Physics World article

Reproduced with permission of Physics World
First published in Physics World, November 1996

Do the electric and magnetic fields created by power
lines cause cancer? Three scientists from the National Grid
Company in the UK think that the evidence is inconclusive
and that more research is needed.

Power lines and health
JOHN SWANSON, DAVID RENEW AND NIGEL WILKINSON

The human race has evolved in the presence of a static magnetic field for almost a million years -the magnetic field produced by the Earth itself. Since the early 1880s,when public electricity supplies were introduced, we have also lived increasingly in electric and magnetic fields that vary with time. And just as the geomagnetic field is generally thought to have no effect on humans, for most of the past century extremely low-frequency low-level fields have also been regarded as harmless.

This assumption has been challenged in the last two decades. In 1979 Nancy Wertheimer of the University of Colorado Medical Centre and Ed Leeper, a freelance physics consultant, published research showing an apparent association between cases of childhood cancer and residential proximity to power lines likely to produce high magnetic fields. A dozen similar studies have followed, most finding some sort of association. However, this is not the same as establishing causation, as we will explain. Associations have also been found between occupations likely to involve exposure to magnetic fields and the occurrence of various cancers.

Tens of millions of pounds are now spent every year on research into the possible links between magnetic fields and cancer. Members of the public frequently express concerns about health risks when new transmission lines are proposed. Legal actions have been commonplace in the US for some years, although no court has yet ruled that magnetic fields cause cancer. And last year a writ served on a British electricity company alleged that a child’s leukaemia was caused by exposure to electromagnetic fields, the first legal action of its kind in the UK.

This issue is a multidisciplinary one, involving statisticians, biologists, physicists and engineers, alongside experts in risk assessment and epidemiology (the study of the pattern of disease in a population). The challenge is to decide appropriate public policy in the face of uncertain scientific evidence, and to communicate what are complex and emotive scientific issues to the public. Physicists bring key skills to bear: understanding exposures and how to measure them, assessing the strength of scientific evidence, assessing likely sources of bias or error and judging the plausibility of any suggested interaction mechanisms.

Fields from power lines
Power systems generally operate at either 50 or 60 Hz. These correspond to wavelengths of 6000 and 5000 km, respectively, so most studies of the health effects of electromagnetic fields are carried out very much less than one wavelength away from the source. The radiation from the power system in this so-called near-field regime is negligible. For example, a 100 km length of high-voltage transmission line, which might carry a gigawatt and dissipate several megawatts in heat, radiates only a fraction of a watt. Instead, we are dealing with the effects of separate electric and magnetic fields.

The power carried by a transmission line, calculated as voltage times current, is equal to the power carried in the electric and magnetic fields around the line, calculated from the component of the Poynting vector (the vector product of the electric and magnetic field vectors) parallel to the line. The entire power of the line can be pictured as being carried in the fields with the conductors merely providing the boundary conditions.

An infinitely long circuit with parallel live and neutral conductors constitutes a “linear dipole” source of magnetic field. The field lines form a dipolar pattern in the plane perpendicular to the conductors, and the field strength falls as the reciprocal of distance, r, squared (figure 1a). Practical electricity systems use three live conductors or phases, with the currents 120o apart in each. A single three-phase circuit also forms a linear dipole. When two circuits are carried on one set of pylons, which is normal practice in many countries, the field depends on the order of the phases on the two circuits. In the most common phasing used on the National Grid (the 275 kV and 400 kV transmission network in England and Wales), the dipoles produced by the two circuits are almost antiparallel. This phasing acts in effect as a quadrupole source and the field strength falls off as r-3 (figure 1b).

When a transmission circuit is placed underground instead of overhead, the conductors are closer together and the dipole strength is reduced. However, the circuit will typically be only 1 m below ground - as opposed to 10 m or more above ground - and may therefore be closer to people and houses. The field directly above a 400 kV underground cable is higher than from an equivalent overhead line (figure 2). With underground cables at lower voltages, such as the 230 V phase-to-neutral cables that form the final distribution circuits, the conductors are very close together and the external magnetic field from the load currents is negligible. The conductors are also wound helically, which causes the field to fall off exponentially with distance.

The background power-frequency field in the majority of homes in the UK comes from the non-zero vector sum of the currents in the cables, known as the net current. Net currents arise because of the way distribution circuits are earthed. There will also be regions of elevated field around any appliances that are operating (figure 3). Most appliances form three-dimensional dipoles, with the field strength falling as r-3. Small appliances, such as electric can openers, tend to produce higher magnetic fields than the high-current devices, such as electric fires. This is because the motors and transformers in small appliances tend to contain only small amounts of iron, which allows more of the magnetic field to “leak” from the appliance.

