Getting Risk Right

The question of whether exposure to radiofrequency energy from cell phones is carcinogenic erupted into the public arena in 1993 when David Reynard, a resident of St. Petersburg, Florida, brought a lawsuit against a mobile phone manufacturer, alleging that his wife’s fatal brain cancer had been caused by her using a cell phone, which she held on the same side of her head as that on which the cancer developed. Although he lost his suit, Reynard went on Larry King Live, voicing his certainty that his wife’s cancer had been caused by her prolonged conversations on her cell phone and attracting widespread media attention. “She held it against her head, and she talked on it all the time,” he explained.¹ Owing to the novelty of wireless technology that was poorly understood by the public and the fact that little is known about what causes brain cancers, this dramatic anecdote was to have an enormous impact.² Among the misunderstandings that circulated then and now and that inflame public concern is the notion that electromagnetic fields from mobile phones are “radiation,” namely, ionizing radiation, like X-rays and gamma rays. Health and scientific agencies know that electromagnetic fields from mobile phones are non-ionizing and so are distinctly different from ionizing radiation sources.

Within a week of the broadcast a congressional hearing was held urging the federal government and the wireless industry to undertake studies to examine the health effects of cell phone radiofrequency emissions. The National Cancer Institute was already engaged in studying the causes of brain cancer, and researchers there added questions regarding cell phone use to their questionnaire. The Cellular Telecommunications Industry Association set up a $27 million project to examine the health effects of cell phone use.

Several early studies appeared between 1996 and 2001 showing little indication of a link between use of cellular or mobile telephones and brain cancer, and these studies might have laid to rest concerns about hazards of the new technology. In fact, in an editorial in the Journal of the National Cancer Institute in 2001, the physicist Robert L. Park cited the failure of a large prospective study of Danish cell phone subscribers, along with basic biophysical considerations, as settling the issue.³ A number of factors, however, were to ensure that the question of carcinogenic cell phones would remain in the public consciousness: First, use of cell phones was still limited, and the average duration of calls was low, but their use was expanding at a rapid pace, and the technology was undergoing change. Second, the fact that new studies were planned or in progress in many different countries assured that new results would be appearing and getting attention. Finally, the question of cell phones and cancer arose at the precise moment when fear of extremely low-frequency electromagnetic fields (EMFs) from power lines and electric motors and appliances was at its height,⁴ and anxiety about these two forms of non-ionizing radiation shaded into each other.

By 2013 there were 300 million subscribers to cell phone service in the United States and nearly six billion subscribers worldwide. Commentators raised the possibility that availing ourselves of this incredibly useful—and now all but indispensable—technology could cause an “epidemic of brain cancer” in the future, and some characterized the expansion of wireless technology, with its unknown effects, as the greatest uncontrolled experiment ever conducted on human beings. Given the unprecedented uptake of this new technology, whose potential effects on health were poorly understood and difficult to study, many scientists and health officials voiced concern about possible consequences. As the epidemiologist Kenneth Rothman, who has been involved in studying the health effects of mobile phone use since the issue first arose in the mid-1990s, put it, “The safety of mobile telephones is a pressing question, now that the brains of nearly half the humans on the planet have become exposed within a short span of time to a physical agent to which their ancestors’ genes could not have adapted.”⁵

The question of cell phones and cancer provides a prime example of how what appears to be a purely scientific question can be influenced by factors that have little or nothing to do with science and by the reciprocal interactions between the scientific community—and even within subgroups of the scientific community—and the larger society, including the general public, special interest groups, and health and regulatory agencies. Perhaps the most important bias of all those discussed in the preceding chapter is that of pretending that “the science is the science” and can be insulated from the fears, misconceptions, and agendas that are set in motion by the publication of “scientific findings” and dramatic anecdotal reports, such as that of David Reynard. It is disingenuous of scientists in the field of public health to pretend that their research is not affected by, and does not have effects in, the wider public sphere. However, before examining the science—or, rather, different versions of science—relating to cell phones and brain cancer, we need to briefly examine the basic facts about the two phenomena that are being linked.

Tumors of the brain, which can be benign or malignant, are extremely rare. Each year in the United States there are approximately 13,000 new cases in men and 10,000 cases in women.⁶ For comparison, there are roughly 230,000 new cases of prostate cancer and an equal number of breast cancer cases. Cancers of the brain account for 1.4 percent of all cancer. Little is known about what causes these cancers. The only established risk factors are, in fact, exposure to ionizing radiation and certain rare hereditary conditions, which account for only a small proportion of cases.

To understand what is meant by cell phone “radiation,” we need to situate this type of emissions in the electromagnetic spectrum (fig. 4.1). Electromagnetic energy consists of electromagnetic waves that are oscillations of electric and magnetic fields that travel at the speed of light. Electromagnetic energy can be characterized by its frequency (the number of oscillations per second) or wavelength (the distance between the crests of two waves). The higher the frequency, the greater the energy carried by the waves and the shorter the wavelength. Thanks to quantum mechanics, we know that the energy of the waves comes in discrete packets called photons. A given total energy of lower-frequency waves is made up of many very low-energy photons. The energy of higher-frequency waves is made up of fewer higher-energy photons.

