We found a dose response association between higher
We found a dose-response association between higher birth order and risk of childhood N-octanoyl-L-Homoserine lactone tumors. Previous case-control studies have reported similar results for overall childhood brain tumors [11,, , , ] and particularly for astrocytomas , and embryonal tumors , but this is not consistent in the literature [20,, , , , ]. Our analysis by tumor subtypes was underpowered but showed that the effect might be specific to astrocytomas. Birth order is traditionally used in epidemiologic studies as a surrogate marker of frequency and timing of exposure to infections in early life [36,37]. Specifically, later-born children are considered to be exposed to a larger burden of infections at an earlier age, as compared to their older siblings [36,37]. Hence, earlier exposure to infections possibly associates with an earlier maturation of the immune system that might act protectively against tumorigenesis . However, other mechanisms including different hormonal exposure of later conceived fetuses  and microchimerism  might also be involved in the observed association.
History of living in a farm was associated with a 5-fold higher risk of brain tumors, which was consistent for both astrocytomas and embryonal tumors. This finding might be related to exposure to pesticides early in life. A meta-analysis has shown that paternal exposure to pesticides either during pregnancy or early in life after birth is associated with increased risk of childhood brain tumors . Individual studies have further shown that residential use of pesticides is particularly associated with astrocytomas  and embryonal tumors , which might also relate to the genetically determined capacity of the child to metabolize toxic pesticide substances [44,45]. Pesticides are designed to act in the nervous system and some of them have been shown to be carcinogenic in animal models [5,46]. Alternative explanations could include a lower risk of allergies, socioeconomic disparities, and exposure to animals, but none of these factors were associated with brain tumors in our analysis.
Alcohol consumption was further associated with higher risk of brain tumors. While this finding is in accordance with studies in other childhood neoplasms, including leukemia  and neuroblastoma , it contradicts the results from a combined analysis of two population-based French studies that showed no evidence of an association . Alcohol consumption might simply be an indicator of other lifestyle choices during pregnancy which could explain the increase in the risk of brain tumors and possibly also the differences between the two studies.
Finally, our results did not support the associations of birth weight and size for gestational age with the risk of childhood brain tumors, which we recently showed in a meta-analysis, possibly because of restricted statistical power in this analysis . This might relate with the very high proportion of caesarean section deliveries in Greece, which leads to infants born on average at an earlier gestational week than expected and consequently with lower but still appropriate for their gestational age birth weights, as compared to other settings [26,50,51]. Due to compliance with the MOBI-KIDS questionnaire, gestational age could not be precisely determined for all participants, thus possibly leading to misclassifications in size for gestational age, which could attenuate a potentially significant effect. Nevertheless, the size of the adjusted for gestational age effect estimate for birth weight in the current case-control study was comparable to the pooled estimate derived in our meta-analysis , albeit not reaching statistical significance due to low power issues.