Toxicology 198 (2004) 255–262

Risk analysis of drinking water microbial contaminationversus disinfection by-products (DBPs)

Nicholas John Ashbolt∗

CRC for Water Quality and Treatment, School of Civil and Environmental Engineering, UNSW, Sydney, NSW 2052, Australia

Abstract

Managing the provision of safe drinking water has a renewed focus in light of the new World Health Organization (WHO) watersafety plans. Risk analysis is a necessary component to assist in selecting priority hazards and identifying hazardous scenarios,be they qualitative to quantitative assessments. For any approach, acute diarrhoeal pathogens are often the higher risk issuefor municipal water supplies, no matter how health burden is assessed. Furthermore, potential sequellae (myocarditis, diabetes,reactive arthritis and cancers) only further increase the potential health burden of pathogens; despite the enormous uncertaintiesin determining pathogen exposures and chemical dose-responses within respective microbial and chemical analyses. Theseinterpretations are currently being improved by Bayesian and bootstrapping approaches to estimate parameters for stochasticassessments. A case example, covering the health benefits of ozonation forCryptosporidium inactivation versus potential cancersfrom bromate exposures, illustrated the higher risks from a pathogen than one of the most likely disinfection by-products(DBPs). Such analyses help justify the industries long-held view of the benefits of multiple barriers to hazards and that microbialcontamination of water supplies pose a clear public health risk when treatment is inadequate. Therefore, efforts to reduce potentialhealth risks from DBP must not compromise pathogen control, despite socio-political issues.© 2004 Elsevier Ireland Ltd. All rights reserved.

Keywords: Waterborne pathogens; Risk assessment; DBP; QMRA

1. Introduction

Traditionally, the human health impact of drinkingwaters has been assessed via epidemiology (Hunteret al., 2002), and in both developing and developedregions, pathogens are generally considered a higherhealth risk than chemicals (Craun, 1993; Downs et al.,1999); with arsenic being a notable exception fromsome groundwaters. Yet, given the relatively low lev-

∗ Tel.: +61-2-9385-5946; fax:+61-2-9385-6139.E-mail address: n.ashbolt@unsw.edu.au (N.J. Ashbolt).

els of risks still considered important (in the rangeof one annual infection per 1000–10,000 consumers;Haas, 1996), it is both very costly and difficult todemonstrate epidemiologic links between drinkingwater contaminants and health outcomes. There hastherefore been an increasing interest in the applicationof quantitative risk assessment (National Academyof Sciences, 1983) (formalized process outlined inTable 1andFig. 1), for both chemical and microbialhazard assessment (Haas et al., 1999; Benford, 2001).Furthermore, risk assessment is the cornerstone ofthe recent guideline initiatives from the World Health

0300-483X/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.tox.2004.01.034

256 N.J. Ashbolt / Toxicology 198 (2004) 255–262

Table 1Risk assessment paradigm for human health effects (adapted fromHaas et al., 1999)

Step Aim

Problem formalization andhazard identification

To describe acute and chronic human health effects associated with any particular hazard, includingtoxicity, carcinogenicity, mutagenicity, developmental toxicity, reproductive toxicity, and neurotoxicity

Exposure assessment To determine the size and nature of the population exposed and the route, amount, and duration ofthe exposure

Dose-response assessment To characterize the relationship between various doses administered and the incidence of the healtheffect

Risk characterization To integrate the information from exposure, dose-response, and health steps in order to estimate themagnitude of the public health problem and to evaluate variability and uncertainty

Chemical and Microbial Risk Assessment

PROBLEM FORMULATION and SOURCE CHARACTERIZATION

Characterizationof

Exposure

Characterizationof

Effects

RISK CHARACTERIZATION

RISK MANAGEMENT

Dat

a A

cqui

sitio

n,V

erifi

catio

n, M

onito

ring

Ris

k C

omm

unic

atio

n

ANALYSIS

Fig. 1. Generic approach to undertaking chemical or microbial risk assessment (adapted fromUS-EPA, 1989).

