Science of the Total Environment 496 (2014) 54–62

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Human health risk assessment of the mixture of pharmaceuticals inDutch drinking water and its sources based on frequent monitoring data

Corine J. Houtman a,⁎, Jan Kroesbergen a,1, Karin Lekkerkerker-Teunissen b, Jan Peter van der Hoek c,d

a The Water Laboratory, P.O. Box 734, 2003 RS, Haarlem, The Netherlandsb Dunea Dune and Water, Zoetermeer, The Netherlandsc Delft University of Technology, Delft, The Netherlandsd Waternet, Amsterdam, The Netherlands

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• We assessed the presence and risks of(mixtures of) pharmaceuticals indrinking water.

• Lifelong water consumption led topharmaceutical exposure b10% of adaily medical dose.

• Calculated risks of adverse healtheffects of single pharmaceuticals werenegligible.

• This was confirmed for the mixture ofpharmaceuticals simultaneously present.

• Drinking water treatment plantsreduced the (already negligible) riskup to 80%.

⁎ Corresponding author. Tel.: +31 23 5175900.E-mail addresses: corine.houtman@hetwaterlaborator

(K. Lekkerkerker-Teunissen), j.p.vanderhoek@tudelft.nl (J.1 Tel.: +31 23 5175900.

http://dx.doi.org/10.1016/j.scitotenv.2014.07.0220048-9697/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 26 March 2014Received in revised form 7 July 2014Accepted 7 July 2014Available online xxxx

Editor: D. Barcelo

Keywords:Mixture toxicityThreshold of toxicological concernExposure assessmentPharmaceuticalsDrinking water

The presence of pharmaceuticals in drinking water is a topic of concern. Previous risk assessments indicate thattheir low concentrations are very unlikely to pose risks to human health, however often conclusions had to bebased on small datasets and mixture effects were not included.The objectives of this study were to a) investigate if pharmaceuticals in surface and polder water penetrate indrinkingwater, b) assess the lifelong exposure of consumers to pharmaceuticals via drinking water and c) assessthe possible individual and mixture health risks associated with this exposure.To fulfill these aims, a 2-year set of 4-weekly monitoring data of pharmaceuticals was used from three drinkingwater production plants. The 42 pharmaceuticals that were monitored were selected according to theirconsumption volume, earlier detection, toxicity and representation of the most relevant therapeutic classes.Lifelong exposures were calculated from concentrations and compared with therapeutic doses. Health riskswere assessed by benchmarking concentrations with provisional guideline values. Combined risks of mixturesof pharmaceuticals were estimated using the concept of Concentration Addition.The lifelong exposure to pharmaceuticals via drinking water was calculated to be extremely low, i.e. a fewmg, intotal corresponding to b10% of the dose a patient is administered on one day. The risk of adverse health effectsappeared to be negligibly low. Application of Concentration Addition confirmed this for the mixture of

ium.nl (C.J. Houtman), jan.kroesbergen@hetwaterlaboratorium.nl (J. Kroesbergen), k.lekkerkerker@dunea.nlP. van der Hoek).

55C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

pharmaceuticals simultaneously present. The investigated treatment plants appeared to reduce the (already neg-ligible) risk up to 80%. The large available monitoring dataset enabled the performance of a realistic risk assess-ment. It showed that working with maximum instead of average concentrations may overestimate the riskconsiderably.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

The presence of pharmaceuticals in the aquatic environment hasbeen reported numerous times. They have been detected at a nanogramto microgram/L range in waste water, groundwater and surface water(Daughton and Ternes, 1999; de Jongh et al., 2012; Godfrey et al.,2007; Halling-Sorensen et al., 1998; Houtman, 2010; Houtman et al.,2013; Kasprzyk-Hordern et al., 2008; Kolpin et al., 2002). The emissionof sewage treatment plant (STP) effluent on surface waters has beenidentified as a major source of pharmaceuticals in the environment. Inrural areas without connection to main sewage systems, septic tankscan be another source of exposure (Du et al., 2014; Stanford andWeinberg, 2010), however, not in densely populated countries (e.g.98% of the Dutch households is connected to the main sewage system(De Moel et al., 2004)). Pharmaceuticals are intentionally designed toaffect biochemical functions in humans and livestock. Possible risks ofexposure for human health are a subject of concern, scientifically andin the general population. This is especially the case in countries thatdepend strongly on surface waters as a source for drinking water, suchas the Netherlands. In this country, 40% of the produced drinkingwater is prepared from surface water, predominantly from the riversRhine and Meuse.

Some studies have indeed reported the presence of pharmaceuticalsin drinking water in several countries (e.g. (Benotti et al., 2009b; Bruceet al., 2010; Versteegh et al., 2007)). Concomitant risk assessments(RAs) indicate that the low concentrations of pharmaceuticals detectedin drinking-water are very unlikely to pose risks to human health.However, currently published studies had to deal with a number oflimitations.

Firstly, RAs havemost often been based on limited sets ofmonitoringdata (Benotti et al., 2009b; Jones et al., 2005; Spencer Williams andBrooks, 2012; Villanueva et al., 2014), because research projects gener-ally do not have the possibility to monitor frequently or over long timeperiods. Therefore, conclusions had to be based on a relatively fewdata, only on maximum concentrations in drinking water, or on modelestimations (Benotti et al., 2009b; Bruce et al., 2010; Cunninghamet al., 2009).

Secondly, some studies had to use concentrations measured insurface waters instead of drinking water (Cunningham et al., 2010;Johnson et al., 2008; Vulliet and Cren-Olivé, 2011). This assumes thatdrinking water treatment schemes do not remove any of the pharma-ceuticals in the raw water. With the exception of some groundwater plants with virtually no other treatment than aeration, this is anunrealistic worst case, as demonstrated e.g. in Benotti et al. (2009a)and Lekkerkerker-Teunissen et al. (2012).

