Non-ionizing Radiation Protection: Summary of Research and Policy Options

List of Contributors

Michael Bangay

Michael Bangay Consulting

Melbourne

Australia

Rodney J. Croft

University of Wollongong

Illawarra Health & Medical Research Institute

Wollongong

Australia

Anna Dalecki

University of Wollongong

Illawarra Health & Medical Research Institute

Wollongong

Australia

Michael Dolan

Australian Legal Practitioner

Melbourne

Australia

Thanh Dovan

SP AusNet

Melbourne

Australia

Mark Elwood

Epidemiology and Biostatistics

University of Auckland

School of Population Health

Auckland

New Zealand

Paul Flanagan

Aurecon

Neutral Bay

Australia

Kenneth R. Foster

Department of Bioengineering

University of Pennsylvania

Philadelphia

Pennsylvania

USA

Peter Gies

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

Sue Heward

SunSmart

Cancer Council Victoria

Melbourne

Australia

Bruce Hocking

Specialist in Occupational Medicine

Camberwell

Victoria

Australia

Steve Iskra

Department of Health and Medical Sciences

Faculty of Health, Arts and Design

Swinburne University of Technology

Hawthorn

Victoria

Australia

Telstra Corporation

Melbourne

Australia

John Javorniczky

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

Ken Karipidis

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

Sarah P. Loughran

University of Wollongong

Illawarra Health & Medical Research Institute

Wollongong

Australia

Claire Lyngå

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

Robert L. McIntosh

Department of Health and Medical Sciences

Faculty of Health, Arts and Design

Swinburne University of Technology

Hawthorn

Victoria

Australia

Telstra Corporation

Melbourne

Australia

Ray McKenzie

Faculty of Health, Arts and Design

Swinburne University of Technology

Hawthorn

Victoria

Australia

Garry Melik

Magshield Products (Aust.) International

Melbourne

Australia

Stephen Newbery

Radiation Protection Unit

Department of Health and Human Services

Hobart

Australia

Kevin Nuttall

Energex Limited

Newstead

Queensland

Australia

Colin Roy

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

Rick Tinker

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

David Urban

Australian Radiation Protection and Nuclear Safety Agency

Melbourne

Australia

Adam Verrender

University of Wollongong

Illawarra Health & Medical Research Institute

Wollongong

Australia

Andrew Wood

Department of Health and Medical Sciences

Faculty of Health, Arts and Design

Swinburne University of Technology

Hawthorn

Victoria

Australia

Foreword

Nonionizing radiation, or NIR for short, is a type of radiation that is defined by what it does not, that is, it does not cause ionization of molecules. While this definition conveniently separates NIR from ionizing radiation emitted by radioactive substances, during fission and from certain equipment, it is somewhat unsatisfactory to use a negative definition and it is not even always correct (a certain part of the ultraviolet, UV, spectrum may cause ionization). Furthermore, it is not always radiation in a strict sense (e.g., static fields and ultrasound).

Just as we differentiate between different sources and types of ionizing radiation, we may be better off dealing with NIR on the basis of its specific characteristics, which is very different depending on which form of NIR we are considering. For example, radiofrequency radiation, microwaves, laser, and UV have their own very specific characteristics. This leads to different biological actions and responses and ultimately to different types of effects on the health of people and the environment. Understanding the nature of NIR, its biological actions, health effects, and associated risks is vital when deciding on the need for, and nature of, protective measures. Such protective measures also depend on whether the exposure category is the public, workers in their occupational setting, patients undergoing medical examination involving NIR, or the environment.

The real or potential health implications of NIR exposure for both people and environment is a legitimate concern for the community. Exposure to NIR is ubiquitous and exposure to certain forms of NIR has increased with the advent of technologies such as broadcasting and telecommunication. The ability to opt out of such exposures is sometimes limited; examples include outdoor work or other outdoor activities leading to UV exposure. The everyday environment in virtually all population centers and workplaces also involves exposure to radiofrequency radiation. At the same time, policymakers and the public in general need to take informed decisions to mitigate risks when they are evident and not invest resources in mitigation of risks that on the basis of current evidence are negligible. Unsubstantiated health concerns could itself cause symptoms of ill health and prevent the beneficial use of technologies involving NIR.

While health effects in both the short and long terms can be clearly attributed to certain forms of NIR exposure in what can be considered normal situations (exposure to UV radiation outdoors is one example), there are contrasting views in society as regards the health implications of everyday exposure to, for example, radiofrequency radiation, microwaves, and static fields. Decisions on limitation of exposure and precautionary approaches are often made under uncertainty; one major factor influencing the debate among the public as well as between specialists (whether these are in radiation science, risk communication, or ethics) is how uncertainty, or the unknown, should frame decisions on exposure limits and justification of technologies leading to exposure to NIR.

This volume explains and explores, based on scientific norms and methodologies, the characteristics of different forms of NIR, analyzes the relationship between exposure and biological effects and the associated dose–response relationships, and explores health effects and inferred and established health risks. It takes a holistic approach to the concept of health, building on the World Health Organization's definition of health: which is not only a state of absence of disease but includes also the physical, mental, and social well being of individuals and the population.

It finally addresses awareness, communication, and consultation, all of which are important factors in making it possible for any citizen to form an informed view and for society to take decisions based on the current state of knowledge – including uncertainties. This volume will assist in such judgments. I recommend it to everyone who wants to learn more about the different forms of NIR, the current knowledge on effects of NIR exposure on the health of people and the environment, and the evaluation and mitigation of risks associated with NIR in our everyday environments.

Carl-Magnus Larsson

CEO of ARPANSA

Acknowledgments

We are indebted to the following people who read individual chapters and provided comments: Dr Alireza Lajevardipour and Lydiawati Tjong.

