Fig. 2. Hearing thresholds based on cadmium level quartiles. For the univariate analysis, the 1Q, 2Q, 3Q, and 4Q groups had mean Low/Mid-Freq values of 15.3±0.5, 16.5±0.5, 16.7±0.4, and 17.9±0.5, respectively; mean High-Freq values of 30.3±0.7, 31.2±0.7, 29.9±0.6, and 31.9±0.7, respectively; mean AHT values of 17.2±0.5, 18.3±0.5, 18.2±0.5, and 19.7±0.5, respectively. For the multivariate analysis, the 1Q, 2Q, 3Q, and 4Q groups had mean Low/Mid-Freq values of 15.1±0.5, 16.2±0.4, 16.6±0.4, and 18.0±0.4, respectively; mean High-Freq values of 28.0±0.7, 30.3±0.6, 30.7±0.6, and 33.7±0.6, respectively; mean AHT values of 16.6±0.5, 17.9±0.5, 18.3±0.5, and 20.1±0.5, respectively. The data are expressed as mean and standard error. The model 1 for hearing thresholds was adjusted for age, sex, presence of DM and hypertension, smoking habit, alcohol intake, and occupational and explosive noise exposure. The model 2 for hearing thresholds was adjusted for age, sex, presence of DM and hypertension, smoking habit, alcohol intake, occupational and explosive noise exposure, and lead level. a)p<0.05 vs. 1Q, b)p<0.05 vs. 2Q, c)p<0.05 vs. 3Q. 1Q, first quartile; 2Q, second quartile; 3Q, third quartile; 4Q, fourth quartile; Low/Mid-Freq, low or middle frequency; High-Freq, high frequency; AHT, average hearing threshold; DM, diabetes mellitus.