Powerlines and Cancer FAQs

Organizational notes:

- Cross references to other questions are indicated by the letter Q followed by the question number; for example, (Q16A) indicates that further information is found in Question 16A.
- Bibliographic references are shown in brackets, for example [M2] is a reference to the second entry in section M of the annotated bibliography.
- This FAQ is three documents: the table of contents (toc.html), the Q and A section (QandA.html) and the annotated bibliography (biblio.html).

Most of the concern about power lines and cancer stems from studies of people living near power lines (Q12) and people working in "electrical" occupations (Q15). Some of these studies appear to show a weak association between exposure to power-frequency magnetic fields and the incidence of cancer.

However, epidemiological studies done in recent years show little evidence that power lines are associated with an increase in cancer (Q19A and Q19B, Q19H thru Q19K), laboratory studies have shown little evidence of a link between power-frequency fields and cancer (Q16), and a connection between power line fields and cancer remains biophysically implausible (Q18).

A 1996 review by a prominent group of scientists at the U.S. National Academy of Science concluded that:
"No conclusive and consistent evidence shows that exposures to residential electric and magnetic fields produce cancer, adverse neurobehavioral effects, or reproductive and developmental effects."(Q27E).

A 1999 review by the U.S. National Institutes of Health concluded that:
"The scientific evidence suggesting that [power-frequency electromagnetic field] exposures pose any health risk is weak."(Q27G).

A 2001 review by the U.K. National Radiation Protection Board (NRPB) concluded that:
"Laboratory experiments have provided no good evidence that extremely low frequency electromagnetic fields are capable of producing cancer, nor do human epidemiological studies suggest that they cause cancer in general." (Q27H)

The largest studies of childhood leukemia and power lines ever done reported in 1997-2000 that they could find no significant evidence for an association of power lines with childhood leukemia (Q19H through 19K). In contrast, a pair of studies published in 2000 [C54, C57] reported that if all the studies in which magnetic fields could be measured or estimated were pooled, a statistically significant association could be found for childhood leukemia in the children with the highest average fields.

On the other hand, a series of studies have shown what life-time exposure of animals to power-frequency magnetic fields does not cause cancer (Q16B).

Overall, most scientists consider the evidence that power line fields cause or contribute to cancer to be weak.

X-rays, ultraviolet (UV) light, visible light, infrared light (IR), microwaves (MW), radio-frequency radiation (RF), and magnetic fields from electric power systems are all parts of the electromagnetic (EM) spectrum. The parts of the electromagnetic spectrum are characterized by their frequency or wavelength. The frequency and wavelength are related, and as the frequency rises the wavelength gets shorter. The frequency is the rate at which the electromagnetic field goes through one complete oscillation (cycle) and is usually given in Hertz (Hz), where one Hz is one cycle per second.

The Electromagnetic Spectrum

Power-frequency fields in the US vary 60 times per second (60 Hz), and have a wavelength of 5,000 km. Power in most of the rest of the world is at 50 Hz. Broadcast AM radio has a frequency of around 10^6 (1,000,000) Hz and a wavelength of around 300 m. Microwave ovens have a frequency of 2.54 x 10^9 Hz, and a wavelength of about 12 cm. X-rays have frequencies above 10^15 Hz, and wavelengths of less than 100 nm.

This FAQ sheet will use the term "power frequency" to refer to both the 50- and 60-Hz alternating current (AC) frequencies used in electric power systems, and the term "power frequency field" to refer to the sinusoidal electric and magnetic fields produced by 50- and 60-Hz lines and devices. The phrase "EMF" will be avoided since it is an imprecise term that could apply to many very different types of fields, and because the term has a long-standing usage in physics to refer to an entirely different quantity, electromotive force. The terms "electromagnetic radiation" and "nonionizing radiation" will be avoided since power-frequency sources produce no appreciable radiation (see Q5).

Power-frequency fields are also properly referred to as extremely low frequency (or ELF) fields. In strict electrical engineering terms, ELF refers to frequencies between 30 and 300 Hz, but the term is often used in the biological and occupational health literature to cover the range from above 0 Hz to 3000 Hz (everything above static fields and below radio-frequency).

The interaction of biological material with an electromagnetic source depends on the frequency of the source. We usually talk about the electromagnetic spectrum as though it produced waves of energy. However, sometimes electromagnetic energy acts like particles rather than waves, particularly at high frequencies. The particle nature of electromagnetic energy is important because it is the energy per particle (or photons, as these particles are called) that determines what biological effects electromagnetic energy will have [A12].

At the very high frequencies characteristic of "vacuum" UV and X-rays (less than 100 nanometers), electromagnetic particles (photons) have sufficient energy to break chemical bonds. This breaking of bonds is termed ionization, and this part of the electromagnetic spectrum is termed ionizing. The well-known biological effects of X-rays are associated with the ionization of molecules. At lower frequencies, such as those characteristic of visible light, radio-frequency radiation, and microwaves, the energy of a photon is very much below those needed to disrupt chemical bonds. This part of the electromagnetic spectrum is termed non-ionizing. Because non-ionizing electromagnetic energy cannot break chemical bonds there is no analogy between the biological effects of ionizing and nonionizing electromagnetic energy [A12].

Non-ionizing electromagnetic sources can produce biological effects. Many of the biological effects of ultraviolet (UV), visible, and infrared (IR) frequencies depend on the photon energy, but they involve electronic excitation rather than ionization, and do not occur at frequencies below that of infrared (IR) light (below 3 x 10^11 Hz). Radio-frequency and microwaves sources can cause effects by inducing electric currents in tissues, which cause heating. The efficiency with which a nonionizing electromagnetic source can induce electric currents, and thus produce heating, depends on the frequency of the source, and the size and orientation of the object being heated. At frequencies below that used for broadcast AM radio (about 10^6 Hz), electromagnetic sources couple poorly with the bodies of humans and animals, and thus are very inefficient at inducing electric currents and causing heating [A12].

Thus in terms of potential biological effects the electromagnetic spectrum can be divided into four portions (see diagram of electromagnetic spectrum):

1. The ionizing radiation portion, where direct chemical damage can occur (X-rays, "vacuum" ultraviolet light).
2. The non-ionizing portion of the spectrum, which can be subdivided into:
   1. The optical radiation portion, were electron excitation can occur (ultraviolet light, visible light, infrared light)
   2. The portion where the wavelength is smaller than the body, and heating via induced currents can occur (microwaves and higher-frequency radiowaves).
      
   3. The portion where the wavelength is much larger than the body, and heating via induced currents seldom occurs (lower-frequency radiowaves, power frequencies fields and static fields).

In general, electromagnetic sources produce both radiant energy (radiation) and non-radiant fields. Radiation travels away from its source, and continues to exist even if the source is turned off. In contrast, some electric and magnetic fields exist near an electromagnetic source that are not projected into space, and that cease to exist when the energy source is turned off.

The fact that exposure to power-frequency fields occurs at distances that are much shorter than the wavelength of 50/60-Hz radiation has important implications, because under such conditions (called "near-field"), the electric and magnetic fields can be treated as independent entities. This is in contrast to electromagnetic radiation, in which the electric and magnetic fields are inextricably linked.

To be an effective radiation source an antenna must have a length comparable to its wavelength. Power-frequency sources are clearly too short compared to their wavelength (5,000 km) to be effective radiation sources. Calculations show that the typical maximum power radiated by a power line would be less than 0.0001 microwatts/cm^2, compared to the 0.2 microwatts/cm^2 that a full moon delivers to the Earth's surface on a clear night. The issue of whether power lines could produced ionizing radiation is covered in Q21B.

This is not to say that there is no loss of power during transmission. There are sources of loss in transmission lines that have nothing to do with "radiation" (in the sense as it is used in electromagnetic theory). Much of the loss of energy is a result of resistive heating; this is in sharp contrast to RF antennas, which "lose" energy to space by radiation. Likewise, there are many ways of transmitting energy that do not involve radiation; electric circuits do it all the time.

Ionizing electromagnetic radiation carries enough energy per photon to break bonds in the genetic material of the cell, the DNA. Severe damage to DNA can kill cells, resulting in tissue damage or death. Lesser damage to DNA can result in permanent changes which may lead to cancer. If these changes occur in reproductive cells, they can also lead to inherited changes (mutation). All of the known human health hazards from exposure to the ionizing portion of the electromagnetic spectrum are the result of the breaking of chemical bonds in DNA. For frequencies below that of hard UV, DNA damage does not occur because the photons do not have enough energy to break chemical bonds. Well-accepted safety standards exist to prevent significant damage to the genetic material of persons exposed to ionizing electromagnetic radiation [M2].

A principal mechanism by which RF and MW sources cause biological effects is by heating (thermal effects). This heating can kill cells. If enough cells are killed, burns and other forms of long-term, and possibly permanent tissue damage can occur. Cells which are not killed by heating gradually return to normal after the heating ceases; permanent non-lethal cellular damage is not known to occur. At the whole-animal level, tissue injury and other thermally-induced effects can be expected when the amount of power absorbed by the animal is similar to or exceeds the amount of heat generated by normal body processes. Some of these thermal effects (also see Q9) are very subtle, and do not represent biological hazards [A12].

It is possible to produce thermal effects even with very low levels of absorbed power. One example is the "microwave hearing" phenomenon; these are auditory sensations that a person experiences when his head is exposed to pulsed microwaves such as those produced by radar. The "microwave hearing" effects is a thermal effect, but it can be observed at very low average power levels.

Since thermal effects are produced by induced currents, not by the electric or magnetic fields directly, they can be produced by fields at many different frequencies. Well-accepted safety standards exist to prevent significant thermal damage to persons exposed to MWs and RF (see Q31C), and also for persons exposed to lasers, IR and UV light [M3].

The electric fields associated with the power-frequency sources exist whenever voltage is present, and regardless of whether current is flowing. These electric fields have very little ability to penetrate buildings or even skin. The magnetic fields associated with power-frequency sources exist only when current is flowing. These magnetic fields are difficult to shield, and easily penetrate buildings and people. Because power-frequency electric fields do not penetrate the body, it is generally assumed that any biologic effect from residential exposure to power-frequency fields must be due to the magnetic component of the field, or to the electric fields and currents that these magnetic fields induce in the body [A12].

The argument that biological effects of power-frequency fields must be due to the magnetic component of the field has been the subject of recent debate [A14]. In particular, King [F27] has argued that the electrical fields from power lines do penetrate most buildings, and that the electrical currents induced in the body by power line electrical fields may be greater than those induced by power line magnetic fields. This issue is discussed further in Q16G and Q19L.

At power frequencies, the photon energy is a factor of 10^10 smaller than that needed to break even the weakest chemical bond. There are, however, well-established mechanisms by which power-frequency electric and magnetic fields could produce biological effects without breaking chemical bonds [A12, F3, F23, M6]. Power-frequency electric fields can exert forces on charged and uncharged molecules or cellular structures within a tissue. These forces can cause movement of charged particles, orient or deform cellular structures, orient dipolar molecules, or induce voltages across cell membranes. Power-frequency magnetic fields can exert forces on cellular structures, but since biological materials are largely nonmagnetic these forces are usually very weak.

Power-frequency magnetic fields can also cause biological effects via the electric fields that they induce in the body. These electric and magnetic forces occur in the presence of random thermal agitation (thermal noise) and electric noise from many sources; and to cause significant changes in a biological system applied fields must generally far exceed those that exist in typical environmental exposure conditions [A12, F3, F17, F23, F34, M6].

In general, the fields or currents that are induced in the body by power-frequency electric or magnetic fields are too low to be hazardous; and well-accepted safety standards exist to protect persons from exposure to power-frequency fields that would induce hazardous currents [M4, M5, M6, M8]. These safety standards for fields (as opposed to those that protect against shock from contact with conductors) are set to limit induced currents in the body to levels below those that occur naturally in the body. The well-known hazards of electric power, shock and burns, generally require that the subject directly contact a charged surface (e.g., a "hot" conductor and ground) allowing current to pass directly into the body.

One distinction that is often made in discussions of the biological effects of non-ionizing electromagnetic sources is between "nonthermal" and "thermal" effects. This refers to the mechanism for the effect: non-thermal effects are a result of a direct interaction between the field and the organism (for example, photochemical events like vision and photosynthesis); and thermal effects are a result of heating (for example, heating with microwave ovens or IR light). There are many reported biological effects of non-ionizing electromagnetic sources whose mechanisms are totally unknown, and it is difficult (and not very useful) to try to draw a distinction between "thermal" and "nonthermal" mechanisms for such effects [A12].

In the US magnetic fields are often still measured in Gauss (G) or milligauss (mG), where:

Electric fields are measured in volts/meter (V/m).

Measurement techniques are discussed in Q29 and Q30.

Within the path of a power line (known in the U.S. as a right-of-way or ROW) of a high-voltage (115-765 kV, 115,000-765,000 volt) transmission line, fields can approach 10 microT and 10,000 V/m. At the edge of a high-voltage transmission ROW, the fields will be 0.1-1.0 microT and 100-1,000 V/m. Ten meters from a 12 kV (12,000 volt) distribution line fields will be 0.2-1.0 microT and 2-20 V/m. Actual magnetic fields depend on distance, voltage, design and current; actual electric fields are affected only by distance, voltage and design (not by current flow) [F7].

Fields within residences vary from over 150 microT and 200 V/m a few cm from certain appliances to less than 0.02 microT and 2 V/m in the center of many rooms. Appliances that have the highest magnetic fields are those with high currents or high-speed electric motors (e.g., vacuum cleaners, microwave ovens, electric washing machines, dishwashers, blenders, can openers, electric shavers) [F22]. Electric clocks, and clock radios, which have been mentioned as major sources of night-time exposure of children, do not have particularly high magnetic fields (0.04-0.06 microT at 50 cm [F22]). Appliance fields decrease rapidly with distance [F7, F22]. Of the appliances assessed in British homes, only microwave ovens, electric washing machines, dishwashers and can openers produced fields greater than 0.20 microT at 1 meter [F22].

Because electric fields from powerlines have little ability to penetrate buildings, there is little correlation between electric and magnetic fields within homes [C11, C12]. In particular, while magnetic fields are elevated inside buildings near powerlines, electric fields do not appear to be similarly elevated [C11, C12].

Occupational exposures in excess of 100 microT and 5,000 V/m have been reported (e.g., in arc welders and electrical cable splicers). In "electrical" occupations typical mean exposures range from 0.5 to 4 microT and 100-2,000 V/m [F7, F11, F16, D19]. Exposure to power-frequency electric and magnetic fields are poorly correlated in occupational settings [F16].

Electric trains can also be a major source of exposure, as power-frequency fields at seat height in passenger cars can be as high as 60 microT [F28]

There are engineering techniques that can be used to decrease the magnetic fields produced by power lines, substations, transformers and even household wiring and appliances [F2, F29]. Once the fields are produced, however, shielding is very difficult. Small areas can be shielded by the use of Mu metal (a nickel-iron-copper alloy) but Mu metal shields are very expensive. Larger area can be shielded with less expensive metals; but such shielding is still expensive, and generally successful use requires considerable technical knowledge.

Increasing the height of towers, and thus the height of the conductors above the ground, will reduce the field intensity at the edge of a power line corridor [F2F29]. The size, spacing and configuration of conductors can be modified to reduce magnetic fields, but this approach is limited by electrical safety considerations. Placing multiple circuits on the same set of towers can also lower the field intensity at the edge of the ROW, although it generally requires higher towers. Replacing lower voltage lines with higher voltage ones can also lower the magnetic fields.

Burying transmission lines can substantially reduce their magnetic fields. The reduction in the magnetic field occurs because the underground lines use rubber, plastic or oil for insulation rather than air; this allows the conductors to be placed much closer together and allows greater phase cancellation. The reduction in magnetic fields for underground lines is not due to shielding. Placing high voltage lines underground is very expensive, adding costs that may exceed one million US dollars per mile.

The reduction in magnetic fields from burying a line is greatest at a distance from the line. At the center of a transmission line corridor, fields from a buried line can actually be higher than those from an overhead line [F29]. For example, in a comparison of overhead and underground 400 kV lines [F29], the fields at the center of the corridor were 25 microT for the overhead line and 100 microT for the buried line; but at 20 m, the fields were 10 microT for the overhead line and 1-2 microT for the buried line.

Different methods of household wiring can greatly affect magnetic fields inside houses. For example, the tube-and-knob method of wiring older houses produces higher fields than modern methods that use conduit or other methods that put the wires very close together; the fields are lower because the conductors are closer together and there is greater phase cancellation. Other strategies for reducing fields from household wiring include avoidance of ground loops, and care in how circuits with multiple switches are wired. In general conformance with modern electrical wiring codes will result in decreased magnetic fields.

Some studies have reported that children living near certain types of power lines (high-current distribution lines and high-voltage transmission lines) have higher than average rates of leukemia [C1, C6, C12, C19, C46], brain cancers [C1, C6] and/or overall cancer [C5, C17]. The correlations are not strong, and the studies have generally not shown dose-response relationships. When power-frequency fields are actually measured, the association generally vanishes [C6, C12, C19, C35, C44]. Many other studies have shown no correlations between residence near power lines and risks of childhood leukemia [C3, C5, C9, C10, C16, C17, C33, C35, C44, C45, C48, C51, C53], childhood brain cancer [C5, C9, C16, C17, C19, C28, C29, C33], or overall childhood cancer [C16, C19, C33].

All but one of the recent studies of powerlines and either childhood leukemia or brain cancer [C28, C29, C33, C35, C43, C44] have failed to show significant associations. The exception is a Canadian study [C45, C46] which showed an association between the incidence of childhood leukemia and some measures of exposure (see full discussion in Q19J).

With two exceptions [C2, C32] all studies of correlations between adult cancer and residence near power lines have been negative [C4, C7, C9, C13, C18, C21, C31, C32, C38, C40, C47]. The exception are Wertheimer et al [C41] who reported an excess of total cancer and brain cancer, but no excess of leukemia; and Li et al [C33] who reported excess leukemia, but no excess breast cancer or brain cancer.

The excess cancer found in epidemiologic studies is usually quantified in a number called the relative risk (RR). This is the risk of an "exposed" person getting cancer divided by the risk of an "unexposed" person getting cancer. Since no one is unexposed to power-frequency fields, the comparison is actually "high exposure" versus "low exposure". A RR of 1.0 means no effect, a RR of less the 1.0 means a decreased risk in exposed groups, and a RR of greater than one means an increased risk in exposed groups. Relative risks are generally given with 95% confidence intervals. These 95% confidence intervals are almost never adjusted for multiple comparisons (see Q21E) even when multiple types of cancer and multiple indices of exposure are studied (see Olsen et al, [C17], Fig. 2 for an example of a multiple-comparison adjustment).

No simple overview of the epidemiology is possible because the epidemiologic techniques and the exposure assessment in the various studies are so different. Meta-analysis, a method for combining studies [L15], has been attempted [A7, B3, B5, B9, B12, B18, C54, C57], but the results are problematical because of a lack of consensus as to the correct way to measure exposure. Meta-analyses also tend to get out-dated rather quickly. A 1999 meta-analysis of childhood cancer [B18], for example, was already missing the 4 big 1999 studies at the time it was published.

The following table summarizes the relative risks (RR) for the studies of residential exposure.

As a base-line for comparison, the age-adjusted cancer incidence rate for adults in the United States is 3 per 1,000 per year for all cancer (that is, 0.3% of the population gets cancer in a given year), and 1 per 10,000 per year for leukemia.

Most public and scientific attention has focused on childhood leukemia, with lesser attention given to adult leukemia, childhood and adult brain cancer, lymphoma and overall childhood cancer (see table in Q13A). The original studies which suggested an association between power lines and childhood cancer used a combination of the type of wiring and the distance to the residence as a surrogate measure of exposure, a system called "wire codes" [C1, C3, C6]. Other studies have used distance from transmission lines or substations as measures of exposure, and some studies have used contemporary measured fields or calculated historic fields. In general, the different methods of exposure assessment do not correlate well with each other, or with contemporary measured fields; none of these measures of exposure is obviously superior, and none is common to all the major studies (see figure below).

