Residential Pesticides and Childhood Leukemia: A Systematic Review and Meta-Analysis

Objective We conducted a systematic review and meta-analysis of previous observational epidemiologic studies examining the relationship between residential pesticide exposures during critical exposure time windows (preconception, pregnancy, and childhood) and childhood leukemia. Data sources Searches of MEDLINE and other electronic databases were performed (1950–2009). Reports were included if they were original epidemiologic studies of childhood leukemia, followed a case–control or cohort design, and assessed at least one index of residential/household pesticide exposure/use. No language criteria were applied. Data extraction Study selection, data abstraction, and quality assessment were performed by two independent reviewers. Random effects models were used to obtain summary odds ratios (ORs) and 95% confidence intervals (CIs). Data synthesis Of the 17 identified studies, 15 were included in the meta-analysis. Exposures during pregnancy to unspecified residential pesticides (summary OR = 1.54; 95% CI, 1.13–2.11; I2 = 66%), insecticides (OR = 2.05; 95% CI, 1.80–2.32; I2 = 0%), and herbicides (OR = 1.61; 95% CI, 1.20–2.16; I2 = 0%) were positively associated with childhood leukemia. Exposures during childhood to unspecified residential pesticides (OR = 1.38; 95% CI, 1.12–1.70; I2 = 4%) and insecticides (OR = 1.61; 95% CI, 1.33–1.95; I2 = 0%) were also positively associated with childhood leukemia, but there was no association with herbicides. Conclusions Positive associations were observed between childhood leukemia and residential pesticide exposures. Further work is needed to confirm previous findings based on self-report, to examine potential exposure–response relationships, and to assess specific pesticides and toxicologically related subgroups of pesticides in more detail.


Review
Leukemia is the most common form of child hood cancer in Canada and the United States, accounting for > 30% of new cancer cases (American Cancer Society 2009; Canadian Cancer Society/National Cancer Institute of Canada 2009). During 2000-2004, there were nearly 1,400 new cases of leukemia among children 0-14 years of age in Canada, with incidence rates highest among those 0-4 years of age (Agha et al. 2006; Canadian Cancer Society/National Cancer Institute of Canada 2009). Acute lymphoblastic leukemia (ALL) accounts for most (~ 80%) childhood leukemia cases, followed by acute myelogenous leukemia (AML) (Canadian Cancer Society/National Cancer Institute of Canada 2009). Although much progress in treating childhood leukemia has been achieved, treatment entails substantial morbidity, and elevated morbidity and mortal ity outcomes continue to be observed among survivors compared with children who have not developed the disease (MacArthur et al. 2007;Speechley et al. 2006).
Acute leukemias are heterogeneous, charac terized by different genetic and chromosomal abnormalities, with differing frequency by age (Greaves 2002). The twostep model for childhood leukemia proposes that leukemia development occurs after both a first mutation, usually a chromosomal translocation occur ring in utero, and a second mutation occurring after birth (Greaves 2002;Rossig and Juergens 2008). Children with Down syndrome experi ence an elevated risk for the disease (Alderton et al. 2006;Ross et al. 2005). Although a vari ety of environmental and chemical exposures have been suggested to play a role in the etiol ogy of the disease, ionizing radiation remains the sole environmental risk factor established to date (Belson et al. 2007). Other potential risk factors that have received some attention in the scientific literature include parental smoking and alcohol consumption, electro magnetic field exposure, hydrocarbons, socio economic factors, immunity and infection, and pesticides (Belson et al. 2007;Greaves 2006;InfanteRivard and ElZein 2007;Lee et al. 2009;Rossig and Juergens 2008;Schuz and Ahlbom 2008).
Several studies examining the potential association between childhood leukemia and both parental occupational and residential pesticide exposure have been conducted over the past several decades, with positive associations observed (InfanteRivard and Weichenthal 2007). Partly because of con cerns surrounding potential adverse child health impacts, several Canadian provinces and municipalities have recently banned the cosmetic use of pesticides on public or private property (Arya 2005; Ontario Ministry of the Environment 2008). Similar bans are also being considered elsewhere.
