 (0.995–1.221) 1.072 (0.943–1.217)
       Elementary occupations (including cleaning and security) 1.134 (1.056–1.218) 1.124 (1.014–1.246)
      Survey cycle
       3rd 1 1
       4th 0.605 (0.581–0.629) 0.601 (0.572–0.632)
       5th 0.490 (0.468–0.514) 0.494 (0.465–0.525)
      Sex
       Male 1 1
       Female 1.243 (1.197–1.290) 1.204 (1.149–1.262)
      Age (years)
       19–29 1 1
       30–39 1.149 (1.086–1.215) 1.071 (1.001–1.146)
       40–49 1.166 (1.104–1.232) 1.095 (1.025–1.169)
       50–59 1.187 (1.121–1.257) 1.072 (1.000–1.148)
       60–69 1.135 (1.064–1.211) 0.959 (0.882–1.043)
       ≥70 1.157 (1.075–1.244) 0.964 (0.867–1.071)
      Education
       ≤Elementary 1 1
       Middle school 1.000 (0.945–1.059) 0.921 (0.856–0.992)
       High school 1.009 (0.958–1.062) 0.987 (0.923–1.055)
       ≥College 1.051 (0.986–1.121) 0.988 (0.909–1.073)
      Monthly household income (KRW)
       <2 million 1 1
       2–3 million 1.089 (1.037–1.144) 1.071 (1.003–1.144)
       3–5 million 1.026 (0.977–1.076) 1.009 (0.947–1.075)
       ≥5 million 1.053 (1.001–1.108) 1.026 (0.960–1.097)
      Region
       Urban 1 1
       Rural 0.931 (0.887–0.977) 0.938 (0.883–0.996)
       Coastal 1.001 (0.933–1.075) 1.037 (0.952–1.130)
      Traffic exposure
       Low 1 1
       Medium 1.015 (0.979–1.052) 0.986 (0.942–1.032)
       High 1.006 (0.960–1.054) 0.988 (0.932–1.047)
      Smoking
       Never/former 1 1
       Current smoker 1.286 (1.193–1.386) 1.296 (1.185–1.417)
      SHS exposure
       None 1 1
       1–4 times/week 0.985 (0.934–1.038) 0.997 (0.939–1.058)
       ≥5 times/week 1.075 (1.004–1.151) 1.061 (0.982–1.146)
      COTCr level
       ≤1.0 1 1
       1.0–p50 1.018 (0.976–1.062) 1.019 (0.966–1.076)
       p50–p75 1.058 (1.006–1.112) 1.003 (0.940–1.070)
       p75–p95 1.224 (1.133–1.322) 1.206 (1.097–1.325)
       >p95 1.701 (1.536–1.885) 1.643 (1.457–1.854)
      Alcohol drinking
       Non-drinker 1 1
       Occasional (≤2/month) 1.032 (0.991–1.075) 1.018 (0.964–1.075)
       Frequent (≥1/week) 1.013 (0.971–1.057) 0.959 (0.908–1.012)
      Exercise
       No 1 1
       Yes 0.984 (0.953–1.017) 0.974 (0.936–1.014)
      Grilled meat consumption
       Rarely 1 1
       Monthly 0.994 (0.957–1.032) 0.961 (0.916–1.010)
       Weekly or more 1.003 (0.954–1.055) 0.975 (0.915–1.038)
      Grilled fish consumption
       Rarely 1 1
       Monthly 1.011 (0.975–1.049) 1.017 (0.972–1.064)
       Weekly or more 1.009 (0.941–1.082) 1.061 (0.972–1.158)
      Occupation group No. Full population Economically active population
      COTCr ≤1.0 COTCr > 1.0 COTCr ≤1.0 COTCr > 1.0
      Homemaker/unemployed 3,937 1 1
      Agriculture/fishery 919 1.412 (1.118–1.784) 0.945 (0.868–1.029) 1 1
      Service/sales 1,305 1.073 (0.931–1.236) 1.031 (0.975–1.091) 0.802 (0.605–1.063) 1.075 (0.972–1.189)
      Manager/professional/clerical 2,389 1.031 (0.921–1.153) 0.987 (0.940–1.037) 0.786 (0.590–1.048) 1.022 (0.924–1.130)
      Driver/transport 456 1.019 (0.778–1.333) 1.059 (0.972–1.154) 0.795 (0.552–1.144) 1.095 (0.976–1.228)
      Chemical/metal/machinery manufacturing 566 1.129 (0.925–1.378) 1.069 (0.992–1.152) 0.827 (0.602–1.137) 1.094 (0.981–1.221)
      Food/textile/other manufacturing 280 1.361 (1.018–1.819) 1.072 (0.965–1.191) 0.966 (0.654–1.426) 1.117 (0.979–1.275)
      Construction/mining 285 1.622 (1.216–2.164) 1.012 (0.908–1.127) 1.168 (0.804–1.696) 1.050 (0.918–1.201)
      Elementary occupations (including cleaning and security) 649 1.177 (0.989–1.401) 1.123 (1.039–1.213) 0.912 (0.682–1.219) 1.185 (1.050–1.322)
      Model Covariates added R² ΔR² % R² ΔR² formula Remarks
      Sequential model comparison
       M1 Occupation only 0.0044 Unadjusted baseline
       M2 M1 + Survey cycle 0.1207 0.1163 65.3 R²(M2) − R²(M1) Dominant: secular temporal trend
       M3 M1 + Region 0.0053 0.0009 0.5 R²(M3) − R²(M1) Marginal (cycle absent)
       M4 M1 + Cycle + Region 0.1217 0.1173 65.9 R²(M4) − R²(M1) Time + space together
       M5 M4 + Demographics, lifestyle, traffic exposure 0.1326 0.0109 6.1 R²(M5) − R²(M4) Sociodemographic/behavioral (entered before smoking)
       M6 M4 + Smoking variables 0.1622 0.0405 22.7 R²(M6) − R²(M4) Tobacco exposure (unadjusted for demographics)
       M7 Full model (all covariates) 0.1780 0.0454 25.5 R²(M7) − R²(M5) Final adjusted model
       M8a M7 excluding occupation dummies 0.1762 0.0018 1.0 R²(M7) − R²(M8) Occupation unique (fully adjusted)
      Variance decomposition; sequential partition of M7 R² (= 0.1780) ΔR² % of M7 R² % ΔR² formula Remarks
      Occupation 0.0044 2.5 R²(M1) M1 alone
       + Survey cycle 0.1163 65.3 R²(M2) − R²(M1) M2 vs. M1
       + Region 0.0010 0.6 R²(M4) − R²(M2) M4 vs. M2, cycle controlled
       + Demographics, lifestyle, traffic exposure 0.0109 6.1 R²(M5) − R²(M4) M5 vs. M4, before smoking
       + Smoking variables (after demo/lifestyle) 0.0454 25.5 R²(M7) − R²(M5) M7 vs. M5, after demo/lifestyle
       Total explained variance R²(M7) 0.1780 100.0
      Occupation (adjusted) 0.0018 1.0 R²(M7) − R²(M8) M7 vs. M8 (unique)
      Unexplained variance (1 − R²(M7)) 0.8220
      Table 1. Characteristics of study participants by occupation group: UttMACr GM and weighted percentages