Historically, one of the more puzzling features of the childhood leukemia studies was that the correlation of "exposure" with cancer incidence appeared to be higher when wire codes or proximity to power lines were used as an exposure metric, rather than when fields were directly measured in the homes (see figure below). This has led to the suggestion that the association of childhood cancer with residence near power lines might be due to a factor other than the power-frequency field. For example, it has been suggested that socioeconomic class might be a confounder, since socioeconomic class is associated with cancer risk, and "exposed" and "unexposed" groups in some studies may be from different socioeconomic classes. This is of particular concern in the U.S. residential exposure studies that are based on wire codes, since the types of wire codes that are correlated with childhood cancer are found predominantly in older, poorer neighborhoods, and/or in neighborhoods with a high proportion of rental housing [A7, C20, C25]. However, in 1997 and 1999 the largest studies to date of power lines and childhood leukemia [C35, C44] found no association of leukemia with either wire codes or measured fields, and the most recent studies of brain cancer [C28, C29] have found no correlation with wire codes. These latest studies indicate that the "wire code paradox" does not actually exist.

The figure below shows the variety of endpoints that have been used in the childhood leukemia studies. Because of the lack of consensus as to the correct exposure metric, and the lack of an exposure metric that is common to most of the studies, no simple overview of the epidemiology can be provided. Attempts to provide an overview of these diverse data have been frustrated by the fact that no "unique" analysis can be produced. Rather one gets a family of analyses based on different definitions of exposure, most of which exclude some of the studies, and no one of which can be assumed to be the best. For example, in 1997 the U. S. National Research Council [A7] conducted a complex meta-analysis and concluded that: "wire codes are associated with an approximately 1.5-fold excess of childhood leukemia, which is statistically significant". This conclusion is based on just one of the eight separate meta-analyses of the childhood leukemia data performed by the NRC Committee, an analysis that excludes seven of the 11 studies and uses an arbitrary cut-point for defining who was exposed. A second analysis of the same four studies used a higher cut-point, and found a smaller excess that was "non-significant". The other six analyses done by the NRC committee yielded summary RRs that ranged from 0.8 to 1.7.

The childhood leukemia studies as a whole show no consistent association between residence near power lines and the incidence of leukemia.

However, a pair of studies published in 2000 [C54, C57] found that if certain reports were pooled and certain exposure metrics were chosen, there appeared to be an increased risk of leukemia in the highest exposure group.

• In the first of the analyses of pooled data, Ahlbom et al [C54] reported that if the nine studies that included long-term measurements of magnetic fields were pooled, a statistically significant association (relative risk = 2) could be found for childhood leukemia in the children with average exposures of 0.4 microT or greater. For children with lower average exposures, no significant elevation of childhood leukemia was found in the pooled studies. Average magnetic fields of greater than 0.4 microT are found in about 0.8% of homes [C54]. If this analysis is taken literally, then exposure to power-frequency magnetic fields could account for about 1% of childhood leukemia deaths (that is, 6-8 cases per year in the United States).
• In the second of the analyses of pooled data, Greenland et al [C57] reported that if the 15 studies for which magnetic fields were measured (or could be estimated) were pooled, a statistically significant association (relative risk = 1.7) could be found for childhood leukemia in the children with average exposures of 0.3 microT or greater. For children with lower average exposures, no significant elevation of childhood leukemia was found in the pooled studies. According to the authors this data indicates that exposure to power-frequency magnetic fields could account for 0-8% of childhood leukemia deaths in the United States.

Relative Risk of Childhood Leukemia

Relative risk (RR) of childhood leukemia and exposure to power-line fields. RRs are shown with 95% confidence intervals and the expected number of exposed cases (a measure of the statistical power of the study) is shown in parentheses. Where more than one exposure cut-point was used by the authors, the highest cut-point with more than 5 expected exposed cases is shown. The summary weights each study on the basis of the numbers of expected exposed cases, and treats all exposure measures equally. Pooled 1980-1994 data is from Moulder [A12].

The studies that show a relationship between cancer and power lines do not provide any consistent guidance as to what distance or exposure level is associated with increased cancer incidence. The studies have used a wide variety of techniques to measure exposure, and they differ in the type of lines that are studied. The US studies have been based predominantly on neighborhood distribution lines, whereas the European studies have been based strictly on high-voltage transmission lines and/or transformers.

Field measurements: A number of studies have measured power-frequency fields in residences [C6, C7, C12, C19, C21, C29, C34, C35, C44, C45, C46, C59]. Both one-time (spot), peak, 24-hour and 48-hour average measurements have been made. Two of the studies [C46, C59] using measured fields have shown a statistically-significant relationship between exposure and childhood leukemia. No other types of cancer in either adults of children have been show to be associated with measured fields.

A report published in 2000 [C54] calculated that if all the studies that included long-term measurements of magnetic fields were pooled, a statistically significant association could be found for children with 24-48 hr average exposures of 0.4 microT or greater. A second study published in 2000 [C57] reported that if all the studies for that included estimated or measured magnetic fields were pooled, a statistically significant association could be found for children with exposures of 0.3 microT or greater. For children with lower average exposures, no significant elevation of childhood leukemia was found in either analysis of the pooled studies.

Proximity to lines: Many studies have used the distance from the power line corridor to the residence as a measure of power-frequency fields [C4, C5, C9, C10, C13, C19, C20a, C21, C32, C33, C53, C58]. When something we can measure (distance to the line), is used as an index of what we really want to measure (the magnetic field), it is called a surrogate (or proxy) measure. Three [C5, C19, C32] of the 12 studies that have used distance from power lines as a surrogate measure of exposure have shown a relationship between proximity and cancer. The most notable are a childhood study [C19] that showed an increase in leukemia incidence for residence within 50 m of high-voltage transmission lines, and an adult study [C32] that showed an increase in leukemia incidence for residence within 100 m of high-voltage transmission lines. The largest study of proximity to power lines and childhood cancer found no association with any kind of cancer in children living within 50 m of power lines or substations [C58].

Depending of the type of line and its current, magnetic fields from power lines become less than those produced by the typical residence at a distance of 20-70 meters.

Wire codes: The original US power line studies used a combination of the type of wiring (distribution vs transmission, number and thickness of wires) and the distance from the wiring to the residence as a surrogate measure of exposure [C1, C2, C3, C6, C7, C12, C28, C29, C35, C44, C45, C46]. This technique is known as "wire coding" [F21]. Three studies using wire codes [C1, C6, C12] have reported a relationship between childhood cancer and "high-current configuration" wire codes. Two of these studies [C6, C12] failed to show a relationship between exposure and cancer when actual measurements were made, the third study [C1] made no actual measurements. The most recent studies of wire codes and childhood cancer [C28, C29, C35, C44, C45, C46] have found no significant associations.

Wire codes are stable over time [F6], but correlate poorly with measured fields [A7, F6, F7, F10, F21]. The wire code scheme was developed for urban areas in the U.S., and is not readily applicable elsewhere. It has been suggested that wire codes might be a better measure of long-term magnetic fields than actual magnetic field measurements, but analyses have shown that this is unlikely [A7, F21]. A more serious problem with using wiring codes to estimate magnetic field exposure is that wire codes correlate strongly with things that have nothing to do with magnetic fields (such as age of houses, traffic density and socioeconomic status) [C40].

Calculated Historic Fields: Many recent studies (Q19) have used utility records and maps to calculate what fields would have been produced by high voltage power lines in the past [C16, C17, C19, C21, C26a, C31, C32, C32, C33, C44]. Typically, the calculated field at the time of diagnosis or the average field for a number of years prior to diagnosis are used as a measure of exposure. These calculated exposures explicitly exclude contributions from other sources such as distribution lines, household wiring, or appliances. There is no way to check the accuracy of these calculated historic fields. See Jaffa et al [F36] for a discussion of some of the reasons to question the accuracy of these calculations.

Several studies have reported that people who work in some electrical occupations have higher than expected rates of some types of cancer. The original studies [D1, D2] were only of leukemia. Some later studies also implicated brain, lymphoma and/or breast cancer. As with the residential studies, there are many negative studies, weak correlations, and no consistent dose-response relationships. Additionally, many these studies are based on job titles, not on measured exposures.

Meta-analysis [L15] of the occupational studies is even more difficult than for the residential studies. First, a variety of epidemiologic techniques are used, and studies using different techniques should not be combined. Second, a wide range of definitions of "electrical occupations" are used, and very few studies actually measured exposure. Lastly, there is little consensus as to the appropriate exposure metric. The following table summarizes the relative risks for the studies of occupational exposure.

See Q19 for a more detailed discussion of the recent studies [also B11, B12, B13, B17, B20, B20].

While the causes of specific cancers in individuals are still poorly understood, the mechanisms of carcinogenesis are sufficiently well understood that cellular and animal studies can provide information relevant to determining whether an agent causes or contributes to cancer [A8, A9, A12, A13, K5, L26, L28]. Current research indicates that carcinogenesis is a multi-step process driven by a series of injuries to the genetic material of cells. Not surprisingly, this model of carcinogenesis is referred to as the multi-step carcinogenesis model.

The Multi-Step Carcinogenesis Model

This multi-step model replaced an earlier model, called the initiation-promotion model. The initiation-promotion model proposed that carcinogenesis was a two-step event, with the first step being a genotoxic injury (called initiation) and the second step being a non-genotoxic event (called promotion). It is now clear that this two-step model was too simple. In particular, it is clear that multiple genotoxic injuries are involved in many (in not all) types of cancer; and that promotion may not be involved in all types of cancer.

Our current understanding of cancer is that it is initiated by damage to the genetic information of a cell (the DNA). Agents which cause such injury are called genotoxins. It is extremely unlikely that a single genetic injury to a cell will result in cancer; rather it appears that a series of genetic injuries are required. Genotoxic carcinogens may not have thresholds for their effect; so as the dose of the genotoxin is lowered the risk of cancer induction gets smaller, but it may never reach zero. Genotoxins may affect many types of cells, and may cause more than one kind of cancer. Thus, evidence for genotoxicity of an agent at any exposure level, in any recognized test for genotoxicity, is relevant to assessing carcinogenic potential in humans [A12, A13, A8, A9, L26, L28].

There are many approaches to measuring genotoxicity. Studies of occupational-exposed humans can be done to look for genotoxic injury in white blood cell (Q16A). Animal exposure studies can be used to see whether exposure causes cancer, mutations or chromosomal injury (Q16B). Cellular studies can be done to detect DNA or chromosomal damage (Q16C) or neoplastic cell transformation (Q16D). In reviewing the genotoxicity literature, non-mammalian as well as mammalian systems have been included. The coverage of exposure conditions has also been broad, since any evidence for genotoxicity from any system exposed to any related type of field could be relevant to the question of carcinogenicity.

There are also many different types of laboratory tests that can be used to look for evidence of genotoxic activity:

It also appears that non-genotoxic (epigenetic) agents can contribute to the development of cancer, even though they may not be able to cause cancer by themselves. Epigenetic agents (non-genotoxic carcinogens) affect carcinogenesis indirectly, by increasing the probability that other genotoxic agents will cause genotoxic injury, or that genotoxic injury caused by other agents will lead to cancer. For example, an epigenetic agent might inhibit repair of potentially-genotoxic damage, affect the DNA in such a way as to make it more vulnerable to genotoxic agents, allow a cell with genotoxic injury to survive, or stimulate cell division in a previously non-dividing cell that had genotoxic injury [A8, A9, A12, L26, L28].

The actions of epigenetic agents may be tissue- and species-specific, and evidence exists that epigenetic agents have thresholds for their effects. Thus evidence that an agent has epigenetic activity must be evaluated carefully for its relevance to human carcinogenicity under real-world exposure conditions. This is significant for the issue of possible cancer risks from power-frequency fields, as the evidence, to the extent that it implicates such fields at all, suggests an epigenetic rather than genotoxic mechanism [A9, L26, L28].

Promoters are a specific class of epigenetic agents. In a classical promotion assay, animals are exposed to a known genotoxin at a dose that will cause cancer in some, but not all animals. Another set of animals are exposed to the genotoxin, plus the agent to be tested for promotional activity. If the agent plus the genotoxin results in more cancers than are seen for the genotoxin alone, then that agent is a promoter. Promotion assays are discussed in Q16E. Some types of cellular studies are relevant to the carcinogenic potential of agents, but are neither classic genotoxicity nor promotion tests. For example, cellular systems have been used to test whether an agent enhances the activity of known genotoxins, or whether an agent inhibits repair of DNA damage. These cellular studies of epigenetic activity can be regarded as the cellular equivalent of a promotion study, and are discussed in Q16D and Q16F.

Note: The majority of agents that are known to be carcinogenic in humans are genotoxins; and no role for epigenetic carcinogens have yet been identified in leukemia or brain cancer, the types of cancer most often associated with exposure to power-frequency fields in epidemiological studies.

In studies which blur the boundary between epidemiology and laboratory science, the white blood cells (lymphocytes) from workers with occupational exposure to an agent can be examined for chromosome aberrations, sister chromatid exchanges (SCEs) or micronuclei formation. The interpretation of these studies is complex, as they have all of the problems of exposure assessment, confounding and bias that characterize epidemiological studies. A number of such studies have been published [E2, E3, E5, E11, E12, E13, E14]. At first glance these studies appear very contradictory with some studies reporting "significant" effects and others not.

A major statistical issue that must be considered is that all of the studies examine multiple endpoints and subgroups, creating a massive multiple comparison problem (see Q21E). Skyberg et al [E12], for example, reports chromosomal damage in exposed workers; but this increase was found in only one subgroup, only for one of several assays, and has a p-value of only 0.04. With any adjustment for multiple comparison, the statistical significance of the genotoxicity effect reported by Skyberg et al vanishes. The multiple comparison problem also applies to the "positive" findings reported by Valjus et al [E11].

Even with the multiple comparison problems, several patterns emerge. The effects that are reported are predominantly seen in smokers, groups in which excess chromosomal abnormalities are expected. The effects are also seen predominantly in workers exposed to spark discharges [spark discharges are a phenomena that is unique to the electrical environment of high-voltage sources, where electric fields can reach intensities of up to 20 kV/m, and body currents can reach several amps]. Finally, the reported increases are limited to increased chromosomal aberrations, with no effects on SCEs; this is somewhat surprising, as the SCE assay is generally considered to be more sensitive to genotoxic agents than the chromosome aberration assay.

In summary, the cytogenetic studies of workers exposed to strong power-frequency electric and magnetic fields provides no consistent evidence that these fields are genotoxic. The unreplicated evidence for genotoxic effects is largely confined to current and former smokers, and to workers exposed to spark discharges.

Animal carcinogenesis studies: Until 1997, the biggest gap in the range of genotoxicity endpoints that have been assessed for power-frequency fields was that relatively few long-term whole animal exposure studies had been published.

Bellossi et al [G14] exposed leukemia-prone mice to 6000 microT fields for 5 generations (lifetimes) and found no effect on leukemia rates; however, the study used 12 and 460 Hz pulsed fields, so the relevance of this to power-frequency exposure is unclear.

Rannug et al [G23] reported that exposure of mice for 2 years to 50 and 500 microT fields did not significantly increase the incidence of skin tumors, lung tumors, or leukemia.

Beniashvili et al [G16] reported that exposure of mice for two years at 20 microT resulted in an increased incidence of mammary tumors. However, the study has been reported only in preliminary form with incomplete information about exposure conditions and experimental design.

Fam and Mikhail [G53] reported that mice exposed for three generations to a 60-Hz field at 24,000 microT had an increased incidence of lymphoma. The experiments were not conducted blind (that is, the experimenters knew which animals had been exposed and which had not), and the controls may not have been housed under conditions comparable to those of the exposed animals. When these data were presented at scientific meetings, concerns about noise, hyperthermia (overheating) and vibration were raised.

In 1997, Yasui et al [G66] reported the absence of increased cancer incidence and mortality in male and female rats after two years of exposure to 50-Hz fields at 500 and 5000 microT. In addition to finding no changes in overall cancer rates, they found no differences in the rates of individual types of cancer, including leukemia, lymphoma, brain cancer and breast cancer.

Also in 1997, Mandeville et al [G67] reported that two years of exposure of female rats to 60-Hz fields at 2, 20, 200 or 2000 microT had no effect on survival, leukemia incidence or solid tumor incidence. In addition to finding no overall changes in survival or cancer incidence, Mandeville et al found no evidence for any dose-related trends in survival or cancer incidence.

In 1998, Harris et al [G70] found that 1.5 years of exposure of lymphoma-prone mice to 50-Hz fields at 1, 100 or 1000 microT had no effect on lymphoma incidence. In addition to testing continuous exposure, Harris et al also showed that exposure of mice to intermittent (15 min on, 15 min off) fields at 1000 microT had no effect on lymphoma incidence. Similar results were reported by McCormick et al [G36]. Interestingly, these studies use the same animal model in which Repacholi et al (Rad Res, 1997) reported that exposure to 900 MHz radiofrequency (RF) radiation resulted in an increase in lymphoma incidence.

Also in 1998-1999, the U.S. National Toxicology Program (NTP) reported that two years of exposure of mice (McCormick et al [G72B]) and rats (Boorman et al [G72A]) to 60-Hz fields at 2, 200 or 1000 microT had no effect on survival or cancer incidence. In addition to testing continuous exposure, NTP showed that exposure to intermittent (1 hr on, 1 hr off) fields at 1000 microT had no effect on cancer incidence. No effects on overall cancer, leukemia, brain cancer, lymphoma or breast cancer were observed, and no exposure-response trends were found.

In a study published in late 1999 Kharazi et al [G88] reported that life-time exposure of mice to a 1420 microT field had no effect on brain tumor incidence.

In 2000 Babbitt et al [G84] reported that life-time expose of mice to a 1420 microT field had no effect on lymphoma incidence. The study also found that this field had no effect on the incidence of lymphoma induced by ionizing radiation (see Q16E).

In summary, the long-term animal exposure studies conducted to date provide no replicated evidence that long-term exposure to power-frequency fields causes cancer in animals, and no evidence at all that long-term exposure of animals to power-frequency fields is associated with leukemia, brain cancer or breast cancer.

For further discussion of the animal carcinogenesis studies see McCann et al [K7] and Boorman et al [K10].

The long-term animals exposure studies with power-frequency fields are summarized in the following figures.

Animal Carcinogenesis Studies (Total Cancer or Overall Life Span)

Summary of animal carcinogenesis studies using power-frequency magnetic fields that assessed total malignant tumors or overall survival. The figure shows the ratios (exposed/sham) of the number of animals with tumors at the end of the experiment, or the number of deaths during the experiment. All data are shown with 95% confidence intervals. Typical 24-hour average residential fields are shown for comparison [F7, F22].

Animal Carcinogenesis Studies (Leukemia and Lymphoma Only)

Summary of animal carcinogenesis studies using power-frequency magnetic fields that assessed lymphoma and/or leukemia. The figure shows the ratios (exposed/sham) of the number of animals with lymphoma or leukemia at the end of the experiment. All data are shown with 95% confidence intervals. Typical 24-hour average residential fields are shown for comparison [F7, F22].

Whole organism mutagenesis and genotoxicity studies: Whole organism exposure studies can be relevant to carcinogenic potential even when the end point is not cancer. The ability of an agent to cause mutations or chromosome aberrations in an organism is an indication that the agent is genotoxic, and hence potentially carcinogenic.

Benz et al [G4] reported that mice exposed for multiple generations 300 microT (plus 15 kV/m) or 1,000 microT (plus 50 kV/m) showed no increase in mutation rates, fertility, or sister chromatid exchanges (SCEs). Similarly, Kowalczuk and Saunders reported that mice exposed to 10,000 microT fields [G43] showed no increase in mutations; and Zwingelberg et al [G24] reported that a 30,000 microT field did not increase SCE rates in mice.

