【作者】 何永华；

【导师】 傅华； 叶细标；

【作者基本信息】 复旦大学 ， 劳动卫生与环境卫生学， 2005， 博士

【摘要】 接触评估（exposure assessaent）是根据研究对象的职业史,定性或定量估算其职业生活中通过各种方式接触一种或多种物质（或因素）的程度或强度,是职业流行病学的重要组成部分。接触评估研究应该有严格、完整的方法,才能得出正确、可靠的评估结果,得到合理解释的剂量—效应（或剂量—反应）关系,保护工人的安全和健康,预防职业损害。随着学科的发展和测定资料的逐步积累,对接触评估的要求越来越高,建立准确定量评估方法也越来越紧迫。目前,主要评估方法有职业接触矩阵法、专家法、物理模型法、生物样品法等。但各种评估方法均有待完善,如专家评估法的主观性较强,灵敏度变化较大,可重复性不高。接触评估还有很多重要问题亟需解决。如,评估结果的有效检验和比较,使用何种接触剂量指标表示接触水平,选用何种代表值进行计算。对这些问题没达成一致,所以很多评估结果相差较大,其无偏性、有效性、一致性都亟需提高。 苯是重要的芳香族化工原料,用途广泛。接苯工人健康状况与苯接触可能密切相关。研究发现,与非接苯人群相比,接苯工人的外周血白细胞、红细胞、血小板计数下降,较高浓度的苯接触还导致血红蛋白含量降低。但也有与此不一致的研究结果。由于大多数研究的接触水平估算欠准确,或者所采用的剂量指标有限,或未充分考虑生物效应的发展过程,所以结果不能完全吻合。 针对上述问题,本研究将以某橡胶厂车间空气中苯浓度的历史监测资料为基础,选择合适的浓度典型值,用资料法、阶段法、模型法分别评估该厂工人的区域浓度、时间加权平均浓度（TWA₈）、年度接触水平、累积接触水平,并以现场测定结果对三种方法评估的区域浓度、TWA₈浓度的结果进行直接检验。同时根据三种评估方法的结果,采用混合效应模型研究接苯水平对工人历年外周血白细胞计数、血红蛋白含量的影响,并以模型分析的信息值为标准,间接比较以上三种回顾性接触评估方法的准确性,选择最佳接触剂量指标。 本研究的资料法即为直接使用该厂的职业卫生区域浓度监测资料,结合现场工厂水平调查、岗位水平调查划分相似接触组（SEG）,估算各岗位的区域浓度,如果当年资料缺失,则以其上下各一年的均值替代;阶段法在SEG的基础上,划分相似接触时期（SEP）,取该时期SEG监测资料的均值为区域浓度;模型法则在阶段法的基础上,结合了专家主观概率、监测资料,进行贝叶斯统计推断,估算更多还原

