Table 6. Mediation Analysis^a,b.

oxidative stress marker	POPs	indirect effect HR (95% CI)c	direct effect HR (95% CI)c	total effect HR (95% CI)c	estimated percent mediated (%)d
SOD	HCB	1.07 (0.85, 1.33)	1.09 (1.01, 1.21)	1.17 (0.95, 1.44)	43
	β-HCH	1.10 (0.88, 1.39)	1.22 (1.09, 1.40)	1.34 (1.09, 1.66)	33
	PCB-138	1.04 (0.78, 1.39)	1.60 (1.29, 1.89)	1.66 (1.21, 1.85)	8
GRd	HCB	1.01 (0.79, 1.31)	1.19 (1.08, 1.36)	1.20 (0.98, 1.49)	6
	β-HCH	1.14 (0.90, 1.47)	1.16 (1.05, 1.32)	1.32 (1.07, 1.65)	47
	PCB-138	1.02 (0.81, 1.27)	1.05 (1.00, 1.12)	1.07 (0.86, 1.33)	29

^aEffect estimates (95% CIs) of each natural log-unit increase in AT POP concentrations and the estimated percentage mediated by selected in situ oxidative stress biomarkers on the risk of non-hormone-dependent cancers (n = 247).

^bNon-hormone-dependent (NHD) cancers (n = 17), excluding hormone-dependent cancers (n = 6) from the analysis. Rest of the study population (n = 224). Superoxide dismutase (SOD); glutathione reductase (GRd). Models were adjusted for age (years), sex (male/female), BMI (kg/m²), smoking (yes/no), alcohol consumption (yes/no), place of residence (urban vs semi-rural), and education (lower than primary education, primary education, or higher than primary).

^cThe direct effect, indirect effect, and total effect reflect the natural log hazard ratios (HR) and 95% confidence intervals (95% CI). The indirect effect represents the mediated effect.

^dPercent mediated = indirect effect/(direct effect + indirect effect) × 100.