Introduction

This is a book about appropriate ways to protect people (and perhaps the environment) against harmful effects of nonionizing radiation (NIR). NIR includes forms such as ultraviolet, visible light, infrared, microwaves, radio waves, and the electric and magnetic fields associated with electric power lines, magnetic resonance imaging (MRI) machines, and other electromagnetic technologies. There are many books about ionizing radiation (IR) protection because the link between X-rays, subatomic particles, and gamma radiation and serious illness such as cancer or in the case of high dose/high-dose rate death within days has been known about for over a century. NIR has always been viewed as a benign form of radiation, with MRI and ultrasound preferred over X-ray, CT, and PET modalities of imaging. Some radiation protection practitioners have labeled NIR as not interesting radiation because it seems that there is nothing very much to talk about in terms of dangers to human health. And yet, in many countries, the radiation source that is responsible for the largest numbers of morbidity and mortality is a NIR source, namely the sun. In other areas, the public outcry over the siting of mobile (cell) phone towers, electric transmission lines, and the roll out of Wi-Fi and smart metering services indicates that in the minds of many, NIR is not benign and is a potent and widespread source of illness, particularly cancer. Many have gone as far as labeling these technologies as the new tobacco smoking or asbestos that are established carcinogens. In addition, a section of the community attribute their being unwell to exposure from NIR sources and some have moved away from urban settings and have sought to shield their homes from man-made NIR fields in an attempt to alleviate symptoms.

The public are in general much more aware of NIR in their environment (since much of modern technology is based on electric power and electronics) than IR, which is perceived to be encountered only in specialist hospital departments (or nuclear power plants). The ubiquity of certain types of NIR coupled with the steady stream of media articles about their possible dangers to health have made sections of the community distrustful of authority reassurances and perplexed as to why there seem to be such differing views among scientists. This has been amplified by a number of legal challenges to planning approvals and personal injury cases on health grounds, which have tended to pit scientific expert witnesses against each other. Rather than trying to explore through public engagement an appropriate way to deal with scientific uncertainty, government agencies have sometimes bowed to community pressure by introducing unrealistically low exposure limits, which are not science based, believing this to be a precautionary approach. The availability of cheap monitoring instruments has also contributed to media coverage, with activists contacting journalists, having made do-it-yourself NIR measurements (often incorrectly). The principle of not in my backyard (NIBY) has often been a potent factor in these debates, with possible health effects used as a weapon against the true concerns: negative visual impacts and property devaluation.

Allied to this has been the question of who to go to for unbiased information and advice. The industries involved are best placed to devote resources for producing public information material, but face a significant challenge to appear credible. This having been said, there are now a range of national and international brochures and web-based materials to provide information on the nature of particular forms of NIR, the rationale for standards, summaries of relevant scientific investigations, and possibly also ways to reduce personal exposure. These have been produced by government agencies as well as the industry organizations involved. The fact that they tend to give very similar advice indicates, in general, a willingness of industry to tell it as it is.

The source of research funds has also been raised as a possible reason for the disparity of conclusions of scientists, with frequent claims that those who accept research funding from industry are tainted and thus unreliable. However, it should be acknowledged that those who believe there to be an unrecognized problem with low-level NIR exposure are also prone to selectivity when quoting earlier scientific studies. The quality of individual studies does vary enormously but is very hard to quantify. International agencies have tended to use the weight of evidence approach, in which relevant peer-reviewed studies are identified by bibliographic searches and then the outcomes compared for consistency and coherence. Isolated findings that lack replication or confirmation by independent teams of investigators tend not to be given great weight in this approach. However, finding consensus is not always easy and uncertainties remain, particularly where underlying mechanisms have not been identified.

This is not to deny that at sufficiently high intensities of NIR the health effects are immediate and serious: intense beams of ultraviolet and laser light cause tissue burning; radiofrequency (RF) fields at high-power levels can also cause excessive heating and extremely low-frequency (ELF) electric and magnetic fields can induce currents sufficient to cause alteration or cessation of heart or breathing rhythm if high enough. NIR standards are formulated to give a high margin of protection against established effects.

This book attempts to summarize the scientific findings regarding the safety of NIR, the rationale behind prevailing standards, the appropriate instrumentation to monitor this radiation, and the options for handling the associated issues in terms of policy and public information. The first chapter is an overview of the nonionizing portion of the electromagnetic spectrum, to describe the features of the way this energy can be propagated with associated electric and magnetic fields. Some of the NIR spectrum is not strictly radiated and this distinction will be made in this chapter. The remainder of the book is divided into nine sections as follows:

Part I deals with generic issues of how to identify hazard, both by studies in the laboratory (short-term and long-term) and by studying relevant human populations, by the methods of epidemiology. It covers the strengths and weaknesses of the experimental method for determining thresholds above which harmful effects are possible in humans. Those who are already familiar with these methods can skip parts of these chapters.

Part II covers aspects of appropriate protection against ultraviolet (UV) light. The most common source of UV exposure to humans is from the sun, which is an unregulated source. The modification of human behavior is the chief way to limit exposure, which may include obligations in the part of employers or business owners to implement these modifications.

Part III considers the visible part of the spectrum and infrared. Again, the sun is a potent source or radiation in this region, but lasers probably represent the greatest potential hazard, because of their high intensity. As well as coherent sources (lasers) a number of incoherent sources such as high-powered light-emitting diodes (LEDs) require consideration for possible eye or skin damage.

Part IV looks at the RF part of the spectrum (which includes microwaves and terahertz (THz) radiation). Although the portions of the spectrum used in telecommunications, broadcasting, and radar represent the most fully studied, lower RFs, used in welding, smelting, and heat-sealing operations, also need consideration.

Part V covers the ELF portion of the spectrum, which includes the electric and magnetic fields associated with the generation, transmission, distribution, and domestic use of electric power at 50 or 60 Hz. Although the most common, ELF fields are also associated with transportation systems, certain forms of welding and smelting are also involved.

Part VI is about static electric and magnetic fields: the former associated with high-voltage direct current (DC) transmission systems and the latter mainly with MRI machines in hospitals. Static electric fields are also encountered in the atmosphere (especially before and during thunderstorms), and the Earth has a familiar magnetic field.

