The news yesterday and today has been full of reports that the World Health Organization – or more specifically the International Agency for Research on Cancer (IARC) is reporting that cellphones could cause cancer. The reaction is based on a press release; the full IARC evaluation has not yet been released. Based on what is in the press release, should we be worried?
The physics of how electromagnetic fields from cellphones interact (or don’t interact) with the human body is not my area of expertise, so I’ll leave the discussions about the plausibility based on known or suspected mechanisms of carcinogenicity – the energy from cellphones doesn’t break chemical bonds in DNA, but that doesn’t mean there is absolutely no possible mechanism of carcinogenicity. Rather, I’ll look at what the IARC press release is really telling us, which is not necessarily the same thing as how the media are spinning it.
First off, I consider IARC to be pretty much the gold standard when it comes to evaluating the carcinogenicity of substances. They use international teams of experts to review all of the relevant data, including human epidemiological studies, animal studies, tests done on bacteria and cell cultures, and studies on potential mechanisms of carcinogenicity. Based on the weight of evidence, they classify a substance as one of the following:
Group 1 – carcinogenic to humans. These are substances for which there is enough evidence of cancer in humans that the evaluators are confident the effect is not due to random chance, biases in study design, or potential confounding factors.
Group 2A – probably carcinogenic to humans. These are substances where there is convincing evidence of carcinogenicity in animals, generally more than one species, and at least some evidence in humans – either studies showing increased cancer in humans but not enough data for the evaluators to be confident in them, or evidence that the cancer in animal studies is caused by a mechanism that is also relevant for humans.
Group 2B – possibly carcinogenic to humans. These are substances are generally those for which there is some evidence of carcinogenicity in animals, but usually not much evidence (if any) in humans, or strong evidence of a potential mechanism by which it could cause cancer without enough data in human and animal studies.
Group 3 – not classifiable as to its carcinogenicity in humans. These are substances for which there isn’t enough data to draw any conclusions, or rarely where there is evidence of carcinogenicity in animals but the mechanism of carcinogenicity in animals does not operate in humans.
Group 4 – probably not carcinogenic to humans. These are substances for which there is reasonably solid evidence showing that the substance doesn’t cause cancer.
Of the substances that have been evaluated by IARC, more than half are classified as Group 3, indicating a lack of good data; the next most common classification is Group 2B. Only 1 substance, caprolactum, is currently classified in Group 4.
The IARC press release indicates that they have classified radiofrequency electromagnetic fields such as those from cellphones as Group 2B – possibly carcinogenic to humans. In this case it was based on “limited” evidence for glioma and acoustic neuroma in humans, which is a bit different from the typical substances given a 2B classification which are more often based on animal studies. “Limited” evidence, for the purposes of an IARC assessment, means at least one study showed a positive association between exposure and cancer, and the evaluators believed that an effect was plausible, but they couldn’t rule out the effect being caused by random chance, bias in the study design, or confounding factors.
Until the full evaluation report is released it is difficult to make to many conclusions about whether or not this classification is supported. In particular the press release suggests that the evaluators had access to a few studies that haven’t been published yet. However, a Group 2B classification is generally not considered to be a major cause for concern – most regulatory agencies would not treat such a substance as a carcinogen. Rather it is an indication that further study is warranted to confirm or refute the limited evidence.
In addition to looking at whether or not a substance can theoretically cause cancer, it is also important to consider what the actual incidence rate is. I looked at Canadian statistics; the Public Health Agency of Canada has a website where you can generate charts of cancer incidence. Unfortunately it doesn’t let you look at glioma individually, but you can look at all brain cancers:
The scale isn’t that legible, but it shows the age-adjusted rate of all brain cancers has been fairly stable at around 6 to 6.5 cases per 100,000 people from 1992 to 2007. You can get similar data for the US from the SEER web site; it shows very similar rates and over the last several years rates are stable or if anything decreasing. I’ve seen other data suggesting the rate of glioma could be around 3 per 100,000, so if there was a significant trend in glioma rates there should be some indication of a trend in total brain cancer rates. Now this doesn’t mean that there couldn’t be an effect on a small population subgroup (e.g. heavy users of cellphones) that doesn’t show up in the population-level statistics. Indeed, one of the studies mentioned in the IARC release is the Interphone study. One of the papers published from this study showed a marginal increase in glioma in the heaviest users of cellphones, but the results were of borderline significance and not strong enough to be convincing (consistent with IARC’s classification of the evidence as “limited”). The authors noted that some of the users reported what they considered to be implausible levels of cell phone use, suggesting some level of recall bias. The cancers involved are also relatively rare.
So in the end, what do we know? We know there is at least one study that showed suggestive evidence of carcinogenicity, but the data were not very conclusive, and the effect being due to random chance, study bias, or confounding factors cannot be ruled out. The cancers involved are also relatively rare, indicating the actual incidence would not be very high even if this effect is real, and the effect, dubious as it is, was only identified in the very heaviest users of cellphones. This means that there is still no reason for most people to be worried. If you are an extremely heavy user of cellphones, then maybe you want to take a precautionary approach and take steps to reduce your exposure – the odds of getting brain cancer aren’t high and the evidence that cellphone use is associated is not convincing, but given that malignant brain cancer is pretty serious and unpleasant taking reasonable steps like reducing cellphone use to more normal levels or using a headset is probably reasonable. I’d also be careful about young children making excessive use of cellphones – again there’s no convincing evidence of an effect, but children are often more susceptible to this sort of effect, since their brains are still developing, and since the extensive use of cellphones by children is a relatively recent phenomenon if there are any effects they might not be apparent yet. Besides, does a young child really need to be using a cellphone all the time? I don’t think the odds of an adverse effect are very high, between the dubious nature of the link and the rarity of the particular cancers involved, but given the severity of the potential effect I personally don’t see a problem with taking at least a few basic precautions, particularly if they don’t significantly offset the benefits of cellphones. For most people, though, I don’t see any need to panic or change habits at this time. If the actual IARC monograph or the summary to be published in the Lancet change my opinion, I’ll address that when they are published.