This study was performed to investigate the effectiveness of the questionnaire.

Workers'health examination records, work environment record, and questionnaires of selected industries as samples were analysed using logistic regression analysis and discrimination analysis Results'Cases of bearing impairment (Dl+C) as dependent variables, and age, work duration and level of environmental noise as independent variables were selected for multiple unconditional logistic regression analysis. Odds ratio was 4.04 in hearing difficulty, 2.78 in tlnnitus and 1.08 in age. In the second analysis Noise induced hearing loss is selected as dependent variable. The OR in hearing difficulty was 3.67(95 % C.1. : 1.61 8.61), and was 1.09(95 % C.1. : 1.05-1.14) in age. Conditionnal multlple logistic regression analysis was performed. In hearing impairment as dependent variable, the OR of age was 1.02(95 % C.1. : 1.00-1.04) and other variables were not significant. However, NIHL as dependent, the OR of hearing difficulty was 4.57(95 % C.1. : 1.43-14.67). According to multiple logistic regression adopting each items of questionnaire as dependent variable, the only item of hearing difficulty showed significant difference with hearing ability. The discrimination analysis was performed with the class variable of hearing impairment, and discrimination variables of age, work duration, and environment noise level. The sensitivity of 59 %, and specificity of 88 %, and average error count of 23 % were obtained. When the numbers of items answered in questionnaire were assumed as the parameter of judgement for noise induced hearing loss (NIHL), the highest sensitivity and specificity were 33.5% and 49.0% in cases of more than two items answered.

The development of items that is simple and discriminative mainly consisted of chronic and specific diseases related items is needed. Computerization for newly developed items might be needed to establish effective surveillance system for NIHL in the future.