Part VII moves on to community issues: these are of two types, firstly the nature of perceived hypersensitivity to electric technologies and secondly the types of policy options aimed at making proactive changes or limitations ahead of clear scientific conclusion of hazard at commonly encountered levels of exposure, the so-called cautionary approach (or Precautionary Principle). A chapter deals with examples of how to decide on whether or not to spend money on certain mitigation measures, based on cost-benefit analyses.

Part VIII covers the question of how to avoid injury (by occupational training or public awareness programs) and in the event of suspected NIR injury, how a medical assessment could be carried out.

Part IX includes what can be done to involve the public in decision-making in regard to exposure to NIR and what can be done to mitigate or attenuate the exposure at source. A chapter covers some of the public debates that have marked the development of NIR protection strategies in recent decades, with a concluding chapter to set NIR protection into perspective and to predict what may happen in the next decades.

A glance at the list of contributors on the next pages will reveal that all except two are from Australia. This is because the idea for the book arose from collaborations on developing standards and guideline documents for use in Australia. Since these were based, in the main, on international documents and inputs, the collaborators felt well placed to write a book for an international readership. We have sought wherever possible to avoid local references, but in places illustrative examples or practices are local, purely because they are most familiar to the authors.

The editors would like to express their gratitude to Dr Colin Roy, who until his retirement was one of the original editors of this book. We would also like to thank Dr Carl-Magnus Larsson for agreeing to provide a preface.

Many of the chapters have at the end of them a selection of tutorial problems, with answers given at the back of the book.

Chapter 1

Overview: The Electromagnetic Spectrum and Nonionizing Radiation

Andrew Wood¹ and Colin Roy²

¹Department of Health and Medical Sciences, Faculty of Health, Arts and Design, Swinburne University of Technology, Hawthorn, Victoria, Australia

²Australian Radiation Protection and Nuclear Safety Agency, Melbourne, Australia

1.1 What Is Nonionizing Radiation (NIR)?

By definition, nonionizing radiation (NIR) does not cause atoms and molecules to be ionized, that is, electrons are not removed from the atom or molecule leaving it with an electrical charge. Before describing the particular features of NIR, it is instructive to consider some of the general properties of electromagnetic radiation, which comprises both NIR and ionizing radiation.

Radiation can be thought of being both wave-like and particulate (this is often referred as the wave-particle duality). Ionization occurs when the energy in individual particles (or quanta) is sufficiently high to remove an electron, by transferring all of the energy of an individual quantum. Because of the wave-particle duality just referred to, each quantum can be associated with a particular wavelength. The wavelength of X-rays (a form of ionizing radiation) is approximately a nanometer (or a millionth of a millimeter), and other forms of ionizing radiation have wavelengths even shorter. NIR is regarded primarily as electromagnetic radiation whose wavelength is longer than 100 nm or 0.1 µm (see Figure 1.1). This is in the ultraviolet (or UV) part of the spectrum. To get this into perspective, a biological cell is around 10 µm in diameter and a single molecule of hemoglobin is 6 nm in diameter. Other forms of NIR have longer wavelengths, several thousands of kilometers in the case of waves associated with the domestic electricity supply. The wave itself is made up of two components, an electrical field (E-field) and a magnetic field (H-field), at right angles to each other and both of these quantities at right angles to the direction of propagation (see Figure 1.2). The wavelength is the physical distance between one peak and the next for either the E or the H field. The speed of propagation in vacuum is the same for all forms of electromagnetic radiation, whether ionizing or NIR, and is 300,000 km/second (or 3 × 10⁸ m/second). It should perhaps be remembered that in media (such as human tissue), the speed will be somewhat less than this value and will contribute to the phenomenon of refraction or deviation in the direction of propagation when going from one medium to another (e.g., air to tissue). This phenomenon is most familiar in the case of visible light (optical radiation), but applies to NIR generally. In a vacuum, the ratio of the magnitude of the E-field to that of the H-field has a fixed value for positions more than a few wavelengths from the generator. The fields are said to be coupled. When the wavelengths are of the order of kilometers, most positions of interest are much closer than one wavelength, and the fields are then said to be uncoupled.

Figure 1.1 The electromagnetic spectrum, from power frequencies through to γ-rays. Top: wavelength in meters; middle: relative sizes of wavelengths, names, and typical sources; bottom: frequency in waves per second or hertz (Hz) and the relative energy of each type. Source: K. Karipidis, ARPANSA, Australia.

Figure 1.2 A propagating electromagnetic wave, showing electric (E) and magnetic (H) vectors (arrows), the direction of propagation (k), and the wavelength (λ). Note that the E and H vectors are at right angles to each other and also to the direction of propagation. See diagram as supplied for location of all of these symbols (λ, E, k).

The fundamental unit of measurement of E-field is the volt per meter (or V/m) and of H-field is amperes per meter (A/m). The ratio E/H has units of resistance (ohms) and for a vacuum has the value 377 Ω, which is related to fundamental electrical constants. In a medium such as body tissue, the value reflects more complex interactions and is referred to as impedance. In fact, 377 Ω is usually referred to as the impedance of free space.

The E/H ratio is analogous to Ohms law (i.e., that electrical resistance is the ratio of voltage to current), so in the same way that the power dissipated by a resistor is the product of the voltage and the current (in W), the product of E × H is a measure of power density of the fields in watts per meter squared (W/m², see Figure 1.3). These measures have relevance throughout the spectrum of NIR, but at the longest wavelengths, the individual E and H field values are more important than power density and at the shortest wavelengths the opposite applies.

Figure 1.3 The relationship between power density and power. The sphere represents an expanding wavefront from the origin. Alternatively, it can represent an imaginary spherical surface across which the radiated power is flowing. Power density is expressed as power per unit area, so if the area considered is A in the diagram, proportion of the total power P crossing A will be P⋅A/(4πr²) watts, since 4πr² is the surface area of the entire sphere. Dividing by A gives the power density in W/m².