Electromagnetic energy exhibits an enormous range from gamma rays, which carry the highest-energy photons, to 60-Hertz waves from the electric power distribution system, which are made up of many very low-energy photons. Gamma rays and X-rays have very short wavelengths and very high photon energies, and for this reason they can damage molecules in our cells by “ionization”—that is, by knocking an electron out of an atom. The region of the spectrum that is involved in wireless telecommunications is in the kilohertz to gigahertz frequency range, spanning FM radio waves and microwaves. (For this reason, both terms are used in discussing the health effects of cellular telephones in the scientific literature.) This band of the spectrum is well below that of visible light and infrared energy, and the photon energies are far too weak to cause ionization. Microwaves can cause heating—as in microwave ovens—but only at much higher power levels than those used in wireless communications. In fact, the highest frequencies used in wireless communications, roughly 2000 MHz, are such that the photon energies are much smaller than the relevant biological energies.

A cell phone user’s main source of exposure to radiofrequency energy comes through the antenna embedded in the phone, and his or her degree of exposure depends on a number of factors, including the characteristics of the phone and the distance of the antenna from the head; the greater the distance, the lower the exposure. Exposure is also influenced by the strength of the signal sent to the nearest base station, which is determined by how strong a signal the handset is receiving (i.e., the weaker the signal from the base station, the higher the exposure from the handset).⁷

The main effect of microwaves on living organisms is heating. A number of other effects (not related to heating) have been demonstrated, but these require high exposure levels. Standards for exposure to RF and microwave fields for workers and the general population have been in existence for decades, and these are based on microwaves’ ability to heat tissues by transferring energy to molecules. The rate at which energy is absorbed by human tissues is measured by the specific absorption rate (SAR), and it is this value that is regulated to limit a cell phone user’s exposure. In the United States, the Federal Communications Commission (FCC) has adopted the standard of 1.6 watts per kilogram, averaged over 1 gram of tissue for the head, recommended by the Institute of Electrical and Electronic Engineers. In Europe, the limit is 2 watts per kilogram averaged over 10 grams of tissue set by the International Commission on Non-Ionizing Radiation Protection (ICNIRP). These standards are conservative, providing a margin of safety that takes into account “worst-case scenarios.” The underlying assumption is that with an adequate safety margin below the level at which microwaves can heat tissues, the likelihood of adverse biological effects is essentially zero.

Still, questions remain about the impact of exposure to RF on human tissues. One of these is the concern, voiced by some, that the limit for the amount of RF energy set by the FCC is based on the dimensions of an average adult male and therefore may be inadequate to protect women and particularly children, whose heads are smaller. Another area of concern stems from extensive laboratory studies that attempt to gauge the effects of RF exposure on cells and test animals. These studies are difficult to perform, and the aspects of exposure that are most relevant are unknown.⁸ In this situation, there have been many claimed effects, and concerned groups have seized on these as evidence that RF may indeed be subjecting cell phone users to imperceptible damage that could lead to cancer or other diseases.

But the greatest problem confronting both types of studies (case-control and cohort) is that of accurately assessing an individual’s exposure to radiofrequency energy. To date, epidemiologic studies have relied on rather crude proxies for actual exposure.⁹ Typically, study participants in a case-control study are asked questions about when they first used a mobile phone, how many calls they make and receive each day or week, how many minutes they usually talk on the phone per day, and on which side of the head they usually hold the phone, if they have a preference. This requires respondents to recall their usage pattern going back as much as a decade or more. But usage patterns may have changed, and recall may be faulty. Moreover, people can hold the device in different ways that may affect their exposure, and they may have used hands-free attachments for some portion of their usage history. Another strategy that has been used in cohort studies has been to use billing records of mobile phone subscribers to estimate usage. While this approach avoids the problem of recall bias, it has other limitations as a measure of RF exposure (e.g., billing records may not capture actual exposure).

Another complication is that cellular technology has continued to develop rapidly over the past twenty years and has gone through four generations, including analog, digital, and digital UMTS. This further complicates the assessment of an individual’s exposure over time.

Finally, brain tumors can take several decades to develop. But cell phones have only been widely used for the past ten to fifteen years. This means that not enough time has elapsed to gauge the full effects, if any, of exposure to RF. What we really would like to know is the effect of lifetime use of these devices, starting at an early age and at current levels of use (which for some people can amount to hours per day). Most currently available studies provide information on only a relatively short duration of use, at generally lower levels, and to earlier generations of cell phones.

These points are crucial to bear in mind when interpreting the results of studies conducted to date, and so their interpretation and the weight of different considerations requires care and an awareness of their substantial limitations.

As mentioned earlier, the timing of this new “radiation” hazard was significant in that it arose at the height of the furor surrounding the possibility that extremely low-frequency electromagnetic fields from power lines and electric appliances posed a threat of childhood leukemia, brain tumors, and other cancers. Many scientific studies of EMFs were in progress at that time (in the 1990s), and both the federal government and the electric power industry had programs devoted to research and education in this area. The media reported the results of each new study that seemed to hint that exposure to EMFs was associated with an ever-growing list of diseases—various types of cancer, heart disease, depression, Alzheimer’s, Parkinson’s disease, miscarriage, and so on. It was common to have scientists interviewed on the nightly news about the results of the latest study, with the ominous image of high-voltage transmission lines as the backdrop. Within a few years, as better studies were published, the notion of a threat from EMF exposure to the general population lost support in the scientific community.¹⁰ The fear instilled in the public was longer-lived, however, and the focus on this new form of non-ionizing radiation—radiofrequency fields—picked up where EMFs had left off.