      UttMACr: creatinine-adjusted urinary trans,trans-muconic acid; GM: geometric mean; ttMACr: creatinine-adjusted trans,trans-muconic acid; AM: arithmetic mean; SD: standard deviation; GSD: geometric standard deviation; KRW: Korean Won; SHS: secondhand smoke; COTCr: creatinine-adjusted urinary cotinine; p50, p75, and p95: the 50th, 75th, and 95th percentiles, respectively.

      Occupation: the two-digit sub-major groups of the Korean Standard Classification of Occupations (KSCO) were reorganized into nine groups according to potential benzene exposure. A = homemaker/unemployed; B = agriculture/fishery; C = service/sales; D = manager/professional/clerical; E = driver/transport; F = chemical/metal/machinery manufacturing; G = food/textile/other manufacturing; H = construction/mining; I = elementary occupations (including cleaning and security).

      Table 2. Stepwise adjusted GMRs (95% CIs) for UttMACr by occupation group

      Models were CSGLM with LogUttMACr as the dependent variable; exponentiated coefficients are reported as GMRs. M1: occupation group only; M2: M1 + survey cycle (3rd/4th/5th); M3: M1 + residential region (urban/rural/coastal); M4: M1 + survey cycle + region; M5: M4 + sex, age, education, monthly household income, alcohol drinking, grilled meat/fish consumption, regular exercise, and traffic exposure; M6: M4 + smoking status, secondhand smoke exposure, and COTCr category; M7: M4 + all covariates from M5 and M6.

