Prospective Study of Pregnancy Outcomes After Parental Cell Phone Exposure

The Norwegian Mother and Child Cohort Study

Abstract

Background: 

Research about prenatal exposure to electromagnetic fields from cell phones among expectant parents and reproductive outcome is limited. The aim of this article is to investigate the association between pregnancy outcome and parental cell phone exposure in a large prospective study.

Methods: 

The study was based on the Norwegian Mother and Child Cohort Study conducted during the decade 1999–2009. In that study, pregnant women were recruited before a routine ultrasound examination during gestational week 15; they answered a questionnaire at that time and again around gestational week 30. The expectant father was invited to answer a questionnaire during gestational week 15 (2001–2009). The forms contained questions regarding cell phone use. The response rate was 38.7% and the cohort comprised 100,730 singleton births. Pregnancy outcomes were obtained by linkage to the Medical Birth Registry of Norway.

Results: 

The risk of preeclampsia was slightly lower among women with medium and high cell phone exposure compared with low exposure after adjusting for potential confounders. Fathers with testis exposure when using cell phones had a borderline increased risk of perinatal mortality among offspring and a slightly decreased risk of partner developing preeclampsia during pregnancy compared with no cell phone exposure of head or testis. None of the other pregnancy outcomes was associated with cell phone exposure.

Conclusions: 

We found no association between maternal prenatal or paternal preconceptional cell phone exposure and any of the studied pregnancy outcomes. The only risk estimate suggesting a potential increased risk was not consistent with other findings.

Cell phone use has increased dramatically in the past 25 years, transforming from a rare occurrence to a common one; in many contexts, the cell phone has replaced the stationary phone. Cell phone technology is based on the transmission of voice, text, and images by radiofrequency electromagnetic fields, to which cell phone users are exposed during calls. The development of new 3G cell phone technology has largely improved this situation through the reduction of radio frequency exposure when the cell phone is used.

Brain cancer has been the main health concern regarding cell phone use due to the placement of the cell phone close to the head during calls, but other issues such as headache, concentration, and behavioral problems have also raised concern. In 2011, the International Agency for Research on Cancer (IARC) classified radiofrequency fields as possibly carcinogenic to humans due to an increased risk of glioma and acoustic neuroma (Group 2B).^1,2 The levels of radiofrequency electromagnetic fields exposure from mobile phones on the human brain have been examined using simulations and phantoms as well as postmortem studies.² It has, for instance, been shown that exposure of the brain regions of young children to radiofrequency electromagnetic fields can be 1.6 to three times higher than exposure of regions in adult brains.² Such studies have led to implementation of strategies to reduce radiofrequency electromagnetic fields exposure during wireless phone calls; for instance, the use of hands-free devices for cell phones has reduced exposure from the phones to the head. However, this has led to new questions regarding possible exposure to other parts of the body, primarily the gonad region. Electric and magnetic fields induced in tissues depend on several factors, among them anatomy, distance, and dielectric tissue properties. Therefore, resulting induced fields are strongly nonuniform, with hotspots and variations in field strengths over several orders of magnitudes.² Therefore, the exact tissue exposure is difficult to assess. The report from IARC² also discusses the possible mechanism for adverse health effects from radiofrequency electromagnetic fields. Biological tissue heating is a recognized effect of the radiofrequency fields that may cause adverse health effects. However, there is no agreement about adverse health effects related to radiofrequency fields at exposure intensities that do not cause a detectable increase in tissue temperature, except for reactions mediated by free radical pairs. Still, the IARC report states that it is likely that not all mechanisms of interaction between weak radiofrequency fields and biological structures have been discovered or characterized.

Safety standards for physical exposure are often lowered for vulnerable groups such as the elderly, the ill, and pregnant women (ICNIRP),³ and exposure limits are five-fold lower for the general public than for occupational exposure settings. Studies of cell phone exposure among pregnant women are sparse, but a few regarding behavioral problems in children after mothers’ cell phone use during pregnancy have been published.^4,5 Studies on pregnancy problems or reproductive outcomes in relation to cell phone exposure have to our knowledge not been published in English.

