I doubt that any of us are recommending exposing children to pesticides - it doesn't take a medical degree to guess that nothing good would come of that. However, it's tougher to quantify the risks of exposure to environmental toxins. This study offers some insight, but also has some serious flaws.
Source: Chen M, Chang C-H, Tao L, et al. Residential exposure to pesticide during childhood and childhood cancers: a meta-analysis. Pediatrics. 2015;136(4):719-729; doi:10.1542/peds.2015-0006. See AAP Grand Rounds Commentary by Dr. Mary-Jane Staba Hogan (subscription required).
PICO Question: Among children exposed to indoor or outdoor residential pesticides, what is the risk of developing childhood cancer?
Question type: Prognosis
Study design: Meta-analysis
I've often said that meta-analytic studies are the most dangerous of study designs, because so few clinicians are sufficiently aware of the nuances of methodology and statistical arguments that can alter conclusions. However, the authors of this study did a pretty good job of stating their methodology and the study limitations. It's worth reading the study carefully, anyone can learn a lot about meta-analysis from it. They did found a small but statistically significant association of indoor pesticides and childhood leukemia and lymphomas, as well as a possible herbicide-leukemia association.
Stated up front, in the abstract, is the major limitation: very few high-quality studies have looked at this problem. So, if you try to combine the results in a meta-analysis, you're skating on thin ice. With so few studies, I'm surprised the authors limited their review to only English-language reports. They might have found more studies to include if they had branched out to other languages (easy for me to say - the researchers would need multiple translators to clear this hurdle).
If you take the time to read the article, pay attention to a few key points in meta-analysis:
1. Heterogeneity - because of the low number of articles, the authors appropriately used the I squared statistic, and found that there was a low degree of heterogeneity. Translation - the studies were probably similar enough in design that any differences observed in the meta-analysis were unlikely to be due to chance alone.
2. Publication Bias - the authors checked this using funnel plots and a statistical method called Egger's test, both appropriate. This helps assess whether the published studies are biased towards one point of view - in this case it might be more likely for studies showing a cancer risk to be published (particularly in English-language journals) compared to studies that did not show a risk, because those results are less interesting. The authors found no trend towards publication bias.
Dr. Staba Hogan's AAP Grand Rounds commentary is spot on to draw attention to other study limitations, including recall bias (parents of children with cancer might remember things about pesticides a bit differently than would parents of healthy children) and lack of analysis of other environmental carcinogens in some of the studies.
I see this report as interesting, but nothing I would mention to parents unless they happen to ask about this particular study. With the study limitations, it doesn't reach my threshold for usefulness to parents. Suffice to say it's a good idea to keep kids away from pesticides. The AAP policy statement on pesticide exposure is a good reference point.