Kikuchi et al [G95] reported that exposure of fruit flies to 500 or 5000 microT fields for 40 generations had no effect on the mutation rate.

The only positive report of genotoxicity from whole organism studies is Lai and Singh [G60] who reported that 100-500 microT fields caused DNA strand breaks in rat brain cells.

In summary, the long-term animals exposure studies conducted to date provide no replicated evidence that long-term exposure to power-frequency fields causes cancer or genotoxic injury in animals.

The traditional cellular test systems for genotoxicity have been mutagenesis assays in bacteria, yeast, and mammalian cells. A variety of other mammalian test systems for genotoxicity also exist, including chromosome aberration assays, SCE assays, DNA strand break assays, and micronuclei formation assays.

Cellular genotoxicity studies of power-frequency and ELF fields have been massive in scope. Published studies have spanned many different models, from plasmids and bacteria to human cells. All major genotoxicity endpoints have been assessed in multiple models and multiple labs. A wide range of exposure conditions have also been assessed, including combined electric and magnetic fields, pulsed as well as sinusoidal fields, non-power-frequency fields and field intensities ranging from less than 1 microT to greater than 1000 microT.

Mutagenesis assays: Studies using a wide range of exposure conditions and assay systems have shown that power-frequency fields are not generally mutagenic. Five studies have found that power-frequency electric and magnetic fields are not mutagenic in bacteria or yeast [G3, G19, G21, G51, G101]. Studies of power-frequency fields and mutagenesis in mammalian cells done at field intensities of 50,000 microT and below have also been negative [G21, G58, G83, G92, G94]; but some studies [G56, G83] have suggested that 400,000 microT fields may be mutagenic.

Chromosome aberration assays: Of eleven studies of the ability of power-frequency fields to cause chromosome aberrations, eight [G1, G8, G38, G40, G41, G75, G96, G99] have found no consistent evidence of genotoxic effects. The remaining three studies showed some unreplicated evidence that power-frequency fields could cause chromosome aberrations. In 1984, Nordenson et al [E3] reported that exposure of human lymphocytes to spark discharges caused chromosome aberrations; but in 1995, Paile et al [G40] found no evidence for this effect. In 1991, Khalil and Qassem [G17] reported that a pulsed 1050 microT field caused chromosome aberrations in humans lymphocytes, but a similar 1994 study by Scarfi et al [G38] found no such effect. Finally, in 1994 Nordenson et al [G34] reported that exposure of mammalian cells to an intermittent 30 microT field caused chromosome aberrations, but that continuous exposure did not.

Sister chromatid exchanges (SCEs): Of the nine studies of the ability of power-frequency fields to cause SCEs, eight [G2, G5, G8, G12, G40, G42, G99, G102] have found no evidence of genotoxic effects. The only "positive" study is Khalil and Qassem [G17] who reported that a pulsed 1050 microT fields caused an increase in SCE's in humans lymphocytes; the study has never been replicated.

DNA strand breaks: None of the five studies of the ability of power-frequency fields to cause DNA strand breaks in cultured mammalian [G6, G20, G37, G99, G104] have found evidence of genotoxic effects.

Micronucleus formation assays: Of the 11 studies of the ability of power-frequency fields to enhance micronucleus formation, six [G12, G38, G40, G63, G65] found no evidence for such effects. Tofani et al [G45] reported that exposure of human lymphocytes to a 32-Hz field enhanced micronucleus formation; this effect was not found at 50-Hz or if the Earth's static geomagnetic field was eliminated. Scarfi et al [G68] reported that strong (1300 microT) pulsed fields enhanced micronucleus formation in human lymphocytes.

More recently, Simkó et al [G76, G93] reported that 48-72 hours of exposure to 800-1000 microT fields enhanced micronucleus formation in human tumor cells, but that no such effects were found for lower field intensities, shorter exposure times or in normal human cells. In a separate study, Simkó et al [G78] reported that 1000 microT fields enhanced micronucleus formation under some conditions, but not under many others. The scattered positive genotoxicity results reported by Simkó et al [G76, G78, G93] show no obvious pattern.

Pulsed fields: A number of studies have also examined pulsed ELF fields. Pulsed fields do not cause leukemia in leukemia-prone mice [G14], do not cause mutation in bacteria [G21, G62] or mammalian cells [G21], do not cause SCEs [G5, G17], do not cause DNA strand breaks [G37], do not cause micronucleus formation [G38], and do not cause cell transformation [G62]. One study has reported that 1050 microT pulsed fields cause chromosome aberrations [G17], but the report cannot be replicated [G38, G62].

Summary of genotoxicity studies: There are over 60 published studies of power-frequency fields and genotoxicity that include over 150 separate tests for genotoxicity activity. These assays are overwhelmingly negative, despite the fact that many have used huge field amplitudes. Of the studies that do report evidence for genotoxicity, most contain either a mix of positive and negative results, or ambiguous results. Since most of these publications contains multiple sub-studies, the presence of some studies with positive or mixed results would be expected from random chance. None of the positive reports of genotoxicity have been replicated, and several have failed direct attempts at replication. Many of the positive reports have also used exposure conditions (e.g., spark discharges, pulsed fields, fields of 20,000 microT and above) that are very different from those encountered in real-world exposure conditions.

Cell transformation assays have been widely used to study mechanisms of carcinogenesis. In a transformation assay, normal cells (typically fibroblasts) growing in cell culture undergo a set of changes when exposed to a carcinogen. These changes include loss of density-dependent inhibition of cell growth (loss of "contact inhibition") which causes cells to pile up ("focus formation"), and acquisition of the ability to grow in soft agar ("anchorage-independent cell growth"). The ability of an agent to induce transformation is a indication that the agent is a genotoxic carcinogen. The ability of an agent to enhance transformation by a known carcinogen is an indication of epigenetic activity.

In 1993, Cain et al [G29] reported that a 60-Hz field at 100 microT did not induce transformation, but that the field enhanced transformation induced by TPA (a known promoter). However, at meetings in 1993 and 1994 Cain reported that the observation of enhanced TPA-induced transformation could not be repeated (see Q21D). West et al [G35, H29] reported that 60-Hz fields induced cell transformation at field intensities from 1 to 1100 microT, but Saffer et al [G64] could not replicate this result. In addition, Balcer-Kubiczek et al [G55] reported that a 200 microT 60-Hz field did not cause transformation in two different transformation models, even with co-exposure to TPA; and in 1999 Snawder et al [G81] reported a similar lack of effects of 100 and 960 microT fields on cell transformation.

In 2000, Miyakoshi et al [G90] reported a lack of effect on cell transformation for fields of 5000 to 400,000 microT; but that these fields could inhibit transformation induced by ionizing radiation.

Jacobson-Kram et al [G62] have reported that pulsed magnetic fields do not cause cell transformation.

In an assay that is closely related to the transformation assay, Gamble et al. [G87] showed that exposure to 10-1000 microT fields did not "immortalize" normal cells or enhance the ability of ionizing radiation to "immortalize" cells.

In summary, there is no replicated evidence that power-frequency fields can induce or enhance neoplastic cell transformation.

Promotion of mammary tumors: The literature on promotion of chemically-induced breast cancer is extensive, but inconclusive. In 1991, Beniashvili et al [G16] reported that a 20 microT field could promote mammary tumors induced in rats by a chemical carcinogen (NMU). This unreplicated study is difficult to evaluate, as it has been published only in preliminary form, and critical experimental details are missing.

Loscher, Mevissen and colleagues [G26, G27, G32, G39, G49, G50, G86, K5] have conducted a series of breast cancer promotion studies in rats using a different chemical carcinogen (DMBA) (see Figure below).

Interpretations of the studies of Loscher, Mevissen and colleagues is complicated by several factors (see also Boorman et al [K8] and Anderson et al [K11]):

1. The dose of DMBA used in most of these studies is so high that essentially all animals develop breast cancer, even without promotion. As a result, the studies must be stopped before all tumors induced by DMBA have appeared, making it difficult to distinguish between induction of more tumors (promotion) and an increase in the growth rate of the tumors.

2. The authors use multiple endpoints for determining the presence of a promoting effect. In all studies, they assess the number of animals that have macroscopically-visible tumors. By this standard (see Figure below), one study using a 100 microT field [G26] shows significant promotion; the study using a more intense field, and the four studies using less intense fields do not show significant promotion. In some studies, the animals have also been examined histopathologically for the presence of smaller tumors (see Figure below). Two of these studies [G50, G86] indicate that 50-100 microT fields produce marginally-significant promotion that is not observed when only macroscopically-visible tumors were assessed. However, the study which showed promotion at 100 microT based on macroscopically-visible tumors [G26] did not show promotion when the assessment was based on histopathological determinations [G39].

3. The authors often use a test for significance that assesses the time to development of tumors, rather than the number of animals with tumors. In some cases, the authors report that tumors develop sooner in the animals exposed to power-frequency fields even though the number of animals with tumors was not significantly different. While such an effect may indicate an influence on tumor growth, it is not evidence for promotion (see Q17A).

4. The data has been summarized in a potentially misleading fashion. In 1995 Löscher and Mevissen [K5] published a summary claiming a linear relationship between magnetic flux density and breast cancer promotion. However, a comparison of that summary to their publications shows that the data in the summary is highly selected (see Figure below). First, the 30,000 microT experiment [G27] (which shows no promotion) was excluded; inclusion of that point destroys the "linear" relationship. Second, where data for both macroscopically-visible and histopathologically-determined tumor incidence were available, the "best" result was plotted; a consistent use of either end point destroys the linear relationship.

In 1998, Mevissen et al [G74] published a replication of their 100 microT experiment, in which they found an excess of "macroscopically-visible" tumors in the exposed group. In 1999, the group published a second replication [G74] of their 100 microT experiment, in which they found an excess of tumors in the exposed group based on histopathology that was not significant when only "macroscopically-visible" tumors were assessed.

In 1998, Ekström et al [G69] reported on the first independent attempt to replicate the Löscher and Mevissen studies. They found no evidence of breast cancer promotion at either 250 or 500 microT. Their data has been added to the figure which follows.

Also in 1998, the U. S. National Toxicology Program [G73] reported on a second independent attempt to replicate the Löscher and Mevissen studies. NTP found no evidence of breast cancer promotion at either 100 or 500 microT, with 3-4 independent studies at each exposure level. Their data has been added to the figure which follows.

In 1999 a third independent replication attempt by Anderson et al [G85] found no significant promotion of mammary tumors at either 100 or 500 microT.

See Boorman et al [K8] and Anderson et al [K11] for a detailed review of the animal breast cancer studies.

Breast Cancer "Promotion" in Rats

The breast cancer promotion studies of Löscher, Mevissen et al [G26, G27, G32, G39, G49, G50], Ekström et al [G69], the U. S. National Toxicology Program [G73], and Anderson et al [G85]. The figure shows the ratios (exposed/sham) of the number of rats with tumors at the end of each study (with 95% confidence intervals). Where Löscher, Mevissen et al reported data for both macroscopic and pathologically-confirmed tumors, both are shown. The dashed line is the "linear" relationship shown in the 1995 Löscher and Mevissen summary [K5] (the line is curved here because the field strength is shown on a log-scale). Typical 24-hour average residential fields are shown for comparison [F7, F22].

Promotion of skin tumors: Of the published studies of promotion of chemically-induced skin cancer [G11, G18, G23, G31, G44, G59, G77, G82], only one [G44] has reported statistically-significant promotion. The negative studies have used field intensities from 40 to 2,000 microT and exposure durations from 21-105 weeks, have tested both continuous and intermittent fields, and have used both promotion and co-promotion endpoints. The one positive study, by McLean et al [G44], exposed animals to 2,000 microT fields for 30 hours per week for 52 weeks.

Kumin et al [G71] reported that exposure of rats to 100 microT fields for 10.5 months enhanced UV-induced skin carcinogenesis. In contrast, Heikkinen et al [G105] reported that life-time exposure of mice to 1-130 microT fields did not increase the incidence of skin cancer induced by X-rays.

See figure below for a summary of the skin cancer promotion data.

Promotion of lymphoma: Studies of promotion of chemically-induced lymphoma by 2-1000 microT have found no evidence for promotion [G36, G61]. The two studies of promotion of lymphoma induced by ionizing radiation have also found no evidence for promotion at 130-1420 microT [G84, G105]. The Babbitt et al study [G84] is sufficiently large that promotion of lymphoma by greater than a factor of 1.10 can be ruled out. See figure below for a summary of the lymphoma promotion data.

Promotion of liver cancer: Multiple studies of promotion of chemically-induced liver cancer by 0.5 to 500 microT fields have found no evidence for such promotion [G28, G25]. See figure below for a summary of the liver cancer promotion data.

Promotion of brain cancer: In a study published in late 1999 Kharazi et al [G88] reported that life-time exposure of mice to a 1420 microT field did not promote brain cancers induced by ionizing radiation, however the number of brain tumors in all groups (exposed and unexposed) was very low. In 2000, Mandeville et al [G89] reported that 65 weeks of exposure of rats to 60 Hz fields at 2-2000 microT did not promote chemically-induced brain cancer.

Promotion of Lymphoma, Liver Cancer, Skin Cancer and Brain Cancer in Animals

Summary of the skin cancer, lymphoma, liver and brain cancer promotion studies. The vertical axis shows the ratio (exposed/sham) of the number of animals with tumors at the end of the experiment (except for the liver cancer promotion data where the ratio is the number of cancer foci at the end of the experiment). Skin tumor promotion data are from McLean and colleagues [G11, G18, G30, G44, G59], Rannug et al [G23, G31], Kumlin et al [G71], and Sasser at al. [G77]. Lymphoma promotion data are from Shen et al [G61], McCormick et al [G36], Babbitt et al [G84], and Heikkinen et al [G105]. Liver tumor promotion data are from Rannug et al [G25, G28]. Brain tumor promotion data are from Mandeville et al [G89]. All data are shown with 95% confidence intervals. Typical 24-hour average residential fields are shown for comparison [F7, F22].

Co-promotion: It has been suggested that power-frequency fields might be co-promoters; that is, that they could enhance the activity of other promoters, even though they have no genotoxic or promotional activity on their own. Published studies of co-promotion have shown little evidence for such activity [G11, G25, G30, G59, G77].

Promotion vs. growth enhancement: Interpretation of the tumor promotion studies is complicated by the observation in several studies [G17, G39] that exposure to power-frequency fields appears to speed the growth of chemically-induced tumors, or decrease the latent period for their appearance [G50, G84], rather than increase the actual number of tumors. Such an effect on growth would be of interest if it occurred at the field intensities to which people were actually exposed, but it would not be evidence for promotion [see Q17A].

Summary of promotion studies: There is no replicated evidence that power-frequency fields are promoters or co-promoters, and the few studies that have shown evidence for promotion have used field intensities well above those encountered in real-world settings.

Inhibition of DNA repair: The five published studies of the ability of power-frequency fields to inhibit the repair of DNA damage [G9, G10, G19, G47, G52] have found no evidence for such activity. These studies have used magnetic fields from 0.2 to 2500 microT, electric fields from 0.001 to 20 kV/m, and combined electric and magnetic fields. Both pulsed and sinusoidal fields have been assessed, and exposure durations have ranged from 10 minutes to 6 days.

In 2000, Chow et al [G97] reported that 400-1200 microT fields could enhance the repair of chemically-induced DNA damage in bacteria (this is the opposite of what an epigenetic carcinogen would do).

Enhancement of genotoxicity: Of the 13 published studies of the ability of power-frequency fields to enhance genotoxic damage produced by known chemical carcinogens, 12 [G3, G21, G45, G58, G65, G78, G83, G93, G94, G99, G101, G102] have found no consistent evidence for such activity. In 1989, Rosenthal and Obe [G8] reported that intense (2500 to 5000 microT) fields enhanced the cytogenetic damage produced in human lymphocytes by some chemical carcinogens; no such enhancement was seen at lower field intensities or for other chemical carcinogens.

Lagroye and Poncy [G63] reported that a 100 microT field enhanced cytogenetic damage produced in two of three mammalian cell lines by high doses of ionizing radiation. Walleczek et al [G79] reported a similar effect at 230-700 microT, and Miyakoshi et al [G92, G104] reported enhancement of x-ray induced mutagenesis at 5000 to 400,000 microT. In contrast, Ansari and Hei [G94] found no enhancement of x-ray induced mutagenesis at 100 microT, Maes et al [G99] found on evidence that 62-2500 microT fields enhanced x-ray induced chromosomal damage, and Nakasono et al [G101] found that a 14,000 microT field did not enhance UV-induced mutagenesis. Three studies have also found that life-time exposure to animals to power-frequency fields did not enhance the incidence of cancer induced by ionizing radiation [G84, G88, G105].

Enhancement of neoplastic transformation: See Q16D.

Other: In 2000, Chen et al [G98] reported that exposure of leukemia cells to 5-100 microT fields inhibited chemically-induced differentiation (an indicator of possible epigenetic activity); a 1993 study of the same system by Revoltella et al [Electro.Magnetobio. 1993; 12:135-146] had found no such effect at 200 microT.

In summary, there is little evidence that power-frequency fields have epigenetic activity in cell culture, and no evidence at all for epigenetic activity under real-world exposure conditions.

The magnetic fields associated with power lines, transformers and electrical appliances easily penetrate buildings or tissue and are difficult to shield. By contrast, power-frequency electric fields are easily shielded by conductive objects and have little ability to penetrate buildings or tissue. Because power-frequency electric fields have little ability to penetrate, it is generally assumed that any biologic effect from residential exposure to power-frequency fields must be due to the magnetic component of the field, or to the electric fields and currents that these magnetic fields induce in the body (for an opinion to the contrary, see King [F27]). In addition, the epidemiology that suggests that power-frequency fields might be associated with some types of cancer implicates the magnetic, rather than the electric, component of the field (see Q19L). As a result, most laboratory research has focused on power-frequency magnetic rather than electric fields, although there are some [L31, F27] who advocate that the electric, rather than the magnetic fields might be causally associated with cancer incidence.

Nevertheless, there have been laboratory studies of the genotoxic and epigenetic potential of power-frequency electric fields and combined power-frequency electric and magnetic fields [A14].

Genotoxicity Assays: There have been over a dozen studies of whether power-frequency electric or electric plus magnetic fields have genotoxic activity. Within this body of work, there is no replicated evidence for genotoxicity. These studies include:

• Benz et al [G4]; Kowalczuk and Saunders [G10a]: electric fields or electric plus magnetic fields are not mutagenic in mice.
• Morandi et al [G51]; Jacobson-Kram et al [G62]: electric fields or electric plus magnetic fields do not cause mutations in bacteria.
• Nordenson et al [E3]; Jacobson-Kram et al [G62]; Cohen et al [G1, G2]: electric fields or electric plus magnetic fields do not cause chromosome aberrations or SCEs in mammalian cells.
• Reese et al [G6]; Fiorani et al [G20]; Novelli et a [G13]; D'Agruma et al [G30b]: electric fields or electric plus magnetic fields do not cause DNA strand breaks in mammalian cells.
• Scarfi et al [G30A]: exposure of human lymphocytes to electric fields does not enhance micronucleus formation.
• Jacobson-Kram et al [G62]: electric fields do not cause transformation in mammalian cells.
• Nordenson et al [E3]: exposure of human lymphocytes to spark discharges caused chromosome aberrations, but Paile et al [G40] found no evidence for this effect in a replication study.

Assays for Epigenetic Activity: The studies of power-frequency electric or electric plus magnetic fields show no evidence of epigenetic activity. These studies include:

• Whitson et al [G0]: electric fields do not inhibit repair of DNA damage induced by UV radiation.
• Frazier et al [G10]: electric fields and electric plus magnetic fields do not inhibit repair of DNA damage induced by ionizing radiation.
• Cantoni et al [G47, G52]: electric fields and electric plus magnetic fields do not inhibit repair of DNA damage induced by peroxides, UV radiation or chemical carcinogens.
• Scarfi et al [G30A]: exposure of human lymphocytes to electric fields does not increase the incidence of micronuclei induced by a chemical carcinogen.