Cell phone use and the possible effect on semen parameters have been studied among men attending fertility clinics.^6–9 These studies suggest an association between prolonged cell phone use and adverse sperm motility, whereas results concerning other sperm parameters differed among the studies. However, due to methodological limitations, uncertainties are attached to the results. As far as we know, studies of reproductive outcome after paternal preconceptional exposure to cell phones have not been published.

Occupational studies among both men and women exposed to radiofrequency electromagnetic fields show some adverse reproductive outcomes, but the results are not homogeneous.^10–15 A problem with all previous studies regarding radiofrequency electromagnetic fields exposure and reproductive outcome has been the uncertainties in exposure assessment and dosimetry, and, in some studies, uncertainty in reproductive outcomes assessed through questionnaires.

This study aims to test the association between maternal cell phone use during pregnancy and adverse pregnancy outcomes, as well as paternal preconceptional cell phone use and adverse pregnancy outcomes. The study is based on a large cohort where data on cell phone use was obtained from questionnaires filled out during gestational weeks 15 and 30 and linked to the Medical Birth Registry of Norway (MBRN) to achieve the pregnancy outcome variables.

METHODS

The article is based on the Norwegian Mother and Child Cohort Study (MoBa), which is a prospective population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health.¹⁶ During the period of 1999–2009, pregnant women from at least 15 weeks of gestation were recruited by mail from all over Norway before a routine ultrasound examination. The cohort includes more than 100,000 pregnancies. Of the total number of invited women, 38.7% consented to participate.¹⁷ The MoBa participants received three questionnaires during pregnancy. In this study, the self-administered questionnaires from around gestational age 15 and 30 weeks were used. At the end of 2001, expectant fathers were also invited to participate by answering a questionnaire around week 15 of gestation. The participation rate among fathers was 31.8%.¹⁷ The MoBa has been described in earlier publications.¹⁸ This study was based on Version 7 of the quality-assured data files. Informed consent was obtained from each MoBa participant upon recruitment. Follow-up was conducted by linkage to the MBRN to obtain information about all pregnancies and offspring at birth. Only singleton births were included and the current database consists of 100,434 births with maternal information and 74,908 births with paternal information (Table 1). The Regional Committee for Medical Research Ethics in South-Eastern Norway approved the study.

TABLE 1: Contribution of Data From the Different Questionnaires in the Norwegian Mother and Child Cohort (1999–2009)

Maternal Cell Phone Exposure

Women who participated in the MoBa answered two questions regarding cell phone use in both gestational weeks 15 and 30. The questions were identical in the two questionnaires. The women were asked “How often do you talk on a cell phone?” and the options to respond were “Seldom/never,” “A few times per week,” “Daily,” and “On average more than 1 hour per day.” The second question was “Do you talk on your cell phone for longer than 15 minutes at a time?” and the response alternatives were “Never,” “Seldom,” and “Often.” These two questions were combined to reflect cell phone exposure as Low, Medium, and High, as shown in Table 2. There was substantial agreement between the answers regarding cell phone use for gestational weeks 15 and 30 (Cohen’s κ equals 0.61 and 0.58 for the first and the second question, respectively). In the analysis, we used a variable that was assigned the highest exposure from the two questionnaires. If exposure information was available from only one questionnaire, that was used. In addition, subanalyses were done for gestational weeks 15 and 30 separately. Due to no answers regarding cell phone use, 0.2% was set to missing.

TABLE 2: Cell Phone Use in Gestational Week 15 Among Women Participating in the Norwegian Mother and Child Cohort (1999–2009), Definition of the Combined Cell Phone Exposure Variable

Paternal Cell Phone Exposure

The expectant fathers answered a questionnaire at around week 15 of gestation. The questions regarding cell phone use were changed through different versions of the paternal questionnaire. Common to the different versions was the question regarding cell phone use before conception: “How often did you talk on a cell phone during the last 6 months before your partner got pregnant?” But the response alternatives differed between different versions of the questionnaires. In the first two versions (n = 41,397, 0.3% was set to missing due to no answer), the alternatives were “Seldom/never,” “A few times per week,” “Daily,” or “On average more than 1 hour per day.” For the last versions of the questionnaire (n = 33,511, 4.0% did not answer), the alternatives were “Less than once a week,” “1–2 times a week,” “3–6 times a week,” “1–4 times a day,” or “More than 5 times a day.” A common variable to express cell phone exposure with three categories was created as follows:

Low: Answering “Seldom/never” or “A few times per week” in the first versions and “Less than once a week,” “1–2 times a week,” or “3–6 times a week” in the last versions of the questionnaire.

