PERSPECTIVE

published: 10 August 2021

doi: 10.3389/frcmn.2021.721925

Needed: More Reliable Bioeffects

Studies at “High Band” 5G
Frequencies
Kenneth R Foster 1* and Vijayalaxmi 2
1
Department of Bioengineering, University of Pennsylvania, Philadelphia, PA, United States, 2Department of Radiology, University
of Texas Health Science Center, San Antonio, TX, United States

One major source of controversy related to possible health effects of radiofrequency

radiation (RFR) is the large number of reported statistically significant effects of
exposure, over the entire RF part of the spectrum and over a wide range of exposure
levels, even as health agencies do not find clear evidence for health hazards of
exposure at levels within current IEEE and ICNIRP exposure limits. This
Perspective considers 31 studies related to genetic damage produced by
exposure to RFR at frequencies above 6 GHz, including at millimeter-wave (mm-
wave) frequencies. Collectively, the papers report many statistically significant
effects related to genetic damage, many at exposure levels below current
Edited by: exposure limits. However, application of five risk of bias (RoB) criteria and other
Luca Chiaraviglio, considerations suggest that the studies in many cases are vulnerable to false
University of Rome Tor Vergata, Italy
discovery (nonreplicable results). The authors call for improvements in study
Reviewed by:
Marta Parazzini,
design, analysis and reporting in future bioeffects research to provide more reliable
National Research Council (CNR), Italy information for health agencies and regulatory decision makers. This Perspective
Ahmed Elzanaty,
is a companion to another Perspective by Mattsson et al. elsewhere in this volume
King Abdullah University of Science
and Technology, Saudi Arabia (Mattsson et al., 2021)1.
*Correspondence: Keywords: millimeter waves, genetic damage, bioeffects, 5G NR, risk of bias (RoB), significance testing
Kenneth R Foster
[email protected]
mailto:[email protected] INTRODUCTION

Specialty section: The possible biological and health effects of radiofrequency (RF) energy from wireless
This article was submitted to communications have been debated by scientists and the public for many years, with
Networks, particularly vociferous public debate about the safety of 5G (more accurately, 5G New Radio or
a section of the journal 5G NR) systems that are currently being rolled out around the world. While several thousand
Frontiers in Communications and bioeffects studies have been conducted, nearly all of have been done at frequencies below 6 GHz
Networks
where most present communications systems operate.
Received: 07 June 2021 Some scientists have pointed to the many reported statistically significant effects of RF exposure as
Accepted: 29 July 2021
evidence of proof that RF fields over wide ranges of exposure parameters damage genetic material.
Published: 10 August 2021
For example, Ruediger (2009) commented:
Citation:
Foster KR,
Vijayalaxmi (2021) Needed: More
Reliable Bioeffects Studies at “High
Band” 5G Frequencies.
1
Front. Comms. Net 2:721925. Mattsson, M.-O., Simkó, M., and Foster, K. R. (2021). 5G New Radio Requires the Best Risk Assessment Possible: Experiences
doi: 10.3389/frcmn.2021.721925 Gained from Previous Studies and Assessments of RF EMF Health Effects. Under review.

Frontiers in Communications and Networks | www.frontiersin.org 1 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