For further details on these and other studies of power-frequency electric fields, see Moulder and Foster [A14].

There are biological effects other than genotoxicity and promotion that might be related to cancer. In particular, agents that have dramatic effects of cell growth, on the function of the immune system, or on hormone balances might contribute to cancer without meeting the classic definitions of genotoxicity or promotion [A12, A8, A9, E4, L18].

There have been reports that power-frequency fields can enhance cell or tumor growth, but most studies have shown no effect. Many essentially harmless agents (e.g., temperature, pH, nutrients) affect the growth rates of cells and tumors, so effects of cell growth, by themselves, are not evidence for hazards [A8, A9, L18, L26]. However, the presence of certain types of effects on cell growth would be relevant to an evaluation of carcinogenic potential. It would be of particular relevance to cancer if an agent caused previously non-dividing normal (as opposed to tumor or transformed) cells to begin to divide, if the growth stimulation effect persisted after the agent was removed, and/or if the effect occurred at levels to which people were actually exposed.

Most studies of the effects of power-frequency magnetic fields on tumor growth have shown no effect [G7, G11, G25, G27, G28, G49, G57, G100, G103]; but four studies have reported enhanced tumor growth after exposure to fields of 50-2000 microT [G18, G26, G39, G50].

Of particular note are the studies by Sasser et al [G57], Morris et al [G80], Devevey et al [G91] and Anderson et al [G103] which found that prolonged exposure of leukemic animals to 2-2000 microT 50- or 60-Hz fields had no effect on leukemia progression or animal survival.

Most studies of effects of power-frequency magnetic fields on cell growth have also shown no effect [G1, G12, G20, G24, G40, G54, H1, H7, H27, H37, H38, H57, G93, G99]; but some have shown increased [G8, G42, G102] or decreased [G13, G48, J20] cell growth after exposure to strong (greater than 1,000 microT) fields.

Kwee and Rasmark [G46] reported increased mammalian cell growth after 30 minutes of exposure to 80-130 microT fields; but higher or lower field intensities, and longer or shorter exposure times, gave no effect. Wei et al. [H59] reported increased mammalian cell growth after long (6+ hours) exposures to fields of 90-120 microT, but no effect when the field was reduced to 60 microT. Chen et al [G98] reported stimulation of proliferation at 100 and 1000 microT.

Of particular interest is a study by Zhao et al [H45] which found that both sham exposure and exposure to 100-800 microT fields enhanced cell growth. The effect was shown to be due to a 0.1-0.8 °C rise in temperature caused by the double-wound coils used for the sham exposure. Whether other reports of effects on cell growth might be due to heating is unknown, but temperature rises from sham exposures have been reported by others (e.g., Rosenthal and Obe [G8]).

In summary, there have been no reported effects on cell proliferation or tumor progression that suggest a potential for carcinogenesis, and there have been no reports of effects at all for fields below 50 microT.

In the early 1970's there was speculation that the immune system had a major role in preventing the development of cancer; this theory was known as the "immune surveillance hypothesis" [E4]. If this hypothesis were true, then damage to the immune system could effectively cause cancer. Subsequent studies have shown that this hypothesis is not generally valid [E4, E7]. Suppression of the immune system in animals and humans is associated with increased rates of only certain types of cancer, particularly lymphomas [E7]. Immune suppression has not been associated with an excess incidence of leukemia, except for viral-induced leukemia in animals; and has not been associated with brain or breast cancer in either animals or humans [E4, E7].

Some studies have shown that power-frequency fields can have effects on cells of the immune system [K1], but no studies have shown the type or magnitude of immune suppression that is associated with an increased incidence of lymphomas. Of particular relevance are four recent studies:

• a study of human volunteers that showed no effects of a 10 microT field on immune function [E19];
• a study of primates that found that combined electric (6 or 30 kV/m) and magnetic (50 or 100 microT) fields had no consistent effects on the immune system [H23];
• a comprehensive study in mice [H32] that found that neither continuous (2-1000 microT) nor intermittent (1000 microT) fields had any effect on immune function;
• a study of mice [H33] that found some effects on immune function at 2000 microT, less effects at 200 microT, and no significant effects at 2 or 20 microT.

The "power line-melatonin" hypothesis: Some investigators have hypothesized that power-frequency fields might suppress the production of hormone melatonin, and that melatonin might have "cancer-preventive" activity [H7, L4]. There are reports that electric fields and static magnetic fields can affect melatonin production, but studies using power-frequency magnetic fields have largely shown the absence of such effects. The second component of the hypothesis, that decreased melatonin levels are associated with increased cancer, is also unproven.

Effects of power-frequency magnetic fields on melatonin in non-human primates and in humans: In a large study in baboons, Rogers et al [H24] found that exposure to combined 60-Hz electric (6 or 30 kV/m) and magnetic fields (50 or 100 microT) had no affect on night-time melatonin. However, in a two-monkey pilot study, they found some evidence that the exposure might be effective is decreasing night-time melatonin if the fields were turned on and off very rapidly [H24].

Five studies of exposure of human volunteers [E18, E19, E20, E23, E25] have found no evidence that either continuous or intermittent fields at 1-28 microT affected night-time melatonin levels. A sixth study [E21] reported evidence that the night-time peak was delayed, but that overall melatonin levels were not affected. A study of occupational exposure to power-frequency fields in female garment workers showed some ambiguous evidence for a decrease in night-time melatonin production [E24].

Effects of power-frequency magnetic fields on melatonin in non-primates: In a series of four studies in rats, Kato et al [H8, H11] reported that 1 microT fields caused small (20-25%) but inconsistent decreases in night-time melatonin levels. Also in rats, Loscher and Mevissen have reported that 0.3-1.0 and 10 microT fields produced small (15-25%) decreases in night-time melatonin [G32, G49], but that larger fields did not [G50]. In contrast, Selmaoui and Touitou, [H20, H48] reported that 1 microT fields had no effect of melatonin levels while 100 microT caused a 25-40% decrease in young, but not in adult rats; and Bakos et al [H19, [H49] reported that 1, 5, 100 and 500 microT fields had no affect on night-time melatonin levels in rats.

In 1998, John et al [H39] reported that 1000 microT fields had no effect of night-time melatonin levels in rats exposed for periods of time ranging from 1 hr to 6 weeks. Also in 1998, Löscher et al [H39] reported that 100 microT fields had no effect of night-time melatonin levels in rats exposed for periods of time ranging from 1 day to 13 weeks.

In the only study to date in mice, Heikkinen et al [H47] found that 17 months of exposures to 1.3, 13 or 130 microT fields at 50 Hz had no effect on melatonin levels.

In a series of publications, Yellon and colleagues [H13, H30, H31, H35, H42] studied the effects of 10 and 100 microT fields on melatonin levels in the Djungarian hamster. In some experiments, decreases of night-time melatonin of 20-50% were observed; but in most experiments no effects at all were seen, and in one experiment an increase was observed. Niehaus et al [H34], working with the same hamsters, found that neither sinusoidal or pulsed fields affected night-time melatonin levels in these hamsters. In 1999, Wilson et al [H46] reported that some exposure regimens caused decreases in night-time melatonin at 100 microT, but found no effects at 50 microT.

The most recent study from Yellon et al [ H42] concludes that:
"recent evidence in the Siberian hamster suggests that magnetic field exposure effects on the melatonin rhythm... cannot be distinguished from normal variation between replicate studies in sham-exposed controls."

In two studies with sheep, Lee et al [H9, H16] found that 4 microT plus 6 kV/m had no effect on night-time melatonin levels.

Melatonin and anti-cancer activity: In the 70's and 80's there was interest in using melatonin as an anti-cancer agent, but clinical trials of melatonin continue to show that it is largely ineffective. There are reports that melatonin levels are decreased in some cancer patients, particularly those with breast cancer, but there is no evidence for a causal link.

There is some evidence that melatonin can inhibit the induction of breast cancer by chemical carcinogens; and that inhibition of melatonin production can enhance the induction of breast cancer by chemical carcinogens. However, some studies have not found one or both of these effects, and at least one group has reported that melatonin enhanced the chemical induction of breast cancer. There is also evidence that melatonin can retard the growth of transplanted immunogenic tumors, and that inhibition of melatonin production can enhance the growth of such tumors. However, there are also reports of stimulation of the growth of immunogenic tumors by melatonin. There are no reports that melatonin affects the development of spontaneous tumors, or that it affects the induction or progression of leukemia.

In cell culture there is evidence that melatonin can inhibit cell growth in some breast cancer cell lines [H7, H62], but melatonin does not appear to have a general growth inhibitory effect on tumor cells [H41]. There is also evidence that melatonin is an effective free-radical scavenger and that it can protect cells from the genotoxic effects of ionizing radiation and chemical carcinogens [H27].

In summary: Neither component of the melatonin hypothesis, that power-frequency fields suppress melatonin, or that decreased melatonin causes an increase in cancer, have strong experimental support. In humans, there is essentially no evidence to support either component of the hypothesis. What evidence exists, suggests that any effects would be confined to breast cancer, and possibly to other hormonally-sensitive cancers such as prostate cancer.

While the laboratory evidence does not suggest a link between power-frequency magnetic fields and cancer, numerous studies have reported that these fields do have "bioeffects", particularly at high field strength [A7, K1, M4, M6]. Power-frequency fields intense enough to induce electric currents in excess of those that occur naturally (above 500 microT, see Q8) have shown reproducible effects, including effects on humans [M4, M6].

If a reproducible biological effect is defined as one that has been reported in the peer-reviewed literature by more than one laboratory, without contradictory data appearing elsewhere; then there may be no reproducible effects below 50 microT [A7, A12, A15, K9]. While there are reports of effects for fields as low as about 0.5 microT, none of these reports have been validated.

The lack of validation of the "positive" laboratory studies could be due to many factors:

• Some reports on the biological effects of power-frequency fields have never been published in the peer-reviewed literature, and cannot be scientifically evaluated or replicated.
• No attempts have ever been made to replicate some of the published reports of biological effects; and one positive report, standing in isolation, is impossible to evaluate.
• When attempts have been made to replicate many of the published studies, these replications have often failed to show the effect [H1, H4, H10, H13, H14, H15, H22, H44, H50, A15, H55-H58, H60, K9].
• The investigators in this field use a wide variety of biological systems, endpoints, and exposure conditions, which makes studies extremely hard to compare and evaluate.
• Variation in exposure systems, plus the lack of adequate exposure details [F19] make many reports impossible to reproduce.
• The possibility that some of the positive reports were fabrications in the first place cannot be overlooked [L34, L35, L39].

The known biological mechanisms through which high-amplitude (greater than 500 microT) power-frequency magnetic fields cause biological effects are not relevant to fields below about 50 microT. These high-field effects involve induced electric currents, and the currents induced in the body by fields of less than 50 microT are qualitatively similar to, but much weaker than, the currents that occur naturally [A7, A12, A14, F3, F23, F34 and see Q8].

If sinusoidal power-frequency fields below 5 microT do actually have biological effects, the mechanisms must be found, in Adair's [F3, F12] words: "outside the scope of conventional physics".

The considerations discussed in Q18B show that the interactions of sinusoidal power-frequency fields with the human body are very weak at typical environmental field levels. Numerous investigators have speculated about how power-frequency fields might overcome signal-to-noise problems via resonance or signal amplification mechanisms [F4, F17,H26].

Induced currents: Power-frequency electric and magnetic fields can induce electric currents, and these currents can cause biological effects if they are sufficiently strong [F23, M6, M8]. However, the currents induced in the body by fields of less than 1 kV/m or 50 microT are weaker than those that occur naturally in the body [F3, F17, F23, M6, M8]. Therefore, if sinusoidal power-frequency fields of the magnitude encountered in residential settings do have biological effects, they are unlikely to be mediated by induced electric currents.

Magnetic Biological Material: Small magnetic particles (magnetite, Fe3O4) have been found in bacteria that orient in the Earth's static magnetic field, and these particles may also exist in fish, honeybees and birds [F4]. The presence of magnetite in mammalian cells is still unproven. Kirschvink [F4] has suggested that power-frequency magnetic fields could cause biological effects by acting directly on such particles. However, calculations show that this would require 50/60 Hz fields of 2-5 microT or above [F4, F12, H11, F23].

Free Radical Reactions: Static (DC) magnetic fields can affect the reaction rates of chemical reactions that involve free radical pairs [F18, F37]. Since the radicals involved have lifetimes in the microsecond range, and power-frequency fields have a cycle time in the millisecond range, a power-frequency field acts like a static field during the time scale in which these reactions occur. The effects of the power-frequency field would be additive with the Earth static field (30-70 microT), so no detectable biological effects would be expected below about 50 microT [F18, F23, F33]. In addition, if one were to hypothesize that biological effects mediated by such free radical reactions were involved in carcinogenesis, the relevant studies would be those using static fields; and studies of the genotoxic and epigenetic activity of static fields have been overwhelmingly negative (see Static Electromagnetic Fields and Cancer FAQs).

Eichwald and Walleczek [F32] have made a theoretical argument which suggests that biochemical effects mediated by the radical-pair mechanism could account for effects of power-frequency fields of 1000 microT or more; and Eveson et al [F37] have shown experimental evidence that magnetic fields as low as 1000 microT can have effects on free radical reactions. Adair [F33], on the other hand, has presented theoretical arguments that effects due to the radical-pair mechanisms are wildly implausible at levels of 5 microT or below.

Resonance Theories: Some of the biophysical constraints could be overcome if there were resonance mechanisms that could make cells (or organisms) uniquely sensitive to power-frequency fields. Several such resonance mechanisms have been proposed, most recently by Lednev and Blanchard/Blackman [H26]. So far, none of these theories have survived scientific scrutiny [F3, F5, F23], and much of the experimental evidence that prompted the speculations cannot be independently reproduced [H1, H4, H10, H17]. There are also severe incompatibilities between known biophysical characteristics of cells and the conditions required for such resonances [A7, F3, F5, H26, F23, F26]. Note also that resonance theories would predict that biological effects would be different in North America (60 Hz) than in Europe (50 Hz).

The biophysical barriers to biological effects discussed in Q18B and Q18C presume that 50/60-Hz sinusoidal power-frequency fields are the only time-varying electromagnetic fields found in conjunction with the transmission, distribution, and use of electric power. If this presumption is not true, and large transients and/or higher-frequency harmonics are present; then it is possible that electric currents stronger than those that occur naturally in the body could be induced at field levels that are present in residential and occupational settings. Such large currents might provide a route to biological effects.

A 2000 study of transients in US homes [F35] found that transients do occur, but did not directly address the issue of whether they might be powerful enough or frequent enough to cause biological effects.

New studies, particularly epidemiologic studies, appear frequently. When these studies show "positive" effects they generate considerable media coverage. When they fail to show "positive" effects they are generally ignored. This section will cover the more recent (1995 to present) studies in some detail.

In 1993-94 five European residential exposure studies were published [C16, C17, C18, C19, C21]. The childhood study from Sweden [C19] showed the highest relative risks, and drew the most attention. In contrast to the earlier US studies which assessed exposure from both distribution and transmission lines, these new studies were restricted to high voltage power lines and substations. Exposure was assessed by spot measurements [C19, C21], calculated retrospective assessments[C16, C17, C19, C21], and distance from power lines [C18, C19, C21].

The authors of the three Scandinavian childhood cancer studies [C16, C17, C19] have produced a combined analysis of their data [B4]. This analysis is based on retrospective calculated fields, the only measure of exposure common to all three studies. The range of RRs from this meta-analysis are shown below in comparison to prior and subsequent studies.

Two 1996 studies of childhood brain cancer and residence near powerlines show no evidence for an association with either measured fields [C29] or wire codes [C28, C29]. A 1997 European study [C33] of childhood leukemia, lymphoma, brain cancer, and overall cancer shows no evidence for an association with either distance from transmission lines or calculated fields. In 1997 a second European study [C34] found a non-significant elevation of leukemia in children whose bedrooms had average fields above 0.2 microT. A third 1997 study [C35], which is discussed in detail in Q19H, found no association of childhood leukemia with either measured fields or wire-codes. A 1999 study [C44], which is discussed in detail in Q19J, found no association of childhood leukemia with either measured fields or wire-codes.

A 2001 German study [C59] found no significant association of 24-hour average magnetic fields and childhood leukemia; but when pooled with previous German studies [C34], a statistically-significant association was seen for 24-hour average magnetic fields of 0.4 microT and above.

Also see the discussion of the childhood leukemia studies in Q13B.

The Scandinavian studies of adults living near high voltage lines show no increases in overall cancer, leukemia, or brain cancer [C18, C21, C31]. Only the 1997 study from Taiwan [C32] shows any evidence for an association of adult cancer and residence near transmission lines.

Since 1994, at least 20 major occupational studies of cancer and occupational exposure to power-frequency fields and cancer have been published. These studies deal with:

• leukemia [D21, D24, D25, D26, D26a, D28, D29, D31, D40]
• brain cancer [D21, D24, D25, D26, D27, D28, D31, D35, D42]
• male and female breast cancer [D22, D23, D31, D33, D34, C41]
• lymphoma [D25, D26, D26a, D31, D39]
• lung cancer [D25, D26, D26a, D30, D31]
• other cancers [D24, D25, D26, D31]
• overall cancer rates [D21, D25, D26, D26a, D31]

Of the 10 studies of leukemia published in 1995 or later, one [D28] showed some evidence for a statistically significant increase in at least one group that was "exposed to power-frequency magnetic fields". Two other studies [D25, D40] reported an increased risk for electric, but not for magnetic field exposure; the other recent studies of occupational exposure to electric fields contradict this finding [D26, D29]. For the studies as a whole the median RR was 1.2, but values as high as 1.8 or as low as 1.0 (no effect) are compatible with the data.

Of the 5 studies of lymphoma published in 1995 or later, none found evidence for a statistically significant increase in any groups exposed to power-frequency magnetic fields, but one study [D39] found an increase in workers exposed to power-frequency electric fields. For the studies as a whole the median RR was 1.2, but values as high as 1.5 or as low as 1.0 (no effect) are compatible with the data.

Of the 10 studies of brain cancer published in 1995 or later, one [D21] showed evidence for a statistically significant increase in at least one group that was "exposed" to magnetic fields, and one [D26] reported an increase for "exposure" to electric fields. For the studies as a whole the median RR was 1.15, but values as high as 1.5 or as low as 1.00 (no effect) are compatible with the data. Also see the 2001 review by Kheifets et al [B19].

Of the 5 studies of overall cancer published in 1995 or later, only one [D21] showed evidence for increase in overall cancer in at least one "exposed" group. For the studies as a whole the median RR was 1.05, but values as high as 1.10 or as low as 0.95 (protection) are compatible with the data.

The new studies of lung cancer (Q19D) and breast cancer (Q19C) are covered separately.

In 1999 Kheifets et al. [B17] published a combined reanalysis of 3 earlier (1994-1995) [D10, D12, D21] occupational exposure studies. The combined analysis (see Figure below) shows a weak association between exposure to power-frequency fields and both brain cancer and leukemia. However, even in the most highly-exposed groups, the associations are not strong or statistically significant.

Leukemia and Brain Cancer in Electric Utility Workers

Brain cancer and leukemia as a function of cumulative expose to power-frequency fields in the electric utility industry, based on a combined analysis [B17] of three separate studies [D10, D12, D21]. The study by Thériault et al. [D12] included two distinct group of workers in Ontario and Quebec. The data is shown as relative risks with 95% confidence intervals. Adapted from Kheifets et al. [B17].

There are some laboratory studies [G16, G26, G50] that suggest that power-frequency fields might promote chemically-induced breast cancer (Q16B), and a biological mechanism has been proposed that could explain such a connection (Q17C).

McDowall et al [C4] found no excess female breast cancer (and no male breast cancer at all) in adults living near transmission lines or substations; and Li et al [C32] found no excess female breast cancer in adults living near transmission lines. More recently, Feychting et al [C38, C52] found no significant excess of male or female breast cancer in adults living near transmission lines; and Coogan et al [C41] found no excess breast cancer in women with occupational and/or residential exposure to power-frequency fields.