In general, the quantal nature of electromagnetic radiation is less important for NIR, but for the very shortest wavelengths of the NIR spectrum (UV and visible light), more quantal energy-specific modes of interaction (photoreactions) become very important. Unlike the ionizing radiation case, where an electron can leave a molecule, specific wavelengths of NIR can induce electron transitions to produce excited molecular states. Thus, as well as power density, the precise wavelengths of the radiation are very important in determining the precise biological effects which could follow exposure. At longer wavelengths (radiofrequencies or RF), the applied E and H fields cause ions and charged molecules to try to align with them. In the resulting motion, the friction between charged particles and surrounding molecules gives rise to an increase in temperature, the so-called thermal effect of RF. At longer wavelengths (longer than a few tens of meters), mobile charge carriers in living tissue, such as ions, contribute to induced currents, which may have direct influences on cellular function.

There are two types of NIR that do not strictly belong to the spectrum: static fields, because they are in no way radiated, and ultrasound, which is not an electromagnetic wave (it is a mechanical wave). Health effects evaluation for static fields are similar to that for fields in the extremely low-frequency (ELF) range, so this is included in this book. On the other hand, since ultrasound is, in some ways, a specialist area (and since outside of medical uses and vehicle parking technologies, ultrasonic beams are not commonly encountered), this area will not be covered.

1.2 Types of NIR

Table 1.1 summarizes the main features of the types of NIR dealt with in this book.

Table 1.1 Types of nonionizing radiation.

Note: 1 mm = 1000 and 1 µm = 1000 nm.

1.3 How Dangerous is NIR?

Since the nature of NIR is not to cause ionization, the damage to biomolecules such as strand breaks in DNA, normally associated with ionizing radiation, is not expected. There is a perception among the scientific community that most health issues from NIR are media generated, concerned as they are with sources that are part of the normal home or work environment. This contrasts with ionizing sources, which are limited to specialist environments such as hospital diagnostic and therapy units and uranium mining. Relatively few people are exposed on a regular basis to these sources (ignoring, for the moment, the background natural ionizing radiation which we are all exposed to continuously). The notion of the electromagnetic environment of home appliances, mobile phones, and electrical power lines being possibly harmful to health is something that patently affects everybody and makes for commanding headlines in the media. It should be said that, except for consequences of UV and laser exposure, scientific evidence for harm at levels of other forms of NIR commonly encountered is considered by most of the scientific community to be inconclusive and inconsistent. Moreover, there is the suspicion that scarce research funds are being diverted from more pressing health needs because of public misconceptions. Community campaigns against particular siting of powerlines and mobile phone base stations tend to marshal multiple arguments, including adverse visual impacts, property values, and loss of amenity, as well as possible health effects. Unfortunately, in this process, the language of ionizing radiation hazard (being made to glow in the dark near to powerlines and nuking food in a microwave oven) is borrowed to enhance the dread factor of certain types of NIR. There is also a tendency to lump different forms of NIR together, for example, by using research pertaining to powerline safety to mobile phone base station safety and vice versa, despite the widely different interaction mechanisms with tissue and the billion-fold difference in frequency.

This being said, the health detriment due to overexposure to a natural and ubiquitous form or NIR, namely, the sun, is underrated by the public and represents an area in which more effective control and public awareness could lead to significant improvement in health outcome, in certain areas of the world at least. Again, drawing a comparison with ionizing radiation, which is usually controlled via strict licensing and registration provisions at state legislature level, the question of whether or not to regulate NIR exposures is an interesting one to answer. This will be covered in Chapter 33. Penalties for infringing radiation legislation are fairly severe in the case of ionizing radiation. While it is advisable to encourage responsible behavior, including the avoidance of NIR overexposure, solar UV, for example, is uncontrollable at source, so legislation, if there is to be any, has be directed, in this and other cases, toward exposure limitation rather than source control.

Public perceptions of the seriousness of certain types of NIR exposure cannot be ignored. Apart from the community actions on siting of infrastructure just mentioned, there are many in the community who sincerely believe themselves to be adversely affected by such NIR sources. In surveys investigating the risk perception of NIR sources, the majority of participants did not rate, for example, mobile phone handsets or base stations as being a major health concern; however, over half were at least a little bit concerned (Schreier, Huss, and Roosli, 2006; Siegrist et al., 2005). Similar perceptions were reported for fields from powerlines and electrical devices (Schreier, Huss, and Roosli, 2006). An early study from Carnegie Mellon University (Morgan et al., 1985) showed that university students were more concerned about powerline than electric blanket exposure, even though in terms of actual exposure the latter is higher than the former. This also showed that supplying detailed information actually increased the concern over the risk, which has been born out in more recent studies on risk perceptions of mobile phones (Wiedemann et al., 2006).

Part VII will explore the notion of risk having two components, the probabilistic assessment of hazard by risk professionals and the outrage factor, which is a measure of how upset people feel (Sandman 1987). This impinges on how the Health of an individual is defined. According to the World Health Organization (WHO), Health is a state of complete physical, mental, and social well being and not merely the absence of disease or infirmity. On this definition, health is about not being upset or outraged over NIR exposure. Note that how upset people feel is related more to feelings of the exposure being involuntary and mysterious than the strength or even the existence of any credible scientific evidence of harm. A successful strategy for managing NIR exposures must pay particular attention to public perception. It could be argued that this is true for most environmental agents for which there is concern, but certainly in the case of ELF and RF, the absence of a generally accepted mechanism of interaction at low levels of exposure has been a reason for sections of the scientific and regulative community dismissing such concerns as groundless. While most of those expressing a belief that phones and electrical appliances do not find this affects their quality of life, a significant number declare themselves to be electrically sensitive. Chapter 24 will discuss the characteristics of this syndrome and will suggest effective strategies for its management.

