Dialling up an old panic

How should we respond to yet another piece of obscure research suggesting health problems with mobile phones, or in this case mobile phone masts?

On 17 May 2005, UK newspapers reported that using a digital mobile phone in rural areas may pose a greater risk of developing brain tumours than it does in urban settings. These claims were based on a Swedish study by Dr Lennart Hardell, which looked at 1,400 adults aged 20 to 80 who had been diagnosed with a malignant or benign brain tumour and compared them to healthy adults living in the same area. The Times (London) reported that the risk could be as much as eight times greater, even if it was ‘based on so few cases the result couldn’t be relied upon’.

So should you start thinking about dumping your mobile before it gives you cancer (especially if you’re not a ‘townie’)? The simple answer is ‘no’ – and this is not just in the sense that ‘more research is needed’. Behind the headlines, the sceptic would say that what’s really going on is that the enterprising Dr Hardell has found a new ‘sexy’ angle (an urban/rural split) through which to recycle old research that barely stood up in the first place.

The British media seems uncritically to report anything that suggests harm from mobiles. Some reports didn’t even include the critical voice of a mobile industry spokesman indicating problems with the research. Nor did experts in the field from the Health Protection Agency take the opportunity to explain or contextualise the research, but just suggested that interpretive caution was appropriate and reiterated the precautionary line that it is advisable to limit usage, particularly for children.

In the judgement of other scientists in this area, Hardell’s work has significant flaws. His latest paper is a statistical reanalysis of data collected for other purposes by this group; a reanalysis with many comparisons and small numbers in some categories that even the authors say have to be interpreted with caution. But more than caution is called for in this case; the very validity of such an endeavour is open to question.

The most important methodological point was expressed succinctly to me by American expert Professor Kenneth R Foster from the University of Pennsylvania, who said that Hardell’s latest piece might set a new international record for ‘slice and dice’ – an expression used to refer to the practice of extracting ever-more articles from the same limited data set. In this case it’s the fifth analysis of material recycled elsewhere. The problem with this is that research results only make any sense in relation to the question they were originally addressing.

This study’s original hypothesis was whether overall brain tumours are associated with radiofrequency exposure from mobile phones in use in certain areas of Sweden (1). That is the only question this case-control study was designed to test, rather than anything to do with any ideas researchers may subsequently have developed about rural/urban differences. Such experiments are necessarily targeted at addressing very particular issues, such as looking at the impact upon a particular type of tumour when subjected to a particular radiofrequency signal. To then try to twist any such results in order that they shed light on an entirely different type of problem isn’t meaningfully useful. Hardell has generated five papers from the same data over the past few years – all are post-hoc (thoughts after the event), and therefore hypothesis-generating only. The result is hypotheses that, at best, might act as useful starting points for future research, rather than definitive conclusions.

This is a general problem with research into the effects of mobile phones. Epidemiological research into the effects of radiofrequency fields generated by mobile phones is largely speculative enquiry, which has little chance of coming to clear conclusions – only a wealth of hypotheses and uncertainties that are unlikely to be resolved.

Even in its own terms, Hardell’s latest use of his data is very limited. The ‘increased risk’ for rural residents was small, and given that the association is so weak and close to non-statistical significance, the result is very unreliable. This points to a significant methodological problem with this approach to research. If you make enough comparisons (‘slice and dice’ enough), and neglect to take multiple comparisons into account in the statistical analysis, one comparison out of 20 will be ‘statistically significant’. There is also the fact that such a presentation conveniently avoids any consideration of real scale – a doubling or quadrupling of a tiny risk or a tiny number of cases tells you little about statistical data. In this case, the risk is barely statistically significant, so any eight-fold increase is particularly misleading.

Aside from the particulars of this latest study, alarm bells should have been set ringing from the fact that Hardell is so willing to adopt a political stance on this highly technical issue. In a report on BBC News, for example, he declared: ‘My advice would be to operate the precautionary principle. Use hands-free kits when possible and children should only be using mobiles when necessary.’ If the conclusions are based upon too few studies to be in any way reliable, how can they possibly warrant the dramatic step of limiting our mobile phone usage?

While such a stance has become normalised in the UK because of the precautionary stance established by the Stewart Committee on mobile phones, it remains an unusual one – particularly for a scientist working in the field. Most scientists would say that even if there were some uncertainties about the effects of mobile phones, this in itself is only grounds for further research, not for adopting a precautionary approach to usage. We are uncertain about many things in the natural world, but only where there is clear evidence of potential widespread harm is there reason to adopt a restrictive policy.

Hardell is one of only a handful of individuals with scientific credibility who advocate the idea that harm results from mobile phone radiation. It is for this reason that he has appeared as an expert witness for American litigants who claim that mobiles have caused cancer. These are significant cases not just because of their continued failure, but because it was the first of these cases – the case of David Reynard, back in the early 1990s – that launched the whole issue of harm from mobile phones. Hardell appeared as an expert witness in a case heard in Maryland in 2002, for example, where Christopher Newman and his wife claimed that use of a Motorola mobile had caused his brain cancer.