“101 publications . . . have studied genotoxicity of Protection (ICNIRP), 2020 (Simkó and Mattsson, 2019;
radiofrequency electromagnetic fields (RF-EMF) in Leszczynski, 2020).
vivo and in vitro. Of these 49 reported a genotoxic This Perspective addresses the potential reasons for the
effect and 42 do not. In addition, 8 studies failed to disparity in viewpoints, between concern for “many effects” of
detect an influence on the genetic material but showed exposure to RFR on one hand, with conclusions of other experts
that RF-EMF enhanced the genotoxic action of other and health agencies that fail to find convincing evidence for
chemical or physical agent . . . there is ample evidence harmful effects of RFR at exposure levels below current safety
that RF-EMF can alter the genetic material of exposed limits. The present focus is on technical weaknesses in study
cells in vivo and in vitro and in more than one way.” design and analysis. It is not intended as a critical or systematic
(Ruediger 2009) review, for which a different analytical approach would be
needed.
Lai (2021) in a comprehensive review of genetic damage
studies commented:
METHODS
“[I]n the studies reviewed . . . approximately 70% of
them showed effects. One could say that EMF exposure We presently consider 31 genetic damage studies on animal
can lead to genetic changes. Some genetic damages and human cells exposed in vitro or in vivo to RF energy over
could eventually lead to detrimental health effects. . .. a wide range of exposure parameters at frequencies
knowing the mechanism is not necessary to accept that above 6 GHz. The papers had been extracted from a
the data are valid.” recent review by one of us of English-language papers on
genetic damage studies involving exposures to RFR between
By contrast, in reviewing the same evidence, officially- and 0.3 MHz and 300 GHz (Vijayalaxmi and Prihoda, 2019), and
health agency-sponsored expert reviews have expressed a more published between 1990 and 2017. The papers had been
cautious view. A critical review by the Scientific Committee on identified from an extensive search of standard databases,
Emerging and Newly Identified Health Risks (Scientific and at the time of the study were as complete as possible a
Committee on Emerging and Newly Identified Health Risks collection of genetic damage studies involving human and
(SCENIHR), 2015, under the auspices of the European animal cells.
Commission) concluded in 2015: The 31 papers described a total of 175 different experiments
involving RFR exposures to animals (10 studies, in vivo or in vitro
“. . .taken together, the in vitro studies differ greatly for exposures) or humans (21 studies, in vitro exposures). Most of the
exposure characteristics and duration, cell type, experiments compared RFR-exposed to sham controls; a few
biological endpoint and do not allow for any compared RFR + a known genotoxic agent such as X-rays. RFR
conclusion. Concerning genotoxicity, due to the close exposures were restricted to frequencies >6 GHz, which is the
correlation between DNA damage and cancer transition frequency in both the IEEE and ICNIRP limits at which
occurrence, and the importance of genomic the dosimetric quantity changes from specific absorption rate
instability in assessing the potential health effects of (SAR) to absorbed power density at the surface of the tissue,
radiation, the conflicting results presented here deserve reflecting the increasingly shallow penetration depth of the
future attention”. (Scientific Committee on Emerging radiation in tissue at higher frequencies.
and Newly Identified Health Risks (SCENIHR), 2015, p. We summarized effect sizes for 157 individual experiments in
69, p. 69) terms of Cohen’s d (the remaining papers did not provide
sufficient information to determine d). Cohen’s d, a standard
Similarly, critical reviews by individual scientists find at best measure of effect size, is defined as the difference in means of the
weak evidence of genotoxic effects of exposure to RFR even as they exposed and control groups divided by a pooled standard
note many reports of such effects in the literature (Verschaeve et al., deviation:
2010; Vijayalaxmi and Prihoda, 2019; Karipidis et al., 2021 for meanexposed − meancontrol
laboratory studies; for a much broader review see; IARC Working d



















(1)
Group on the Evaluation of Carcinogenic Risks to Humans, 2013). 0.5SD2exposed + SD2control 
Demonstration of genotoxicity of RFR at exposure levels within
current safety limits (International Commission on Non-Ionizing where mean and SD refer to the mean and standard deviation of
Radiation Protection (ICNIRP), 2020; Institute of Electrical and the respective group.
Electronics Engineers (IEEE), 2019) would be extremely important For a two-sample t-test with n observations and equal
for carcinogenic risk assessment. standard deviations in each group, Cohen’s d is related to the
Other recent reviews of the bioeffects literature above 6 GHz t-statistic by
show many reports of effects of exposure for many endpoints, √

t
d n (2)
many at exposure levels below international limits such as
Institute of Electrical and Electronics Engineers (IEEE), 2019 In addition, we applied the Student’s t-test (one-sided for
or International Commission on Non-Ionizing Radiation independent samples, assuming equal variances in each group)

Frontiers in Communications and Networks | www.frontiersin.org 2 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

TABLE 1 | Risk of bias (RoB) criteria.