Thirdly, RA in the case of mixtures in drinking water is stillchallenging as knowledge gaps keep existing in terms of assessing risksassociatedwith long-termexposure to lowconcentrations of pharmaceu-ticals and especially with possible synergistic mixture effects caused bythe simultaneous presence of traces of multiple pharmaceuticals (Boxallet al., 2012; Villanueva et al., 2014; World Health Organization, 2012).

The aim of this study was to a) investigate if pharmaceuticals insurface and polder waters penetrate in drinking water, b) assess thelifelong exposure of consumers to individual pharmaceuticals and thetotal mixture via drinking water and c) assess the possible health risksassociated with this exposure.

The above mentioned limitations were tackled by collecting anextensive 2-year dataset of 4-weekly monitoring data for three Dutchdrinking water production plants, using a very sensitive UPLC-MS/MS

analysis method for 42 pharmaceuticals. Data were available forthe source waters and the produced drinking waters, thus enablingcomparison of the sources for drinking waters, and their correspondingdrinkingwaters towhich consumers are exposed. RAwas performed bybenchmarking measured concentrations with provisional Guide LineValues (pGLV), representing concentrations that do not result in anysignificant risk to health over a lifelong consumption (World HealthOrganization, 2006).

To include possible mixture effects, the concept of ConcentrationAddition (CA) was applied. Although developed for mixtures ofsimilarly acting compounds, it is suggested also to be applicable as aprecautious first tier for mixture constituents with various modes ofaction (Backhaus and Faust, 2012), such as the investigated set ofpharmaceuticals.

2. Materials and methods

2.1. Study sites

Water sampleswere collected from three drinkingwater productionplants in theNetherlands. The Rhine plant produces drinkingwater fromthe river Rhine. After the intake the water is pretreated and infiltrat-ed into the dunes. After abstraction and post treatment the produceddrinking water is distributed to Amsterdam and surroundings. TheMeuse plant takes in surface water from the Enclosed Meuse, adead-end side stream of the river Meuse, to produce drinkingwater for the inhabitants of the city of The Hague and surroundings.The surfacewater is pretreated, infiltrated into thedunes, abstracted andis distributed after post treatment. The Polder plant uses seepage waterfrom the Bethune Polder. This water is a mixture of old groundwater,run-off of rainfall and seepage water from the surrounding lakes of thispiece of reclaimed land. After treatment, the drinking water is distribut-ed to Amsterdam and its surroundings. For amore extensive descriptionof the plants see the Supplemental Material.

2.2. Sampling and analysis of pharmaceuticals

Grab samples (100 mL) were taken in pre-rinsed bottles of greenglass every four weeks from August 2010 till December 2012 fromwater at the three intake sites. Drinking water samples were takenfour-weekly from January 2011 till December 2012. In total 26samples were taken for each water source. Samples were immediatelytransported in a refrigerated van to the laboratory and kept at 4 °Cuntil processing. Analysis was performed after solid phase extractionof the samples on an Ultra Performance Liquid Chromatograph(UPLC), combined with a triple quadrupole Mass Selective Detector asdescribed in Houtman et al. (2013). The analysis method contained 42pharmaceuticals selected according to their consumption volume,earlier detection, toxicity and representation of the most relevant ther-apeutic classes. The method was validated by calculating the recoveryand standard deviation of surface water samples from eight differentlocations and sampled on different days spiked with pharmaceuticalsat a low level (resp. 0.5, 2.5 or 15 ng/L) to determineMRLs and at a higherlevel (resp. 0.5, 2.5 or 15 μg/L) extracted and analyzed at different daysto determine reproducibility. Most (32) compounds had a minimumreporting limit (MRL) of 5 ng/L or lower, of which 18 compoundshad an MRL between 0.1 ng/L to 1 ng/L. Highest MRL was obtained forclofibrate (85 ng/L). The average recovery found was 91 ± 14% (errorincludes whole process from sampling to analysis).

Fig. 1. Pharmaceuticals detected in drinking water sampled at three Dutch drinking water plants, respectively the Rhine plant (upper panel), Meuse plant (middle panel) and the Polderplant (lower panel). Total concentrations of all pharmaceuticals are shown according to their therapeutic classes.

56 C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

3. Calculations

3.1. Abstracted loads

Daily loads of pharmaceuticals in the intake water were calculat-ed from themeasured concentrations and abstracted water volumes.Abstracted water volume data were provided by the drinking watercompanies in m3/d. From the abstracted water volumes four-weekaverages were calculated (two weeks before till two weeks afterthe sample for pharmaceutical analysis was taken) and multiplied

with the detected concentration, to obtain four-weekly averageddaily loads.

3.2. Average concentrations in drinking water

Most pharmaceuticals were detected only in part of the samples.Removing non-detects would overestimate the calculated averageconcentration, whereas setting all non-detects at ‘zero’would underes-timate average concentrations. The latter was confirmed by the fact thatall analyzed pharmaceuticals, except oxacillin, were detected in some of

Fig. 2. Concentrations (averages ± standard deviations of 26 4-weekly measurements) of individual pharmaceuticals detected in drinking water prepared from the river Rhine.

57C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

the surface water and polder water samples, and therefore theirpresence in (part of) the drinking water cycle was demonstrated,indicating that the presence of pharmaceuticals in undetectably lowconcentrations in drinking water could not be excluded. Thereforenon-detects in drinking water were set at the nominal value of ¼ ofthe individualMRL of each pharmaceutical for the calculation of averageconcentrations of the 26 samples taken in 2011 and 2012. In this way,data are still biased, however more realistic than in case values wouldhave been set at ‘zero’ or measurements would have been ignored.

3.3. Lifelong exposure

Lifelong exposures (mg) to pharmaceuticals via drinkingwaterwerecalculated as: concentration of pharmaceutical (ng/L) ∗ drinking waterconsumption (L/d) ∗ 65.25 (d/y) ∗ 70 (y)/1,000,000, with the followingassumptions:

• The consumption of drinking water is 2 L/d. This is a standard assump-tion used by the Dutch drinking water sector and as such e.g. also usedby de Jongh et al., (2012), Schriks et al. (2009), and Versteegh et al. (2007).

