Health risk in the offspring of female semiconductor workers

(1)

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Occupational Medicine doi:10.1093/occmed/kqn046

SHORT REPORT

Health risk in the offspring of female semiconductor

workers

Ching-Chun Lin1, Jung-Der Wang1,2,3,4, Gong-Yih Hsieh5, Yu-Yin Chang4and Pau-Chung Chen1,4

Background There are no published studies focusing on adverse birth outcomes or infant mortality in the semi-conductor industry.

Aim To investigate whether female workers have higher risks of any adverse birth outcome or death from congenital malformation.

Methods A total of 27 610 female workers had been employed in eight semiconductor companies in Taiwan between 1980 and 2000. Using the national birth registry, their live born children were identified, and then any deaths under 5 years of age with or without congenital malformations were identified by linking with the national death registry. Periconceptional exposure was defined as the mother having been employed in the semiconductor industry 3 months before and 3 months after conception of the live born infants.

Results A total of 24 223 live births were included. No significant association between adverse birth outcomes or death with congenital malformation and maternal employment in semiconductor industry was found either in the period of 1980–94 or 1995–2000.

Conclusions There is no convincing evidence that female workers employed during the periconceptional period in the semiconductor industry had higher risks of having adverse birth outcomes or death due to con-genital malformations. However, prospective research is warranted to confirm these findings. Key words Birth outcome; congenital malformation; female worker; occupational exposure; semiconductor

industry.

Introduction

The semiconductor industry is one of the most important worldwide high-technology industries. Many chemicals which were regularly used in the manufacturing have been reported or suspected to have reproductive toxicity. For example, ethylene glycol ethers which were associated with reproductive toxicity were widely used in the 1980s [1]. In our previous studies [2,3], we found that ethylene glycol

ethers were still used in the semiconductor manufacturing industry of Taiwan. Previous research found that the fe-male workers in semiconductor manufacturing have ad-verse reproductive health effects. They may have higher risk of having menstrual irregularity [2,4], prolonged time to pregnancy, diminished probability of conception [3,5] or spontaneous abortion [6–11]. To our knowledge, there are no studies focusing on adverse birth outcomes or infant mortality in children born to mothers employed in the semiconductor industry. Therefore, the objective of this study was to investigate whether female workers were at higher risk of any adverse birth outcome or death from congenital malformation.

Methods

A total of 27 610 female workers were employed at the eight major semiconductor companies in Taiwan during 1980–2000. Identifying particulars were obtained from the Bureau of Labour Insurance and included the national identification number, sex, date of birth, company and in-sured wage and its effective date of labour insurance. The female worker’s identification number was used to retrieve

1_{Institute of Occupational Medicine and Industrial Hygiene, National Taiwan}

University College of Public Health, Taipei, Taiwan.

2_{Department of Internal Medicine, National Taiwan University Hospital, Taipei,}

Taiwan.

3

Department of Environmental and Occupational Medicine, National Taiwan University Hospital, Taipei, Taiwan.

4_{Center for Health Risk Assessment and Policy, National Taiwan University}

College of Public Health, Taipei, Taiwan.

5_{Department of Industrial Management, Aletheia University, Taipei County,}

Taiwan.

Correspondence to: Pau-Chung Chen, Institute of Occupational Medicine and Industrial Hygiene, National Taiwan University College of Public Health, 17 Syujhou Road, Taipei 10055, Taiwan. Tel:1886 2 3322 8088; fax: 1886 2 2358 2402; e-mail: pchen@ntu.edu.tw

(2)

data about their offspring from the Taiwan National Birth Registration. A total of 24 223 live born children were iden-tified from the registry from 1978 to 2000. Information on sex, date of birth, single/multiple pregnancy, gestational age and birth weight and parental national identification numbers, dates of birth, educational levels, marital status and maternal parity was obtained for each child.

The deaths under 5 years of age in the offspring of the cohort were then examined by using the identification number to link with the Taiwan National Death Registry from 1970 to 2000. Information on date of birth and death, places of residence, education and underlying cause of death [coded according to the ninth revision of the International Classification of Diseases (ICD-9)] was obtained by reviewing death certificates which had been maintained by the national death registry.

Female workers who were employed at the semicon-ductor companies in the 3 months preceding and the 3 months after the conception of the live born children were classified as exposed, otherwise female workers were classified as non-exposed group. The date of conception was estimated as the date of birth minus the length of gestation plus 14 days. When length of gestation was not known, it was estimated as 40 weeks.

