We conducted a case -control study to investigate the role of early infections in the aetiology of childhood acute leukaemias. The study included 280 incident cases (240 acute lymphoblastic leukaemia and 40 acute non-lymphoblastic leukaemia) and 288 hospital controls, frequency matched by age, gender, hospital, catchment area of the hospital and ethnic origin. Data were obtained from standardised face-to-face interviews of the mothers. The interviews included questions on early common infections, day-care attendance, breast-feeding, birth order and infantile diseases. Odds ratios were estimated using an unconditional regression model including the stratification variables, parental socio-economic status and perinatal characteristics. Birth order was not associated with childhood leukaemia (acute lymphoblastic or acute non-lymphoblastic). A statistically-significant inverse association was observed between childhood leukaemia and day-care attendance (odds ratio=0.6, 95% Confidence Interval=(0.4 -1.0)), repeated early common infections (54 per year before age two, odds ratio=0.6 (0.4 -1.0)), surgical procedures for ear -nose -throat infections before age two (odds ratio=0.5 (0.2 -1.0)) and prolonged breast-feeding (56 months, odds ratio=0.5 (0.2 -1.0)). In the multivariate model including day-care attendance, early common infections and breast-feeding, results concerning breast-feeding remained unchanged. A statistically significant interaction between day-care attendance and repeated early common infections was observed. When the interaction was taken into account, the simple effects of day-care and early common infections disappeared (odds ratio=1.1 (0.5 -2.3) and odds ratio=0.8 (0.5 -1.3), respectively) while the joint effect of day-care attendance and early common infections was negatively associated with childhood leukaemia (odds ratio=0.3 (0.1 -0.8)). All the above associations were observed both for acute lymphoblastic leukaemia and acute non-lymphoblastic leukaemia. Our results support Greaves' hypothesis, even though they are not specific of common leukaemia.

The results presented support the hypothesis that prolonged breast-feeding may protect against childhood acute leukaemia and suggest an association with a maternal history of repeated miscarriages.Several recent studies have suggested that breastfeeding reduces the risk of childhood acute leukaemia and some prenatal factors, such as maternal history of fetal loss, have also been considered to play a role. This paper reports on perinatal data generated by a French casecontrol study which, overall, was designed to assess the role of perinatal, infectious, environmental and genetic factors in the aetiology of childhood acute leukaemia.A hospital-based, case-control study was conducted between 1995 and 1999 in the hospitals of Lille, Lyon, Nancy and Paris (France). The hospital-based design of the study was chosen since case and control blood samples were required. In order to be eligible for the study, the cases were required to present with newlydiagnosed primary leukaemia, be aged 15 years or less, and reside in the hospital catchment area. The controls were children hospitalised in the same hospital as the cases, mainly in orthopaedic and emergency departments, and residing in the same area as the cases. However, children hospitalised for cancer or major congenital birth defects were not eligible for the study. For both cases and controls, only French speaking mothers were eligible for the interview. The mothers of two cases and two controls refused to participate. Thus, a total of 280 cases of acute leukaemia (AL) and 288 controls were included in the study. Recruitment was frequency matched by age, gender, hospital, hospital catchment area and ethnic origin.Face-to-face interviews were carried out with the mothers of the cases and of the controls, using a standard questionnaire. Case mothers were interviewed when the index child was in complete remission or in good condition (on average, 2 months post-diagnosis). Interviews of case and control mothers were performed at about the same time in order to minimise recall bias. The mothers were asked to provide certain birth characteristics, lifestyle and medical history of the index child (breast-feeding) and a detailed maternal reproductive history (miscarriages, therapeutic or voluntary abortions, extra-uterine pregnancies, stillbirths). All analyses were performed using the SAS software package. Odds ratios (OR) were estimated using an unconditional logistic regression model including stratification variables. The analyses of breast-feeding were conducted on the children aged over 2 years in order to be certain that weaning had already taken place for both cases and controls.