• A lifelong exposure equals 70 years, in agreementwith the average life ex-pectancy at birth of 71.0 years over the period 2010–2013 according toUnited NationsWorld Population Prospects 2012 Revision andwith earlierrisk assessment studies (Schriks et al., 2009; Versteegh et al., 2007; Webbet al., 2003).

In addition, lifelong exposures were expressed as percentages oftherapeutic doses used by patients treatedwith the specific pharmaceu-ticals. As lifelong exposures were expressed as percentage of one singledefined daily dose (DDD), lifelong exposure via drinking water wascompared with a one-day exposure via medical treatment. To this aimDDDs were used that are defined by the World Health Organization(WHO) as reported on http://www.whocc.no/atc_ddd_index/. Forcompounds for which no DDD was defined, following alternativeswere used: caffeine: the Acceptable Daily Intake (ADI) as reported byVersteegh et al. (2007), clofibric acid and fenofibric acid: zero, as thesecompounds are metabolites of clofibrate and fenofibrate respectively

and are not administered as such, iopromide: zero, because this is anX-ray contrast agent that is only administered for diagnostic purposesand not on a regular basis as treatment against a disease, and tiamulin:zero, as this pharmaceutical is only registered for veterinary applica-tions. In these latter cases in which the alternative for a DDD was zero,compounds were left out of the corresponding exposure calculation.

The influence of the extensiveness of the dataset on the outcome ofthe determined exposures, was determined by:

1) calculating the exposure using concentrations averaged over the twoyears of data collection, and besides that also with two worst caseand less realistic scenarios:2a) using concentrations asmeasured on the sampling daywith the

highest total concentration (or DDD equivalent)2b) using for each pharmaceutical the maximum concentration

observed in the 2 year monitoring period.

To calculate the exposure of the total mixture of pharmaceuticals, life-long exposureswere summedassuming the applicability of the concept ofConcentrationAddition (Backhaus and Faust, 2012) as a practical estimate.

3.4. Possible human health risks of pharmaceuticals

The human health risk of exposure to pharmaceuticals via drinkingwater was assessed by comparing concentrations of pharmaceuticalswith pGLVs, for “threshold chemicals”, which are believed to have asafe dose below which no adverse effects will occur, an AcceptableDaily Intake (ADI) or Tolerable Daily Intake (TDI) value from literaturewas used as Point of Departure (PoD) to calculate a pGLV. If notavailable, a pGLV directly from literature was used. If also no pGLV wasavailable, a provisional ADI was calculated using the lowest therapeuticdose or DDD as PoD, using an Uncertainty Factor (UF) of 100, as inSchriks et al. (2009). The pGLV was calculated as: pGLV (mg/L) = A(T)DI ∗ bodyweight ∗ P/drinking water consumption (L/d), with:

• 70 kg taken as bodyweight; as in this study concentrations in Dutchsampleswereused, a value closest to the European averagebodyweight

Table 1Lifelong exposure to pharmaceuticals via drinking water and benchmark quotients indicating the risk of adverse health effects. DDD = Defined Daily Dose.

Lifelong exposure Exposure (mg/70 years) Exposure (% of 1 DDD/70 years) Benchmark quotienta

Production plant Rhineplant

Meuseplant

Polderplant

Rhineplant

Meuseplant

Polderplant

Rhineplant

Meuseplant

Polderplant

Most contributing pharmaceuticalsAverage ± standard deviation(N = 26 water samples)Bisoprolol 0.05 ± 0.10 0.03 ± 0.05 0.06 ± 0.13 0.49 ± 0.97 0.26 ± 0.47 0.55 ± 1.27 0.0010 ± 0.0019 0.0005 ± 0.0009 0.0011 ± 0.0025Caffeine 0.68 ± 1.19 1.08 ± 1.45 0.39 ± 0.47 2.27 ± 3.98 3.60 ± 4.83 1.31 ± 1.56 0.0000 ± 0.0000 0.0000 ± 0.0000 0.0000 ± 0.0000Clofibrate(only detected in dw Meuse)

0.06 ± 0.00 0.63 ± 2.90 0.06 ± 0.00 0.00 ± 0.00 0.03 ± 0.14 0.00 ± 0.00 0.0000 ± 0.0000 0.0004 ± 0.0019 0.0000 ± 0.0000

Diclofenac 0.09 ± 0.16 0.2 ± 0.78 0.05 ± 0.00 0.09 ± 0.16 0.20 ± 0.78 0.05 ± 0.00 0.0002 ± 0.0004 0.0005 ± 0.0020 0.0001 ± 0.0000Furosemide 0.31 ± 1.35 0.31 ± 1.35 0.05 ± 0.06 0.78 ± 3.38 0.10 ± 0.00 0.12 ± 0.15 0.0003 ± 0.0013 0.0000 ± 0.0000 0.0000 ± 0.0001Metformin 1.31 ± 1.20 1.31 ± 1.56 1.13 ± 0.84 0.07 ± 0.06 0.07 ± 0.08 0.06 ± 0.04 0.0001 ± 0.0001 0.0001 ± 0.0001 0.0001 ± 0.0001Paroxetine 0.34 ± 1.31 0.09 ± 0.16 0.08 ± 0.16 1.68 ± 6.55 0.44 ± 0.79 0.39 ± 0.82 0.0006 ± 0.0025 0.0002 ± 0.0003 0.0002 ± 0.0003Propranolol 0.20 ± 0.56 0.09 ± 0.18 0.10 ± 0.15 2.04 ± 5.62 0.85 ± 1.81 0.98 ± 1.55 0.0008 ± 0.0022 0.0003 ± 0.0007 0.0004 ± 0.0006Salicylic acid 0.62 ± 1.30 0.14 ± 0.00 0.28 ± 0.58 0.62 ± 1.30 0.14 ± 0.00 0.28 ± 0.58 0.0005 ± 0.0010 0.0001 ± 0.0000 0.0002 ± 0.0005

Sum of all measured pharmaceuticalsAverage ± standard deviation(N = 26 water samples)

6.3 ± 10.8 6.1 ± 9.7 4.4 ± 5.0 9.2 ± 23.5 7.1 ± 11.1 4.8 ± 7.1 0.006 ± 0.006 0.005 ± 0.003 0.005 ± 0.004

At date with maximum sum 16 18 9 47 25 13 0.024 0.012 0.017Based on all maximumconcentrations

51 48 25 113 48 33 0.054 0.039 0.032

a Only “threshold” compounds included.