Five studies [C23, C39, C41, C55, C56] have found no excess breast cancer in women who used electric blankets. A number of studies have reported an elevated incidence of male breast cancer in electrical workers [D4, D5, D6, D20]; but other studies have found no such excess [D7, D11, D12, D14, D18, D33].

In 1994, Loomis et al [D15] reported an elevated incidence of female breast cancer in occupations with presumed exposure to power-frequency fields. The occupations that showed an excess incidence of breast cancer were "male-dominated". Breast cancer mortality is known to be elevated among women in professional and technical jobs in general; this is because women working in male-dominated jobs tend to have reproductive histories (for example, no pregnancies, delayed child-bearing, not breast-feeding) that increase their risk for breast cancer. Cantor et al [D22], analyzing the same database, found no evidence for an elevated incidence of female breast cancer in occupations with presumed exposure to power-frequency or radio-frequency fields.

A 1996 study in this area [D23] was preceded by a rather misleading press release, whose title was "Occupational exposure to magnetic fields increases risk of breast cancer". The study itself does not support the title of the press release. This study is based on breast cancer registry information, with exposure assessed on the basis of the "most representative job". Occupations were grouped in categories according to "potential for exposure to 60-Hz magnetic fields", and no estimates of actual exposure levels or exposure duration were made. Less than 1% of the women were in jobs with "high potential for exposure". The RR for the "high potential exposure" group was elevated, but the elevation was not statistically significant. For medium and low potential exposure, RRs were not elevated.

In 1998, Johansen et al [D31], Coogan et al [C41], and Petralia et al [D34] reported that occupational exposure to power-frequency fields was not associated with excess female breast cancer. In 2000, Feychting et al [C52] reported that neither occupational nor residential, nor a combination of residential and occupational exposure, to power-frequency fields were associated with increased risk of female breast cancer.

This area of research was reviewed in detail in 1999 by Kheifets and Matkin [B15] and Brainard et al [B16] and in 2001 by Erren [B21]. All three reviews concluded that no human health risk has been proven, but that the data was insufficient to prove that a small effect could not exist.

In 1994, Armstrong et al [D16] reported that utility workers exposed to short-duration pulsed electromagnetic fields (PEMFs) had increased lung cancer. The association of lung cancer with PEMF was moderately strong, and there was evidence for a dose-response relationship. The workers with the highest exposure to PEMFs had an elevated lung cancer risk compared to workers with lower levels of exposure; but they had a lower lung cancer rate than members of the general public. No relationships were found between PEMF exposure and any other type of cancer.

Previous studies of power-frequency fields and lung cancer had found no association. In a summary of pre-1992 occupational studies, Hutchison [B2] reports a summary relative risk of 0.8 (0.7-0.9), indicating that workers with exposure to power-frequency fields have less lung cancer than expected. Similarly, Theriault et al [D12] reported a RR of 1.0 (0.7-1.5) for lung cancer in electrical workers with the highest magnetic field exposure.

A 1997 study by Savitz et al [D30] found no association of lung cancer with either exposure to power-frequency magnetic fields or exposure to PEMFs.

The most difficult issue with the Armstrong report [D16], is the definition of "PEMF" exposure. The dosimetry is based on readings from a dosimeter that was designed to respond to signals having an electric field component greater than 200 V/m at 2-20 MHz; but this isn't what the dosimeter actually responds to [D17]. In the utility environment this dosimeter is exquisitely sensitive to radio transmissions near 150 MHz, a band that is now (but only in the 1990's) used for portable radio communication [D17]. So the job categories in which the Armstrong report [D16] found excess lung cancer are actually the jobs that involve proximity to the use of portable radios; and the vast majority of the reported excess lung cancer occurred before the use of these radios became common.

The fields close to appliances that contain AC electric motors can exceed 100 microT and 200 V/m. If these appliances are used very close to the body, as electric razors and hair dryers are, there can be large exposures of small parts of the body. There have been epidemiologic studies that have looked at the relationship between the use of electric appliances and cancer [C6, C8, C11, C12, C22, C23, C28, C29, C30, C37, C51, C55, C56]. These studies have shown little consistent association between the use of electric appliances and cancer incidence; although one of these studies [C22] has actually shown a decrease in adult leukemia among users of personal electric appliances.

A recent large study in this area is Hatch et al [C37], run in parallel with the Linet et al [C35] power line study discussed in Q19H. As with other studies, this study show no consistent association of childhood leukemia with use of electrical appliances.

It is frequently said that Sweden or Denmark have decided to regulate the magnetic fields produced by power lines, or have decided to move power lines away from schools. However, statements over the years from officials in both countries [L9, L12, L19, L27] show no evidence that they are either regulating fields from the lines nor ordering lines to be moved away from schools.

In 1996, the Swedish government announced a "precautionary principle" [L27]:
- "The [Swedish] national authorities recommend a precautionary principle based primarily on non-discountable cancer risks..." - "The research findings presented hitherto afford no basis for and cannot be said to justify any limit values or other compulsory restrictions on low-frequency electrical and magnetic fields..." - "The national authorities join in recommending the following precautionary principle: If measures generally reducing exposure can be taken at reasonable expense and with reasonable consequences in all other respects, an effort should be made to reduce fields radically deviating from what could be deemed normal in the environment concerned. Where new electrical installations and buildings are concerned, efforts should be made already at the planning stage to design and position them in such a way that exposure is limited..."

The Swedish statement includes a number of examples where attempts to measure the costs of mitigation were made. Assuming a childhood leukemia incidence of 1 case per 25,000 per year, and a RR of 2.7, the cost per case avoided varies from 200,000 US$$ to 50,000,000 US$$. To put this in some perspective, the document notes that it is generally considered "reasonable" to spend up to 1,000,000 US$$ to avoid a death due to ionizing radiation exposure.

The inherent biophysical problems (see Q18B) with explaining how environmental power-frequency fields could cause biological effects might be overcome if a biological mechanism for amplifying the fields could be identified. A number of such amplification models (see Q18C) have been proposed, most of which are based on some type of resonance between the power-frequency field and the Earth's static geomagnetic field.

In 1995, Bowman et al [C27] hypothesized that the risk of childhood leukemia might be related to specific combinations of static (geomagnetic) and power-frequency fields. Childhood leukemia data from the Los Angeles were analyzed on the basis of these combinations. No correlation of cancer with measured static or power-frequency fields were found; but the authors do claim a positive trend for the combined power-frequency and static field data. An issue not addressed by the authors is that all resonance theories require a specific orientation between the power-frequency and the static field. Thus it should not be the total static field that matters, but only the component of the static field with the right orientation to the power-frequency field.

A case-control study of power-lines and childhood cancer, done by the U.S. National Cancer Institute, was published in July 1997 [C35]. This was the largest such study to that date (the 1999 McBride et al [C44] study discussed in Q19J is even larger), and finds no association between measured fields and childhood leukemia, or between wire-codes and childhood leukemia.

• For a time-weighted average bedroom field above 0.2 microT, the study found a RR of 1.2 (0.9-1.8), with no statistically-significant dose trend.
• For a "very-high current configuration" wire code (as defined by Wertheimer and Leeper [C1]), the study found a RR of 0.9 (0.5-1.6).

From the authors' abstract [C35]

Previous studies found associations between childhood leukemia and surrogate indicators of exposure to magnetic fields (the power line classification scheme known as wire coding), but not between childhood leukemia and measurements of 60-Hz residential magnetic fields...

We enrolled 638 children with acute lymphoblastic leukemia (ALL)... and 620 controls in a study of residential exposure to magnetic fields generated by nearby power lines. In the subjects current and former homes... [we] measured magnetic fields for 24 hours in each child's bedroom... A computer algorithm assigned wire-codes to the subject main residence... and to those where the family has lived during the mother's pregnancy with the subject...

The risk of childhood ALL was not linked to time-weighted average residential magnetic fields... The odds ratio for ALL was 1.24 (95% confidence interval 0.86-1.79) at exposures of 0.2 microT (2 mG) or greater... The risk of ALL was not increased among children whose residence was in the highest wire code category [odds ratio of 0.88 (0.48-1.63)]...

Our results provide little evidence that living in homes characterized by high measured magnetic field levels or by the highest wire code category increases the risk of acute lymphoblastic leukemia in children.

In recent years, several commissions and expert panels have concluded that there is no convincing evidence that high voltage power lines are a health hazard or a cause of cancer. And the weight of the better epidemiological studies, including that by Linet et al now supports the same conclusion. It is sad that so many hundreds of millions of dollars have gone into studies that never had much promise of finding a way to prevent the tragedy of cancer in children. The many inconclusive and inconsistent studies have generated worry and fear and have given peace of mind to no one. The 18 years of research have produced considerable paranoia, but little insight and no prevention. It is time to stop wasting our research resources. We should redirect them to research that will be able to discover the true causes of the leukemia clones that threaten the lives of our children.

Two separate Canadian studies of power line exposure and childhood leukemia were published in 1999. McBride et al [C44], the larger of the two studies, found no associations between any measures of exposure and the incidence of childhood leukemia. Green et al [C45 and C46], a smaller study, did find an association between childhood leukemia incidence and some measures of exposure.

McBride et al [C44] is the largest study to date (399 cases and 399 matched controls), and it finds no evidence for any association between power lines and childhood leukemia. The study is notable for its size and for the wide range of exposure metrics tested. In conjunction with the 1997 US-NCI study [C35] discussed in the previous question, this new study eliminated essentially all grounds for suggesting a causal association between exposure to power line fields and the incidence of childhood leukemia.

The findings of the McBride et al [C44] study:

• Fields measured with personal monitors (48-hr averages) were not associated with childhood leukemia, with:
  • a RR of 0.6 (0.3-1.2) for those with the highest magnetic field exposures (greater than 0.27 microT).
  • a RR of 0.8 (0.5-1.5) for those with the highest electric field exposures (greater than 25 V/m).

• Contemporary measured fields in residences were not associated with childhood leukemia, with:
  • a RR of 0.7 (0.4-1.3) for those with the highest magnetic field exposures (greater than 0.27 microT).

• Historic magnetic field reconstructions were not associated with childhood leukemia, with:
  • a RR of 0.6 (0.3-1.1) for those with the highest exposures two years prior to diagnosis (greater than 0.27 microT).
  • a RR of 1.0 (0.6-1.9) for those with the highest average life-time exposure (greater than 0.27 microT).

• Wire codes were not associated with childhood leukemia, with:
  • a RR of 1.2 (0.6-2.3) for those living at the time of diagnosis in a house with a "very high current configuration" (as defined by Wertheimer and Leeper [C1]).
  • a RR of 0.8 (0.4-1.6) for those living two years prior diagnosis in a house with a "very high current configuration" (as defined by Wertheimer and Leeper [C1]).
  • a RR of 1.2 (0.7-1.9) for those living at the time of diagnosis in a house with a "high current configuration" (as defined by Kaune and Savitz [F6]).

The specific findings of the Green et al [C45, C46] study:

• Fields measured with personal monitors (48-hr averages) were associated with childhood leukemia, with:
  • a RR of 2.4 (1.0-5.5) for those with the highest magnetic field exposures (greater than 0.14 microT).
  • a RR of 4.5 (1.3-16) for those with the highest magnetic field exposures (greater than 0.14 microT), when the data was "adjusted for average power consumption"
  • a RR of 0.3 (0.1-0.9) for those with the highest electric field exposures (greater than 12 V/m).

• Contemporary measured fields in residences were not associated with childhood leukemia, with:
  • a RR of 1.1 (0.3-4.1) for those with the highest bedroom magnetic fields (greater than 0.13 microT).
  • a RR of 1.5 (0.4-4.9) for those with the highest residential magnetic fields (greater than 0.15 microT).

• Contemporary measured fields outside residences were associated with childhood leukemia, with:
  • a RR of 3.5 (1.1-10.5) for those with the exterior measured magnetic fields (greater than 0.15 microT).

• Wire codes were not associated with childhood leukemia, with:
  • a RR of 0.8 (0.2-3.0) for those living prior to diagnosis in a house with a "very high current configuration" (as defined by Wertheimer and Leeper [C1]).
  • a RR of 0.9 (0.3-2.1) for those living prior to diagnosis in a house with a "high current configuration" (as defined by Kaune and Savitz [F6]).

These studies are particularly important in view of the conclusion in the 1996 U.S. National Academy of Science (NAS) report (Q27E) that the only epidemiological evidence for a link between power lines and cancer was the association between high wire codes and leukemia. The NAS report quoted a RR of 1.5 (1.2-1.8) for this association based on the four then-available studies. Merging NAS data with the four subsequent wire-code studies [C35, C43, C44, C45] gives a summary RR of 1.05 (0.90-1.22), with very high heterogeneity.

It should also be noted that some (such as the NIEHS "working group" [A11] discussed in Q27F) have reinterpreted the 1997 NCI study [C35] as positive, by reanalyzing the data based on 0.3 microT measured residential fields as the "cut-point" for determining who was exposed. An analogous assessment of the McBride et al [C44] data gives a RR of 0.7 (0.4-1.2). Green et al [C45] cannot be analyzed in this fashion, because data is not provided for cut-points above 0.15 microT.

The December 4 1999 issue of Lancet carried a report on a large study of powerlines and childhood cancer from the UK [C49], and a summary of a smaller study of power lines and childhood leukemia from New Zealand [C48, C51]. Both studies report that there is no significant association of childhood cancer with exposure to power line fields. In November 2000, the investigators published a follow-up study in which they looked at additional cases and at all external sources of power-frequency fields (that is, substations and distribution lines as well as transmission lines) [C58].

The UK study [C49, C58] is a case-control study of 3380 children with cancer and a similar number of matched controls. Power-frequency magnetic fields were measured in residences and schools, and this was used to calculate the average exposure for the year prior to diagnosis.
According to the authors [C58]:

"Our results provide no evidence that proximity to electricity supply equipment or exposure to magnetic fields associated with such equipment is associated with an increased risk for the development of childhood leukemia nor any other childhood cancer."

• Total leukemia: 0.4 (0.1-1.9)
• Brain cancer: 0.5 (0.1-3.8)
• Other cancer: 0.9 (0.3-3.0)
• Total cancer: 0.6 (0.2-1.6)

• Total cancer in children exposed to fields of 0.4 microT and above: 0.5 (0.2-1.6)

• Total leukemia: 0.8 (0.5-1.3)
• Brain cancer: 1.1 (0.6-2.1)
• Total cancer: 0.9 (0.6-1.3)

The New Zealand study [C48, C51] was much smaller (121 cases and matched controls), assessed only leukemia, and assessed exposure to both electric and magnetic fields. The relative risks were:

• Leukemia and magnetic fields greater than 0.2 microT: 3.3 (0.5-24)
• Leukemia and electric fields greater than 14 volts/meter: 1.3 (0.2-7)

In a commentary accompanying the first part of the study [C50], Repacholi and Ahlbom from the World Health Organization "EMF Project" argue that the UK study is not "definitive" because it did not assess "transients", because only a relatively small number of children were found who were exposed to average fields above 0.4 microT, and because the study "was unlikely to affect the results of previous meta-analyzes and reviews suggesting a weak link between power-frequency magnetic field exposures and childhood cancer."

When these new results are added to all previous studies, the summary relative risk for childhood leukemia and exposure to power-frequency fields is about 1.2 if the original Wertheimer and Leeper study [C1] is included and 1.1 if it is excluded.

Because power-frequency electric fields have little ability to penetrate, it is generally assumed that any biologic effect from residential exposure to fields from power lines must be due to the magnetic component of the field, or to the electric fields and currents that these magnetic fields induce in the body. For this reason, most epidemiological studies have focused on magnetic field exposure. However, there are some [L31, F27] who have advocated that the electric, rather than the magnetic fields might be causally associated with cancer incidence.

The existing residential epidemiology provides even less support for an association with electric fields than for an association with magnetic fields [A14]. First, residences along high-current distribution lines, where excess rates of childhood leukemia have been reported in some U.S. studies, do not have elevated electric fields [C6, C12, F7]. Second, all but one of the residential epidemiological studies that have looked at both electric and magnetic fields have found that the association (where there is any) is with the magnetic, not the electric field [C6, C12, C33, C44, C46, C48, C51].

The exception is a 1996 study by Coghill et al [C42], which measured electric and magnetic fields in bedrooms of 56 boys who had developed leukemia and an equal number of healthy controls. The investigators reported that the 24-hour mean electric fields in the bedrooms of the leukemic children was 14±13 V/m, compared with 7±13 V/m for the controls. The validity of the Coghill et al [C42] study can be questioned on several grounds. First, the study had an unblinded design, so that those doing the field measurements knew whether the homes were those of cases or controls. Second, the study recruited its subjects through media requests, and because of the great media attention to the possible hazards of power line fields, it is quite possible that parents of children with cancer, who lived near high voltage lines, would have been more likely to volunteer for the study. Finally, the huge standard deviations in the measured electric fields is an indication of extreme variability in exposure.

The latest studies of residential exposure to electric fields and childhood leukemia [C44, C46] found average electrical field exposures as high as 25-65 V/m, but found no excess leukemia risk, and no trend for leukemia risk to increase with increasing electrical field strength.

The existing occupational epidemiology also does not generally support an association of cancer with power-frequency electric fields [A14]. Exposure to power-frequency electric and magnetic fields are poorly correlated in occupational settings [F16], so that evaluation of electric fields as a causal agent requires examination of studies that have looked at electric field exposure separate from magnetic field exposure. Miller et al [D25] has reported an increased risk of leukemia, but not brain cancer, for occupational exposure to power-frequency electric fields. Guénel et al [D26], on the other hand, reported an increased risk of brain cancer, but not leukemia, for similar occupational exposure to power-frequency electric fields. Villeneuve et al [D39, D40] reported an association of occupational electric field exposure with leukemia and lymphoma. Other studies of occupational exposure to power-frequency electric fields have not found associations with leukemia [D13, D18, D26, D26a, D29], brain cancer [D13, D18, D25, D26a], lymphoma [D18, D25, D26, D26a], or overall cancer [D18, D25, D26, D26a].

The suggestion that power-frequency cause cancer via the electric, rather than the magnetic component of the field, is a speculation that is not only poorly supported by epidemiological and laboratory studies; but is actually contradicted by a substantial body of epidemiological and laboratory (see Question 16G) evidence. For further details see Moulder and Foster [A14]

There are certain widely accepted criteria that are weighed when assessing epidemiologic and laboratory studies of agents that may pose human health risks [A8, A9, A12, A13, E1]. These are often called the "Hill criteria" [E1]. Under the Hill criteria one examines the strength (Q20A) and consistency (Q20B) of the association between exposure and risk, the evidence for a dose-response relationship (Q20C), the laboratory evidence (Q20D) and the biological plausibility (Q20E).

The Hill criteria should be applied with caution:

• Examine the entire published literature; it is not acceptable to pick out only those reports that support the existence of a health hazard.
• Directly review the important source documents; it is not adequate to base judgments solely on academic or regulatory reviews.
• Satisfying the individual criteria is not a yes-no matter; support for a criterion can range from strong to moderate to weak to nonexistent.
• It is important to distinguish lack of support for a criteria (e.g., relevant data does not exist) from data which indicates that the criteria is not met (e.g., data showing biological implausibility or laboratory data contradicting the existence of a hazard).
• The Hill criteria should be viewed as a whole; no individual criterion is either necessary or sufficient for concluding that there is a causal relationship between exposure to an agent and a disease.

Overall, application of the Hill criteria shows that the current evidence for a connection between power-frequency fields and cancer is weak to non-existent [A7, A8, A9, K6, K7, A10, A11, A12, A15]. A detailed evaluation of the criteria follows.

The first Hill criterion is the strength of the association between exposure and risk. That is, is there a clear risk associated with exposure? A strong association is one with a relative risk (RR) of 5 or more. Tobacco smoking, for example, shows a strong association, with a RR for lung cancer 10-30 times that of non-smokers. A RR of less than about 3 indicates a weak association. A RR below about 1.5 is essentially meaningless unless it is supported by other data.