Low and high birth weights babies were those with birth weight ,2500 g and .4000 g, respectively. Pre-term and post-Pre-term babies were defined as those born before 37 and after 42 completed weeks of gestation, respectively, as measured from the first day of the last menstrual period. Small and large gestational ages were

defined as a birth weight below the 10th percentile and above the 90th percentile of gender-specific birth weight for gestational age [12].

Underlying causes of deaths were classified according to ICD-9 into five groups: congenital malformations, perinatal causes (including prematurity, birth trauma and asphyxia, maternal conditions classified as fatal and non-infectious respiratory disorders), infectious disor-ders, malignant neoplasms, and other causes. The anom-alies were further classified into two subgroups: heart and others.

Because the potential exposure to chemical agents may have been higher in earlier manufacturing processes, we divided the course of time into two periods: 1980 to 1994 and 1995 to 2000. Multiple logistic regression was used to estimate odds ratio and 95% confidence interval for ad-verse birth outcomes and causes of death in the live born children of the female workers according to their status of periconceptional exposure. The potential confounding variables included infant sex, maternal age, paternal ed-ucation, parity and multiple birth. Statistical analysis was performed using the SAS software, release 8.0.

Results

Of the 24 223 live born children, 11 256 and 12 967 were born between 1980 and 1994 and between 1995 and 2000, respectively. Mothers in the exposed group were slightly older than those in the non-exposed group during the earlier time period. Mothers in exposed groups had

Table 1. Parental characteristics of live births in female semiconductor workers

Characteristics 1980–94 1995–2000

Exposed Non-exposed P value Exposed Non-exposed P value

Total 1645 (%) 9611 (%) 3933 (%) 9034 (%) Maternal age (years)

,25 665 (40) 4978 (52) ,0.001 1360 (35) 3057 (34) NS 25–29 784 (48) 3875 (40) 1736 (44) 4054 (45)

$30 196 (12) 758 (8) 837 (21) 1923 (21) Paternal age (years)

,30 1183 (72) 7372 (77) ,0.001 2437 (62) 5561 (62) NS 30–34 408 (25) 1937 (20) 1207 (31) 2716 (30)

$35 54 (3) 302 (3) 289 (7) 757 (8) Maternal education (years)

#9 327 (20) 2475 (26) ,0.001 1118 (28) 2958 (33) ,0.001 10–12 1087 (66) 6373 (66) 1913 (49) 4490 (50)

$13 231 (14) 763 (8) 899 (23) 1586 (18) Paternal education (years)

#9 368 (22) 3054 (32) ,0.001 956 (24) 2694 (30) ,0.001 10–12 705 (43) 4475 (47) 1540 (39) 3791 (42) $13 572 (35) 2082 (22) 1437 (37) 2549 (28) Parity Primipara 1021 (62) 5248 (55) ,0.001 2120 (54) 4248 (47) ,0.001 Multipara 624 (38) 4363 (45) 1813 (46) 4786 (53) NS, not significant.

(3)

a higher proportion with ‘education .12 years’ than those in non-exposed groups during each time periods (Table 1). However, there were no statistically significant findings between exposed and non-exposed groups in either time period for vital status, infant sex, adverse birth outcomes and causes of death (Table 2). There were no significant differences between exposed and non-exposed mothers in each time period after adjusting for potential confounding factors (Table 3).

Discussion

In our study, no significant association between adverse birth outcomes or death with congenital malformation and work in the semiconductor industry was found either in the period of 1980–94 or 1995–2000.

Previous studies in the semiconductor industry have focused on irregular menstrual cycles [2], subfertility [3,5] and spontaneous abortion [6–11]. Due to lack of

data in the national registries on spontaneous abortion, stillbirth and congenital malformations, we used birth and death registries only and then evaluated adverse birth outcomes and deaths due to congenital malformations. Although the reasons for the negative results are unclear, adverse birth outcomes which may not be too specific to detect any exposure effect need to be considered. The small number of fatal congenital malformations whose mothers worked in the semiconductor industry during pregnancy is also worth mentioning in this study. More-over, one might speculate that exposed pregnancies could cause infertility or increase spontaneous abortion that might reduce the occurrence of births. Thus, the expo-sure effects on adverse birth outcomes and fatal congen-ital malformations might be underestimated.

