Environment International 48 (2012) 109–120

Contents lists available at SciVerse ScienceDirect

Environment International

j ourna l homepage: www.e lsev ie r .com/ locate /env int

Environmental exposure to organophosphorus and pyrethroid pesticides in SouthAustralian preschool children: A cross sectional study

Kateryna Babina a,⁎,1, Maureen Dollard b, Louis Pilotto c, John W. Edwards a

a Health and Environment, School of the Environment, Flinders University, Australiab School of Psychology, Social Work and Social Policy, University of South Australia, Australiac Rural Clinical School, Faculty of Medicine, University of NSW, Australia

Abbreviations: SA, South Australia; OP, organophospMS–MS, high performance liquid chromatography tanddialkylphosphates; DMP, dimethyl phosphate; DEP, diethiophosphate; DMTP, dimethyl thiophosphate; DEDDMDTP, dimethyl dithiophosphate; TCPy, 3,5,6-trichltrans-3-(2,2-dichlorovinyl)-2,2-dimethyl-cyclopropane-3-(2,2-Dibromovinyl)-2,2-dimethyl-cyclopropane-1-car3phenylbenzoic acid; PBA, 3-phenoxybenzoic acid; Ointerval; SE, standard error of the mean; ADI, acceptableTotal Diet Survey; CV, coefficient of variation; LOD, limi⁎ Corresponding author at: Scientific Services Public

Rundle Mall, Adelaide SA 5000, Australia. Tel.:+61 8 822E-mail address: kateryna.babina@health.sa.gov.au (K

1 Please note, at the time of the study, Dr Babina wasand as such, her current affiliation with SA Health bearsstudy.

0160-4120/$ – see front matter © 2012 Elsevier Ltd. Allhttp://dx.doi.org/10.1016/j.envint.2012.07.007

a b s t r a c t

a r t i c l e i n f o

Article history:Received 20 February 2012Accepted 21 July 2012Available online 11 August 2012

Keywords:Environmental exposureOrganophosphatePyrethroidNeurotoxic insecticidesChildrenSouth Australia

Organophosphorus (OP) and pyrethroid (PYR) compounds are the most widely used insecticides. OPs andPYRs are developmental neurotoxicants. Understanding the extent of exposure in the general populationand especially in young children is important for the development of public health policy on regulationand use of these chemicals. Presented here are the results of the first investigation into the extent of environ-mental exposure to neurotoxic insecticides in preschool children in South Australia (SA).Children were enrolled from different areas of SA and assigned into urban, periurban and rural groups according totheir residential address. Residential proximity to agricultural activity, parental occupational contact to insecticidesand use of insecticideswithin the householdwere investigated as potential indirect measures of exposure.We usedliquid chromatography/tandem mass spectrometry to measure the following metabolites of OPs and PYRs inurine samples as direct indicators of exposure: dialkylphosphates, p-nitrophenol, 3-methyl-4-nitrophenol, 3,5,6-trichloro-2-pyridinol, cis- and trans-3-(2,2-dichlorovinyl)-2,2-dimethyl-cyclopropane-1-carboxylic acid,cis-3-(2,2-dibromovinyl)-2,2-dimethyl-cyclopropane-1-carboxylic acid, 2-methyl-3phenylbenzoic acid and3-phenoxybenzoic acid. Results were analysed to assess factors affecting the risk and level of exposure. Resultswere also compared to the published data in similar age groups from US and German studies.The results of this study demonstrate that there was widespread chronic exposure to OPs and and PYRs in SAchildren. OP metabolites were detected more commonly than PYR. Exposure to more than one chemical andcontemporaneous exposure to chemicals from both OP and PYR groups was common in the study population.There were some differences in risks and levels of exposure between the study groups. Exposure to some re-stricted use of chemicals, for example, fenitrothion, was higher in periurban and rural children. There was nodifference among the study groups in exposure to chlorpyrifos, used commonly in agriculture and in domesticsettings and most frequently found OP pesticide in food in Australia. South Australian children appear to havehigher levels of exposure compared their peers in US and Germany.

© 2012 Elsevier Ltd. All rights reserved.

horus; PYR, pyrethroid; HPLC/em mass spectrometry; DAP,thyl phosphate; DETP, diethylTP, diethyl dithiophosphate;oropyridinol; DCCA, cis- and1-carboxylic acid; DBCA, cisboxylic acid; MPA, 2-methyl-R, odds ratio; CI, confidencedaily intake; ATDS, Australiant of detection.Health SA Health, PO Box 6,6 7824; fax: +61 8 8226 7102.. Babina).not an employee of SA Healthno relevance to the presented

rights reserved.

1. Introduction

Insecticides are chemicals used to control insect pests that damage ordestroy crops or transmit diseases among humans and animals. Organo-phosphorus (OP) and pyrethroid (PYR) compounds are the most widelyused groups of insecticides in Australia and worldwide. OP and PYR in-secticides are neurotoxins. Neurotoxicity is defined as any permanentor reversible adverse effect on the structure or function of the nervoussystem. Themainmode of action of OPs in humans is inhibition of acetyl-cholinesterase, while the main mode of action of PYRs is modulation ofvoltage gated ion channels. Apart from main modes of action, bothgroups act on other biochemical and molecular targets within the ner-vous system. Extensive experimental data demonstrate that low-levelexposure to some commonly used OPs and PYRs can negatively affectthe development of nervous system in young laboratory animals via“non-classical” mechanisms that are independent of the main mode ofaction. International scientific literature shows that there is a concern

110 K. Babina et al. / Environment International 48 (2012) 109–120

and need for research into the effects of chronic low level (environmen-tal) exposures to neurotoxic insecticides on the neurobehavioral andemotional development of young children as well as on children's healthin general (Bjørling-Poulsen et al., 2008). Understanding the extent ofexposure in the general population is essential for evaluating whetherchemicals may be linked to adverse health outcomes and for the devel-opment of scientifically sound public health policy on regulation anduse of these chemicals.

South Australia has a well-developed agriculture sector, which in-cludes wide acreage crops, livestock and horticulture. An estimated54.1 million ha or 55% of South Australia's land area is used for agri-cultural activity and there were 14,262 agricultural establishmentsin the 2002–2003 financial years (ABS, 2003). In addition, ‘lifestylefarming’ on small acreage and backyard gardening are very popularin Australia.

The Australian Pesticide and Veterinary Medicine Authority(APVMA) is Australia's federal regulatory body whose roles are to re-view and allow (or disallow) chemicals onto the Australian market, todevelop regulatory residue limits and to make recommendations onthe use of registered chemicals, mostly by means of information col-lated on the chemical label. The APVMA maintains an online publiclyaccessible database of registered products, Public Chemical Registra-tion Information System (PUBCRIS) [http://www.apvma.gov.au/pubcris]. The state health authorities are responsible for trainingand licensing restricted-use chemical users. However, Australianauthorities neither have effective control over pesticide use beyondthe point of sale nor do they collect information on where, when,how or how much product is being used.

Overall, thirty OP actives and twenty two synthetic PYR actives areincluded in over 300 and over 1000 of registered commercial formu-lations respectively (APVMA, 2012). Six OP actives and sixteen PYRactives are allowed for public use in Australia. Publicly accessible OPactives include: chlorpyrifos, diazinon, dichlorvos, fenthion, maldisonand omethoate (APVMA, 2012). The OP and PYR formulations areavailable for domestic use by the general public as indoor surfacesprays (fly, spider, cockroach), outdoor plant and barrier sprays, in-door and outdoor powder preparations, concentrate solutions, insecttraps, flea bombs, mosquito coils, lice control lotions and shampoos aswell as veterinary products. These formulations can be purchasedover the counter in hardware stores, supermarkets and pharmaciesin Australia. In addition, Australian homes are routinely treated withvarious insecticides including chlorpyrifos (an OP insecticide) andbifenthrin (a PYR insecticide) for termite control, both before andafter construction.

This widespread availability of pesticides to the general public,large variety of registered products (both for public and restricteduse) and extensive mixed agricultural land use pattern of SouthAustralia suggests that there is a potential for significant widespreadexposure in the general population. The extent of environmental ex-posure to OPs and PYRs in the Australian general population and itspossible health effects so far have not been addressed in the scientificliterature and in particular the extent of environmental exposure inchildren in Australia is unknown.

This study aimed to explore the extent of exposure to OP and PYRspesticides in South Australian children and to investigate the factorsthat influence the risk of being exposed and the levels of exposurein the study population.