New studies continued to come out in the first decade of the new millennium. A number of these were part of a large and ambitious collaborative project, which was conceived in the late 1990s, when several groups of scientists recommended that the International Agency for Research on Cancer, an arm of the World Health Organization, investigate the relationship between cell phone use and brain tumors. After conducting a feasibility study, the agency determined that a multicountry case-control study would be both feasible and informative. The resulting INTERPHONE study was a population-based case-control study carried out in thirteen countries (four Scandinavian countries, France, Germany, Italy, the United Kingdom, Australia, New Zealand, Canada, Israel, and Japan), using a common protocol. The study focused on cases of brain tumors occurring in younger people, 30–59 years of age, since this group was expected to have the highest prevalence of cell phone use in the previous five to ten years.¹¹ The four tumor types included in the study were those occurring in tissues most likely to absorb RF energy emitted by cell phones: tumors of the brain (glioma and meningioma), acoustic nerve, and parotid gland. The data collection phase of the study ran from 2000 to 2004.

The results from a number of individual countries or groups of countries participating in INTERPHONE were published in the 2000s, and these gave some insight into the overall study results. It was not until May 2010, however, that the combined results for all participating countries were finally published. To appreciate the significance of the results when they finally came out, it is important to describe views on the question of cell phones and brain tumors held by both scientists and advocates who were to articulate the messages that influenced the regulatory as well as the public discussion.

By 2009 nearly thirty studies of cell phone use and risk of various tumors of the brain and head had been published. A critical assessment of this body of evidence by the Standing Committee on Epidemiology of the International Commission for Non-Ionizing Radiation Protection (ICNIRP) concluded that:

This paper, written by a number of epidemiologists who have been involved in research on the health effects of non-ionizing radiation for decades, is noteworthy for its thoroughness and clarity. It attempts to provide an impartial summation of the human evidence, acknowledging the many limitations and potential biases inherent in the studies but at the same time providing some measure of reassurance, owing to the lack of any strong or consistent signal indicating a carcinogenic hazard. For this reason, the ICNIRP assessment stands in contrast to what has been made of epidemiologic and other data relating to cell phones by a small but vocal group of advocates, including some scientists, who take a very different view of the cell phone issue and who have had a disproportionate influence on the tone of the public discussion. The view articulated by ICNIRP in periodically updated publications is embraced by the majority of mainstream cancer epidemiologists and is in agreement with assessments by other groups.¹³

Studies from one group of investigators, led by the Swedish oncologist Lennart Hardell, stand out from the majority of studies and have been referred to as “outliers” or as discordant in a number of critical assessments of the overall evidence.¹⁴ Starting in 1999, Hardell and colleagues published a series of studies that appear to show evidence of an increased risk of certain tumor types among long-term users of mobile phones. For example, in a paper published in 2006, these researchers reported that users of digital cell phones had a near doubling of risk of glioma (odds ratio 1.9) and that long-term users had more than a threefold increased risk (odds ratio 3.6). In more recent publications, Hardell and colleagues have reported that tumor risk was increased on the side on which was cell phone was held, and Hardell has declared RF energy to be a known carcinogen. However, cautious interpretation of these results is warranted for a number of reasons. First, as is the case for all case-control studies, there is concern about possible recall bias (that is, that cases with brain tumors may recall their past use differently from “controls,” and that they may emphasize their exposure in order to account for why they developed their illness). It is also noteworthy that in the 2006 report even “short-term” users of digital cell phones (those who reported using them for one to five years) showed evidence of an increased risk (odds ratio 1.6). As we will see shortly, this result is at variance with the results of the much larger INTERPHONE study, as well as with the results of other studies from Sweden. Furthermore, it has been pointed out that, if such short duration of use were associated with brain cancer, this would be apparent in nationwide Swedish cancer rates after 2002–03 (the years cases were diagnosed in the Hardell study).¹⁵ In fact, Swedish brain tumor rates have shown no increase through 2009. And there are similar data for the United States and the Nordic countries as a whole.¹⁶

In addition, questions have been raised about the methodology of the Hardell studies. For example, although the researchers used mailed questionnaires to obtain information from cases and controls, they reported participation rates of nearly 90 percent, which are highly unusual for studies using mailed surveys. Finally, the results reported by Hardell and colleagues contrast with those from almost all other studies (including cohort studies, where recall bias is not an issue). In the ICNIRP assessment from 2009 of epidemiologic studies, the authors pointed out that, if one analyzed the data on glioma omitting the Hardell studies, there was no evidence of an increased risk for either short-term or long-term use. Other recent overviews have made the same point.¹⁷

In spite of the questions surrounding the Hardell studies, activists have seized on his findings as confirming their conviction about the adverse effects of RF exposure. Starting in 2007 groups in the United States published a number of reports alerting the public to the possible dire effects of cell phones on human health. First to appear was the “BioInitiative Report,” which declared that “existing public safety limits” on the radiation from phones and other wireless technologies are “inadequate.”¹⁸ Next, in the summer of 2008 the head of the University of Pittsburgh Cancer Institute, Dr. Ronald Herberman, issued an unusual alert to the school’s faculty, staff, and students citing new, but unpublished, evidence that cell phone use causes brain cancer.¹⁹ In the summer of 2009 two additional reports appeared: Cell Phones and Brain Tumors—15 Reasons for Concern: Science, Spin and the Truth Behind Interphone and a report from the Environmental Working Group entitled Cell Phone Radiation: Science Review on Cancer Risks and Children’s Health.²⁰

