Modified Health Effects of Non-ionizing Electromagnetic Radiation Combined with Other Agents Reported in the Biomedical Literature

Abstract

Ionizing and non-ionizing electromagnetic field (EMF) radiation, either stand-alone or in combination with other agents, exert health effects on biological systems. The present chapter examines the scope of non-ionizing EMF radiation combined effects; i.e., identifies effects on biological systems from combined exposure to non-ionizing electromagnetic fields/radiation and at least one other agent. Only articles in which the presence of non-ionizing EMF radiation had some effect (beneficial or adverse) on the biological system were selected. A comprehensive and novel query was developed using an iterative hybrid approach, whereby articles related by common text and by citation linkages were retrieved. This retrieved literature was: (1) clustered algorithmically into 32 biomedical sub-themes (assigned by the authors); (2) grouped through factor analysis into 32 factors; and (3) subsequently grouped manually (by the authors) into an effects-based taxonomy. The common principles within each thematic cluster/group that accounted for the combined effects were identified.

Non-ionizing EMF radiation plays a supportive role in a wide range of beneficial and adverse effects. Major beneficial effects include (1) accelerated healing of wounds and injuries in concert with other agents and (2) treatment of cancer by combining chemotherapy with radiation. Major adverse effects, on the other hand, include (1) enhanced carcinogenesis, (2) enhanced cellular or genetic mutations, and (3) teratogenicity. It should be noted that community consensus (unanimity among papers published in peer-reviewed journals) does not exist on these potential effects, either beneficial or adverse, although there is substantial credible scientific evidence supporting the above effects (as described in this chapter).

In daily living, the body is exposed to multiple external agents simultaneously, e.g., myriad non-ionizing EMF radiations, pesticides, food additives, heavy metal, legal and illegal drugs, ionizing radiation, and air pollution. The number of combinations of potential external agents is large. Each combination could potentially have synergistic or antagonistic, and beneficial or adverse, effects. However, non-ionizing EMF radiation exposure safety standards are based primarily on stand-alone radiation exposures. When combined with other agents, the adverse effects of non-ionizing EMF radiation on biological systems may be more severe. Much work remains to be done before definitive statements about non-ionizing EMF radiation exposure safety can be made.

Appendices

Appendices

4.1.1 A.1 Appendix 1 – Text Mining Query Used for Information Retrieval

Appendix 1 contains: (a) brief overview of text mining; (b) brief summary of past EMF co-promoter studies; (c) text mining methodology used to identify, retrieve, extract, and analyze the relevant EMF co-promoter articles from the premier biomedical literature; (d) details of the iterative relevance feedback component of the query used, and of the subsequent citation network traversing component of the query.

4.1.1.1 A.1.1 Text Mining Overview

Text mining is the extraction of useful information from large volumes of text (Hearst 1999; Feldman et al. 1998; Kostoff et al. 2001a). Its component capabilities of computational linguistics and information retrieval were the main analytical techniques used in the present chapter. A typical text mining study of the published literature involves the development of a query for comprehensive information retrieval, an analysis of the database using computational linguistics and bibliometrics, and an integration of the processed information.

Computational linguistics identifies the main technical/medical themes of the database(s) being examined as well as the relationships among these themes. Computational linguistics has been used to enhance information retrieval and increase awareness of the global technical literature (Kostoff et al. 1997; Greengrass 1997), as well as to track the impact of a specific research area across time and applications areas (Davidse and VanRaan 1997; Kostoff et al. 2001b). Upgraded versions of these techniques were used for enhanced text-based and citation-based information retrieval in the present chapter.