Quality criterion Definition and importance of criterion Number of studies Number of studies
meeting criterion meeting criterion that
(total reported one or
of 31 studies; more statistically
studies met significant
multiple increase in genetic
criteria) damage in RF-exposed
cells

B: Blinded study design Whether or not the researchers state that they have used blind evaluations to avoid 20 (65%) 12 (60%)
individual bias which could interfere with more appropriate assessments
D: Appropriate dosimetry Whether or not the researchers have included a detailed/adequate description of 18 (58%) 7 (41%)
dosimetry in the publication sufficient to allow replication/confirmation studies in their own
and/or independent laboratories
P: Use of positive controls Whether or not the researchers have included positive controls in the experiment(s) to 13 (42%) 3 (23%)
confirm the sensitivity of the experimental methods to assess the chosen endpoint
S: Use of sham controls Whether or not the researchers have used sham-exposed controls, i.e., exposing the 18 (58%) 7 (41%)
controls in exactly the same equipment without turning-on the radiofrequency fields to
simulate identical exposure. In the absence of sham controls, it is not possible to attribute
changes to effects of exposure
T: Adequate temperature Whether or not the researchers have recorded the temperature during the exposure of the 26 (83%) 15 (58%)
control cells to radiofrequency fields. This is especially important for in vitro experiments since
higher temperatures are shown to induce genetic damage

FIGURE 1 | Distribution of size of exposed and control groups (typically equal) for the 175 experiments over 31 studies. The numbers refer to the number of
statistically independent samples representing different exposures to animals (not to numbers of cells from individual animals scored by investigators).

using the group means and standard deviations obtained from the in the Supplementary Material. Three other reviews [(Simkó and
papers using p < 0.05 as the criterion for statistical significance. Mattsson (2019), Karipidis et al. (2021), and Wood et al. (2021)]
Calculations were done using Matlab (The Mathworks, Natick used the same set of criteria to evaluate different sets of bioeffects
MA). The results were partly recalculated using a statistics studies at frequencies >6 GHz. While these evaluations
software package to confirm the results from Matlab. necessarily involve expert judgment and might vary somewhat
Finally, we evaluated each of the 31 studies using the five risk with assessor, all three studies had similar distributions of scores
of bias (RoB) criteria defined in Table 1; with results summarized across all papers evaluated.

Frontiers in Communications and Networks | www.frontiersin.org 3 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

FIGURE 2 | Effect size (Cohen’s d) vs incident power density from 157 experiments in 31 studies. International Commission on Non-Ionizing Radiation Protection
(ICNIRP), 2020 reference levels (occupational) shown by vertical dotted line. Experiments with statistically significant/not statistically significant increases in damage
relative to controls are indicated by + and o respectively. For a list of endpoints see the Supplementary Material.

RoB criteria have come into increasing use to assess internal (2008). Comparison of the studies is limited by the diversity of
validity of studies for use in systematic reviews and meta-analyses assays used (Supplementary Material), which however are all
of both laboratory and epidemiology studies (National sensitive measures of genetic damage.
Toxicology Program Office of Health Assessment and
Translation (OHAT), 2015). Criteria similar (but more
extensive than) those presently used have been applied by an RESULTS
expert group at Aachen University in systematic reviews of the
bioeffects literature (e.g., Bodewein et al., 2019) and, less Distribution of Study Sizes
formally, by health agencies such as the Swedish Radiation Figure 1 shows the distribution of number of statistically
Safety Authority (Swedish Radiation Safety Authority (SSM), independent samples or animals exposed in the 175
2019) in triaging excessively flawed studies from their literature experiments. Consistent with Organisation for Economic Co-
reviews. operation and Development (OECD) (2016) recommendations,
The RoB assessments and the statistical significance testing the exposed animal was considered to be the experimental unit;
had significant uncertainties, for two main reasons. typically investigators scored many cells per exposed animal. The
First, many of the papers lacked sufficient documentation to median “n” was 3, indicating that most of the studies had
permit more than a rough evaluation of RoB–a problem that extremely limited statistical power.
is hardly unique to this set of papers. Exposure characterization,
in particular, is a difficult technical problem at frequencies Effect Sizes
>6 GHz due to the short energy penetration depth in Figures 2, 3 show the effect sizes (Cohen’s d) for the 157
tissue and other factors. It is difficult to evaluate a study that experiments for which d could be extracted vs incident power
simply states that exposure (the specific absorption rate or SAR) density (Figure 2) and frequency of exposure (Figure 3).
was calculated using a software package without further Statistically significant/no significant effects are indicated by
elaboration (e.g., Karaca et al., 2012). Second, many of the +/o. Statistically significant effects are scattered over the whole
papers lacked sufficient information to allow an assessment range of exposures. Most of the results correspond to twofold or
of the correctness of the statistical analysis or even, in some less variations in damage measures between exposed and control
cases, an independent application of a significance test. For a samples. This is comparable to variations in spontaneously
fuller discussion of the problems in extracting statistical data occurring chromosome abnormalities and micronuclei
from genetic toxicology studies see Vijayalaxmi and Prihoda endpoints in human cells (Vijayalaxmi and Prihoda, 2012).

