The safety of visual display units (VDUs) in terms of reproductive outcomes has been questioned since the widespread introduction of VDUs in the work environment during the 1970s. Concern for adverse pregnancy outcomes was first raised as a result of numerous reports of apparent clusters of spontaneous abortion or congenital malformations among pregnant VDU operators (Blackwell and Chang 1988). While these reported clusters were determined to be no more than what could be expected by chance, given the widespread use of VDUs in the modern workplace (Bergqvist 1986), epidemiologic studies were undertaken to explore this question further.
From the published studies reviewed here, a safe conclusion would be that, in general, working with VDUs does not appear to be associated with an excess risk of adverse pregnancy outcomes. However, this generalized conclusion applies to VDUs as they are typically found and used in offices by female workers. If, however, for some technical reason, there existed a small proportion of VDUs which did induce a strong magnetic field, then this general conclusion of safety could not be applied to that special situation since it is unlikely that the published studies would have had the statistical ability to detect such an effect. In order to be able to have generalizable statements of safety, it is essential that future studies be carried out on the risk of adverse pregnancy outcomes associated with VDUs using more refined exposure measures.
The most frequently studied reproductive outcomes have been:
- Spontaneous abortion (10 studies): usually defined as a hospitalized unintentional cessation of pregnancy occurring before 20 weeks of gestation.
- Congenital malformation (8 studies): many different types were assessed, but in general, they were diagnosed at birth.
- Other outcomes (8 studies) such as low birthweight (under 2,500 g), very low birthweight (under 1,500 g), and fecundability (time to pregnancy from cessation of birth control use) have also been assessed. See table 1.
Table 1. VDU use as a factor in adverse pregnancy outcomes
Objectives |
Methods |
Results |
|||||
Study |
Outcome |
Design |
Cases |
Controls |
Exposure |
OR/RR (95% CI) |
Conclusion |
Kurppa et al. |
Congenital malformation |
Case-control |
1, 475 |
1, 475 same age, same delivery date |
Job titles, |
235 cases, |
No evidence of increased risk among women who reported exposure to VDU or among women whose job titles indicated possible exposure |
Ericson and Källén (1986) |
Spontaneous abortion, |
Case-case |
412 |
1, 032 similar age and from same registry |
Job titles |
1.2 (0.6-2.3) |
The effect of VDU use was not statistically significant |
Westerholm and Ericson |
Stillbirth, |
Cohort |
7 |
4, 117 |
Job titles |
1.1 (0.8-1.4) |
No excesses were found for any of the studied outcomes. |
Bjerkedal and Egenaes (1986) |
Stillbirth, |
Cohort |
17 |
1, 820 |
Employment records |
NR(NS) |
The study concluded that there was no indication that introduction of VDUs in the centre has led to any increase in the rate of adverse pregnancy outcomes. |
Goldhaber, Polen and Hiatt |
Spontaneous abortion, |
Case-control |
460 |
1, 123 20% of all normal births, same region, same time |
Postal questionnaire |
1.8 (1.2-2.8) |
Statistically increased risk for spontaneous abortions for VDU exposure. No excess risk for congenital malformations associates with VDU exposure. |
McDonald et al. (1988) |
Spontaneous abortion, |
Cohort |
776 |
Face-to-face interviews |
1.19 (1.09-1.38) |
No increase in risk was found among women exposed to VDUs. |
|
Nurminen and Kurppa (1988) |
Threatened abortion, |
Cohort |
239 |
Face-to-face interviews |
0.9 |
The crude and adjusted rate ratios did not show statistically significant effects for working with VDUs. |
|
Bryant and Love (1989) |
Spontaneous abortion |
Case-control |
344 |
647 |
Face-to-face interviews |
1.14 (p = 0.47) prenatal |
VDU use was similar between the cases and both the prenatal controls and postnatal controls. |
Windham et al. (1990) |
Spontaneous abortion, |
Case-control |
626 |
1,308 same age, same last menstrual period |
Telephone interviews |
1.2 (0.88-1.6) |
Crude odds ratios for spontaneous abortion and VDU use less than 20 hours per week were 1.2; 95% CI 0.88-1.6, minimum of 20 hours per week were 1.3; 95% CI 0.87-1.5. Risks for low birthweight and intra-uterine growth retardation were not significantly elevated. |
Brandt and |
Congenital malformation |
Case-control |
421 |
1,365; 9.2% of all pregnancies, same registry |
Postal questionnaire |
0.96 (0.76-1.20) |
Use of VDUs during pregnancy was not associated with a risk of congenital malformations. |
Nielsen and |
Spontaneous abortion |
Case-control |
1,371 |
1,699 9.2% |
Postal questionnaire |
0.94 (0.77-1.14) |
No statistically significant risk for spontaneous abortion with VDU exposure. |
Tikkanen and Heinonen |
Cardiovascular malformations |
Case-control |
573 |
1,055 same time, hospital delivery |
Face-to-face interviews |
Cases 6.0%, controls 5.0% |
No statistically significant association between VDU use and cardiovascular malformation |
Schnorr et al. |
Spontaneous abortion |