58 C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

of 70.8 kg (Walpole et al., 2012) was used and not e.g. a globalaverage.

• P = fraction of the ADI or TDI that is allocated to drinking water,set at 10% (de Jongh et al., 2012; Mons et al., 2012; Schriks et al.,2009; Versteegh et al., 2007). In this way, it is taken into accountthat drinking water is not the only source of exposure possible,thus ensuring that the total daily intake from all sources doesnot exceed the ADI or TDI.

For genotoxic carcinogens, it is often considered that exposure maytheoretically at any dose lead to an adverse health risk. Therefore, theyare referred to as “non-threshold compounds” and no safe ADI isderived. Instead, guideline values representing an estimated upper-bound lifelong cancer risk at a lifelong exposure (e.g. of 1 in 106 people)are usually derived. In the absence of data to do so,we applied the Thresh-old of Toxicological Concern concept. This approach sets a target valueof 0.01 μg/L in drinking water as conservative limit for both singlegenotoxicants as well as mixtures thereof (Mons et al., 2012). Thehuman health risk of exposure was assessed by benchmarking theconcentrations of pharmaceuticals against the pGLVs. Compounds witha so called Benchmark Quotient (BQ), i.e. the ratio of concentration overpGLV, ≥1 in drinking water may be of potential human health concernif exposure lasts a lifelong. Compounds with 0.1 ≤ BQ ≤ 1 may warrantfurther investigation, whereas for compounds with BQ b 0.1 the risk ofadverse health effects is considered negligibly low (Schriks et al., 2009).

To calculate the possible health risk of the total mixture of pharmaceu-ticals present in water samples, BQs of “threshold” chemicals weresummed, according to the CA concept, for water samples taken at intakepoints for drinking water production and for drinking water samples forall three plants. Genotoxicants, being “non-threshold” chemicals, were notincluded in the calculated BQs, as for these compounds risks could possiblynot be excluded, even if the calculated BQ is b0.1. Statistical significance ofremoval of mixture BQ during drinking water production was assessedusing the two sample paired T-test in the software package XLStat.

4. Results

4.1. Pharmaceuticals in water abstracted for drinking water production

Large volumes of water are abstracted daily for the production ofdrinking water; in 2011 daily volumes of (0.28 ± 0.02) ∗ 106 and

(0.21 ± 0.02) ∗ 106 m3 surface water at the Rhine and Meuse plants.Loads of the analyzed pharmaceuticals in the abstracted water were,respectively, 320 ± 165 and 165 ± 82 g daily (Supplemental Material,Fig. S1). The majority of the loads existed of the antidiabetic metformin(approximately half of the load), X-ray contrast agent iopromide andstimulant caffeine (both nearly 10%). Largest loads occurred in spring2011 and 2012 when high abstracted volumes coincided with highconcentrations of metformin. The Polder plant abstracted in 2011(0.09 ± 0.01) ∗ 106 m3 water daily, contaminated with 8 ± 7 g of phar-maceuticals, of which the analgesic phenazone accounted for one third.The Rhine plant received the largest loads of pharmaceuticals of the threeplants, and also encountered the highest concentrations in its intakewater. In 2011, in total 117 kg of the analyzed pharmaceuticals werepresent in the abstracted water, of which 51 kg consisted of metformin.

4.2. Concentrations of individual pharmaceuticals in drinking water

Fig. 1 provides total concentrations of pharmaceuticals in drinkingwater summed according to the therapeutic classes they belong to (in-dividual concentrations are shown in the Supplementary Material,Table S1). The 4-weekly monitoring data of drinking water collectedin 2011 and 2012 show that traces of pharmaceuticals penetratefrom the sources into the prepared drinking water. Concentrations ofpharmaceuticals are higher in drinking water prepared from the riverRhine and Meuse than in water prepared from polder water (Fig. 1).Highest total concentration was 292 ng/L (Meuse plant, May 2012,due to an incidental high concentration of clofibrate). Total concen-trations in drinking water averaged in time are for all investigatedplants ≤ 100 ng/L, with the Polder plant having the lowest concentra-tion in drinkingwater (61± 40 ng/L). Comparison with concentrationsin the sources (data used to calculate loads in Supplemental Material,Fig. S1) indicates that during treatment considerable reduction ofconcentrations of pharmaceuticals is achieved (89–92% for the Rhineand Meuse plants and 23% for the Polder plant).

Average concentrations of individual pharmaceuticals detected indrinking water from the Rhine plant are shown in Fig. 2. Individualpharmaceuticals were often detected at the (low) ng/L range. All inves-tigated pharmaceuticals, except five, were detected in one or moredrinking water samples. The concentration of metformin is 26 ±31 ng/L (Fig. 2), which is 95% less than in the Rhine (Fig. 2).

59C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

4.3. Lifelong exposure of consumers

Calculated lifelong exposure doses of drinking water consumersrange from≤5 μg/70 years for cyclophosphamide, enalapril, ifosfamideandphenazoneup to 1.3 mg/70 years formetformin (selected results inTable 1, complete overview in Supplemental Material, Table S2). Thispicture changeswhen expressed as % of 1 DDD, due to themutual differ-ences in the heights of the DDDs. Highest exposures are now obtainedfor propranolol, paroxetine, furosemide, and caffeine - compoundswith low DDDs between 10 and 40 mg/d. Lifelong consumption to 2 Lof drinking water from the Rhine plant leads to the exposure to 2 ±6% of 1 DDD of propranolol (Table 1). The highest lifelong exposurewas 4 ± 5% of 1 single DDD (value of caffeine in drinking water pre-pared from the river Meuse).