Residential pesticide use is associated with elevated child exposures. Use of pyrethroid insecticides in the household was found to be a significant predictor of urinary pyrethroid metabolite levels in children in a recent longi tudinal study (Lu et al. 2009). Child urinary concentrations of two organophosphorus pesticide metabolites (dimethyl and diethyl dialkyl phosphate compounds) were found to be higher with parental garden pesticide use but not with pet treatment or indoor pesti cide use in a Seattle study (Lu et al. 2001).
We conducted a systematic review and metaanalysis of previous observational epi demiologic studies examining the relation ship between residential pesticide exposures during critical exposure time windows (pre conception, pregnancy, and childhood) and childhood leukemia and explored potential methodological and clinical sources of hetero geneity in results. Although there have been previous reviews, none have included a quan titative synthesis of the results available to date. Results of an analysis examining the association between childhood leukemia and parental occupational pesticide exposure are presented in a separate, companion review (Wigle et al. 2009).

Materials and Methods
This systematic review and metaanalysis was conducted according to a protocol designed by M.C.T. and D.T.W.
Literature search. The search strategy was designed to identify previous observational epidemiologic studies examining the rela tionship between residential pesticide expo sures during critical exposure time windows (preconception, pregnancy, childhood) and childhood leukemia. Preliminary searches using Ovid MEDLINE were conducted to inform the design of the final search strategy detailed below. An information specialist at the University of Ottawa was also consulted in finalizing the search strategy.
The search strategy was first developed to search the Ovid MEDLINE (1950-March week 3, 2009  All titles and abstracts identified were independently examined by two of us (M.C.T. and D.T.W.) in order to deter mine their potential suitability for inclusion in the systematic review. After this primary screen, the complete articles were obtained and the inclusion/exclusion criteria applied. Discrepancies were resolved by consensus. No language criteria were applied. Where abstracts were identified or further details required, particularly relating to the designa tion of pesticide exposure as residential or occupational, the corresponding author was contacted to ascertain further details of the study. In addition to searching the databases listed above, the reference lists of all included studies and journal Web sites were also hand searched; studies identified manually were evaluated in the same manner as above.
Inclusion and exclusion criteria. Original epidemiologic studies of childhood leukemia using a case-control or cohort design with an assessment of at least one index of residential/ household pesticide exposure/use were included here. Reports were excluded if they were review articles, ecologic studies, case reports, cluster investigations, or studies of adults or if they examined residential exposure or proximity to agricultural pesticides. Where there were mul tiple publications, the most relevant report was retained (usually the most recent).
Data abstraction. After identification of all relevant studies, data abstraction was performed independently by the same two reviewers (M.C.T. and D.T.W.). A stan dard data abstraction form was prepared and piloted to collect relevant data related to referencing, study design, subject selection, exposure assessment, statistical analysis, and results. A single exposure index was identi fied for each original study and, where data were available, each combination of exposure time window and pesticide type (unspecified, insecticides, herbicides). Quality assessment. All included studies underwent independent quality assessment by the same two reviewers (M.C.T. and D.T.W.). We used a modified version of the Downs and Black (1998) checklist for the assessment of the methodological quality of randomized and non randomized studies of health care interventions [Supplemental Material, Table 2 (doi:10.1289/ ehp.0900966.S1)]. Before conducting the qual ity assessment, the two reviewers discussed the individual items on the checklist to clarify their interpretation. No attempt was made to blind the reviewers of the authorship or publication status of the original studies. Differences in quality assessment were resolved by consensus.
Analysis. We conducted metaanalyses using Review Manager (RevMan) version 5.0 (Nordic Cochrane Centre, Cochrane Collaboration, Copenhagen, Denmark). Generic inverse variance data were combined using random effects models to obtain a sum mary odds ratio (OR) and 95% confidence interval (CI) for the relationship between resi dential pesticide exposures (unspecified, insec ticides, herbicides) and childhood leukemia by exposure time window (preconception, preg nancy, childhood). Heterogeneity across indi vidual studies was quantified by the I 2 statistic (Higgins et al. 2003). Low, moderate, or high degrees of heterogeneity may be approximated by I 2 values of 25%, 50%, and 75%, respec tively (Higgins et al. 2003). We conducted subgroup analyses according to total quality score (≥ median) and individual quality score components (> median for external validity and exposure measurement), study design (hospital based or populationbased case-control study), cell type (ALL, AML), location (indoor, out door), maternal residential pesticide use (vs. household use or exposure) only, year of publi cation (studies published in 2000 or later only), and publication status (studies published in the peerreviewed literature only). Where multiple exposure indices were reported per exposure time window, pesticide type, and study, sen sitivity analyses were undertaken using expo sure or time window definitions different from those used in the main analysis. Finally, we also examined the impact of removing studies with extreme ORs or the highest weight in analy sis, as well as removing individual studies in a sequential manner. Because of the small num ber of included studies, we assessed publication bias by visual inspection of inverted funnel plots, based on the main finding from all stud ies (Ioannidis and Trikalinos 2007).