Organization (WHO) (Fewtrell and Bartram, 2001;Davison et al., 2003) and allows rationale comparisonof risk across a wide range of hazards and hazardousevents (such as rain induced pathogen/turbidity in-creases). In addition, application of disability adjustedlife years (DALY) provides a means to compare healthoutcomes for very different hazards (Murray andLopez, 1996), such as pathogens versus disinfectionby-products (DBPs) (Havelaar et al., 2000a).

2. Hazard identification (key pathogens anddisinfection by-products)

Some 150 known waterborne pathogens fit into fourmajor groups, which due to difficulty/cost in mea-surement are often assessed by means of index or-ganisms (Table 2) (Ashbolt et al., 2001) during theexposure assessment phase of risk assessment. En-teric viruses are increasingly seen as the most impor-tant pathogens in drinking waters in developed regions(from a health outcome point of view), versusCryp-

tosporidium that is recognized to cause the majorityof outbreaks in developed regions. In developing re-gions, enteric bacteria such asV. cholerae andCampy-lobacter spp. are probably more important (Ashbolt,2004). Yet for those who live beyond their childhoodyears, in addition to gastrointestinal diseases, thereare various potential sequellae following enteric infec-tions, such as type 1 diabetes mellitus from coxsackieA and echo viruses (Frisk et al., 1992), myocardi-tis from various enteric viruses and reactive arthritisand Guillain–Barré syndrome following campylobac-ter infections (Niklasson et al., 1998; Havelaar et al.,2000b). It is therefore no surprise that a disease burdenapproach is being adopted rather than simply identify-ing potential infections via exposure to contaminateddrinking waters (Melse and Havelaar, 2002).

Of the disinfection by-products, there are over400 potential chemical species (Fawell et al., 1997;Krasner, 1999), but like pathogens, these hazardsare grouped into several major classes (Table 3).The first DBPs reported in drinking water were tri-halomethanes (THMs) (Bellar et al., 1974; Rook,

N.J. Ashbolt / Toxicology 198 (2004) 255–262 257

Table 2Major groups of disease-causing agents, examples and possible index representatives for their environmental forms

Group Example Index Environmental stage, size (�m) and shape

Virus Rotavirus F-RNA bacteriophagea Virion (0.02–0.08) sphericalBacteria Salmonella typhi E. coli Cell or dormant cell (0.1–2) cocci-rodParasitic protozoa Cryptosporidium parvum Clostridium perfringensb Cyst or oocyst (4–40) oval-sphericalHelminth Ascaris lumbricoides Ascaris Ova (egg, 30–80) variable

a As an index of persistence and transport in the environment, not numbers in the source material for which adenoviruses are probablythe most numerous culturable human enteric virus group.

b If the source is predominantly human sewage.

1974). THMs were shown to be produced by chlori-nation reactions with natural organic matter in water,such as humic and fulvic acids. The next major groupof chlorination DBPs identified were the haloaceticacids (HAAs), and both of these DBP groups form thebasis of existing DBP regulations globally. Yet it nowlooks like we have the wrong DBP indicators. Despitedecades of research there is no clear evidence thatthe very low concentrations of either THMs or HAAsin drinking waters are the primary agents responsiblefor increased renal, bowel and other cancers proba-bly resulting from exposures to chlorinated drinkingwater (Hrudey, 2003).

For example, evaluating dose-response relation-ships with specific THM or HAA by-products doesnot increase the odds ratio for bladder cancer, al-though THMs and HAAs increase with increasingtotal organic carbon in waters. Even for organic richrecycled water the concentration of DBPs and es-timated THM cancer risks are low (Table 4). In ameta-analysis of epidemiological data,Morris (1995)looked at the statistically significant associations forbladder cancers (relative risk(RR) = 1.21) and rec-

Table 3Disinfectants and disinfection by-products (adapted fromICPS, 2000)

Disinfectant Significant organohalogen DBP Significant inorganic DPBs Significant non-halogenated DPBs