Medium: “Daily” and “1–4 times a day.”

High: “On average more than 1 hour per day” and “More than 5 times a day.”

For the two last versions of the questionnaire, four questions regarding current cell phone habits were also included. These questions were more detailed and were used as proxy for exposure before conception. The questions with response alternatives in parentheses were as follows: “Do you use a cell phone?” (“No” or “Yes”), “Do you use a ‘hands-free’ device?” (“Seldom/never,” “Only during longer calls,” or “Normally”), “If/when you use a ‘hands-free’ device, where do you mainly have the phone during the call?” (“In the trouser pocket in front,” “On the belt in front on the body,” “Other places on the body,” or “Away from the body”), and “How long do you talk on average in total on days with cell phone calls?” (“Less than 1 minute,” “1–10 minutes,” “11–30 minutes,” “31–60 minutes,” or “More than 60 minutes”). These questions were combined to create a variable that describes regular cell phone exposure and comprises what part of the body was most exposed during a cell phone call.

No exposure: No use of cell phone or less than 1 minute per exposure day on average, or use of hands-free device “Other places on the body” or “Away from the body” when talking on the phone.

Head exposure: Cell phone use more than 1 minute per exposure day and use of hands-free device seldom or never.

Testis exposure: Cell phone use more than 1 minute per exposure day and use of hands-free device normally or during long calls and had the cell phone in front either in the trouser pocket or on the belt.

The combined variable was missing in 3.8% of men.

Reproductive Outcomes

The MBRN is based on compulsory notification of all live births and stillbirths from 12 weeks of gestation. In a standardized notification form, data on demographic variables, maternal health before and during pregnancy, complications during pregnancy, and delivery and pregnancy outcome are reported by the attending midwife and/or the physician. In this study, data from singleton births were abstracted from the birth registry for participants in the MoBa study. Adverse reproductive outcomes analyzed were congenital malformation, perinatal mortality, low birth weight, preterm birth, small for gestational age (SGA), preeclampsia, abruptio placentae, and change in sex ratio.

Congenital malformations are diagnosed during the medical examination of newborns at the birth clinic or at the neonatal wards and are reported to the MBRN. Classification of malformation is based on the International Classification of Diseases (ICD10). We analyzed all malformations together without regard for specific malformations. Perinatal mortality was defined as stillbirth and death within the first week of life. Low birth weight was defined as a birth weight of less than 2,500 g; 80 births did not have a registered weight, and an additional 165 birth weights were set to missing because the weight was lower than 500 g due to uncertainty regarding the weight. Gestational age was based on ultrasound; if the ultrasound was missing, then the last menstrual date was used. Preterm birth was defined as a gestational age of fewer than 37 completed weeks. To exclude obvious misclassifications of preterm birth, preterm pregnancies with gestational age-specific birth weight calculated z score greater than 3.5¹⁹ (absolute value) was excluded, also those without a registered gestational age was excluded (together n = 663). SGA was defined as a birth weight of less than the 10th percentile for gestational age¹⁹ and 470 births were set to missing due to unknown gestational age or birth weight. The definition of pregnancies with preeclampsia was in accordance with the MBRN’s definition.²⁰ The sex ratio was measured as the ratio of the number of boys to the number of girls. We analyzed abruptio placentae for maternal exposure only.

Covariates

We obtained information on maternal smoking habits from the questionnaire in week 15 or 30 of gestation, whereas paternal smoking information was obtained from the paternal questionnaire. Both were defined as follows: no smoking was defined on the basis of answering “never having smoked” or “no smoking during pregnancy.” Smoking sometimes and smoking daily were categories from the questionnaire. Maternal smoking was set to the most severe smoking habits reported in gestational week 15 or 30 or, if answered only in one of these weeks, we used that answer. Maternal smoking was set to missing for 1,863 births and 921 births for paternal smoking. From MBRN, maternal and paternal age, as well as parity, was obtained. Maternal age was in years while paternal age was in categories (below 19 years, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, and more than 50 years). Parity was dichotomized in groups: first pregnancy and second or later pregnancies.

