Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In a study of childhood cancer and possible exposure to benzene from traffic and petrol stations Harrison et al 1 found a non-significant increase in childhood leukaemia. In their conclusion they state that “… the data are overall rather reassuring in showing that any such effect, if real, is likely to be small in magnitude”. We think that this conclusions cannot be drawn from such limited material.
Firstly, they merge all types of leukaemia, whereas studies in adults have indicated that non-lymphocytic leukaemia is most strongly associated with benzene. An ecological Swedish study found an association between car density and NLL but not with other types of leukaemia or lymphoma.2
Secondly, the misclassification of exposure may be substantial in the study by Harrison et al. In Sweden wood burning makes a major contribution to the exposure to benzene in some areas. Probably the burning of other fuels for heating will also contribute. Children are transported in cars, stay in kindergartens, schools, sports areas, shopping areas, etc for much of their days. Even if stationary sampling shows a difference in concentration of pollutants close to roads, that may not mean that the exposure of the children differs in the same way. A Danish study found that although front door concentrations of benzene were significantly higher in a city than in rural areas (8.9 v 1.9 μg/m3), there was almost no difference in the mean personal exposure of these children (5.4 v4.5 μg/m3).3 Such misclassification of exposure leads to an underestimation of the relative risk.
The important findings of Harrison et al and other studies2 4 (see London Research Centre1 for further references) is that living in highly populated areas, with a lot of traffic or high car density, means a higher risk of childhood leukaemia, which may be preventable. We think that as benzene is a known cause of non-lymphocytic leukaemia in adults, much more thorough studies are needed in children before we can rule out that benzene is an important factor in childhood leukaemia.
So even if we disagree with Harrison et alabout the importance of the risk, we agree that further studies are necessary. They should focus on different types of leukaemia and use much better precision in the estimates of benzene and other possible exposures.
Authors' reply—Järvholm and Forsberg quite rightly point out that studies in adults have indicated that exposure to benzene is associated with an excess of non-lymphocytic leukaemia (NLL) and implicitly criticise us for having merged all types of leukaemia. The reason for merging the leukaemias was because lymphocytic leukaemias accounted for 176 of 225 leukaemias in the West Midlands in the age group 0–14 years in the period 1993–7, and therefore non-lymphocytic types accounted for only a very few cases, with only 42 registrations of acute myeloid leukaemia over this period. It seemed very unlikely that we would be able to obtain a significant result from so few cases.
Järvholm and Forsberg comment also on the potential for misclassification of exposure in our study. We do not think this very likely in the United Kingdom. Wood burning is almost unknown in the West Midlands conurbation, and indeed would not be consistent with United Kingdom smoke control laws. The inventory of air pollutant emissions for the conurbation1-1 indicates that 99% of benzene emissions to atmosphere arise from road traffic. Our paper gives some indication of the likely gradients between houses at the roadside and those in urban background locations. Although children are also exposed to benzene when travelling in cars, the duration of exposure is relatively short and our earlier paper1-2 on personal exposures to aromatic hydrocarbons suggests that this will be a minor contributor to overall exposure. It is exposure in the home and workplace (or in the case of children, in the school) which is dominant. It is likely that other sources of exposure are relatively similar on average for the more exposed and less exposed groups and therefore the main difference relates to the point of residence and the influence of outdoor air on indoor concentrations.
Järvholm and Forsberg cite a Danish study which showed significantly higher front door concentrations of benzene in a city than in rural areas. In the United Kingdom West Midlands the risk of childhood cancer is higher in rural than in urban areas1-3 which suggests that benzene exposure is unlikely to be a major factor and that other causal agents such as population mixing may be far more influential.1-4 None the less our results do suggest a slight excess of cancers for children living close to major roads and petrol stations and we wholly agree with Järvholm and Forsberg that this merits further investigation.