Articulo3

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

Review

Chlorination disinfection by-products, public health risktradeoffs and me

Steve E. Hrudeya,b,*aSteve E. Hrudey & Associates Ltd, Canmore, Canada T1W 3C2bDepartment of Laboratory Medicine & Pathology, University of Alberta, 10-102 Clinical Sciences Building, Edmonton, Canada T6G 2G3

a r t i c l e i n f o

Article history:

Received 5 September 2008

Received in revised form

5 February 2009

Accepted 9 February 2009

Published online 20 February 2009

Keywords:

Risk assessment

Risk management

Drinking water safety

Cancer

Carcinogens

Adverse reproductive effects

Disinfection by-products

Environmental epidemiology

Environmental toxicology

* Department of Laboratory Medicine & PatTel.: þ1 780 492 6807; fax: þ1 780 492 6382.

E-mail address: [email protected]/$ – see front matter ª 2009 Elsevidoi:10.1016/j.watres.2009.02.011

a b s t r a c t

Since 1974 when trihalomethanes (THMs) were first reported as disinfection by-products

(DBPs) in drinking water, there has been an enormous research effort directed at under-

standing how DBPs are formed in the chlorination or chloramination of drinking water,

how these chlorination DBPs can be minimized and whether they pose a public health risk,

mainly in the form of cancer or adverse reproductive outcomes. Driven by continuing

analytical advances, the original DBPs, the THMs, have been expanded to include over 600

DBPs that have now been reported in drinking water. The historical risk assessment

context which presumed cancer could be mainly attributed to exposure to environmental

carcinogens played a major role in defining regulatory responses to chlorination DBPs

which, in turn, strongly influenced the DBP research agenda. There are now more than 30

years of drinking water quality, treatment and health effects research, including more than

60 epidemiology studies on human populations, directed at the chlorination DBP issue.

These provide considerable scope to reflect on what we know now, how our understanding

has changed, what those changes mean for public health risk management overall and

where we should look to better understand and manage this issue in the future.

ª 2009 Elsevier Ltd. All rights reserved.

Contents

1. The title and my topic – why read any further? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20582. Evolution of disinfection by-products (DBPs) as a public health issue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2059

2.1. A journey into public health and environmental risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20592.2. Context of the times – the environment as a major cause of cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20592.3. Environmental carcinogens in drinking water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20602.4. Revelations in 1974 raise DBPs as a water quality and public health issue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20602.5. Guidelines and standards emerge while recognized DBPs proliferate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20612.6. Continuing emergence of new DBPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2062

3. Chlorination DBPs as a cancer risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2076

hology, University of Alberta, 10-102 Clinical Sciences Building, Edmonton, Canada T6G 2G3.

er Ltd. All rights reserved.

mailto:[email protected]

http://www.elsevier.com/locate/watres

www.sciencedirect.com

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22058

3.1. Cancer risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20763.2. Regulated chlorination DBPs as carcinogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20793.3. Epidemiologic evidence for chlorination DBPs and cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20793.4. Other chlorination DBPs and cancer risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2082

4. Chlorination DBPs as a reproductive risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20824.1. Regulated chlorination DBPs as reproductive toxic agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20824.2. Epidemiologic evidence for adverse reproductive outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2083

5. Risk management and public health . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20835.1. Strength of evidence and risk tradeoffs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20835.2. Risk management options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20855.3. The public in public health risk management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2085

6. Where we are and the way forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20866.1. Major lessons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20866.2. Unresolved issues and future needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20876.3. Closing thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2087Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2088Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2088References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2088

1. The title and my topic – why read any issue provides an excellent illustration of managing uncertain

further?

Having agreed with much enthusiasm to realize this oppor-

tunity to share my career perspectives on a truly fascinating

topic, I must confess to considerable discomfort with using

the personal review series title format – ‘‘.. and me’’. Disin-

fection by-products (DBPs) and public health risk provide

a topic to which many excellent scientists and engineers have

dedicated substantial portions of their careers, generating

masses of knowledge about a subject that was unknown only

35 years ago. Thus, I worry about anyone hinting by means of

a title including ‘‘me’’ at making any claims of even partial

ownership of the topic. I certainly make no such ownership

claims!

With the following account, I seek to provide my career

perspective on a remarkably complex and challenging topic

which has, over the past three decades, dramatically changed

how we view drinking water quality and safety. DBPs and

public health also provide an excellent case study of the

broader issue of risk tradeoffs in environmental health.

I believe that Water Research readers can gain useful insights

about why things have happened as they have.

I must be clear that this review is not intended to be an

account of the specific knowledge that we have amassed

about what DBPs are known, how they are formed and how

they can be managed. The reader seeking primarily such

technical background and detail is referred to the classic

treatise edited by Singer (1999), an engineering overview by

Xie (2003), a comprehensive review of the chemistry, toxi-

cology and epidemiology by the International Programme on

Chemical Safety (ICPS, 2000) and more recent updates on

current knowledge about new DBPs by Richardson et al. (2007)

and Krasner et al. (2006).

My account addresses how chlorination DBPs have

emerged as a public health issue, how the knowledge about

health risks has been interpreted and where our current state

of knowledge and residual uncertainty leaves us in deciding

upon appropriate risk management. The chlorination DBP

public health risks attributed to environmental exposures

with additional complexity and character arising from the

distinct health risk tradeoff involved.

Because we are discussing a subject that involves consid-

erable scientific evidence, we should acknowledge while

aspiring to the ideal of scientific research being the purest form

of inquiry for seeking the truth, scientific research is inevitably

conducted by imperfect humans who must rely on funding and

support from social and political institutions that need not

subscribe to all those ideals. This reality brings to mind a few

salient observations from one of the most thoughtful scientists

and science writers of our age, Sagan (1996):

‘‘Science is far from a perfect instrument of knowledge. It’s

just the best we have. In this respect, as in many others, it’s

like democracy.’’

‘‘Science by itself cannot advocate courses of human

action, but it can certainly illuminate the possible conse-

quences of alternative courses of action.’’

The need to distinguish clearly science from advocacy is

a recurring theme in my review. Advocates, who may also be

scientists, will hopefully call upon evidence generated from

careful scientific inquiry to support their positions. If our

knowledge, generated by the best available science, remains

highly uncertain, risk management decisions cannot be

determined strictly by an objective analysis of the evidence.

We also need to understand some key features of scientific

inquiry that are essential for it to be capable of revealing

truths about nature (Sagan, 1996):

‘‘Of course we must be willing to change our minds when

warranted by new evidence. But the evidence must be

strong. Not all claims to knowledge have equal merit.’’

‘‘.at the heart of science is an essential balance between

two seemingly contradictory attitudes

– an openness to new ideas, no matter how bizarre or

counterintuitive,

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2059

– and the most ruthlessly sceptical scrutiny of all ideas, old

and new.

This is how deep truths are winnowed from deep

nonsense.’’

With that sage advice from Carl Sagan as a beacon for our

journey, there are some mundane issues of terminology that

we need to address to avoid potential confusion and misun-

derstanding. First, this account will focus on chlorination

DBPs which includes any unintended chemical product

formed as a result of chlorination or chloramination of water.

This terminology distinguishes my scope from being only

chlorinated DBPs, which would restrict us to DBPs that

contain chlorine; a number of important DBPs arising from

chlorination do not contain chlorine. Likewise, we will not

focus on DBPs from other disinfection processes like those

involving ozone, chlorine dioxide or ultraviolet (UV) radiation.

