Dorsten-rf Cause Cancer

Do power-frequen y elds ause an er? Matthew P. Dorsten

Department of Physi s

The Ohio State University, Columbus, OH 43210

2 De ember 1999

Abstra t Although publi fear of a onne tion between an er and low-frequen y ele tromagneti elds has in reased re ently, publi poli y regulating the pla ement and density of residential power lines reating these elds is unne essary. Data from over 500 studies support this on lusion. Spe i ally, the relative risk asso iated with these elds is small, studies produ ing a onne tion are in onsistent, there is no \dose-response" relationship between eld exposure and in reased an er rate, and laboratory tests of eld exposure show no risk of an er. In addition, the onne tion is theoreti ally unlikely, for no plausible biologi al me hanisms linking eld exposure and an er exist.

1

Power-frequen y elds are the extremely low-frequen y (usually about 60 Hz) ele tromagneti (EM) elds that are produ ed by residential power lines and ele tri al devi es used in the home (e.g., ele tri shavers). Con ern about the adverse health ee ts of these elds arose in 1979, when a study showed that hildren living lose to

ertain types of ele tri al wires were 50% more likely than average to develop hildhood leukemia [1℄. Other studies sin e have found onne tions as well. Despite these results, most re ent studies make the onne tion very unlikely. Indeed, a ommittee of the National Resear h Coun il examined 17 years of data ontained in over 500 studies to

on lude that \no on lusive and onsistent eviden e shows that exposure to residential ele tri and magneti elds produ es an er [2℄." Thus, publi poli y regulating the pla ement and density of residential power lines is unne essary. The reasons are both epidemiologi al and theoreti al. Those epidemiologi al are that the relative risk asso iated with these elds is small, studies produ ing a onne tion are in onsistent, there is no \dose-response" relationship between eld exposure and in reased an er rate, and laboratory tests of eld exposure show no risk of an er. The theoreti al reason is that there is no physi al or biologi al motivation for an er development resulting from EM eld exposure. The rst epidemiologi al eviden e against the onne tion between EM elds and

an er is that the relative risk of exposure (RR)|that is, the risk of an exposed person getting an er relative to that of an unexposed person|is very small. The relative risk quanti es the ex ess an er found in epidemiologi al studies, and a RR value of greater than 5.0 generally is onsidered a strong asso iation [3℄. A RR value of 1.0 means the 2

exposed groups show no ee t. The re ent epidemiologi al studies almost all produ ed RR values of less than 2.0 [3℄. Establishing absolute risk requires nding a high relative risk or, if the relative risk is low, on rming the RR in many independent studies. Be ause 2.0 is a relatively small RR value, a on rmation of any orrelation between an er and EM elds would require onsistent results from a large number of experiments. Su h results do not exist. Even most of the positive studies (i.e., those studies nding RR values greater than 1.0) nd RR values less than 2.0. Only a few have RR values around 5.0. The initial hildhood leukemia study found only a RR value of 2.54 [1℄. The in onsisten y of studies laiming a onne tion is a se ond pie e of eviden e against that onne tion. Many of these studies are both in onsistent with ea h other and internally in onsistent [3℄. There are studies that demonstrate onne tions between low-frequen y elds and ertain types of an er, but there are also others denying these

onne tions while laiming onne tions with other types of an er. For example, a 1993 Swedish study found a positive onne tion (RR = 2.45) with hildhood leukemia but none with hildhood lymphomas, while a Danish study using similar measures and methods found pre isely the opposite: a onne tion with lymphomas but not with leukemia (RR = 1.0) [4, 5℄. These dis repan ies undermine the validity of positive studies and make the existen e of numerous positive studies meaningless, sin e they are not onsistent with one another. Internal in onsisten ies make many of the studies laiming a onne tion espe ially dubious. For example, the Swedish study mentioned above found a positive asso iation 3

(RR = 2.45) of hildhood leukemia with al ulated elds ( elds estimated from the

on guration of surrounding power lines) but a negative asso iation (RR = 0.40) with measured elds [4℄. Clearly, either the estimation pro ess or the eld measuring pro ess has aws. A number of other positive studies reveal similar defe ts. This fa t alone for es one to dis ount the results of ertain positive studies. The ina

ura y of the methods used in many studies to estimate eld strengths resolves the apparent ontradi tion above. Field estimates in many studies were based on fa tors su h as the size of the power lines going past homes and the distan e between the power lines and homes. Estimating elds in this way is a heap substitute for a tually measuring the elds, but often it is unreliable. Part of the problem is that many studies used histori al wire sizes and on gurations that had hanged by the date of the studies [3℄. The ompli ated nature of al ulating elds from power lines also makes estimation ina

urate, even if the wire statisti s are orre t. These fa ts indi ate that poor experimental methods have ontaminated the results of ertain positive studies. Another epidemiologi al fa tor that questions the validity of positive studies is the la k of a \dose-response" relationship between measured elds and an er rates. For example, while the initial hildhood leukemia study found a orrelation between living near power lines and developing leukemia, it found no asso iation between the measured strengths of the elds and leukemia in iden e [1℄. Just as smoking two pa ks of igarettes a day is more likely to ause lung an er than smoking one pa k, stronger elds or more exposure to the same elds should produ e a greater biologi al ee t, if 4

