半導體員工生殖健康及癌症風險之探討

Abstract

背景 半導體產業在過去研究中針對職業暴露與生育力下降及罹患癌症之風險一直有它的限制及爭議性存在。目前有關半導體工業的健康研究，在國內並無生殖及癌症相關研究。本研究的目的是探討本國半導體產業員工生育力下降及死於癌症等可能性。基於研究資料之需求，而作死因資料庫準確性之確認與補正。 方法 生殖健康研究：本研究針對是在女性生殖健康，主要包含月經功能與女性員工生育力。在生育力方面，本研究中使用受孕所需時間（time to pregnancy）做為生物效應標記，以便可以偵測受孕前後的潛在生殖危害存在。 癌症死亡研究：本研究為一回溯性世代設計來探討半導體產業癌症之風險，從勞工保險資料選取半導體職業世代，並連結死亡登記檔獲得癌症資料，並使用標準死亡比來作分析。 死因資料庫準確性之確認與補正：本研究擬以衛生署死因資料檔為主體，而以內政部死因資料檔之身分證字號及戶政資料來補正。 結果 生殖健康研究：對某一晶圓製造廠女性員工進行回溯性世代研究，配合該公司1997年健康檢查進行問卷訪視，女性員工人數842人，共回收719份（回收率85.5%），月經功能有效問卷606份，進入懷孕分析共292孕次(173個女性員工)。在月經週期方面，黃光區月經平均週期長度危險勝算比為非無塵室高，潛在暴露於二醇醚類和異丙醇（危險勝算比=5.0，95%信賴區間=1.7-14.1）與氫氟酸、異丙醇和磷化物（危險勝算比=3.5，95%信賴區間=1.1-10.9）女性員工月經週期平均長度異常亦有增多。在生育力方面，黃光區（受孕力比=0.77，95%信賴區間=0.45-1.32）女性員工受孕所需時間較長，且在二醇醚類（受孕力比=0.59，95%信賴區間=0.37-0.94）暴露組受孕所需時間也是增加。 癌症死亡研究：連結死亡登記檔篩選出191個死亡個案，其中有50個死於癌症。我們發現白血病(男性: 標準死亡比=3.33, 95% 信賴區間=1.08-7.77; 5個個案)，其入廠至最後觀察時間至少5年。在本研究中共有7個白血病個案，其中有6個個案工作年資少於5年，而在1995年之後進廠的有3個個案。 死因資料庫準確性之確認與補正：第一部分，自公元1980至1997年以身分證字號與戶政資料為連結碼，連結不上之個案利用身分證字號與戶政資料分別比對後，可補正衛生署。另外，兩資料庫可連結之部分，教育及村里變項補正。第二部分，公元1980至1984年，以戶政資料為連結碼可以填補身分證字號。 結論 在黃光區內含二醇醚類的多種化學物質暴露將導致月經週期延長。這結果在半導體女性員工受孕所需時間延長扮演一個重要的角色。因此，本研究發現二醇醚類是導致女性員工生育力下降原因之一。雖然本研究沒有充分證據來證實白血病與半導體職業暴露有關，但有潛在風險存在，是一個未來值得探討的議題。

Background Previous research on reproductive health and cancer risks of working in the semiconductor industry was limited and has produced conflicting results. No studies about reproductive health and cancer risks in semiconductor industry had been published in Taiwan. The aim of this study was to determine reproductive health and the cancer mortality of workers in semiconductor industry to follow up the concern that had been expressed. Methods Reproductive health study: Reproductive health outcomes included menstrual cycle function and subfertility measured by the interview questionnaire. The subfertility biomarker used time to pregnancy in this study. Cancer mortality study: This retrospective cohort study that was established from Taiwan labor insurance database was designed to investigate the cancer risks in the semiconductor workers of Taiwan. The other information was founded by record linkage of National Mortality Registry dataset. We used the cancer standard mortality ratio (SMR) in the semiconductor worker cohort. Verification and correction of error for death registration data: The method of this study is to verify the accuracy of the dataset of death registration for the DOH with the dataset from the MOI. Results Reproductive health study We conducted a retrospective cohort study of pregnancies between March 1990 and July 1997. Of a total 842 female workers, 719 agreed to be interviewed (a response rate of 85.5%). Among the participants which 173 workers were reported 292 pregnancies were analyzed in the subfertility study, and 606 questionnaires were valid in measuring menstrual cycle function. By using workers in non-fabrication as referents, workers in photolithography and diffusion had higher risks for long menstrual cycles. Workers exposed to ethylene glycol ether (EGEs) and isopropanol (OR = 5.0; 95% CI = 1.7 to 14.1) and exposed to hydrofluoric acid, isopropanol, and phosphorous compounds (OR = 3.5; 95% CI = 1.1 to 10.9) also showed increased risks of a long menstrual cycle. Waiting time to pregnancy of female workers in the photolithography area was longer than that of those in the nonfabrication area (fecundability ratio (FR) = 0.77; 95% confidence interval (CI) = 0.45–1.32), and those who were potentially exposed to ethylene glycol ethers showed longer time to pregnancy compared with those not exposed (FR = 0.59; 95% CI = 0.37–0.94). Cancer mortality study From the linkage of dataset of National Mortality Registry, we found 191 deaths including 50 cancer cases. We found that leukemia (male: SMR=3.33, 95% CI =1.08-7.77; No=5 cases) in lag period for five years had potential risks. In this study, there were seven leukemia cases, among them, six leukemia cases were less five years working duration, there were three cases were employed after 1995. Verification and correction of error for death registration data Verification of the data from 1985-97, we refiled the missing demographic data and the missing ID no. into the dataset of the DOH. After linking the two datasets with the same demographic data, we were able to refile the missing IDs for all the death certificate data of the DOH during 1980-84. Conclusion Exposure to multiple chemicals including EGEs in photolithography might be associated with long menstrual cycles. This may play an important role in a prolonged time to pregnancy in the wafer manufacturing industry. This study provides further evidence that ethylene glycol ethers may cause female subfertility. Although, this study provided no substantial evidence that occupational exposure in semiconductor industry was associated with human leukemia risk, its potential risk needs to be followed up in the future.