Study identification.
The results of the search strategy and study selection process are detailed in Figure 1. Of the 1,776 studies identified using our search algorithm, 112 were retained from the primary screening process. Most stud ies were excluded during primary screening because they were irrelevant (n = 1,178), a duplicate record (n = 380), or a review article (n = 93). After the secondary screening process, 17 studies were retained (listed in Table 1). Major reasons for exclusion during secondary screening were irrelevance (n = 36), examina tion of occupational or residential exposure to agricultural pesticides exposure only (or unclear) (n = 27), or a letter or editorial with no results presented (n = 14).

Study characteristics.
Of the 17 identified case-control studies, 6 were hospital based, Included studies (n = 17) 10 were population based, and 1 reported results separately for both hospital and popu lation controls (Table 1). Studies were con ducted in the United States, Canada, Mexico, Japan, France, Brazil, and Germany. Most of the studies were published in the peer reviewed literature; however, three doctoral dissertations presented results not published elsewhere (Davis 1991;Dell 2004;Steinbuch 1994). Although most studies examined both ALL and acute nonlymphoblastic leukemia cases among children and adolescents up to a maximum age of 19 years, one study examined infantile acute leukemia (PombodeOliveira et al. 2006) and another examined both ALL and AML in children with Down syndrome (Alderton et al. 2006).
Studies varied in size, ranging from a total of 49 leukemia cases (with 7-25 cases ever exposed, depending on exposure category and time window) in the dissertation by Dell (2004) up to 1,184 cases (with 25-164 exposed) in the German study by Meinert et al. (2000). All studies conducted to date relied on parental reports of residential pes ticide exposure or use inside or outside of the home, either by themselves or by profes sional exterminators [Supplemental Material, Appendix 1 (doi:10.1289/ehp.0900966. S1)]. Although most studies assessed use of, or exposure to, pesticides or specific pesticide subgroups (insecticides, herbicides, fungicides), some studies also attempted to collect informa tion on pesticide names and formulations or on target organism (Davis 1991;Dell 2004;InfanteRivard et al. 1999;Ma et al. 2002;Steinbuch 1994). Nine studies considered both residential and occupational pesticide expo sures (Buckley et al. 1989;Dell 2004;Kishi et al. 1993;Lowengart et al. 1987;Meinert et al. 1996Meinert et al. , 2000Menegaux et al. 2006;Rudant et al. 2007;Steinbuch 1994), and the remaining eight studies were exclusively resi dential. Five studies clearly specified (or explic itly assumed) whether residential pesticide exposure during pregnancy was attributable to maternal use (Davis 1991;Lowengart et al. 1987;Ma et al. 2002;Menegaux et al. 2006;Rudant et al. 2007) as opposed to household use or exposure. Virtually all studies assessed pesticide expo sures during separate preconception, preg nancy, and childhood time windows; however, time window definitions differed somewhat by study (Appendix 1). Leiss and Savitz (1995) considered only the last 3 months of preg nancy. Ma et al. (2002) considered the first 3 years of age in a separate manner. Davis (1991) considered the first 6 months of age separately from the remainder of the child hood period. Some studies also combined results from different exposure time windows in analysis and reporting (Meinert et al. 1996(Meinert et al. , 2000Rudant et al. 2007;Steinbuch 1994).
