Lack of exposure data is the most commonly cited factor preventing identification of causal association between environmental and occupational risk factors and health effects.1 Environmental exposure assessment presents more challenges than occupational exposure assessment, particularly when carried out retrospectively; large proportions of populations exposed to low levels of ubiquitous substances make consideration of all possible sources of exposure necessary.

Brosselin et al in their study of acute childhood leukaemia (see page 598) used a questionnaire to collect information on businesses in the vicinity of the home from pregnancy to date of diagnosis, focusing on proximity to repair garages and petrol stations.2 They find a significant association with ever having lived in proximity to these two businesses and suggest that benzene exposure may be a putative factor.

Occupational studies have shown a clear relationship between acute non-lymphoid leukaemia and high levels of benzene exposure and have served as a basis for risk assessment and regulatory rule making.3 4 A more common scenario such as that in the study by Brosselin et al, is more …