Computational linguistics has been used in three modes to identify potential innovation and discovery: (a) co-occurrence of disease/technical problem and potential treatment/problem solution in the same article for identifying innovation (Kostoff and Briggs 2008; Kostoff et al. 2008b, c, d; Kostoff 2012); (b) linking disease/technical problem and disparate potential treatment/problem solution literatures directly (Kostoff and Briggs 2008; Kostoff et al. 2008a, b, c, d, e; Kostoff 2008, 2011, 2012; Swanson 1986; Swanson et al. 2001; Kostoff and Patel 2015; Kostoff and Los 2013); linking disease/technical problem and disparate potential treatment/problem solution literatures indirectly (Kostoff and Briggs 2008; Kostoff et al. 2008b, c; d). While these three modes each have their own distinctive features, they share the common requirement for an ‘intelligent’ query that targets documents with specific characteristics while filtering out ‘noise’. That was also the main requirement for the present chapter, and the experiences from the above references were directly applicable to the requirements of the present chapter. The focus of the present chapter was on co-occurrence of two or more promoters/stimuli in the same article (mode a), but modes b) and c) were used informally to help further understand the mechanisms linking the promoters/stimuli to the health impacts. Modes b) and c) could be used to start with non-ionizing EMF radiation promoters/stimuli or combinations ab initio, and identify direct and indirect linkages/mechanisms to potential health impacts, beneficial or adverse.

4.1.1.2 A.1.2 EMF Co-promoter Studies

Most non-ionizing EMF radiation health/biological impact studies focus on one non-ionizing EMF radiation form, with no identifiable co-promoters/stimuli. The initial comprehensive non-ionizing EMF radiation retrieval performed by the first author used a text-based query to search the Medline/Science Citation Index databases for a wide range of EMF radiation records, and retrieved over 6000 documents. The present chapter used a more intense hybrid text-based and citation-based query to search for non-ionizing EMF radiation combinations and their effects, and retrieved under 500 records. Thus, perhaps 5–7% of non-ionizing EMF radiation health/biological impact studies (probably less) are concerned with combination-type effects.

There are relatively few comprehensive review articles covering the combination effects. Juutilainen (2008; Juutilainen et al. 2006) examines cocarcinogens. Stam (2010) examines non-ionizing EMF radiation effects on the blood-brain barrier; this study is typical of reviews that may examine non-ionizing EMF radiation effects on an organ or system. Lee et al. examine the carginogenic potential of 60 Hz ELF-MF (1 mT) alone or in combination with ionizing radiation (IR), hydrogen peroxide (H₂O₂), or c-Myc overexpression (Lee et al. 2012).

Most of these previous studies that focused on combination effects of non-ionizing EMF radiation with other stimuli are like the articles referenced in the body of this chapter, namely, selected non-ionizing EMF radiation ranges in the frequency spectrum combined with usually one or a few co-stimuli. However, none of these review articles covers the wide range of disciplines as the present chapter, as well as both adverse and beneficial combined effects . It is the scope of coverage and the comprehensiveness of retrieval using the hybrid search technique, as well as the integration across disciplines, which makes the present chapter unique.

4.1.1.3 A.1.3 Methodology: Document Identification, Retrieval, Extraction, Analysis

4.1.1.3.1 A.1.3.1 Definitions of Combined/Interactive Effects

The goal of this chapter is to examine the scope of the non-ionizing EMF radiation combined effects on biological systems; i.e., identify effects on biological systems from combined exposure to non-ionizing electromagnetic fields /radiation and at least one other agent. These interactive effects include:

• A.1.3.1.1 Additive effects (the combined effect of two or more agents acting in the same general direction approximates the sum of the effects of the agents administered separately, subject to the maximum possible effects in biological systems);

• A.1.3.1.2 Antagonistic effects (the combined effects of two agents acting in different/opposite directions are smaller than the effect of any one of them in stand-alone mode);

• A.1.3.1.3 Potentiative effects (the increased effect of an agent by concurrent action of another agent that does not have a stand-alone effect); and

• A.1.3.1.4 Synergistic effects (the combined effect of two or more agents is significantly greater than the sum of the effects of each agent administered alone, subject to the maximum possible effects in biological systems).

Other terminology is used in the documents, such as co-promotional , co-mutagenic , co-carcinogenic , etc., but these terms tend to be sub-sets of the more general terms defined above.

The approach in this chapter is to: (1) select the most credible global databases of research articles; (2) develop a query that will retrieve the relevant combined effects literature comprehensively; (3) identify the key biomedical thrusts in this retrieved literature; and (4) extract the mechanisms and principles that describe the influence of the non-ionizing EMF radiation component on the final combined effects . These four approach components are now described in more detail.