Statistical Analyses

We compiled characteristics of the cohort with regard to maternal and paternal age, parity, and smoking habits for the different cell phone exposures. Cohen’s κ statistics were calculated as a measure of agreement between maternal cell phone use in gestational weeks 15 and 30. To estimate relative risk (RR) and 95% confidence interval (CI) for adverse reproductive outcome after cell phone exposure, we applied log-binomial regression. Initially, all analyses were performed with and without including year of birth as a potential confounding factor. Because this factor did not influence the association with cell phone exposure for any of the outcomes, we did not include it in the final analysis. The analyses for maternal exposure were adjusted for maternal age, maternal smoking, and parity. Due to the established knowledge of higher risk of preeclampsia in nulliparous women, subanalysis with stratification according to parity was performed for preeclampsia after mother’s cell phone exposure. Paternal exposure was adjusted for maternal and paternal age, maternal and paternal smoking and parity. The data were analyzed using STATA SE 12.

RESULTS

This study includes the cohort of 100,231 births with information on maternal prenatal cell phone exposure; 15,139 (15%) of the pregnant women had high exposure to cell phone use (Table 3). There was a difference in cell phone use among women expecting their first child compared with those expecting a second or later child. Among low cell phone exposure, there were 28% nulliparous women, whereas among high cell phone exposure, there were 61% nulliparous women.

TABLE 3: Maternal Cohort Characteristics by Prenatal Cell Phone Exposure

Among 74,908 men answering the questionnaire, 73,467 men answered questions regarding cell phone exposure during the 6 months before conception. Only the last two versions of the questionnaire contained the questions concerning regular cell phone habits, and 32,102 reported no exposure of head or testis, any head exposure, or any testis exposure. Paternal cohort characteristics are shown in Table 4.

TABLE 4: Paternal Cohort Characteristics by Cell Phone Exposure 6 Months Before Conception

Table 5 gives the RR for the perinatal outcome variables after maternal cell phone use, with low exposure as a reference. After adjustment, the risk of preeclampsia was lower among those with medium and high cell phone exposure compared with those with low exposure. Stratified analysis of parity revealed an adjusted RR = 0.86 (95% CI = 0.77, 0.96) for medium and RR = 0.85 (0.75, 0.97) for high exposure relative to low cell phone exposure in first pregnancy (data not tabulated). The corresponding RRs for later pregnancies were close to the null (0.92 [0.82, 1.03] and 0.93 [0.77, 1.13]). None of the other reproductive outcome variables were associated with cell phone use (Table 5). Abruptio placentae were also not associated with cell phone exposure (data not shown). All analyses were also done separately for gestational age 15 and 30, and the results were similar.

TABLE 5: Percent and Adjusteda Relative Risk of Pregnancy Outcomes by Maternal Prenatal (Week 15 or 30 Gestation) Cell Phone Exposure

The adjusted RR for adverse reproductive outcome after paternal cell phone exposure during the 6 months before conception showed similar risk estimates for low, medium, and high cell phone use (Table 6). Regarding paternal regular cell phone habits, exposure to the testes was associated with an increased risk of perinatal mortality among offspring and decreased risk of partner developing preeclampsia in pregnancy compared with no exposure to the head or testes (Table 6).

TABLE 6: Percent and Adjusteda Relative Risk of Pregnancy Outcomes by Paternal Cell Phone Exposure 6 Months Before Conception

DISCUSSION

In this prospective study, we could not observe any association between maternal cell phone use during pregnancy and pregnancy outcomes: congenital malformation, perinatal mortality, low birth weight, preterm birth, SGA, abruptio placentae, or change in sex ratio. The results concerning preeclampsia suggested a reduced risk but this was not consistent with the other findings. We also did not find any clear and consistent association between the adverse pregnancy outcomes and paternal preconceptional cell phone use.

