A new study has found no link between exposure to mobile phone masts while in the womb and the risk of developing early childhood cancers, newspapers have reported.
During the study, scientists used complex transmitter data to estimate the signal exposure levels that almost 1,400 children with childhood cancers had experienced before birth, comparing them to the exposure levels of approximately 5,600 children not affected by cancer. The researchers specifically looked at three different measures of exposure – distance to nearest base station, total power output from nearby base stations, and estimated power density from nearby base stations. None of these measures suggested any association with the likelihood of cancer.
The rarity of childhood cancers and the practical constraints of individually measuring women’s exposure mean that the study authors had to make various assumptions about exposure, which may have affected the results seen. However, the study seems robustly planned and executed overall. Another limitation is that the study only looked at exposure during pregnancy and early childhood cancers, meaning it cannot tell us about exposure during childhood, or about longer-term outcomes.
Where did the story come from?
The study was carried out by researchers from the School of Public Health at Imperial College London, and funded by the UK Mobile Telecommunications Health Research (MTHR) Programme, an independent body set up to fund research into the possible health effects of mobile telecommunications. The MTHR is jointly funded by the UK Department of Health and the mobile telecommunications industry. The study was published in the peer-reviewed British Medical Journal.
This research was well reported by The Guardian and The Independent.
What kind of research was this?
This was a case-control study looking at whether there was a relationship between mothers’ exposure to mobile phone masts in pregnancy and early childhood cancers in their offspring.
This study design takes a group of individuals with the condition of interest (a case group of children with early childhood cancers) and compares their past exposures with a group of individuals who do not have the condition of interest (a control group). This study design is often used when the condition of interest is rare – as is the case with early childhood cancers – as a cohort study would have to be very large to detect enough individuals with the condition to allow a meaningful analysis.
One of the limitations to this study design is that the exposures being assessed occurred in the past, and therefore it can be difficult to assess them accurately, particularly if researchers rely only on people’s recall of events. However, in this study, researchers did not have to rely on people remembering or estimating their exposure to mobile phone masts, instead they used data on where individuals lived and known locations of mobile phone masts. This increases the reliability of the information about exposure.
What did the research involve?
The researchers analysed data from 1,397 children aged up to four years old who had cancer (the case group). They were compared with 5,588 children without cancer (the control group) who were matched to the cases for gender and date of birth. They determined where the children’s mothers had lived during their pregnancy, and how near that was to a mobile phone mast. They then compared the cases and controls to see if their mothers had lived at different distances from mobile phone masts, or whether they were exposed to different levels of power output from these masts.
To gather a suitable case group the researchers had identified all children in Great Britain aged up to four years old who were registered as having cancer in the national cancer registries for 1999 to 2001. They also noted which types of cancers these children had. For the 1,926 cases of early childhood cancer identified, there was sufficient data to include 1,397 of the children in the analyses (73%). For each child with cancer, they used national birth registries for Great Britain to identify four matched controls: children of the same sex born on the same date, and who were not recorded as having cancer in the national cancer registries.
For each child the researchers used their registered address or postcode at time of birth. They excluded children without a valid birth address or postcode. The four national mobile phone operators at the time of the study (Vodafone, O2, Orange and T-Mobile) provided information on all 81,781 mobile phone antennae in use from January 1 1996 to December 31 2001. This included where the antennae were, how many there were at each site (base station), dates at which they started and ended transmission, and features including the type of antennae, orientation, height above ground level, beam width, power output and frequency.
The researchers excluded 4,891 low-power antennae covering limited areas (called microcells, and accounting for 6% of the antennae). In total the researchers had full data on 66,790 (87%) of the 76,890 remaining antennae. Where data was missing, it was estimated using the data that the researchers had on other antennae, or was assigned the average (median) value for the company.
For each child, the researchers calculated the distance from the nearest base station, the total power output from all base stations within 700m (at ground level power density is reported to drop off rapidly after 500m). They also calculated ‘power density’ for base stations within 1,400m, essentially how much power was concentrated in a given area (exposures from over 1,400m away were considered to be at background levels).
The researchers had based their calculations of power density in a given area on measurements taken in a survey of a rural area (151 sites around four base stations) and an urban area (50 sites). These calculations used complex mathematical models, which were checked against data obtained from other surveys and measurements. The model appeared to perform better in predicting power density in rural areas than in urban areas. Pregnancies were assumed to last nine months, and exposure over the nine months prior to birth was estimated for each child.
The researchers looked at how mobile phone mast exposure in the womb related to an outcome of any childhood cancer and to specific cancers (brain and central nervous system cancers, leukaemia, and non-Hodgkin’s lymphomas). They took into account factors that could affect results, including socioeconomic deprivation, population density and population mixing (migration into the area in the previous year). Data on these factors was obtained from the 2001 census for the small area containing the birth address (census output area).
What were the basic results?
Of the 1,397 cancer cases, 527 were leukaemia or non-Hodgkin’s lymphoma (38%), and 251 were cancers of the brain or central nervous system (18%). Cases and controls were similar in terms of social and demographic characteristics.
The researchers also found that:
- Children who had cancer had birth addresses 1,107m from the nearest base station on average.
- Controls had birth addresses 1,073m from the nearest base station on average.
- There was no significant difference between children with early childhood cancer and controls in distance of birth address from the nearest base station.
- There was no significant difference between the children with cancer and the controls in terms of total power output or modelled power density exposure at their birth addresses while in the womb.
Distance from nearest base station, total power output and modelled power density did not differ between healthy controls and children with specific cancer types (either leukaemia and non-Hodgkin’s lymphoma, or brain and central nervous system cancer).
How did the researchers interpret the results?
The researchers concluded that they found “no association between risk of childhood cancers and mobile phone base station exposures during pregnancy”. They say that their results “should help to place any future reports of cancer clusters near mobile phone base stations in a wider public health context”.
This study appears well conducted. Its strengths include:
- Analysis of data from children born across Great Britain and inclusion of a high proportion (73%) of all registered early childhood cancer cases in Great Britain for the period assessed (1999-2001). This reduces the possibility that the area or children selected may not be representative of most cases.
- The use of three different measures to assess exposure to mobile phone base stations during pregnancy, none of which showed an association between exposure and childhood cancer.
The study’s limitations include:
- Only assessing the effects of exposure during pregnancy on early childhood cancers (up to age four). Longer-term effects or effects of later exposure during infancy and childhood were not assessed.
- Researchers did not measure individual exposure and therefore had to use surrogate measures of exposure – these may not fully capture or reflect individual exposure. Although measuring individual exposure would have been more accurate, doing so with a large cohort of pregnant women would be unlikely to be feasible.
- The researchers had to make certain assumptions in order to carry out their analyses. For example, they assumed that all pregnancies lasted nine months and calculated exposures based on registered birth address. In some cases, pregnancies may have been shorter or slightly longer than nine months, and mothers may have moved house or have spent significant amounts of time in other areas (eg for work). The accuracy of the assumptions may affect the results.
- The researchers were not able to assess radiofrequency exposure from other sources, such as low power mobile phone antennae, maternal use of mobile phones in pregnancy, radio or TV transmitters, or cordless phone base stations.
- The technology used in mobile phone masts may have changed since the study assessment period (1996-2001), therefore results may not be representative of modern exposure levels
- Although the researchers took into account factors that could affect results, these or other factors may still be having an effect.
Analysis by Bazian
Edited by NHS Website
Links to the headlines
The Guardian, 23 June 2010
The Independent, 23 June 2010
Links to the science
British Medical Journal, 2010;340:c3077
British Medical Journal, 2010;340:c3015