Petroleum Attendants and Mechanics Show Highest Benzene Levels in Nigerian Biomonitoring Study

A cross-sectional human biomonitoring study of 675 people in Abuja, Nigeria, has found that serum benzene concentrations differ markedly across petroleum-related occupational groups, with fuel station attendants and motor mechanics showing the highest adjusted levels compared with controls. Benzene is an established human carcinogen and a recognised hazard in petroleum-related work, but environmental air measurements alone can miss how much of the chemical actually enters the body. The study set out to complement environmental monitoring by measuring serum benzene directly, comparing five groups of 135 participants each: petroleum attendants, motor mechanics, truck drivers, staff of the Nigerian National Petroleum Company Limited (NNPCL), and controls with no petroleum-related occupation.

Participants provided serum samples and completed a structured questionnaire covering demographic, occupational, and behavioural factors, including age, years of occupational exposure, smoking status, and use of personal protective equipment. The researchers used non-parametric tests to compare groups, and multivariable linear regression with HC3 heteroskedasticity-robust standard errors to estimate associations with log-transformed serum benzene concentrations. The differences between groups were statistically robust: the overall comparison across occupational categories yielded a Kruskal-Wallis statistic of 256.88 (p < .001).

After adjustment for age, years of exposure, smoking, and protective equipment use, petroleum attendants had significantly higher log-transformed serum benzene concentrations than controls (B = 0.651, 95% CI 0.555–0.748), followed by motor mechanics (B = 0.236, 95% CI 0.113–0.359). NNPCL staff, by contrast, had lower concentrations than controls (B = −0.269, 95% CI −0.377 to −0.161), while truck drivers did not differ significantly from the control group. A secondary analysis classified participants above the study-specific upper quartile of serum benzene.

Because complete separation prevented the use of conventional logistic regression, the team applied Firth bias-reduced logistic regression. This analysis found that petroleum attendants had roughly 12.8 times the adjusted odds of falling in the highest exposure category (OR = 12.77, 95% CI 6.52–25.01), and motor mechanics about 3.6 times the odds (OR = 3.64, 95% CI 1.83–7.24) relative to controls. Beyond the group comparisons, the researchers explored the covariance structure of the serum measurements using principal component analysis, which identified four components explaining 66.12% of the variance.

They also trained machine learning models to predict exposure patterns from the measured variables. A Random Forest classifier achieved a cross-validated area under the curve of 0.837, and an XGBoost model reached 0.819, indicating that the measured variables carried substantial predictive information about occupational group membership and exposure status. Applying SHAP (SHapley Additive exPlanations) values to the XGBoost model, the team found that petroleum-attendant status was the largest mean absolute contributor to the model's predictions.

This machine learning result aligned with the regression findings, reinforcing that working at fuel dispensing points was the strongest single signal in the dataset. The authors emphasise that these predictive analyses were complementary tools rather than causal evidence. The study's design imposes important limits on interpretation.

As a cross-sectional snapshot conducted in a single city, it cannot establish temporal patterns, identify specific exposure sources, or demonstrate health outcomes. Serum benzene reflects exposure at a point in time, and the authors note that longitudinal studies incorporating repeated biomonitoring alongside workplace air measurements are needed to characterise exposure sources, temporal variation, and potential health implications more precisely. Even so, the findings carry practical weight for occupational health surveillance in Nigeria and comparable settings.

The authors argue that human biomonitoring should complement, not replace, environmental monitoring, and that combining both with task-based surveillance would strengthen occupational exposure assessment where regulatory data remain limited. For public health authorities, the study provides a methodological template: a modestly sized, carefully stratified sample, robust statistical handling of skewed exposure data, and transparent reporting of uncertainty.

Subject of Research: Human biomonitoring of occupational benzene exposure among petroleum-related workers in Abuja, Nigeria

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

Article References: Adejumo, B. I. G., Badejo, D. A., Oyakhire, F. O., Obasuyi, G. E., Usman – Onoruvie, V. O., Fajemidagba, G. A., Adegboye, A. A., Afeikhena, O. E., Dele – Ochie, P. E., Adebowale, M. O., Abolarin, O. D., Igbinovia, E. O., Ogunmola, E. O., Drisu, U. I., Udemezue, N. N., & Efenarhua, S. (2026). Human biomonitoring of occupational benzene exposure among petroleum-exposed workers in nigeria and its implications for public health surveillance. BMC Public Health. https://doi.org/10.1186/s12889-026-29238-4

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29238-4

Keywords: benzene, occupational exposure, human biomonitoring, Nigeria, petroleum workers, serum benzene, public health surveillance, cross-sectional study, BMC Public Health, fuel attendants, motor mechanics, machine learning