Two of these authoritative-appearing documents had as “endorsers” or “participants” diverse rosters of PhDs and MDs, and others, including neurosurgeons, general practitioners, politicians, lawyers, educators, and a firefighter.²¹ Two of the documents were adorned with suggestive images adding to their scientific cachet—in one case, what looked like a radiological scan in vivid colors showing the radiation from a cell phone penetrating the brain; in another, assorted images of cell phone towers, high-voltage transmission lines, and zeros and ones projected on the back of a man’s skull.²²

The basic thrust of these reports was to argue that there is credible evidence that mobile phone use is associated with increased risk of brain cancer and nonmalignant tumors of the brain, then invoke the “precautionary principle” and counsel “prudent avoidance” to reduce one’s risk, and particularly that of children.

To the lay reader, and even to many scientists and physicians who were not familiar with this subject, these reports were likely to appear to be serious and impartial assessments coming from independent-minded scientists concerned about the public’s welfare. Their authors, we are given to believe, are speaking out in order to expose the flaws of industry-funded research and inadequate government regulation, and to expose the truth that is being suppressed by powerful interests. And, as they were intended to do, these alarming reports received widespread coverage in the media.

What the reports have in common, and what is most striking to anyone who is conversant with the scientific evidence concerning the health effects of cell phone use, is the highly selective and slanted presentation they give of the relevant evidence. While pointing to the findings of various studies as cause for alarm, the reports studiously avoid a number of crucial considerations that would help put the matter in perspective. In other words, their agenda involves something other than attempting to provide the kind of critical and demanding assessment of the totality of the relevant evidence that, as we saw in chapter 2, is the only way to arrive at a sound, if provisional, judgment about a public health issue.

In citing results from the epidemiologic studies, they fail to discuss the weaknesses and limitations of the studies in an impartial way. One element of such a discussion is to recognize that these limitations and biases could act in opposing directions—they could result in a failure to detect evidence of a real hazard, but they could also spuriously create the appearance of a hazard where none exists. One of the glaring symptoms of the activists’ tendentious approach to the evidence is that they tend to pay attention only to the former possibility and not the latter.

It is less Hardell’s results than his style of argumentation that calls into question his objectivity. He and his colleagues write as if their positive results are to be taken at face value, that is, as evidence of a risk. This reflects an uncritical approach to data from observational studies in general and from studies on the question of cell phone use and brain cancer in particular. Hardell and colleagues seem to fall into the trap of equating “association” with “causation.” It appears that they are believers, and they marshal the evidence in a selective manner to support their belief. Given his aggressive advocacy, it is highly significant that Hardell was one of the “participants” involved in the “BioInitiative Report.”

In addition to the overview by the ICNIRP group, two other meta-analyses on cell phones and brain tumors were published in 2009. One of these, by V. G. Khurana, Hardell, and colleagues, concluded that “the results indicate that using a cell phone for >10 years approximately doubles the risk of being diagnosed with a brain tumor on the same (“ipsilateral”) side of the head as that preferred for cell phone use …. The authors conclude that there is adequate epidemiologic evidence to suggest a link between prolonged cell phone usage and the development of an ipsilateral brain tumor.”²³ The ICNIRP report presents a table displaying similar results, but some of the studies included by Khurana/Hardell and ICNIRP differ, and the ICNIRP had refrained from presenting a pooled estimate of risk for ipsilateral glioma because it felt that the data showed evidence of recall bias. In other words, people with a brain tumor may be more likely to misreport the side on which they held the phone as the side on which the tumor occurred compared to healthy controls.

The second meta-analysis was performed by researchers from South Korea and from the University of California at Berkeley.²⁴ The senior author is Joel Moskowitz of the University of California at Berkeley, and I will refer to this as the Moskowitz paper. This paper represents a curious exercise. Without providing any rationale for their approach, the authors divided studies into those that used “blinding” (i.e., the interviewers and researchers did not know who was a brain tumor case and who was a control) and those that did not. In the latter group were fifteen studies by different research groups in the United States, Europe, Israel, and Japan; in the former group were eight studies, seven of which were by Hardell and colleagues. When the studies using blinding were analyzed as a group, they showed a statistically significant association between any use of mobile phones and the risk of brain tumor, whereas there was no association in the group of studies that did not use blinding. However, the authors never justified their use of blinding as the primary criterion to judge the quality of the studies, and it was never made clear how the absence of blinding would mask an association. In reality, the existence of blinding in the Hardell studies was due to the fact that they used mailed questionnaires to obtain information from participants, whereas in most other studies cases and controls were interviewed in person, making it difficult for the interviewer to be unaware of who was a case. The Moskowitz meta-analysis confirms the observation that the results of the studies by Hardell and colleagues differ from those of other studies, and, while it provides no insight into the reasons for this difference, its authors make the subjective claim that the Hardell studies are superior owing to their “low bias.”