EM elds produ e any ee t at all. Be ause ause-ee t relationships generally obey this prin iple, this is one of the most onvin ing pie es of eviden e against the initial leukemia studies. A nal reason to doubt the orrelations found by some epidemiologi al studies is that these orrelations annot be reprodu ed in the laboratory. Naturally, any biologi al ee t produ ed by residential elds should be reprodu ible in the laboratory. However, there is no laboratory eviden e for an asso iation between exposure to lowfrequen y elds and the risk of an er [3℄. Experiments using elds similar in strength to residential elds produ e no dete table ee t on biologi al material. In fa t, ells do not show any response to elds at all until the eld levels are near 1,000 times stronger than ommon residential levels [6℄! The ell response at this level is only a slight modi ation of the way ells ex hange hemi al signals, well short of the DNA damage that ar inogens generally produ e. These fa ts suggest the most signi ant short oming of the ontention that EM eld produ e an er: a orrelation between low-frequen y eld exposure and the risk of an er la ks a plausible biologi al me hanism to support it. A s ienti onne tion is diÆ ult to formulate be ause su h low-frequen y EM elds appear to have no ee t on ells. In short, there simply is no reason to suspe t that these elds pose us any risk. The me hanism through whi h an er develops in the body is well understood. Current resear h indi ates that an er generation is a multi-step pro ess involving a series of injuries to the DNA [3℄. In this pro ess, some normal ells suer DNA 5

damage to be ome pre- an erous ells. This DNA damage is the result of environmental in uen es, whi h may or may not in lude exposure to EM elds. Exposure to geneti in uen es over time then turns these pre- an erous ells into an erous ells, whi h develop into tumors. Consider the way EM radiation produ es biologi al ee ts. The intera tion between EM radiation and matter is dependent on the energy of the radiation. Fig. 1 shows the EM energy spe trum and the energies required for ertain types of intera tions. At high frequen ies, the energy arried by elds is relatively large, and radiation intera ts strongly with matter. It often has enough energy to break hemi al bonds (i.e., ionize mole ules). Thus, high-energy radiation an damage DNA and lead to an er. For example, X-rays are high-energy radiation, and too mu h exposure an lead to an er, as the in reased in iden e of foot an er resulting from shoe-store X-rays earlier this

entury demonstrates [7℄. EM elds of middling energy produ e thermal ee ts through relatively large indu ed urrents, a me hanism that allows mi rowave ovens to heat food. These thermal ee ts an kill ells by heating, but these EM elds appear to produ e no permanent DNA damage. For this reason, they are not ar inogen andidates. Power-frequen y elds are at the very end of the spe trum, indi ating that they

arry very little energy. These elds have too little energy, by a fa tor of 1010 , to break even the weakest hemi al bond [3℄. They do indu e urrents in the body, but these

urrents are very small. To experien e indu ed urrents only as big as those naturally produ ed as the heart beats and nerves fun tion, one would have to be half a meter 6

Figure 1: The ele tromagneti spe trum. Power-frequen y elds are in the range of the spe trum where elds intera t imper eptibly with biologi al material. This makes it diÆ ult to nd a biologi al me hanism linking the ee ts of these elds with an er.

away from a typi al residential power line [2℄! As this is the living on guration of very few people, natural urrent densities in the body are usually mu h bigger than those indu ed by power lines (by a fa tor of 103 ). Thus low-frequen y elds annot ause DNA damage either by indu ed urrents or by breaking bonds. Although they do not have enough energy to break hemi al bonds or indu e large enough urrents to produ e heat, low-frequen y ele tri elds an exert for es on

harged mole ules within tissues. These for es an deform ells, orient polar mole ules, or indu e voltages a ross ell membranes. Su h ee ts on eivably ould damage DNA, but those due to power lines are virtually negligible be ause ele tri elds from power 7

lines have very little ability to penetrate houses or even skin [3℄. The magneti omponent of the eld also annot damage DNA. Magneti elds easily penetrate homes and humans, but they exert only small dire t for es on ellular stru tures, whi h are primarily non-magneti . Magneti elds only ae t ells through the urrents that they indu e in the body. As mentioned above, indu tion from power lines is insigni ant. Thus low-frequen y EM elds do very little to ells. They do not ontribute to DNA damage, one of the hallmarks of ar inogens, and so there is no need for publi poli y regulating the pla ement of power lines. Support for this on lusion omes both from the epidemiologi al data of many studies over many years and from s ienti hypothesis. The positive results of the initial hildhood leukemia studies most likely

an be explained by fa tors other than low-frequen y elds, but new resear h is needed to identify the spe i ause or auses, both out of s ienti uriosity and to allay publi fears.

8

Referen es [1℄ N. Wertheimer and E. Leeper, Am. J. Epidem. 109, 273 (1979). [2℄ National Resear h Coun il, Possible health ee ts of exposure to residential ele tri

and magneti elds, (National A ademy Press, Washington, D.C., 1997). [3℄ John E. Moulder, \Ele tromagneti elds and human health," (Online: http://www.m w.edu/g r / op/powerlines- an er-FAQ/to .html), 10 September 1999. [4℄ M. Fey hting and A. Ahlborn, Am. J. Epidem. 7, 467 (1993). [5℄ J.H. Olsen et al., B.M.J. 307, 891 (1993). [6℄ S ienti

Ameri an,

\Ask

the

Experts,"

(Online:

http://www.s iam. om/askexpert/medi ine/medi ine25.html), 4 August 1997. [7℄ This is ommon knowledge, at least for those of a ertain age.

9