Epidemiological review on the health impacts of semiconductor production

Since the early 1980s, there has been a concern over the harmful health effects of semiconductor production (cancer and reproductive problems). In the context of legal disputes, several epidemiological studies were conducted in Korea, Taiwan, UK and the US. Regarding cancer, retrospective cohort studies based on employment cohort estimated standardized mortality or incidence ratio with a reference to the general population. Internal comparisons between potential exposure classes based on job records were also made. One exception was the proportional mortality ratio study. Although no consistent findings were observed across studies, elevated risk for some types of cancer was found: breast (Taiwan), lung (UK), brain and hematologic (US), and hematologic cancer (Korea). However, all studies were subject to selection bias due to healthy worker effect and loss to follow-up, information bias from misclassification of exposure status, and insufficient statistical power, all of which could bias the results toward the null. Regarding reproductive outcomes, menstrual cycle aberration, decreased fecundability and increase of spontaneous abortion were consistently observed across studies in various designs. In addition, elevated risk of congenital anomalies and some cancer among offsprings were found in Taiwan. International collaboration and in-person survey is required to overcome insufficient statistical power and to minimize information bias from the exclusive use of secondary data. Above all, disclosure of corporate data is required against trade secret and workers' participation in planning and conducting studies should be guaranteed for further understanding of the actual working conditions.