2. Study design and methods

2.1. Subjects

The study took place during 2003–2006. The subjects of this studywere 340 healthy children aged 2.5 to 6 years old residing in metro-politan Adelaide (urban group), Adelaide Hills area (peri-urbangroup) and agricultural areas of South Australia including the Yorke

Peninsula, mid-north region, Barossa Valley, Riverland, McLarenVale and the Murray-Hills area (rural group) (Fig. 1). Study groupswere chosen based on a priori descriptive knowledge of pesticideuse patterns and the knowledge of land use and agricultural activityin each area. Differences in population density meant that largerareas needed to be covered to achieve the target sample size inrural areas. The description of pesticide use and agricultural activitypatterns in each sample group's area of residence are summarisedin Table 1.

The study employed stratified random sampling approach. Chil-dren were recruited through local preschools (child care centres, kin-dergartens and reception year in primary schools). The contact detailsof the preschools within areas of interest were taken from the “Yel-low Pages” phone directory. In peri-urban and rural areas all pre-schools listed in the phone directory were contacted, while urbanpreschools were chosen within a sample of 30 postcodes drawn inurban Adelaide. The urban postcode sample was drawn to representthe diversity of the socio-economic status (SES) of the urban Adelaidepopulation based on SES profiles by Glover and Tennant (1999). Atthe time of initial contact, brief introduction to the project and sam-pling process were given to the preschool directors/principals. A pro-portion of preschools did not consent to participate in the study. Theconsented preschools were then asked to distribute the project mate-rials among the parents of eligible children. The parents were givenan opportunity to meet/contact the project officer to discuss anyquestions or concerns.

The sampling was conducted through early September to lateNovember in 2003 and in 2004. Sampling at the same time of the yearallowed accounting for possible seasonal variations in pesticide use.

Approvals were gained from human research ethics committees(including Flinders University, University of South Australia and theSouth Australian Department of Education).

2.2. Sample size estimation

An earlier study by Loewenherz et al. (1997) compared urinaryDAP levels in children from pesticide applicator families and from ref-erence families in a rural community in the USA. The difference inmeasured concentrations between the two groups reached up to 4fold for one of the metabolites, DMTP. For the purpose of the samplesize calculation for this study, the target difference of 50% between thegroups was assumed with standard deviation of 1.0 (log-transformedconcentrations). A similar approach has been used in other studies(Wilson et al., 2004). To detect the difference in exposure levels be-tween the sample groups with at least 80% power (α=0.05, 20% prob-ability of Type II error, and 5% probability of Type I error) (Cohen, 1988),a sample size of at least 100 subjects in each sample group (urban,peri-urban and rural) would need to be achieved.

2.3. Indirect measures of exposure

Parents were asked to complete a General Questionnaire (GQ)that was designed to acquire information about pesticide use insideand outside each child's residence, parental occupational contactwith pesticides and parental smoking habits. The GQ provided the in-formation that formed the basis of several indirect measures of expo-sure, the effects of which were tested in the statistical analysis. TheGQ also provided information on the both mother's and father's occu-pation, which was used to estimate combined family income based onthe ANU4 scales (Jones and McMillan, 2001). The subjects wereassigned into “high income”, “medium income” and “low income”categories based on commonly used income classification approach:the 25th percentile of the population distribution in income com-prises low income subjects, while 75th percentile consists of highincome subjects (ABS, 2004). Proximity of every child's residenceto agricultural activity was measured using the web-based South

Fig. 1. Map of study groups and sample areas.

111K. Babina et al. / Environment International 48 (2012) 109–120

Australian Government's Report on Generalized Land Use for 2003(Atlas SA, 2004). This allowed the location of the residence to be classi-fied aswithin 50 m,within 200 m, or beyond 200 m from agriculturally

Table 1Description of pesticide use patterns by study group.

Study groupand location

Pesticide use patterns Examples of chemicals used

Urban groupUrbanAdelaide

• Indoor/outdoor pest (ant,spider, fly, cockroach) controlby the household members

• Pre- and post constructiontermite control by thepest control operators

• Head lice control• Veterinary products

• Indoor surface sprays,garden sprays, c (PYRs,OPs, carbamates,herbicides)

• Termite control(chlorpyrifos, bifenthrin,deltamethrin and others)

• Shampoos and lotions(malathion, PYRs)

• Pet shampoos, collars(OPs and PYRs)

Peri-urban groupAdelaide Hills(Mount LoftyWatershed)

• As in urban Adelaide, plus:• High density of agriculturalactivity within populatedresidential area

• “Lifestyle” and small scalefarming (orchards, vineyards,sheep and dairy farms)

• Large number of residents arelicensed to use restricted usechemicals

• Rugged terrain makes aerialspray common, especiallyfor pastures

• As in urban Adelaide• Anecdotal evidence ofwide spread use andoveruse of generallyavailable and restricteduse chemicals

Rural groupRegional SA

• As in urban Adelaide, plus:• Rural townships and farmresidences surroundedby large agriculturalestablishments

• Wide acreage crops (barley,wheat), pastures, orchards,vineries

• As in urban Adelaide• Use of restricted chemicals• Aerial spray on thepastures and fields

zoned areas. These distances were chosen mostly intuitively, since thisstudy was the first of a kind in Australia and there were no preliminarydata upon which the decisions of grouping by distance could be based.

2.4. Direct measures of exposure: Urinary metabolites

The extent of human exposure to chemicals may be characterisedusing biological monitoring—measurement of chemicals or their me-tabolites in body fluids or tissues, such as blood and urine. Biologicalmonitoring of exposure to neurotoxic insecticides has been used ex-tensively in occupational health in Australia and worldwide. Mea-surement of OP and PYR insecticides and/or their metabolites inurine samples is a widely accepted tool in occupational and environ-mental exposure assessment (Aprea et al., 1994; Heudorf et al., 2004;NCEH, 2009; Smith et al., 2002).

The parents of the study participants were asked to collect a singlefirstmorning void urine sample from their child. Urine sampleswere in-dividually coded and stored frozen (−20 °C) until analysis.

The urine samples were analysed for the following metabolites ofOP insecticides: dialkylphosphates (DAPs), p-nitrophenol, 3-methyl-4-nitrophenol, 3,5,6-trichloro-2-pyridinol (TCPy) and followingmetabolitesof PYR insecticides: cis- and trans- 3-(2,2-dichlorovinyl)-2,2-dimethyl-cyclopropane-1-carboxylic acid (DCCA), cis- 3-(2,2-dibromovinyl)-2,2-dimethyl-cyclopropane-1-carboxylic acid (DBCA), 2-methyl-3phenylbenzoic acid (MPA) and 3-phenoxybenzoic acid (3-PBA). Thesemetabolites have been used for the purposes of biological monitoringof exposure in a number of studies (Aprea et al., 2000; Barr et al.,2002; Esteban et al., 1996; Heudorf et al., 2004; Morgan et al., 2005;NCEH, 2009; Saieva et al., 2004; Smith et al., 2002). The metaboliteswere chosen based on their informative value and availability of thechemical standards. Table 2 summarises information on themetabolitesmeasured in this study and their respective parent compounds.

While a number of gas chromatography coupled with mass spec-trometry (GC/MS) methods have been developed and are thoroughlyvalidated, recent developments in liquid chromatography coupledwith mass spectrometry (LC/MS) allowed the development of a new

112 K. Babina et al. / Environment International 48 (2012) 109–120

generation of methods. A number of new methods, which employ theuse of LC combined with tandem mass spectrometry (LC/MS–MS)have been recently reported in the literature (Baker et al., 2004;Hernandez et al., 2002; Olsson et al., 2004; Sancho et al., 2000). OPand PYR metabolites are polar non-volatile compounds, which re-quires them to be derivatised into volatile compounds for analysisby GC/MS using complex chemical processes (Hardt and Angerer,2000; Oglobline et al., 2001). At the same time, polar non-volatilecompounds can be easily detected by LC/MS–MS without the needfor extensive sample preparation (Hernandez et al., 2005). Limits ofdetection (LODs) for LC/MS–MSmethods are less than 1 μg/L, makingthem applicable for monitoring of low-level environmental exposurein large population studies (Baker et al., 2004).

A Waters Alliance 2695 HPLC system (Waters, USA) interfacedwith a Micromass Quattro tandem mass spectrometer (Micromass,UK) was used for the urine sample analysis in this study.

2.4.1. Method 1—DAPsChemical standards of dimethyl phosphate (DMP), diethyl phosphate

(DEP), diethyl thiophosphate (DETP), dimethyl thiophosphate (DMTP),diethyl dithiophosphate (DEDTP) and dimethyl dithiophosphate(DMDTP) were kindly donated by WorkCover Laboratories, Sydney,Australia. HPLC grade acetonitrile and methanol were purchasedfrom BDH Laboratory Supplies (UK). Pairing reagent, tetraethyl ammo-nium acetate (TEA) was purchased from SigmaAldrich (USA). Nitrogenand argon were purchased from BOC Gases (Adelaide, South Australia).All chemicals were of an analytical grade.