Finally, I will use the acronym THM4 to represent the sum

of the four chlorinated/brominated THMs in any given

sample. Much of the literature uses TTHM (total trihalo-

methanes) for this purpose, and others have just used THM.

Although not commonly used in the literature, THM4 seems

a less ambiguous term than TTHM especially since iodinated

THMs have been identified in treated drinking water.

2. Evolution of disinfection by-products(DBPs) as a public health issue

2.1. A journey into public health and environmental risk

My career journey started in the fall of 1969 going into my last

undergraduate year, after Neil Armstrong set the first human

foot on the moon July 20, 1969. This was fortuitous timing for

my engineering class because we were able to use a drawing of

this remarkable feat of science and engineering as the centre-

piece of our graduating class picture in the spring of 1970. The

prevailing mood among engineers of that day was that we

could literally do anything with technology; if we could put

a man on the moon, there were no boundaries to technology.

I had been planning on doing graduate work in the

emerging field of biomedical engineering, so I had been taking

extra undergraduate options in the life sciences. The late

1960s had been a turbulent time marked by growing protests

focused on U.S. involvement in the Viet Nam war that spilled

over to include youthful opposition to authority of any kind,

growing use of mind-altering drugs among mainstream youth

and an emerging concern over the environment and human

impact on our planet. I was fortunate in having my commit-

ment captured by the last of these.

I read with great concern the dire global predictions of Dr.

Paul Erhlich, a population biologist at Stanford University, in

his book The Population Bomb (Ehrlich, 1968). While the form

and detail of problems we currently face differs from many of

his predictions, we have certainly experienced some of the

major disasters he raised including: a major viral epidemic

(i.e. AIDs) causing massive mortality worldwide, extensive

loss of arable land causing major famine (i.e. sub-Saharan

Africa) and over-fishing of the oceans leading to major fish-

eries collapse (i.e. the Atlantic cod fishery). Ehrlich’s warnings

largely convinced me to pursue graduate work in environ-

mental science and public health by entering a graduate

program in Public Health Engineering at the Imperial College

of Science and Technology in London.

Against this backdrop of future environmental disaster,

much of my graduate level training in public health engi-

neering seemed painfully mundane. Eventually I did come to

accept that such drab topics like sewage treatment allowed us

to control pollution of our water resources. Fascinating

lectures from the London School of Hygiene and Tropical

Medicine greatly piqued my interest in public health. Partic-

ularly memorable was learning that some major water

resource projects of the 1960s in Africa and Asia, undertaken

with the best of intentions to support development and

improve quality of life, had also served to cause the endemic

spread of terrible chronic waterborne diseases like schisto-

somiasis (infecting hundreds of millions of victims, causing

a severely debilitating parasitic disease which had no effective

medical treatment at that time). This was my first exposure to

an environmental risk versus risk scenario and the inherent

danger of ignoring unintended consequences arising from

well-meaning interventions.

I spent a few years working for provincial and federal

environmental regulatory agencies, in industrial water

pollution control and waste management before I started my

academic career at the University of Alberta in 1975. I focused

on hazardous waste management which brought me squarely

back to issues of public health risk from environmental

contaminant exposures. Initially, I had not taken much

interest in drinking water issues. This was likely the result of

a combination of perceiving a lower public health profile for

drinking water quality during the 1970s and recalling that our

instructor teaching water treatment technology at Imperial

College had told us that we already knew all that we needed to

know to make drinking water safe. The combination of

coagulation, filtration and disinfection, mainly by chlorina-

tion, was generally considered full and sufficient treatment

for assuring safe drinking water.

My years in environmental engineering revealed that the

issues I found most interesting were those that needed

a clearer definition and understanding of the problem rather

than a focus on developing a technological solution. Conse-

quently, I moved to the Faculty of Medicine in 1988 where I

was able to focus more on understanding the public health

risk dimensions that were driving environmental issues like

chlorination DBPs.

2.2. Context of the times – the environment as a majorcause of cancer

To understand better the context of discovering chlorination

DBPs in drinking water, we need to understand the prevailing

beliefs about the environment and human health at that time.

The 1960s had laid the foundation for the major expansion of

environmental regulation in the 1970s with the creation of

government departments of environment around the world

and the Environmental Protection Agency in the United

States. Major environmental research journals like Water

Research and Environmental Science & Technology (ES&T) had

only begun publication in 1967.

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22060

Silent Spring by Carson (1962) stimulated public concern

with the environment. This book became a best seller with

a potent message about the unintended consequences of

environmental pollution and indiscriminate chemical pesti-

cide use. The title referred to the killing of song birds by

pesticides aimed at controlling insect pests. This extremely

popular and influential book included an entire chapter on the

connection between environmental contaminants and

human cancer.

About the same time, a World Health Organization report

(WHO, 1964) stated that three quarters of all cancers were

caused by extrinsic factors (meaning other than genetic

predisposition). A later report by Dr. John Higginson, the

founding Director of the International Agency for Research on

Cancer (Higginson, 1969) raised the proportion attributed to

extrinsic factors to 90%. These authoritative references to

most cancers being caused by non-genetic factors were

translated by various environmentalists to mean that most

cancers were caused by environmental contamination. For

example, in his book The Politics of Cancer Epstein (1978),

a physician concerned with occupational carcinogens,

described environmental cancer as the ‘‘plague of the twentieth

century’’.

A feature article titled ‘‘The specter [sic] of cancer’’ (Ember,

1975) in ES&T opened with the statement: ‘‘This year 600,000

people in the U.S. contracted cancer, and more than 300,000 have

died from it. From 60to 90% of all human cancers are caused by

environmental factors, including cigarette smoking.’’ Environ-

mental science readers would likely have read environmental

factors to mean environmental contamination.

The ongoing confusion about this premise prompted Dr.

Higginson to speak out in an interview published in Science

(Maugh, 1979) in which he explained his perspective on cancer

and how it had been misinterpreted by those who claimed

that environmental carcinogens are primarily responsible for

cancer:

‘‘Environment is what surrounds people and impinges on

them. The air you breathe, the culture you live in, the

agricultural habits of your community, the social cultural

habits, the social pressures, the physical chemicals with

which you come in contact, the diet and so on. A lot of

confusion has arisen in later days because most people

have not gone back to the early literature, but have used

the word environment purely to mean chemicals.’’

When asked if his conclusions had been misinterpreted,

Higginson said:

‘‘They have been misinterpreted, funnily enough, not among the

majority of scientists with whom I have contact, but by the chemical

carcinogen people and especially by the occupational people.’’

Although Higginson did not mention Epstein by name, his

book, The Politics of Cancer promoting the premise that envi-

ronmental contamination was responsible for an epidemic of

cancer had recently been published.

Higginson went on to illustrate his point by noting: ‘‘ .we

are probably being exposed to so many carcinogens all the time that

what happens is mostly incomplete at the target-cell level; the cells

die and nothing ever happens. Because you and I walk across the

street, we are exposed to sunlight, a well-recognized carcinogen, but

we only develop one skin cancer in a lifetime, or even none.., despite

the thousands or millions or billions of cells that have been exposed to

sunlight. Only a very rare cell goes on to a cancer.’’

In addition to the popular views about environmental

carcinogens, subscribed to by many, if not most, environ-

mental scientists of the day, including me, a revolution in our

ability to detect trace chemicals in the environment was

underway. During the late 1960s and early 1970s, trace organic

analysis with gas chromatography (GC), linked to electron

capture and mass spectrometry detection (ECD and MSD),

dramatically improved analytical sensitivity, allowing for the

detection of numerous trace organic compounds in treated

drinking water supplies.