Chlorine THM, HAA, HAN, CH, CP, CPh,N-chloramines, halofuranones,bromohydrins

Chlorate (mostly from hypochlorite use) Aldehydes, cyanoalkanoic acids,alkanoic acids, benzene,carboxylic acids

Chlorine dioxide Chlorite, chlorate UnstudiedChloramine HAN, cyanogens chloride, organic

chloramines, CH, chloramino acids,haloketones

Nitrate, nitrite, chlorate, hydrazine Aldehydes, ketones, nitrosamines

Ozone Bromoform, MBA, DBA,dibromoacetone, cyanogens bromide

Chlorate, iodate, bromate, hydrogenperoxide, HOBr, epoxides, ozonates

Aldehydes, ketoacids, ketones,carboxylic acids

THM: trihalomethanes, HAA: haloacetic acids, HAN: haloacetonitriles, HK: haloketones, MBA-CP: chlorophenols, CH: chloral hydrade.

tal cancers (RR= 1.38) from drinking US waters.If a causal link existed between these cancers andDBP exposure then about 8% of bladder cancers and18% of rectal cancers in the US may be attributableto drinking chlorinated surface waters (equivalent toabout 4200 cases of bladder cancer and 6500 casesof rectal cancer per year). But such meta-analyses arecriticized base on the likelihood of chance associa-tions (Egger et al., 1998).

Hence, what is clear is that precursor content andcharacter are important for DBP production, but thatthe specific compounds of concern can only be spec-ulated. Further, more reactive organics, such as humicacids are considered a greater risk than slower-reactingones (e.g. fulvic acids). Recent evidence points tobrominated compounds and nitrosamines (e.g. ni-trosodimethylamine, NDMA) as the carcinogen DBPsof concern (Bull, 2003). Nitrosamines at concentra-tions of 0.1–1�g/l could account for the reportedbladder cancer risks, along with aromatic aminesand nitrogen heterocycles. Should DBPs like NDMAprove to be the carcinogens in drinking water it hasmajor ramifications for the water industry, as NDMA

258 N.J. Ashbolt / Toxicology 198 (2004) 255–262

Table 4Carcinogenic risk assessment for DBP at Aqua III advanced wastewater treatment works (adapted fromOlivieri et al., 1998)

Chemical Concentration (�g l−1) Cancer risk

50th percentile Mean 95th percentile

Bromodichloromethane 0.48 1.3× 10−7 4.9 × 10−7 1.8 × 10−6

Chloroform 0.51 6.1× 10−9 2.2 × 10−8 8.9 × 10−8

Chloromethane 0.11 1.6× 10−9 6.1 × 10−9 2.5 × 10−8

Assumed 2 l per day consumption of water. Slope factors derived form a linearized multistage model with a 95% upper confidence limit(US-EPA IRIS database) used to estimate risk; median value of the cancer potency distribution is approximately 0.072 times the EPAupper confidence limited, and the distribution was considered to be reasonably represented by a log-normal distribution with a standarddeviation of 4.93 (Hattis and Wasson, 1987).

is generated during chloramination of drinking waters(chlorine + ammonium yielding monochloramine),but not during chlorination (Najm and Trussell,2001); noting that many agencies have changed fromchlorination to chloramination in an effort to reduceTHM/HAA DBPs. However, due to a lack of sensi-tive enough analytical methods generally used by thewater industry (Ells et al., 1999), there is very littleinformation on nitrogen precursors (other than addedammonium) in source waters within the sub-�g/lrange.

In summary, it is generally agreed that DBPs haz-ards may cause cancers and potentially birth defects(King et al., 2000) following exposures to disinfecteddrinking waters, but a range of environmental fac-tors are also important, including type of organic mat-ter, disinfection type and dose, temperature, pH, andbromine concentration.