1.4 Overview Summary of NIR Health Effects Evaluation: Status

The following represents a brief overview of the status of NIR health issues at the time of writing.

UV: The immediate effects on skin are well established – delayed effects (skin cancer and cataract) are also well researched and accepted. Safety limits are based on the notion of preventing sunburn. The assumption is made that if the skin is not allowed to burn, the risks of malignant and nonmalignant melanoma are not elevated.

Lasers: The thermal and photochemical effects on the retina, eye lens, and skin are well established. The blink and other aversion reflexes, which normally prevent overexposure of the retina are well understood. The frequency range of lasers and intense light emitting diode sources is constantly extending further into the invisible part of both UV and IR regions, where more hazard evaluation may need to be done.

RF: The thermal effects stemming from temperature rise within certain sensitive organs is well understood – the evidence for the so-called nonthermal effects is currently not compelling; however, research in this area is continuing. Some epidemiological studies have reported an association between heavy mobile and cordless phone use and brain cancer; however, other research has not confirmed these results.

ELF: Protection levels for ELF exposure are based on the prevention of activation of nervous tissue in the most sensitive areas of the body. Epidemiological research has left open the possibility of a raised risk of childhood leukemia in homes where average magnetic fields are within the top few percent of the total range; however, these results have not been shown to be causal.

Static fields: The mechanisms by which static fields affect the human body are well understood. Exposures within guideline limits will protect against established effects. Specific procedures will minimize transient effects experienced in certain situations, for example, moving within a strong magnetic field such as when undergoing a magnetic resonance imaging examination.

Tutorial Problems

1The wavelengths given in Table 1.1 relate to the frequencies of the radiated waves via the relationship f (frequency in Hz) = c (velocity of propagation in m/second)/λ (wavelength in m). What frequency corresponds to the following: 1 m waves; 1 mm waves; 1 km waves. What regions of the NIR spectrum do these correspond to?

2 At a point 100 m from a radiofrequency transmitter at 1 GHz frequency, a survey meter measures a steady power density of 1 mW/m². What is the value of the electric field (E-field) vector at this point (in V/m). What is the magnetic field vector (in A/m)? (Assume in air).

3 Estimate the power of the source in question 2 (in watts).

4 What interaction mechanism with the human body is the chief concern for the following types of NIR: UV, microwaves, and power frequency magnetic fields?

References

Morgan MG, Slovic P, Nair I, Geisler D, MacGregor D, Fischhoff B, Lincoln D, Florig K. 1985. Powerline frequency electric and magnetic fields: a pilot study of risk perception. Risk Analysis: An Official Publication of the Society for Risk Analysis5:139–149.

Sandman P. 1987. Risk communication: facing public outrage. EPA Journal13:21–22.

Schreier N, Huss A, Roosli M. 2006. The prevalence of symptoms attributed to electromagnetic field exposure: a cross-sectional representative survey in Switzerland. Sozial- und Präventivmedizin51:202–209.

Siegrist M, Earle TC, Gutscher H, Keller C. 2005. Perception of mobile phone and base station risks. Risk Analysis: An Official Publication of the Society for Risk Analysis25:1253–1264.

Wiedemann PM, Thalmann AT, Grutsch MA, Schutz H. 2006. The impacts of precautionary measures and the disclosure of scientific uncertainty on EMF risk perception and trust. Journal of Risk Research9:361–372.

Part I

Hazard Identification and Assessment: What are the Dangers and How are the Sources Dangerous?

Chapter 2

Hazard Identification: Laboratory Investigation

Andrew Wood

Department of Health and Medical Sciences, Faculty of Health, Arts and Design, Swinburne University of Technology, Hawthorn, Victoria, Australia

2.1 Introduction

As a preliminary to discussing investigations into possible health effects from various specific forms of nonionizing radiation (NIR), this chapter and Chapter 3 will deal with more general aspects of identifying the conditions under which an agent could be considered hazardous. This chapter considers investigations carried out in a laboratory, whereas Chapter 3 deals with studies of groups of the human population going about their normal duties at home or at work. This general introduction will set bioeffects research into a proper perspective regarding limitations on setting safe levels with any degree of certainty. What will become apparent in later chapters is a lack of consistency in experimental evidence of harm from environmental NIR, particularly those forms with a frequency less than 300 GHz. In this chapter, some discussion of the strengths and weaknesses of various forms of laboratory studies in this hazard identification process is thus in order, starting with human volunteer experiments down through various levels of biological complexity to studies of individual molecular components and mathematical modeling of interaction mechanisms. Those with some familiarity with the so-called scientific method could skim over the following few sections.

2.2 The Scientific Method

Ideally, this is a method for distinguishing truth from falsehood. It applies to the physical world, since it involves making observations or making measurements on processes amenable to measurement. It also involves hypothesis making, in order to characterize these measurements or observations. These hypotheses, once formulated, will suggest further sets of observations that should be done to give further support to the hypothesis. If these observations do not do this, then the hypothesis is further modified to take account of this divergence. As the hypothesis is refined, the subsequent observations will be more and more consistent with it. Moreover, if independent scientists were to repeat these measurements under similar conditions, they too should find behavior consistent with the hypothesis. Indeed, sufficient details should be reported to allow such replication to take place. Logically, it is a process of induction, rather than deduction, since it is arguing that specific observed behavior can be generalized.

To take an example, if we observe people sunbathing on the beach, we will notice that at the end of a period of 2 hours, some will be sunburnt, others not. We can then hypothesize that, for example, the lighter the hair color, the more extensive the sunburn. We may then construct a meter to measure (i) reflectivity of hair and (ii) severity of sunburn. We may want to modify the hypothesis in the light of further measurements to allow for similar reflectivities of hair but different color (i.e., ginger vs blonde). We may wish to analyze skin and hair for specific compounds such as melanin, to see if they are correlated. We may wish to measure the differential expression of genes in those who sunburn versus those who do not, further modifying the hypothesis.