Frontiers in Communications and Networks | www.frontiersin.org 4 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

FIGURE 3 | Effect size (Cohen’s d) vs frequency of exposure for 157 experiments reported in 31 studies. Vertical dotted lines indicate the 5R NR “high band” at
25–39 GHz. Experiments reporting statistically significant/not statistically significant effects of exposure are indicated by + and o respectively.

Figure 4A shows the distribution of the d values for all 157 Risk of Bias
experiments. The effect sizes clustered near d ≈ 0 (no effect) with Table 1 and the Supplementary Material summarizes results of
the large majority between (−2 < d < 2), possibly due to the RoB analysis. Most of the 31 studies satisfied one of the
sampling effects. However, a number of outliers appear with criteria (temperature control). Each of the remaining four criteria
higher d values. These are older studies (Garaj-Vrhovac et al., were satisfied by about half of the studies. The most common
1991; Zotti-Martelli et al., 2000; Kesari and Behari, 2009; deficiency was lack of positive controls, which are needed to
Shckorbatov et al., 2010; Karaca et al., 2012). Their results assess the proper functioning of an assay. Failure to satisfy other
are not consistent with subsequent studies on similar RoB criteria (lack of appropriate sham controls, blinded study
endpoints and may be in error. design and adequate dosimetry) would be fatal to the validity of
Figure 4B shows the distribution of d for the 30 experiments a study.
that showed statistically significant effects, which are consistently While failure to meet RoB criteria raises concerns about the
higher than those in the full set of experiments in Figure 4A. In possibility of systematic errors, the converse is not true: the RoB
part this reflects a well-known tendency of null hypothesis criteria do not establish internal validity of a study. To reliably
significance testing to exaggerate effect sizes (Gelman and measure effects of the magnitude reported in most of these
Carlin 2014). This is a trivial effect of selecting “statistically studies, which were comparable to natural background
significant” results, which selects datasets with t (and hence d) variation, would require extraordinary measures to control
above a critical level (Eq. 2). With underpowered studies, as in the experimental errors. If such measures were taken in any of the
present case, this exaggeration can be quite large. In addition, presently considered studies they were not described in the
some of the studies may have appreciable systematic errors. papers. Some studies found quite large effects, but their
Most of the data in Figure 4A are consistent with no effect (or consistency with other studies and replicability would need to
at best small effects relative to the natural background variability be considered.
of the endpoints) of exposure. A more detailed analysis
(systematic review or meta-analysis) with detailed evaluation
of the individual studies with respect to each endpoint is Other Criteria: Flexibility in Data Collection
clearly needed, but the sparse and very uneven quality of the and Statistical Analysis
presently considered literature limits what can be concluded from Simmons et al. (2011) attributed a major cause of nonreplicable
such an analysis. Karipidis et al. (2021) found “no confirmed science to “flexibility in data collection and analysis” which
evidence” for genotoxic or other hazardous effects of RFR > “allows presenting anything as significant.” This refers to
6 GHz. investigator degrees of freedom in arranging the conduct of a

Frontiers in Communications and Networks | www.frontiersin.org 5 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

for chemical toxicity testing [Organisation for Economic Co-

operation and Development (OECD) 2014; Organisation for
Economic Co-operation and Development (OECD) (2016) for
genetic toxicology studies] include extensive precautions related
to study design and evaluation and reporting of results to reduce
this flexibility. None of the 31 studies considered here appear to
have been compliant with OECD guidelines (although some
authors did follow OECD guidelines in the number of animals
used and/or in the number of cells examined).
A final consideration is the high rate of false discovery due to
naïve use of null hypothesis significance testing (NHST), a
problem that has been pointed out many times by statisticians
but remains the default approach to analyzing many experimental
studies. [For an extensive recent review see Colling and Szucs,
2021]. Gelman (2018) noted:

“Null hypothesis significance testing (NHST) only

works when you have enough accuracy that you can
confidently reject the null hypothesis. You get this
accuracy from a large sample of measurements with
low bias and low variance. But you also need a large
effect size. Or, at least, a large effect size, compared to
the accuracy of your experiment.”

