Various modern technologies are known to bear risks to the human health. In many cases, the society learns about these risks only once negative consequences have become obvious, for example in the cases of asbestos, DDT and, most recently, the broad-spectrum systemic herbicide Glyphosate.

In the area of drug development, there is a risk mitigation strategy enforced by governmental agencies regulating market access for new drugs or new therapeutic uses of existing drugs. This strategy typically includes a thorough analysis of safety risks investigated in laboratory animals using a set of guidelines under conditions of Good Laboratory Practice (GLP).

In the area of environmental influences on human health, however, the current approach seems to be different. For example, in the US, although the Federal Communications Commission (FCC) and FDA are jointly responsible for the regulation of wireless communication devices, the FDA does not require safety evaluations based on GLP or applied OECD standards. Current safety regulations are essentially engineering standards that do not take into account potential impact on human health or physiology other than short-term heating risks (https://www.fcc.gov/general/radio-frequency-safety-0).

This is quite concerning as there is a growing body of publications suggesting that the radio frequency radiation (RFR) may have harmful biologic or health effects from exposure to RFR at intensities too low to cause significant heating.

Thus, we set out to characterize the internal validity of the results of published studies using laboratory animals that evaluated the effects of whole-body RFR at the intensities and frequencies (2G-5G) relevant to the current and emerging use of cell phones. 

In a scoping review, we screened the literature collated from a source that is commonly cited in the public discussion of the potential harms of RFR (https://www.powerwatch.org.uk) to support the development of a search strategy and the definition of extraction terms for a subsequent systematic review. Further, this exploratory analysis has allowed us to formulate a research hypothesis: the internal validity of preclinical studies on whole-body effects of RFR does not enable policy making. 

Search and screening strategy

Studies of animal models were identified from the PowerWatch database of 525 peer-reviewed scientific papers (Section: Mobile and Cordless Phones) about electromagnetic fields (EMF). Articles included after the title screening underwent concurrent full-text screening for definitive inclusion.

Inclusion and exclusion criteria

Publications were included that: a) described biological effects on whole-body RFR exposure in rodent model systems published in English language, peer-reviewed journals. Excluded were studies describing the effects of RFR exposure on in vitro or ex vivo test systems), Reviews, conference presentations, slides, posters and articles for which no full-text version could be obtained were also excluded from the analysis.

Study quality and risk of bias assessment

Publications identified were assessed against a four-point study quality checklist comprising the following internal validity criteria: (1) random allocation to group, (2) blinded assessment of outcome, (3) sample size calculation, and (4) inclusion / exclusion criteria. We recorded the number of checklist items scored. 

Further, we evaluated reporting of a) compliance with animal welfare legislation, b) adherence to the ARRIVE guidelines, c) use of AAALAC accredited facilities, d) application of OECD or GLP standards, e) study of genotoxicity outcomes, f) study of carcinogenesis outcomes, and g) measurement of temperature/heating effects. 

Each reference was evaluated by two independent reviewers. Disagreements between reviewers were resolved by a third reviewer and a consensus was reached. 

The interpretation of the results provided in the PowerWatch database – that studies were positive (any effects from the radiation exposure, whether harmful or not) or negative (no effects) – was also recorded.

Results

There was a total of 81 studies that met inclusion criteria.  Of those, 60 studies reported effects (any) of RFR (according to the PowerWatch database) and we refer to them as “positive”.

The figure below presents numbers of “positive” and “negative” studies (% within corresponding category) that met none, 1, 2, 3, or all four interval validity criteria (LO, L1, L2, L3, L4, respectively).

Due to an exploratory nature of the analysis, we can only describe these results as suggesting greater internal validity in “negative” studies.

Out of all 81 studies, only one study met all Landis 4 criteria.

Out of 60 “positive” studies, only 2 studies met three out of four Landis criteria.

As summarized in the table below, while randomization (typically no method or details described) and blinding (typically limited to blinded outcome assessment) are mentioned in about 50% of reports, very few publications provided information on sample size calculation and/or inclusion/exclusion criteria.

	All studies	Studies with negative results	Studies with positive results
Total number of publications	81	21	60
Reports followed ARRIVE guidelines	0.0	0.0	0.0
Conducted in AAALAC accredited facilities	2.5	0.0	3.3
Compliance with international or national animal welfare legislation stated	34.6	42.9	31.7
Any use of randomization mentioned	55.6	66.7	51.7
Any use of blinding mentioned	46.9	57.1	43.3
Any data inclusion or exclusion criteria mentioned	11.1	14.3	10.0
Sample size justified	7.4	28.6	0.0
Any OECD or GLP standards applied	1.2	4.8	0.0
Genotoxicity outcomes used	8.6	0.0	11.7
Carcinogenesis outcomes used	7.4	23.8	1.7
Heating / temperature effects measured	19.8	33.3	15.0

Conclusions and Outlook

While we would like to emphasize the exploratory descriptive analysis that was applied in our pilot study, results summarized above suggest that the internal validity of currently available studies on whole-body effects of radiofrequency radiation (technologies used in the 2G/3G cell phones) is too low to enable any policy making.

Together with our colleagues from QED Biomedical (https://www.qedbiomedical.com/), we have prepared a protocol for a systematic review that will be preregistered in a publicly accessible repository shortly (and this article will be updated with a link).