3. Dose-response assessment

Fig. 2provides an indication of the relative toxicitiesof different DBPs, and it is clear why there is mount-ing concern over compounds like NDMA. However,in contrast to acute effects from exposure to pathogens(Haas et al., 1999), Bull (2003) has proposed a farmore complex interaction of chemical, environmen-tal, genetic and lifestyle factors that may need to befulfilled for cancer onset (Fig. 3). The consequenceof such a model to DBP risk assessment is currentlyrather daunting, and as far as the author is concerned,has yet to be attempted. Current dose-response modelsfor carcinogens assume no threshold, which has leadto a water industry goal of zero DBPs. Knowledge of

modes of actions for THMs however, would suggestthat the low levels encounter in drinking waters pro-vide an insignificant environmental pathway for THMslike chloroform (Table 4). The dose-response modelfor genotoxic carcinogenic contaminants involves fit-ting the linearized multistage model to data from ani-mal studies (WHO, 1996). The limit concentration isthen derived from the upper 95% confidence bound ofthe fitted model.

On the other hand, microbial risk assessment pro-cedures have recently improved, and dose-responsemodels are based on human feeding trials (Haas et al.,1999). Looking at the dose-response curves in moredetail, the single hit models (e.g. only one hazardparticle needed) were first used in the fifties to de-scribe processes associated with radioactivity, andwas based on a simple exponential relationship whichholds for some pathogens (Eq. (1)). In the sixties,two virologists developed what is currently termedthe “beta-Poisson” model to describe tobacco mosaicvirus infection via plant leaves (Eq. (2); Furumoto andMickey, 1967). Since the introduction of quantitative

Fig. 2. Relative potency of DBPs as carcinogens (Bull, 2003).

N.J. Ashbolt / Toxicology 198 (2004) 255–262 259

Fig. 3. Alternate paths to bladder cancer (Bull, 2003).

microbial risk assessment to drinking water regula-tions in the US (Regli et al., 1991), it has now evolvedto be a major tool for assessing health outcomes inthe harmonization of the World Health Organization’swater quality guidelines (Teunis and Havelaar, 1999;WHO, 1999; Bartram et al., 2001).

Exponential model : probabilityinfection

= 1 − exp(−rD) (1)

where D is the pathogen dose;r the fraction ofpathogens that survives to produce an infection.

Beta-Poisson model : probabilityinfection

= 1 −(

1 +(

D

β

))−α

(2)

whereD is the pathogen dose;α andβ are parametersof the beta-distribution used to describe variability ininfectivity.

Whilst the conceptual framework for both chemicaland microbial risk assessments is the same (Fig. 1),pathogens differ from toxic chemicals in several keyways by their:

• variability within strains of the one pathogen tocause disease (differing virulence);

• virulence that can also evolve as the pathogen passesthrough various infected individuals;

• generally not evenly distribution in water;• ability to be passed from one person to many (sec-

ondary spread), from either healthy but infected(asymptomatic) or ill (symptomatic) hosts (e.g. ra-tio of secondary to primary cases of 0.33 forCryp-tosporidium parvum to over 1.0 forGiardia lambliaand Norwalk virus (Haas et al., 1999)); and

• whether a person becomes infected or ill dependsnot only on the health of the person, but also ontheir pre-existing immunity and pathogen dose.

4. Hypothetical case study

Havelaar et al. (2000a)were one of the first todemonstrate a thorough approach to estimate healthburden and compare apples and oranges, being hy-pothetical ill effects expected fromCryptosporidiumin drinking water versus its removal by ozonation butproduction of the carcinogenic DBP, bromate. Theyused the common metric of DALYs to account formorbidity and mortality and based their theoretical as-sessment on health statistics from The Netherlands.

Oocysts ofCryptosporidium parvum are highly re-sistant to water treatment and considered by the water

260 N.J. Ashbolt / Toxicology 198 (2004) 255–262

industry to be one of the most important waterbornepathogens following numerous waterborne outbreaks(Smith and Rose, 1998). While ingestion of oocystsmay cause self-limiting gastroenteritis in immuno-competent individuals, it can cause life-threateningdiarrhea in those immunocompromized, particularlyHIV-AIDS patients or the malnourished in developingregions.