2.3. Environmental carcinogens in drinking water

A study done for the Environmental Defense Fund (Harris and

Page, 1974) suggested higher cancer mortality for those

consuming treated drinking water from the Mississippi River

than for those consuming drinking water from groundwater

sources. That finding was quickly reinforced by a U.S. EPA

report that the New Orleans water supply drawn from the

Mississippi River contained a number of trace organics, many

of which were suspected carcinogens (USEPA, 1974). These

events triggered a feature article in Science (Marx, 1974) titled:

‘‘Drinking Water: Another Source of Carcinogens?’’ This article

began: ‘‘As if life was not already hazardous enough, there is now

one more environmental alert with which we have to contend:

Drinking water may cause cancer.’’

Coincidently, Marx reported that these events concerning

New Orleans preceded by only five days a House of Repre-

sentatives vote on the new Safe Drinking Water Act, providing

a vote margin of 296–85, sufficiently large to override

a threatened Presidential veto. On December 16, 1974, the

President signed the Safe Drinking Water Act (SDWA) into law,

including a specific requirement for the U.S. EPA to conduct

a national survey of municipal water supplies for the presence

of halogenated organics. This was the background for the

research reporting the discovery of DBPs in 1974.

2.4. Revelations in 1974 raise DBPs as a water qualityand public health issue

Two publications (Rook, 1974; Bellar et al., 1974) changed

forever the earlier perspective that drinking water safety was

only about waterborne disease. The background story of these

discoveries has been recounted in some detail by Symons

(2001a,b) who was Chief of the Drinking Water Research

Division of the U.S. EPA.

In Europe, Johannes Rook, a Dutch water chemist, had

reported that chloroform and the other trihalomethanes

(THMs) were found at higher concentrations in chlorinated

drinking water than in raw surface water supplies. He

provided meticulous evidence for his hypothesis that the

THMs were produced by reactions between chlorine and

natural organic matter (NOM) in water. Rook was researching

the ability of the Rotterdam water treatment works to remove

trace organic pollutants from the Rhine River, its raw water

source. He had consistently identified chloroform in treated,

but not raw water samples. Rook chose not to publish the

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2061

identity of the large chloroform peak until he had figured out

what was causing its formation. He was not troubled about

consumer health risk, noting (Symons, 2001a): ‘‘Our health

officer told us chloroform was a normal constituent of cough syrups

and was not known to be particularly toxic.’’ Rook would later

question how the chloroform issue was handled by the U.S.

EPA, writing in a letter to ES&T (Rook, 1975): ‘‘The sudden

tumultuous publicity around possible adverse effects of chlorine

disinfection of water in my opinion is wide of the mark.’’ He noted

that ‘‘EPA research director Robeck went on the warpath against the

use of chlorine.’’

As an aside, Rook’s landmark 10 page paper should be

required reading for any young water chemist determined to

make an impact. Not only did Rook demonstrate a vitally

important new technique (head-space analysis) for trace

volatiles in water which made his other discoveries possible,

he also demonstrated that the brominated THMs were formed

by chlorine converting bromide to the oxidizing agent HOBr

and that it was the polyhydroxybenzene structures under-

stood to be the building blocks of NOM which were likely the

key precursors for THM formation. These discoveries were

validated countless times by numerous other researchers to

varying degrees. However, his discoveries remain among the

most critical insights in our understanding of THM formation

by chlorination. Rook’s original article has been cited 964

times as of December 2008.

Rook’s discoveries were independently corroborated by

U.S. EPA scientists (Bellar et al., 1974). The U.S. EPA also found

higher levels of THMs with increasing chlorine contact during

disinfection, but Bellar et al. speculated that ethanol was the

likely precursor. Symons (2001a) explained that there was

originally little health concern about chloroform at U.S. EPA

because of the widespread use of chloroform in consumer

products, such as toothpaste, and the likely limited occur-

rence of ethanol, their proposed (but incorrect) precursor for

THMs in raw water. Symons only became more concerned

about chloroform after consulting with Rook and becoming

convinced by him that THMs were being formed from the

reaction of chlorine with NOM, a constituent that is ubiqui-

tous in surface water supplies. The resulting expectation of

widespread THM formation was soon confirmed in the

national survey of halogenated organics mandated by the

SDWA (Symons, 1975).

Shortly after the growing body of evidence showing chlo-

roform appearing in chlorinated drinking water supplies, the

National Cancer Institute (NCI) published results of a rodent

cancer bioassay on chloroform NCI, 1976. This bioassay was

conducted in accordance with the practices of that day, i.e.

they were designed to determine the potential for chemical

substances to cause cancer in mammals and were designed to

maximize the ability of the experiment to reveal any carci-

nogenic effect. That approach meant finding the highest

possible chemical dose, the so-called maximum tolerated

dose (MTD) that could be delivered without killing or severely

injuring the test animals so they could survive for 2 years to be

able to develop cancer.

Dosing in this experiment was done as a daily slug (bolus)

dose of chloroform dissolved in corn oil. The initial high dose

in female rats of 250 mg/kg(bw)-d had to be reduced to 180 mg/

kg(bw)-d after 22 weeks because of the frank toxic effects that

were observed. Mice proved more tolerant to chloroform so

that their initial doses of 200 and 400 mg/kg(bw)-d were

increased after 18 weeks to 300 and 500 mg/kg(bw)-d. A

human dose equivalent to the highest experimental dose rate

would be more than 25,000 times that achieved by consuming

2 L per day of drinking water containing the Canadian limit

(100 mg/L) for chloroform daily over a lifetime.

The results of this high dosing showed strong evidence of

liver tumors in mice (98% of males and 95% of females at

lifetime average doses of 277 mg/kg-d and 477 mg/kg-d,

respectively, and 36% of males and 80% of females at lifetime

average doses of 138 and 238 mg/kg-d, respectively) in the

mouse experiments. These enormous dose levels were from

27 to 115% of published median lethal doses (LD50) for the

mouse (Hill et al., 1975), suggesting that the B6C3F1 strain of

mouse used in these cancer bioassays was remarkably

tolerant of chloroform. The rats dosed at up to 200 mg/kg-d

failed to show a significant excess of liver tumors relative to

controls. Rats did show a significant increase in kidney

tumors, but mice did not.

Within 4 months of the publication of the NCI bioassay

results, the U.S. Food and Drug Administration banned the use

of chloroform in cosmetics. This was a dramatic change for

chloroform which had been widely used as an anaesthetic

from the mid 1800s into the early 1900s. In fact, Dr. John Snow,

the public health icon who established with epidemiologic

evidence that fecal matter-contaminated drinking water was

responsible for cholera epidemics in London, earned his live-

lihood as an anaesthetist using chloroform after he had

perfected the means to deliver an effective anaesthetic, but

non-toxic dose (Vinten-Johansen et al., 2003). He campaigned

to overcome chlorophobia (fear of chloroform) because Lon-

don was experiencing negative publicity that chloroform was

being used for mugging, rape and murder. This publicity had

resulted in proposals to make illegal the possession of chlo-

roform, or the tools for administering it. Snow recognized the

need to filter water as a means to remove or to boil water to

inactive the, as yet, unidentified causal agent of cholera. Had

he known about the ability of chlorine to inactivate the

cholera bacterium in water, he would very likely have been

a strong advocate of chlorination. The continuous use of

chlorine for drinking water disinfection (in Middelkerk,

Belgium) did not begin until almost 45 years after Snow’s

death in 1858.