An essential feature of a scientific hypothesis is falsifiability, that is, it must be capable of being shown to be false by experimentation. An individual may have a strong conviction that, for example, wearing a copper wristband will alleviate pain. This is falsifiable by substituting a look-alike plastic band for a copper one and asking the individual to rate the severity of pain. However, if there turns out to be a placebo effect, where the plastic band is more effective than no band and may be of similar effectiveness as a copper band, it becomes unclear what the hypothesis actually is.

2.3 Human Volunteer Experiments

This type of experiment consists of recruiting groups of volunteers, representative of the general community or sections of the community (such as electrical utility workers or regular mobile phone users), to determine what immediate effects result from short-term exposure to forms of agent; in this case, types of NIR. Sometimes, these are termed provocation studies because responses are deliberately evoked, in contrast to survey experiments, in which responses to everyday exposures are studied. In a provocation experiment, once a level has been established for the occurrence of a certain bioeffect and this bioeffect becomes the basis for safety standards, subsequent experimentation will concentrate on identifying any previously undetected effects at lower levels, since it would be unethical to continue to expose volunteers to levels deemed unsafe. Some of the endpoints investigated to establish safety limits and at lower levels are summarized in Table 2.1.

Table 2.1 Human (provocation) studies.

Human volunteer studies have certain strengths: if effects can be clearly demonstrated, they can be directly assessed in terms of general human performance, without the need for extrapolation from animals, in vitro studies or from simulations. Secondly, the exposure conditions can be accurately controlled and varied by the experimenters. In fact, the laboratory environment can be accurately monitored and standardized to eliminate many confounding variables encountered in survey studies outside the laboratory. Thirdly, volunteers can be categorized into subgroups according to age, gender, education, and so on, to account for covariation. On the other hand, there are clear limitations on this type of study, the main one being that only immediate or short-term effects can be identified, the effective upper limit on exposure duration being the time constraints on volunteers. It is rare for participants to be able to devote more than a 24-hour period to be in a laboratory. Even if a participant makes several return visits to the laboratory over a longer period, it is hard, if not impossible to control incidental exposures when away from the laboratory. Another limitation concerns volunteer compliance: under most ethics provisions, participants can withdraw at any time, even after several segments of a repeated-measures trial. Usually, as part of initial experimental design, numbers in subgroups (such as age ranges) are carefully planned to provide adequate statistical power. Volunteers who fail to keep appointments or who withdraw sometimes cause the group sizes becoming less than optimal. Post hoc analysis and multiple comparisons both weaken statistical design, and unfortunately, these are often a feature of human volunteer studies.

Double-blind experiments are those in which neither the volunteer participants nor the experimenters attending the participants are aware of whether the exposure is real or sham. This is done each time the test is conducted, for example, maybe with the exposure coded A or B. This way of conducting trials is considered to be best practice because subjective responses and biases are removed. In these experiments, a second experimenter maintains a key or list of the actual real/sham exposure status on each occasion. At the conclusion of the experiment, the key to the codes is revealed, preferably after the group analyses have been carried out.

2.4 Whole Organism Experiments

This type of experiment is fundamental to bioeffects research and for establishing the nature of adverse health effects. A traditional health effects assessment will involve exposing populations of typically laboratory-bred rodents to various levels of agent and comparing health-related outcomes in these groups with those in a separate group of animals maintained in exactly the same conditions, except that the agent is not given (sham exposure). With respect to NIR, the sham exposure usually consists of the generator of NIR (of whatever type) being placed in the same position as for the other parts of the experiments, but with the power supply turned off. For nonvisible forms of NIR, the experimenter is often made unaware of the level of exposure given to avoid bias in the interpretation of results. The size of groups of animal is determined by the magnitude of the effects expected: for small effects, the groups have to be large – see Section 2.9. The duration of these experiments is typically over the lifetime of the animal, which for rodents is around 2 years. Exposure systems have to be such that they are representative of human exposures and ethical considerations preclude experimenting in ways which would cause pain or distress. However, ethical standards have become more stringent in recent years and data from early work, which cannot now be repeated, provide a backdrop that can be correlated with or predictive of effects of accidental overexposure in humans. The issue of coexposure (e.g., the question of whether UV exposure preconditions an animal to subsequent ionizing radiation effects or vice versa) is one that has been widely investigated using animal models. In most cases, the disease type of major concern is cancer (i.e., the carcinogenic or mutagenic potential of the agent) but possible effects on birth outcomes (teratogenicity) have also been extensively studied. In shorter term experiments, effects of NIR agents on physiological and behavioral responses can be studied (as shown in Table 2.2)

Table 2.2 Experiments carried out on experimental animals to determine existence of health effects and level of exposure these occur (if they do occur).

The main disadvantage in this type of research is extrapolation, that is, of projecting to the animal model what an appropriate level of exposure should be, to be equivalent to human exposure, and the extrapolation of outcomes of animal models of disease to the human. For example, a human will absorb RF energy strongly at 80 MHz (where there is a resonant condition for the whole body), whereas the equivalent resonant condition for a mouse is several hundred megahertz. A four times stronger electric field applied to a human compared to that of a pig will induce the same amount of current in the legs of the respective organisms. As an example of the second, the Eμ Pim-1 strain of mouse, which has an abnormally high lymphoma rate, cannot easily be compared to human susceptibility to lymphoma because the Pim-1 oncogene (gene whose inappropriate activation leads to cancer formation) appears to be associated with different cancer types in the human. The other disadvantages of this type of experiment are the relatively large cost and the length of time required to plan and execute the experiment and analyze the results. The standard duration of a rodent experimental phase is 2 years, which corresponds to the life expectancy of the animals. It is customary to use several levels of exposure of the agent in order to establish a dose–response relationship, if one exists. For each level, a cohort of 100 animals or more is required, bringing the total number of animals per health effects assessment to around 1000 or more. One way around this problem of cost and time is to use whole organisms that are more primitive and whose lives are shorter. Because of recent advances in genomics, organisms such as the nematode worm (Caenorhabiditis elegans) with a 3-week lifespan have recently become popular. It consists of roughly 1000 cells, has a nervous system and brain, and exhibits behavioral characteristics akin to learning. The genome, that is the molecular sequence of the 20,000 genes, is now known. Other simple organisms, which are often used as a standard biological preparation, are the fruit fly (Drosophila melanogaster) and the Zebra fish. These are perhaps not the best models for human cancer unless, for example, human oncogenes are inserted into the host genome.

