Cellphones, Brain Cancer and the Interphone Study
The topic of cellphones and brain cancer has come up a couple of times in our forum over the last few months. In one thread, Hang up your Cell Phone in TO, BillC discussed a CBC article about Toronto Public Health issuing a health advisory warning parents to reduce cellphone use by children, and in another thread, Is “Big Cell Phone” burying studies?, aDam talked about the Walrus article “Cellphone Games“, which questioned the cellphone industry’s commitment to researching the health effects of their product. The Globe and Mail revisited the cellphone-cancer issue on Friday 24 October with the article “Cancer and cellphones: The jury’s still out“. I found it odd that the Globe and Mail bothered to write about a non-finding, but when I looked closer, I noticed that all three of these discussions had something in common. It turns out that they are all based on the Interphone Study.
For William S. Burroughs fans, Interphone is not a communication device used by William Lee in Naked Lunch, and for Ottawa Skeptics regulars, it is not the next phone fad that Jon and Pat will chase when they realize they are in denial about their iPhones. Interphone is an international epidemiological case-control study that is “set-up to evaluate possible associations between RF exposure from mobile telephones and risk of [certain] tumours” of the head and neck, namely glioma, meningioma, acoustic neurinoma and salivary gland tumours. It is a study that I had never heard of before but one that the cellphone industry and many public health professionals have been anticipating for a number of years.
Because these cancers are rare, the study has had to cast a wide net for cases. The resulting international collaboration involves close to 50 researchers and 14,000 subjects at 16 study centres in 13 countries – Australia, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan, New Zealand, Norway, Sweden, and the United Kingdom. It is being coordinated by the World Health Organization’s (WHO’s) International Agency for Research on Cancer (IARC), and the principal research coordinator is Dr. Elizabeth Cardis, formerly of IARC and currently at the Center for Research in Environmental Epidemiology (CREAL) in Spain. Dr. Cardis has been affiliated with U Ottawa over the years, and in fact, U Ottawa’s Institute of Population Health (IPH) is a research participant in the study.
The multi-centric study has a common core protocol. Case subjects were selected from residents of the study regions aged 30 to 59 whose cancers were diagnosed during the period 2000 to 2004 and subsequently confirmed by histology or diagnostic imaging. Control subjects were selected randomly from the same source population and “individually- or frequency-matched to cases, with the number of controls varying according to tumour type. … [I]ndividual matching was conducted post hoc, with cases being assigned controls chosen to have been interviewed as close as possible in time to the case, from among those who fit the matching criteria.”
The study’s consolidated results were supposed to have been published in 2006 but have been held up while design problems with the study are resolved. Preliminary statistical analysis of the various data sets has been available since 2004 and can be found on the U Ottawa site, as well as in the form of an IARC summary, a CREAL presentation and a WHO presentation. (To review these results, you might want to brush up on the definitions of the terms odds ratio (OR), confidence interval (CI) and ipsilateral.)
It is perhaps understandable that the various national study teams have been unable to sit on their individual work and forego publishing until the collective study is produced. Many of the national studies have made their way into journal articles as preliminary results, as summarized in Table 1 of IARC’s Interphone Study Results Update 7 February 2008. These partial results do not show a risk of cancer for cellphone use under a period of 10 years, but a few studies hint at a correlation for over that time. These few preliminary positive findings are the genesis of the Toronto Public Health advisory.
Study Design Problems
The problem with these individual studies is that they suffer from small sample sizes and so are not definitive on their own. The Interphone study was conceived as an international collaboration in order to deal with this small sample size problem. However, merely combining the individual results into a consolidated study will not provide the definitive answers that the research team aspires to deliver. A number of study design problems are being realized, many of which are discussed in The Economist article “The Interphone study: Mobile madness” and in the researchers’ own article on the study’s design and methodology.
These problems should not come as a surprise. Attempting to study the effects of a common, pervasive and changing technology in a large population over a long period of time is no easy task, and the Interphone study team itself has not only acknowledged these study limitations but is actively conducting supplementary studies and analyses to correct for them.
Participation Bias – The study team admits having had difficulty recruiting control participants, especially in the younger age groups that were needed for this study. The low participation rate of 53 percent that was experienced runs the risk of imposing participation biases in what should be a random population group. In fact, The Economist article reports that, of those who declined to be control subjects, 34 percent were regular cellphone users.
“That meant those in the control group were more likely than average to be regular users, and therefore were not representative of the population at large.”
Amongst potential case subjects, people who were severely diseased tended not to participate in the study for understandable reasons, but this relation of refusal rate and disease severity causes another type of selection bias.
Recall Bias – Cellphone use in the study was estimated based on retrospective self-reporting, a methodology for which data fidelity is somewhat dubious, especially for a period of recall of 10 or more years. In fact, the researchers did a side study that determined that subjects tended to underestimate their number of calls per month and overestimate the duration of the calls, both significantly. They concluded that:
“Volunteer subjects recalled their recent phone use with moderate systematic error and substantial random error. This large random error can be expected to reduce the power of the Interphone study to detect an increase in risk of brain, acoustic nerve, and parotid gland tumours with increasing mobile phone use, if one exists.”
In addition, because case subjects likely tended to contemplate the cause and progression of their disease, a rumination bias of cellphone head side use and frequency probably resulted. As well, because brain tumours can cause cognitive impairment, case subjects may have made errors in recalling their cellphone use.
Technological Differences – Over the last 10+ years, cellphones have experienced significant technological changes. They once were all analog and now are digital except when ranges require reverting to analog mode. Different modulation schemes have prevailed over time. Even the use of headsets has become common when it was once rare. All these differences have an effect on the amount of RF power to which users’ heads and necks have been exposed. In fact, all things being equal, it is not necessarily safe to assume that two different cellphone makes have a similar radiated RF pattern or strength. As a result, the researchers are doing side studies into RF exposure gradients and the precise location of subject tumours.
Participant-based Risks – Some possible participant-based confounders, such as region, age and sex, were taken into account in the design of the study. However, other potential contributions to these diseases, such as the medical and family history of the subjects as well as their occupational and environmental exposure to other types of radiating devices and to certain chemicals, are the subject of other side studies.
The Interphone study, which is supposed to be the study that is going to provide a definitive answer to whether cellphones cause brain cancer, is now the study that will not deliver. Recent articles, such as the Globe and Mail story, focus on this non-delivery and the apparent split amongst the researchers. Part of the study team expects that a long-term causal link will bear out, part believes that the few long-term preliminary positive results are study design artefacts, and the rest are not talking.
The delay in publication of the final report can be seen to be due to the complicated nature of the study and not, as the Walrus article suggests, the result of an industry-based conspiracy to hide the results. If a causal link had been definitively uncovered or dismissed, the results would have been undoubtedly published by now. Unfortunately, the results so far are ambiguous and contaminated by biases.
The researchers seem intent on correcting for all the admitted biases of the study. However, doing this type of post hoc correction to the degree that is needed undermines the validity of the study. In reality, what the team can claim to have accomplished with Interphone is to have completed a thorough preliminary study that has identified methodological limitations and confounders that need to be addressed in subsequent studies. Preliminary studies are an invaluable part of the scientific process.
Unfortunately, it will be almost impossible for those responsible for a study of the scope and duration of Interphone to put down their pens and admit that they ended up with an inconclusive preliminary study. Instead, they will likely publish highly manipulated results that will ambiguously show or not show a weak causal link for brain cancer with regular cellphone use over a 10-year period but with a caveat that more studies are definitely required. Then most people will read into the report what they want to hear, ignoring the caveat in the process.