Subsequently, exposures summed over all pharmaceuticals werecalculated (Table 1). The highest lifelong exposure is found fordrinking water from the Rhine; 6 ± 11 mg/70 years, equivalent to notmore than 9 ± 24% of 1 DDD. The calculated exposures for the twoworst case scenarios are also shown in Table 1. Lifelong exposures areconsiderably higher using maximum concentrations instead of averageconcentrations. At the date with the maximum sum, the estimated ex-posure is, averaged over the three plants, 2.6± 0.4 fold higher for expo-sure in mg/70 years and 7.3 ± 1.3 fold higher using maximumconcentrations in the whole period. Expressed as % of DDD, estimatedexposures are respectively 3.8 ± 1.2 and 8.7 ± 3.2 fold higher. This ap-proach delivers for drinking water from the Rhine a lifelong combinedexposure just above one DDD (1.13).

4.4. Assessment of possible health risks

4.4.1. Individual compoundspGLVs for 20 pharmaceuticals could be derived or obtained from

toxicological literature data, for the remaining 20 compounds, datawere lacking and pGLVs had to be derived from therapeutic doses(Supplemental Materials, Table S3). The pGLVs ranged from 2.0 mg/Lfor the analgesic/NSAID naproxen to 0.0005 mg/L for atorvastatin. The

Fig. 3. Calculated Benchmark Quotients indicating the risk of adverse health effects of exposurewater prepared from the Rhine plant. Benchmark Quotients b 0.1 (dotted line) indicate that th

two cytostatics cyclophosphamide and ifosfamide were evaluated asgenotoxic carcinogens (Health Council of the Netherlands, 2008; Inter-national Agency for Research on Cancer, 1981; International Agency forResearch on Cancer, 2012), and as such considered as “non-threshold”chemicals. Their average concentrations were between 0.0 ± 0.1 and0.1 ± 0.2 ng/L. The maximum measured concentrations were 0.5 ng/Lcyclophosphamide and 0.8 ng/L ifosfamide in August 2011 at theRhine plant (Supplemental Material, Table S1). This implies that norfor the individual compounds, neither for the mixture of genotoxicantsthe TTC target value of 0.01 μg/L (Mons et al., 2012) was exceeded.

BQs of the “threshold” chemicals were calculated using averageconcentrations of pharmaceuticals (Supplemental Material, Table S4).Fig. 3 depicts the results for all individual pharmaceuticals in drinkingwater from the Rhine plant. All BQs are b0.01 indicating that for therisk of adverse health effects due to lifelong exposure can be considerednegligibly low and further investigations are not warranted.

4.4.2. RA of combined exposureThepossible risk of exposure to traces ofmultiple pharmaceuticals in

drinking water was estimated applying the CA concept to the deter-mined BQs (lower righter part of Table 1). Mixture BQs using averageconcentrations were 0.006 ± 0.006 or smaller for all plants confirmingformixtures that the drinkingwater produced at the investigated plantsdoes not pose a health risk with respect to contamination with traces ofpharmaceuticals. Usingmaximum concentrations as worst case scenar-ios instead of average concentrations, the mixture BQ was up to 8-foldhigher, but still well below the 0.1 threshold.

4.5. Reduction of risk during drinking water production

To get an impression of the performance of the drinking watertreatment processes with respect to the reduction of risks, mixtureBQs were also calculated for the sources (Fig. 4). Highest mixture BQwas found for the surface water from the Rhine, in agreement withthe relatively high load of pharmaceuticals this river receives (Hutet al., 2013; Ter Laak et al., 2010). However, the mixture BQ is below

to traces of pharmaceuticals due to lifelong consumption of drinking water from drinkingis risk can be considered negligibly low.

Fig. 4. Calculated Benchmark quotients for the total mixture of pharmaceuticals in drinking water sources and drinking water from three drinking water plants. *Significant (p= 0.0003and p b 0.0001, respectively) reduction of BQ with respect to the source water.

60 C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

the threshold (0.04 ± 0.04). During the treatment, the mixture BQ ofthe Rhine water is significantly reduced by 84%. Also in the Meuseplant, reduction of – already negligible – risk is achieved, with 70%.The lower contamination of the Polder plant water is reflected by alow mixture BQ in the source and a reduction in the treatment that islower than those of the other plants.

5. Discussion

5.1. Pharmaceuticals in sources for drinking water production

This study confirmed the presence of pharmaceuticals in the aquaticenvironment, and more specifically in sources for drinking waterproduction, as reported earlier (Houtman et al., 2013; Richardson andTernes, 2011). Except oxacillin, all investigated pharmaceuticals weredetected in at least one of the three sources examined in this study,showing the variety of pharmaceuticals in the environment. The antidi-abeticmetforminwas responsible for half of the load of pharmaceuticalsabstracted from the rivers Rhine andMeuse. This compound is excretedin non-metabolized form, and although bacterial degradation in STPs ishigh, the enormous quantities administered to diabetes type II patients(DDD= 2 g/d) are a cause of detection in the range of μg/L concentra-tions in surface waters, e.g. Houtman et al. (2013) and Scheurer et al.(2009). Due to its degradation during drinkingwater treatment and rel-atively high pGLV (0.22 mg/L) it did not translate into a health risk fordrinking water consumers.