4.1.1.3.2 A.1.3.2 Select Global Databases

The two premier biomedical research article databases are the Web of Science (WOS-Science Citation Index/Social Science Citation Index/Arts and Humanities Citation Index-SCI/SSCI/A&HCI) and Medline. Each has its unique strengths. WOS has the capability for citation linkages (references, citing papers, papers that share at least one reference), while Medline has a unique taxonomy/keyword structure called MeSH. Both databases were used in this chapter for query development.

4.1.1.3.3 A.1.3.3 Develop Retrieval Query

The first step in query development is to define the scope of the study topic. The scope selected was effects on biological systems from combined exposure to non-ionizing EMF radiation and at least one other agent. A novel hybrid iterative relevance feedback technique (based on Kostoff et al. 1997) was used to develop the query. The non-ionizing EMF radiation component of the full query (based on Medline and the SCI) used by the first author for the comprehensive non-ionizing EMF radiation retrievals was intersected with the terms “synerg* and ‘combined effect *’” to form an initial test query (see A.1.3.6.1. for complete initial test query). This initial query was inserted into the SCI search engine, further filtering was performed by restricting to biomedical Subject Areas (each SCI record has one or more Subject Area keywords assigned to it), and about eighty records were retrieved. These were termed the core records. The query was then expanded by examining the local citation network for each of the core records.

First, all the records that cited the core records were retrieved (~700), and were examined manually to select relevant non-duplicative records (~90). These were termed the core citing records. Second, all the references to the core records and core citing records that were in the SCI were examined manually to select relevant non-duplicative records (~150). These were termed the core record references and core citing record references. Third, some of the records that shared references with the core records and core citing records were examined, as follows.

The SCI has a feature called Related Records. For a record of interest, the Related Records feature will display all records in the SCI database that share at least one reference in common with the record of interest. These Related Records can be ordered by the number of references in common. The numbers of Related Records can range from zero (all the references in the record of interest were cited only once) to hundreds of thousands for a large block of highly cited references. Typically, the numbers of Related Records for a record of interest range from hundreds to tens of thousands. In practice, only a few of the Related Records for each record of interest can be examined for relevance, due to the large volumes involved.

The protocol used was to examine the twenty-forty records with the most shared references for each of the core records and core citing records, and extract those non-duplicative records deemed relevant (~75). These were termed the core records related records and the core citing records related records. Many of the Related Records tended to display repeatedly on the twenty-forty records with the most shared references, and the marginal utility of this approach decreased with time. This can be visualized as a well-connected network, where the same material is being accessed repeatedly. Like any network problem, the path to new information is less through complete-link type approaches and more through single-link type approaches. Unfortunately, the numbers of records with two or one shared references are large compared to the numbers of records with many shared references, and there were too many of these low shared reference records to examine manually.

Fourth, all the relevant records retrieved above were combined, and imported into the Vantage Point (VP) software (Vantage Point 2015). Text patterns in the Abstracts and Titles were examined, and were added to the initial test/text query. The additional query terms are shown in A.1.3.6.2. These terms were inserted into the SCI search engine, with further filtering done by Subject Area. About 500 records were retrieved and examined manually, and those deemed relevant (~135) were extracted. The total of about 530 records were then examined in detail, stricter criteria were applied for relevance, and 436 records with Abstracts were judged to be relevant. Obviously, the citation and text linkages could have been continued in an iterative manner, and more relevant records would have been found. However, the marginal utility of both approaches was beginning to decrease, especially for the citation linkages, and the manual selection approach was becoming infeasible.

4.1.1.3.4 A.1.3.4 Identify Key Biomedical Thrusts

The 436 retrieved records were inserted into the VP text mining software package and into the CLUTO document clustering software package (CLUTO 2015). A factor analysis was performed in VP using 32 factors, and a hierarchical taxonomy was generated in CLUTO using 32 clusters. Text mining was performed on each factor and cluster, to identify the key biomedical phrases representative of the group and the titles of papers in the group. Based on reading the titles and phrases, and reading of many paper Abstracts in each group as well, the theme of each group was identified.

4.1.1.3.5 A.1.3.5 Extract the Mechanisms and Principles that Describe the Influence of the Non-ionizing EMF Radiation Component on the Final Combined Effects

This is the key analytic step. Based on the two groupings identified by the clustering and factor analysis described above, and a reading of all the Abstracts, a final taxonomy was generated. The records were assigned manually to each taxonomy category. The non-ionizing EMF radiation-co-promoter-mechanism-disease ‘signatures’ were extracted by integrating the relevant factors, relevant clusters, and relevant sections of each record in the cluster. These relevant sections are displayed in the narrative section following the clustering summaries and taxonomy (Figs. 4.1 and 4.2).