2.5. Guidelines and standards emerge while recognizedDBPs proliferate

Health concerns associated with chloroform and THMs

rapidly led to the adoption of drinking water guidelines and

standards; Canada was first in 1978 to adopt a guideline

maximum value for THM4 of 350 mg/L. Then in 1979, the U.S.

adopted a regulatory standard for THM4 under the Safe

Drinking Water Act of 100 mg/L as a running annual average of

four quarterly samples. This format for the U.S. standard

reflected knowledge from intervening water research on

THMs showing that they fluctuated seasonally because of

variation in the NOM precursors and in temperature affecting

formation kinetics. Likewise, the running annual average

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22062

reflected concern for managing long-term exposure for the

purposes of reducing cancer risk.

In 1984, WHO proposed a guideline for chloroform of 30 mg/

L based on an estimate that this would assure less than a 1 in

100,000 (10�5) lifetime cancer risk. The method used for

calculating this cancer risk estimate is an important story in

its own right that we will visit in the next section.

These regulatory limits became drivers for an explosion of

science and engineering research seeking to understand how

THMs were formed and what measures, short of abandoning

chlorination for disinfection, could be pursued to reduce

THMs sufficiently to satisfy the THM4 regulatory limits. At the

same time, chemists began discovering an ever-growing list of

other chlorination DBPs. Table 1 shows the main DBPs known

and commonly found in chlorinated drinking water by 1990.

Toxicologists began studying THMs and other chlorination

DBPs while epidemiologists began studying human pop-

ulations in search of evidence for a linkage between exposure

to chlorinated drinking water (and by implication chlorination

DBPs) and human cancer.

Understanding that the more we look for new DBPs

the more we will find raises another interesting lesson in

environmental science research from the pioneering work of

Rook. Prior to his development of head-space analysis (and

the related purge-trap method of Bellar et al. (1974), the

standard method for analysis of trace organic matter in

water was by carbon–chloroform extract (EPA-Staff-Report,

1975). In this method, the trace organics were concentrated

by passing a large volume of water through an activated

carbon adsorbent column, with subsequent elution of the

adsorbed trace organics in a smaller volume of chloroform

which was then analyzed by GC/ECD or GC/MS (the latter

only available in very specialized research laboratories by the

early 1970s). Not surprisingly, this method was blind to the

presence of chloroform in the water being analyzed, as well

as being blind to the many polar DBPs, such as haloacetic

acids (HAAs) which do not adsorb effectively to activated

carbon.

2.6. Continuing emergence of new DBPs

A summary of the occurrence of a number of more recently

identified chlorination DBPs in U.S. water utilities has been

provided by Krasner et al. (2006) and an exhaustive review of

their mutagenicity and carcinogenicity has been provided by

Richardson et al. (2007). These are summarized in Table 2.

Two classes of these emerging DBPs have been known

since the late 1980s, the halofuranones (initially 3-chloro-4-

(dichloromethyl)-5-hydroxy-2(5H)-furanone, MX and related

compounds) and the nitrosamines (initially nitrosodimethyl-

amine, NDMA). These two classes of DBPs are noteworthy

because of their potential relevance for carcinogenicity as

chlorination DBPs. A third class of chlorination DBPs, odorous

aldehydes, is noteworthy because its members play an

important role in consumer reactions to drinking water.

MX was first isolated as the main source of mutagenicity

(in the Ames Salmonella assay) in waste effluent from chorine-

bleached kraft pulp mills (Holmbom et al., 1984), an obviously

enriched source of halogenated compounds given the high

organic precursor content available in pulp mill effluents. The

occurrence of chlorinated organics in chlorine-bleached pulp

mill effluents became one of the major battlegrounds of the

1980s, particularly after finding that the chlorinated

compounds included 2,3,7,8 tetrachlorodibenzodioxin

(TCDD), a contaminant which became widely regarded, erro-

neously, as the most toxic agent ever identified (Leiss and

Hrudey, 1997). Concerns over the huge bioaccumulation

potential of TCDD causing contamination of fisheries forced

the pulp and paper industry to change its practices drastically

to eliminate chlorine bleaching in order to minimize TCDD

production, a conversion that was widely adopted by the

1990s. Having chlorination DBPs associated with TCDD and

numerous other persistent organic pollutants (POPs) such as

chlorinated pesticides like DDT did little to enhance the

reputation of DBPs.

MX was soon isolated in chlorinated drinking water

extracts, accounting for 20–50% of Ames mutagenicity (Kron-

berg et al., 1988). MX was also shown to be formed in chlori-

nated water high in NOM (accounting for 50–100% of Ames

mutagenicity).

NDMA was first recognized as a drinking water contami-

nant in Ontario in 1989 from results of a province-wide survey

which included non-target compounds. NDMA was identified

in treated water at Ohsweken with concentrations up to

0.115 mg/L (Taguchi et al., 1994; Jobb et al., 1994). NDMA was

also under investigation as a groundwater contaminant

because a chemical manufacturer in the area released NDMA.

Subsequent laboratory investigations confirmed that NDMA

was produced by drinking water treatment processes so it

came to be recognized as a DBP. Ontario established a stan-

dard for NDMA of 0.009 mg/L (ACES, 1992) and from 1994

through 1997, 20 water samples from 13 drinking water

sources exceeded this standard.

In March 1998, a drinking water well in Sacramento County,

California, sampled in response to concerns about ground-

water contamination from an aerospace industrial plant,

showed NDMA at w0.15 mg/L (CDHS, 2006). California Depart-

ment of Health Services set a notification level for NDMA of

0.002 mg/L in April 1998, which was below the analytical

detection level available at that time. NDMA was subsequently

found in other groundwater supplies. In 2000, two wells in

Orange County, where highly treated wastewater is re-injected

in a groundwater recharge project for ultimate potable re-use,

revealed NDMA levels of 0.03 and 0.04 mg/L which led to those

contaminated wells being taken out of service. A comprehen-

sive review of the emergence of NDMA as a water contaminant

(Mitch et al., 2003) has noted that NDMA is found at much

higher concentrations in chlorine-disinfected wastewaters

than the levels found in drinking water.

The NDMA story attracted my interest for three major

reasons. First, there are no halogens in NDMA illustrating the

need to consider ‘‘chlorination DBPs’’, rather than just

‘‘chlorinated DBPs’’. Second, nitrosamines are well estab-

lished as carcinogens and NDMA has a cancer slope factor

(CSF), a measure of how much cancer risk is predicted for

a given level of exposure, which is the highest among known

chlorination DBPs. Finally, NDMA is only one among a family

of nitrosamine compounds that might be found in drinking

water, but all the attention during the 1990s was focused on

NDMA alone.

Table 1 – Classes of established chlorination DBPs (adapted from Krasner et al., 1989; Froese et al., 1999).