The study reflects that surface water is a more vulnerable sourcefor the preparation of drinking water with respect to contaminationwith traces of anthropogenic micro pollutants than ground water(Bauerlein et al., 2013; Vulliet and Cren-Olivé, 2011). This was exempli-fied by the much lower contamination with pharmaceuticals andconcomitant much lower abstracted loads found for the Polder plant.Whilemetforminwas almost absent in this source (average concentration6 ng/L) the analgesic phenazone (concentration 26±4 ng/L) contributedmost to the abstracted load. The polder water is less influenced by STP ef-fluent as emission route for polar, persistent and high consumption

pharmaceuticals such as metformin than the Rhine and Meuse.Phenazone has been in use for a long time in humanmedicine and veter-inary purposes in the Bethune polder. It could therefore have penetratedinto the ground water in the polder that was abstracted via seepage(Bauerlein et al., 2013). Seepageof surfacewater surrounding theBethunepolder could for the small quantities of other pharmaceuticals detected inthe polder water be an important emission source.

5.2. Exposure and RA

Lifelong exposure to pharmaceuticals via drinking water con-sumption appeared to be very low in comparison with therapeuticdoses. This is in agreement with previous investigations (Vullietand Cren-Olivé, 2011; Webb et al., 2003). Exposure also appearedlow with the simultaneous exposure to multiple pharmaceuticalstaken into account (b10% of 1 DDD). Of course, one has to be awarethat the DDD in the latter case refers to a non-existing “mixturepill” composed of small quantities of all detected pharmaceuticals.Comparison with DDDs is sometimes used as a method to assess therisk associatedwith the exposure. However, DDDs refer to doses neededfor pharmacological effects, mostly in adult patients undergoing acuteexposure, whichmay deviate considerably from doses giving toxicolog-ical effects in e.g. sensitive subpopulations and after chronic exposure(Snyder et al., 2008). Therefore, it was decided to perform a separateRA using pGLVs based on toxicological data such as ADI values. For 20compounds, no such data were found, and there was no alternativethan using pharmacological effect data and therefore UFs were appliedin the ADI derivation. All determined BQs were b0.1 indicating thatlifelong exposure to studied pharmaceuticals via drinking water isnot likely to pose a health risk to consumers. With an average BQ of0.0002 there was even a margin of 500 between the actual value andthe 0.1 threshold. The highest BQs were found for atorvastatin (notdetected, so its RA is based solely on the censored concentrations of ¼the MRL) and gemfibrozil, both carcinogens, due to their very lowpGLVs. Apart from the carcinogenic compounds, no therapeutic groupcould be assigned as in general having higher or lower BQs.

61C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

5.3. Data quality

In this study, a large dataset was at our disposal of measurementsboth in sources and in produced drinking water. The measured concen-trations showed considerable variation in time, indicating that frequentmonitoring gives much more insight in the presence of pharmaceuticalsthan single or fewmeasurements. The opportunity ofworkingwith aver-age concentrations led to estimated exposures and risks that were up to8-fold lower than those based on maximum concentrations, indicatingthat the latter probably leads to substantial overestimation of risks.

5.4. Mixture RA

Possiblemixture effects are often one of the open endsmentioned inRA studies (SpencerWilliams and Brooks, 2012; Villanueva et al., 2014;Vulliet and Cren-Olivé, 2011). Application of the concept of CA in thisstudy confirmed the picture of negligible risk for combined exposure.CA and Independent Action (IA) are the two classical concepts for thecalculation of expected mixture toxicities. CA assumes that, if thechemicals in themixture act according to a similarmechanism of action(MoA), equal levels of effect can be achieved by (partial) replacement ofa component by another component. Its applicability has been shownfor e.g. estrogens with the activation of endogenous estrogen receptorsas the sole MoA (Houtman et al., 2006; Payne et al., 2000) and for amixture of four non-steroidal anti-inflammatory drugs acting by non-polar narcosis (Cleuvers, 2004). IA is applicable when compounds in amixture act by different MoAs. In IA, effects, instead of concentrations,are added. The pharmaceuticals investigated in this study are a largeand very diverse group, somewith similarMoAs, but otherswithmutu-ally different mechanisms. This has as consequence that the outcome ofthemixture BQ calculation contains some uncertainty as the CA conceptis in fact used inappropriately. However, CA is the more conservativeapproach of the two as in CA the added concentrations could togetherexceed the effect threshold, whereas if individual concentrations arebelow thresholds for effects, IA predicts no combined effect. In addition,Boobis et al. (2011) observed that in 6 studies with dissimilarly actingmixtures the effect deviated less than a factor of 4 from the effectpredicted by CA. Moreover, the calculated mixture BQs in our studyare that far below the BQ threshold that if in reality any non-additive/synergistic effects might appear due to dissimilarity in MoA the 0.1threshold is not quickly exceeded. CA appeared thus to be a usefulmeans to get an impression of the risk of combined exposure tomultiplepharmaceuticals in drinking water.

Even although a large number of pharmaceuticals was taken intoaccount, the mixture RA performed here is not complete, as drinkingwater may also contain traces of other contaminants (Houtman et al.,2013), such as metabolites of pharmaceuticals (de Jongh et al., 2012)and sweeteners (Scheurer et al., 2010). Furthermore, we know wecannot even measure a lot of contaminants. It will therefore always bea challenge to be as complete as possible in RA studies using environ-mental monitoring data from target analyses. To this purpose the useof panels of bioassays with a variety of MoAs providing the combinedresponses of compounds in the samples besides target analyses wouldgive valuable information.

5.5. Outlook

Due to their presence, but also due to the sensitivity of the appliedanalysis method, ng/L concentration of pharmaceuticals was detectedin drinkingwater. During treatment considerable reductions of concen-trations and mixture BQs were observed. The resulting mixture BQs indrinking water were for all plants that low that no appreciable healthrisk is foreseen due to exposure to current levels of pharmaceuticals indrinking water. Nevertheless, led by the precautionary principle,drinking water companies anticipate to possible future developments.Climate change (Bannink, 2012) and aging of populations in the

catchments of drinking water sources (Van der Aa et al., 2011) are fac-tors that might cause concentrations of pharmaceuticals in the sourcesto rise in the future. Van der Aa et al. estimated an expected rise of thetotal pharmaceutical use in the Netherlands of 37% by 2050, comparedwith 2007. Margins between (mixture) BQs estimated in this studyand the 0.1 BQ threshold, however, indicate that even such rising ofconcentrations will not immediately lead to health risks. Besidespossible health risks, other factors, such as customer trust, may activatedrinking water companies to take measures. Preventive measures canbe taken to lower concentrations in the intake water, and additionaltreatment steps can be considered to remove more pharmaceuticals,especially e.g. in case of calamities leading to temporary high concentra-tions in the sources.