Quality assessment. Quality scores are presented in Supplemental Material, Table 3 (doi:10.1289/ehp.0900966.S1). For hospital based studies, total scores ranged from 7 to 12, with a median value of 9, of a possible maxi mum score of 20. For populationbased stud ies, quality scores were higher, with a range of 9-14 and a median of 11. More recent stud ies tended to have higher quality scores. In assessing external validity, questions remained regarding the representativeness of subjects (both selected and participating), particularly for earlier hospitalbased studies. Only Buckley et al. (1989) reported that interviewers were blind to case/control status; however, the eth ics of such practices have also been questioned (InfanteRivard et al. 1999). Because of the selfreported nature of exposure data, no study received a point for avoidance of bias from misclassification, since the possibility for dif ferential misclassification remained. Only Dell (2004) and Ma et al. (2002) reported results for a clearly defined preconception exposure time window. There were few data regarding frequency or duration of pesticide use, with most studies reporting only "ever/never" use of/exposure to the pesticide of interest. Six studies attempted to examine potential expo sure-response relationships (Alderton et al. 2006;Buckley et al. 1989;Dell 2004;Infante Rivard et al. 1999;Ma et al. 2002;Meinert et al. 2000). Although confounding is difficult to assess because there are few established risk factors for childhood leukemia, most studies examined or adjusted for at least a range of sociodemographic and maternal characteris tics. Four studies, however, explicitly assessed the potential confounding influence of mater nal or childhood Xray exposure (Dell 2004;FajardoGutierrez et al. 1993;Kishi et al. 1993;Lowengart et al. 1987).
Publication bias. To assess the possibility of publication bias, we examined the main findings from all included studies in an inverse funnel plot [Supplemental Material, Figure 1 (doi:10.1289/ehp.0900966.S1)]. Although limited by the small number of individual studies, there was some evidence for asym metry, with a lack of small studies found with effect sizes smaller than those from larger stud ies. Asymmetry may also be due to a range of other factors, including study quality, method ological differences, or the study populations per se. We attempted to identify all relevant original studies possible, including three doc toral dissertations (Davis 1991;Dell 2004;Steinbuch 1994) and two studies published in a language other than English (Fajardo Gutierrez et al. 1993;Kishi et al. 1993).
Data synthesis. Of the 17 identified studies, we excluded two from the quan titative data synthesis due to a lack of CIs (Schwartzbaum et al. 1991) or a unique study population (Down syndrome cases only) (Alderton et al. 2006). Supplemental Material, Appendix 2 (doi:10.1289/ehp.0900966.S1) lists the individual studies included in each overall and subgroup analysis by exposure time window and pesticide type.
Exposure to residential herbicides during pregnancy also had a significant positive asso ciation with childhood leukemia (OR = 1.61; 95% CI, 1.20-2.16; I 2 = 0%) when combin ing the results from five studies ( Figure 2C, Table 4). Again, we observed little difference in the summary OR according to study qual ity, study design, or publication status. The combined relative risk estimate increased somewhat for ALL (OR = 1.73; 95% CI, 1.28-2.35; I 2 = 0).
Results for the pregnancy exposure time window were fairly robust to sensitivity analy ses: removing studies with extreme ORs or with the highest weight or including addi tional studies with wide or illdefined expo sure time windows. However, removing the study of AML by Steinbuch (1994) from the unspecified pesticide analysis did result in a somewhat stronger association (OR = 1.74; 95% CI, 1.36-2.24; I 2 = 31%).
Finally, we observed no association between exposure to residential herbicides dur ing childhood and childhood leukemia when  (2004), which did not collect frequency information for all control groups. Where results were reported for leukemia overall as well as for specific cell types, the overall results were selected here. Where results were reported for indoor or outdoor pesticide use only, the indoor value was used here. Where results were reported for either owner-applied or professionally applied pesticides, the owner-applied value was used here. For Kishi et al. (1993), we selected results using population controls, except for the subgroup of hospital-based studies. For Buckley et al. (1989), unmatched OR of 1.47 (95% CI, 0.72-3.04) was used, calculated by collapsing the two highest exposure categories for pregnancy exposure. For the childhood time window, where studies reported results for different childhood time periods, the earliest was selected [for Ma et al. (2002), results for year 1 were used; for Davis (1991), results for 0-6 months were used; for Leiss and Savitz (1995), results from birth to 2 years before diagnosis were used]. b OR for Kishi et al. (1993) corrected to 1.80 for hospital controls, childhood exposure. c Using results for ALL, instead of overall leukemia, for Ma et al. (2002). d Unmatched OR of 3.52 (95% CI, 1.11-11.11) calculated from data in Lowengart et al. (1987). e Removing studies with the highest and lowest ORs. f Removing the study (or studies, in the case where there are two with identical values) with the highest weight in analysis. g Including studies with wide or ill-defined exposure time windows.
