Researchers publishing in the Journal of Environment and Public Health investigate the association between risk of brain tumour and lifestyle factors e.g. mobile phones.
A roundup accompanied this analysis.
Title, Date of Publication & Journal
‘Brain tumours: rise in Glioblastoma Multiforme incidence in England 1995–2015 suggests an adverse environmental or lifestyle factor’ by Alasdair Philips et al
Journal of Environment and Public Health
Thursday 3 May 2018.
Study’s main claims – and are they supported by the data
The paper claims that there has been a rise in the occurrence of an aggressive type of brain tumour (glioblastoma multiforme, GBM) from 1995 to 2015. Whilst the data do appear to show an increase specifically in this type of tumour, the methods are not sufficiently clear to follow precisely what the authors have done, and it would be very difficult to replicate the analysis in the paper from the information provided here.
The paper also claims that the percentage rise in this brain tumour is similar across different age groups, and suggests that widespread environmental or lifestyle factors must be the cause. This claim is quite poorly supported by the data (see Figure 5), as there are much higher relative changes between earlier and later time periods (in both 15-19 year-olds and in older adults).
The press release claims that the study points to lifestyle factors including excessive phone use as a likely causal factor in the rise in aggressive brain tumours. This statement is a huge exaggeration of both the results and conclusions of the paper, as the data analysed only refer to the number of cases for brain tumours and do not tell us anything about causal factors, such as mobile phone use, or any of the other environmental factors discussed by the authors. The comments in the press release (most of which refer to mobile phones as a potential cause) are therefore also not supported by the data.
The increase in the rate of GBM tumour that the authors claim to have found is highly statistically significant in all parts of the brain and at all ages, suggesting that this reported increase in incidence is reasonably robust.
The authors note that the impact of improved imaging techniques and increased use of multidisciplinary teams may explain the rise in numbers of GBM tumours found in older people.
The authors use a large dataset from the Office of National Statistics, which therefore should be appropriately representative of the English population, and they highlight some potential errors that may exist within the dataset.
The statistical methods are not sufficiently explained in the paper. It looks as though there may be some more details in the supplementary information, but I do not have access to this. In either case, key details that should be easily visible are not given in the paper.
The authors have not highlighted the difference between absolute and relative rates. The 1.5% relative change between 1995-1999 and 2011-2015 sounds like a substantial increase, but as these types of tumour are very rare this corresponds to approximately only 2 additional cases per 100,000 people per year.
The authors have given some thoughts in the discussion on possible causal factors for the statistical trend they claim to have found, and the press release mentions mobile phone use as a possible causal factor, but these thoughts are not based on any of the data and analyses in this paper, so they are not at all supported by any evidence.
The authors have not clarified whether the groupings they have decided for their analysis were agreed in advance of looking at the data.
The authors have a section in the paper on ‘confounding’ but they do not fully explain what their confounders are nor how they have accounted for them in their analysis. Other confounding factors that they have not statistically accounted for may have led to the results seen here (e.g. improved imaging, changes in tumour classification, meaning more brain tumours are now picked up). Whilst these other factors are discussed by the authors, they are not taken into account in the analysis itself.
There is no adjustment for performing multiple statistical tests – performing multiple statistical tests makes it more likely that a statistically significant trend will be found. However, the statistical significance of the increase in GBM incidence is so strong that had they accounted for multiple testing, the results would still have been found to be statistically significant.
There is no data at all presented to support any suggestion of mobile phone use being responsible for an increase in incidence of GBM brain tumours.
Glioblastoma multiforme (GBM): an aggressive type of brain tumour, typically with a very poor prognosis.
Confounding: When a variable influences both the dependent variable (tumour incidence) and the independent variable (time) and may therefore lead to a spurious association.
*‘Brain tumours: rise in Glioblastoma Multiforme incidence in England 1995–2015 suggests an adverse environmental or lifestyle factor’ by Alasdair Philips et al., in press in the Journal of Environment and Public Health on Wednesday 21 March 2018 and published Monday 24 June 2018.
Before The Headlines is a service provided to the SMC by volunteer statisticians: members of the Royal Statistical Society (RSS), Statisticians in the Pharmaceutical Industry (PSI) and experienced statisticians in academia and research. A list of contributors, including affiliations, is available here.