In the presently considered collection of studies, those

conditions are clearly not satisfied.
Needless to say, NHST with p < 0.05 is the default statistical
approach used in virtual all RF bioeffects studies to identify
“effects” of exposure. It is profoundly misleading to
retrospectively focus on statistically significant results in a
collection of studies without concern for nonsignificant results
that were also reported, study validity, and the size and biological
significance of reported effects.

DISCUSSION AND CONCLUSION

FIGURE 4 | (A) Distribution of effect sizes (Cohen’s d) across 157
experiments reported in 31 studies, including both statistically significant and Because of their small size and other limitations, many of the 31
not significant changes. One outlier with d
63.7 (Kesari and Behari, 2009) is
off scale in the figure. (B) Distribution of effect sizes (Cohen’s d) across
presently considered studies can best be described as pilot studies.
30 experiments reported in 31 studies. All of these experiments reported Because of the public interest in possible health hazards of 5G NR
statistically significant effects of exposure. One outlier with d
63.7 (Kesari communications and the paucity of quality bioeffects studies at
and Behari, 2009) is off scale in the figure. The increases in d compared to (A) frequencies >6 GHz, further studies are warranted (but not
arises in part from the exaggeration in effect size due to NHST; the possibility
necessarily a full range of studies).
of experimental errors cannot be excluded.
In its 2019 review of the bioeffects literature related to possible
health effects of 5G NR technology in three bands, the French agency
Agence nationale de sécurité sanitaire de l’alimentation and de
study, selecting data to present and analyze, choosing which l’environnement et du travail (ANSES), 2019 (Agence nationale
comparisons to make, etc. A simple (but unethical) example de sécurité sanitaire de l’alimentation, de l’environnement et du
would be to disregard data as they are being collected that appear travail) concluded that (in English translation) “the data are not
inconsistent with what the investigator believes are reasonable sufficient to conclude on the existence or not of health effects related
results, without a formal procedure for managing erroneous data. to exposure to electromagnetic fields in the band of frequencies
These authors recommended reducing this flexibility by: “list around 26 GHz”. ANSES offered a laundry list of suggested studies
[ing] all variables collected in a study . . . and report [ing] all in this band, emphasizing studies on the skin, in vitro genotoxicity
experimental conditions, including failed manipulations. . . .. . .. studies, possible behavioral and neurophysiological effects, all to be
If observations are eliminated, authors must also report what the done with “rigorous quality methods”.
statistical results are if those observations are included.” None of For such studies, major improvements in quality are needed
the 31 papers described such precautions. The OECD protocols relative to the presently considered set of studies. These include:

Frontiers in Communications and Networks | www.frontiersin.org 6 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