Stock solutions of all six DAP standards were prepared in 100%acetonitrile at concentration 1 g/L. Mixed standard solution (MSS)was prepared in 50% methanol with resulting concentration of10 mg/L for each of the compounds. MSS was then used to preparecalibration dilutions of DAPs in blank urine at the following levels: 5,10, 25, 50, 75, 100, 125 μg/L. Blank urine was collected from unexposedindividuals and was heat treated (60 °C for 30 min) in order to mini-mise amount of DAPs potentially present in blank urine. External

Table 2Urinary metabolites measured in the study and respective parent compounds.

Metabolite Parent compound—use andavailability in Australia

Dialkylphosphates (DAPs):

• Dimethyl phosphate (DMP)*• Diethyl phosphate (DEP)• Diethyl thiophosphate (DETP)• Dimethyl thiophosphate (DMTP)• Diethyl dithiophosphate (DEDTP)• Dimethyl dithiophosphate (DMDTP)*

Non-specific group of metabolitesindicative of exposure tomost OP insecticides.*Due to the analytical methodlimitations these DAPswere not measured.

p-Nitrophenol Parathion, parathion-methyl—formally restricted to licenseduse, currently phased out.

3,5,6-Trichloropyridinol (TCPy) Chlorpyrifos—most commonlyused OP insecticide availableto public in Australia;Chlorpyrifos-methyl—restricted use.

3-Methyl-4-nitrophenol Fenithrothion—mostly restricted use.cis- and trans-3-(2,2-dichlorovinyl)-2,2-dimethyl-cyclopropane-1-carboxylic acid (DCCA)

Permethrin, cypermethrin andcyfluthrin—common PYR insecticidesavailable to the general publicin Australia.

cis 3-(2,2-Dibromovinyl)-2,2-dimethyl-cyclopropane-1-carboxylic acid (DBCA)

Deltamethrin—common PYRinsecticides available to thegeneral public in Australia.

2-Methyl-3phenylbenzoic acid (MPA) Bifenthrin—common PYRinsecticides available to thegeneral public in Australia.

3-Phenoxybenzoic acid (PBA) Non-specific PYR metaboliteindicative of exposure tomany PYR insecticides.

standard method was used for compound detection and quantification.The calibration dilutions of DAPs prepared in blank urine at the levels: 0,5, 10, 25, 50, 75, 100, 125 μg/Lwere run prior to each run in order to ob-tain calibration curves. Linear calibration curves (R2 ranging between0.97 and 0.99) were obtained for each run.

LC analysis employed Wakosil II 5C18RS 5 μm×50×2.0 mm ana-lytical column coupled with Wakosil C18RS 5 μm guard cartridge.Both were purchased from SGE (SGE International Pty Ltd).

Urine samples and standards were thawed at room temperature,spun at 2500 rpm for 10 min, then 1 mL of samples/standard wastransferred into a vial, already containing 50 μL TEA. 10 μL was direct-ly injected into LC system. Water/methanol mobile phase was usedfor gradient elution (0% methanol between 0 and 2 min, then increas-ing to 10% methanol by 7 min and to 100% methanol by 9 min, hold-ing at 100% methanol until 10 min, then decreasing to 0% methanolby 12 min and holding at 0% methanol until 17 min to allow for col-umn equilibration). Compounds retention times were between 1.5and 4 min, making this method fast and convenient for analysis of alarge number of samples.

The mass spectrometer was used in negative ionisation mode(ES-) in 8 channel multiple reactionmonitoring (MRM)method. A se-ries of infusion experiments were run for each compound in order todevise MS–MS parameters optimal for both specificity and selectivity.Two transitions were selected for each compound (with the excep-tion of DEDTP). Nitrogen was used as a desolvation gas and argonwas used as a collision gas. Desolvation gas flow rate was set at800 L/h; collision cell pressure was maintained at 4–4.5 e−4mBar.

Only four of six DAPs gave reliable retention, elution and signal.DMP was detected as an unretained compound, thus quantificationof DMPwas not possible. DEDTP coeluted with unidentified urine ma-trix interference and was excluded from the final method. The quan-tification of only four out of six DAPs still provided informativeassessment of the general exposure to a wide range of OPs in thefocus population. It would not though be sufficient if the assessmentof exposure to particular OP compounds were the focus of the study.

To determine within-series variability of the method, a calibrationstandard (50 or 100 μg/L) was run after every 10 injections. Within-series coefficient of variation (CV) ranged from 4.3 to 10.4%. Between-day variability was determined by running duplicates of samplesanalysed the previous day. Between-day CV ranged between 6.2 and12.7%. The overall mean recovery ranged between 89 and 103 (%±SD). The overall method limit of detection was 0.1 μg/L.

2.4.2. Method 2—PYRs and specific OP metabolitesHPLC grade acetonitrile and methanol were purchased from BDH

Laboratory Supplies (UK). Glacial acetic acid was purchased fromSigmaAldrich (USA). Chemical standards of cis- and trans-DCCA, DBCA,3-PBA, 3-PBA-13C6 where purchased from AccuStandard (USA). Chemi-cal standard of MPA was synthesised at the Environmental HealthLaboratories, Flinders University, South Australia (Smith et al., 2002).Chemical standards of TCPy and p-nitrophenol were purchased fromSigmaAldrich (USA). Nitrogen and argon were purchased from BOCGases (Adelaide, South Australia). All chemicals were of an analyticalgrade.

The analysis employed a Wakosil II 5C18RS 5 μm×50×2.0 mmanalytical column coupled with Wakosil C18RS 5 μm guard cartridge.Both were purchased from SGE (SGE International Pty Ltd).

Sample preparation procedure was conducted as described inOlsson et al. (2004). In brief, urine samples were thawed at roomtemperature, 2 mL of sample was used in the analysis. Availableinternal standard (3-PBA 16C6) was added to achieve resulting concen-tration of 50 ppb. To release the glucuronide and/or sulphate conjugat-ed compounds, β-glucuronidase type H-1 (SigmaAldrich, USA) withspecific activity of ~500 U/mg was used. To each sample, an amountof enzyme giving ~800 U of activity was added in 0.2 M acetate buffer

113K. Babina et al. / Environment International 48 (2012) 109–120

(3.1 mL of glacial acetic acid, 9.7 g of sodium acetate, 1 L of water).Samples were incubated at 37 °C overnight (Olsson et al., 2004).

Sample extraction was carried out next day using solid-phase ex-traction (SPE) procedure. SPE cartridges containing 50 mg of C18media (United Chemical Technologies) were preconditioned using1 mL methanol and then 1 mL of 1% acetic acid. Samples were filteredthrough cotton wool and then passed through the SPE cartridge. Toreduce interferences, cartridges were washed with 5% methanol in1% acetic acid (Olsson et al., 2004) and the compounds of interestwere eluted with 2 mL 100% methanol. 1.5 mL of the eluent was trans-ferred into LC vial and 10 μL of that was injected into the LC/MS–MSanalytical system.

Stock solutions of analytical standards were prepared at concentra-tion of 500 μg/L inmethanol. Standard curve solutionswere prepared inblank urine at concentrations: 5, 15, 30, 45, 60, 75, and 100 μg/L. Thecalibration solutions were subjected to the same sample preparationas the test samples. The blank urine only had 1.5 mL of 0.2 M aceticbuffer added to it (without β-glucuronidase) before the overnight in-cubation. Addition of β-glucuronidase would release compounds po-tentially present in the blank urine and therefore introduce additionalinterference.

Calibration solutions were run at the beginning of each run andresulting calibration curve (linear, R2 ranged between 0.93 and 0.98)was used to estimate the concentrations of the compounds of interestin urine samples from study participants.

Mobile phase composed of water and acetonitrile was used (20%acetonitrile between 0 and 0.5 min, then increasing to 100% acetoni-trile by 5 min holding at 100% acetonitrile until 10 min, then decreas-ing to 20% acetonitrile by 12 min and holding at 20% acetonitrile until19 min to allow for column equilibration). Samples and column werekept at room temperature. Injection volume was 10 μL. The flow ratewas 0.2 mL/min.

Mass spectrometer was used in negative ionisation mode (ES-) in13-channel (MRM) method. Nitrogen was used as a desolvation gasand argon was used as a collision gas. Desolvation gas flow rate was setat 800 L/h; collision cell pressure was maintained at 4–4.5 e−4mBar. Aseries of infusion experiments were run for each compound in order todevise MS–MS parameters optimal for both specificity and selectivity.

To determine within-series variability of the method, a calibrationstandard (15 or 30 μg/L) was run after every 10 injections. Within-series CV ranged from 4.3 to 9.8%. Between-day variability was deter-mined by running duplicates of samples analysed the previous day.Between-day CV ranged between 6.4 and 12.1%. The overall mean recov-ery ranged between 98 and 113 (%±SD). The method limit of detectionwas 0.1 μg/L.