This reproductive health study may be subject to a se-lection bias inherent to such studies. Female workers with a poor obstetric history may quit their jobs or those in a privileged position may find it easy to combine

Table 2. Birth outcomes, vital status and cause of death of live births to female semiconductor industry workers

Characteristics 1980–94 1995–2000

Exposed Non-exposed P value Exposed Non-exposed P value

Total 1645 (%) 9611 (%) 3933 (%) 9034 (%) Infant sex Male 852 (52) 5015 (52) NS 2049 (52) 4676 (52) NS Female 793 (48) 4596 (48) 1884 (48) 4358 (48) Multiple birth Singleton 1611 (98) 9490 (99) ,0.01 3839 (98) 8820 (98) NS Twin or triplet 34 (2) 121 (1) 94 (2) 214 (2)

Gestation age (weeks)

,37 71 (4) 349 (4) NS 228 (6) 548 (6) NS 37–42 1568 (95) 9223 (96) 3699 (94) 8484 (94)

.42 6 (0.4) 39 (0.4) 4 (0.1) 12 (0.1) Birth weight (grams)

,2500 83 (5) 445 (5) NS 217 (6) 487 (5) NS 2500–3999 1509 (92) 8791 (92) 3625 (92) 8306 (92)

$4000 53 (3) 375 (4) 91 (2) 251 (3) Fatal growth

Small for gestational age 205 (12) 1237 (13) NS 398 (10) 864 (10) NS Normal 1309 (80) 7472 (78) 3181 (81) 7315 (81)

Large for gestational age 131 (8) 902 (9) 354 (9) 855 (10) Vital status Live 1635 9526 3914 8976 Dead 10 85 19 58 ,28 days 7 24 9 22 28–364 days 2 34 7 20 1–4 years 1 27 3 16 Cause of death Congenital malformations Heart 1 15 3 9 Others 1 6 4 9 Perinatal causes 4 11 5 15 Infectious disorders 1 7 2 4 Malignant neoplasms 0 0 0 4 Other causes 2 43 5 17 NS, not significant.

(4)

motherhood and a career. The risk could thus be under-estimated due to such healthy worker or privilege effects. Conversely, female workers who were infertile or those under financial pressure may remain at work. These in-fertile worker or desperation effects [13] would affect the risk in the other direction. Altogether, it would seem that no obvious selection bias existed in this study.

This study has a number of potential limitations. First, due to lack of abortion and stillbirth registries, spontane-ous abortion or stillbirth from congenital anomalies was not taken into account. Second, only severe congenital malformations could be identified from the death registry in this study, suggesting that non-fatal anomalies could have been neglected. Third, it is possible that the foetus with known congenital anomalies was electively termi-nated before birth due to the advanced prenatal diagnos-tic procedures available in early pregnancy. All the above would mean that the effects could be underestimated.

A further limitation in our study was the absence of adequate exposure measurements and difficulty in defin-ing the critical exposure period. Female workers perform-ing administrative tasks may have also been included in the study. However, there were consistent results after ex-cluding8% managers or office workers in the exposed groups obtained from the Ministry of the Interior data-base. Thus, the size of the misclassification bias in relation to fabrication could be overlooked. Secondly, we used an internal control comparison instead of general population to reduce the difference of lifestyle risk factors and tenta-tively presumed there were no noteworthy confounders such as parental smoking. Thirdly, under registration of neonatal death may underestimate the effect on causes of death but may not be a major impact on interpretation after the complete implementation of the revised birth

registration system since October 1994 and the National Health Insurance programme since March 1995.

In conclusion, there was no convincing evidence that female workers employed during the periconceptional period in the semiconductor industry had higher risks of having adverse birth outcomes or death with congenital malformations. However, prospective research is war-ranted to confirm these negative findings.

Funding

National Science Council, Taiwan (NSC 92-2320-B-002-171).

Conflicts of interest

None declared.

References

1. Koh D, Chan G, Yap E. World at work: the electronics in-dustry. Occup Environ Med 2004;61:180–183.

Key points

• There was no evidence that female workers em-ployed during the periconceptional period in the semiconductor industry had higher risks of having adverse birth outcomes or death with congenital malformations. However, prospective research is warranted to confirm these negative findings.