There is a significant difference between the strength of the fields produced by different sources (e.g. power lines, appliances, background and occupational) and the average exposure of the whole population to these sources (figure 4). This is because only a small fraction of people live close to power lines (0.1% within 50 m in the UK) and because most people spend only small amounts of time very close to appliances.

Overhead lines produce electric as well as magnetic fields, ranging up to around 11 kV m-1 near ground level. However, characterising electric fields as dipoles or quadrupoles is difficult because the ground conducts electricity and image charges beneath the ground must also be considered. In underground power lines a conducting sheath eliminates the electric field. Buildings also screen electric fields, typically by factors of 10-100. Inside homes, the major sources of electric field are appliances and house wiring.

Both electric and magnetic power-frequency fields, when strong enough, can produce a range of effects. For instance, a fluorescent tube held aloft by a person standing underneath a transmission line can have a current of 50 microamps induced in it. This is sufficient to cause the tube to glow, but too small to be felt. An oscillating magnetic field of around 1 microtesla can deflect the electron beam of a VDU sufficiently to cause the picture on the screen to wobble at the beat frequency of the mains frequency and the screen refresh rate. And a magnetic field of 500 microtesla or more can disrupt the stepper motor of a quartz watch, causing the hands to rotate 100 times faster than usual.

In humans, 20 kV m-1 electric fields can induce current densities of 10 mA m-2 in the neck, which is comparable to the currents produced naturally by nerve and muscle action. Magnetic fields of 1600 microtesla can induce similar currents in the torso. These current densities are also comparable to the lowest current densities known to have an effect on humans: the threshold for producing “magnetophosphenes”, a visible flickering in the eye, is a current density in the retina of about 14 mA/m2 at 25 Hz. On this basis the UK’s National Radiological Protection Board recommends that people should not be exposed to fields above 1600 microtesla or 12 kV m-1 at 50 Hz. Other national and international bodies have produced guidance on a similar basis.

Epidemiological studies
The worries about cancer concern not short-term exposures to relatively high fields, but long-term exposures to fields perhaps as low as 0.2 microtesla. The epidemiological studies that give rise to these worries fall into two groups. Some concern residential exposures, mainly of children. These are typically “case-control” studies, in which children with cancer - the “cases” - are compared with children drawn from the same population but without cancer - the “controls”. The proportions of the two groups exposed to fields, or to a surrogate for fields such as distance to power lines, are then compared. Other studies concern occupational exposures. These are typically cohort studies, in which a group of people who are believed to be exposed to fields (eg workers in power stations) is identified. The proportion of this group contracting cancer is compared with the rest of the population.

Not every study has found an association between exposure to magnetic fields and cancer. Moreover, a single study will often yield varying results when different methods are used to analyse the exposure, or when different subgroups are considered. However, there is a tendency for both types of study to find weak statistical associations with leukaemia and, sometimes, with brain tumours.

As in all experiments, errors in epidemiological studies come in two varieties: random and systematic. Random errors are straightforward to assess. They arise from the need to draw inferences from samples, and epidemiology uses methods such as significance testing to allow for the consequent uncertainty. Most studies so far reported have been based on relatively small samples and are thus statistically imprecise. There is always a finite probability that any apparent effect of electromagnetic fields, or any other postulated cause, may have arisen simply by chance in the absence of any real effect. This makes it unwise to draw conclusions from any single study. However, the probability that the results of all the studies on magnetic fields, taken together, would arise purely by chance is small.

What a physicist would describe as “systematic errors” come in several varieties. Epidemiologists are always very conscious of their possible presence, but they are hard to quantify. “Control-selection bias” has certainly occurred in some studies of magnetic fields. For instance, when the control group - the children without cancer - are selected and contacted by telephone, as is common practice in much American epidemiology, there are likely to be socio-economic differences between the case and control groups (e.g. poorer families are less likely to have telephones). This introduces a bias into such studies because exposure to magnetic fields varies with socioeconomic status: for instance, wealthier families tend to live in houses with bigger gardens and are thus further from power lines that, particularly in America, are also more likely to be underground.