General class Name Structure

Trihalomethanes THMs (collectively: THM4) Chloroform

Cl

Cl

Cl

CH

Bromodichloromethane

Cl

Cl

Br

CH

Dibromochloromethane

Cl

Br Br

CH

Bromoform Br

Br

Br

CH

Haloacetic acids HAAs (collectively: HAA9) Monochloroacetic acid Cl

CH2C

O

OH

Dichloroacetic acid

Cl

Cl

CH CO

OH

Trichloroacetic acid

Cl

Cl

Cl

C CO

OH

Bromochloroacetic acid

Br

Cl

CH CO

OH

Bromodichloroacetic acid

Cl

Br

Cl

C CO

OH

Dibromochloroacetic acid

Br

Br

Cl

C CO

OH

(continued on next page)

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2063

Table 1 (continued)


Monobromoacetic acid

Br

CH2 CO

OH

Dibromoacetic acid Br

BrCH C

O

OH

Tribromoacetic acid

Br

Br

BrC C

O

OH

Haloacetonitriles HANs Trichloroacetonitrile

Cl

Cl

Cl

CCN

Dichloroacetonitrile

Cl

Cl

CHCN

BromochloroacetonitrileCHCNBr

Cl

Dibromoacetonitrile Br

BrCHCN

Haloketones HKs 1,1-Dichloroacetone

Cl

3

Cl

CH

O

CCH

1,1,1-Trichloroacetone Cl

CH

O

CCClCl

3

Miscellaneous chlorinated organics Chloral hydrate

OHCl

CHCOH

ClCl

Chloropicrin

Cl

Cl

C

O-

O

Cl

N+

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22064

Table 1 (continued)


Cyanogen halides Cyanogen chloride ClCN

Cyanogen bromide BrCN

Oxyhalides Chlorite OOCl -

Chlorate

ClO

O

O

-

Bromate

O-

BrO

O

Aldehydes (including odorous aldehydes*) Formaldehyde O

CH2

Acetaldehyde O

CH3CH

Glyoxal OO

CHCH

Methyl glyoxal OO

CH3CCH

Isobutyraldehyde*

CH3

CH3CHCH

O

Isovaleraldehyde* O

CHCH2

CH3

CHCH3

2-MethyIbutyraldehyde*

CH

CH3 O

CHCH3 CH2

Phenylacetaldehyde* O

CHCH2


w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2065

Table 1 (continued)


Aldoketoacids Pyruvic acid

CCH3

C

O

OH

O

Ketomalonic acid

C CO

OH

C

OH

OO

Carboxylic acids Formate

CH

O

OH

Acetate

CH C

O

3

OH

Oxalate

C

O

C

OH

OHO

Maleic acids 2-tert-Butylmaleic acid

CH

C

O

OHCO

OH

CH2

CH3CH3CH3C

Chlorophenols CPs (odorous) Chlorophenol

Cl

OH

Dichlorophenols

Cl Cl

OH

Trichlorophenols Cl OH

Cl Cl

Chloroanisoles (odorous) Trichloroanisoles* CH3Cl

Cl Cl

O

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22066

Table 2 – Recently discovered and emerging DBPs after Krasner et al. 2006, Richardson et al. (2007).


Haloacids 3,3-dichloropropenoic acid

Cl

Cl

CHC

C OOH

3-bromo-3-chloro-4-oxypentanoic acid Cl

Br

O

CH2CCCH3

C O

OH

2,3-dibromopropanoic acid

Br Br

CHCH2 C O

OH

3,3-dibromo-4-oxopentanoic acid

Br

BrO

CH2CCCH3C O

OH

3,3-dibromopropenoic acid

Br

BrC CH C

O

OH

cis-2,3-dibromopropenoic acid BrCH C C

O

OH

Br

trans-2,3-dibromobutenedioic acidBr

Br

C C C

O

OH

C

O

OH

tribromopropenoic acid

Br

Br Br

C C C

O

OH

cis-2-bromo-3-methylbutenedioic acid Br CH3

C C C

O

OHC

O

OH

2-bromobutanoic acid Br

CHCH2CH3

C O

OH


w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2067

Table 2 (continued)


(E )-3-bromo-3-iodopropenoic acid

CHI

BrC

C OOH

trans-4-bromo-2-butenoic acid

Br

CHCHCH2 CO

OH

bromoiodoacetic acid

Br OH

ICH C

O

cis-4-bromo-2-butenoic acid BrCO

OHCHCHCH2

(Z )-3-bromo-3-iodopropenoic acid

Br OH

IC C

O

H C

trans-2,3-dibromo-2-butenoic acid

Br OH

CH3 C C C

OBr

(E )-2-iodo-3-methylbutanedioic acid

IOH

CH

CH3

CH C

O

OH

C

O

iodoacetic acid

IOH

CH2 C

O

Haloacetates bromochloromethylacetate ClBr C C

O

CH3 OH

Halo-nitromethanes chloronitromethane ClCH2 N

O

+

O-

dichloronitromethane

Cl+

O-

ClCH N

O

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22068

Table 2 (continued)


trichloronitromethane (chloropicrin)

Cl

Cl

C N+

O

O

Cl -

bromonitromethane BrCH2 N

O

+

O-

dibromonitromethane

Br+

O-

BrCH N

O

tribromonitromethane (bromopicrin)

Br

Br

C N+

O

O

Br -

bromochloronitromethane

Cl

BrCH N+

O

O

-

dibromochloronitromethane

Br

Br

C N+

O

Cl O-

bromodichloronitromethane

Cl

Br

C N+

O

Cl O-

Iodoacids iodoacetic acid

I CH2 C

O

OH

bromoiodoacetic acid ICH C

O

Br OH

(Z )-3-bromo-3-iodopropenoic acid

Br OH

IC CH C

O

(E )-3-bromo-3-iodopropenoic acidBrC CH C

O

I OH


w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2069

Table 2 (continued)


(E )-2-iodo-3-methylbutendioic acid

I

C

CH3

C C

O

OH

C

O

OH

Iodo-tri halomethanes iodoform I

ICH

I

dichloroiodomethane I

ClCH

Cl

bromochloroiodomethane I

BrCH

Cl

dibromoiodomethane I

BrCH

Br

chlorodiiodomethane I

ICH

Cl

bromodiiodomethane I

ICH

Br

Other halomethanes chloromethane ClCH3

bromomethane BrCH3

bromochloromethane Cl

BrCH2

dibromomethane BrCH2Br

carbon tetrachlorideCl

ClC

ClCl

tribromochloromethane

BrBr

C

BrCl

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22070

Table 2 (continued)


Halo-acetonitriles chloroacetonitrile ClCH2CN

bromoacetonitrile BrCH2CN

bromodichloroacetonitrile

ClClC

Br

CN

dibromochloracetonitrile

ClBrC

BrCN

tribromoacetonitrile

BrBrC

BrCN

3-bromopropanenitrile BrCH2CH2CN

Halo-ketones chloropropanoneClO

CH2CH3 C

1,3-dichloropropanone ClCH2CH2 CClO

1,1-dibromopropanone

Br CH3CH C

Br O

1,1,3-trichloropropanone ClCl O

Cl CHCH C 2

1-bromo-1,1-dichloropropanone

BrCl CH3C CCl O

1-bromo-1,3,3-trichloropropanone

ClCl

ClBr

CHCH C

O

1,1,3,3-tetrachloropropanone

ClCl

ClCl

CHCH C

O


w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2071

Table 2 (continued)


