Statistical Analysis

Sociodemographic characteristic of participants was reported as mean (SD) for continuous variables and as number (percentages) for categorical variables. Logistic regressions were performed to evaluate the independent and combined associations of occupational noise exposure and smoking with hearing loss. Then the associations were further evaluated with stratified sex and age, based on previous published reports suggesting that age and sex may be important factors for hearing loss^{21, 38}. The models were conducted with those do not smoke and no occupational noise exposure as the reference group. We chose covariates which could affect hearing loss according to evidence from published literatures^{17, 21}. Covariates included age, sex, race, shift work, drinking status, hypertension, ototoxicity medicine, and chronic diseases history (diabetes mellitus, coronary heart disease, myocardial infarction and stroke). All statistical analyses were performed using SAS version 9.2 software (SAS institute Inc., Cary, NC). The statistical tests were two sided, and significance was set at P < 0.05.