Hardwell’s original data itself is flawed when it attempts to stand as evidence of a causative association between cancer and mobile usage. The minutes of the Maryland court’s ruling describe how in his earlier 1999 paper Hardell ‘found no increased risk of brain tumour associated with mobile phones’. Recalculating in 2001, however, he now suggested an increased risk – but this was based on only 13 cases. Overall no statistically significant increased risk was shown for development of any malignant brain tumor.

However, Hardell supported his theory of causation and thereby stood as the main witness for the litigants, with his purported findings of an increased association between development of malignant brain tumors and ‘ipsilateral’ phone use: that is, persons who developed tumors were more likely to report that their phone use had been on the same side of the head as the tumors. The validity and relevance of this finding is subject to serious criticism by the Maryland court on a number of grounds, including the problem of recall bias by those who had developed brain cancer about their mobile usage, and an overdue emphasis on the findings for isolated subgroups of tumours.

The court added that numerous large-scale studies have rejected any connection between mobiles and brain cancer (2). More recently, the Interphone studies in Denmark and Sweden found no increased brain tumour risk (3). The court complained that Hardell’s assertions in the American court were at odds with the scientific consensus and the rejection of any causal relationship between cancer and mobile usage by more than 10 scientific panels and government bodies. There is also the fact that there has been no significant change in the numbers of tumours since the introduction of the mobile phone. No wonder the case was thrown out. But Hardell’s research continues to be reported in the British media.

It’s remarkable that such a reanalysis of originally weak studies should have been reported in the first place. This might be partly simply down to a lack of news on 17 May, but it also indicates how the UK media has a critical blind spot when it comes to negative mobile phone research.

Papers such as the Daily Express have adopted an explicitly campaigning role around the issue. The British print media in particular chose to pursue this issue with campaigning zeal, ever since the ‘mobiles fry the brain’ story in The Sunday Times back in 1996. Some journalists appear to have convinced themselves that this might be the next big thalidomide-type story of corporate cover up – and have printed virtually any story that suggested ill-effects associated with mobile usage, be it giving kids headaches or cancer, lowering the sex drive or, worryingly, disorientating homing pigeons. And such is the media’s influence upon a highly defensive government, that New Labour commissioned a special scientific committee to consider the problem in response, thereby confirming it as a bona fide concern.

The most interesting aspect of the study arises only indirectly. Hardell suggests that the reason for his urban/rural discrepancy is the fact that masts in rural settings naturally tend to be further away from users (given the large number of masts needed in urban areas to handle the larger amount of traffic). But why might being further away from an allegedly dangerous source suggest greater harm? Like a lot of science it is counter-intuitive, and clearly illustrates the limitations of relying upon common sense when it comes to understanding the natural world. The explanation is that the further away from a mast, the ‘harder’ the mobile has to work, meaning the more powerful the radiofrequency radiation has to be. This is entirely contrary to the assumption of, for example, the parents anxious about masts near their children’s schools.

But then, as experts in the field have explained, the ‘harm’ from a mobile mast is comparable to getting a suntan from the moon. Something that is already too weak to harm human tissue at close range (and hence have any kind of health effect in conventional terms) disperses very quickly near the mast and would appear about as dangerous as light from our moon.

Of course, I don’t want to suggest that the British media should entirely ignore the issue. But at least some of the energies would be better spent on some more investigative journalism; digging deeper and asking more critical questions about the individuals promoting these anxieties, and why institutions and authorities in some countries such as Britain then feel duty bound to respond. Or even an article on why the British media developed such an obsession with mobile phone risks in the first place. Now that would make an interesting story.

Dr Adam Burgess is lecturer in sociology at the University of Kent and author of Cellular Phones, Public Fears and a Culture of Precaution (New York: Cambridge University Press, 2004)

(1) European Journal of Cancer Prevention 2002, 11, pages 377-386

(2) See Peter D. Inskip, et al., Cellular-Telephone Use and Brain Tumors, NEW ENG. J.MED. 344:79-86 (2001) (DX 89) (“Our results do not substantiate the concern that some brain tumors diagnosed in the United States during the mid-1990s were caused by the use of handheld cellular telephones. There was little or no indication of an increased risk of glioma, meningioma, or acoustic neuroma associated with any use, cumulative use, or the laterality of use of these telephones.”); Christoffer Johansen, et al., Cellular Telephones and Cancer – a Nationwide Cohort Study in Denmark, J. NAT’L CANCER INST. 93:203-207 (2001) (“The results of this investigation, the first nationwide cancer incidence study of cellular phone users, do not support the hypothesis of an association between use of these telephones and tumors of the brain or salivary gland, leukemia, or other cancers.”); Joshua E. Muscat, et al., Handheld Cellular Telephone Use and Risk of Brain Cancer, JAMA 284:3001-3007 (2000) (DX 88) (“The use of handheld cellular telephones was unrelated to the risk of brain cancer in the current study.”); Kenneth J. Rothman, et al., Overall Mortality of Cellular Telephone Customers, EPIDEMIOLOGY, 7:303-305 (1996)

(3) [http://aje..oupjournals.org/cgi/content/abstract/161/6/526] and [http://www.neurology.org/cgi/content/abstract/64/7/1189].