“Confounding” is another bugbear of epidemiology. Suppose an association has been found between, in this instance, magnetic fields and cancer. This might mean that fields cause cancer. Alternatively, the cancer might be caused by something else that happens also to be associated with the fields. No-one can be certain whether confounding is occurring or not. However, there is no shortage of other factors associated with magnetic fields. Residential fields are associated with socioeconomic status, as already mentioned, and so will be connected with a number of other lifestyle factors. It has been shown in both the UK and US that residential fields are also associated with traffic density. In occupational settings, the people who are exposed to magnetic fields tend to be industrial workers who are often also exposed to benzene, ionising radiation and so forth. One of the reasons many epidemiologists are reluctant to accept a causal relationship between fields and cancer is because of the potential for confounding.

There is also an apparent inconsistency within the epidemiology. The case-control studies already described suggest a relationship between residential exposure to magnetic fields and childhood leukaemia of the order of a doubling of risk for the more highly exposed section of the population. In most western countries, however, exposure of the population as a whole to fields has increased many-fold during this century as electricity consumption has increased. Assuming a linear relationship between exposure and risk, the case-control studies would therefore imply that leukaemia incidence rates should have increased correspondingly. In fact, they have not. The exposure of the UK population to magnetic fields has increased fourfold or more since the Second World War, whereas incidence rates for childhood leukaemia have increased by a few tens of percent at most.

Assessing the evidence
Is the evidence strong enough to move beyond simply observing an association involving fields and cancer to deducing that one causes the other? It is true that scientists have sometimes felt able to deduce causality on the basis of epidemiology alone without either a mechanism for the observed effect or experimental biological support for it. In one of the pioneering studies of epidemiology, the eminent Victorian physician John Snow established a link between cholera and water supply in London decades before the discovery of the cholera bacterium. In another landmark study, Sir Richard Doll, probably Britain’s foremost epidemiologist, and others demonstrated a convincing link between smoking and lung cancer, without depending on any biological evidence from the laboratory that tobacco smoke contains carcinogens. However, in both of those cases the epidemiological evidence was exceptionally strong. When the epidemiology is as weak as it is in studies of the association between magnetic fields and cancer, it is essential to have either experimental biological support or a plausible mechanism. At present we have neither.

If fields do have any adverse health effect, there will be a sequence of events: the field must interact with atoms or molecules; the interaction must lead to an effect at the cellular level; this effect must cause cell dysfunction; and the cell dysfunction must have an effect on the whole organism. Laboratory experiments worldwide have probed each stage in this chain, looking at individual genes and cells as well as at whole animals. Many positive results have been published, but none has yet been convincingly replicated by an independent laboratory.

Physicists are probably accustomed to repeating other people’s published experiments in order to build on them and move forward. It is therefore rather disturbing to observe how often one group publishes apparently clear evidence for the effect of magnetic fields, and another group then repeats the experiment, sometimes building in improvements that ought to make it more sensitive, and yet fails to find any effect. An explanation that is sometimes advanced is that the effect could be a subtle one that depends on other factors, not yet identified. Since such effects are not known, they cannot be controlled and therefore may differ between different groups. The alternative explanation is that the original effect was never actually real.

Is there a mechanism?
Regardless of whether biologists can demonstrate the later stages of the chain from exposure to disease, physicists have a special interest in the very first stage. If fields are to have any effect, there must be a mechanism of interaction with a simple physical system involving electric charges or magnetic moments. Moreover, that mechanism must produce a “signal” in the biological system that is greater than whatever “noise” level exists naturally in the body. This requirement is not unique to electric or magnetic fields, but applies to any physical stimulus that produces an effect on the whole organism. Specialist sensory organs, such as the eye or ear, allow the human body to detect signals that are comparable to the fundamental noise levels, but it cannot detect signals below that limit.

There is not space here to give a comprehensive survey of every mechanism ever proposed. We will consider some of the more interesting ones, however, in the hope that this might prompt more physicists to get involved in the debate.

• Alternating fields induce currents, and at high fields these currents lead to effects such as magnetophosphenes. But the currents induced by environmental field levels (the “signal” in this context) are less than the naturally occurring current densities present in the body (the “noise”).



• The induced alternating currents could modulate the voltages that exist across cell membranes. In this case the “noise” is partly thermal noise, which can be calculated. For a single cell the voltage produced by the magnetic field is much less than this noise. It might be possible for the signal to exceed the noise if a large number of cells acted cooperatively, but this would suggest a special structure that is optimised for the purpose, analogous to the eye or ear. There is no evidence for such a “magnetic organ” in humans. In contrast, sharks have such structures - the ampullae of Lorenzini - and use them to detect microvolt-per-metre electric fields in water.
  