1. Better exposure assessment, which is a difficult problem at the provide solid, scientific data on whether ELF-EMF
frequency range >6 GHz due to the small penetration depth of exposure represents a human health hazard, and if so,
RFR into tissue at those frequencies. whether risks are increased under exposure conditions in
2. Larger studies with reasonable statistical power the general population.” Given the clear need of the public
3. Stronger study design with due attention to the RoB criteria, for reliable information about 5G NR telecommunications
see also Zeni and Scarfi (2012) and Vijayalaxmi (2016). technologies and the difficult and only partly solved problem
4. Use of currently accepted best practices to reduce effects of of assessing RFR exposure to the user of a handset and from
investigator degrees of freedom, e.g., “publish [ing] pre-study base stations, a program such as EMF-RAPID would be a
power calculations and effect sizes, including negative promising approach.
findings. Hypothesis-testing studies should be pre-registered Short of such large programs, funding agencies and journal
and optimally raw data published.” (Szucs and Ioannidis, editors can increase the reliability of the bioeffects literature by
2017). supporting quality studies with adequate funding, and raising
acceptance standards for bioeffects papers - a “carrot and stick”
INTERPHONE is one potential model for such studies approach (Vijayalaxmi and Foster, 2021).
(Cardis, et al., 2007). This set of coordinated epidemiological Unfortunately, after many years of debate about biological
studies in several countries was funded jointly by industry and effects of RFR, health agencies are clearly losing enthusiasm for
government, and was designed to address public concerns about such a program and they may have to make do with a series of
possible links between use of mobile phones and brain cancer. At smaller investigator-generated studies, hopefully of better quality
present, such controversies have not yet developed with respect to than presently available.
high-band 5G NR handsets (few if any are presently on the
market in any event). However, the model for this program, with
joint industry-government funding but governance free of DATA AVAILABILITY STATEMENT
industry influence, might be useful for more diverse studies on
mm-wave bioeffects such as recommended by ANSES. The original contributions presented in the study are included in
The EMF-RAPID Program is perhaps a more useful model. the article/Supplementary Material, further inquiries can be
The program was set up by an act of [U.S.] Congress in 1992 to directed to the corresponding author.
study the potential health impacts of extremely low frequency
(ELF-EMF) fields from powerlines, responding to a growing
public controversy at that time [The National Academies of AUTHOR CONTRIBUTIONS
Sciences Engineering and Medicine (NASEM), 1999]. The 5-
year program had three basic components: 1) a research program All authors listed have made a substantial, direct, and intellectual
of bioeffects research on a range of endpoints; 2) information contribution to the work and approved it for publication.
compilation and public outreach and 3) a health assessment
for evaluation of any potential hazards arising from exposure
to ELF-EMF. In addition it sponsored extensive surveys of SUPPLEMENTARY MATERIAL
population exposures to ELF-EMF from various sources. The
program had multiple levels of oversight with government The Supplementary Material for this article can be found online at:
officials as well as representatives of public interest groups. https://www.frontiersin.org/articles/10.3389/frcmn.2021.721925/
Studies were selected for support “for their potential to full#supplementary-material

Garaj-Vrhovac, V., Horvat, D., and Koren, Z. (1991). The Relationship between
REFERENCES colony-forming Ability, Chromosome Aberrations and Incidence of Micronuclei
in V79 Chinese Hamster Cells Exposed to Microwave Radiation. Mutat. Res. Lett.
Agence nationale de sécurité sanitaire de l’alimentationde l’environnement et du 263 (3), 143–149. doi:10.1016/0165-7992(91)90054-8
travail (ANSES) (2019). Expositions aux champs électromagnétiques liées au Gelman, A., and Carlin, J. (2014). Beyond Power Calculations. Perspect. Psychol.
déploiement de la technologie de communication « 5G » et effets sanitaires Sci. 9 (6), 641–651. doi:10.1177/1745691614551642
éventuels associés. Available at: https://www.vie-publique.fr/sites/default/files/ Gelman, A. (2018). The Failure of Null Hypothesis Significance Testing when
rapport/pdf/279567.pdf (Accessed July 20, 2021). Studying Incremental Changes, and what to Do about it. Pers Soc. Psychol. Bull.
Bodewein, L., Schmiedchen, K., Dechent, D., Stunder, D., Graefrath, D., Winter, L., 44 (1), 16–23. doi:10.1177/0146167217729162
et al. (2019). Systematic Review on the Biological Effects of Electric, Magnetic IARC Working Group on the Evaluation of Carcinogenic Risks to Humans (2013).
and Electromagnetic fields in the Intermediate Frequency Range (300 Hz to 1 Non-ionizing Radiation, Part 2: Radiofrequency Electromagnetic Energy. IARC
MHz). Environ. Res. 171, 247–259. doi:10.1016/j.envres.2019.01.015 Monogr. Eval. Carcinog. Risks Humans 102 (PT 2), 1. doi:10.1191/
Cardis, E., Richardson, L., Deltour, I., Armstrong, B., Feychting, M., Johansen, C., 0960327103ht394oa
et al. (2007). The INTERPHONE Study: Design, Epidemiological Methods, and Institute of Electrical and Electronics Engineers (IEEE) (2019). Standard for Safety
Description of the Study Population. Eur. J. Epidemiol. 22 (9), 647–664. Levels with Respect to Human Exposure to Radio Frequency Electromagnetic
doi:10.1007/s10654-007-9152-z fields, 0 Hz to 300 GHz, IEEE C95.1-2019. Piscataway NJ: IEEE.
Colling, L. J., and Szűcs, D. (2021). Statistical Inference and the Replication Crisis. International Commission on Non-Ionizing Radiation Protection (ICNIRP)
Rev.Phil.Psych. 12 (1), 121–147. doi:10.1007/s13164-018-0421-4 (2020). Guidelines for Limiting Exposure to Electromagnetic Energy (100

