PROSPECTIVELY ASSESSED MENSTRUAL-CYCLE CHARACTERISTICS IN FEMALE WAFER-FABRICATION AND NONFABRICATION SEMICONDUCTOR EMPLOYEES

Women aged 18-44 years in silicon-wafer fabrication-room (fab) jobs an d frequency-matched women in nonfab jobs were screened for a prospecti ve study of reproductive health (n = 2,639). Among the 739 (28%) eligi ble women, 481 (65%) completed a baseline interview; 402 completed at least one menstrual cycle of follow-up with daily diaries and urinary assays to exclude conceptive cycles. Adjusted mean cycle lengths (MCL) did not differ between fab and nonfab women (p = 0.97). Women working in thin film and ion implantation (TFII) had the highest adjusted MCL (34.8 +/- 1.7 days) compared with nonfab workers (32.5 +/- 1.4 days, p = 0.07). Among women working exclusively in one group, TFII women ha d significantly higher MCL (36.1 +/- 2.04 days) than nonfab women (32. 0 +/- 1.38 days, p = 0.017). TFII women were also more likely to have all cycles >35 days (adjusted relative risk [RR] = 2.45; 95% CI = 0.85 -6.06). Variability was assessed by logarithmic transformation of the mean standard deviation (MLSD) in cycle length per woman and adjusted for age and ethnicity (4.5 days for fab vs. 4.0 days for nonfab, p = 0 .16). Women working exclusively in TFII or photolithography (PHOTO) ha d significantly higher adjusted MLSD in cycle length (6.68 +/- 1.28 an d 5.72 +/- 1.24 days, respectively) than women in nonfab (4.1 +/- 1.16 days, p = 0.013 and 0.019, respectively). Fab and nonfab women did no t differ significantly in mean days of bleeding or risk of having cycl es >35 or <24 days. However, elevated risks of having cycles <24 days were seen in supervisor engineers (adjusted RR = 2.46, 95% CI = 1.19-3 .63) and PHOTO women (adjusted RR = 1.83, 95% CI = 0.94-2.88). (C) 199 5 Wiley-Liss, Inc.