Maternal Exposure

Maternal studies concerning pregnancy outcome related to cell phone use are scarce. We are aware of two case–control studies, both published in Chinese with abstracts available in English. The first study included 200 cases of early abortions and age-matched controls, and reported that after adjusting for risk factors, the multifactorial analysis revealed a significant association between mobile phone use and the risk for early spontaneous abortion.²¹ Another study by the same group used the same study design but with 138 cases of embryo growth ceasing as endpoint.²² The multifactorial analysis resulted in an association with the mother’s mobile phone use (OR = 6.0, 95% CI = 1.9, 18.9). The results of these studies are difficult to interpret, because only the abstracts are available. The mobile exposure information was self-reported and the risk of introducing recall bias is probably high because the information was obtained after the outcomes were known.

Several animal studies have been conducted in which pregnant animals received exposures to fields similar to those from cell phones during pregnancy to assess effects on fetal development. Many of those suffer from lack of control or documentation of exposure. In others, exposure and dosimetry have been characterized with whole body specific absorption rate ranging from about 0.02 to 4.0 W/kg.^23–27 Different types of mobile phone systems and frequencies were used in the studies. Extensive teratological evaluations were performed including end points, such as number of live, dead, or resorbed embryos; sex ratios; weights; abnormalities; physical functioning; and behavioral development. None of these studies indicated any effect of exposure. In three of the studies only 12 animals were exposed, which makes it doubtful whether potentially subtle effects would have been detected. However, Ogawa et al.²⁵ exposed 20 female animals and Sommer et al.²⁶ exposed 34, with similar results.

The majority of epidemiologic studies concerning maternal radiofrequency electromagnetic field exposure and pregnancy outcomes have studied physiotherapists using diathermy. However, there are no consistent results from these studies. One register-based study from Denmark found a lower percentage of boys born,¹⁰ whereas a Swiss survey among female physiotherapists showed no change in sex ratio.¹¹ An Australian survey found lower incidence of congenital anomalies and miscarriage among 824 physiotherapists compared with the general population.¹²

There have been some studies regarding behavioral problems in children after maternal cell phone use during pregnancy.^4,5,28,29 In a study from the Danish National Birth Cohort,⁴ cell phone exposure was obtained from a questionnaire 7 years after birth. Based on this information, the authors found a higher risk of behavior problems after prenatal cell phone exposure; however, they wrote that the results should be interpreted with caution. In 2012, this group performed the same research on another group of participants in the Danish National Birth Cohort and they replicated their original findings.²⁸ The same group of researchers also studied prenatal cell phone use and motor or cognitive/language developmental delays among infants at 6 and 18 months of age based on the same dataset in which where cell phone use was obtained 7 years subsequently; for these outcomes, the authors found no associations.²⁹ Vrijheid and colleagues⁵ used data from a prospective birth cohort to examine the relation between cell phones use during pregnancy and neurodevelopment at 14 months of age, but found no association, nor was cell phone use related to the child’s sex, birth weight, or prematurity. The study was based on only 587 pregnancies.

Paternal Exposure

Concerning paternal exposure, pregnancy outcome may be affected through an adverse effect on semen or sperm cells. Because distance to the phone is important for the radiofrequency exposure level, a potential direct effect is most likely when the cell phone is located close to the testes, whereas an indirect effect might occur if reproductive hormones are affected. In the latter case, head exposure must also be regarded. In our study, we could not observe any effect on reproductive outcomes irrespective of where the father usually held the emitting phone.

In a recent article, the association between cell phone use and semen quality in human, in vitro, and animal studies was reviewed and meta-analyses were performed.³⁰ Even though the association between cell phone use and semen parameters was inconsistent, most of the studies in the review showed a negative effect on semen parameters, in all three types of studies. The authors concluded that even though no defined effect of mobile phone use on semen quality could be drawn, men should not keep the cell phone near the testicles to avoid the potential harmful effect of the radiofrequency field on the male reproductive system. Another systematic review followed by meta-analysis included only human sperm.³¹ Based on 10 studies including 1,492 semen samples, they concluded that exposure to cell phones was negatively associated with sperm motility and viability.