2.4.3. Creatinine correctionA spot sample urinary concentration of pesticide metabolite in the

units of μg/L has uncertain meaning because variable fluid intakesmay result in large variations in the excreted concentrations. Severalmethods are available for correction for urine dilution of spot samplesincluding specific gravity, timed or daily urinary excretion, and creati-nine correction. The creatinine correction approach involves calculatingthe ratio of analyte to creatinine and the resulting concentration isreported as microgram of analyte per gram of creatinine (μg/g creati-nine). Creatinine correction has been used extensively in biomonitoringstudies in both adult and children populations (Aprea et al., 2000;MacIntosh et al., 1999; NCEH, 2009). The urinary creatinine was mea-sured by Jaffe kinetic reaction (Yatzidis, 1974) without deproteinisationon a Boehringer Mannheim Hitachi 917 automated analyser.

2.5. Statistical analysis

Statistical analysis was performed using SPSS statistical softwarepackage version 12.0.1 (SPSS Inc., Chicago, IL).

2.5.1. Data screening and diagnosticsInitially, data frequencies were screened for logical inconsistencies

that may have been the result of erroneous handling of the data. Next,detection of outliers was conducted using the scatter, stem-and-leaf,and error bar plots (Gilbert, 1987). The outliers were investigated inthe regression diagnostics and characterised by centred leverage(cutoff was set at 2 k/N, where k=sample groups number=3), dis-crepancy (measured by externally studentised residuals, cutoff wasset at 4.0) and influence (measured by Cook's distance, cutoff wasset at 1.0) (Cohen et al., 2003). Outliers were detected in this studycomprised less than 5% of the study population and, based on the re-gression diagnostics, they were deemed conceptually and statisticallytolerable (Cohen et al., 2003) and therefore remained unchangedwithin the dataset.

The interactions among predictor variables were monitored foreach metabolite prior to the development of multivariate regressionmodels in stepwise univariate regression analyses. The dependentvariables (metabolite concentrations) were analysed against each in-dependent variable (indirect exposuremeasures) and then new predic-tor variables were added to the model in succession while monitoringfor large unexplainable changes in the value and/or direction of the co-efficients which would suggest serious influence of collinearity (Cohenet al., 2003). There were no signs of unacceptable multicollinearityamong independent variables in our dataset.

2.5.2. Data transformationThe creatinine corrected concentrations of urinary metabolites

(measured in μg/g creatinine) were sorted into two types of variablesto reflect different aspects of exposure: risk of exposure and level ofexposure. Firstly, binary variables were created for each metabolite,allowing assignment of the study participants into either “exposed” or“non-exposed” groups: subjectswhose urine samples containedmetab-olites at levels “below LOD” levels were assigned into “non-exposed”group, while subjects whose urine samples contained “above LOD”levels were assigned into “exposed” group. This approach allowed theanalysis of the factors affecting the risk of being exposed within thewhole study population. The shape of the raw concentrations of urinarymetabolites (measured in μg/g creatinine) was departing significantlyfrom normal distribution (skewness ranging from 2 to 10). The raw datawas therefore logarithmically transformed (log-transformed) to ensureits normality. The log-transformed data was normally distributed(maximum skewness−0.6 for DETP and DMDTP). These log-transformedmetabolite concentrations formed the second type of variables thatallowed investigation of the factors affecting the levels of exposureamong exposed study participants.

After initial descriptive analysis, a series of simple bivariate corre-lations were performed to explore the associative relationshipsamong the variables in the dataset. The final step in the data analysiswas either logistic (binary data) or linear (normally distributed con-tinuous data) regression analysis.

2.5.3. Binary logistic regression analysis: Exploring factors predicting riskof exposure

To explore which of the indirect exposure measures are associatedwith the risk of exposure, binary logistic regression analysis wasperformed. The independent variables included “Area of residence”(rural, peri-urban, urban); “Type of residence” (a house inmetropolitanAdelaide, a house in rural town, a farm, a lifestyle farm in peri-urbanarea); “Parental occupational exposure” and “Residing within 50 mfrom an agricultural activity” and “Residing within 200 m from an agri-cultural activity”. The other variables used in the analysis were “Child'sage” and “Family income”. The results are reported here as an odds ratio(OR) together with corresponding 95% confidence intervals (95%CI)and significance level (p) in Table 5.

114 K. Babina et al. / Environment International 48 (2012) 109–120

2.5.4. Linear regression analysis: Exploring factors predicting levels ofexposure

A series of stepwise linear regressions were run to explore the ef-fects of indirect measures of exposure as independent (predictor)variables on the level of exposure among exposed subjects only esti-mated using log‐transformed creatinine corrected urinary metaboliteconcentrations as dependent variables. The analysis was performedseparately for each metabolite. The independent variables included“Area of residence” (rural, peri-urban, urban); “Type of residence”(a house in metropolitan Adelaide, a house in rural town, a farm, alifestyle farm in peri-urban area); “Parental occupational exposure”and “Residing within 50 m from an agricultural activity” and “Resid-ing within 200 m from an agricultural activity”. The other variablesused in the analysis were “Child's age”, “Child's gender” and “Familyincome”. The results of this analysis, including model summary(standardised β and t statistics) and significance levels (p) are presentedin Table 6.

3. Results

3.1. Study population: Descriptive analysis

The samples were collected from various areas of metropolitanAdelaide, Adelaide Hills and rural areas, aiming to include familiesfrom different socio-economic backgrounds ensuring that the finalsample population reflected the socio-economic gradients existingin the general population in South Australia. A total of 340 children(115 urban, 111 peri-urban, 114 rural) participated in the study. Ofthose preschools that agreed to participate, the response (participa-tion) rate among eligible families ranged between 1% and 40%, withan overall participation rate of 12.1%. Highest response rates were inrural areas and lowest response rates were in some less advantagedSouthern suburbs of metropolitan Adelaide. The gender compositionof the final sample population was comparable to the gender composi-tion of the overall SA population of pre-schoolers (about 55% boys, esti-mated using 2000 SA Demographic (ABS, 2000)). Overall, the resultingsample of 340 children represented 1.41% of the total population of3–6 year olds at the time of the study, satisfied estimated sample sizerequirement and was deemed representative of the general SA

Table 3Study population: descriptive statistics by sample group.

All samples Area of residence p

Urban Peri urban Rural

Number ofsubjects

340 115 111 114

Mean child'sage±SD

4.77(±0.915)

4.48(±0.685)

5.08(±1.04)

4.76(±0.898)

b0.001⁎a

0.033⁎b

Gender % boys 56.5 50.4 57.7 61.4 0.236⁎⁎

% girls 43.5 49.6 42.3 38.6Mean familyincome($×103)±SD

74.61(±46.033)

90.57(±54.06)

75.16(±40.65)

57.833(±35.36)

0.017⁎a

0.000⁎b

% reportedparentaloccupationalcontact withpesticides

29.4 17.4 18.0 52.6 b0.001⁎⁎

% have someagriculturalactivity within50 m ofresidence

36.8 0.9 40.5 69.3 b0.001⁎⁎

⁎a ANOVA, urban—peri urban contrast.⁎b ANOVA, urban—rural contrast.⁎⁎ Kruskal–Wallis test.

population aged 3 to 6 years. The demographic characteristics and in-direct exposure parameters of the total sample and sample groupsare summarised in Table 3.

3.2. Indirect exposure measures—descriptive analysis

Significantly more rural parents (52.6%, pb0.001) reported occupa-tional contact with pesticides than those from urban and peri-urbangroups. Similarly, significantly more children were residing in closeproximity to agricultural activity in peri-urban (40.5%) and rural(69.3%) groups compared to the urban group (0.9%, pb0.001). Meancombined family income was significantly lower in peri-urban andrural groups compared to the urban group (p 0.017 and b0.001 respec-tively) (Table 3). The differences in income and age distribution wereaddressed in further analysis by assessing variables “Family income”and “Child's age” for confounding effect.

Some variables initially intended to be assessed for use as indirectmeasures of exposure were excluded from the analysis. Many parentsmisreported the use of pesticides inside homes due to the misinter-pretation of the term “pesticide”. For example, some parents reporteduse of domestic disinfectants, dishwashing detergents and even airfresheners in question 5 of the GQ (“Do you/your partner use pesti-cides inside the house?”). Therefore the reported use of pesticidescould not be considered to be a true representation of the domesticuse of pesticides in the community and variable “Reported use of pes-ticides” was excluded from further analysis. Similarly, most parents(over 80%) could not recall whether or not their house had been pro-fessionally treated for pest control, which led to the exclusion of thevariable “House treated professionally” from further analysis.