Frontiers in Communications and Networks | www.frontiersin.org 7 August 2021 | Volume 2 | Article 721925
Foster and Vijayalaxmi Reliability of 5G Bioeffects Studies

kHz to 300 GHz). Health Phys. 118 (5), 483–524. doi:10.1097/ The National Academies of Sciences Engineering and Medicine (NASEM) (1999).
HP.0000000000001210 Research on Power-Frequency Fields Completed under the Energy Policy Act of
Karaca, E., Durmaz, B., Altug, H., Yildiz, T., Guducu, C., Irgi, M., et al. (2012). The 1992. Washington, DC: National Research Council.
Genotoxic Effect of Radiofrequency Waves on Mouse Brain. J. Neurooncol. 106 Verschaeve, L., Juutilainen, J., Lagroye, I., Miyakoshi, J., Saunders, R., De Seze, R.,
(1), 53–58. doi:10.1007/s11060-011-0644-z et al. (2010). In Vitro and In Vivo Genotoxicity of Radiofrequency fields. Mutat.
Karipidis, K., Mate, R., Urban, D., Tinker, R., and Wood, A. (2021). 5G mobile Res. Rev. Mutat. Res. 705 (3), 252–268. doi:10.1016/j.mrrev.2010.10.001
Networks and Health-A State-Of-The-Science Review of the Research into Vijayalaxmi (2016). Biological and Health Effects of Radiofrequency fields: Good
Low-Level RF fields above 6 GHz. J. Expo. Sci. Environ. Epidemiol. 31, 585–605. Study Design and Quality Publications. Mutat. Res. Genetic Toxicol. Environ.
doi:10.1038/s41370-021-00297-6 Mutagen. 810, 6–12. doi:10.1016/j.mrgentox.2016.09.007
Kesari, K. K., and Behari, J. (2009). Fifty-gigahertz Microwave Exposure Effect of Vijayalaxmi and Foster, K. R. (2021). Improving the Quality of Radiofrequency Bioeffects
Radiations on Rat Brain. Appl. Biochem. Biotechnol. 158, 126–139. doi:10.1007/ Research: the Need for a Carrot and a Stick. Rad. Res. 196. (, 2021 In press).
s12010-008-8469-8 Vijayalaxmi and Prihoda, T. J. (2019). Comprehensive Review of Quality of Publications
Lai, H. (2021). Genetic Effects of Non-ionizing Electromagnetic fields. and Meta-Analysis of Genetic Damage in Mammalian Cells Exposed to Non-
Electromagn. Biol. Med. 40, 264–273. doi:10.1080/15368378.2021.1881866 ionizing Radiofrequency Energy. Rad. Res. 191 (1), 20–30. doi:10.1667/RR15117.1
Leszczynski, D. (2020). Physiological Effects of Millimeter-Waves on Skin and Skin Vijayalaxmi and Prihoda, T. J. (2012). Genetic Damage in Human Cells Exposed to
Cells: an Overview of the To-Date Published Studies. Rev. Env. Health 35 (4), Non-ionizing Radiofrequency fields: A Meta-Analysis of the Data from 88
493–515. doi:10.1515/reveh-2020-0056 Publications (1990-2011). Mutat. Res. Genetic Toxicol. Environ. Mutagen. 749
National Toxicology Program Office of Health Assessment and Translation (1-2), 1–16. doi:10.1016/j.mrgentox.2012.09.007
(OHAT) (2015). OHAT Risk of Bias Rating Tool for Human and Animal Vijayalaxmi and Prihoda, T. J. (2008). Genetic Damage in Mammalian Somatic
Studies. Washington, DC: US Department of Health and Human Services. Cells Exposed to Radiofrequency Radiation: A Meta-Analysis of Data from 63
Organisation for Economic Co-operation and Development (OECD) (2016). Publications (1990-2005). Radiat. Res. 169 (5), 561–574. doi:10.1667/rr0987.1
OECD Guideline for the Testing of Chemicals. Test Guideline 489: In Vivo Wood, A., Mate, R., and Karipidis, K. (2021). Meta-analysis of In Vitro and In Vivo
Mammalian Alkaline Comet Assay. Available at: https://www.oecd.org/env/ Studies of the Biological Effects of Low-Level Millimetre Waves. J. Expo. Sci.
test-no-489-in-vivo-mammalian-alkaline-comet-assay-9789264264885-en. Environ. Epidemiol. 31, 606–613. doi:10.11038/s41370-021-00307-710.1038/
htm (Accessed July 20, 2021). s41370-021-00307-7