The records were updated by repeating the above process for years 2012–2016, and adding the additional relevant records to the existing taxonomy categories.

4.1.1.3.6 A.1.3.6 Iterative Relevance Feedback Query

4.1.1.3.6.1 A.1.3.6.1 Initial Test Query

Topic=((EMF OR "Electromagnetic Field*" OR "Radio-Frequency Radiation" OR "Radio-Frequency Irradiation" OR "RF-Radiation" OR "RF -Irradiation" OR "Microwave Radiation" OR "Microwave Irradiation" OR "Mobile Phone*" OR "Cell* Phone*" OR "Wireless Phone*" OR "Cordless Phone*" OR "Mobile Telephone*" OR "Cellular Telephone*" OR "Wireless Telephone*" OR "Cordless Telephone*" OR "Base Station*" OR "RF-Transmission Tower*" OR "Cell Tower*" OR (("Magnetic Field*" OR "Electric Field*") AND ("Power Line*" OR "Low Frequency" OR "Power Frequency" OR "Intermediate Frequency" OR "Transmission Line*" OR "Electric Power Transmission"))) AND (synerg* OR "combined effect *")) AND Document Type=(Article OR Review)

Refined by: Subject Areas=(Biology OR Behavioral Sciences OR Biophysics OR Cell Biology OR Environmental Sciences OR Toxicology OR Public, Environmental & Occupational Health OR Orthopedics OR Radiology, Nuclear Medicine & Medical Imaging OR Health Care Sciences & Services OR Developmental Biology OR Pharmacology & Pharmacy OR Neurosciences OR Materials Science, Biomaterials OR Oncology OR Parasitology OR Biotechnology & Applied Microbiology OR Physiology OR Biochemical Research Methods OR Biochemistry & Molecular Biology OR Reproductive Biology OR Rheumatology OR Genetics & Heredity OR Surgery)

4.1.1.3.6.2 A.1.3.6.2 Additional Terms in Refined Query

Topic=(("magnetic field *" OR EMF * OR microwaves OR "microwave radiation " OR "microwave exposure*" OR "microwave irradiation" OR "electromagnetic field*" OR "RF radiation" OR ELF-MF* OR "mobile phone *" OR PEMF* OR EMR OR "Electromagnetic radiation " OR ELF OR "Radiofrequency field*" OR "radiofrequency radiation" OR ELF-EMF OR "cell* phone*" OR "electric power" OR "electromagnetic noise" OR IFC OR "pulsed magnetic field*" OR "CELL TOWER*" OR "static MF" OR "electric field*" OR electricity OR "electromagnetic EM field*" OR "electromagnetic radiation" OR "EM field exposure*" OR geomagnetic OR "power line*" OR electroporation OR electrofusion OR electrochemotherapy OR electropermeabilization))

AND ("combin* effect*" OR potentiat* OR synerg* OR "combin* exposure*" OR co-exposure OR "combin* treatment*")) AND Document Type=(Article OR Review)

Refined by: Subject Areas=(Biophysics OR Biology OR Neurosciences OR Biochemistry & Molecular Biology OR Radiology, Nuclear Medicine & Medical Imaging OR Toxicology OR Oncology OR Cell Biology OR Public, Environmental & Occupational Health OR Medicine, Research & Experimental OR Physiology OR Genetics & Heredity OR Microbiology OR Immunology OR Endocrinology & Metabolism OR Medicine, General & Internal OR Clinical Neurology OR Surgery OR Behavioral Sciences OR Developmental Biology OR Hematology OR Orthopedics OR Psychiatry OR Urology & Nephrology OR Dermatology OR Infectious Diseases OR Psychology, Experimental OR Cardiac & Cardiovascular Systems OR Gastroenterology & Hepatology OR Health Care Sciences & Services OR Neuroimaging OR Respiratory System OR Rheumatology OR Veterinary Sciences OR Virology OR Dentistry, Oral Surgery & Medicine OR Medical Informatics OR Nutrition & Dietetics OR Obstetrics & Gynecology OR Parasitology OR Pathology OR Pediatrics OR Peripheral Vascular Disease OR Psychology OR Psychology, Multidisciplinary OR Reproductive Biology)