Most of the positive power-frequency studies have RRs of two or less. The leukemia studies as a group have RRs of 0.8-2.0, while the brain cancer studies as a group have RRs of 0.8-1.6. This is a weak association. Interestingly, as the sophistication of the studies has increased, the RRs have not increased.

The second Hill criterion is the consistency of the studies. That is, do most studies show about the same risk for the same disease? Using the same smoking example, essentially all studies of smoking and cancer showed an increased risk for lung and head-and-neck cancers.

Many power-frequency studies show increased incidence of some types of cancers and some types of exposures, but many do not (see, for example Q19B). Even the positive studies are inconsistent with each other. For example, while a 1993 Swedish study [C19] shows an increased incidence of childhood leukemia for one measure of exposure, it contradicts prior studies that showed an increase in brain cancer [B3], and a parallel Danish study [C17] shows an increase in childhood lymphomas, but not in leukemia. There are similar contradictions among the studies based on wire-codes.

Many of the studies are internally inconsistent. For example, where a 1993 Swedish study [C19] shows a positive association of childhood leukemia with calculated retrospective fields, it shows a negative association with measured fields. This study also shows no overall increase in childhood cancer. Since leukemia accounts for about one-third of all childhood cancer, this implies that the rates of other types of cancer were less than expected; an examination of the data indicates that this is true.

The third Hill criterion is the evidence for a dose-response relationship. That is, does risk increase when the exposure increases? For example, the more a person smokes, the higher the risk of lung cancer.

No published power-frequency exposure study has shown a dose-response relationship between measured fields and cancer rates, or between distances from transmission lines and cancer rates. However, there is some indication of a dose-response in some of the older childhood leukemia studies when wire codes or calculations of historic fields are used as exposure metrics [B9], or when measured and/or estimated magnetic fields are used as an exposure metric [C54, C57]. The lack of a clear relationship between exposure and increased cancer incidence is a major reason why most scientists are skeptical about the significance of much the epidemiology.

Not all relationships between dose and risk can be described by simple linear no-threshold dose-response curves where risk is strictly proportional to dose. There are known examples of dose-response relationships that have thresholds, that are non-linear, or that have plateaus. For example, the incidence of cancer induced by ionizing radiation in rodents rises with dose, but only up to a certain point; beyond that point, the incidence plateaus or even drops. Without an understanding of the mechanisms connecting dose and effect it is impossible to predict the shape of the dose-response relationship.

The fourth Hill criterion is whether there is laboratory evidence suggesting that there is a risk associated with such exposure. Epidemiologic associations are greatly strengthened when there is laboratory evidence for a risk.

Power-frequency fields show little evidence of the type of effects on cells, tissues or animals that point towards their being a cause of cancer (Q16A thru Q16D), or to their contributing to cancer (Q16D thru Q16G and Q17). In fact, the existing laboratory data provides strong evidence that power-frequency fields of the magnitude to which people are exposed are not carcinogenic.

The fifth Hill criterion is whether there are plausible biological mechanisms that suggest that there should be a risk. When it is understood how something causes disease, it is much easier to interpret ambiguous epidemiology. For smoking, while the direct laboratory evidence connecting smoking and cancer was weak at the time of the Surgeon General's report, the association was highly plausible because there were known cancer-causing agents in tobacco smoke.

From what is known of the physics of power-frequency fields and their effects on biological systems (Q18) there is no reason to even suspect that they pose a risk to people at the exposure levels associated with the generation and distribution of electricity. In fact, the existence of such a health hazard is both physically and biophysically implausible.

There are at least five factors that can result in false associations in the epidemiologic studies: inadequate dose assessment (Q21A), confounders (Q21B), inappropriate controls (21C), publication bias (Q21D), and multiple comparison artifacts (Q21E).

If power-frequency fields are associated with cancer, we do not know what aspect of the field is involved. At a minimum, risk could be related to the peak field, the average field, or the rate of change of the field. The duration of exposure could also be a factor. It has even been suggested that harmonics, transients, and/or interactions with the Earth's static magnetic fields are involved. If we do not know who is really exposed, and who is not, we will usually (but not always) underestimate the true risk [C15].

An additional problem posed by the lack of knowledge of the correct dose metric is that this leads many epidemiological studies to use multiple dose metrics, and thus create a massive multiple comparison problem (see Q21E).

Associations between things are not always evidence for causality. Power lines (or electrical occupations) might be associated with a cancer risk other than magnetic fields. If such an associated cancer risk were identified it would be called a "confounder" of the epidemiologic studies of power-frequency fields and cancer. An essential part of epidemiologic studies is to identify and eliminate possible confounders. Many possible confounders of the power line studies have been suggested, including PCBs, herbicides, ozone and nitrogen oxides, traffic density, and socioeconomic class.

PCBs: Many transformers contain oil that is contaminated by polychlorinated biphenyls (PCBs) and it has been suggested that PCB contamination of power-line corridors might be the cause of the excess cancer. This is unlikely. First, there is little evidence for widespread PCB contamination of power line corridors. Second, transformers are not found along high-voltage transmission lines, so PCBs could not account for the linkage of childhood leukemia with transmission corridors [B4]. Three, the evidence that PCB exposure causes or promotes cancer in people is weak [E10, L2]. Lastly, PCBs predominantly cause and promote liver cancer in animals; leukemia, brain and breast cancer have not been reported.

Herbicides: It has been suggested that herbicides sprayed on the power line corridors might be a cause of cancer. This is also an unlikely explanation. Herbicide spraying would not affect distribution systems in urban areas, where many of the "positive" childhood cancer studies have been done; and would not explain increased cancer in electrical occupations. In addition, evidence that herbicides are carcinogens in humans is weak [L7]; and the studies which suggests that the phenoxy herbicides might be carcinogens suggests an increase in lymphomas [L7], soft-tissue sarcomas [L7] and/or malignant melanoma [L32]; only one study implicates leukemia [D3], and none implicate brain cancer.

Ozone and nitrogen oxides: It has been suggested that ozone and nitrogen oxides created when high voltage lines arc might be responsible for the increased cancer. This is another unlikely explanation. While ozone is a cellular genotoxin, there is no evidence that it causes cancer in humans, and only ambiguous evidence that it causes lung cancer in rats [L6]. There is essentially no evidence that the nitrogen oxides are carcinogens. In addition, this potential confounder would apply only to high-voltage lines and would not explain reports of excess cancer along distribution systems or in electrical occupations.

Traffic density: Transmission lines frequently run along busy roads, and the "high current configurations" associated with excess childhood leukemia in some of the US studies [C1, C6, C12] are associated with busy roads [C40] . It has been suggested that power lines might be a surrogate for exposure to cancer-causing substances in traffic exhaust. This may be a serious confounder of the residential exposure studies, since traffic exhaust contains known carcinogens, and traffic density has been shown to correlate with childhood leukemia incidence [E6, C40].

Socioeconomic class: Socioeconomic class may be an issue in both the residential and occupational studies, as socioeconomic class is clearly associated with cancer risk, and "exposed" and "unexposed" groups in many studies are of different socioeconomic classes [C15, C40]. This is of particular concern in the US residential exposure studies that are based on "wire codes", since the types of wire codes that are correlated with childhood cancer are found predominantly in older, poorer neighborhoods, and/or in neighborhoods with a high proportion of rental housing [C20, C25, C40].

Ionizing radiation from corona: Periodically it is suggested on the net that corona discharges produce ionizing radiation, and that this could explain the association between power lines and cancer. Corona discharges produces heat, light (in form of small sparks), audible noise, radio interferences and a very small amount of ozone. There is no evidence that these discharges produce ionizing radiation, and strong physical arguments to suggest that they could not. Several investigators [F20, F23, F31] have measured ionizing radiation levels around high-voltage powerlines and have shown that they are not elevated. The issue is semantically complicated by the fact that corona discharges can produce ionization of the surrounding air (but ionization and ionizing radiation are two very different phenomena). An added complication is that many types of ionizing radiation monitors produce erratic readings in the presence of strong electric and magnetic fields.

An infectious basis for leukemia: See Q21F.

Other carcinogens: If "other" factors exist that increase the incidence of cancer they need to be controlled for in studies. In other words, you have to make sure that the "exposed" and "unexposed" groups have the same risk factors. Every time a new risk factor is discovered, previous studies need to be reexamined. This is a particular problem for the studies of "electrical occupations", because it would only require the presence of an unknown carcinogen in a few of those occupations to cause a false positive association with electromagnetic fields. The presence of an unidentified carcinogen is some "electrical" occupations would create weak associations, inconsistencies, and a lack of dose-response when such occupations were merged with occupations lacking exposure to this carcinogen.

An inherent problem with many epidemiologic studies is the difficulty of obtaining a "control" group that is identical to the "exposed" group for all characteristics related to the disease except the exposure. This is very difficult to do for diseases such as leukemia and brain cancer where the risk factors are poorly known. An additional complication is that often people must consent to be included in the control arm of a study, and participation in studies is known to depend on factors (such as socioeconomic class, race and occupation) that are linked to differences in cancer rates. See Jones et al [C20] and Gurney et al [C25] for example of how selection bias could affect a power line study.

It is known that positive studies in many fields are more likely to be published than negative studies. This can severely bias meta-analysis studies such as those discussed in Q13 and Q15. Such publication bias will increase apparent risks. This is a bigger problem for the occupational studies than the residential ones.

Several specific examples of publication bias are known in the studies of electrical occupations and cancer. In their review, Coleman and Beral [B1] report the results of a Canadian study that found a RR of 2.4 for leukemia in electrical workers. The British NRPB review [B3] found that further followup of the Canadian workers showed a deficiency of leukemia (a RR of 0.6), but that this followup study has never been published. This is an anecdotal report; but publication bias, by its very nature, is usually anecdotal.

It is also a clear problem for laboratory studies -- it is much easier (and much more rewarding) to publish studies that report effects than studies that report no effects. An example of this can be seen in work by Cain and colleagues. In a 1993 they published a report [G29] that 60-Hz fields were a co-promoter in a cell transformation system. But in 1993 and 1994 the same authors reported at meetings that they could not replicate the co-promotion, and that some subsequent experiments even showed a decrease in transformation when 60-Hz magnetic fields were present. However, the report of failure to replicate is not published, so that only the positive report is currently in the peer-reviewed literature.

A similar phenomena occurred in the early 90's over the issue of whether exposure to power-frequency magnetic fields affected gene transcription. There were published reports as early as 1990 of gene transcription effects (for example, [H3]); but there are also meeting reports as early as 1993 that these studies could not be replicated. The issue was not resolved until the first four reports that the studies could not be replicated [H14, H15, H22, G55] appeared in the peer-reviewed literature starting in late 1995.

There is also "reporting bias", which refers to situations where multiple studies are done but only some are reported, and to situations where abstracts and/or press reports emphasize unrepresentative subsets of the actual study. The "Swedish" studies [C19, C21] provide an example of both types of reporting bias. The original unpublished report used a number of different definitions of "exposure", and studied both children and adults. Of all the comparisons, the strongest associations were found for childhood leukemia and calculated fields. The first published English language version omitted the adult data, and the abstract emphasized the groups, exposure definitions and cancer types for which the associations were the strongest; the press reports were based largely on that abstract. The later publication of the adult portion of the study [C19], which shows no relationship between exposure and cancer incidence in adults has received virtually no press coverage. The result is that a handful of positive associations have been emphasized from a much larger group of overwhelmingly non-significant associations.

A 1996 report on breast cancer and occupational exposure [D23] provides another example of reporting bias. The study found a "modest" but non-significant increase in breast cancer in jobs with "high potential exposure". The publication itself is quite cautious, but the prepublication press release (which came out weeks before the article was actually available) read "Occupational exposure to magnetic fields increases risk of breast cancer", and omits all cautions.

Interpretation of the epidemiologic studies is complicated by multiple comparison issues. When studies include multiple exposure metrics and/or multiple types of cancer, the investigator can compare many different subgroups. A related problem arises when the investigator groups subjects into categories based on arbitrarily chosen exposure cut-points. Each such comparison (by commonly accepted statistical criteria) has a 5% probability of yielding a "statistically significant" difference, even if there were no real differences. Between multiple exposure metrics, multiple cut-points, multiple cancer sites, and subgroup analysis, a study may contain 50 or more calculations of RR, each individually analyzed for significance at 5%. A high incidence of "false positive" associations would be expected from such a study.

An illustrative example is the study by Feychting and Ahlbom [C19, C21], which looked at 12 cancer types (4 in children and 8 in adults), and 3 different exposure metrics (measured fields, calculated historic fields, and distances from lines). Within each exposure metric were further sub-definitions, such as different cut-points for separating unexposed from exposed. Solely because of the multiple cancer types and exposure metrics, 228 RRs were calculated, with values ranging from 0.0 (no cancer in exposed groups) to 5.5 (more cancer in exposed groups). Each RR was separately analyzed to calculate 95% confidence intervals. Eleven of the 228 RR's had lower confidence intervals of 1.0 or above (a crude indication of statistical significance); but even if there were no relationship between power lines and cancer, 5% (or 11.5) of the 288 RRs would been expected to be "significant" by this standard. Similarly, if there were no relationship between power lines and cancer, some "significantly" decreased rates of cancer would be expected, and such examples can be found in the study.

As a result, we are left not knowing whether the "significant correlation" of childhood leukemia with calculated historic fields is an indicator of a real association, or whether it is a piece of statistical noise. The inability of this type of epidemiologic study to prove "statistical significance" is explicitly acknowledged by Feychting and Ahlbom [C26], who point out that they do not even use the term "statistically significant" in their papers. The authors' caveat has been largely ignored by the mass media, and even by many scientific reviews of this field.

The existence of multiple comparisons, combined with post-hoc (after the fact) selection of cut-points and exposure metrics, is also a severe problem for meta-analysis, where it will cause false positives [B8].

The multiple comparison issues is a particular problem for "hypothesis-generating" studies of the type that have dominated the epidemiology of power-frequency fields. Because of the large number of variables, it is almost impossible for such studies to show true "statistical significance". What such studies can do is generate ideas that can be tested in subsequent "hypothesis-testing" studies. The hallmarks of such "hypothesis-testing" is a small set of hypotheses (usually only one) that are stated in advance, and an experimental design that avoids the multiple comparison issue by limiting the comparisons to just those that could disprove the hypothesis. Such hypothesis-testing epidemiology have been rare in studies of power-frequency fields.

The multiple comparison problem is not unique to this type of epidemiology. It is also a pervasive problem in clinical trials, and issues such as multiple endpoints, multiple cut-points, subgroup analysis, and selection of results for summaries have been extensively discussed in the biomedical literature [L1, L13, L14]. Three things are very clear:
- Ignoring multiple comparison issues can lead to a dramatic increase in reports that something is statistically significant when it is, in fact noise.
- Statistical techniques exist for correcting these problems, but it is better to avoid the problems by using proper experimental designs.
- The need for multiple comparison corrections is not accepted by some practicing epidemiologists [L17].

Interpretation of the childhood leukemia studies is greatly complicated by recent evidence that a high rate of "population mixing" (also called "high population mobility") is a strong risk factor for childhood leukemia and lymphoma [L36, L37]. The explanation for the association (called the Kinlen [L16] hypothesis) is that: "childhood leukemia might result from a rare response to a common but unidentified infection and the increased risks would occur when populations were mixed that increased the level of contacts between infected and susceptible individuals."[L36]

The complication for the power line studies, is that it has been a common observation that the "cases" are more residentially mobile than the "controls"[D6, C20, C44, C45], and that people living in high wire-code homes are more residentially mobile than people living in low wire-code homes [C20]. This means that the weak associations seen in some studies could be due to differences in residential mobility and have nothing to do with power-frequency fields.

Even if this confounder turns out to be real, it would not probably not be applicable to studies of adult leukemia, or to studies of other types of cancer.

The best evidence for a connection between cancer and power-frequency fields is probably:

• The four epidemiologic studies that show a correlation between childhood leukemia and proximity to high-current wiring (see Q14) [C1, C6, C12, C19], plus the meta-analysis of the Scandinavian studies [B4].
  Caution: The 1997-1999 studies discussed in Q19H through Q19K have seriously eroded the validity of this argument.

• The pooled analyses [C54, C57] of multiple studies of power line fields which report that for measured or estimated magnetic fields, there is an increased incidence of childhood leukemia in the children in the highest exposure group.

• The suggestion of a dose-response relationship (see Q20C) in some of the childhood leukemia studies [B9, C54, C57].
  

• The epidemiologic studies (see Q13) that show a correlation between work in electrical occupations and cancer, particularly leukemia [D9, D11, D12, D19, B17] and brain cancer [B6, D21, B17].

• The lab studies that show that power-frequency fields do produce bioeffects (see Q18A).
  Caution: Many of these effects have no known relationship to cancer, or have never been replicated, or have failed attempts at replication (see Q18A), or occur only for exposures far above those actually encountered in residential and occupational settings.

• The report [G60] that power-frequency fields can cause DNA strand-breaks in rat brain cells.
  Caution: This group has also reported genotoxicity for microwaves using the same assay system, and the microwave results have recently failed three independent attempts at replication. The five other groups [G6, G20, G37, G99, G104] that have looked for evidence that power-frequency fields cause DNA strand breaks have found nothing.

• The laboratory studies (see Q16E) that provide evidence that power-frequency magnetic fields can promote chemically-induced breast cancer [G16, G26, G50, G86].
  Caution: These studies should be interpreted with great caution, as they have failed three separate attempts at replication [G69, G73, G85]. See the discussion in Q16E; and see Boorman et al [K8] and Anderson et al [K11]) for some of the problems with these studies.

• The studies reporting that intense fields can enhance tumor [G18, G26, G39, G50] and cell [G8, G42, G46] growth rates (see Q17A).

• The studies reporting that fields can cause [G35, H29] or influence [G29] neoplastic cell transformation (see Q16D).
  Caution: These cell transformation studies have failed numerous confirmation attempts (see Q16D).

The best evidence that there is not a connection between cancer and power-frequency fields is probably:

• Hill criteria analysis of the entire body of epidemiologic and laboratory studies, which shows that the evidence for a causal relationship is weak to non-existent (Q20).

• The fact that the epidemiological associations are weak (Q20A) and inconsistent (Q20B); and that they generally fail to show any exposure-response relationship (Q20C).

• The fact that recent epidemiological studies have failed to find any significant evidence for an association between power lines and childhood brain cancer or childhood leukemia (Q19A, Q19H through Q19K).

• The fact that long-term exposure of animals to power-frequency fields does not cause cancer (Q16B).

• The fact that the lab studies of genotoxicity have been overwhelmingly negative (Q16A thru Q16D).

• The fact that most lab studies of epigenetic activity have been negative, and that the few positive studies have used fields far more intense than those to which people are actually exposed (Q16D thru Q16F).

• The biophysical analyses that indicates that "any biological effects of weak (less than 5 microT) ELF fields on the cellular level must be found outside of the scope of conventional physics"(Q18B).

• The fact that multiple comparison problems call into question the statistical significance of all of the "positive" epidemiologic studies (Q21E).

• The consistent rejection of the idea that there is convincing data to support a causal relationship between exposure to power-frequency fields and cancer by all scientific panels that have examined this issue over the past decade [e.g., A1, A2, A3, A4, A7, A11, A15, A16, A17, A19, A20].

• Jackson's [E9] and Olsen's [C17] argument that a connection between cancer and power lines is unlikely because childhood and adult leukemia rates have been stable over a period of time when per capita power consumption has risen dramatically. This argument presumes that "exposure" has risen in parallel with "consumption", but until recently there was little relevant historical data to support this assumption. However, Swanson [F25] has analyzed power use in the UK between 1949 and 1989 and has calculated that average residential exposures have risen by a factor of nearly 5. This gives considerable strength to this argument.

• The fact that the "power line - cancer controversy" has many of the hallmarks of "pathological science" [L29].

