Health risks of radiofrequency electromagnetic fields, and the World Health Organization’s programme

Concerns about health effects of electromagnetic fields (EMF) date back as far as the use of EMF in human activities.¹ In recent decades, adverse health effects have been extensively studied both for extremely low frequency (ELF) fields, used in providing electrical power, and radiofrequency (RF) fields, used in telecommunication.

The health issue intensely studied relates to RF exposures from the use of mobile phones, beginning in 1987 and becoming almost universal in developed countries by 2006.² RF refers to electromagnetic energy at frequencies from 100kHz to 300GHz, being of lower frequency and therefore lower photon energy than visible light. It is “non-ionising”, meaning that there is not sufficient energy to ionise an atom or a molecule. Ionising radiation such as X-rays and atomic radiation are at much higher energy levels. RF at high levels produces heating; internationally accepted exposure limits for RF in communications are set at much lower levels, at which there is no established mechanism for carcinogenicity.³ The International Agency for Research on Cancer (IARC), which is part of the World Health Organization (WHO), reported in 2013 on their extensive review of studies.⁴ The IARC review concluded that there was some weak evidence for several effects of RF from experimental work, but no strong or consistent evidence of any mechanism for carcinogenesis. They considered genetic damage, oxidative stress and effects on the immune system, gene and protein expression, cellular signalling, the blood–brain barrier and cellular replication.⁴

Brain cancers in particular have been studied extensively due to the close proximity of a mobile phone against the head when making a call. At the time of the IARC review, there were some case–control studies showing increased risks of brain cancers related to mobile phone use, and some laboratory studies indicating an increase in some tumours in animals after experimental exposure. While there were also many studies showing no or negligible increases in risk, the IARC classified RF as a possible carcinogen, group 2B.

The exposures to radio frequencies in New Zealand are very similar to those in other developed countries.⁴ Radio and television broadcasting is a long-standing exposure. The use of mobile phones in New Zealand increased rapidly after 1990, to more than 50% of the population using them from about 2000, and almost all the population from 2006.² To assess cancer risks from mobile phones, large international studies have been done to give adequate numbers and international relevance. The Interphone case–control study of brain cancers and mobile phone exposures was performed in 13 countries and co-ordinated by the WHO. In New Zealand, 135 newly diagnosed patients with brain cancers (glioma and meningioma) and 172 controls were included in the combined analyses of 5,190 cases and 7,658 controls.^5,6 The study essentially showed no overall increased risk, although a small increase was seen with the highest category of cumulative call time. Other studies, including very large international collaborative cohort studies such as the Million Women Study in the United Kingdom⁷ and the COSMOS study,⁸ have not shown an increased risk.

Particular concerns relate to RF exposures in children. Earlier work suggested that children might absorb RF more than adults,⁹ but this has now been discounted.³ A further international case–control study, the MOBI-Kids study, was set up in 14 countries to investigate brain cancers in children. The study involved identifying recently diagnosed children with brain cancers, and as a control group children with appendicitis were enrolled. New Zealand contributed 16 cases and 29 controls to the international analysis of 899 cases and 1,910 controls, which demonstrated no increased risk.¹⁰ Studies of the use and some short-term effects of mobile phone use in children have also been done in New Zealand.^11,12

If there was an increased risk of brain cancer from mobile phone exposures, increases in the incidence would be expected. Studies have been done of trends in brain cancer incidence in New Zealand, relating this to mobile phone usage. The incidence of glioma from ages 10 to 69 has shown a small decrease over the last 25 years from 1995 to 2020, during which time the use of mobile phones has become almost universal.^2,13 The data are sufficient to reasonably exclude an increased risk of 15% for all mobile phone users, even assuming a 20-year latent period, or equivalently a 2.5 times increased risk in 10% of the population with maximal intensity of exposure.²

The recent WHO reviews

The WHO is currently developing an Environmental Health Criteria monograph that will assess the safety of RF exposure for humans. As part of the assessment, in 2019 the WHO commissioned a series of major reviews of potential adverse health outcomes related to RF.¹⁴ Thirteen reports have been published, each a review of all relevant studies using a pre-agreed assessment system. For each, a protocol was peer-reviewed, published, and registered. The reviews were conducted by multinational teams that included 85 scientists of various disciplines. All topics have now been reported in peer-reviewed papers. All protocols and reviews were published in Environment International. A further task group set up by the WHO will review the findings, formulate an overall risk assessment and identify good practice interventions and research gaps.¹⁵

While the reviews cover different topics and types of study, findings were assessed using the Grading of Recommendations Assessment Development and Evaluation (GRADE) system.^16,17 Based on this approach, the level of confidence in each exposure/outcome association was classified according to four descriptors:

• High: the true effect is highly likely to be reflected in the apparent relationship.
• Moderate: the true effect may be reflected in the apparent relationship.
• Low: the true effect may be different from the apparent relationship.
• Very low: the true effect is highly likely to be different from the apparent relationship.

This system has been designed and most widely used for therapeutic questions, and the top classification of high confidence is defined as based on “controlled exposure”. This is only applicable to experimental studies such as randomised clinical trials. The highest level of confidence based on human epidemiological studies, even if these are well done, is moderate confidence since these are based on observational investigations and the exposures are not controlled. Observational studies are the best possible design to study potential long-term human hazards such as cancer.¹⁸

The confidence in the results was assessed by incorporating “downgrading factors”, such as risk of bias, unexplained inconsistency, imprecision and publication bias, and “upgrading factors”, such as large effects, dose response and lack of residual confounding.