Unlike chlorine alone, ozone is very effective in in-activatingCryptosporidium oocysts (Corona-Vasquezet al., 2002). The problem with ozonation, however,is that various DBPs are produced, of which bromateis considered the most important. Bromate has beenshown to induce tumors in rat kidney, thyroid andmesothelium, along with being a renal carcinogen inmice (Havelaar et al., 2000a).

A probabilistic approach was used byHavelaaret al. (2000a)to model exposure toCryptosporid-ium and bromate. Such an approach is necessary asthe hazards are generally too low for reliable di-rect detection in drinking waters, and it also allowsan estimation of uncertainty and variability. Hence,modeling of infectious oocysts in drinking water withor without ozone treatment, through Monte Carlosampling of oocyst distributions, provided a prob-ability density function (PDF) of oocyst in water.Assuming a PDF for the ingestion rate of drinkingwater (median 800 ml per person per day), doseswere then applied to a dose-response model. ForCryptosporidium, Teunis et al. (1996)have shownthat the exponential model (Eq. (1)) is suitable,with bootstrapping providing uncertainty in the in-fection parameter,r. It was assumed that a medianof 71% of the general population infected wouldget ill, versus 100% for AIDS patients (Perz et al.,1998). To obtain a full uncertainty distribution ofthe probability of cancer from bromate, the pointestimate of risk should be the median, rather thana conservative upper estimate, as used byWHO(1996). Havelaar et al. (2000a)used the followingtwo-stage model with available data from animalstudies:

Probability of renal cell cancer

= a + (1 − a)(1 − e−bidose−c(bidose)2) (3)

where dose (�g KBrO3 (kg−1 body weight)2/3), a(background response) andc (shape) are common pa-

rameters for all data sets, andbi is the dose-responseparameter fitted to data seti.

In their hypothetical,Havelaar et al. (2000a)esti-mated that the median risk of infection withC. parvumwas 10−3 per person per year without ozonation. Withozonation, the median risk from oocysts was reducedby about seven-fold, but bromate would be produced atlevels above current WHO guideline levels. Nonethe-less, the health benefits of preventing gastroenteritis inthe general population and in patients with acquiredimmunodeficiency syndrome outweighed health lossesby premature death from renal cell cancer by a factorof >10. Therefore, the net benefit from ozonation wasapproximately 1 DALY per million per person-years.Noting, however, that the application of DALYs washampered by a substantial degree of uncertainty, as istypical for most risk assessments.

5. Conclusions

Application of risk assessment to the DBP ver-sus pathogen debate has helped justify the industrieslong-held view of the benefits of multiple barriers tohazards and that microbial contamination of watersupplies pose a clear public health risk when treatmentis inadequate. Therefore, efforts to reduce potentialhealth risks from DBP must not compromise pathogencontrol, despite socio-political issues. Furthermore,current regulatory effort is directed at reducing THMsand HAAs, both of which may be very poor indicesfor adverse health outcomes from DBPs.

References

Ashbolt, N.J., 2004. Microbial contamination of drinking waterand disease outcomes in developing regions. Toxicology, 198,229–238.

Ashbolt, N.J., Grabow, W.O.K., Snozzi, M., 2001. Indicatorsof microbial water quality. In: Fewtrell, L., Bartram, J.(Eds.), Water Quality: Guidelines, Standards and Health. RiskAssessment and Management for Water-Related InfectiousDisease. IWA Publishing, London, Chapter 13, pp. 289–315.

Bartram, J., Fewtrell, L., Stenström, T.A., 2001. Harmonisedassessment of risk and risk management for water-relatedinfectious disease: an overview. In: Water Quality: Guidelines,Standards and Health. Risk Assessment and Management forWater Related Infectious Diseases. IWA Publishing, London,pp. 1–16.

N.J. Ashbolt / Toxicology 198 (2004) 255–262 261

Bellar, T.A., Lichtenberg, J.J., Kroner, R.C., 1974. The occurrenceof organohalides in chlorinated drinking water. J. Am. WaterWorks Assoc. 66 (12), 703–706.

Benford, D., 2001. Principles of risk assessment of food anddrinking water related to human health. International LifeSciences Institute Brussels.