Most scientists who are familiar with the literature consider that the issue has either already been resolved, or that it cannot be resolved (see Q27E and Q27F). Thus, the question is what will it take to convince the public and the media.

In the epidemiologic area, more of the same types of studies are unlikely to resolve anything. Studies showing a dose-response relationship between measured fields and cancer incidence rates would clearly affect thinking, as would studies identifying confounders in the residential and occupational studies.

In the laboratory, more genotoxicity and promotion studies may not be very useful. Further studies of some of the known bioeffects would be useful, but only if they identified mechanisms or established the conditions under which the effects occur (e.g., thresholds, dose-response relationships, frequency-dependence, optimal wave-forms).

While this FAQ sheet, and most public concern, has centered around cancer, there have also been suggestions that there might be a connection between non-ionizing electromagnetic exposure and a variety of other human health problems.

Concern about miscarriages and birth defects has focused as much on video display terminals (VDTs) as on power lines. There is little epidemiologic [J1, J5, J6, J9, J10, J12, J15, J18, J19] or laboratory [J4, J12, J13, J15] support for a connection between non-ionizing electromagnetic exposure and birth defects. Robert [J16], Huuskonen et al [J12] and Brent [J15] have reviewed this field in detail.

In 1999 Ryan et al [J14] reported that exposure of mice to 2, 2000 or 10,000 microT power-frequency fields for multiple generations had no effect on fertility or birth defects. In a 2000 follow-up study Ryan et al [J17] reported that adding harmonics to the exposure also produced no reproductive toxicity.

In 1996, there was is a report of excess Alzheimer's disease in occupations with "probable exposure" to power-frequency fields [E16]. That study reported that dressmakers, seamstresses and tailors had excess rates of Alzheimer's disease, and that these groups were exposed to power-frequency fields from sewing machines; the study found no excess Alzheimer's disease in any other "electrical occupations". More recent studies have found no excess rates of Alzheimer's disease in electrical utility workers [D32, D38] or in other occupations with exposure to power-frequency fields [D38].

In 1998, Sastre et al [Bioelectromag 19:98-106, 1998] reported that exposure of human volunteers to power-frequency magnetic fields caused changes in heart rate. In a 1999 study that was stimulated by the hypothesis advanced by Sastre et al, Savitz et al [D36] reported that occupational exposure to power-frequency fields was associated with an increased incidence of certain types of heart disease. In related studies, Sait et al [E22] reported that exposure of human volunteers to a 15 microT power-frequency field caused a small decrease in heart rate. However, in 2000, Graham, Sastre and colleagues [L44, L45] reported that they could not replicate the 1998 Sastre et al, even at higher field strengths.

A variety of other possible human health effects have been assessed in single studies:

• In 1999, Johansen et al [D37] found no significant association of multiple sclerosis with occupational exposure to power-frequency fields.
• In 1999 Graham et al [L42] reported that exposure of human volunteers to 14 or 28 microT fields at 60-Hz fields did not cause neurophysiological effects, and that there was no evidence that the volunteers could sense the field.
• In 1999 Graham and Cook [L43] reported that exposure of human volunteers to a 28 microT field at 60-Hz fields caused sleep disturbance if the exposure was intermittent, but not if it was continuous.
• in 2000, van Wijngaarden et al [D41] reported an association between suicide and exposure to power-frequency fields in male electric utility workers.

Comprehensive reviews of power-frequency fields and human health:

• The 1996 report from the US National Academy of Science [A7 and see Q27E] is largely restricted to residential exposures, and is now a bit out-dated.
• The report from the NIEHS "working group" [A11 and see Q27F] is comprehensive, but the organization and style of the report makes it very hard to read.
• The 1999 review by the NAS provides an overview of the large body of laboratory work that was done under the US EMF-RAPID program, much of which has not yet been published (but see the May 2000 special issue of Radiation Research [A18]).
• The 1999 NIEHS report to the U.S. Congress [A16] provides a compact review of power-frequency fields and human health and is available on the web at: http://www.niehs.nih.gov/emfrapid/
• The reviews by Davis et al [A1] and Doll et al [B3], and the two French reviews [A3, A4] are good, but were published before many of the important studies were available and are now really of only historic interest.
• The 1998 review by Moulder [A12] is derived directly from an early-1998 version of this FAQ document.
• The 1999 policy statement from the Committee on Man and Radiation (COMAR) of the IEEE [A17], "Possible Health Hazards From Exposure to Power-Frequency Electric and Magnetic Fields", is available on line at: http://homepage.seas.upenn.edu/~kfoster/powerfreq.htm
• The 2000 review by Preece et al. [A19] provides a compact review that focuses on the childhood leukemia issue.

Reasonably up-to-date (1996 or later) reviews of specific areas:

• Meinert and Michaelis [B8] review the residential cancer epidemiology.
• Miller et al [B13] review both the occupational and residential studies.
• Li et al [B10] review the epidemiological studies of powerlines and adult cancer.
• McCann et al [K7] review the animal carcinogenesis studies.
• Moulder [K6] and Lacy-Hulbert et al [A10] review the biological evidence for carcinogenesis.
• Kavet [A8] review the current thinking on carcinogenesis with an emphasis on how it might apply to power-frequency fields.
• Foster et al [A9] review risk assessment and how it applies to exposure to electromagnetic fields.
• Robert [J16], Huuskonen et al [J12] and Brent [J15] review the lab and epidemiological evidence for birth defects associated with power-frequency fields.
• Valberg et al [F23] review the plausibility of proposed mechanisms of interaction of power-frequency fields and biological systems.
• McCann et al [K2] review genotoxicity studies done with power-frequency electric and magnetic fields.
• McCann et al [A13] review cancer risk assessment issues as they apply to power-frequency fields.
• Moulder and Foster [A14] review cancer risk assessment issues as they apply to power-frequency electric (as opposed to magnetic) fields.

Yes, a number of governmental and professional organizations have developed exposure guidelines. The most generally relevant are those issued by the UK National Radiation Protection Board (NRPB-UK) [M4], the International Commission on Non-Ionizing Radiation Protection (ICNIRP) [M6], and the American Conference of Governmental Industrial Hygienists (ACGIH) [M5].

See Bailey et al [M8] for a detailed discussion of the standards, and of the biological basis for these standards.

• NRPB-UK [M4]:
  • 50 Hz: 1,600 microT (16 G) and 12 kV/m
  • 60 Hz: 1,330 microT (13.3 G) and 10 kV/m
  • This document also contains guidelines for other frequencies.

• ICNIRP [M6]
  • 50 Hz: 100 microT (1 G) and 5 kV/m
  • 60 Hz: 84 microT (0.84 G) and 4.2 kV/m
  • This document also contains guidelines for other frequencies.

• NRPB-UK [M4]:
  • 50 Hz: 1,600 microT (16 G) and 12 kV/m
  • 60 Hz: 1,330 microT (13.3 G) and 10 kV/m
  • This document also contains guidelines for other frequencies.

• ACGIH [M5]:
  • At 60 Hz: 1,000 microT (10 G)
  • This document also contains guidelines for other frequencies.

• ICNIRP [M6]
  • 50 Hz: 500 microT (5 G) and 10 kV/m
  • 60 Hz: 420 microT (4.2 G) and 8.3 kV/m
  • This document also contains guidelines for other frequencies.

Pacemaker function can be affected by power-frequency fields. Fields strong enough to interfere with pacemaker function clearly could exist in some occupational settings [L10, L11], and might even exist in some non-occupational settings [L0, L11]. The sensitivity of cardiac pacemakers and the severity of the effects are very dependent on design and model [L0, L10, L11]. This is probably also a situation where the electric field is at least as important as the magnetic field.

ICNIRP [M6] calculated that interference could be caused by power-frequency fields as low as 15 microT, but states that there is "only a small probability" of malfunction below 100-200 microT. NRPB-UK [M4] states that "interference is unlikely to occur" below 20 microT. ACGIH [M5] has a formal occupational limit for pacemaker wearers of 100 microT. Based on above sources it would appear that pacemaker interference from a power line magnetic field would be unlikely (see Q10).

However, two studies of pacemakers reports that power-frequency electric fields as low as 5000-6000 V/m could cause interference with some models [L0, L48]; and another implies that interference might be possible for electric fields as low as 1500 V/m [L10]. Electric fields as high 1,500 V/m would not be encountered in the vast majority of residence or in the vicinity of distribution lines, but this level could be exceeded directly under a high-voltage transmission line (see Q10).

Pacemaker users who work or live in environments where there is equipment capable of causing significant electromagnetic interference should bring this to the attention of the physician who implanted the pacemaker. Pacemaker users would also be advised to exercise some caution when in the close vicinity of high voltage transmission lines, particularly lines with voltages of 230 kV and above. The same words of caution are probably applicable to implanted defibrillators, and might be applicable to other implanted biomedical devices.

The July/August 1995 issue of Microwave News contained extensive quotes from what was said to be a draft report of a committee of the National Commission on Radiation Protection (NCRP). The excerpt(s) published by Microwave News appear to have been written in early 1993. According to the Microwave News article, the NCRP report recommended strict standards for occupational and residential exposure to power-frequency (and other ELF) electric and magnetic fields. The Microwave News report was subsequently picked up by Science and the New Scientist and then by the mass media.

According to an official statement by the NCRP (August 22, 1995), this draft report "has absolutely no standing at this time". The NCRP statement goes on to say that "the draft in question is still undergoing revisions to prepare it for entry into the initial review phase, it exists only as a working draft that should not have been released outside [the Committee]. Thus it should not be copied, quoted, or referenced outside of the NCRP."

A later (October 11, 1995) NCRP statement says that "contrary to many erroneous sources of information, the NCRP has not made recommendations on ELF EMF" and notes that "considering the extensive nature of the review process, it is impossible to predict when the NCRP may have a report on the subject of ELF and it is not possible to know the extent or recommendations that might be made".

The 1999 Annual Report of the NCRP refers to this report as still being in subcommittee SC89-3 with a "draft report being prepared for Council review".

In 1991 the US Congress asked the National Academy of Sciences to review the literature on the possible health risks of residential exposure to power-frequency electric and magnetic fields. In response the National Research Council, the research arm of the National Academy of Sciences, set up a committee of epidemiologists, biologists, chemists, and physicists who were experts in cancer, reproductive toxicology and neurobiological effects. Some members had spent their careers studying the effects of electric and magnetic fields, some where new to the field. The Committee issued its report in November of 1996 [A7]. The following are direct quotes from the executive summary.

• Conclusions of the Committee
  • "Based on a comprehensive evaluation of published studies relating to the effects of power frequency electric and magnetic fields on cells, tissues, and organisms (including humans), the conclusion of the committee is that the current body of evidence does not show that exposure to these fields presents a human-health hazard."
  • "No conclusive and consistent evidence shows that exposures to residential electric and magnetic fields produce cancer, adverse neurobehavioral effects, or reproductive and developmental effects."
  • "At exposure levels well above those normally encountered in residences, electric and magnetic fields can produce biologic effects...but these effects do not provide a consistent picture of a relationship between the biologic effects of these fields and health hazards."
  • "An association between residential wiring configurations (called wire codes) and childhood leukemia persists in multiple studies, although the causative factor responsible for that statistical association has not been identified."
• Epidemiology
  • "The driving force for continuing the study of the biologic effects of electric and magnetic fields has been the persistent epidemiologic reports of an association between a hypothetical estimate of electric- and magnetic-field exposure called the wire-code classification (see Q14) and the incidence of childhood leukemia."
  • "Living in homes classified as being in the high wire-code category is associated with about a 1.5-fold excess of childhood leukemia, a rare disease."
  • "Wire-code ratings correlate with many factors-such as age of home, housing density, and neighborhood traffic density-but the wire-code ratings exhibit a rather weak association with measured residential magnetic fields."
  • "No association between the incidence of childhood leukemia and magnetic-field exposure has been found in epidemiologic studies that estimated exposure by measuring present-day average magnetic fields."
  • "Studies have not identified the factors that explain the association between wire codes and childhood leukemia. Although various factors are known to correlate with wire-code ratings, none stands out as a likely causative factor."
  • "[The] epidemiologic evidence does not support possible associations of magnetic fields with adult cancers, pregnancy outcome, neurobehavioral disorders, and childhood cancers other than leukemia."
• Exposure Assessment
  • "Magnetic fields of the magnitude found in residences induce currents within the human body that are generally much smaller than the currents induced naturally from the function of nerves and muscles."
  • "However, the highest field strengths to which a resident might be exposed (those associated with appliances) can produce electric fields within a small region of the body that are comparable to or even larger than the naturally occurring fields."
  • "The endogenous current densities on the surface of the body (higher densities occur internally) associated with electric activity of nerve cells are of the order of 1 mA/m-sq... Therefore, the typical externally induced currents are 1,000 times less than the naturally occurring currents."
  • "Because the mechanisms through which electric and magnetic fields might produce adverse health effects are obscure, the characteristics of the electric or magnetic fields that need to be measured for testing the linkage of these fields to disease are unclear."
• Cellular and Molecular Effects
  • "Magnetic-field exposures at 50-60 Hz delivered at field strengths similar to those measured for typical residential exposure (0.01 - 1 microT) do not produce any significant in vitro effects that have been replicated in independent studies."
  • "Reproducible changes have been observed in the expression of specific features in the cellular signal-transduction pathways for magnetic-field exposures on the order of 100 microT and higher."
  • "At field strengths greater than 50 microT, credible positive results are reported for induced changes in intracellular calcium concentrations and for more general changes in gene expression and in components of signal transduction."
  • "No reproducible genotoxicity is observed, however, at any field strength."
  • "The overall conclusion, based on the evaluation of these studies, is that exposures to electric and magnetic fields at 50-60 Hz induce changes in cultured cells only at field strengths that exceed typical residential field strengths by factors of 1,000 to 100,000."
• Animal and Tissue Effects
  • "There is no convincing evidence that exposure to 60-Hz electric and magnetic fields causes cancer in animals."
  • "One area with some laboratory evidence of a health-related effect is that animals treated with carcinogens show a positive relationship between intense magnetic-field exposure and the incidence of breast cancer."
  • "There is no evidence of any adverse effects on reproduction or development in animals, particularly mammals, from exposure to power-frequency 50- or 60-Hz electric and magnetic fields."
  • "There is convincing evidence of behavioral responses to electric and magnetic fields that are considerably larger than those encountered in the residential environment; however, adverse neurobehavioral effects of even strong fields have not been demonstrated."
  • "Despite the observed reduction in pineal and blood melatonin concentrations in some animals as a consequence of magnetic-field exposure, studies of humans provide no conclusive evidence to date that human melatonin concentrations respond similarly... In animals with observed melatonin changes, adverse health effects have not been shown to be associated with electric- or magnetic-field-related depression in melatonin."

• "The NIEHS biologic research program made two important conclusions that reduce somewhat the concern about whether the use of electric power might have adverse health effects...
  • The first contribution was the effort to replicate previous reports of biological effects... All the attempted replications in the EMF-RAPID program have had negative or equivocal results...
  • The second important contribution was the completion of several investigations of the relationship between magnetic field exposure and cancer through controlled laboratory experiments in animals. Nearly all the animals studies relevant to the [power-frequency field]-cancer question had negative results, even at field levels that were orders of magnitude greater than levels typical of human exposure."
• "The EMF-RAPID biologic research contributed little evidence to support the hypothesis that a link exists between [power-frequency fields] and cancer...
  • The results of in vivo studies do not support a [power-frequency field] effect on cancer initiation, promotion or progression...
  • Evidence of any robust and replicated effects on the development of cancer is lacking."
• "The results of the EMF-RAPID program do not support the contention that the use of electricity poses a major unrecognized public-health danger"
• "The committee recommends that no further special research program focused on possible health effects of power-frequency magnetic fields be funded."
• See comments by the NAS committee in the NIEHS "working group" report in the next question.

In 1997-1998, NIEHS organized a series of scientific meetings to evaluate "the potential human health effects from exposure to extremely low frequency electric and magnetic fields". The reports generated at those meetings were to be used to assist NIEHS in preparing a report to the U.S. Congress (see Q27G).

The final of the series of meetings organized by NIEHS (called the "working group") evaluated the evidence for effects on human health under the rules of the International Agency for Research on Cancer (IARC). The actual report from the "working group" was released on 30-July-1998 [A11], and is available at: http://www.niehs.nih.gov/emfrapid/

Unlike most modern approaches to risk assessment (see Q20), the IARC rules used by the "working group" (see Table below) place heavy emphasis on epidemiological studies, and pay much less attention to animal and mechanistic studies.

The "working group" unanimously concluded that the power-frequency fields were not an IARC class 1 or class 2A agent; that is, that they were not a "known human carcinogen" or a "probable human carcinogen" (see Table below). The majority of the "working group" concluded that power-frequency fields should be classified as IARC class 2B; that is that they were a "possible human carcinogen". Other agents similarly classified by the IARC as "possible human carcinogens" include coffee, saccharin and automobile exhaust. A substantial minority of the "working group" concluded that the evidence was not even sufficient to place power-frequency fields in IARC class 2B.

According to the report of the "working group", the classification in IARC class 2B was based on "limited epidemiological evidence" that residential exposure to power-frequency fields was associated with childhood leukemia. "Limited epidemiological evidence", in the IARC scheme means: "A positive association has been observed between exposure... and cancer for which a causal interpretation is considered credible, but chance, bias or confounding could not be ruled out with reasonable confidence."

The "working group" also concluded that studies in experimental animals "did not support or refute" the epidemiological studies, and that mechanistic studies provided no support for the epidemiological studies.

The "working group" concluded that the epidemiological and experimental evidence was "inadequate" (see Table below) to suggest that exposure to power-frequency fields was a "possible" cause of any type of cancer other than leukemia. The "working group" also concluded that the epidemiological and experimental evidence was "inadequate" (see Table below) to suggest that exposure to power-frequency fields was a "possible" cause of adverse human health effects other than cancer.

Some have interpreted the conclusions of the "working group" as a contradiction to what was said in 1996 by the National Academy of Sciences (NAS) panel (see Q27E) and in 1999 by the NIEHS in their report to Congress (see Q27G). In fact, the body of the "working group" report [A11] is quite compatible with both the NAS report [A7] and the 1999 NIEHS report [A16]. In particular, all three reports agree that no causal association has been established between cancer and exposure to power-frequency fields. The perceived difference between the reports is due to the approach to risk assessment used by the "working group".

In 1999, the National Academy of Sciences commented on the "working group report" [A15]. They concluded:

When the working group report is considered in more detail, the dramatic contrast between the Research Council committee report [A7] and the NIEHS report [A11] -- "no effect" versus "probable carcinogen" -- is reduced; and when the differences between the two evaluation processes that were used are taken into account, the difference in conclusions is understandable. The current committee concludes, however, that the conclusions of the 1997 Research Council committee report more accurately convey the health implications of the underlying research to the public."

By "possible human carcinogen", the "working group" explicitly meant IARC class 2B. As shown in the Table below, classification in class 2B requires only weak epidemiological evidence of an association. No laboratory confirmation or biological/biophysical plausibility is required to place something in class 2B. In fact, once there is any epidemiological evidence of an association, "possible human carcinogen", may be the lowest designation allowed by the IARC scheme.

It is also important to note that the "working group" unanimously rejected a conclusion that the power-frequency fields were a "probable" (IARC class 2A) or "proven" (IARC class 1) human carcinogens.

On 15 June 1999, the U.S. National Institute of Environmental Health Sciences (NIEHS) issued a report to the U.S. Congress on "Health Effects from Exposure to Power-Line Electric and Magnetic Fields" [A16]. The report is based on:

• The four symposia discussed in Q27F;
• An up-to-date review of the relevant epidemiologic, animal, cellular and biophysical studies (so that even the 1999 Canadian childhood leukemia study, Q19J, is discussed);
• Consideration of the laboratory research sponsored by NIEHS under the program called EMF-RAPID [A18].