volume 118 | number 1 | January 2010 • Environmental Health Perspectives combining results for four studies overall (OR = 0.96; 95% CI, 0.59-1.58; I 2 = 72%) ( Figure 3C, Table 4). We also observed sub stantial hetero geneity. Sensitivity analysis using alternate exposure indices for Ma et al. (2002) (using year 2 of childhood as opposed to year 1) and for Davis (1991) (using garden herbicide use between 7 months of age and age at diagnosis as opposed to yard herbicides from 0 to 6 months of age) resulted in a summary OR for childhood leukemia of 1.38 (95% CI, 1.10-1.72; I 2 = 0%).

Discussion
In this metaanalysis, we examined previous observational epidemiologic studies of the asso ciation between residential pesticide exposure during critical exposure time windows (precon ception, pregnancy, childhood) and childhood leukemia. Overall, exposure to residential pes ticides during pregnancy was positively associ ated with childhood leukemia. We observed the strongest association for insecticides, with little evidence of heterogeneity. Results for childhood exposures were less clear. Although we observed overall positive and significant associations for exposure to unspecified pes ticides and insecticides during childhood, for insecticides the association attenuated among studies with higher total methodological qual ity. Few studies examined childhood exposure to herbicides, and we observed no overall posi tive association. Although we also examined preconceptional residential pesticide exposure, only two studies had clearly defined exposure time windows on which to base an assess ment of the effect of this type of exposure on  (2004), which did not collect frequency information for all control groups. Where results were reported for leukemia overall as well as for specific cell types, the overall results were selected here. Where results were reported for indoor or outdoor insecticide use only, the indoor value was used. Where results were reported for either owner-applied or professionally applied insecticides, the owner-applied value was used here. For Pombo- de-Oliveira et al. (2006), personal correspondence with the study author (13 March 2008) corrected the upper CI reported in the published article from 2.13 to 2.95 and confirmed that pesticide exposure was mainly insecticide exposure. For the childhood time window, where studies reported results for different childhood time periods, the earliest was selected [for Ma et al. (2002), results for year 1 used; for Davis (1991), results for 0-6 months used here; for Leiss and Savitz (1995), results from birth to 2 years before diagnosis used here]. For Fajardo-Gutierrez et al. (1993), personal correspondence with the study author (28 May 2008) confirmed exposure was postnatal exposure to insecticides in the home. b Using results for ALL, instead of overall leukemia, for Ma et al. (2002) and Rudant et al. (2007). c Using results for AML, instead of overall leukemia, for Rudant et al. (2007). d Removing studies with the highest and lowest OR. e Removing the study (or studies, in the case where there are two with identical values) with the highest weight in analysis. f Including studies that reported indoor unspecified pesticide use. g Including studies with wide or ill-defined exposure time windows. a Where studies used multiple indices of exposure categories, the highest was selected, except for Dell (2004), which did not collect frequency information for all control groups. Where results were reported for leukemia overall as well as for specific cell types, the overall results were selected here. For the childhood time window, where studies reported results for different childhood time periods, the earliest was selected [for Ma et al. (2002), results for year 1 used here; for Davis (1991), results for 0-6 months]. b Using results for ALL, instead of overall leukemia, for Ma et al. (2002) and Rudant et al. (2007). c Removing studies with the highest and lowest OR. d Removing the study (or studies, in the case where there are two with identical values) with the highest weight in analysis. e Including studies that reported outdoor unspecified pesticide use. For Leiss and Savitz (1995), the earliest time window from birth to 2 years before diagnosis used for childhood analysis. f Including studies with wide of ill-defined exposure time windows. childhood leukemia. We obtained some differ ences in results in subgroup analysis according to study quality, study design, and publication status or when using alternate exposure indices for some of the associations we observed. Previous reviews have concluded that there is likely to be a positive association between pesticide exposure and childhood leukemia (Daniels et al. 1997;InfanteRivard and Weichenthal 2007). Results from a com panion article revealed positive associations between childhood leukemia and prenatal maternal occupational exposure to pesticides (Wigle et al. 2009). Occupational pesticide exposures are of greater magnitude compared with those from other sources (Bradman et al. 2005;Mandel et al. 2005;Sala et al. 1999). Summary ORs for the relation between prena tal maternal occupational exposure to pesticides and childhood leukemia were larger compared with those here, with an overall summary OR of 2.08 (95% CI, 1.51-2.88), reported for any pesticide exposure, and 2.72 (95% CI, 1.47-5.04), reported for insecticide exposure, lending further credibility to the hypothesis.