4.1.2 A.2 Appendix 2 – Underreporting of Adverse Effects of EMF Radiation

Most of the papers in this chapter (~70%) focused on adverse health effects of non-ionizing EMF radiation combined with other agents. There is a growing literature providing evidence that adverse events in the biomedical literature are underreported; some of this literature is summarized in Chapter 9 of (Kostoff 2015). References 71–125 of (Kostoff 2015) are a modest sample of studies showing how adverse events in the biomedical literature have been, and are being, underreported. What are the reasons for, and consequences of, such underreporting of adverse events from non-ionizing EMF radiation?

4.1.2.1 A.2.1 Full Reporting of Adverse Events

A literature in which adverse events were fully reported would have the following characteristics:

• all critical research problems necessary for credible policy are addressed/ funded;

• all research performed is credible and high quality;

• all research findings are submitted for publication;

• all papers are reviewed by unbiased experts before publication;

• all high quality research submissions are published;

• all published articles are available to the general public;

• all accessible articles are easily retrieved.

4.1.2.2 A.2.2 Underreporting of Adverse Events

Adverse events are not reported fully in part because of the following reasons.

4.1.2.2.1 A.2.2.1 Critical Research Not Funded

Some critical research problems are not addressed/funded, for myriad reasons:

• the combinatorics parameter space is too complex to address all the areas;

• the funds available to the sponsor organization are insufficient to cover all critical research areas;

• the process for setting funding priorities within the sponsor organization is poor;

• selection based on potential for favorable results;

• external pressures effectively limit what topics can be funded, including

  • industry pressure to suppress topics that may have commercial sensitivity, and/or

  • government pressure to suppress topics that may have political sensitivity.

The pressures may operate intra-organizationally or inter-organizationally.

4.1.2.2.2 A.2.2.2 Research Not Submitted for Publication

Some research findings are not submitted for publication, for myriad reasons:

• national security classification, or classification for other reasons;

• organizationally proprietary;

• no organizational or individual publishing tradition, with equally little incentive to publish;

• costs associated with submissions for publication (time and money), which some organizations may not be willing to spend.

Most disturbing is the potentially deliberate suppression of research findings. This may result from:

• negative findings, which many organizations/journals/researchers are reluctant to publish;

• adverse events, which many industrial and governmental organizations in the biomedical community are reluctant to publish;

• commercial sensitivity, which industry would rather not be published;

• political sensitivity, which government would rather not be published; and,

• unethical research, whose performers would rather not be published, and whose quality may be relatively low due to lack of research oversight and lack of reproducibility.

4.1.2.2.3 A.2.2.3 Poor Research Published; Good Research Not Published

Some research that enters the literature may be of low quality, due to:

• poor peer review (where the peer review process and/or the peer reviewers are of low quality) or no peer review;

• contribution to the journal Editor’s pre-determined agenda.

Some high quality research may not get published, due to:

• poor peer review or biased peer review;

• lacking editorial expertise to judge the quality of the research;

• non-contribution to the journal Editor’s pre-determined agenda;

• not viewed as potentially contributing to increasing journal’s Impact Factor

4.1.2.2.4 A.2.2.4 Manufactured Research

Finally, some/much research that enters the published literature may be deliberately distorted or skewed; this research can be termed “manufactured research”. The purpose of this manufactured research is to both

• counter publications showing adverse effects from specific products and

• sow confusion among the public and decision-makers, not allowing the consensus required for policy.

The books Merchants of Doubt (Oreskes and Conway 2011) and Doubt is Their Product (Michaels 2008) describe this ‘research manufacturing’ process quite well. A few illustrative biomedical examples of some of the more egregious misrepresentations of science mentioned above, especially suppressed and manufactured research, are presented in Chapter 9 of (Kostoff 2015).

4.1.2.2.5 A.2.2.5 Published Research Not Easily Accessible

Some good published research may not be easily accessible to the public and the decision-makers, because of:

• publication in relatively obscure media or foreign languages only;

• publication behind high paywalls;

• poor search engines/algorithms.