3.3. Urinary metabolites data: Descriptive analysis

Tables 4 and 5 show the mean concentrations with standard devia-tions, range and major percentiles: 25th (P25), 50th (P50), 75th (P75)and 95th (P95) for each measured metabolite by sample group. Mostcommonly detected metabolites were TCPy (detected in between 92.2and 98.2%), 3PBA (detected in 80.9 to 84.2% of samples) and DAPs(detected in 53.0 to 82.6% of samples). Highest concentrations detectedwere for DMTP (periurban group, range bLOD–1615 μg/g creatinineand P95–697.9 μg/g creatinine), DEP (periurban group, range bLOD–1177 μg/g creatinine and P95–386.5 μg/g creatinine) and DETP (ruralgroup, range bLOD–365.7 μg/g creatinine and P95–141.5 μg/g creati-nine). The least frequently detected metabolite was DBCA (detected inbetween 10.8 and 14.9% of samples). There were differences in thelevels of metabolites detected among the groups (Figs. 2–4), whichwere further explored in regression analyses.

3.4. Binary logistic regression analysis: Exploring factors predicting riskof exposure

Therewere no significant differences in the risk of exposure formostmetabolites. Significantly higher risks of exposure to fenitrothion,as measured by urinary 3-methyl-4-nitrophenol, were detected forperiurban (OR 2.228, 95%CI 1.214–4.09 and p 0.01) and rural (OR2.648, 95%CI 1.426–4.917 and p 0.002) groups as compared to urbangroup. Similarly, farm children were more likely to be exposed tofenitrothion (OR 2.226, 95%CI 1.131–4.381 and p 0.02) than children liv-ing in metropolitan Adelaide. At the same time, children living in ruraltownsweremore likely than children living in urban Adelaide to be ex-posed to deltamethrin based on the detection of deltamethrin specificmetabolite, DBCA (OR 2.097, 95%CI 1.046–4.203 and p 0.037).

Risk of exposure to some OP insecticides (based on the detectionof DEP and DMTP) was higher in children whose parents reportedcontacting pesticides at work (OR 2.839, 95%CI 1.155–6.976 andp 0.023 for DEP and OR 2.403, 95%CI 1.328–4.346, p 0.004 forDMTP). Risk of exposure to the methyl-containing group of OP

Table 4Summary statistics: OP metabolites, mean concentrations (μg/g creatinine) by sample group.

Metabolite Group N % samples with detectable levels Mean±SD Range P25 P50 P75 P95

p-Nitrophenol Urban 115 70.4 5.1±9.5 bLOD–64.8 1.2 2.3 4.3 24.3Periurban 111 69.4 9.5±18.4 bLOD–100.9 1.2 3.2 7.0 50.1Rural 114 67.5 9.1±19.9 bLOD–115.1 1.3 2.9 7.0 50.4

3-Methyl-p-nitrophenol Urban 115 61.7 9.2±21.3 bLOD–139.4 0.7 1.8 4.4 48.1Periurban 111 78.4 19.0±41.9 bLOD–241.0 0.8 5.5 13.9 98.1Rural 114 82.5 17.9±30.7 bLOD–188.1 1.3 6.0 21.4 92.5

TCPy Urban 115 92.2 21.5±28.8 bLOD–217.9 6.9 12.5 24.8 71.1Periurban 111 97.3 27.1±54.9 bLOD–539.8 6.5 13.8 29.9 84.9Rural 114 98.2 16.3±18.3 bLOD–153.9 5.8 12.5 20.5 43.7

DEP Urban 115 82.6 7.4±9.5 bLOD–55.7 1.6 4.1 8.4 29.2Periurban 111 78.4 84.1±175.3 bLOD–1177.1 4.6 15.4 77.7 386.5Rural 114 82.5 79.2±151.8 bLOD–880.9 11.0 24.7 76.7 486.9

DETP Urban 115 71.3 8.3±23.5 bLOD–195.1 1.3 2.7 5.9 30.5Periurban 111 77.5 24.7±49.7 bLOD–309.9 2.8 7.9 25.0 114.0Rural 114 78.1 27.9±48.8 bLOD–365.7 4.8 13.8 27.0 141.5

DMTP Urban 115 53.0 20.2±48.2 bLOD–346.5 1.8 6.8 18.4 79.7Periurban 111 69.4 135.9±259.2 bLOD–1615.0 5.5 24.8 152.9 697.9Rural 114 74.6 100.8±162.4 bLOD–981.3 12.3 37.3 123.3 466.5

DMDTP Urban 115 77.4 12.4±14.9 bLOD–75.5 3.3 7.2 14.5 50.0Periurban 111 73.9 15.7±21.2 bLOD–103.5 3.5 6.9 18.5 73.7Rural 114 72.8 20.0±36.2 bLOD–255.4 2.5 7.2 21.4 96.2

115K. Babina et al. / Environment International 48 (2012) 109–120

insecticides (based on the detection of DMTP in urine samples) washigher in peri-urban (OR 2.459, 95%CI 1.376–4.393, p 0.002) andrural children (OR 3.484, 95%CI 1.885–6.437, pb0.001).

3.5. Linear regression analysis: Exploring factors predicting levels ofexposure

Compared to the urban group, the periurban group had higherlevels of urinary p-nitrophenol (β 0.214, p 0.046), 3-methyl-4-nitrophenol (β 0.242, p 0.028), 3-PBA (β 0.267, p 0.006), DEP (β 0.320,p 0.001) and DMTP (β 0.259, p 0.018). At the same time, compared tourban group, rural group had higher levels of 3-methyl-4-itrophenol(β 0.319, p 0.02), 3-PBA (β 0.479, pb0.001), DEP (β 0.390, p 0.001)and DMTP (β 0.277, p 0.04). Residing within 50 m or less from an agri-cultural activity was significantly associated with higher levels of uri-nary DCCA (β 0.381, p 0.002). Parental occupational exposure wassignificantly associated with higher levels of urinary DEP (β 0.141,p 0.021) and DMTP (β 0.209, p 0.003). Residing on a farm was onlymarginally significantly associated with higher levels of urinary DETP(β 0.223, p 0.05).

4. Discussion

Since the human body metabolises OP and PYR insecticides rapid-ly, urinary metabolites only reflect recent exposure. However, the de-tection of short lived urinary metabolites in a significant proportion ofurine samples from a large group of subjects in a cross-sectional study

Table 5Summary statistics: PYR metabolites, mean concentrations (μg/g creatinine) by sample gro

Metabolite Group N % samples with detectable levels

DCCA (cis-+ trans-) Urban 115 35.7Periurban 111 36.0Rural 114 49.1

DBCA Urban 115 13.9Periurban 111 10.8Rural 114 14.9

MPA Urban 115 25.2Periurban 111 28.8Rural 114 25.4

3-PBA Urban 115 80.9Periurban 111 86.5Rural 114 84.2

would be indicative of ongoing (chronic) exposure in the population.As such, the results of this study demonstrate that there is widespreadchronic exposure to OPs and PYRs in South Australian children. Fur-thermore, exposure to more than one chemical and contemporane-ous exposure to chemicals from both OP and PYR groups is common.

We did not aim to conduct an in depth investigation of the expo-sure pathways in the study population. We did, however, aim to ex-plore some of the potentially most significant pathways. The factthat we had to exclude the only variable that was directly reflectiveof the domestic use of pesticides due to common misinterpretationof the question ‘Do you use pesticides at home?’ is a noteworthy lim-itation in our study. This misinterpretation is still meaningful in that itdemonstrates very low level of ‘chemical literacy’ among the generalpopulation in South Australia. Combined with the significant numberof OP and PYR products being available to the public, this finding war-rants further investigation. Interestingly, the question about occupa-tional contact to pesticides did not cause as much confusion, perhapsdue to the higher awareness of work-related hazards, which stemsfrom occupational health and safety education.

The differences in frequency of detection (reflecting risk of expo-sure) and levels (reflecting levels of exposure) of urinary metabolitesamong the groups were complex, but seemed to be reflective of theland use patterns and chemical availability. There is no way for us toknow what chemicals were used on what type of crops in which areasat the timeof the study. Australian authorities do not have effective con-trols over chemical use beyond the point of sale and there is no system-atic data collection on chemical use in Australia.

up.