It turns out that the person who apparently motivated Ronald Herberman, the founding director of the University of Pittsburgh Cancer Institute, to issue his unusual and unprecedented alert to his staff and community was the cancer researcher, author, and activist Devra Davis, who at the time held a position at the university.²⁵ Davis has long been on the front lines of efforts to identify linkages between environmental exposures and cancer. In the 1990s she was instrumental in persuading the federal government to conduct a study to examine links between environmental exposures and breast cancer on Long Island—which in the end turned up no evidence of any association.²⁶ She has postulated a role for endocrine disruptors in explaining a wide range of phenomena, from effects on wildlife to a role in human breast cancer.²⁷ More recently she has been calling attention to the potential dangers of wireless RF and cell phones in interviews, newspaper columns, and a popular book: Disconnect: The Truth About Cell Phone Radiation, What the Industry Has Done to Hide It, and How to Protect Your Family.²⁸

We need people who draw attention to potential health threats that have received inadequate attention. However, when advocates present themselves as scientists concerned about public health, one is entitled to expect an informed and critical consideration of all relevant evidence. The danger lies in latching on to certain findings that appear to signal the presence of a hazard, while ignoring the totality of the evidence and, equally important, the limitations of the types of studies being cited. Reading the publications of Davis and other cell phone activists provides a textbook exhibit of giving weight only to positive findings because these appear most convincing to people with a strong belief that there is something going on to which regulators and scientists need to pay attention. In this mindset, every elevated risk estimate becomes a signifier of a danger that is being denied by those who take a more critical view.

A few examples from Davis’s Huffington Post article (2010) will show what I mean. In the second sentence, she writes, “This thirteen country report found what every study that has ever examined people who have used phones for a decades [sic] or more has determined—top users of cell phones had a doubled risk of malignant tumors of the brain.” Only that is not what the INTERPHONE study reported. Rather than Davis’s “doubling”—which corresponds to a 100 percent increase—INTERPHONE actually reported a 40 percent increase in the risk of glioma among long-term users with the greatest cumulative number of hours (odds ratio 1.40, 95% confidence interval 1.03–1.89) (table 2). Davis’s reference to a doubling of risk refers to a subanalysis, which partitioned the data by whether the cases reported using their cell phone on the same side as the tumor or on the opposite side (table 5). There, the risk of glioma among “ispilateral” phone users with the highest cumulative call time is doubled, whereas the risk of glioma among “contralateral” users with the greatest call time shows a 25 percent increase (which is not statistically significant). This is just one example of selecting a particular result that supports one’s position, rather than accurately describing the full results. Furthermore, while Davis points to this result as unproblematic evidence, the ICNIRP report cautioned against putting weight on the risk for ipsilateral phone use because of indications that recall of “laterality” of phone use is biased. Davis goes on to cite the work of Moskowitz—again, incorrectly—and to refer to Hardell’s work (“regarded as some of the best efforts in the world on this challenging topic”), which, she tells us, “concur with the INTERPHONE and Moskowitz results.” We have just seen that Hardell’s results do not in fact “concur” with the INTERPHONE results. As for the agreement with Moskowitz, only the results for the “superior” (i.e., blinded) studies “concur,” and this is hardly surprising since they are based, with one exception, on studies by Hardell!

Another key component of the argument put forward by Hardell, Davis, Moskowitz, and others is, as Davis puts it, “that there is a growing experimental literature showing that pulsed micro-wave like radiation from modern cell phones disrupts living cells and causes our DNA to become unstable—signs of cancer and other chronic disease.”²⁹ The problem here—and one that the believers and activists seem blissfully unaware of—is that these studies are extremely hard to do, and it is difficult to know what the relevance of their findings is to actual human exposure. Even more to the point, most studies in this area tend not to have been replicated by independent researchers—the single most important criterion for judging the reliability of scientific evidence. Thus breathless references to these kinds of studies reveal an inability or an unwillingness to assess evidence in a critical manner.

The reports by activist groups invariably invoke the precautionary principle to clinch their argument that there is good reason to expect harm from use of cellular telephones. As was discussed earlier, while sounding eminently reasonable (“better safe than sorry”), the precautionary principle does not offer a clear guide to action.³⁰ Furthermore, when partisans seek to publicize a skewed version of the evidence, playing up certain findings and ignoring other crucial facts, invocation of the precautionary principle amounts to a rhetorical device to give their position a semblance of reasonableness to those who are unfamiliar with the real issues involved. (Another problem that is rarely acknowledged by those who invoke the precautionary principle is that there are risks associated with the status quo—not just with adopting some new technology or industrial process.)³¹

Finally, it is revealing that Hardell and associates involved in the “BioInitiative Report” chose to indict exposure not only to RF energy but also to the much lower frequency electromagnetic fields from power lines, electrical appliances, and other sources. Their reading of the evidence as of 2009—a full thirty years’ worth—shows an ability to screen out vast amounts of research that was done that does not support their contention that exposure to EMFs is a cause of leukemia, breast cancer, and other diseases.³² Reports like the BioInitiative document are really directed at people who are not acquainted with the extensive published literature, starting with the National Research Council report in 1997 and the large National Cancer Institute study in the same year on electromagnetic fields and childhood leukemia, which concluded that there was no persuasive evidence of an association. The activists’ modus operandi is made clear in their treatment of the question of the health effects of EMFs. Basically, they ignore the most powerful studies and the most comprehensive assessments, and in the isolated studies they point to they avoid making the crucial distinction between association and causation (that is they show no awareness of the need for caution in interpreting the results of observational studies); but they readily accept, without hesitation, the results of any study that purports to provide evidence of an effect of EMFs. This same approach is taken with RF.