Reproductive hormones were assessed in addition to semen quality in one study.⁹ No association with cell phone use was found for two hormones (prolactin and follicle stimulating hormone), one (luteinizing hormone) showed a somewhat lower and one (testosterone) a somewhat higher level in the user group. In an experimental study with global system for mobile communications cell phone signals emitted by circularly polarized antenna 40 cm from the head of the 22 male participants, no effect on luteinizing hormone was observed.³²

While we are not aware of any studies on paternal exposure to cell phones with pregnancy outcomes as endpoints, some epidemiologic studies have been performed where fathers had been exposed to radiofrequency fields in occupational settings. A questionnaire survey among male physiotherapists analyzed paternal radiofrequency exposure before conception and congenital malformations in offspring.³³ The authors stated that the total number of congenital anomalies observed appeared to be higher than expected among fathers exposed to nonionizing radiation at any time before conception and exposure 3 months before conception, but the finding was not statistically significant. The study included only live births and the description of the exposure was poor. Two studies with different paternal occupational RF exposure have been linked to MBRN. Mjøen et al.¹³ categorized paternal occupation derived from general population censuses based on job title grouped into three crude radiofrequency exposure classes. Except for a small increased risk of preterm delivery, no other clear association was found with the studied pregnancy outcomes and exposure. The main limitation of the study was the possible misclassification of exposure. The total cohort of male employees in the Royal Norwegian Navy was also linked with MBRN.¹⁵ The exposure was divided into both acute (exposure during the 3-month period before conception) and nonacute (exposure more than 3 months before conception) exposure. Both perinatal mortality among the offspring, and partner developing preeclampsia was associated with paternal service aboard fast patrol boats in the acute period. Based on the calculated radiofrequency exposure a radiofrequency effect was not considered ruled out. However, no other associations with adverse reproductive outcome were found related to the exposure.

Study Methodology Evaluation

The main strength in this study is the prospective study design where cell phone exposure was collected before birth outcome. The challenge of gathering cell phone exposure data in epidemiologic settings has been an issue since the widespread use of cell phone began. Cell phone subscription information versus self-reported phone use as exposure proxies as well as changes in technology throughout the years have been discussed.^34,35 Problems with obtaining subscription information are the numerous network providers, participants have to agree upon collection of the information, amount of use cannot be distinguished between users of the phone, and participants may have access to a separate cell phone at work. Recall bias of self-reported cell phone use was found to be low in an evaluation of the Interphone study; however, there was an underestimation of the numbers of calls and an overestimation of call duration.³⁵ Radiofrequency exposure from cordless phones could be important but unfortunately we did not have such information. We also had no information about whether a cell phone was turned on when it was carried in the pocket or on the belt, but in this situation the radiofrequency exposure has been shown to be negligible.³⁶

Our exposure categorization was based partly on frequency of use and partly on hours of use, with the intent to categorize according to amount of use at the relevant assessment times. Maternal cell phone use was current use was a combined variable of exposure in weeks 15 and 30 of gestation. The agreement between cell phone use in these 2 weeks was good, so we believe that the use of a cell phone did not change throughout pregnancy. Furthermore, we had no hypotheses regarding specific gestational weeks of the fetus development during which the exposure would have any potential effect on pregnancy outcome. The fathers were asked about current use and use during the 6-month period before conception. Some misclassification might have occurred due to the grouping of answers, and especially because of the two different sets of exposure questionnaires provided to the fathers. This did not apply to the questions regarding exposure of testes and head, but there was only information on this for a subgroup of the population. In the case of bias, it would be nondifferential. Another advantage of this study is the large cohort and the application of registry data for all outcomes.

Concerning confounding, we adjusted for age, parity, and smoking habits. Year of birth did not influence the results and so was not included in the final model. These are probably the most essential to include and we had no access to other variables. The main weakness in our study is the low response rate, which might have introduced bias. However, this is mainly a problem when estimating prevalence.¹⁸ A response bias that would have affected the assessment of estimated exposure outcome associations is less likely due to the prospective design of the study and because outcomes were obtained from a registry. Nevertheless, if the response was linked to mobile phone use and to a risk factor for the pregnancy outcomes, this would have introduced a bias which size would depend on the strength of the association between the risk factor and the outcomes.

CONCLUSION

We could not reveal any association between maternal cell phone use during pregnancy, nor could we find any association between paternal preconceptional cell phone use and any of the pregnancy outcomes researched in this study.