Mean±SD Range P25 P50 P75 P95

8.1±22.0 bLOD–137.7 1.5 2.9 4.8 43.29.0±21.3 bLOD–134.6 1.8 4.6 7.8 30.65.6±6.5 bLOD–32.8 2.4 3.6 5.4 24.03.0±2.5 bLOD–8.4 1.1 2.0 5.1 –

4.7±4.5 bLOD–16.5 2.3 2.9 5.4 –

5.2±4.9 bLOD–17.2 1.9 3.7 7.2 –

1.9±3.0 bLOD–16.4 0.7 1.1 1.8 11.23.7±7.0 bLOD–34.7 1.1 1.6 2.6 27.61.1±0.7 bLOD–3.4 0.6 1.1 1.5 2.91.2±3.2 bLOD–30.0 0.2 0.5 1.1 3.01.4±2.6 bLOD–18.6 0.3 0.8 1.3 6.91.6±1.7 bLOD–12.9 0.6 1.1 1.7 4.8

Table 6Risk of exposure: results of the binary logistic regression analysis comparing exposedand non-exposed subjects. Significant associations reported only.

Metabolites Predictor variables OR 95%CI p

Lower Upper

3-Methyl-4-nitrophenol Periurban groupa 2.228 1.214 4.090 0.010Rural groupa 2.648 1.426 4.917 0.002Residing on a farmb 2.226 1.131 4.381 0.020

DBCA Residing in a ruraltownb

2.097 1.046 4.203 0.037

DEP Parental occupationalexposure

2.839 1.155 6.976 0.023

DMTP Periurban groupa 2.459 1.376 4.393 0.002Rural groupa 3.484 1.885 6.437 b0.001Parental occupationalexposure

2.403 1.328 4.346 0.004

a Compared to urban group.b Compared to subjects residing in the metropolitan Adelaide.

Urban Peri-urban group

Rural group

Mea

n (µ

g/g

crea

tinin

e) ±

SE

Fig. 3. Concentrations of specific metabolites of OP insecticides by study group. Y axisrepresents mean concentration in μg/g creatinine±1SE.

116 K. Babina et al. / Environment International 48 (2012) 109–120

The periurban group was comprised of children living in theAdelaide Hills region. This region is known for mixed pastoral andhorticultural (wine growing and orchards) land use mixed withhobby farms, were farmers and many hobby farmers hold restrictedchemical use licences. While no comprehensive data has ever beencollected, this area is anecdotally known for overuse of agriculturalchemicals.

The rural group was comprised of children living on the farms andin small townships in several areas of country SA. These are areas ofmixed pastoral, broad acre agricultural and intensive horticulturalland uses, where a large variety of chemicals would be used.

4.1. Risk of exposure

Risk of exposure to a wide range of PYR insecticides (as measuredby urinary levels of 3-PBA) did not differ among the study groups andwas not affected by the predictor variables used in the analysis. Sim-ilarly, the risk of exposure to chlorpyrifos (as determined by urinarylevels of TCPy) and bifenthrin (as determined by urinary levels ofMPA) was not influenced by the type or area of residence, proximityto agricultural activity or parental occupational exposure. This canbe explained by the fact that chlorpyrifos as well as bifenthrin andmany PYRs are used extensively in agriculture and are freely availableto the general public for domestic use and therefore are potentiallyubiquitous in the environments of the study children. The risk of ex-posure to fenithrothion (as determined by detection of 3-methyl-4-nitrophenol) was strongly associated with rural and periurban areasof residence and living on the farm. Fenitrothion is restricted forlicensed use only. Similarly, the risk of exposure to deltamethrin (as de-termined by urinary DBCA) was higher in children residing in ruraltowns. At the time of the study deltamethrin was not commonly

Urbangroup

Peri-urbangroup

Rural

Mea

n (µ

g/g

crea

tinin

e) ±

SE

Fig. 2. Concentrations of metabolites of non-specific metabolites of OP insecticides bystudy group. Y axis represents mean concentration in μg/g creatinine±1SE.

accessible to the general public. This finding may also be confoundedby low frequency of detection of urinary DBCA (less than 15% ofpopulation).

4.2. Level of exposure

Overall, the best predictors for higher levels of urinary metaboliteswere periurban and rural areas of residence and reported parental oc-cupational exposure. This was true for non-specific metabolites ofPYRs (3-PBA) and OPs (DEP, DETP and DMTP), potentially reflectinghigher exposure to a wide range of PYRs and OPs compared to chil-dren living in urban Adelaide or children whose parents did not re-port occupational contact with pesticides. These findings supportthe previously reported observations of increased exposures inpopulations residing in rural agricultural areas, and in families of pes-ticide handlers (Castorina et al., 2010; Fenske et al., 2002). This didnot however hold true for exposure to commonly available and wide-ly used chemicals, such as chlorpyrifos and bifenthrin. This indicatesthat reliance on indirect exposure assessment without direct expo-sure measurement can lead to exposure misclassification. A similarobservation but in respect of children's exposure to herbicides hasbeen highlighted in an earlier study by Arbuckle et al. (2004).

Of specific metabolites, periurban group had higher levels ofp-nitrophenol and 3-methyl-4-nitrophenol, reflecting higher expo-sure to parathion and parathion-methyl and fenithrothion. Parathionwas not registered for use in SA at the time of the sample collectionwhile parathion-methyl and fenitrothion were registered for licenseduse only. The rural group, similar to the risk of exposure, had higherlevels of urinary 3-methyl-4-nitrophenol reflecting higher levels ofexposure to fenitrothion compared to urban children. Significantlyhigher levels of urinary DCCA in children living within 50 m froman agricultural activity would be reflective of higher exposure to per-methrin, cypermethrin and cyfluthrin, all of which are used widely in

Urban group Periurban group

Rural group

Mea

n (µ

g/g

crea

tinin

e) ±

SE

Fig. 4. Concentrations of metabolites of PYR insecticides by study group. Y axis representsmean concentration in μg/g creatinine±1SE.

117K. Babina et al. / Environment International 48 (2012) 109–120

agriculture and are available for domestic use. No other factors wereassociated with urinary DCCA. Similar to the results of the analysisof the risk of exposure, there were no significant differences in theurinary levels of TCPy and MPA reflecting no significant differencesin levels of exposure to chlorpyrifos and bifenthrin among the studygroups. Unlike the risk of exposure to deltamethrin, the levels of uri-nary DBCA did not differ significantly by study group and were not as-sociated with any other factor. This may be due to low frequency ofdetection of urinary DBCA in the study population.

4.3. Dietary exposure

Apart from chemical use in the area of residence or in the home,consumption of contaminated food may be a significant contributorto the overall exposure. OP and PYR residues have been measuredin foods consumed by young children in many studies (Clayton etal., 2003; Fenske et al., 2002; MacIntosh et al., 2001). Curl et al.(2003a) and Lu et al. (2006) demonstrated lower levels of urinarymetabolite excretion in children on organic diets. Lu et al. (2010)detected eleven OP and three PYR chemicals in their 24-h duplicatefood sample study.

Dietary exposure to agricultural and veterinary chemicals has beenrecently explored in Australia in the 23rd Australian Total Diet Survey(ATDS) (FSANZ, 2011). Three PYR chemicals were reported in the23rd ATDS: allethrin (in two food samples), bifenthrin (in one fruitsample) and permethrin (in one sample). Of the OPs, chlorpyrifos-methyl, chlorpyrifos and dimethoate were detected in several foodsamples, also detected were methamidofos and omethoate (each inone sample). While it appears that occurrence of pesticide residues onfood is low in Australia, the 23rd ATDS also reported that chlorpyrifosresidues in food contributed up to 20% of acceptable daily intake(ADI) in the 2–5 years old group (estimated at 90th percentile). Thiswould account to approximately 0.6 μg/kg bw/day of chlorpyrifos inthe general population of Australian children aged 2–5 years old. Forcomparison, Kawahara et al. (2007), estimated dietary intake ofchlorpyrifos in children aged 3–6 years old residing in Tokyo at0.007 μg/kg bw/day. Similarly, estimates for the US population basedon the results of the US Food and Drug Administration's Total DietStudy were around 0.03 μg/kg bw/day (estimated at 90th percentile)(Wright et al., 2002). Both comparisons seem to imply that, althoughbelow the regulatory levels of concern (belowADI), the Australian pop-ulation of young children has higher levels of dietary exposure to chlor-pyrifos, than their peers in Japan and the US. This may, in part, explainwhy there was little difference in levels of urinary TCPy was commonamong the study groups in our study.

Drinking water may be a contributor to the overall exposure. Con-temporary water treatmentmethods can reliably remove pesticide res-idues from centralised drinking water supplies. Residents in periurbanand rural SA, however, often rely heavily or solely on rainwater fortheir drinking water. We could not find any published reports of inves-tigations into pesticide residues in domestic rainwater tanks in SA.

4.4. Preformed metabolites

The validity of using urinary metabolites as a golden standard ofexposure assessment has been questioned and debated due to thefact that for some chemicals, environmental degradation of parentcompounds may yield the same metabolites as human metabolism.TCPy has been shown to be common in children's environment byMorgan et al. (2005) and has also been detected in the ambient air(Raina and Sun, 2008). DAPs have been detected in fruit juices (Luet al., 2005). It is therefore possible that some proportion of urinarymetabolites can come from ingesting preformed metabolites withfood and environmental media rather than ingesting the parent com-pound (active insecticide). TCPy, DEP and DETP have been shown to

be rapidly absorbed and eliminated with limited metabolism by ratswhen administered orally (Timchalk et al., 2007).