Cohort |
136 |
746 |
Company records measurement of magnetic field |
0.93 (0.63-1.38) |
No excess risk for women who used VDUs during first trimester and no apparent |
Brandt and |
Time to pregnancy |
Cohort |
188 |
Postal questionnaire |
1.61 (1.09-2.38) |
For a time to pregnancy of greater than 13 months, there was an increased relative risk for the group with at least 21 hours of weekly VDU use. |
|
Nielsen and |
Low birthweight, |
Cohort |
434 |
Postal questionnaire |
0.88 (0.67-1.66) |
No increase in risk was found among women exposed to VDUs. |
|
Roman et al. |
Spontaneous abortion |
Case-control |
150 |
297 nulliparous hospital |
Face-to-face interviews |
0.9 (0.6-1.4) |
No relation to time spent using VDUs. |
Lindbohm |
Spontaneous abortion |
Case-control |
191 |
394 medical registers |
Employment records field measurement |
1.1 (0.7-1.6), |
Comparing workers with exposure to high magnetic field strengths to those with undetectable levels the ratio was 3.4 (95% CI 1.4-8.6) |
OR = Odds ratio. CI = Confidence Interval. RR = Relative risk. NR = Value not reported. NS = Not statistically significant.
Discussion
Evaluations of reported clusters of adverse pregnancy outcomes and VDU use have concluded that there was a high probability that these clusters occurred by chance (Bergqvist 1986). In addition, the results of the few epidemiologic studies which have assessed the relation between VDU use and adverse pregnancy outcomes have, on the whole, not shown a statistically significant increased risk.
In this review, out of ten studies of spontaneous abortion, only two found a statistically significant increased risk for VDU exposure (Goldhaber, Polen and Hiatt 1988; Lindbohm et al. 1992). None of the eight studies on congenital malformations showed an excess risk associated with VDU exposure. Of the eight studies which looked at other adverse pregnancy outcomes, one has found a statistically significant association between waiting time to pregnancy and VDU use (Brandt and Nielsen 1992).
Although there are no major differences between the three studies with positive findings and those with negative ones, improvements in exposure assessment may have increased the chances of finding a significant risk. Though not exclusive to the positive studies, these three studies attempted to divide the workers into different levels of exposure. If there is a factor inherent in VDU use which predisposes a woman to adverse pregnancy outcomes, the dose received by the worker may influence the outcome. In addition, the results of the studies by Lindbohm et al. (1992) and Schnorr et al. (1991) suggest that only a small proportion of the VDUs may be responsible for increasing the risk of spontaneous abortion among users. If this is the case, failure to identify these VDUs will introduce a bias that could lead to underestimating the risk of spontaneous abortion among VDU users.
Other factors associated with work on VDUs, such as stress and ergonomic constraints, have been suggested as possible risk factors for adverse pregnancy outcomes (McDonald et al. 1988; Brandt and Nielsen 1992). Failure of many studies to control for these possible confounders may have lead to unreliable results.
While it may be biologically plausible that exposure to high levels of extremely low frequency magnetic fields through some VDUs carries an increased risk for adverse pregnancy outcomes (Bergqvist 1986), only two studies have attempted to measure these (Schnorr et al. 1991; Lindbohm et al. 1992). Extremely low frequency magnetic fields are present in any environment where electricity is used. A contribution of these fields to adverse pregnancy outcomes could only be detected if there was a variation, in time or in space, of these fields. While VDUs contribute to the overall levels of magnetic fields in the workplace, only a small percentage of the VDUs are thought to have a strong influence on the magnetic fields measured in the working environment (Lindbohm et al. 1992). Only a fraction of the women working with VDUs are thought to be exposed to levels of magnetic radiation above that which is normally encountered in the working environment (Lindbohm et al. 1992). The lack of precision in exposure assessment encountered in counting all VDU users as “exposed” weakens the ability of a study to detect the influence of magnetic fields from VDUs on adverse pregnancy outcomes.
In some studies, women who are not gainfully employed represented a large proportion of the comparison groups for women exposed to VDUs. In this comparison, certain selective processes may have affected the results (Infante-Rivard et al. 1993); for instance, women with severe diseases are selected out of the workforce, leaving healthier women more likely to have favourable reproductive outcomes in the workforce. On the other hand, an “unhealthy pregnant worker effect” is also possible, since women who have children may stop work, whereas those without children and who experience pregnancy loss may continue working. A suggested strategy to estimate the magnitude of this bias is to do separate analyses with and without women not gainfully employed.