Bull, R.J., 2003. Are there significant health effects associated withthe use of chemical disinfection of drinking water? In: Sinclair,M. (Ed.), Report on DBPs and Health Effects Seminar andWorkshop, 29–31 October 2001, Melbourne, CRC for WaterQuality and Treatment, Adelaide.

Corona-Vasquez, B., Samuelson, A., Rennecker, J.L., Mariñas,B.J., 2002. Inactivation ofCryptosporidium parvum oocystswith ozone and free chlorine. Water Res. 36 (16), 4053–4063.

Craun, G.F., 1993. Safety of water disinfection: balancing chemicaland microbial risks. International Life Science Institute,Washington, DC.

Davison, A., Howard, G., Stevens, M., Callan, P., Kirby, R.,Deere, D., Bartram, J., 2003. Water safety plans. World HealthOrganization, Geneva.

Downs, T.J., Cifuentes-Garcia, E., Suffet, I.M., 1999. Riskscreening for exposure to groundwater pollution in awastewater irrigation district of the Mexico City region.Environ. Health Perspect. 107 (7), 553–561.

Egger, M., Schneider, M., Smith, G.D., 1998. Meta-analysis:spurious precision? Meta-analysis of observational studies. Br.Med. J. 316 (7125), 140–144.

Ells, B., Barnett, D.A., Froese, K., Purves, R.W., Hrudey,S., Guevremont, R., 1999. Detection of chlorinated andbrominated byproducts of drinking water disinfection usingelectrospray ionization-high-field asymmetric waveform ionmobility spectrometry-mass spectrometry. Anal. Chem. 71,4747–4752.

Fawell, J., Robinson, D., Bull, R., Birnbaum, L., Boorman, G.,Butterworth, B., Daniel, P., Galal-Gorchev, H., Hauchman,F., Julkunen, P., Klaassen, C., Krasner, S., Orme-Zavaleta, J.,Reif, J., Tardiff, R., 1997. Disinfection by-products in drinkingwater: critical issues in health effects research. Environ. HealthPerspect. 105 (1).

Fewtrell, L., Bartram, J., 2001. Water Quality: Guidelines,Standards and Health. Risk Assessment and Management forWater Related Infectious Diseases. IWA Publishing, London.

Frisk, G., Nilsson, E., Tuvemo, T., Friman, G., Diderholm, H.,1992. The possible role of coxsackie A and echo viruses inthe pathogenesis of type 1 diabetes mellitus studied by IgManalysis. J. Infect. 24, 13–22.

Furumoto, W.A., Mickey, R.A., 1967. Mathematical model for theinfectivity-dilution curve of tobacco mosaic virus: theoreticalconsiderations. Virology 32, 216–223.

Haas, C.N., 1996. Viewpoint: acceptable microbial risk. J. Am.Water Works Assoc. 88 (12), 8.

Haas, C.N., Rose, J.B., Gerba, C.P., 1999. Quantitative MicrobialRisk Assessment. Wiley, New York.

Hattis, D., Wasson, J., 1987. A pharmacokinetic/mechanisms-basedanalysis of the carcinogenic risk of butadiene. MIT Center forTechnology, Policy and Industrial Development, CTPID 87-3,Boston.

Havelaar, A.H., De Hollander, A.E.M., Teunis, P.F.M., Evers, E.G.,Van Kranen, H.J., Versteegh, J.F.M., Van Koten, J.E.M., Slob,W., 2000a. Balancing the risks and benefits of drinking waterdisinfection: disability adjusted life-years on the scale. Environ.Health Perspect. 108 (4), 315–321.

Havelaar, A.H., de Wit, M.A., van Koningsveld, R., van Kempen,E., 2000b. Health burden in The Netherlands due to infectionwith thermophilicCampylobacter spp. Epidemiol. Infect. 125,505–522.

Hrudey, S.E., 2003. Drinking water disinfection by-products: when,what and why? In: M. Sinclair (Ed.), Report on DBPs andHealth Effects Seminar and Workshop, 29–31 October 2001,Melbourne, CRC Water Quality and Treatment, Adelaide.