6. Conclusions

This study described the presence of pharmaceuticals in sources andin drinking water of three specific plants. The lifelong exposure topharmaceuticals via drinking water was estimated to be very low, i.e.in the order of a few mg, in total corresponding to less than 10% ofone DDD, the dose a patient is administered on one day. The risk ofadverse health effects, as calculated by benchmarking measuredconcentrations with pGLVs, appeared to be negligibly low. Applicationof the CA concept confirmed this picture for combined exposure to themixture of pharmaceuticals simultaneously present in drinking water.The investigated treatment plants appeared to reduce the -alreadynegligible- risk up to 80%. The large available monitoring datasetenabled the performance of a more realistic RA showing that workingwith maximum instead of average concentrations may overestimatethe risk considerably.

Acknowledgments

The authors are indebted to Rob ten Broek, Jeroen de Jong, Fred vanSchooten, Trude Rutgrink, Tineke Slootweg and Paul Jansen for theircontributions to this study.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.scitotenv.2014.07.022.

References

Backhaus T, Faust M. Predictive environmental risk assessment of chemical mixtures: aconceptual framework. Environ Sci Technol 2012;46:2564–73.

Bannink A. Jaarraport 2011 Maas. Maastricht, the Netherlands: Association of Riverwaterworks Meuse; 2012. p. 1–56.

Bauerlein P, Hofman J, Ter Laak Th. Vóórkomen en voorkómen van geneesmiddelen inbronnen van drinkwater. BTO.2013.049. Nieuwegein, The Netherlands: KWRWatercycle Research Institute; 2013. p. 1–25.

Benotti MJ, Stanford BD, Wert EC, Snyder SA. Evaluation of a photocatalytic reactormembrane pilot system for the removal of pharmaceuticals and endocrine disruptingcompounds from water. Water Res 2009a;43:1513–22.

Benotti MJ, Trenholm RA, Vanderford BJ, Holady JC, Stanford BD, Snyder SA. Pharmaceu-ticals and endocrine disrupting compounds in US drinking water. Environ Sci Technol2009b;43:597–603.

Boobis A, Budinsky R, Collie S, Crofton K, Embry M, Felter S, et al. Critical analysis of liter-ature on low-dose synergy for use in screening chemical mixtures for risk assess-ment. Crit Rev Toxicol 2011;41:369–83.

Boxall AB, Rudd MA, Brooks BW, Caldwell DJ, Choi K, Hickmann S, et al. Pharmaceuticalsand personal care products in the environment: what are the big questions? EnvironHealth Perspect 2012;120:1221.

Bruce GM, Pleus RC, Snyder SA. Toxicological relevance of pharmaceuticals in drinkingwater. Environ Sci Technol 2010;44:5619–26.

Cleuvers M. Mixture toxicity of the anti-inflammatory drugs diclofenac, ibuprofen,naproxen, and acetylsalicylic acid. Ecotoxicol Environ Saf 2004;59:309–15.

Cunningham VL, Binks SP, Olson MJ. Human health risk assessment from the presence ofhuman pharmaceuticals in the aquatic environment. Regul Toxicol Pharmacol 2009;53:39–45.

Cunningham VL, Perino C, D'Aco VJ, Hartmann A, Bechter R. Human health riskassessment of carbamazepine in surface waters of North America and Europe.Regul Toxicol Pharmacol 2010;56:343–51.

62 C.J. Houtman et al. / Science of the Total Environment 496 (2014) 54–62

Daughton CG, Ternes TA. Pharmaceuticals and personal care products in the environ-ment: agents of subtle change? Environ Health Perspect 1999;107(Suppl. 6):907–38.

de Jongh CM, Kooij PJF, de Voogt P, ter Laak TL. Screening and human health risk assess-ment of pharmaceuticals and their transformation products in Dutch surface watersand drinking water. Sci Total Environ 2012;427-428:70–7.

De Moel PJ, Verberk JQJC, Van Dijk JC. Drinkwater — principes en praktijk. The Hague:Delft Technical University/SDU Publishers B.V.; 2004.

Du B, Price AE, Scott WC, Kristofco LA, Ramirez AJ, Chambliss CK, et al. Comparison ofcontaminants of emerging concern removal, discharge, and water quality hazardsamong centralized and on-site wastewater treatment system effluents receivingcommon wastewater influent. Sci Total Environ 2014;466:976–84.

Godfrey E, Woessner WW, Benotti MJ. Pharmaceuticals in on-site sewage effluent andground water, Western Montana. Ground Water 2007;45:263–71.

Halling-Sorensen B, Nielsen SN, Lanzky PF, Ingerslev F, Lutzhoft HCH, Jorgensen SE.Occurrence, fate and effects of pharmaceutical substances in the environment — areview. Chemosphere 1998;36:357–94.

Health Council of the Netherlands. Ifosfamide. Evaluation of the carcinogenicity andgenotoxicity. 2008/06OSH. The Hague, The Netherlands: Health Council of theNetherlands; 2008.

Houtman CJ. Emerging contaminants in surface waters and their relevance for theproduction of drinking water in Europe. J Integr Environ Sci 2010;7:271–95.

Houtman CJ, ten Broek R, de Jong K, Pieterse B, Kroesbergen J. A multi-component‘snapshot’ of pharmaceuticals and pesticides in the river Meuse basin. EnvironToxicol Chem 2013;32:2449–59.