【Abstract】 Exposure assessment （EA） is qualitative or quantitative evaluation of exposure levels of chemicals or factors in one’ s occupational history, which is an important component of occupational epidemiology. EA should have scientific methods to get correct and reliable results, reasonable dose-effect （dose-response） relationships, in order to protect workers’ health and safety. Development of new EA methods is necessary with the increasing requirements on EA. Currently common methods include job-exposure matrix, expert subjective, physical model prediction, biological mark and so on, but there are some shortage in all of the methods, and it is necessary to futher improve. Expert subjective, for example, is prone to subjectivity, large variety of sensitivity and low repetition. In addition, there are many problems on EA which need to be solved, such as validation and comparison of results, choice of reasonable typical value and dose indexes. No agreements on those questions resulted in great differences of estimation for exposure levels and influence its unbias, validity and consistency.Benzene, a kind of widely used aromatic material, is a hazard to health. Studies showed that peripheral counts of white blood cell （WBC）, red blood cell and platelet of workers exposed to benzene decreased significantly compared with that of the unexposed. Furthermore, low hemoglobin （Hb） concentrations were observed on subjects exposed to high level benzene in comparison with the unexposed. Inconsistent results are also reported on the adverse effect of benzene. The inconsistency is explainable giving that most studies had imprecise exposure assessment, or limited dose indexes, or incorrect typical value, or even the neglect of natural cycle of adverse effects.For the problems mentioned above, the present study assessed exposurelevels of subjects exposed to benzene from a rubber factory by data method, period method, model method on basis of historic measurements of area concentration. Personal exposure dose indexes included area concentration, time weighted average of 8 hours （TWA₈）, yearly level, cumulative level. Results of area concentration and TWA₈ were compared and directly tested with field samples. Relationships between exposure levels and WBC, Hb were analyzed with mixed model. Accuracy of the 3 methods and choice of the best dose indexes were conducted by comparison of their information criteria values.Data method in present study meant that area concentrations were calculated by directly using historic measurements in association with similar exposure groups （SEGs） according to 2 level questionnaires on factory and tasks. If no data available for certain years, area concentrations were substitute by that of the nearest years. Period method took geometric means as area concentrations within similar exposure periods （SEPs） that were established on the basis of SEGs. Model method appraised area concentrations by the combination of historic measurements and expert objective with Bayesian statistics model. TWAh, yearly level and cumulative level were respectively assessed according to exposed hour daily and exposed month yearly.515 area concentration samples of historic measurements during past 20 years from 1964 to 2003 were collected, with geometric mean 11. 76mg/m3. 26292 person-months of 94 males and 53 females exposed to benzene, 14112 person-months of 40 males and 26 females unexposed, were observed.Results of the 3 methods showed that tasks of slurry-making, cloth-lining-assembling, slurry-applying had the highest area concentrations and TWA₈ of benzene. Averages of area concentrations of subjects exposed to benzene calculated by the 3 methods were 12. 19 mg/m3, 5. 10 mg/m3, 4. 26mg/m3respectively, and that of TWA₈ were 14. 75 mg/m3, 4. 69 mg/m3, 6.36mg/m3, yearly level 118.41 mg/m3. month, 72.9 mg/m3. month, 72. 63mg/m3. month, cumulative level 2371. 81 mg/m3. month, 783. 7 mg/m3. month, 835.14 mg/m3. month separately.Area concentration and TWA₈ of data method were significant higherthan that of period method and model method in most years. Furthermore, the former was more instable than the others. Numbers of area concentration of the 3 methods which were not obviously different with field measurements of 8 tasks were 2, 4, and 7 respectively and that of TWAh were 6, 6, and 7 compared with field measurements of 9 tasks. The results mentioned above suggested that there were less variety between model method output and field measurements than data method, and period method was medium.Mixed effect analysis on exposure level and WBC as the following were not able to achieve their convergence: area concentration and yearly level of data method; analysis on exposure levels and Hb as the following were not convergent: area concentration and TWA* of data method, cumulative levels of data method and period method. Analysis on all dose indexes of model method and WBC, Hb were convergent. It suggested that exposure assessment of model method fitted well, and that of data method and period method dispersed, fitted bad and was in need to be improved.The present results indicated that data method was practically simple and easy, but adequate historic measurements were in need; period method could save costs and be applicable to subjects of stable exposure and with certain measurements; model method was able to get precise assessment and fitted subjects with sparse instable measurements, but pertain information and profound experts were necessary.The study suggested that area concentration, TWA?, yearly level and cumulative level were all useful dose indexes because they all had dose effect relationships of WBC. Of all 4 dose indexes, TWA* had the smallest information criteria values, which were 4858.09, 4828.51, 4977.97, 5039.98 （according to -2 Restricted Log Likelihood of mixed analysis on TWA? and WBC）. Therefore TWA? should be recommended as the most suitable dose indexes of benzene exposure.The finding in the present study indicated that there were significant negative correlations between benzene exposure levels and peripheral WBC. 3 coefficients of area concentration, TWA*, yearly level, cumulative level are-0.028> -0.022, -0.0019, -0.000076 （according to model method）.更多还原