Longitudinal Trends Of Peripheral Blood Cell Counts Of Benzene Exposed Workers And Correction Of Exposure Measurement Error

Benzene is one of the most important aromatic chemicals, which has been used in industry for more than 150 years and is still widely used. Peripheral blood cell counts examination is one of the most important and convenient indicators of health surveillance for benzene-exposed workers. It has been proved that compared with the non-exposed workers, decreased peripheral counts of white blood cell (WBC), red blood cell (RBC) and platelets can be found in benzene-exposed groups. Literatures have showed consistent reports on the hematological effects of high benzene exposures, while the effects of low levels of benzene exposure are not well understood.There have been few studies, however, which attempt to examine the relationship between a quantitative estimate of benzene exposure and hematological response, with controversial results for this relationship. As natural peripheral blood cell counts variations within individuals, cross-sectional study cannot distinguish the degree of variation in blood cell counts across time (by benzene exposure) for one worker from the variation in blood cell counts among people. Also, little attention to measurement error for exposures was paid. Accurate estimates for retrospective benzene exposure levels are the basis for the assessment of the relationship between occupational exposure and disease. As the nature of retrospective studies, we cannot assess the validity and reliability of the estimates. In such situation, a validation study is considered to assess the measurement error of retrospective estimates and the influence on the relationship between the exposures and the disease.For the purpose mentioned above, a longitudinal data analysis was conducted to study into the changes of blood cell counts of benzene-exposed workers with exposure time. At same time, correction of exposure measurement error was also perfomred for the retrospective benzene estimates and correction method of its effect on regression parameters was studied.192 subjects were selected from a benzene exposed factory named as Rubber Product Facility (95 benzene-exposed and 97 unexposed). Each with 14 years of hematological data from 1990 to 2003. In the exposed group, there were 727 WBC samples, and 724 Hb ones. While in the unexposed group, there were 486WBC samples and 487 Hb ones..Age, sex and smoking status were matched. The average years of benzene exposure for the exposed subjects were 18.5?.5. During retrospective benzene exposure assessment, a workshop-location code was developed instead of job titles which seemed to be too crude. Combined with work location retrospective questionnaire andhistorical benzene monitoring data, levels of four exposure variables, time-weighted average(TWA), peak exposure(PE), average exposure intensity (AEI) and cumulative exposure level(CEL) were estimated for the period between 1990 to 2003 for different work locations. Through one benzene estimate level for one hematological result for one year, a 14 year longitudinal data analysis of the effect of benzene exposure on peripheral blood cell counts were conducted. In order to assess the validity and reliability of the historical benzene estimates which is the measurement error of the estimates and its influence on the regression model of benzene exposure and peripheral blood cell counts, Two week validation survey was conducted in this facility and gold standard of personal monitoring by 3M Samplerwas chosen to link the true exposure and estimate one. Correction of measurement error was also done to correct the parameters of the "na'ive" regression model.In retrospective benzene exposure assessment, the range of TWA, PE, AEI and CELwas0.23~111ppm (0.7-331.2mg/m3), 0.30-221 ppm (0.9~662.4mg/m3), 0.06-110.4ppm (0.20-331.2mg/m3) and12.73~23407.96ppm-month (38.18~70223.89mg/m3month) respectively. The range of PE was a little higher than that of TWA and AEI ?The range of TWA estimates was nearly the same with that of AEI.Cross-sectional study showed that WBC counts were significantly lower in high exposure groups( >...