• The noise level could be reduced by decreasing the bandwidth of the interaction mechanism: in other words the magnetic field frequency would have to be resonant with some process in the body. However, to achieve an adequate signal-to-noise ratio, the bandwidth would probably have to be so narrow that the effect could be tuned to 50 Hz or 60 Hz - the power frequencies on the two sides of the Atlantic - but not both. Various resonant mechanisms involving the earth’s static field have been proposed. One of these, ion cyclotron resonance of, for example, the calcium ion, cannot actually occur in a cell as the orbit would have to be more than 1 m. Others, such as Larmor precession and ion parametric resonance, require unfeasibly infrequent collisions and are open to other objections.



• Magnetic fields exert moments on ferromagnetic particles, which are found in humans. However, when the maximum plausible magnetic moment and the viscosity of the cell matter surrounding it are quantified, the amplitude of oscillation induced by environmental fields is less than the Brownian motion.



• Many biochemical reactions involve free radicals - highly reactive entities produced in pairs, each with unpaired electron spins. Magnetic fields can affect conversion between singlet states (electron spins antiparallel) and triplet states (parallel spins). Hence they can affect the concentration of free radicals by altering the recombination probabilities. The appropriate “signal-to-noise ratio” for this process has never been properly addressed. However, there is another reason why it cannot underlie the epidemiological results. Free-radical reactions typically have timescales of tens of nanoseconds - 50 Hz fields are effectively static on these timescales. The relevant magnetic field is therefore the instantaneous total field, which is usually dominated by the geomagnetic field (50 microtesla in the northern hemisphere), not the power-frequency component.



• It has been suggested that the fields produced by transmission lines could deflect cosmic rays, concentrating them on people living nearby. The two most obvious objections to this are that the deflection in an alternating field will average to almost zero, and that the size of deflection can be estimated as a metre at most.



• Most recently, Denis Henshaw at Bristol University in the UK has pointed out that electric fields can cause movement and concentration of radon daughter products such as polonium, which is also radioactive, and possibly other carcinogens as well. Daughter products attached to polarisable particles will drift towards the source of a non-uniform field. Moreover, daughter products attached to charged particles also oscillate in the field. Whether this could lead to health consequences and in particular to an increase in leukaemia is still controversial. Although particles made to oscillate by an electric field may hit surfaces, such as the lungs, and stick to them more often, the electric field in the lungs is well shielded by the surrounding tissue and seems too small to produce any significant effect. Outside the body, calculations suggest that if significantly increased concentrations of particles occur at all they do so only within millimetres of charged conductors, and these are not the places where people generally breath. Lastly, radon daughter products are not thought to be a major factor in leukaemia.

What now?
Clearly it would be both arrogant and rash of physicists to argue that because we have not yet been able to think of a possible physical mechanism, it is impossible for there to be an effect. However, the absence of both a mechanism and reproducible laboratory results inevitably means that the epidemiological results, already somewhat weak, are viewed with much greater scepticism. Several national or international bodies have surveyed this question and have concluded that the evidence linking fields with cancer is not convincing. However, such bodies often call for further research. That research needs the skills of physicists alongside those of biologists, epidemiologists and engineers.

Further reading
R K Adair 1991 Constraints on biological effects of weak extremely-low-frequency electromagnetic fields, Phys.Rev. A 43 1039-1048

R Doll et al. 1992 Electromagnetic fields and the risk of cancer: report of an advisory group on non-ionising radiation, Documents of the NRPB 3 (1)

D Hafemeister 1996 Biological effects and low-frequency electromagnetic fields Am. J. Phys.
64 974-981

National Radiological Protection Board 1993 Board statement on restrictions to human exposure to static and time varying electromagnetic fields and radiation Documents of the NRPB 4 (5)

Swanson J and Renew D C 1994 Power-frequency fields and people, Engineering Science and Education Journal, 3 71-79

E P Washburn, M J Orza, J A Berlin, W J Nicholson, A C Todd, H Frumkin and T C Chalmers 1994 Residential proximity to electricity transmission and distribution equipment and risk of childhood leukaemia, childhood lymphoma, and childhood nervous system tumours: systematic review, evaluation, and meta-analysis, Cancer Causes and Control 5 299-309

John Swanson, David Renew and Nigel Wilkinson are in the Technology and Science Division, The National Grid Company plc

back