Table 1 summarises the reviews and the results. It is striking that even though there have been many studies, for many topics the final assessment was that the confidence in the results was low or very low. This usually meant that the studies were individually small, the subjects were poorly selected, the RF exposure or the outcomes being studied were poorly measured, confounding by other factors was not controlled adequately or that the results were inconsistent.

For human studies, for nearly all topics there was no consistent evidence of an adverse effect of RF exposure. The strongest evidence relating to human experience was for cancer, where 63 studies and 119 exposure–outcome relationships were examined.¹⁹ The review concluded that there was no increased risk, relating mobile phone use to brain cancers, and radio and television transmissions to leukaemia in children. This evidence was assessed as being of moderate confidence, as it relied on observational studies. For other types of cancer and other exposures, there was low or very low confidence.²⁰

Other outcomes with observational studies in humans included cognitive impairment,²¹ symptoms,²² female reproductive outcomes²³ and male fertility.²⁴ No effects were found, but the evidence was regarded as low or very low confidence.

Experimental studies in humans, such as clinical trials under laboratory conditions in volunteers, give high or moderate confidence of evidence that RF exposure did not produce symptoms.^25,26 For cognitive performance, none of the meta-analyses observed a statistically significant effect of RF/EMF exposure compared with sham exposure.²⁷ Experimental exposures of human sperm gave very low confidence evidence of reductions in sperm motility, but no effects on sperm DNA damage.²⁸

The data on experimental studies in animals are in contrast. The evidence was assessed as high certainty for increases in glioma and in heart schwannoma in male rats after RF exposures. There was moderate certainty for increases in lymphoma, pheochromocytoma, liver cancer and lung cancer in rats or mice.²⁹ No associations were seen with several other types of cancer. The methods used in this review differed from the published protocol and from the methods used in the other reviews. These results have been heavily criticised largely because they selectively emphasise isolated positive findings, overlook the large number of non‑significant results and apply an overly generous risk of bias and confidence assessment—especially for two key studies with well‑documented methodological flaws.^30,31

Animal experimental studies of male fertility showed decreases in pregnancy rates with moderate confidence,²⁸ and in studies of female reproductive outcomes a decrease in foetal weight was found with moderate confidence, although there were no effects on litter size or brain.³² The review of in vitro and in vivo studies of biomarkers of oxidative stress showed no effects, but with very low confidence.³³

Conclusions

These WHO-sponsored reviews are not designed to set policies; their objective is only to present the scientific evidence. Policies regarding safety limits, standards and legal requirements are set by governments, but often follow the recommendations of international groups such as the International Commission on Non-Ionizing Radiation Protection (ICNIRP). ICNIRP is an independent commission set up in 1992, and later linked to WHO, which has developed exposure guidelines that are used as a the basis of legal standards in most countries, including New Zealand and Australia.³ A further task group set up by the WHO will review the findings of these reviews, formulate an overall risk assessment and identify good practice interventions and research gaps.¹⁵ That may include reviewing more recently published work, and will lead to an Environmental Health Criteria monograph. This process will need to consider uncertainty and the limits of evidence, and should provide objective recommendations for policy, which governments can consider.

The reviews looking at observational studies in humans did not find associations between RF and any health effects, particularly cancer. But questions remain. For many topics the reviews concluded that the information available was too limited, of poor quality or too inconsistent to allow firm conclusions.

The relevance of animal experiments to human exposures is a key issue. Epidemiological studies give evidence of actual harms to humans; animal experiments and mechanistic studies give evidence of potential harms. Balancing the evidence from human observational studies and from animal experimental studies is a key challenge in assessing safety. The authors of the review of animal studies noted in their summary: “However, even in cases where the animal evidence demonstrates high confidence of evidence, the extrapolation of risk from cancer bioassays to humans is particularly complex for RF electromagnetic fields.”²⁹ As the mechanism of any effect is unknown, the appropriate dose of RF is uncertain, as well as species differences in absorption and metabolism.

The review process and the results have been criticised; one review states that the reviews “provide no assurance of safety”.³⁴ The reviews show that for many topics, low- or very low–quality evidence was found, making it impossible to demonstrate a risk, but also impossible to rule out a risk.

These reviews will not remove the differences in opinion of scientists, or of the wider public, on the risks of RF fields. Some of these differences are based on closely argued technical discussion, such as the selection of the studies for review and the methods of meta-analysis.³⁴ Much of the dispute can be related to limits of the information and the scarcity of data on many issues.³⁵ But other influences are important. Conflicts of interest are often defined narrowly as relating only to funding and affiliations—and there is empiric evidence that industry-supported studies are less likely to show risks.³⁶ The critique of these WHO reviews³⁴ emphasises the overlap between the review groups and ICNIRP, which is suggested as compromising their independence. Conflicts, however, include wider issues; few scientists or commentators are free from any conflict of interest, if only in their wish to support their own previous results. Assumptions, values and other personal attitudes may influence interpretations, particularly in complex, important and uncertain questions, as described by Woodward et al.³⁷ No science can “prove” the absence of any risk, although that is often what is asked by the public.³⁵

New Zealand has not been large enough to support major studies itself, but its contributions to international case–control studies in adult and childhood brain cancer have been important. The studies of trends in brain cancer have been valuable in themselves, and consistent with overseas studies. Importantly, from involvement in these international studies the level of expertise in this area in New Zealand is sufficient to guide appropriate standard setting and policies.

View Table 1.