Rather than adopting a critical attitude toward evaluating the evidence from epidemiologic studies, and referring to basic considerations that should be part of any informed discussion of the issue, these self-proclaimed experts utilize a number of tactics to argue for the existence of a hazard and to attempt to undermine results that do not support their position. One of these is the recurring claim that studies of cell phones that are supported by the wireless communications industry are less likely to find an association with brain tumors than studies not supported by industry.³³ Like the arbitrary use of blinding as a criterion for rating the quality of studies, this is a red herring, because INTERPHONE was only partly funded from industry sources and, in any event, the study was run by IARC and was well insulated from any influence from the telecommunications industry. To imply that the results of this high-profile study could have been manipulated, when an army of respected scientists was intimately familiar with all aspects of the study, is preposterous and merely reveals the cynicism of the believers.

Perhaps what is most disturbing is how a small group of highly motivated activists can present a distorted picture of the evidence that can have wide influence. They are careful to use the trappings of “science” in order to impress people who have no background in this area and to appeal to people’s unconscious fears about “radiation” and their insecurities about who is telling them the truth.

The long-awaited results of the entire INTERPHONE study—the largest study to examine the link between cell phone use and tumors of the brain and other tumors of the head, which cost twenty-five million dollars—were finally published in May 2010 in the International Journal of Epidemiology, more than ten years after its initiation. It included 2,708 glioma and 2,409 meningioma cases and matched healthy controls. The ambitiousness and complexity of the INTERPHONE study is hinted at by the nearly three pages of acknowledgement of the individuals involved in referring patients and collecting information at multiple collaborating sites within each country. The paper’s authors totaled forty-eight. But perhaps the most telling statistic, which is nowhere mentioned in the twenty-page article, is that it took over four years for the authors to agree on their interpretation of the data and approve a final draft. This conveys some indication of the difficulties of interpreting this type of data, reconciling different viewpoints, and reaching a consensus.³⁴

Overall regular use of a mobile phone, as compared with nonuse, was associated with reduced risk of both glioma and meningioma, a result the authors attributed to possible bias. Long-term users (ten or more years) did not have an increased risk. However, in an analysis that divided cumulative call time into ten categories, those in the highest category had an odds ratio for glioma of 1.40 (95% confidence interval 1.03–1.89), meaning a 40 percent increased risk over nonregular users. The second highest category, though, showed one of the lowest risks. Thus there was no suggestion of a trend toward increasing risk with increasing cumulative call time. The odds ratio for meningioma was 1.15 (95% confidence interval 0.81–1.62), or a nonsignificant 15 percent increased risk. Having reported these findings, the researchers were quick to state that “biases and error prevent a causal interpretation.”

The core message of the paper was thus an ambiguous one—that cell phone use did not appear to increase the risk of brain tumors overall, but that there was some suggestion that users with the greatest number of cumulative hours of use might have a slightly increased risk. The authors cautioned, however, that these positive results could be affected by a number of biases and could not be interpreted as evidence of a causal relationship. The paper received worldwide media coverage.³⁵ An editorial by two epidemiologists associated with IARC that accompanied the article in the journal pointed out that the conclusions of the article tolerated “diametrically opposed readings.”³⁶

The ambiguity of the INTERPHONE results allowed different groups to interpret the findings in conformity with their views. Three different positions are discernable. One group, including the members of ICNIRP, recognizes the problems with the study and fails to find any strong or clear-cut support for an association. While referring to what is known about RF energy, the ICNIRP authors take a balanced and reassuring view of the issue, while acknowledging the need for continued monitoring of brain tumor rates. A second group includes Hardell and scientists and activists aligned with him. This group finds confirmation of a risk in the few isolated blips in the results, which the INTERPHONE authors cautioned about taking at face value. The third, more difficult to characterize, group is composed of epidemiologists associated with IARC who organized the agency’s assessment of cell phones and brain cancer. We will come to this group and their more complicated and ambiguous position shortly.

In view of the biases inherent in case-control studies of cell phone use, the results of a nationwide cohort study from Denmark assume particular importance. Launched in the late 1980s, this study included all Danes thirty-five years of age and older who were born in the country after 1925.³⁷ The population was divided into subscribers and nonsubscribers of mobile phones. Over 358,000 subscribers were followed for eighteen years, and brain tumors were identified through the Danish Cancer Registry. No association of cell phone use, or of long-term cell phone use (ten or more years), with brain tumors was observed in either men or women in this study.