It is not clear what proportion of TCPy and DAPs excreted in urinewould have come from actual parent compounds, their environmen-tal oxon products or the preformed metabolites. This would poten-tially depend on route of exposure, environmental conditions andthe individual's metabolic make up. The preformed metabolites con-tribution needs to be taken into account in the future case–controland cohort studies, where good understanding of all factors affectingintra- and inter-individual variability is essential for investigatingcausative relationships between exposure and health effects. However,it is believed that a population-based cross sectional studies exploringassociative relationships, such as the one presented here, would not dra-matically suffer from potential interference frompreformedmetabolites.

4.5. International comparison

Urinary metabolite concentrations measured in this study do nothave clinical guideline values to compare to in order to ascertain po-tential public health significance of the observed exposure levels.Also, being the first Australian study of its kind, we had no ‘back-ground’ data for Australian population to compare to. The only optionfor conceptualising our findings was to compare the results of ourstudy to the available published international data.

We compared our results with available published data on theconcentrations of urinary DAPs in children under 6 years of age inGermany (Heudorf et al., 2004) and children aged 6 to 11 in theUSA (NCEH, 2009). Figs. 5–7 demonstrate that SA children havehigher exposures than both German and the US children. We usedthe international data in the formats they were available: Germandata were reported in mean μg/L (Heudorf et al., 2004), while USdata were reported in μg/g creatinine by percentile group. While thecomparison is of qualitative nature since it was not possible to con-duct statistical comparison, the differences appear to be substantial.South Australian children appear to have substantially higher expo-sure levels to a wide range of OP and PYR pesticides than Germanand the US children.

The analytical methods employed by this study differed frommethods used in Heudorf et al. (2004) and NCEH (2009) and werenot cross validated against existing methods. Such validation wasnot possible at the time due mostly to resource limitations. Besides,Australia did not and still does not have another laboratory activelyinvolved in biological monitoring of environmental exposures to pes-ticides in the general population, which could be used for cross-validation purposes. We have put every effort to adhere to good lab-oratory practices in our method development. While it is possible thatsome proportion of the differences seen between Australian, US andGerman populations may be due to potential methodological bias,the authors do not believe that such bias would be of significantmagni-tude. There is a need for thorough international cross-validation of theexisting methods for future studies.

To further conceptualise this comparison, we looked at thenumber of registered OP and PYR-containing products in Australia,Germany, the US and the UK. Several international regulatory databaseswere examined in the preparation of this paper (APVMA, 2012; CERIS,2012; HSE, 2012; National Ministry of Food, Agriculture and ConsumerProtection, 2012). Table 7 summarises the numbers of OP and PYR(active constituents only) registered in Australia, UK and USA withnumbers of actives available to the general public. It is clear from thenumbers that Australia has substantially more OPs and PYRs registeredfor both restricted and domestic use by the general public than othercomparable developed countries such as Germany, the USA and UK(Table 8). This difference in chemicals variety and in general availabilityto the general public in particularmay be contributing to the differencesseen in urinary metabolite concentrations.

Fig. 5. Concentration (by percentile groups, μg/g creatinine) of selected DAPs in urine samples from South Australian and US children (US samples collected 2003–2004, NCEH, 2009).

Fig. 6. Concentrations of selected specific metabolites (by percentile groups, μg/g creatinine) in urine samples from South Australian and US children (US samples collected2001–2002, NCEH, 2009).

118 K. Babina et al. / Environment International 48 (2012) 109–120

Fig. 7. Concentration of selected DAPs in urine samples from South Australian and German (Heudorf et al., 2004) children. Y axis represents mean concentration in μg/L±1SE.

119K. Babina et al. / Environment International 48 (2012) 109–120

4.6. Further research

This study raises more questions than provides answers. Furtherresearch is needed to thoroughly investigate exposure pathways inAustralian general population. Further studies should combine environ-mental monitoring, biological monitoring and testing of food for bothpesticide residues and pre-formed metabolites. Repeated samplingspanning different seasons, thorough evaluation of domestic use ofpesticides-containing products and cross-validation of analytical pro-cedures are needed. Participation in an international research effort,such as AGRICOH (Leon et al., 2011) could be very beneficial forAustralia.

5. Conclusions

Our findings demonstrate widespread chronic exposure to OP andPYR pesticides in young children in South Australia. There are differ-ences in exposure risks and levels in different populations, but only

Table 7Level of exposure: results of linear regression analysis of levels of exposure. Comparisonamong exposed subjects only (log-transformed creatinine corrected concentrations).Significant associations reported only.

Metabolites Predictor variables β t p

p-Nitrophenol Periurban groupa 0.214 2.007 0.0463-Methyl-p-nitrophenol Periurban groupa 0.242 2.218 0.028

Rural groupa 0.319 2.344 0.020DCCA 50 m proximity 0.381 3.210 0.0023-PBA Periurban groupa 0.267 2.745 0.006

Rural groupa 0.479 4.051 0.000DEP Periurbana 0.320 3.512 0.001

Rural groupa 0.390 3.476 0.001Parental occupational exposure 0.141 2.315 0.021

DETP Residing on the farmb 0.223 1.971 0.050DMTP Periurban groupa 0.259 2.375 0.018

Rural groupa 0.277 2.062 0.040Parental occupational exposure 0.209 2.964 0.003

a Compared to urban group.b Compared to subjects residing in the metropolitan Adelaide.

for some chemicals, while the exposure seems to be ubiquitous forother chemicals. There appear to be higher levels of exposure in thestudy population as compared to similar populations in the US andGermany. Further research into pesticide exposure in the generalpopulation in Australia is warranted, as is public discussion on thecurrent approaches to pesticide regulation and risk assessment andthe widespread availability of neurotoxic pesticides to the generalpublic.

Competing interests

Authors declare that they have no competing interests.

Acknowledgements

Authors express their appreciation to the participating families. Theurine sample analyses were performed in Flinders Advanced AnalyticalLaboratory with invaluable help fromDr Daniel Jardine. Analytical stan-dards for DAPs were kindly donated by the WorkCover LaboratoriesNSW. Kateryna Babina was supported by the FUSA PostgraduateAward (funded by the Flinders University). The study was funded bythe Financial Markets for Children Foundation.

Table 8Summary of registered OP and PYR chemicals in Australia, USA, Germany and UK (May2012).

Registered chemicals Australia USA Germany UK

OP Active constituents 30 10 4 5Active constituents available to the generalpublic

6 1 2 0

PYR Active constituents 22 15 8 7Active constituents available to the generalpublic

18 9 2 4

120 K. Babina et al. / Environment International 48 (2012) 109–120

References

ABS. Australian Bureau of Statistics: Demography: South Australia. [3311.4,], http://www.abs.gov.au.

ABS. Australian Bureau of Statistics: Agricultural State Profile, South Australia. AgriculturalState Profiles: 7123.4.55; 2003. [[http://www.abs.gov.au], accessed 2005].

ABS. Australian Bureau of Statistics: the measures of financial hardship. AustraliaNowMeasures of Australia's Progress; 2004. [[http://www.abs.gov.au], accessed2005].

Aprea C, Sciarra G, Sartorelli P, Desideri E, Amati R, Sartorelli E. Biological monitoring ofexposure to organophosphorus insecticides by assay of urinary alkylphosphates:influence of protective measures during manual operations with treated plants.Int Arch Occup Environ Health 1994;66:333–8.

Aprea C, Strambi M, Novelli MT, Lunghini L, Bozzi N. Biologic monitoring of exposure toorganophosphorus pesticides in 195 Italian children. Environ Health Perspect2000;108:521–5.

APVMA. Australian Pesticide and Veterinary Medicine Authority: Public ChemicalRegistration Information System (PUBCRIS) http://www.apvma.gov.au/pubcris.[accessed May 2012].

Arbuckle TE, Cole DC, Ritter L, Ripley BD. Farm children's exposure to herbicides. Com-parison of biomonitoring and questionnaire data. Epidemiology 2004;15:187–94.

Atlas SA. Report on Generalized Land Use for 2003. Web-based resource by SouthAustralian Government. www.atlas.sa.gov.au. [accessed 2004].

Baker SE, Olsson A, Barr D. Isotope dilution high-performance liquid chromatography–tandemmass spectrometry method for quantifying urinary metabolites of synthet-ic pyrethroid insecticides. Arch Environ Contam Toxicol 2004;46:281–8.