Among the potential limitations of the present analysis is the possibility for publica tion bias. Although such bias can be difficult to assess, we found several small studies that were either unpublished (PhD dissertations) (Davis 1991;Dell 2004;Steinbuch 1994) or writ ten in languages other than English (Fajardo Gutierrez et al. 1993;Kishi et al. 1993). The magnitude of the association observed between unspecified pesticides and childhood leukemia tended to strengthen, and the heterogeneity reduce, on restriction to studies published in the peerreviewed literature only.
Original studies may be subject to limita tions related to exposure assessment and reporting. Typically, the quality of environ mental epidemiology studies is influenced by the quality of exposure measurement (Hertz Picciotto 1998). The studies in the present metaanalysis measured residential pesticide exposure entirely by parental report, and only in some instances were detailed data collected on specific types of pesticides or frequency of use. Although based on small numbers of exposed subjects, some limited evidence sup ports a positive exposure-response relation ship between childhood leukemia and both pregnancy and childhood household pesticide or insecticide exposure (Alderton et al. 2006;Buckley et al. 1989;InfanteRivard et al. 1999;Ma et al. 2002). Although there may be dif ferential misclassification of exposure among cases, it has also been suggested that nondif ferential misclassification may be of greater concern (InfanteRivard and Jacques 2000). Although none of the studies we included here appear to have attempted to validate self reported residential pesticide exposure infor mation, Meinert et al. (1996Meinert et al. ( , 2000 examined the risk of both childhood leukemia and solid tumors in the same study. They found a posi tive association between pesticides and child hood leukemia, but not solid tumors, possibly suggesting that the extent of recall bias by parents may be limited. The concordance of pesticide exposure among farmers, as meas ured via either selfreport or biomonitoring, was poor (Arbuckle et al. 2004;Perry et al. 2006). Recently, Ward et al. (2009) exam ined exposure to persistent organochlorine pesticides in residential carpet dust samples in the Northern California Childhood Leukemia Study. They observed no positive associa tions with childhood leukemia for chlordane, DDT (dichlorodiphenyltrichloroethane) or its metabolite DDE (dichlorodiphenyldichloro ethylene), methoxychlor, or pentachloro phenol concentrations.
Studies differed in the precise exposure time windows captured and reported. Some studies reported results only for all time win dows combined, which may obscure the potential association linked to specific exposure time windows; however, because high cor relations have been found between pesticide exposures in different exposure time windows (Alderton et al. 2006;Buckley et al. 1989), the extent to which such obfuscation might occur is difficult to determine. Sensitivity analyses that included studies reporting results in wide or illdefined exposure time windows tended to increase the degree of heterogeneity we observed, as quantified by the I 2 statistic. In models comparing pesticide exposures occur ring during pregnancy, in childhood, and in both pregnancy and childhood, Menegaux et al. (2006) observed the strongest associa tions with childhood leukemia when exposures were experienced during both exposure time windows, as opposed to during one exposure time window only.
