In the present prospective cohort study, we found that inverse associations between prenatal ELF-MF exposure and fetal growth were only in girls. Compared with girls whose prenatal MF level was below median, girls with prenatal MF level above medians of TWA, P50 and P75 had a lower birth weight, thinner skinfolds of back, triceps, and abdomen, and smaller circumferences of head, up-arm, and abdomen. The results indicated a sex specific effect of prenatal ELF-MF on fetal growth.
Our study was supported by several previous findings [12, 16], but not all. Vocht F et al. found a decreased birth weight for living close (≤50 m) to ELF-MF sources compared with those living > 200 m during pregnancy, and the decrease was larger for female than male babies(− 251 g vs -137 g) [12] . Another study found a 5% increased risk of SGA for living < 50 m from transmission lines, compared to those living ≥400 m [10]. A cross-sectional study on 780 children aged between 0 and 12 years found that exposure to ELF-MF from high voltage electric power lines was associated with suboptimal growth profile of children, including lower circumferences of the head, chest and height at all studied ages [16]. However, one previous study showed no difference on the average birth weight following exposure to electric heated beds during pregnancy [11]. Another study did not find significant association between exposure to high voltage electricity towers and cables and neonatal head circumference [13]. The difference among these studies might lie in the ability to accurately measure actual ELF-MF exposure during pregnancy. Most previous studies used indirect prenatal MF measurements [10,11,12] and the only study which measured a 7-day MF exposure used a single-axis device rather than a triple-axis device [11], all of which might compromise the accuracy of MF measurements and therefore led to misclassification of prenatal MF exposure. Such misclassification generally led to attenuation of an underlying MF effect because of the more likely nature of non-differential. Moreover, gender differences were not well addressed in some studies and the confounding could not be appropriately considered.
Although the biological mechanisms of the effect of prenatal ELF-MF exposure on fetal growth are currently unclear, ELF-MF has been reported to be able to influence the oxidated state and intracellular Ca2+ signaling patterns at a cellular level [17, 18], which could in turn result in placental vascular function changes and subsequently suboptimal fetal growth [19]. Another study showed that MF exposure could impact cellular glucose activity level [20], which provided another possible mechanism of our observed associations. In addition, we had previously reported associations between maternal ELF-MF exposure and increased risk of shorter embryonic bud length, miscarriage, and obesity. Therefore, it is biologically plausible that prenatal ELF-MF exposure have an adverse effect on fetal growth.
The study had several strengths. The most important strength was the prospective nature of the study design. Information on prenatal ELF-MF exposure was collected before child birth, which reduced information bias compared to measurements done retrospectively. Second, we measured actual personal ELF-MF exposure with measuring instrument quantitatively, which provided a more accurate ELF-MF measurement and thus a better opportunity to detect the association between prenatal ELF-MF exposure and fetal growth, if existed. Third, the present study was able to control for a number of potential confounders such as parental BMI before pregnancy [21] and maternal passive smoking during pregnancy [22], which were reported to play an important role in birth weight.
Several potential limitations should be kept in mind when the results of this study were interpreted. First, the single 24-h measurement might limit the capability to acquire the complete profile of ELF-MF exposure during the time period of interest, resulting in misclassification of the MF exposure level. However, the design that woman wore meter in a typical day and was monitored consecutively for 24 h, increased measurement validity of subjects’ exposure during the 3rd trimester. In addition, the misclassification would most likely be non-differential in prospective study, and would bias the results towards null. Moreover, the measurement of MF in a typical day might reduce the potential misclassification to a large extent [23]. Second, only MF exposure from extremely low frequency was collected by EMDEX Lite meter; Thus, our finding may or may not apply to MF in other frequencies. Third, the participation rate was somewhat low in the study (about 56%). It was a prospective cohort study and MF exposure level was largely unknown to the general public, so that the low participation rate was unlikely to be associated with MF exposure. Fourth, a chance finding could exist since the sample size in boys or girls was relatively small. However, the consistent associations observed across various MF matrices (TWA, P50, and P75) and fetal growth indices might alleviate the concern. Fifth, the information of maternal smoking during pregnancy was not collected. However, only about 2.4% of women smoke in China [24]. The percentage of women who smoked during pregnancy was expected to be lower than 2.4% in this cohort of pregnant women and was not expected to bias our estimates largely. Alternatively, passive smoking might play a more important role in confounding our results and had already been adjusted for throughout the analyses. Finally, the small sample size did not allow evaluating the dose-responsive effect with finer categories of exposure, and restricted the ability to verify the associations by high exposure. Further prospective cohort studies with larger sample size will be more promising to verify the potential adverse effect of prenatal exposure to MF.
In conclusion, this study indicated that prenatal exposure to higher ELF-MF could result in reduction of birth weight, circumference of head, upper arm, and abdomen and skinfold thickness of back, triceps and abdomen in girls, but not in boys. Future studies with larger sample sizes are warranted to confirm our findings.