      GMR: geometric mean ratio; CI: confidence interval; UttMACr: creatinine-adjusted urinary trans,trans-muconic acid; CSGLM: complex-samples general linear models; LogUttMACr: log-transformed creatinine-adjusted urinary ttMA; COTCr: creatinine-adjusted urinary cotinine; Occupation: A = homemaker/unemployed [Ref]; B = agriculture/fishery; C = service/sales; D = manager/professional/clerical; E = driver/transport; F = chemical/metal/machinery manufacturing; G = food/textile/other manufacturing; H = construction and mining-related trade occupations (construction/mining); I = elementary occupations (including cleaning and security).

      Incremental R² test (ΔR²): an additional reduced model with the same specification as M7 but excluding all occupation indicators (k = 8) was estimated. The incremental R² = R²(full) − R²(reduced) = 0.1780 − 0.1762 = 0.0018. F(8, 10745) = 2.976, p = 0.0025.

      Sampling weights were rescaled within each cycle (mean = 1.0) to balance cycle contributions.

      Table 3. GMRs from the fully adjusted model (M7) for UttMACr by occupational group

      Exp[β] from log-linear regression. Weights were rescaled to mean 1 within each survey cycle. Agriculture/fishery is the reference group in the economically active population model.

      GMR: geometric mean ratio; UttMACr: creatinine-adjusted urinary trans,trans-muconic acid; CI: confidence interval; KRW: Korean Won; SHS: secondhand smoke; COTCr: creatinine-adjusted urinary cotinine; p50, p75, and p95: the 50th, 75th, and 95th percentiles, respectively.

      Table 4. Fully adjusted (M7) GMRs stratified by COTCr (μg/g creatinine) level

      Model 7 (M7, full model) adjusted for survey cycle, region, sex, age, education, monthly household income, alcohol drinking, grilled food consumption, regular exercise, traffic exposure, smoking status, secondhand smoke exposure, and urinary cotinine-to-creatinine ratio category where applicable. Full population: COTCr ≤ 1.0 μg/g creatinine, R2 = 0.1724; COTCr > 1.0 μg/g creatinine, R2 = 0.1891. Economically active population: COTCr ≤ 1.0 μg/g creatinine, R2 = 0.1718, n = 1,270; COTCr > 1.0 μg/g creatinine, R2 = 0.1992, n = 5,579.

      GMR: geometric mean ratio; COTCr (μg/g creatinine): creatinine-adjusted urinary cotinine.

      Table 5. Hierarchical sequential model comparison of weighted R2 for LogUttMACr

      n = 10,786. Outcome: LogUttMACr; weighted least squares (WLS) with cycle-rescaled weights. R2 = weighted R2; ΔR2 = incremental R2 [ΔR2(Mj←Mi) = R2(Mj) − R2(Mi)]; %R2 = ΔR2/R2(M7) × 100. Covariates were added sequentially, so each ΔR2 is the variance gained given the variables already in the model; demographics/lifestyle were entered before smoking because socioeconomic factors are distal antecedents of tobacco use. M6 (M4 + smoking only) is shown for reference and lies off the main sequential path used in the variance decomposition (Table 6).

      LogUttMACr: natural logarithm of creatinine-adjusted urinary trans,trans-muconic acid.

      M8 = the full model (M7) with all eight occupation indicators removed; occupation as a block was jointly significant, F(8, 10745) = 2.976, p = 0.0025.

      Table 6. Variance decomposition for LogUttMACr

      The covariates in M7 explain only 17.8% of the variance in LogUttMACr (R2(M7) = 0.1780); the remaining 82.2% is attributable to factors not captured by this model. The M7 R2 (0.1780) is partitioned cumulatively to sum to 100%. The top "Occupation" row is occupation's unadjusted contribution (R2(M1) = 0.0044, 2.5%), whereas "Occupation (adjusted)" is its unique contribution after all other covariates, computed as R2(M7) − R2(M8) = 0.0018 (1.0%); the drop from 2.5% to 1.0% reflects variance occupation shares with survey cycle and smoking. M7 was adjusted for survey cycle, region, sex, age, education, monthly household income, alcohol drinking, grilled-food consumption, regular exercise, traffic exposure, smoking status, secondhand-smoke exposure, and urinary cotinine-to-creatinine ratio category. For model and variable definitions, the ΔR2 formula, and the M8 joint-significance test, see the footnote to Table 5.

      LogUttMACr: natural logarithm of creatinine-adjusted urinary trans,trans-muconic acid.

      Table 1.

      Table 2.

      Table 3.

      Table 4.

      Table 5.

      Table 6.

      Ann Occup Environ Med : Annals of Occupational and Environmental Medicine
      © Korean Society of Occupational & Environmental Medicine.
      Close layer
      TOP