Table 3. Crude and adjusted odds ratios (OR) and 95% confidence intervals (CI) for adverse birth outcomes and causes of death

1980–94 1995–2000

Crude OR (95% CI) AdjustedaOR (95% CI) Crude OR (95% CI) AdjustedaOR (95% CI)

Birth outcome

Pre-term 1.20 (0.93–1.56) 1.19 (0.92–1.55) 0.95 (0.81–1.12) 0.95 (0.81–1.12) Post-term 0.90 (0.38–2.13) 0.98 (0.41–2.34) 0.77 (0.25–2.37) 0.77 (0.25–2.39) Low birth weight 1.10 (0.86–1.40) 1.17 (0.92–1.49) 1.02 (0.87–1.21) 1.02 (0.86–1.20) High birth weight 0.82 (0.61–1.10) 0.78 (0.58–1.04) 0.83 (0.65–1.06) 0.84 (0.66–1.08) Small for gestational age 0.97 (0.82–1.13) 1.03 (0.88–1.21) 1.06 (0.94–1.21) 1.08 (0.95–1.23) Large for gestational age 0.84 (0.69–1.01) 0.80 (0.66–0.97) 0.95 (0.83–1.08) 0.96 (0.84–1.10) Cause of death Congenital malformations 0.56 (0.13–2.37) 0.56 (0.13–2.40) 0.89 (0.37–2.14) 0.88 (0.37–2.11) Heart 0.39 (0.05–2.95) 0.40 (0.05–3.01) 0.77 (0.21–2.83) 0.76 (0.20–2.82) Other 0.97 (0.12–8.10) 0.95 (0.11–8.04) 1.02 (0.31–3.32) 0.97 (0.30–3.18) Perinatal causes 2.13 (0.68–6.69) 2.30 (0.72–7.30) 0.77 (0.28–2.11) 0.77 (0.28–2.14) Infectious disorders 0.83 (0.10–6.79) 0.88 (0.11–7.22) 1.15 (0.21–6.27) 1.03 (0.19–5.72) Malignant neoplasms – – – – Other causes 0.27 (0.07–1.12) 0.30 (0.07–1.23) 0.68 (0.25–1.83) 0.62 (0.23–1.70)

(5)

2. Hsieh GY, Wang JD, Cheng TJ et al. Prolonged menstrual cycles in female workers exposed to ethylene glycol ethers in the semiconductor manufacturing industry. Occup Environ Med 2005;62:510–516.

3. Chen PC, Hsieh GY, Wang JD et al. Prolonged time to preg-nancy in female workers exposed to ethylene glycol ethers in semiconductor manufacturing. Epidemiology 2002;13: 191–196.

4. Gold EB, Eskenazi B, Hammond SK et al. Prospectively assessed menstrual cycle characteristics in female wafer-fabrication and non-wafer-fabrication semiconductor employees. Am J Ind Med 1995;28:799–815.

5. Eskenazi B, Gold EB, Samuels SJ et al. Prospective assess-ment of fecundability of female semiconductor workers. Am J Ind Med 1995;28:817–831.

6. Pastides H, Calabrese EJ, Hosmer DW Jr et al. Spontaneous abortion and general illness symptoms among semiconduc-tor manufacturers. J Occup Med 1988;30:543–551. 7. Beaumont JJ, Swan SH, Hammond SK et al. Historical

co-hort investigation of spontaneous abortion in the Semicon-ductor Health Study: epidemiologic methods and analyses

of risk in fabrication overall and in fabrication work groups. Am J Ind Med 1995;28:735–750.

8. Swan SH, Beaumont JJ, Hammond SK et al. Historical co-hort study of spontaneous abortion among fabrication workers in the Semiconductor Health Study: agent-level analysis. Am J Ind Med 1995;28:751–769.

9. Eskenazi B, Gold EB, Lasley BL et al. Prospective monitor-ing of early fetal loss and clinical spontaneous abortion among female semiconductor workers. Am J Ind Med 1995;28:833–846.

10. Correa A, Gray RH, Cohen R et al. Ethylene glycol ethers and risks of spontaneous abortion and subfertility. Am J Epidemiol 1996;143:707–717.

11. Elliott RC, Jones JR, McElvenny DM et al. Spontaneous abortion in the British Semiconductor Industry: an HSE in-vestigation. Am J Ind Med 1999;36:557–572.

12. Hsieh WS, Wu HC, Jeng SF et al. Nationwide singleton birth weight percentiles by gestational age in Taiwan, 1998–2002. Acta Paediatr Taiwan 2006;47:25–33.

13. Joffe M. Biases in research on reproduction and women’s work. Int J Epidemiol 1985;14:118–123.