2.5 Studies on Isolated Cells, Organs, or Subcellular Organelles

The development of reliable and replicable tissue and organ culture techniques has made it possible to carry out tests for toxicity on colonies of living cells maintained in special flasks. Since the labware was originally glass, the name in vitro (in glass) is applied to this experimentation. It is now possible to maintain slices of brain taken from a freshly killed rodent in an artificial biological fluid (media) for several weeks with some degree of functionality (e.g., electrical activity) persisting over thisperiod. It is also possible to grow colonies of individual nerve cells in such a way that they form spontaneous connections and show rudiments of memory formation. Other cells types will continue to secrete hormones and neurotransmitters in response to stimuli over several cell generations, more or less indefinitely. These so-called immortalized cell lines continue to divide for many decades. An example is the HeLa cell line, which was established from a human cervical tumor in 1951 and is still used around the world, despite the fact that the donor has long since died. HeLa cells continue to possess similar characteristics to the original colony, so provide a stable in vitro model, which can be compared between laboratories and between countries. Colonies of these cells (which are derived from cancers and are referred to as transformed cells) and hybrid cells (which have the nucleus of one cell transplanted into another) can be obtained from recognized suppliers, such as the American Type Culture Collection or ATTC (http://www.atcc.org/). Other commonly used single-cellular organisms include bacteria (Escherichia coli or E. coli), yeasts, and amphibian eggs (oocytes). The advantages of exposing single-cell organisms to physical agents are that it is relatively easy to perform a large number of replicated trials, the basic materials are cheap, interlaboratory comparisons are easy to specify, and many ambient conditions are readily controllable. In studying basic cellular processes, such as membrane channels and pumps, the regulation of enzymic or cell signaling pathways, or the development of chromosomal abnormalities, these single-cell preparations are ideal. For example, various types of white blood cells (T or B lymphocytes) are widely used to investigate possible alterations in immune system responses.

There are two main drawbacks in the use of this type of data in human risk assessment are (i) the design of an exposure system that will produce a measured dose relevant to human exposure and (ii) the interpretation of any changes in biological function at the cellular level in terms of implications for human health. Since some of the cell lines used are transformed (i.e., already showing characteristics of cancer cells), extrapolation of findings to considerations of carcinogenicity is not easy. The mere fact that these cellular systems are isolated from their usual environment (which would control for temperature, pH, and nutrient supply) implies that the processes studied may not be the same as those in vivo.

At the lowest level of complexity, various components can be isolated from individual cells, such as enzymes (ornithine decarboxylase and ATP-ase), proteins (hemoglobin, myosin, and heat-shock protein), and nuclear material (DNA and RNA), to discover whether there are any modes of interaction at the level of single molecules, as there is, for example, in UV-induced dimerization of DNA components.

The determination of an equivalent dose (i.e., equivalent to human exposure) is not straightforward. For example, the rate of absorption of radiofrequency (RF) energy in a test tube or culture flask varies from position to position, making it hard to design an in vitro exposure system in which nonuniform heating effects can be eliminated. Questions of dosimetry have been reviewed in a supplement to Bioelectromagnetics (Guy, Chou, and McDougall, 1999) and more recently (Paffi et al., 2015).

2.6 Sources of Artifact and Importance of Independent Replication and Quality Control

An artifact is an effect that does not occur in the undisturbed organism but is the result of the way in which the experiment was conducted or in the preparation of the organism for investigation. An experiment that appeared to show reduced drinking behavior in animals (exposed in a laboratory) to high electric fields, but in which it was later revealed that the animals were suffering microshocks from the metallic feeding troughs is an example of an effect being an artifact. However, in this case, it was the specific aspect of the exposure leading to a bioeffect that needed proper identification – it was not electric fields per se, but their interaction with the experimental setup that caused the observed bioeffect. Other examples are unintentional statistical bias (caused by skew-symmetric data, inappropriate choice of reference or control observations, omission of outlying data, inconsistent inclusion criteria, etc.); systematic error (due to drift in measuring apparatus, electromagnetic interference, as examples); observer bias (if the experimenter is not blind to the exposure status); and, in the case of human experimentation, subject bias. Biological experiments are inherently variable, due to the stochastic nature of many biological processes: to a certain extent, this is reflected in the standard deviation in repeated measurements, but an outcome may still achieve statistical significance via an inexplicable anomaly.

The concept of statistical significance needs further clarification: an appendix to this chapter gives the rudiments of statistical testing and power calculations, but it needs to be kept in mind that the accepted criterion for a change to be significance is that the odds for the effect being due to chance variation is 1 in 20 or 5%. Many would argue that this is not conservative enough and that odds of only 1% or even 0.1% for the observed change being due to chance alone represent a better basis for identifying effects. This emphasizes the need for corroboration or replication of experiments reporting bioeffects, particularly if the levels of exposure chosen are so low as to appear out of step with everyday experience. Replication should be carried out ideally by independent groups following similar if not identical procedures and outcomes should be consistent across several levels of investigation (e.g., if an agent causes an increased release of hormone in a cell culture, the same agent should produce similar rises in hormone levels in blood of human subjects). It is not unusual for a scientific paper to contain flaws. In an Editorial (2003), the prestigious scientific journal Nature noted that it is regrettable but inevitable that the scientific record contains errors. Several of its issues contain retractions of papers in which flaws were discovered subsequent to publication. With this in mind, the above would represent a reasonable check list, before an effect can be considered as a reliable piece of information in an overall risk assessment process Table 2.3).