First, a crucial piece of evidence that would help to put the cell phone question in perspective is information about the rates of brain tumors and cancers over the past twenty to thirty years in different countries. Cell phone usage has increased at a geometric rate over the past twenty years in the United States (fig. 4.2). In contrast, brain tumor incidence increased during the late 1970s and 1980s, owing to improvements in screening, but has remained flat and even decreased in recent years.³⁸ Furthermore, when rates are broken down by age category, no increase is seen in younger age-groups, those with the heaviest cell phone usage. An analysis of long-term trends in glioma and meningioma incidence in four Scandinavian countries failed to detect any clear change between 1974 and 2003.³⁹ Similar results have been reported from other advanced countries.⁴⁰ A further analysis of glioma incidence trends in the United States between 1992 and 2008—a period during which cell phone use increased from close to 0 percent to almost 100 percent—concluded that the rates were not compatible even with the lowest risk estimate (odds ratio of 1.5) reported by Hardell after ten years of cell phone use.⁴¹ While not enough time has elapsed to gauge the full effects of long-term cell phone use, nevertheless these statistics showing no change in the incidence/mortality from brain cancers in the face of a dramatic increase in mobile phone use provide some reassurance. The fact that there is no suggestion of an uptick in rates of brain cancer over a twenty-year period is important, since within twenty years of the increase in cigarette smoking in the United States following World War I, there was already a noticeable increase in rates of lung cancer in men. The extraordinary expansion of cell phones has occurred at an even greater rate.

Second, hundreds of experimental studies have been carried out to understand the effects of RF waves on animals and cells. The most informative of these studies are long-term experiments in which one group of test animals is exposed to RF energy of defined characteristics, and their “tumor yield” is compared to that of control animals, who were not exposed. One early study looked positive; however, the majority of studies show no evidence that RF of the type emitted by mobile phones is carcinogenic in laboratory rodents. In its 2015 comprehensive review of the health effects of non-ionizing radiation, the European Commission’s Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR) concluded that these “well-performed,” negative studies “provide strong evidence for the absence of an effect.”⁴²

Third, it is crucial to understand the properties of the agent in question, that is, the radiofrequency energy used in mobile phone technology. The first-generation mobile phones used frequencies in the 800 and 900 megahertz range, whereas the newer, digital technology uses frequencies in the 1800–1990 megahertz range. These frequencies are too low to induce or promote cancer by damaging DNA directly. While other “epigenetic” effects are theoretically possible, there is no reproducible evidence of such effects.⁴³

One person who has attempted to put concerns about RF and electromagnetic fields in perspective is John Moulder, a professor of radiation oncology at the Medical College of Wisconsin, who has followed the research on these forms of energy for many years and produced assessments of the evidence. In 2005 he and his colleagues published a thoroughgoing, critical assessment of the evidence bearing on mobile phones and cancer. Their conclusions are worth quoting in full:

Biophysical considerations indicate that there is little theoretical basis for anticipating that RF energy would have significant biological effects at the power levels used by modern mobile phones and their base station antennas. The epidemiological evidence for a causal association between cancer and RF energy is weak and limited. Animal studies have provided no consistent evidence that exposure to RF energy at non-thermal intensities causes or promotes cancer. Extensive in vitro studies have found no consistent evidence of genotoxic potential, but in vitro studies assessing the epigenetic potential of RF energy are limited. Overall, a weight-of-evidence evaluation shows that the current evidence for a causal association between cancer and exposure to RF energy is weak and unconvincing.⁴⁴

Another figure who has attempted to understand the biophysical basis of possible health effects of RF and EMFs is Robert K. Adair, who is Sterling Professor Emeritus of physics at Yale University. Adair’s career, spent at Yale and Brookhaven National Laboratory, had focused on high-energy physics, but in the early 1990s he turned his attention to the biophysics of the interaction between electromagnetic fields and biology. In 2003 he published an article entitled “Biophysical Limits on Athermal Effects of RF and Microwave Radiation” in the journal Bioelectromagnetics, in which he pointed out that there were no reproducible effects on biology of exposure to RF or microwave fields below the level at which heating occurs.⁴⁵ After considering a “complete set of possible biological interactions” involving possible athermal effects of low-intensity RF and microwave electromagnetic fields on human physiology, he concluded that it was “quite unlikely” that any mechanism could transfer energy exceeding the normal thermal noise of the human body. Hence he concluded that it is “most unlikely that RF or microwave fields of an intensity less than 10 mW/cm² incident on humans, can affect physiology significantly.” But Adair went a step further. Referring to the eighteenth-century English mathematician Thomas Bayes, he argued that, when gauging the probability of an effect, one needs to take into account prior knowledge bearing on its likelihood. This is a much more sophisticated approach than simply examining each new finding in isolation, as if there were no previous relevant knowledge. Pointing to the body of published experimental findings purporting to show physiological effects of low-intensity fields, Adair commented that, given the theoretical implausibility of such effects, the results would have to be “especially definitive,” since “remarkable conclusions—which seem to violate well considered principles—require remarkably strong evidence.” In his judgment, the existing studies do not meet this standard. Note that Adair is not saying that it is impossible that RF could cause cancer. He is merely saying that, given everything that we know, it is extremely unlikely.

Although one cannot prove a noneffect, taken together, the results of the epidemiologic studies, the trends over time in incidence and mortality rates for brain tumors in many different countries, experimental evidence from studies in animals, and the biophysical considerations all converge in suggesting that cell phone RF is not carcinogenic.

Then in May 2011 the IARC Working Group on cell phones and brain cancer issued a report declaring that radiofrequency energy from cell phones was “possibly carcinogenic.”⁴⁹ Coming only a year after publication of the results of the INTERPHONE study, which showed no convincing evidence of a hazard, IARC’s announcement left the public as well as many scientists nonplussed.

