June 20, 2016

socioeconomic status and brain tumours

Socioeconomic status and risk of developing brain tumours is investigated in a paper published in the Journal of Epidemiology and Community Health with the authors reporting a higher risk of glioma for those with measures of higher socioeconomic status. Roundup comments accompanied this analysis.

Title, Date of Publication & Journal

Socioeconomic position and the risk of brain tumour: a Swedish national population-based cohort study.  20 Jun 2016, J Epidemiol Community Health

Study’s main claims – and are they supported by the data

The paper does not support the basic claim in the press release that a university degree is linked to a heightened risk of developing a brain tumour. This study does not compare having a degree to not having a degree; instead, it compared the people with the most basic level of education with four other groups of increasing education.

Also, the paper and the press release talk about increased risk, but that implies there is something causal about having a degree. In fact, all this study looks at is the incidence of tumours. In other words, it does not demonstrate that more education is more risky.

The authors themselves don’t suggest this is causal and they admit that the entire effect is likely to be for other reasons altogether (see summary below).

For glioma:

The men with the lowest level of education in the study were found to have lower incidence of glioma when compared to the three highest groups.

In women, only the highest group (university degree) showed increased incidence when compared to the most basic group.

In summary: risk of glioma for 3+ years of higher education (highest group) was 19% higher in men and 16% in women compared to those with <9 years of education (lowest group).

For meningioma:

There are relatively few meningioma cases

In men there weren’t enough to be statistically significant

In women, incidence of meningioma was 16% higher in those with 3+ years of higher education compared to those with only primary education, but the other groups were not statistically significant.

The paper also presents evidence that those in middle and high non-manual occupations (men) or low, middle or high non-manual occupations (women) have higher risk of glioma than those in lower manual occupations. Results for women or other types of brain tumour were generally less consistent than for glioma in men.

In summary:

All of these observed associations, as the authors themselves state, may not be directly causal. This study does not show that the more education you have, the more likely you are to get a brain tumour. In other words, having education is likely to be a proxy for something else.

Possible explanations include confounding by lifestyle factors (such as smoking or BMI, which were not recorded) and biases in cancer registration (i.e. some people are more likely than others to be diagnosed).

The people in the group with the lowest education are likely to be much older than the others. The authors have tried to adjust for confounding here but even so this group could be markedly different in other ways – people born in 1911 are going to be very different to people born in 1961 in terms of the risk factors they have been exposed to in their lifetimes.

Absolute risks of brain tumour associated with socioeconomic status are anyway likely to be very small because brain tumours are rare. This is not discussed in the paper or press release.

Strengths/Limitations

Strengths:

This was a very large study of more than 4 million individuals on the Swedish Total Population Register. 5735 and 7101 brain tumours were identified in men and women respectively. Highest education, occupation and current income were ascertained from other linked databases. Outcomes were obtained from the Cancer Register.

Limitations:

There are no tables of descriptive statistics. So, for example, we can’t tell if those in one group are older than those in another. This makes it hard to see the whole picture.

The analyses are presented ONLY with respect to the group with lowest socioeconomic status. A test for a linear trend would have helped us see if there’s anything really going on.

The study did not have information on lifestyle factors such as BMI or smoking which may have been confounders in the observed associations.

Brain tumours are rare which makes it hard to spot a pattern. Despite the large size of this study there were only about 400 acoustic neuromas each in men and women. Effect sizes were also small and associations were inconsistent, particularly for acoustic neuromas and meningioma. This makes interpretation of study result difficult.

Glossary

Confounding: apparent association between an exposure and outcome that is actually due to something else.

Any specific expertise relevant to studied paper (beyond statistical)?

I have experience in analysis of cohort studies but no specific knowledge of brain tumours.

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.