1,1,1,3-tetrachloropropanone

ClCH2C CCl

ClCl O

1,1,3,3-tetrabromopropanone

BrBr

BrBr

CHCH C

O

1,1,1,3,3-pentachloropropanone

ClCl

ClCl

CHC

O

CCl

hexachloropropanone

ClCl

Cl

ClClCl

CC

O

C

Halo-aldehydes chloroacetaldehyde

Cl CH2 C

O

H

dichloroacetaldehyde ClCH C

O

Cl H

bromochloroacetaldehyde ClCH C

O

Br H

tribromoacetaldehyde

BrBrC C

O

Br H

Haloamides monochloroacetamide

Cl CH C

O

2

NH2

monobromoacetamide

Br CH C

O

2

NH2

dichloroacetamide ClCH C

O

Cl NH2

dibromoacetamide BrCH C

O

Br NH2

trichloroacetamide

ClClC C

O

Cl NH2

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22072

Table 2 (continued)


Carbonyls 2-hexenal OCH3CH2

CH2CH

CHCH

5-keto-1-hexanal

OC

OCH3 CH2CH2CH2CH

cyanoformaldehyde OCHCN

methylethylketone OCH3 C CH2 CH3

6-hydroxy-2-hexanone OHCH2CH2

OCH2CH3 C CH2

dimethylglyoxal

C

O

C

CH3

CH3O

VOCs & m DBPs 1,1,1,2-tetrabromo-2-chloroethane

ClBr

Br

Br

Br

C CH

1,1,2,2-tetrabromo-2-chloroethane

BrClBr

CH C

BrBr

methyl-tert-butyl ether

CH3OCH3

CH3CH3 C

benzyl chloride

ClCH2

Halopyrrole 2,3,5-tribromopyrroleNH

BrBr

Br

Nitrosamines NDMA: nitrosodimethylamineONN

CH3

CH3

n-nitrosopyrrolidine

N

N O


w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2073

Table 2 (continued)


n-nitrosomorpholine

N O

O

N

n-nitrosopiperidineN ON

n-nitrosodiphenylamine

ONN

Halogenated furanones MX: 3-chloro-4-(dichloromethyl)-5-hydoxy-2(5H) – furanone

ZMX: (Z )-2-chloro-3-(dichloro-methyl)-4-oxobutenoic acid

EMX: (E )-2-chloro-3-(dichloro-methyl)-4-oxobutenoic acid

red-MX: 3-chloro-4-(dichloro methyl)-2-(5H)-furanone

ox-MX: (E )-2-chloro-3-(dichloro methyl)

butenedioic acid

MCA: 2,3-dichloro-4-oxobutenoic acid

BMX1: 3-chloro-4-(bromochloromethyl)-5-hydoxy-2(5H) – furanone

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22074

Table 2 (continued)


BMX2: 3-chloro-4-(dibromomethyl)-5-hydoxy-2(5H) – furanone

BMX3: 3-bromo-4-(dibromomethyl)-5-hydoxy-2(5H) – furanone

BEMX1: (E ) 2-chloro-3-(bromochloromethyl)-4-oxobutenoic acid

BEMX2: (E ) 2-chloro-3-(dibromomethyl)-4-oxobutenoic acid

BEMX3: (E ) 2-bromo-3-(dibromomethyl)-4-oxobutenoic acid

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2075

In order to pursue our own research on nitrosamines we

first had to develop an analytical method that was sufficiently

sensitive and reliable for drinking water detection but did not

require using a high resolution MS (Charrois et al., 2004). With

an excellent method in hand, we were able to study 20

communities (Charrois et al., 2007) for eight nitrosamines,

finding one community using chloramination having NDMA

levels at 0.1 mg/L, two other chloramination supplies with

NDMA at 0.006 and 0.008 mg/L and a chlorinated groundwater

supply with 0.012 mg/L. In addition, N-nitrosomorpholine

(NMor) was found at one supply and N-nitrosopyrrolidine

(NPyr) was found at another. We undertook an evaluation of

the chlorination practice (Charrois and Hrudey, 2007) at the

water treatment plant showing the highest levels of NDMA

using bench scale experiments. We found that providing

a free chlorine contact time of 2 h before adding ammonia for

chloramination reduced NDMA formation by up to 93%. In this

case, longer free chlorine contact time would improve disin-

fection efficiency but would also increase THM formation.

Avoidance of THM formation in this situation came at the

expense of better overall disinfection and dramatically lower

NDMA formation, an illustration of the danger of focusing

only on satisfying a THM regulation when THMs (as we shall

soon see) are not likely the cause of potential health concerns.

This is an example of a risk tradeoff where actions to avoid an

uncertain regulated risk may increase an uncertain, unregu-

lated risk.

We undertook further refinements on the analytical

method (Zhao et al., 2006) to use liquid chromatography (LC)

with tandem mass spectrometry (MS/MS) which allowed for

the analysis of sub ng/L levels of 9 nitrosamines, leading to the

first detections of N-nitrosopiperidine (NPip) and N-nitro-

sodiphenylamine (NDPhA) in drinking water. Furthermore, we

studied nitrosamine formation in eleven different disinfection

systems treating seven different source waters, finding

NDMA, NMor, NDPhA and N-nitrosodiethylamine (NDEA)

from various combinations of disinfection processes (Zhao

et al., 2008).

Odorous aldehydes represent an interesting class of chlo-

rination DBPs, not because of their health effects, but because

of the impact that they have on consumers. In 1985, I was

approached to conduct an independent inquiry into the water

quality and safety of Edmonton’s drinking water supply on

behalf of the City, the provincial regulator and the chief public

health officer. Edmonton had in 1981/82 experienced, what I

was later to discover, the second largest outbreak of giardiasis

ever reported in the developed world (Hrudey and Hrudey,

2004), yet the public water quality concerns were linked to

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22076

carcinogens and odour. The carcinogen concern had precipi-

tated the crisis in public confidence about Edmonton’s water

because one of the city’s newspapers published a front page

story under the headline: ‘‘Carcinogens in water supply’’ based on

a commercial laboratory analysis reporting detectable (mg/L)

levels of benzene in treated water. Ironically, this news inci-

dent occurred months after the benzene-contaminated sample

was taken, but at a time when the water supply was experi-

encing an annoying but unrelated odour episode. Water treat-

ment plant officials were heard reassuring the public ‘‘We don’t

know what is causing the odour, but we are sure the water is safe’’.

The full story about this situation is entertaining but is too

detailed to address in this account. What is directly relevant is

that I discovered, in reviewing analytical reports from one of

the odorous periods when raw water color rose as high as

80 TCU and dissolved organic carbon (DOC) was as high as

24 mg/L, that four aldehydes had been tentatively identified in

the treated water: isobutyraldehyde (2-methyl propanal), iso-

valeraldehyde (3-methyl butanal), 2-methyl butyraldehyde (2-

methyl butanal) and phenylacetaldehyde (Hrudey et al., 1988).

Although the raw water had a musty odour likely caused by

geosmin that was identified at about its odour threshold

(0.02 mg/L), the treated water had a more intense, somewhat

pungent swampy odour. Literature searching revealed that

the aldehydes identified had relatively low odour thresholds

of 0.9–2.3 mg/L for 2-methyl propanal, 0.15–2.0 mg/L for

3-methyl butanal, 12.5 mg/L for 2-methyl butanal and 4 mg/L for

phenylacetaldehyde. Furthermore, these aldehydes were

recognized in the food science literature as aroma compounds

in baked foods. Amoore et al. (1976) speculated that humans

may be sensitive to them as indicators of the presence of

essential amino acids (valine, leucine, isoleucine and

phenylalanine) which served as their precursors via a thermal

oxidation reaction.