Houtman CJ, Van Houten YK, Leonards PE, Brouwer A, Lamoree MH, Legler J. Biological val-idation of a sample preparation method for ER-CALUX bioanalysis of estrogenic activityin sediment using mixtures of xeno-estrogens. Environ Sci Technol 2006;40:2455–61.

Hut R, Van de Giesen N, Houtman CJ. Medicinal Footprint of the population of the Rhinebasin. Environ Res Lett 2013;8. [044057-, 7 pp.].

International Agency for Research on Cancer. IARC monographs on the evaluation of thecancer risk of chemicals to humans. Some antineoplastic and immunosuppressiveagents, vol. 26. Lyon France: IARC; 1981. p. 1–396 [Geneva, Switzerland, WorldHealth Organization Press].

International Agency for Research on Cancer. IARC monographs. Pharmaceuticals. Areview of human carcinogens, vol. 100A. Lyon France: IARC; 2012. p. 1–462 [Geneva,Switzerland, World Health Organization Press].

Johnson AC, Juergens MD, Williams RJ, Kuemmerer K, Kortenkamp A, Sumpter JP. Docytotoxic chemotherapy drugs discharged into rivers pose a risk to the environmentand human health? An overview and UK case study. J Hydrol 2008;348:167–75.

Jones OA, Lester JN, Voulvoulis N. Pharmaceuticals: a threat to drinking water? TrendsBiotechnol 2005;23:163–7.

Kasprzyk-Hordern B, Dinsdale RM, Guwy AJ. Multiresidue methods for the analysis ofpharmaceuticals, personal care products and illicit drugs in surface water and waste-water by solid-phase extraction and ultra performance liquid chromatography–electrospray tandem mass spectrometry. Anal Bioanal Chem 2008;391:1293–308.

Kolpin DW, Furlong ET, MeyerMT, Thurman EM, Zaugg SD, Barber LB, et al. Pharmaceuticals,hormones, and other organic wastewater contaminants in US streams, 1999–2000: anational reconnaissance. Environ Sci Technol 2002;36:1202–11.

Lekkerkerker-Teunissen K, Chekol ET, Maeng SK, Ghebremichael K, Houtman CJ, VerliefdeARD, et al. Pharmaceutical removal during managed aquifer recharge with pretreat-ment by advanced oxidation. Water Sci Technol Water Supply 2012;12:755–67.

MonsMN, HeringaMB, van Genderen J, Puijker LM, BrandW, van Leeuwen CJ, et al. Use ofthe Threshold of Toxicological Concern (TTC) approach for deriving target values fordrinking water contaminants. Water Res 2012;47:1666–78.

Payne J, Rajapakse N, Wilkins M, Kortenkamp A. Prediction and assessment of the effectsof mixtures of four xenoestrogens. Environ Health Perspect 2000;108:983–7.

Richardson SD, Ternes TA. Water analysis: emerging contaminants and current issues.Anal Chem 2011;83:4614–48.

ScheurerM, Sacher F, BrauchHJ. Occurrence of the antidiabetic drugmetformin in sewageand surface waters in Germany. J Environ Monit 2009;11:1608–13.

Scheurer M, Storck FR, Brauch HJ, Lange FT. Performance of conventional multi-barrierdrinking water treatment plants for the removal of four artificial sweeteners. WaterRes 2010;44:3573–84.

Schriks M, Heringa MB, van der Kooi MM, de VP, van Wezel AP. Toxicologicalrelevance of emerging contaminants for drinking water quality. Water Res2009;44:461–76.

Snyder SA, Trenholm RA, Snyder EM, Bruce GM, Pleus RC, Hemming JDC. Toxicologicalrelevance of EDCs and pharmaceuticals in drinking water. Denver, California, U.S.A.:Awwa Research Foundation; 2008

SpencerWilliams E, Brooks BW. Human health risk assessment for pharmaceuticals in theenvironment: existing practice, uncertainty, and future directions. In: Brooks BW,Huggett DB, editors. Human pharmaceuticals in the environment. Current and futureperspectivesNew York: Springer; 2012. p. 167–224.

Stanford BD, Weinberg HS. Evaluation of on-site wastewater treatment technology toremove estrogens, nonylphenols, and estrogenic activity from wastewater. EnvironSci Technol 2010;44:2994–3001.

Ter Laak TL, van der Aa M, Houtman CJ, Stoks PG, van Wezel AP. Relating environmentalconcentrations of pharmaceuticals to consumption: a mass balance approach for theriver Rhine. Environ Int 2010;36:403–9.

Van der Aa NGFM, Kommer GJ, Van Montfoort JE, Versteegh JFM. Demographicprojections of future pharmaceutical consumption in the Netherlands. Water SciTechnol 2011;63:825–31.

Versteegh JFM, Van der Aa NGFM, Dijkman E. Geneesmiddelen in drinkwater endrinkwaterbronnen. Resultaten van het meetprogramma 2005/2006. RIVM Report703719016/2007Bilthoven: Dutch Ministry for Public Health and the Environment;2007.

Villanueva CM, Kogevinas M, Cordier S, Templeton MR, Vermeulen R, Nuckols JR, et al.Assessing exposure and health consequences of chemicals in drinking water: currentstate of knowledge and research needs. Environ Health Perspect 2014;122(3):213–21.

Vulliet E, Cren-Olivé C. Screening of pharmaceuticals and hormones at the regional scale,in surface and groundwaters intended to human consumption. Environ Pollut 2011;159:2929–34.

Walpole S, Prieto-Merino D, Edwards P, Cleland J, Stevens G, Roberts I. The weightof nations: an estimation of adult human biomass. BMC Public Health 2012;12:439.

Webb S, Ternes T, Gibert M, Olejniczak K. Indirect human exposure to pharmaceuticals viadrinking water. Toxicol Lett 2003;142:157–67.

World Health Organization. Guidelines for drinking-water quality. incorporating first andsecond addenda, vol. 1. Geneva, Switzerland: World Health Organization; 2006.

World Health Organization. Pharmaceuticals in drinking water. Geneva, Switzerland:World Health Organization; 2012.