The NIEHS report to Congress [A16] differs from the "working group" report [A11] in several respects:

• The report to Congress gives more weight to animal, cellular and biophysical studies than did the "working group" report.
• The report to Congress does not focus on the IARC criteria and language [see Table] that dominated the "working group" report.
• The report to Congress is much shorter than the "working group" report, and uses language that should be far easier for most people to understand.

The report is available at: http://www.niehs.nih.gov/emfrapid/html/EMF_DIR_RPT/Report_18f.htm

From the Executive Summary:

The scientific evidence suggesting that [power-frequency electromagnetic field] exposures pose any health risk is weak. The strongest evidence for health effects comes from associations observed in human populations with two forms of cancer: childhood leukemia and chronic lymphocytic leukemia in occupationally exposed adults. While the support from individual studies is weak, the epidemiological studies demonstrate, for some methods of measuring exposure, a fairly consistent pattern of a small, increased risk with increasing exposure that is somewhat weaker for chronic lymphocytic leukemia than for childhood leukemia. In contrast, the mechanistic studies and the animal toxicology literature fail to demonstrate any consistent pattern across studies although sporadic findings of biological effects (including increased cancers in animals) have been reported. No indication of increased leukemias in experimental animals has been observed...

Epidemiological studies have serious limitations in their ability to demonstrate a cause and effect relationship whereas laboratory studies, by design, can clearly show that cause and effect are possible. Virtually all of the laboratory evidence in animals and humans and most of the mechanistic work done in cells fail to support a causal relationship between exposure to [power-frequency electromagnetic fields] at environmental levels and changes in biological function or disease status. The lack of consistent, positive findings in animal or mechanistic studies weakens the belief that this [epidemiological] association is actually due to [power-frequency electromagnetic fields], but it cannot completely discount the epidemiological findings.

The NIEHS concludes that [power-frequency electromagnetic field] exposure cannot be recognized as entirely safe because of weak scientific evidence that exposure may pose a leukemia hazard. In our opinion, this finding is insufficient to warrant aggressive regulatory concern. However, because virtually everyone in the United States uses electricity and therefore is routinely exposed to [power-frequency electromagnetic fields], passive regulatory action is warranted such as a continued emphasis on educating both the public and the regulated community on means aimed at reducing exposures.

As part of the EMF-RAPID Program's assessment of [power-frequency electromagnetic field]-related health effects, an international panel of 30 scientists met in June 1998 to review and evaluate the weight of the scientific evidence [see Q27F]. Using criteria developed by the International Agency for Research on Cancer [see Table], none of the Working Group considered the evidence strong enough to label [power-frequency electromagnetic field] exposure as a "known human carcinogen" or "probable human carcinogen." However, a majority of the members of this Working Group (19/28 voting members) concluded that exposure to power-line frequency [electromagnetic fields] is a "possible" human carcinogen. This decision was based largely on "limited evidence of an increased risk for childhood leukemias with residential exposure and an increased occurrence of CLL (chronic lymphocytic leukemia) associated with occupational exposure." For other cancers and for non-cancer health endpoints, the Working Group categorized the experimental data as providing much weaker evidence or no support for effects from exposure to [power-frequency electromagnetic fields].

The NIEHS agrees that the associations reported for childhood leukemia and adult chronic lymphocytic leukemia cannot be dismissed easily as random or negative findings. The lack of positive findings in animals or in mechanistic studies weakens the belief that this association is actually due to [power-frequency electromagnetic fields], but cannot completely discount the finding. The NIEHS also agrees with the conclusion that no other cancers or non-cancer health outcomes provide sufficient evidence of a risk to warrant concern...

The National Toxicology Program routinely examines environmental exposures to determine the degree to which they constitute a human cancer risk and produces the "Report on Carcinogens" listing agents that are "known human carcinogens" or "reasonably anticipated to be human carcinogens." It is our opinion that based on evidence to date, [power-frequency electromagnetic field] exposure would not be listed in the "Report on Carcinogens" as an agent "reasonably anticipated to be a human carcinogen." This is based on the limited epidemiological evidence and the findings from the EMF-RAPID Program that did not indicate an effect of [power-frequency electromagnetic field] exposure in experimental animals or a mechanistic basis for carcinogenicity.

The NIEHS suggests that the level and strength of evidence supporting [power-frequency electromagnetic field] exposure as a human health hazard are insufficient to warrant aggressive regulatory actions; thus, we do not recommend actions such as stringent standards on electric appliances and a national program to bury all transmission and distribution lines. Instead, the evidence suggests passive measures such as a continued emphasis on educating both the public and the regulated community on means aimed at reducing exposures. NIEHS suggests that the power industry continue its current practice of siting power lines to reduce exposures and continue to explore ways to reduce the creation of magnetic fields around transmission and distribution lines without creating new hazards. We also encourage technologies that lower exposures from neighborhood distribution lines provided that they do not increase other risks, such as those from accidental electrocution or fire.

On 6 March 2001, the U.K. National Radiation Protection Board (NRPB) issued a report on power-frequency fields and cancer [A20]. The report is: "a comprehensive review of experimental and epidemiological studies relevant to an assessment of the possible risk of cancer resulting from exposures to power-frequency electromagnetic fields... It is not concerned with exposures to high frequencies nor with other potential effects of exposure to power frequencies..."

The main conclusion of the report was that:

Laboratory experiments have provided no good evidence that extremely low frequency electromagnetic fields are capable of producing cancer, nor do human epidemiological studies suggest that they cause cancer in general. There is, however, some epidemiological evidence that prolonged exposure to higher levels of power frequency magnetic fields is associated with a small risk of leukaemia in children. In practice, such levels of exposure are seldom encountered by the general public in the UK. In the absence of clear evidence of a carcinogenic effect in adults, or of a plausible explanation from experiments on animals or isolated cells, the epidemiological evidence is currently not strong enough to justify a firm conclusion that such fields cause leukaemia in children. Unless, however, further research indicates that the finding is due to chance or some currently unrecognized artifact, the possibility remains that intense and prolonged exposures to magnetic fields can increase the risk of leukaemia in children.

At the cellular level, there is no clear evidence that exposure to power frequency electromagnetic fields at levels that are likely to be encountered can affect biological processes...

There is no convincing evidence that exposure to such fields is directly genotoxic nor that it can bring about the transformation of cells in culture and it is therefore unlikely to initiate carcinogenesis...

Those results that are claimed to demonstrate a positive effect of exposure to power frequency magnetic fields tend to show only small effects, the biological consequences of which are not clear. Many of the positive effects reported involve exposures which are unlikely to been encountered.

Overall, no convincing evidence was seen from a review of a large number of animal studies to support the hypothesis that exposure to power frequency electro-magnetic fields increases the risk of cancer.

Most studies report a lack of effect of power frequency magnetic fields on leukaemia or lymphoma in rodents...

Further studies found no effect on the progression of transplanted leukaemia cells in mice or rats...

A recent large-scale study reported a lack of effect of exposure to power frequency magnetic fields on chemically induced nervous system tumours in rats. In addition, the low incidence of brain cancers in three recent large-scale rat studies was not elevated by magnetic field exposure.

With regard to studies of tumours [other than leukemia and brain cancer], the evidence is almost uniformly negative.

Most evidence from human volunteer studies suggests that melatonin rhythms are not delayed or suppressed by exposure to power frequency magnetic fields...

The evidence for an effect of exposure to power frequency magnetic fields on melatonin levels and on melatonin-dependent reproductive status in seasonally breeding animals is largely negative.

There is no consistent evidence of any inhibitory effect of power frequency magnetic field exposure on those aspects of immune system function relevant to tumour suppression...

Recent large and well-conducted studies have provided better evidence than was available in the past on the relationship between power frequency magnetic field exposure and the risk of cancer. Taken in conjunction they suggest that relatively heavy average exposures of 0.4 microT or more are associated with a doubling of the risk of leukaemia in children under 15 years of age. The evidence is, however, not conclusive...

Data on brain tumours come from some of the studies also investigating leukaemia and from others concerned exclusively with these tumours. They provide no comparable evidence of an association...

There is no reason to believe that residential exposure to electromagnetic fields is involved in the development of leukaemia or brain tumours in adults.

Although recently published studies of occupational exposure to electromagnetic fields and the risk of cancer are, in the main, methodologically sound, and some of them have considerable statistical power, causal relationships between such exposure and an increase in tumour incidence at any site are not established. The excesses, where they exist, are generally modest and are largely restricted to leukaemia and cancer of the brain.

The evidence of any risk for brain cancer is conflicting, even that from the most powerful of the studies.

There is very little hard data on this issue. There have been "comparable property" studies, but any studies done prior to about 1991 (when London et al [C12] was published) might be irrelevant. One comparable value study has been published since 1991 [L5], and another has been presented at a meeting [L8]. Neither study shows evidence for an impact of power lines on property values. However, both studies indicate that many owners think that there will be an impact, particularly if concerns about health effects become widespread.

It appears that the presence of obvious transmission lines or substations can adversely affect property values if there has been recent local publicity about health or property value concerns. It appears less likely that the presence of "high current configuration" distribution lines of the type correlated with childhood cancer in the US studies (see Q14) would affect property values, since few people would recognize their existence. If buyers start requesting magnetic field measurements, no telling what will happen, since while measurements are relatively easy to do (Q29), they are essentially impossible to interpret (see Q14).

Power-frequency fields are measured with a calibrated gauss meter. The meters used by environmental health professionals are too expensive for "home" use. A unit suitable for home use should meet the following criteria:
- a reasonable degree of accuracy and precision (plus/minus 20% seems reasonable for home use);
- true rms detection, otherwise readings might be exaggerated if the wave form is non-sinusoidal;
- a tailored frequency response, because if the unit is too broad-band, higher frequency fields from VDTs, TVs, etc. may confound the measurements
- the correct response to overload; if the unit is subjected to a very strong field, it should peg, not just give random readings;
- a strong electric field should not affect the magnetic field measurement.

Meters meeting these requirements are expensive, and inexpensive meters may be unreliable. A 1994 review of meters by Iowa State [F15], found a $450 meter they rated a suitable for use by a lay-person. For the expert or non-expert who has a good multimeter, and knows how to use a spread sheet, the Iowa State report indicated that a suitable unit could be gotten for as little as $115.

The suggestion is sometimes made that one could wind a coil and use headphones or a high impedance multimeter to measure power-frequency fields. This is misguided; while a clever physicist or engineer could anticipate and correct for non-linearity and interference, this is unreasonable approach for the average person, even one technically trained.

Measurements must be done with a calibrated gauss meter (Q29) in multiple locations over a substantial period of time, because there are large variations in fields over space and time. Fortunately, the magnetic field is far easier to measure than the electric field. This is because the presence of conductive objects (including the measurer's body) distorts the electric field and makes meaningful measurements difficult. Not so for the magnetic field.

It is important for the person who is making the evaluation to understand the difference between an emission and exposure. This may seem obvious, but many people, including some very smart physical scientists, stick an instrument right up to the source and compare that number with an exposure standard. Also, if the instrument is not isotropic, measurement technique must compensate for this.

In the case of power distribution line and transformer fields, the magnetic fields may vary considerably over time, as they are proportional to the current in the system. A reasonable survey needs to be done over time, with anticipated and actual electricity usage factored in.

This FAQ sheet concerns itself primarily with sinusoidal fields at frequencies of 50 or 60 Hz. However, certain general issues are relevant to some other types of electromagnetic sources.

The basic principles and data discussed in the FAQ sheet are generally applicable to electromagnetic sources with frequencies between 1 Hz and 30,000 Hz (30 kHz). The major issue encountered when dealing with low-frequency sources other than power-frequency is that the currents induced by time-varying magnetic fields depend on frequency and wave-form, as well as field intensity. As the frequency increases, so do the induced currents. Thus safety guidelines change with frequency [M4, M5]. For example, the NRPB magnetic field exposure guideline [M4], which is 1,330 microT at 60 Hz, rises to 80,000 microT at 1 Hz, but falls to 80 microT at 3 kHz.

Estimating the currents induced by non-sinusoidal ELF wave forms is more complex, because the magnitude of the induced current depends on the rate at which the magnetic field changes. Thus a square wave of the same frequency and amplitude of a sinusoidal wave will induced a much greater current.

Static electric and magnetic fields, and ELF fields with frequencies below 1 Hz are covered in a companion FAQ sheet called "Static Electromagnetic Fields and Cancer FAQs" (http://www.mcw.edu/gcrc/cop/static-fields-cancer-FAQ/toc.html). For standards and regulations concerning occupational and environmental exposure to static fields see the ICNIRP guidelines [M7].

Above 30 kHz, one moves into the radiofrequency (RF) and microwave (MW) range, and biophysical and biological issues arise [M1, M3] that are not within the scope of this FAQ sheet. First, as the wavelength gets shorter, there is non-ionizing radiation as well as electric and magnetic fields to consider. Second, as the frequency rises into the MHz range, heating due to induced electric currents may no longer be negligible.

To our knowledge, there are no on-line resources on RF/MW bioeffects and human health issues, except for "FAQs about Cell Phone Base Antennas and Human Health" (http://www.mcw.edu/gcrc/cop/cell-phone-health-FAQ/toc.html). Some of the general issues involved with RF and MW exposure are covered in Q2, Q3 and Q7. For standards and regulations concerning occupational and environmental exposure to RF and MW sources see the ICNIRP guidelines [M3].

Henshaw and colleagues [H25, H52] speculated that the radioactive decay products of radon [H25], and other potentially-carcinogenic airborne particles [H52], might be attracted to strong power-frequency electric field sources, and that there could be enhanced exposure to such carcinogenic agents near high-voltage power lines. They went on to theorize that this provided a mechanism for an association between power lines and childhood leukemia.

In 1999, Henshaw and colleagues [H53] amended their hypothesis to suggest that ions produced by corona from high voltage power lines might attach to aerosol pollutants (for example, motor vehicle exhaust) and increase the probability that these pollutants would be deposited in the lung. The authors have so far presented no evidence that this increased pollutant exposure actually occurs; and have offered no plausible mechanism whereby any such increase, if it occurred, would lead to an increase in childhood leukemia.

The basic observation of increased deposition of radon daughter containing aerosols on very strong electric (not magnetic) field sources is plausible [H54]. However, there are major theoretical problems with the Henshaw/Fews hypotheses which indicate that the postulated mechanisms are extremely unlikely to produce adverse human health effects under real-world exposure conditions [H28, H40, H54, L47, H61].

There are particular problems with the suggestion that the Henshaw/Fews hypotheses could explain the alleged connection between powerlines and childhood leukemia:

• Residences along powerlines do not appear to have elevated electric fields [C6, C12], and it is elevated electric (rather than magnetic) fields that the Henshaw thesis requires.
• There is no evidence that children who live along high-voltage power lines spend enough time out-doors in very high intensity electric fields for any particle deposition effects to be biologically significant [C44, C46, H54, L47, H61].
• The residential epidemiological studies that have looked at both electric and magnetic fields have found that the association (where there is any) is for the magnetic, not the electric field [C6, C12].
• Elevated radon exposure is linked with adult lung cancer (not reported in excess near power lines) [L20], but it is not associated with childhood leukemia [L33, L40, L41].
• Outdoors, power line electric fields might be strong enough to concentrate radon daughter aerosols, but the outdoor concentrations of radon is generally very low.
• Martinson et al [F20], using solid-state dosimeters, have shown that ionizing radiation levels are not elevated around high-voltage powerlines; and Burgess et al [F23] reported similar results.
• Miles and Algar [F31] and McLaughlin and Gath [L46] also measured radon daughters under high voltage power lines and found that the concentration was not elevated.

• No one has found a consistent association of cancer with occupational exposure to electric fields.
• Increased exposure to radon and aerosol pollutants would be expected to increase lung, skin and oral/throat cancer [L20], none of which have generally been found in excess in "electrical occupations".

In a letter to the journal in which Henshaw published his original hypothesis, Jeffers [H40] commented:

"Although the phenomena demonstrated by Henshaw et al are interesting... their own data show that DC fields are far more effective in producing [radon-containing] aerosol plate-out than AC fields. The DC fields that occur naturally and the intensity of man-made AC field strengths are well documented and lead to the view that, even for people who are occupationally exposed to high average AC fields, the additional plate-out [of radon-containing aerosols] is unlikely to exceed a few per cent..."

A syndrome, now called "sensitivity of electricity" or "electrosensitivity" first appeared in Norway in the early 1980's among users of VDTs. In Sweden "the problem has grown to epidemic proportions" according to one author [L25]; but until recently, there are few reports of the syndrome from other parts of the world [L38]. Initial reports were largely of a transient skin reaction, but in more recent years the syndrome has included central nervous system, respiratory, cardiovascular and digestive symptoms [L25, L38]. In double-blind studies published to date, patients with self-reported "sensitivity of electricity" have been unable to consistently sense whether a masked VDT was on or off [L25, L30]. Some consider that the syndrome is most likely a psychosomatic disease [L25].

In a 1999 review, Silny [L38] observes that:

1. The phenomena of "electrical hypersensitivity" cannot be explained by any known mechanisms, as the threshold for known interactions are at least 50 times higher than actual exposures levels.
2. The prevalence of the syndrome varies by a factor of 1000 or more between countries that have comparable exposure situations (for example, over 1000 cases per million people in Sweden versus less that 2 cases per million people in Italy, France and Britain).
3. The pattern of symptoms varies from country to country (For example, in Sweden most subjects report only skin symptoms, whereas in Denmark a wide range of symptoms are reported).
4. The types of exposures alleged to cause the syndrome varies from country to country (For example, in Sweden and Finland the syndrome is associated largely with work on video display terminals, whereas in Germany the syndrome is associated with power-frequency sources and radio/TV transmission towers).

This is not a question for which FAQ provides a direct answer. Rather, the goal of the FAQ is to suggest approaches to answering the question, and to provide a referenced and up-to-date summary of what is known and what is not known about the science.

Certain general conclusions can be drawn from the science:

• There is a broad consensus in the scientific community that no causal association has been established between residential exposure to power-frequency fields, and human health hazards.

• There is a broad consensus that exposure to these fields has not been, and cannot be proven to be absolutely safe.

• There is also a growing consensus that if there is a human health hazard, it is either very small or restricted to small subgroups; that is, that the possibility of a large and general hazard has been ruled out.

• The scientific controversy is over whether power-frequency fields might be shown to be hazardous by some future set of studies; and the related issue of what additional studies should be done, and what priority those studies should be given.

Notice

This FAQ is Copyright©, 1993-2001, by John E. Moulder, Ph.D. and the Medical College of Wisconsin, and is made available as a service to the Internet community. Portions of this FAQ are derived from the following four articles, and are covered by the Copyrights on those articles:

• JE Moulder and KR Foster: Biological effects of power-frequency fields as they relate to carcinogenesis. Proc Soc Exp Med Biol 209:309-324, 1995.
• JE Moulder: Biological studies of power-frequency fields and carcinogenesis. IEEE Eng Med Biol 15 (July/Aug):31-49, 1996.
• KR Foster, LS Erdreich, JE Moulder: Weak electromagnetic fields and cancer in the context of risk assessment. Proc IEEE 85:733-746, 1997.
• JE Moulder: Power-frequency fields and cancer. Crit Rev Biomed Engineering 26:1-116, 1998.
• JE Moulder: Une approache biomédicale: le point de vue d'un chercheur en cancérologie. In: J Lambrozo, I Le Bis (Eds), Champs Électriques et Magnétique de Très Basse Fréquency: Electricité de France, 1998.
• JE Moulder KR Foster: Is there a link between exposure to power-frequency electric fields and cancer? IEEE Eng Med Biol 18(2):109-116, 1999.
• JE Moulder: The Electric and Magnetic Fields Research and Public Information Dissemination (EMF-RAPID) Program. Radiat Res 153:613-616, 2000.
• JE Moulder: The controversy over powerlines and cancer, III Jornadas sobre Líneas Eléctricas y Medio Ambiente, Red Eléctrica de España, Madrid, 2000, pp. 159-168.