Barr D, Turner WE, DiPietro E, McClure PC, Baker SE, Barr JR, et al. Measurement ofp-nitrophenol in the urine of residents whose homes were contaminated withmethyl parathion. Environ Health Perspect 2002;110:1085–91.

Bjørling-Poulsen M, Andersen HR, Grandjean P. Potential developmental neurotoxicityof pesticides used in Europe. Environ Health 2008;7:50.

Castorina R, Bradman A, Fenster L, Barr DB, Vedar MG, Harnly ME, et al. Comparisonof current-use pesticide and other toxicant urinary metabolite levels amongpregnant women in the CHAMACOS cohort and NHANES. Environ HealthPerspect 2010;118(6):856–63.

CERIS. Centre for Environmental and Regulatory Information Systems. National Pesti-cide Information retrieval System (NPIRS) http://ppis.ceris.purdue.edu/htbin/ppisprod.com. [accessed May 2012].

Clayton CA, Pellizzari ED, Whitmore RW, Quackenboss JJ, Adgate J, Sefton K. Distribu-tions, associations, and partial aggregate exposure of pesticides and polynucleararomatic hydrocarbons in the Minnesota Children's Pesticide Exposure Study(MNCPES). J Expo Anal Environ Epidemiol 2003;13:100–11.

Cohen J. Statistical power analysis for the social sciences. 2nd edition. Hillsdale, NewJersey: Erlbaum; 1988.

Cohen J, Cohen P, West SG, Aiken LS. Applied regression/correlation analysis for the be-havioral sciences. 3rd edition. Mahwah, New Jersey: LEA; 2003.

Curl CA, Fenske RA, Elgethun K. Organophosphorus pesticide exposure of urban andsuburban preschool children with organic and conventional diets. Environ HealthPerspect 2003;111:377–82.

Esteban E, Rubin C, Hill RH, Olson D, Pearce K. Association between indoor residentialcontamination with methyl parathion and urinary p-nitrophenol. J Expo Anal Envi-ron Epidemiol 1996;6:375–87.

Fenske RA, Lu C, Barr D, Needham L. Children's exposure to chlorpyrifos and parathionin an agricultural community in central Washington State. Environ Health Perspect2002;110:549–53.

FSANZ. Food Standards Australia New Zealand. The 23rd Australian Total Diet Survey.Food Standards Australia New Zealand. [Available] www.foodstandards.gov.au.[accessed December 2011].

Gilbert RO. Statistical methods for environmental pollution monitoring. New York:VNR; 1987.

Glover J, Tennant S. South Australia. A social Atlas of Australia, 2nd edition, Vol. 5.Australia: Openbook Publishers; 1999.

Hardt J, Angerer J. Determination of dialkyl phosphates in human urine using gaschromatography–mass spectrometry. J Anal Toxicol 2000;24:678–84.

Hernandez F, Sancho JV, Pozo OJ. Direct determination of alkyl phosphates in humanurine by liquid chromatography/electrospray tandem mass spectrometry. RapidCommun Mass Spectrom 2002;16:1766–73.

Hernandez F, Sancho JV, Pozo OJ. Critical review of the application of liquidchromatography/mass spectrometry to the determination of pesticide residues inbiological samples. Anal Bioanal Chem 2005;382:934–46.

Heudorf U, Angerer J, Drexler H. Current internal exposure to pesticides in children andadolescents in Germany: urinary levels of metabolites of pyrethroid and organo-phosphorus insecticides. Int Arch Occup Environ Health 2004;77:67–72.

HSE. Health and Safety Executive. Pesticides register https://secure.pesticides.gov.uk/pestreg/. [accessed May 2012].

Jones FL, McMillan J. Scoring occupational categories for social research: a review ofcurrent practice, with Australian examples. Work Employ Soc 2001;15:539–63.

Kawahara J, Yoshinaga J, Yanagisawa Y. Dietary exposure to organophosphorus pesti-cides for young children in Tokyo and neighboring area. Sci Total Environ2007;378:263–8.

Leon ME, Beane Freeman LE, Douwes J, Hoppin JA, Kromhout H, Lebailly P, et al.AGRICOH: a consortium of agricultural cohorts. Int J Environ Res Public Health2011;8(5):1341–57.

Loewenherz C, Fenske RA, Simcox NJ, Bellamy G, Kalman D. Biological monitoring oforganophosphorus pesticide exposure among children of agricultural workers incentral Washington State. Environ Health Perspect 1997;105:1344–53.

Lu C, Bravo R, Caltabiano LM, Irish RM, Weerasekera G, Barr D. The presence ofdialkylphosphates in fresh fruit juices: implications for organophosphorus pesti-cide exposure and risk assessment. J Toxicol Environ Health A 2005;68:209–27.

Lu C, Toepel K, Irish RM, Fenske RA, Barr D, Bravo R. Organic diets significantly lowerchildren's exposure to organophosphorus pesticides. Environ Health Perspect2006;114:260–3.

Lu C, Schenk FJ, Pearson MA, Wong JW. Assessing children's dietary pesticide exposure:direct measurement of pesticides residues in 24-hr duplicate food samples. Envi-ron Health Perspect 2010;118(11):1625–30.

MacIntosh DL, Needham LL, Hammerstrom KA, Ryan PB. A longitudinal investigation ofselected pesticide metabolites in urine. J Expo Anal Environ Epidemiol 1999;9:494–501.

MacIntosh DL, Kabiru CW, Ryan PB. Longitudinal investigation of dietary exposure toselected pesticides. Environ Health Perspect 2001;109:145–50.

Morgan MK, Heldon LS, Croghan C, Johnes P, Robertson G, Chuang JC, et al. Exposure ofpreschool children to chlorpyrifos and its degradation product 3,5,6-trichloro-2-pyridinol in their everyday environments. J Expo Anal Environ Epidemiol2005;15:297–309.

National Ministry of Food, Agriculture and Consumer Protection. List of approvedpesticides https://portal.bvl.bund.de/psm/jsp/. [accessed May 2012].

NCEH. National centre for Environmental Health. Fourth National Report on HumanExposure to Environmental Chemicals. NCEH Pub. No. 05‐0570. Department ofHealth and Human Services, Centers for Disease Control and Prevention; 2009[http://www.cdc.gov/exposurereport/pdf/FourthReport.pdf].

Oglobline AN, O'Donnell GE, Geyer R, Holder GM, Tattam B. Routine gas chromato-graphic determination of dialkylphosphate metabolites in the urine of workersoccupationally exposed to organophosphorus insecticides. J Anal Toxicol 2001;25:153–7.

Olsson A, Baker SE, Nguyen J, Romanoff LC, Udunka SO, Walker RD, et al. A liquidchromatography-tandem mass spectrometry multiresidue method for quantifica-tion of specific metabolites of organophosphorus pesticides, synthetic pyrethroids,selected herbicides, and DEET in human urine. Anal Chem 2004;76:2453–61.

Raina R, Sun L. Trace level determination of selected organophosphorus pesticides andtheir degradation products in environmental air samples by liquid chromatography-positive ion electrospray tandem mass spectrometry. J Environ Sci Health B 2008;43:323–32.

Saieva C, Aprea C, Tumino R, Masala G, Salvini S, Frasca G, et al. Twenty-four-hoururinary excretion of ten pesticide metabolites in healthy adults in two differentareas of Italy (Florence and Ragusa). Sci Total Environ 2004;332:71–80.

Sancho JV, Pozo OJ, Hernandez F. Direct determination of chlorpyrifos and its mainmetabolite 3,5,6-trichloropyridinol in human serum and urine by coupled-column liquid chromatography/electrospray-tandem mass spectrometry. RapidCommun Mass Spectrom 2000;14:1485–90.

Smith PA, Thompson MJ, Edwards JW. Estimating occupational exposure to the pyre-throid termiticide bifenthrin by measuring metabolites in urine. J ChromatogrB Analyt Technol Biomed Life Sci 2002;778:113–20.

Timchalk C, Busby A, Campbell JA, Needham LL, Barr DB. Comparative pharmacokinet-ics of the organophosphorus insecticide chlorpyrifos and its major metabolitesdiethylphosphate, diethylthiophosphate and 3,5,6-trichloro-2-pyridinol in therat. Toxicology 2007;237(1–3):145–57.

Wilson NK, Chuang JC, Iachan R, Lyu C, Gordon SM, Morgan MK, et al. Design and sam-pling methodology for a large study of preschool children's aggregate exposures topersistent organic pollutants in their everyday environments. J Expo Anal EnvironEpidemiol 2004;14:260–74.

Wright JP, Shaw MC, Keeler LC. Refinements in acute dietary exposure assessments forchlorpyrifos. J Agric Food Chem 2002;50(1):235–41.

Yatzidis H. New method for direct determination of “true” creatinine. Clin Chem1974;20:1131–4.