4.1.2.2.6 A.2.2.6 Incentives for Inadequate Literature

There are many disincentives for an adequate literature, including.

• Industry: financial advantages for concealing the adverse effects of their products and services.

• Government: supporting corporate and large donor interests (through selective topic sponsorship and suppressed/distorted research findings) to lay the groundwork for future industry employment.

• Journal Editors: maintaining industry-sponsored professional society and/or advertising support through selective publication favorable to sponsors.

• Research performers: receiving and maintaining grants by working on topics of interest to, and producing results desired by, corporate and government sponsors; producing publications aligned with the interests of journal sponsors or advertisers in order to increase publication likelihood; laying the groundwork for future industry employment and/or consultancies by not publishing findings antithetical to the interests of industry.

In this section, we have identified the disincentives (for an adequate literature) for four classes: Industry, Government, Journal Editors, Research Performers. There are individuals who span multiple classes. For example, a person who works in government may also be a research performer and a journal Editor. The incentive (for an inadequate literature) associated with e.g. their government function may ‘spill over’ to their journal Editor and research performer roles. So, even though the journal may not have industry or government financial support as a source of potential bias, the potential biases arising from the government or research performer affiliations of the Editor could (in theory) influence the journal Editor role.

4.1.2.2.7 A.2.2.7 Policy Implications of Inadequate Literature

There are myriad important drivers of government policy; the three critical drivers of policy considered now will be technical literature, interests of political donors, and interests of the electorate.

Three options that relate policy to technical literature are:

• Option 1: The topical area is non-sensitive commercially or politically (e.g., weather satellite research, age of universe research). There is little incentive for much ‘manufactured research’ in these topical areas. Donors and voters would agree with, or be indifferent to, policy dictated by adequate literature; donors and voters agree with policy dictated by inadequate literature; policy reflects literature.

• Option 2: The topical area is sensitive commercially and/or politically (e.g., EMF health impacts). There is incentive for much ‘manufactured research’ in these topical areas, and our own studies have confirmed this. In this case, donors and voters would disagree with policy dictated by adequate literature. The donors are driven by profit, and the voters are addicted to the specific technology in this case (e.g., wireless communications). Thus, donors and voters agree with policy dictated by inadequate literature.

• In the case of EMF health impacts, the policy on EMF exposures that would be required as the result of an objective reading of the credible technical literature (severe restrictions on the use of wireless communications, etc) would not be acceptable to the vast majority of donors and voters. Thus, the policy in practice reflects the interests of the donors and voters, not the dictates of an adequate technical literature.

• Option 3: The topical area is sensitive commercially and/or politically (e.g., exposures/treatments that cause disease). There is incentive for much ‘manufactured research’ in this case, and our own studies have confirmed this. In this case, donors would disagree with policy dictated by an adequate literature, whereas the voters would agree with policy dictated by adequate literature. The donors are driven by profit, whereas the voters are driven by the benefits of technology in this specific case. Unlike the previous option, the voters are not addicted to the technology, since its application may be unpleasant in many cases. The donors still agree with policy dictated by inadequate literature, whereas the voters agree with policy dictated by inadequate literature, only because they believe it is adequate. This means that some literatures may be highly manufactured to maintain voter support. The policy reflects donors, not adequate technical literature.

In conclusion, the published technical literature is inadequate for myriad reasons, and the degree of inadequacy is unknown and may be unknowable. The fraction of inadequacy due to deliberate misinformation is unknown, but may be large for topical areas with commercial or political sensitivity.

4.1.2.2.8 A.2.2.8 Alleged Under- and Distorted Reporting of Non-ionizing EMF Radiation Adverse Effects

In the course of the present research, we have received numerous oral and written statements by non-ionizing EMF radiation researchers, and have seen many anecdotes reported, describing how adverse effects from non-ionizing EMF radiation have been suppressed as topics of research by the sponsoring agencies, and suppressed or distorted in the published research literature. This is not surprising; many government operations depend on the use of wireless communications, many industrial organizations are involved in producing wireless communications devices and systems, and many utilities involve high non-ionizing EMF radiation as part of their daily operations. If non-ionizing EMF radiation levels were required to be drastically reduced for safety purposes, the effect on our economies and many government operations would be devastating.

