Human biomonitoring of occupational benzene exposure among petroleum-exposed workers in nigeria and its implications for public health surveillance

Benzene is an established human carcinogen and an important occupational hazard in petroleum-related work. Human biomonitoring can complement environmental monitoring by characterising internal exposure, yet biomonitoring evidence remains limited in many Sub-Saharan African occupational settings. This study compared serum benzene concentrations and other measured serum compounds across petroleum-related occupational groups in Abuja, Nigeria, and examined factors associated with serum benzene concentration. A cross-sectional human biomonitoring study included 675 participants, comprising 135 petroleum attendants, 135 motor mechanics, 135 truck drivers, 135 Nigerian National Petroleum Company Limited (NNPCL) staff, and 135 controls. Serum benzene and other compounds were measured, and demographic, occupational, and behavioural information was obtained using a structured questionnaire. Occupational-group differences were assessed using non-parametric comparisons. Associations with ln(benzene + 1) were estimated using multivariable linear regression with HC3 heteroskedasticity-robust standard errors. Firth bias-reduced logistic regression was used for a secondary study-specific upper-quartile classification because complete separation affected conventional logistic regression. PCA examined covariance among serum measurements, while Random Forest, XGBoost, and SHAP were used as complementary predictive analyses. Serum benzene concentrations differed across occupational groups (H = 256.88, p < .001). After adjustment for age, years of occupational exposure, smoking status, and personal protective equipment use, petroleum attendants (B = 0.651, 95% CI 0.555–0.748; p < .001) and motor mechanics (B = 0.236, 95% CI 0.113–0.359; p < .001) had higher log-transformed serum benzene concentrations than controls, whereas NNPCL staff had lower concentrations (B = − 0.269, 95% CI − 0.377 to − 0.161; p < .001). Truck drivers did not differ significantly from controls. In the secondary Firth analysis, petroleum attendants (adjusted OR = 12.77, 95% CI 6.52–25.01) and motor mechanics (adjusted OR = 3.64, 95% CI 1.83–7.24) had higher odds of being above the study-specific upper-quartile threshold. Four PCA components explained 66.12% of variance. Random Forest and XGBoost achieved cross-validated AUCs of 0.837 and 0.819, respectively; SHAP identified petroleum-attendant status as the largest mean absolute contributor to XGBoost predictions. Serum benzene concentrations varied across petroleum-related occupational groups in Abuja, with higher adjusted concentrations among petroleum attendants and motor mechanics than controls. These cross-sectional findings support strengthening occupational exposure assessment through complementary human biomonitoring, environmental monitoring, and task-based surveillance. Longitudinal studies incorporating repeated biomonitoring and workplace air measurements are needed to characterise exposure sources, temporal patterns, and potential health implications more precisely.