Table 2.3 Check list for inclusion of research report in overall risk assessment process.

The last two considerations will be explored further in the following two sections. Peer reviewed refers to the method used by most reputable scientific journals of only accepting articles after extensive review and approval by two or more independent experts in the field. Although this method of review does not guarantee that the methodology has been appropriate and the conclusions drawn from the results valid, it is a method that reduces the risk of this.

2.7 Difference Between Effects and Harmful Effects: Extrapolation to Human Health Outcomes

Many effects of physical agents (such as raised skin temperature in response to RF or UV exposure), although statistically significant, represent a change that is within the normal range of changes produced by everyday experiences. The unclothed human body can cope comfortably with environmental temperatures ranging between around 15 and 35 °C. Shivering and sweating are normalphysiological adjustments designed to maintain core temperature at 37 °C. While prolonged exposure to excessive heat or cold can be fatal, there are various degrees of discomfort that can lead to impaired physical or mental performance. The point at which a biological effect becomes a health effect may be somewhat difficult to pinpoint, but taking the World Health Organization definition of health referred to in Chapter 1, it would be wise to err on the side of conservatism. The identification of discomfort is relatively easy in the case of human volunteer experiments, in the short term at least. Extrapolation from other types of experiment to health implications for humans requires considerable scientific judgment based on careful quantitative argument. It is regrettable that many scientific papers engage in vague speculation on possible health outcomes without due regard to the plausibility or consistency of the assertions made.

2.8 Role of Mathematical Modeling and Mechanism Studies

One way to place laboratory findings in a clearer context of health effects research is to be able to model the interaction of a physical agent at a molecular or tissue level and then to integrate this model to assess effects in the complete human body. We will see this as being particularly relevant in the case of regional or whole body exposure to RF fields, in which the rate of energy absorption can be represented point by point, with the effects of different tissue and blood flow properties being taken fully into account. Some of these models of the human body are extremely sophisticated, involving resolution down to a few millimeters. Another area where modeling is crucial is in the evaluation of putative forms of interaction of agents with biological system beyond those currently accepted. Even where an interaction mechanism has been accepted, there are often considerable areas of uncertainty, and it is in these areas that elucidation can be provided using this approach. An example is the area of extremely low-frequency (ELF) fields, where understanding of the relationship of induced electrical current to cell stimulation is still incomplete. A number of interactions of ELF fields with tissue (not involving induced current) have been proposed. This is in order to account for incomplete experimental evidence that low-level fields give rise to effects that cannot be explained by accepted mechanisms. By modeling these proposed mechanisms, their plausibility can be gauged and further experimental tests can be formulated.

Appendix: Statistical Concepts

A.1 Averaging

If we need to get a reliable estimate of, say, resting blood sugar concentration or blood pressure, a single measurement is inadequate. This is because there is variation throughout the day, even for someone sitting in a relaxed state, due to food intake, diurnal variations, and so on. The measuring instrument is also subject to variation in terms of the care the measurer takes in making the measurement, the calibration of the instrument, random variations due to natural electronic or chemical reaction noise, drift in instrument response, and so on. For this reason, several measurements are taken and then an average or mean determined. A large spread of the measurements would suggest than a large number of separate determinations should be made. The amount of spread is estimated by subtracting each determination from the average, squaring this difference, and then finding the average squared difference (root mean square). Actually, for reasons we do not need to go into here, rather than dividing the sum of the square differences by the number of observations (before taking the square root), the number of observations is reduced by 1. This is called the Standard Deviation or SD, and the formula for calculating it can be written as

(A.2.1) equation

where c02mathml_1 is the mean value and xi is the particular value, with n observations.

A.2 Standard Error of the Mean

A way of estimating how confident we are that the estimated average or mean value is close to the actual mean value (which is what we would get if we took an infinitely large set of measurements) is to compute the standard error of the mean (or SEM). Effectively, this is the SD divided by the square root of the number of observations (√n).

A.3 When Is a Difference Significant?

Often, we want to know if a certain treatment (exposure to NIR for instance) produces a difference that is statistically significant. Given the spread of measured values, what we want to know is if the change is within normal variation or not. There are ways of estimating what the chances are of a difference of 10%, for example, being due to normal variation. The variation could be quite low (0.1% of the mean, perhaps), in which case we would regard this 10% change as being significant. A useful test for significance is the "Student t-test (if we are sure the observations fit within what is known as a normal distribution). The way of estimating the value of t" varies according to the type of experiment done, but in an example where each person involved in a trial can be compared before and after exposure (paired t-test), the t value is given by the difference in means (before and after) divided by the standard error (SE) of the difference in the means. If all the changes have about the same magnitude, the SE will be small so the t value will be large. Finally, the number of paired observations minus 1 is then used in a standard table of values of t to determine whether a particular value is greater than a critical value for the difference to be significant at the 5% or 1% level. It is usual for anything below 5% to be taken as being sufficient to declare a change is being significant. This is equivalent to saying that the odds of the change being due to natural fluctuation are less than 5%.

A.4 Correlations

Instinctively, one would expect that the weight of humans is correlated with their height. However, since people of the same height can be fat or thin, the correlation is expected to be moderate rather than good. The coefficient of correlation (or "r) is a measure of how good this correlation is, ranging between 0 (no correlation at all) to ±1, being a perfect correlation (the minus sign indicating that as the first variable increases, the second decreases, negative correlation"). The formula for finding r is given in standard books on statistics, but in order to determine whether the correlation is significant or not, a t value can be calculated using the following formula:

(A.2.2) equation

where n is the number of pairs of observations. The t value is then compared in the table as before to determine whether the critical value has been exceeded for the correlation to be considered significant (usually at the 5% or 0.05 level). Of course, correlation does not indicate causation since both factors may be dependent on a third factor, which contributes to variation in both of them. In our example of height and weight, the weight is related to volume (weight divided by density) and the height is a