Friday October 21 2011
Users and non-users had the same rate of brain cancer
“A large study of mobile phone users has found no evidence that longer-term users are at an increased risk of developing brain tumours,” The Daily Telegraph has today reported.
The study in question looked at national records and mobile phone subscription registries for all adults aged 30 and over in Denmark between 1987 and 2007. Researchers used the data to compare the risks of getting brain cancer among those who were mobile phone subscribers and those who were not. It found no increased risk of brain cancer among either male or female mobile phone users, even among those who had used them for the longest period (13 years or over).
The study had some major strengths, including its use of a large and unselected population and not having to rely on people estimating their past mobile use. Its main limitation though, is that it used the fact of subscription to a mobile phone as a measure of mobile phone use, rather than the amount of time a person spent on a mobile phone. This could misclassify people, particularly those who used a work mobile.
The researchers also note that brain cancers are rare, meaning that the study cannot completely rule out a small-to-moderate increase in risk for heavy users or risks with use over more than 15 years.
Although on its own this study cannot be seen as proof, its results offer some reassurance that mobile phone use over 10-15 years appears not to be linked to an increased risk of brain cancer in adults. The key messages to remember are that brain tumours are rare, in both mobile phone users and non-users, and that studies have yet to detect any large effect on risk.
Where did the story come from?
The study was carried out by researchers from the Danish Cancer Society and the International Agency for Research on Cancer (IARC). It was funded by Danish Strategic Research Council, the Swiss National Science Foundation and the Danish Graduate School in Public Health Science. The study was published in the peer-reviewed British Medical Journal.
The story was covered by several news sources, with BBC News giving a good summary of the study and providing some context about the World Health Organization (WHO) and Department of Health’s positions on mobile phones. Several newspapers also pointed out the study’s strengths as well as its limitations, which the researchers themselves acknowledge.
What kind of research was this?
This was a nationwide cohort study that looked at whether mobile phone use increased the risk of cancer across the Danish population.
As it would not be feasible to conduct a randomised controlled trial on long-term mobile use, a cohort study is the best way to assess this question. Most other studies assessing this question used a case-control design, where people who developed cancers were compared with a healthy control group to see if their mobile usage in the past differed. Selecting an appropriate control group for such studies can be difficult, and the current study removed this difficulty by using a nation’s entire population as its potential study group.
Many previous studies have also relied on self-reported mobile use. This may not be reliable and case-control studies may be influenced by a person’s perception of whether their mobile phone use could have contributed to their cancer.
As with all cohort studies, mobile users and non-users may differ in other characteristics that may influence results, and researchers need to take these into account in their analyses where possible.
What did the research involve?
The researchers identified all adults aged 30 and over in Denmark who were born after 1925 and still alive in 1990 and whether they were mobile phone subscribers before 1995. They then identified all of the people who developed any cancers up to 2007, and analysed whether they were more common in mobile phone subscribers than non-subscribers.
The researchers only included people for whom they could obtain information about their socioeconomic status (education and disposable income). They excluded offspring of immigrants to the country as information on their education abroad was not systematically recorded. The researchers obtained mobile phone subscription records for 1982 to 1995, and excluded corporate subscriptions. They were only interested in subscriptions from 1987, when handheld mobiles first became available in Denmark.
The researchers also excluded people who had cancer before the start of the study. They also did not include the first year of a person’s subscription in the analysis in case these people already had a tumour when they first started using their mobiles. This left 358,403 mobile users for analysis, and between them they had a total of 3.8 million years of mobile exposure.
The researchers used the Danish Cancer Register to identify any cases of cancer between 1990 and 2007. They were mainly interested in cancers of the brain and spine (the central nervous system, or CNS), including benign tumours. They also looked at all cancers as a whole and cancers related to smoking.
In their analyses the researchers looked at cancers per year among mobile phone subscribers with different periods of mobile use and compared these rates with the cancer rates seen among people who were not mobile phone subscribers or who had less than a year’s subscription. The figures they calculated are called ‘incidence rate ratios' (IRRs), a measure that expresses how the rates of cancer incidence among two groups relate to each other. These figures were calculated by dividing the rate of cancer per person-year of follow-up in the mobile subscribers by the rate in the non-subscribers. An incidence rate ratio of 1 would indicate that the rates of cancer were identical in both groups. The analyses took into account other factors that could potentially affect their results, including calendar year in which the cancer was diagnosed, and markers of socioeconomic status including education and disposable income.
The analyses were carried out separately for men and women.
What were the basic results?
Between 1990 and 2007, the researchers identified 122,302 cases of cancer in men, and 5,111 of these cases were cancers of the CNS. They identified 133,713 cases of cancer in women in this period, and 5,618 of these cases were cancers of the CNS.
The researchers then calculated the incidence rate ratios (IRR) of CNS cancers for subscribers and non-subscribers, a measure expressing how the risk in each group compared. An IRR of one indicates that the risk in the two groups is equal. They found there was no difference in the overall risk of CNS cancers between mobile subscribers and non-subscribers, either in men or women:
- incidence rate ratio in men 1.02 (95% confidence interval [CI] 0.94 to 1.10)
- incidence rate ratio in women 1.02 95% CI 0.86 to 1.22).
This was also the case if researchers looked at people with different lengths of mobile subscription: 1-4 years, 5-9 years, 10 years or more, 10-12 years, or 13 years or more.
When looking at individual types of CNS cancer, mobile users and non-users showed no significant difference in the rates of glioma, meningioma, or other and unspecified kinds of CNS cancer. There was also no evidence that risk increased with increasing length of mobile phone use, or of an increased risk of gliomas in the areas of the brain nearer to where the phone would be held.
How did the researchers interpret the results?
The researchers concluded that in their large nationwide cohort study there was no association between tumours of the CNS or brain and mobile phone use.
This large, nationwide Danish study has found no link between mobile phone use in adults and risk of brain cancers. Its strengths included its size, which allowed a reasonable number of brain cancers (a rare form of cancer) to be identified for analysis. It also included the majority of the eligible Danish population, with only a low proportion lost to follow-up (2.2%), as it used population registries.
The study also provided information of longer periods of mobile phone use than many previous studies, and does not rely on people to report their own mobile use in the past, which may not be reliable, particularly in case-control studies. There are some points to note:
- The study used personal mobile phone subscription as a measure of mobile phone use. People who had mobile phone subscriptions may have had differing levels of use, and some of those without a subscription may have used someone else’s phone or a work phone only. Thus, misclassification could have affected the results.
- The researchers note that as misclassification errors could occur in both directions (users classified as non-users and vice versa). This should not bias results in one direction or the other, but would instead make any effects appear smaller. However, they also report that the analyses looking at the longest period of exposure found no increase in risk and say this supports their conclusions as these particular analyses should be the least affected by low-level misclassification of exposures.
- The researchers only had mobile phone data up to 1995, and use may have changed after this point. However, analyses that only looked at cancer diagnoses up to the end of 1996 had similar results to the overall analyses, suggesting that they were robust.
- The researchers did take into account some factors (other than phone use) that could influence their results, but that does not completely remove the possibility that the results could have been affected by factors other than phone use.
This study offers some reassurance that mobile phone use over 10-15 years appears not to be linked to an increased risk of brain cancer in adults, but for a number of reasons the study cannot in itself be regarded as ‘proof’.
As brain cancers are so rare, the researchers note that even their large study cannot rule out a small-to-moderate increase in risk for heavy users. Similar studies from other countries would help to increase the amount of brain cancer cases that can be analysed to determine whether this is a possibility. The authors note that even longer-term follow up studies are also needed.