Exposure to different types of pesticides may also be correlated, and few studies have attempted to disentangle the independent effects of specific pesticides. Menegaux et al. (2006) incorporated different insecticide exposures simultaneously and found that the Figure 3. Analysis of the association between childhood leukemia and exposure to (A) unspecified residential pesticides during childhood, (B) residential insecticides during childhood, and (C) residential herbicides during childhood. Squares indicating ORs from individual studies are proportional in size to the weight assigned to each estimate.  Buckley et al. 1989Infante-Rivard et al. 1999Kishi et al. 1993Dell 2004Davis 1991Leiss and Savitz 1995Ma et al. 2002Menegaux et al. 2006Steinbuch 1994Total Dell 2004Infante-Rivard et al. 1999Fajardo-Gutierrez et al. 1993Davis 1991Ma et al. 2002Leiss and Savitz 1995Menegaux et al. 2006 Total Davis 1991Ma et al. 2002Menegaux et al. 2006Infante-Rivard et al. 1999 Total volume 118 | number 1 | January 2010 • Environmental Health Perspectives positive associations remained. Lowengart et al. (1987) reported that the positive associa tions observed for parental exposure to either household pesticides or garden pesticides/ herbicides during pregnancy remained after mutual adjustment. Davis (1991) reported little change in pesticide ORs after adjustment for other pesticide use. However, Rudant et al. (2007) reported that associations with paternal pesticide use were confounded by maternal use. Residential pesticides also rep resent only one potential pathway through which parental or childhood pesticide expo sure may occur, with food, occupation, and the transport of agricultural pesticides repre senting other potentially important exposure pathways (Lu et al. 2008;National Research Council 1993;Reynolds et al. 2005;Ritz and Rull 2008;Ward et al. 2009). Among studies of residential pesticides that also col lected data on maternal occupational pesticide exposures, the prevalence of occupational pes ticide exposure was low, and there is little information on the potential interrelation ships of occupational and residential pesticide exposures for childhood leukemia. However, Rudant et al. (2007) reported that excluding children with occupationally exposed par ents did not change results, and Buckley et al. (1989) reported that the positive associations observed for residential pesticide exposure remained in multivariate models contain ing parental occupational pesticide exposure. Residential pesticides may be an important exposure source even in agricultural areas (Quandt et al. 2004).
In terms of other potential confound ing variables, as noted above, most studies examined or adjusted for at least a range of sociodemographic and maternal character istics. Leiss and Savitz (1995) also consid ered magnetic field exposure. Menegaux et al. (2006) examined early common infections, child care attendance, and residence near a gas station/garage as potential confound ers, with no change in results. Rudant et al. (2007) also reported that early infections and daycare attendance did not change results for residential pesticides. Lowengart et al. (1987) reported that the positive associations observed for both residential pesticide expo sure and paternal occupational exposure to chlorinated solvents remained after mutual adjustment. Among the studies that assessed the potential influence of maternal or child hood Xray exposures, Lowengart et al. (1987) and FajardoGutierrez et al. (1993) reported no change in findings; however, Dell (2004) reported that the positive association observed between pregnancy exposure to yard pesti cides and childhood leukemia disappeared after adjusting for maternal Xray exposure and use of antibiotics during pregnancy.
For childhood leukemia, the pregnancy exposure time window may be of particular importance (Belson et al. 2007). Most child hood leukemia cases occur in the first few years of life (Agha et al. 2006). Most child hood leukemia cases have gross chromosomal abnormalities, including translocations; how ever, little is known regarding their underlying cause (Wiemels 2008). A study of routinely collected blood samples in neonates revealed leukemia clones with specific chromosomal translocations in children who later developed ALL (Gale et al. 1997). Preleukemic clones may persist throughout childhood and may require postnatal exposures for leukemia pro gression (Maia et al. 2004). In a small study of infants born in an agricultural region in the Philippines, the prevalence of a common AML translocation [t(8;21)] in cord blood samples was about 2fold higher among those with detectable meconium levels of the methyl carbamate insecticide propoxur (Lafiura et al. 2007;Raimondi et al. 1999). The prenatal origin of AML may be less frequent than that of ALL (Burjanivova et al. 2006).

Conclusions
This systematic review and metaanalysis reveals positive associations between expo sure to residential pesticides in pregnancy and childhood and childhood leukemia, with the strongest associations observed for insecti cides. Further work is needed to confirm pre vious findings based on selfreport, to better describe potential exposure-response relation ships, to assess specific pesticides and toxi cologically related subgroups of pesticides in more detail, and to assess the potential role of preconceptional paternal exposures. Large prospective studies of children with biomoni toring data and discovery of biomarkers of past exposure (especially for rapidly excreted pesticides) would aid in this regard (Metayer and Buffler 2009). Additional studies are needed in order to better understand poten tial mechanisms of action and gene-pesticide interactions. In terms of precautionary public health implications, cosmetic pesticide bylaws implemented in various Canadian jurisdic tions typically do not address the use of pesti cides indoors or for essential purposes, such as to intervene in a health hazard or infestation to property. Further consideration of the need to reduce prenatal and childhood exposure to residential pesticides may be warranted.