Methods

Aus-ALL was a population-based case–control study of ALL in children aged under 15 years in Australia, which took place between 2003 and 2006. Full details of the study population and recruitment methods used have been published elsewhere.6 ,7 Briefly, cases and their families were identified and recruited from 10 paediatric oncology centres in Australia and were eligible if they were diagnosed between 1 July 2003 and 31 December 2006 and were in remission. Three controls per case were recruited in seven recruitment waves over the same period, by random digit dialling, and were frequency matched to cases on age, sex and state of residence. The biological mother was required to have sufficient English language skills to complete the questionnaires for any potential case or control child to be eligible to participate. Human research ethics committee approval was obtained from all participating hospitals.

Written questionnaires seeking information on possible confounders including maternal and paternal age, smoking history etc. were completed by the parents. In addition, a lifetime occupational history up to the birth of the child for the father and up to 1 year after the birth of the child for the mother was collected. Information provided for each job included the year started and finished, the job title, the employer, main tasks and the number of hours worked each week. If the parent reported ever working in a job in which pesticide exposure was possible (eg, farmer), trained interviewers telephoned and asked the parent further specific questions about their job. This included information about the crops grown or the livestock handled, whether they used pesticides and the pest that they were used to control. The circumstances of use were also noted, for example, did they mix or spray the pesticide and what protective equipment did they use while handling the pesticides. The names of any pesticides that they recalled were noted.

An expert (DCG) reviewed the job histories and the answers to the job-specific questions to assess exposure to pesticides. She was blinded to case/control status and determined the likelihood of exposure (no exposure, possible exposure or probable exposure) to organophosphate insecticides, organochlorines, phenoxy herbicides, other herbicides, other pesticides. She also assessed the level of that exposure as high, medium or low. depending on the likelihood of individual inhalational and dermal exposure. Exposure was considered high if pesticides were mixed or sprayed without adequate PPE (gloves, respiratory protection and overalls). The use of adequate PPE, no mixing and no spraying was considered low exposure. The number of years in the job (duration of exposure) was determined from the job history.

Exposure variables were created for specific time periods (before and up to the child's birth and after the birth of the child (mothers only)) and for specific jobs (exposure in the jobs held 2 years before the birth of the child (mothers only) and exposures in the jobs held 1 year before the birth of the child (mothers and fathers)), in order to capture the exposure that occurred during the periods of pre and periconception, during gestation and post natally. Exposures for each time period were coded as none (no exposure), low (probable low exposure and possible low/medium/high exposure) and medium/high (probable medium or high exposure). Exposure variables in specific jobs up to 2 years and 1 year before the birth of the child were created by identifying the exposures in the job that was closest to the specified time period. Because of smaller numbers these variables were grouped as ‘not exposed’ and ‘exposed’. A variable indicating exposure to any pesticide was also created using the highest level of exposure to any of the individual pesticide groups.

The indicator used to assess socioeconomic status was the index of relative socioeconomic disadvantage (IRSD), one of the area-based measures calculated by the Australian Bureau of Statistics. The address at the time of study entry was linked to the 2006 census collection district (the smallest unit of population for which census information is available) and then to the corresponding IRSD. Further details of this process have been published elsewhere.6

Demographic and exposure variables were compared between cases and controls using χ² tests and Fisher's exact tests.

The associations between maternal and paternal exposure and the risk of ALL in the offspring were investigated using logistic regression and adjusting for potential confounders (child sex, child age at diagnosis, socioeconomic status, parental smoking during birth year, and parental drinking alcohol 1 year before pregnancy, professional domestic pest control treatment and the maternal models also adjusted for maternal age). Models examined the relationship between occupational exposure at any time before and up to the birth of the child, in the job 2 years before the birth of the child, in the job 1 year before the birth of the child and in the period up to 1 year after the birth of the child. All analyses were undertaken using Stata V.10.1.