Oh no! My cell phone’s going to kill me! (The revenge)

ResearchBlogging.orgHere we go again.

I’ve written a few times before about the controversy over whether cell phones (a.k.a. mobile phones in most of the rest of the world) cause brain cancer, concluding on more than one occasion that the evidence does not support a link. For example, there has not been a large increase in brain cancer or other cancers claimed to be due to cell phone radiation in the 15 to 20 years since the use of cell phones took off back in the 1990s, nor has any study shown a convincing correlation between cell phone use and brain cancer.

Of course, one would not expect a priori, based on what is known about basic science, that cell phone radiation would cause cancer. After all, the development of cancer in general ultimately requires mutations in critical genes regulating cell growth and development. For an outside treatment to cause such mutations, as far as we know, requires the ability to cause DNA damage through the breaking of chemical bonds. Ionizing radiation can do this, as can certain cehmicals and chemotherapeutic agents. Indeed, that’s how these agents work against cancer because cancer cells tend to be more sensitive to DNA damaging agents than normal cells due to defective DNA repair mechanisms. Thus, it is highly implausible based on basic science that cell phone radiation could cause cancer. It’s not homeopathy level-implausible, but it’s pretty implausible. Nor is it impossible, as has been claimed, because there may be biological mechanisms behind cancer that we do not yet understand, and it’s almost always physicists with little knowledge of epigenetics and other mechanisms of cancer development who make such dogmatic claims. Still, such physicists are not too far off; if cell phones could cause cancer, it would have to be through a previously unknown physiological or genetic mechanism. Absent compelling evidence of a link between cell phones and cancer, then, it is not unreasonable to rely on the basic science and consider the possibility of such a link to be remote.

Still, anything having to do with “radiation” causes fear, because most people don’t understand the different wavelengths and varieties of radiation. There’s also a cottage industry that’s sprung up to take advantage of people’s lack of knowledge about basic physics and chemistry by selling useless “cell phone radiation shields.” Much like research into various highly implausible forms of “alternative medicine,” though, research into a possible link between cell phone use and brain cancer continues unaffected by considerations of prior plausibility. So does the hysteria, sometimes even infecting prominent, high-ranking cancer researchers who really, really should know better.

The latest volley in this fray was released yesterday in the form of a new report of the results of an ongoing study examining whether there is a correlation between cell phone use and cancer. For once, news reports seem to be getting it right in that the results are “inconclusive.” Of course, I would have been shocked if the results had been conclusive. Based on this study, there are two things I can say with confidence. First, it will settle nothing, and, second, it will be attacked by those who, despite all the evidence against it and the incredible implausibility of a link between cell phones and cancer, deeply believe that there is just such a link. No doubt such attacks will include a mention that part of the funding for the study came from the Mobile Manufacturers’ Forum (MMF) and the GSM Association, both industry groups. True, the funding from these organizations went first through a “firewall mechanism,” but that won’t stop the criticisms.

The study was the INTERPHONE study, which involves 13 countries looking for any sort of link between cell phones and two types of brain cancer, glioblastoma and meningioma. Partial results from the study have already been published, but the current study1 represents the first time that results from all 13 countries have been reported. The study itself is a case control study including 2,708 glioma patients and 2,409 meningioma patients, along with matched controls. I’ll basically cut to the chase here (unusual for me) and reveal the outcome: There was no compelling evidence of an association between cell phone use and either of these cancers. Surprise, surprise.

There were, however, a couple of rather interesting findings in subgroups. Now I’ll say one thing about subgroup analysis. Basically, subgroup analysis is something that researchers do when their overall results are negative to try to salvage a “positive” result out of a study. For instance, if a study shows that, for example, factor X is not associated with an increased risk of developing breast cancer, the next thing to do is to see if X is associated with breast cancer in women under 40, in smokers, in drinkers, or in any subgroup the investigators can think of. The problem is, when you slice and dice the subject group into ever smaller groups looking for subgroup effects, you will almost always find one. Whether the result is spurious or not is impossible to tell without a further study, which is why positive results from subgroup analysis should almost always be considered as hypothesis generating rather than hypothesis confirming. The one major exception is when these subgroup analyses were built into the research plan and explicitly included in the statistical power analysis performed before the study begins.

So what were these seemingly “positive” results? The first was actually counter to the hypothesis of the study. The odds ratio related to ever having been a regular cell phone user was seen for glioma (OR 0.81; 95% confidence interval (CI) 0.70-0.94) and meningioma (OR 0.79; 95% CI 0.68-0.91). This is a rather puzzling result because, if it’s incredibly implausible that cell phone radiation causes cancer, it’s even more implausible that it protects against cancer somehow. The second was that in the highest 10% of recalled cumulative call time, the OR was 1.40 (95% CI 1.03-1.89) for glioma, and 1.15 (95% CI 0.81-1.62) for meningioma; but there are implausible values of reported use in this group. In othe words, if these results were to be believed, the heaviest users of cell phones had a barely statistically significant increase in glioma incidence (a 40$ increased risk with wide error bars) and a non-statistically significant increase in meningioma. In other words, none of these subgroup analyses are convincing, as one of the epidemiologists involved in the study explains:

On completing their analysis, the researchers found that being a regular user of a mobile phone seemed to reduce the risk of glioma or meningioma by around 20%. But Anthony Swerdlow, an epidemiologist at the Institute of Cancer Research in London who was involved in the UK arm of the study, says that this result is highly likely to be down to problems that were inherent in the study design.

“We have evidence that the people who refused to be controls are people who didn’t use phones,” says Swerdlow. This meant that the control group, consisting of people without cancer, was rather skewed, appearing to have more mobile-phone use than would be found in a representative sample from the general population. “The controls were over-represented with phone users,” he adds.

Equally, some of those individuals in the top 10% of reported phone usage gave what Swerdlow calls “incredibly implausible values”, such as an average of 12 hours of mobile use per day, every day.

Studies to validate the data-collection methods used in INTERPHONE found that asking participants about the number of calls they had made provided more accurate information than asking about how much time participants spent on the phone. When researchers analysed the number of calls made, the top 10% of participants showed no increased risk of cancer.

Such are the problems with using recall of cell phone usage to measure extent of exposure to radio waves. I would also point out that the results of this study appear to be consistent with random noise, where different measures appear positive or negative at random due to random chance alone.

Obviously, this current study has many limitations. There is no data about exposures longer than 15 years, and, as has been pointed out before, using recall to estimate cell phone use is prone to biases and other problems. Even so, it is yet another study adding to the accumulation of evidence that is failing to find an association between cell phone use and cancer. Given the extreme biological implausibility of the hypothesis that cell phones cause cancer, this result is not at all surprising. In fact, I’ll go beyond that and predict that future updates of the INTERPHONE study will fail to find any evidence of a significant correlation between cell phone use and cancer. I’ll also predict that those who believe in such a link will dismiss the results.

Now where can I pick up my “big mobile” check to go along with my big pharma check?


1 . (2010). Brain tumour risk in relation to mobile telephone use: results of the INTERPHONE international case-control study International Journal of Epidemiology DOI: 10.1093/ije/dyq079