I wondered that if thermal oxidation of common amino acids

could produce these aldehydes, then perhaps chlorination

oxidation could do likewise in water. I soon learned that before

thinking you may have discovered a new organic chemical

reaction, you must check the German chemistry literature. In

the early 1900s, Dakin (1916) had reported the observations of

Langheld (1909) that sodium hypochlorite would react with

a-amino acids to produce the corresponding aldehydes.

LeCloirec and Martin (1985) reported phenylacetaldehyde

being produced by chlorination of phenylalanine and acetal-

dehyde from alanine, so the theory that these odorous alde-

hydes were chlorination DBPs of common amino acids was

a reasonable hypothesis which we verified qualitatively

(Hrudey et al., 1988). Later we evaluated what factors governed

the yield of these reactions (Froese et al., 1999). An interesting

feature of the aldehyde odour was that it was remarkably

reminiscent of the chlorine swimming pool odour, even in the

complete absence of any chlorine residual or other reaction

products (Hrudey et al., 1989). Given the levels of amino acid

precursors and chlorine available in swimming pools, this all

made sense.

Although these odorous aldehydes pose no particular

health threat, as chlorination DBPs they are interesting

because they also contain no chlorine and they can play an

important role in the aesthetic quality of water, a primary

concern for consumers. Bruchet et al. (1992) also reported that

such formation of aldehydes as chlorination DBPs was able to

explain previously unidentified transient odours in a water

supply. Subsequent studies by Freuze et al. (2004, 2005)

confirmed the production of odorous aldehydes from the

chlorination of amino acids as well as finding the production

of even more odorous N-chloroaldimines which also produced

a characteristic swimming pool odour.

3. Chlorination DBPs as a cancer risk

3.1. Cancer risk assessment

Water Research readers who have focused on how chlorination

DBPs are formed and how they can be removed rather than

how the regulatory agenda for them has been developed may

find a primer on risk assessment concepts, definitions and the

weighing of evidence to be helpful. Accordingly, such a primer

has been provided in the Supplementary Data available for

this manuscript.

The prevailing belief that environmental contamination

was a major cause of human cancer was a common U.S. EPA

perspective following its creation in 1972. The U.S. EPA initi-

ated a regulatory agenda to prohibit or restrict the use of

carcinogenic pesticides, which were an obvious, preventable

exposure. The motivation for a focus on pesticides can likely

be traced back to the genesis of public concern about envi-

ronmental contamination that was initiated by Silent Spring

(Carson, 1962). In facing an early court challenge, U.S. EPA

lawyers sought to achieve court adoption of a set of ‘‘princi-

ples’’ on cancer as ‘‘officially noted facts’’ (Albert, 1994). These

‘‘principles’’ mixed prevailing knowledge about cancer with

supposition and regulatory policy, thereby attracting

a scathing editorial in the high profile medical journal Lancet

(Anon., 1976) which described the ‘‘principles’’ as varying ‘‘from

the innocuous to the absurd’’ and that ‘‘as a medico-scientific

statement the E.P.A. principles are about as useful as a law to

prohibit cancer or to make p¼ 3.’’ The editorial challenged the

U.S. EPA ‘‘principle’’ that: ‘‘cancer incidence is increasing, and that

cancer is mainly caused by exposure to external factors such as

chemicals’’ arguing instead that ‘‘The age-specific incidence of

some cancers is rising while for others, such as stomach cancer, it is

falling. There is no evidence that the variation in cancer incidence

between countries is due to variation in exposure to industrial

chemicals. It is far more likely to be due to variations in lifestyle – for

instance, variations in dietary fat intake.’’

The foregoing contrast in views about cancer reveals an

important insight that many without epidemiology training

may not appreciate. Of course, the total number of cases of

cancer in our populations has been increasing. Our population

continues to grow so that comparisons across different years

must be expressed as a population rate, rather than a total

number of cases. Likewise, cancer is a disease of age and older

age groups are an increasing proportion of our total pop-

ulation, so rate comparisons from one year to another must be

adjusted to a common age distribution. Once these critical

factors are considered, the Lancet position is correct because

overall, age-specific incidence rates of major cancer sites have

not been rising, except for lung cancer in women where the

cause, smoking, is well known.

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2077

The U.S. EPA received pushback from industry and agri-

culture when it began proposing pesticide bans on the basis

that pesticides were labelled as carcinogens. Consequently,

the U.S. EPA developed its first guidelines for cancer risk

assessment (USEPA, 1976). These guidelines and those which

replaced them a decade later (USEPA, 1986) adopted a ‘‘no

threshold’’ model for any substance that was deemed to be

a carcinogen, usually on the basis of an animal bioassay to test

whether the substance produced an excess of tumors in

exposed vs. control animals. The premise that any level of

exposure (i.e. even one molecule) carries a non-zero risk of

developing cancer some day cannot be proven or disproven

experimentally. Unfortunately, some environmental regula-

tors conveyed a perspective that this was a scientific fact

rather than a precautionary default assumption based on

some plausible scientific inference.

The 1986 cancer risk assessment guidelines included as

a major default assumption (USEPA, 1986): ‘‘Chemicals act like

radiation at low exposures (doses) in inducing cancer, i.e. intake of

even one molecule of a chemical has an associated probability for

cancer induction that can be calculated, so the appropriate model for

relating exposure-response relationships is the linearized multistage

model’’ (LMS)

The authors of the LMS model (Crump et al., 1976) argued

that ‘‘In our opinion, linear dose-response relationships are likely to

be approximately correct for many environmental carcinogens, and

this should be publicly agreed for such substances, as it was for

radiation 20 years ago.’’ Crump (1979) acknowledged: ‘‘Conclu-

sive experimental evidence is not available nor is it likely to be

available in the near future.’’

The default LMS model provides an exponential expansion

equation to fit the bioassay data points (normally limited to

only 2 or 3 doses in a cancer bioassay). The equation for the LMS

model at very low dose (d ), simplifies to:

ERðdÞ ¼ q�1d

where: ER(d ) is the extra lifetime (70 years) cancer risk above

background, d is the lifetime average daily dose of the

carcinogen, q�1 is the upper 95% confidence interval on the

cancer slope factor, CSF.

This means that the cancer risk at low dose is calculated by

a simple linear equation with a CSF times the dose. The LMS

linear low dose risk extrapolation approach means that the

estimated CSF is the critical number for risk assessment

derived from modeling a cancer bioassay. The regulatory

default of the LMS takes the upper bound fit of the limited

bioassay dose-response data (only 2 or possibly 3 dose levels)

and essentially connects the estimated upper bound for the

lowest responding dose level linearly with the zero dose

origin. That default assumption is derived from the possibility

that a single molecule of a genotoxic (DNA-damaging)

carcinogen could damage the DNA of a single cell. If the DNA

damage occurred in exactly the right manner to initiate

a tumor and if that damaged cell survived to replicate, the

damaged DNA (mutation) could be made irreversible by the

replication of daughter cells that could reproduce exponen-

tially to ultimately develop into a tumor.

A review of low dose models by Charles Brown of the U.S.

National Cancer Institute observed that several of the

available models fit the data within the experimental range

equally well, but that at low dose levels (like those corre-

sponding to 1 in 10�6 lifetime cancer risk) ‘‘differences of 3–4

orders of magnitude are not uncommon. The proposal of ‘new’

models, unless based upon strong mechanistic information, will not

alleviate the difficulties.’’ (Brown, 1984). He further noted: ‘‘The

contribution from statisticians and model-builders has reached an

impass [sic], and more accurate extrapolations are not possible

without additional information on the mechanisms of action of the

toxic agents.’’