Organisation for Economic Co-operation and Development (OECD) (2014). Zeni, O., and Scarfì, M. R. (2012). Experimental Requirements for in Vitro Studies
Organization for Economic Co-operation and Development. Guidance Aimed to Evaluate the Biological Effects of Radiofrequency radiationMicrowave
Document on Revisions to OECD Genetic Toxicology Test Guidelines. Materials Characterization. Rijeka, Croatia: InTech, 121–138.
In vivo Mammalian erythrocyte Micronucleus Test. Available at: http:// Zotti-Martelli, L., Peccatori, M., Scarpato, R., and Migliore, L. (2000). Induction of
www.oecd.org//GeneticToxicologyGuidance/2014 (Accessed July 20, 2021). Micronuclei in Human Lymphocytes Exposed In Vitro to Microwave
Ruediger, H. W. (2009). Genotoxic Effects of Radiofrequency Electromagnetic Radiation. Mutat. Res. Genetic Tox. Env. Mutagen. 472 (1-2), 51–58.
fields. Pathophysiology 16 (2-3), 89–102. doi:10.1016/j.pathophys.2008.11.004 doi:10.1016/s1383-5718(00)00112-1
Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR)
(2015). Potential Health Effects of Exposure to Electromagnetic fields (EMF). Conflict of Interest: KF has received minor support for research on an unrelated
Brussels: European Commission. doi:10.2772/75635 topic (thermal dosimetry) by Microwave and Wireless Forum, an industry group.
Shckorbatov, Y. G., Pasiuga, V. N., Goncharuk, E. I., Petrenko, T. P., Grabina, V. A.,
Kolchigin, N. N., et al. (2010). Effects of Differently Polarized Microwave The remaining author declares that the research was conducted in the absence of
Radiation on the Microscopic Structure of the Nuclei in Human Fibroblasts. any commercial or financial relationships that could be construed as a potential
J. Zhejiang Univ. Sci. B 11, 801–805. doi:10.1631/jzus.b1000051 conflict of interest.
Simkó, M., and Mattsson, M-O. (2019). 5G Wireless Communication and Health
Effects-A Pragmatic Review Based on Available Studies Regarding 6 to 100 Publisher’s Note: All claims expressed in this article are solely those of the authors
GHz. Int. J. Environ. Res. Public Health 16 (18), 3406. doi:10.3390/ and do not necessarily represent those of their affiliated organizations, or those of
ijerph16183406 the publisher, the editors and the reviewers. Any product that may be evaluated in
Simmons, J. P., Nelson, L. D., and Simonsohn, U. (2011). False-Positive this article, or claim that may be made by its manufacturer, is not guaranteed or
Psychology. Psychol. Sci. 22 (11), 1359–1366. doi:10.1177/0956797611417632 endorsed by the publisher.
Swedish Radiation Safety Authority (SSM). Scientific Council on Electromagnetic
Energy. Recent Research on EMF and Health Risk. Fourteenth Report from Copyright © 2021 Foster and Vijayalaxmi. This is an open-access article distributed
SSM’s Scientific Council on Electromagnetic Energy (2019). 14. Available at: under the terms of the Creative Commons Attribution License (CC BY). The use,
www.stralsakerhetsmyndigheten.se (Accessed 04 11, 2021). distribution or reproduction in other forums is permitted, provided the original
Szucs, D., and Ioannidis, J. P. A. (2017). When Null Hypothesis Significance author(s) and the copyright owner(s) are credited and that the original publication
Testing Is Unsuitable for Research: a Reassessment. Front. Hum. Neurosci. 11, in this journal is cited, in accordance with accepted academic practice. No use,
390. doi:10.3389/fnhum.2017.00390 distribution or reproduction is permitted which does not comply with these terms.

Frontiers in Communications and Networks | www.frontiersin.org 8 August 2021 | Volume 2 | Article 721925