Hunter, P.R., Waite, M., Ronchi, E., 2002. Drinking Water andInfectious Disease: Establishing the Links. IWA Publishing,London.

ICPS 2000. Disinfectants and Disinfectant By-Products. WorldHealth Organization, Geneva.

King, W.D., Dodd, L., Allen, A.C., 2000. Relation betweenstillbirth and specific chlorination by-products in public watersupplies. Environ. Health Perspect. 108 (9), 883–886.

Krasner, S.W., 1999. Chemistry of disinfection by-productformation. In: Singer, P.C. (Ed.), Formation and Control ofDisinfection By-Products in Drinking Water, American WaterWorks Association, Denver, CO, Chapter 2, pp. 197–210.

Melse, J.M., Havelaar, A.H. 2002. Quantifying public health risksin guidelines for drinking-water quality: a burden of diseaseapproach. RIVM, Bilthoven.

Morris, R.D., 1995. Drinking water and cancer. Environ. HealthPerspect. 103, 225–232.

Murray, C.J.L., Lopez, A.D., 1996. The Global Burden ofDisease, Global Health Statistics: A Compendium of Incidence,Prevalence and Mortality Estimates for Over 200 Conditions,vol. II. Harvard School of Public Health on behalf of the WorldHealth Organization and The World Bank, Cambridge, MA.

Najm, I., Trussell, R.R., 2001. NDMA formation in water andwastewater. J. Am. Water Works Assoc. 93 (2), 92–99.

National Academy of Sciences, 1983. Risk Assessment in theFederal Government: Managing the Process. National AcademyPress, Washington, DC.

Niklasson, B., Hörnfeldt, B., Lundman, B., 1998. Couldmyocarditis, insulin-dependent diabetes mellitus and Guillain–Barré, syndrome be caused by one or more infectious agentscarried by rodents? Emerging Infect. Dis. 4 (2), 187–193.

Olivieri, A.W., Eisenberg, D.M., Cooper, R.C., 1998. City of SanDiego health effects study on potable water reuse. In: Asano,T. (Ed.), Wastewater Reclamation and Reuse. TechnomicPublishing Co. Inc., Lancaster, PA, pp. 521–579.

Perz, J.F., Ennever, F.K., Leblancq, S.M., 1998.Cryptosporidiumin tap water—comparison of predicted risks with observedlevels of disease. Am. J. Epidemiol. 147 (3), 289–301.

Regli, S., Rose, J.B., Haas, C.N., Gerba, C.P., 1991. Modeling therisk from Giardia and viruses in drinking water. J. Am. WaterWorks Assoc. 83 (11), 76–84.

Rook, J.J., 1974. formation of haloforms during chlorination ofnatural water. Water Treatment Examination 23, 234–243.

262 N.J. Ashbolt / Toxicology 198 (2004) 255–262

Smith, H.V., Rose, J.B., 1998. Waterborne cryptosporidiosis—current status. Parasitol. Today 14 (1), 14–22 (review).

Teunis, P.F.M., Havelaar, A.H., 1999.Cryptosporidium in drinkingwaters: evaluation of the ILSI/RSI quantitative risk assessmentframework. National Institute of Public Health and theEnvironment (RIVM), Bilthoven, The Netherlands, p. 97.

Teunis, P.F.M., van der Heijden, O.G., van der Giessen, J.W.B.,Havelaar, A.H., 1996. The dose response relation in humanvolunteers for gastro-intestinal pathogens. National Institute ofPublic Health and the Environment (RIVM), Bilthoven.

US-EPA, 1989. Risk assessment guidance for superfund. In:Human Health Evaluation Manual, Part A. Interim Final, Officeof Emergency and Remedial Response. US EnvironmentalProtection Agency, Washington, DC.

WHO, 1996. Guidelines for drinking-water quality. World HealthOrganization, Geneva, p. 973.

WHO, 1999. Draft principles and guidelines for the conduct ofmicrobiological risk assessment. World Health Organization,Geneva.