IARC is a prestigious agency that since the early 1970s has been producing respected monographs evaluating the evidence for a wide range of chemical, physical, and hormonal agents in the carcinogenic process. Later in this book we will see the important role played by IARC in identifying human papillomavirus as the cause of cervical cancer. However, in recent years several of the agency’s assessments have been questioned on the grounds that IARC ignores real-world exposure and overemphasizes positive findings.⁵⁰

What accounts for how IARC could evaluate all the relevant scientific evidence and come to an ambiguous conclusion, which meant one thing to scientists and quite another to a public concerned about the possibility of dire effects from talking on a cell phone?

IARC evaluates all available evidence bearing on whether a given compound causes cancer, including animal experiments, laboratory (mechanistic) studies, and human (epidemiologic) studies. However, as the agency makes clear in its preamble, it focuses on “hazard”—that is, any potential indication of harm, no matter how tenuous or under what artificial conditions—rather than “risk”—that is, the likelihood that significant exposure in real-world situations increases cancer in humans.

In addition to the agency’s ignoring of the distinction between “hazard” and “risk,” a number of irregularities in the committee’s decision process regarding RF appear to have contributed to its baffling conclusion. The working group included Lennart Hardell, whose studies stood out from other epidemiologic studies in suggesting the existence of a risk. Two other epidemiologists on the committee resigned, one over an alleged “conflict of interest” and the other apparently in disgust at the proceedings.

Furthermore, although IARC routinely takes into account positive evidence of carcinogenicity from animal studies in its assessments, in the case of RF, panel members were instructed to restrict their attention to the epidemiology, thereby ignoring the high-quality, long-term animal experimental evidence that did not support a risk.⁵¹

The combined effect of these decisions was to give undue weight to the questionable findings of Hardell, who, in a clear conflict of interest, was voting on his own results.

All this raises the question: why did IARC have to carry out another exhaustive evaluation of the evidence? And once having carried out such a review, by any set of criteria for evaluating evidence, the conclusion should have been that, although we have not monitored the effects of cell phone use for long enough, the substantial evidence currently available provides no suggestion that cell phone use increases the risk of brain tumors. The Talmudic label “possible carcinogen” is unfortunate because it means one thing to scientists working for IARC and something quite different to the general public when trumpeted in the headlines.⁵²

In classifying RF as a “possible carcinogen,” IARC aligned itself with the precautionary principle, which sounds perfectly reasonable, except that it is often used conjure up the existence of a possible threat in the face of extensive and solid evidence suggesting the nonexistence of a threat.

We rely on health and regulatory agencies to provide impartial assessments of potential health risks. Unfortunately, the IARC cell phones report demonstrates that these institutions can be subject to the same political and professional pressures at work in society generally. The IARC review and the scientists who led it represent a third position intermediate between ICNIRP and the Hardell group. IARC straddles both camps and seeks to enhance its position by keeping the door open to a causal interpretation of very problematic epidemiologic data, but in its attempt to straddle both positions, it falls into incoherence, illogic, and ambiguity.

Rather than sticking to the science relevant to assessing a carcinogenic risk from cell phone use—which is IARC’s stated mission—and reaching a conclusion along the lines of ICNIRP and SCENIHR, bowing to public opinion, the agency chose to venture into the territory of “risk management.”⁵³ It succumbed to the temptation to convey a public health message rather than make a more boring statement about what we know. In doing so, it has confused the issue, since, if there is no convincing evidence of a risk, there is no compelling case for setting policy and alarming people unnecessarily. By its actions, IARC has only added to the confusion, since, as could have been expected, the public interpreted the “possible carcinogen” classification differently from the committee, that is, that cell phones can cause brain cancer. As the statistician Donald Berry is quoted as saying in response to the IARC announcement: “Anything is a possible carcinogen. This is not something I worry about and it will not in any way change how I use my cellphone,” he said—speaking from his cellphone.⁵⁴

There seems to be a paradox—or rather, a number of interlocking paradoxes—at the heart of epidemiologic research examining the health effects of cell phone use. In this day and age, such studies are going to be done, and new studies of the cell phone question are in progress.⁵⁵ Given the existence of the field of epidemiology, the widespread and high-stakes concern about the potential carcinogenicity of radiofrequency energy, and the fact that evidence from human studies is considered the most relevant and valuable information for health risk assessment, agencies like the Environmental Protection Agency, the Food and Drug Administration, and the International Agency for Research on Cancer see it as their responsibility to consider these studies. Once studies are done, it goes without saying that the media will report their results. But even more important, once the results are published—once they are out there—those people disposed to find evidence of an effect will seize on certain results, even when extreme caution is in order in interpreting them. On the other hand, when evaluated critically by scientists with no axe to grind, the severely limited ability of these studies to answer the ultimate question they are designed to answer comes clearly into focus. In the worst case, studies like INTERPHONE can appear to be what the literary theorist Stanley Fish called “self-consuming artifacts.” A rigorous and honest assessment of their limitations and potential biases can completely undercut the credibility of their results. And yet the one thing that is certain is that new studies will be undertaken. The problem with research in this area is not that it is worthless, but that all too frequently it is interpreted naively and uncritically and used for partisan rather than scientific purposes.