Two examples will be discussed in this section: alleged publication bias in a leading radiation journal (Slesin 2006) and the Bioinitiative Report (2012).

4.1.2.2.8.1 A.2.2.8.1 Alleged Journal Bias on Publishing Adverse Effects

As the reader will see from the following example, obtaining data to support and validate allegations of journal bias is extremely difficult. Dr. Louis Slesin has been publishing a newsletter addressing myriad issues related to microwave radiation , and it is aptly entitled ‘Microwave News’ (MN). We came across this newsletter during the course of our initial non-ionizing EMF radiation health impacts study (Kostoff and Lau 2013), and found the MN articles were quite accurate in the areas where they overlapped our study.

In 2006, MN published an article entitled “Radiation Research” and The Cult of Negative Results (Slesin 2006). It was a unique study with major contributions from Dr. Henry Lai, a leading researcher in the technical area of the article. The study’s focus was essentially to ascertain how reflective (of the microwave-induced genotoxicity publications in the larger technical literature) were those articles published in the journal Radiation Research on this topic.

In short, MN found that:

• [In the larger technical literature on microwave-induced genotoxicity] “There is just about an even split between effect and no-effect papers”;

• “A clear—and disconcerting—pattern emerges: 32 of the 35 studies that were paid for by the mobile phone industry and the U.S. Air Force show no effect. They make up more than 75% of all the negative studies. You don’t need to be a statistician to infer that money, more often than not, secures the desired scientific result”;

• “A similar loss of balance occurs when you look at only the papers published in Radiation Research.....Over the last 16 years, only one positive paper on microwave genotoxicity has appeared in Radiation Research. During the same time, the journal has published 21 negative genotox papers. (Australia’s Pam Sykes, the lead author of the lone positive paper, was denied money for a follow-up and soon moved on to other research areas.)....80% of the negative papers (17 out of 21) published in Radiation Research were paid for by either industry or the U.S. Air Force.” (Kostoff and Lau 2013).

At this point, the statements in MN are only allegations. There could be journal bias, or the best papers submitted to the journal happen to be the ones showing the absence of an impact of microwaves on genotoxicity. How could this issue be resolved?

One could (in theory) re-evaluate the original peer reviews of all the manuscripts submitted to the journal on this topic for bias. Unfortunately, we would then have the issue of determining the biases of the second group of reviewers, a difficult task. Additionally, even for reviewers who are unbiased, there is not always complete agreement. Scientists can sometimes have very differing opinions on the value of the same concept. Proving deliberate bias for a journal is extremely difficult, and may border on the impossible in practice.

Further, while funding source was used as an important metric to assess potential bias in the above study, it is by no means definitive. The determining factor is ‘intent’; were the researchers incentivized to bias their findings? If a researcher had no observable funding sources that could be interpreted as predisposing to potential bias, but wanted to obtain either (1) funding from industry or (2) future employment with industry, then one could argue the researcher had some predisposition to bias in favor of industry. How would we ever know this? Even if the researcher eventually received a grant from industry, or employment with industry, how would we know whether the researcher ‘intended’ to pursue these funding opportunities at the time the research was being performed and published?

Finally, how well this particular example reflects all, or any, other technical/ biomedical journals relative to the presence or absence of potential bias, is unknown.

4.1.2.2.8.2 A.2.2.8.2 The Bioinitiative Report

The Bioinitiative Report (2012) was written by a large number of world-class experts in science and public health policy in 2007, and updated in 2012. It assessed safety of non-ionizing EMF radiations at myriad frequencies. Its main message is that the present permissible non-ionizing EMF radiation exposure limits are orders of magnitude too high to protect against potential adverse health effects from non-ionizing EMF radiation. The Report recommends permissible chronic non-ionizing EMF radiation exposure limits of about one milligauss in the power frequency range (~60 Hz), and about ten microwatts/m² in the outdoor RF range. Contrast that with the ICNIRP (International Commission on Non-Ionizing Radiation Protection; an independent scientific organization) reference levels for general public exposure of about 800 milligauss (time varying magnetic field unperturbed rms value) in the 60 Hz power frequency range (ICNIRP n.d.), and the FCC (a regulatory agency) MPE power density limits of about 6.0 × 10⁶ microwatts/m² at ~900 Mhz cell phone RF frequencies (Federal Communications Commission Office of Engineering and Technology 1997). These are three to six orders of magnitude higher than recommended in the Bioinitiative Report.

Why is there such a difference, and how does this difference relate to the underreporting of adverse health effects from non-ionizing EMF radiation? The answer is partially stated and partially implied within Section 3 of the Bioinitiative Report (The Existing Public Exposure Standards), as follows:

Professional bodies from technical societies like IEEE and ICNIRP continue to support “thermal-only” guidelines routinely defend doing so a) by omitting or ignoring study results reporting bioeffects and adverse impacts to health and wellbeing from a very large body of peer-reviewed, published science because it is not yet “proof” according to their definitions; b) by defining the proof of “adverse effects” at an impossibly high a bar (scientific proof or causal evidence) so as to freeze action; c) by requiring a conclusive demonstration of both “adverse effect” and risk before admitting low-intensity effects should be taken into account; e) by ignoring low-intensity studies that report bioeffects and health impacts due to modulation; f) by conducting scientific reviews with panels heavily burdened with industry experts and under-represented by public health experts and independent scientists with relevant low-intensity research experience; g) by limiting public participation in standard-setting deliberations; and other techniques that maintain the status quo.

Much of the criticism of the existing standard-setting bodies comes because their contributions are perceived as industry-friendly (more aligned with technology investment and dissemination of new technologies) rather than public health oriented.

There are many published studies that bolster the above conclusions, and address the influence of industry funding on scientific results. For example (Huss et al. 2007): “We examined the methodologic quality and results of experimental studies investigating the effects of the type of radiofrequency radiation emitted by handheld cellular telephones. We hypothesized that studies would be less likely to show an effect of the exposure if funded by the telecommunications industry, which has a vested interest in portraying the use of mobile phones as safe. We found that the studies funded exclusively by industry were indeed substantially less likely to report statistically significant effects on a range of end points that may be relevant to health. Our findings add to the existing evidence that single-source sponsorship is associated with outcomes that favor the sponsors’ products....... Most previous studies of this issue were based on studies of the efficacy and cost-effectiveness of drug treatments. A recent systematic review and meta-analysis showed that studies sponsored by the pharmaceutical industry were approximately four times more likely to have outcomes favoring the sponsor’s drug than studies with other sources of funding......The influence of the tobacco industry on the research it funded has also been investigated......To our knowledge, this is the first study to examine this issue in the context of exposure to radiofrequency electromagnetic fields .”

Chapter 9 of reference (Kostoff 2015) shows this problem is endemic to many industries and government agencies. Given the difficulty of obtaining this type of information, much of which is provided by ‘whistleblowers’, we should not under-estimate the amount of distorted non-ionizing EMF radiation adverse effects findings published in the literature, nor under-estimate the amount of non-ionizing EMF radiation adverse effects findings suppressed from publication by government, industry, and the journals themselves.

To conclude, this sub-section has provided reasons for adverse health effects of myriad substances (including non-ionizing EMF radiation) being underreported in the premiere biomedical literature, or entering this literature in distorted form. Since there is no way to gauge the extent of this under/distorted-reporting, the quality and credibility of the ‘premiere’ biomedical/non-ionizing EMF radiation adverse effects literature is unknown. Therefore, any types of meta-analyses or scientometric analyses of this literature will have unknown quality and credibility. The most sophisticated scientometric analysis cannot compensate for a highly-flawed database.

Equally damaging is the effect of this flawed database on the larger scientific enterprise. Science can be viewed as a never-ending construction project, where the building blocks and support structures are the documents from past scientific studies. If some, or many, of these building blocks are flawed, the upper parts of the structure will be weakly supported, and may collapse. Through the citation process, the misleading findings at the lower parts of the structure are promulgated to the upper portions, and the dilution of quality increases. While the propensity for misconduct is greatest in areas of commercial and political sensitivity, the broad reach of basic science will have a ‘spill-over’ adverse impact on myriad directly and indirectly related areas of science.

Disclaimer

The views in this chapter are solely those of the authors, and do not necessarily represent the views of Georgia Institute of Technology or any of its components, or of the Institute for Defense Analyses.