An extremely insightful finding about the meaning of the

CSF estimated in this manner was revealed by Krewski et al.

(1993). This analysis considered the relationship between the

CSF calculated using the LMS or equivalent no threshold

model and the MTD, the maximum tolerated dose. For most

carcinogens tested, the MTD is a very high dose, which may

not be far removed from an acutely toxic dose. Krewski et al.

(1993) analyzed bioassay results for 191 individual carcino-

gens by plotting the upper bound estimate for the CSF versus

the MTD. These values were highly negatively correlated

(r¼�0.941) for values that spanned 9 decades in MTD and 10

decades in CSF, a result that could not conceivably be ach-

ieved from 191 truly independent experiments (Fig. 1).

This finding shows that carcinogens with a very high MTD

(low toxicity) had a very shallow CSF and carcinogens with

a very low MTD (high toxicity) had a very steep CSF. A primary

determinant of the CSF for any of these chemicals was its

MTD, given the procedure used for determining the CSF. In

July, 1995 I was honored with an invitation from Sir Frederick

Warner to participate in a one day symposium at the Royal

Society in London (Warner, 1995) addressing extrapolation of

dose–response data for risk assessment to share my

perspective (Fig. 2) on what was driving the remarkable

outcome revealed by Krewski et al. (1993).

Fig. 2 shows that the assumption which anchors the linear

low dose slope at the origin of the dose–response curve

combines with the point of departure (the MTD or a fixed

increment of it, MTD/2 or MTD/4) for the linear, low dose

modeling to determine the CSF. A chemical which has a very

low toxicity, i.e. a very large MTD, so that the low dose slope,

anchored at the origin and extrapolated down from a point of

departure (MTD/2 or MTD/4), will inevitably correspond to

a shallow CSF. On the other hand, a chemical which has very

high toxicity, i.e. a very low MTD, will have its point of

departure much closer to the origin, thereby yielding a very

steep CSF.

The result of this analysis is that the critical factor derived

from the cancer bioassay for calculating cancer risk, the CSF,

is determined by the method that is driven by a combination

of the assumption used to anchor the slope at the origin and to

use MTD and high fixed fractional doses thereof as the point of

departure for the linear model. This reality means that the

predicted cancer risk should only be interpreted as an upper

bound estimate of the worst cancer risk possibility. The CSF

prediction certainly does not provide for estimation of

expected (likely) cancer risk. The foregoing, combined with

the intentional use of an upper 95% confidence interval

prediction for the CSF, assures that any prediction of cancer

risk with this methodology is unlikely to underestimate the

cancer risk. However, the CSF prediction may substantially

Fig. 1 – Association between cancer slope factors and maximum tolerated dose (MTD) used in rodent carcinogen bioassays

(Krewski et al., 1993). Reprinted with permission from the National Academies Press, Copyright 1993, National Academy of

Sciences.

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22078

overestimate the cancer risk, even predicting a cancer risk

where none exists.

The 1986 U.S. EPA guidelines for carcinogen risk assess-

ment explicitly stated about the CSF: ‘‘It should be emphasized

that the linearized multistage procedure leads to a plausible upper

limit to the risk that is consistent with some proposed mechanisms of

carcinogenesis. Such an estimate, however does not necessarily give

a realistic prediction of the risk. The true value of the risk is

unknown, and may be as low as zero.’’ Although this clear

statement was not included in subsequent cancer risk

Fig. 2 – Explanation of remarkably strong association

between CSF and MTD (Hrudey, 1995).

assessment guidelines, its validity was acknowledged in

a more recent examination of U.S. EPA risk assessment

procedures (USEPA, 2004).

The rationale was explained by one of the key players

involved in the adoption of this approach, Rodricks (2007):

‘‘The linearized multistage model was selected because it seemed to

have some basis in the leading mechanistic hypotheses regarding the

carcinogenic process, and also because it seemed highly likely that

the model – because of its ‘linearization’ at low dose – would not

underestimate low dose risk, that it would, in fact, place an upper

bound on low-dose risk. Actual risk might be as large as the upper

bound, but could be lower and could even be zero. It is not the case

that risk assessors, at least those who truly understood the problem

of low-dose extrapolation, have ever claimed that risks predicted in

this fashion are known to be accurate, even ignoring the uncer-

tainties introduced by the fact that most risk assessments are based

on animal, not human data.’’

Another critical caution in cancer risk assessment is that not

all chemicals that produce a carcinogenic response in a cancer

bioassay do so by a genotoxic (DNA-damaging) mechanism. The

current cancer risk assessment guidelines (USEPA, 2005), first

proposed in 1996, have acknowledged non-genotoxic carcino-

gens. If a substance produces excess tumors in the exposed

animals by some mechanism other than DNA damage, it is

wrong to assume that there is no threshold, as was originally

done for chloroform, and to invoke the low dose linear extrap-

olation to estimate a CSF for risk assessment. The most common

example of a mechanism that is not genotoxic is cytotoxicity,

i.e. the killing of cells, which can lead to an organ response of cell

proliferation to replace the killed cells. This proliferation

response yields an increased chance of naturally-occurring

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2079

DNA replication errors, some of which may lead to tumors (Gold,

1993). However, the cytoxicity will normally exhibit a threshold

and will cause no cancer risk below that threshold. As will be

discussed in the next section, this error was demonstrated in the

case of chloroform administered at high bolus doses in corn oil

by gavage (Fawell, 2000).

The U.S. EPA no threshold assumption has also been

extended to assert that there was no safe level of exposure to

a carcinogen, e.g. a senior U.S. EPA official stated with respect

to low level carcinogen exposure that: ‘‘Such a dose–response

pattern implies that a safe level of exposure is non-existent’’ (Albert

et al., 1977). This view has commonly appeared in other rele-

vant settings, such as the fact sheet for benzene in the 1996

Australian Drinking Water Guidelines (NHMRC and ARM-

CANZ, 1996) which stated under the rationale for developing

the guideline: ‘‘Benzene is a genotoxic human carcinogen, and there

is no safe or acceptable concentration for it in drinking water.’’

My intuition that there was widespread misunderstanding

about the risks of low level exposure to a carcinogen moti-

vated me to seek a collaboration with Dan Krewski to write an

analysis of this issue entitled, appropriately: ‘‘Is there a safe

level of exposure to a carcinogen?’’ (Hrudey and Krewski, 1995).

For this purpose, we chose to accept at face value the calcu-

lations of risk provided by the controversial LMS model,

noting that there was at least general agreement that this

model was likely to give an upper bound or overestimate of

the cancer risk for any particular carcinogen. We dealt with

the issue of whether a safe level of exposure could exist by

calculating the upper bound cancer risk estimate for four

carcinogens based upon exposure to the smallest indivisible

daily dose of one molecule a day for a lifetime.

The steepest available CSF by almost 27,000 fold over the

next steepest CSF we considered was the CSF of 2,3,7,8 TCDD,

even though if it is not a genotoxic carcinogen which should

have a threshold but it provided the most pessimistic basis for

calculating cancer risk. For the exposure to 1 molecule a day

for a lifetime, the calculated cancer risk according to the LMS

model we found is 1.4� 10�15. This tiny number means that

exposing the entire population of the planet (6.7 billion

people) at this level would yield less than 0.00001 cases of

cancer in the entire world over 70 years, surely a ‘‘safe’’ dose.

For benzene, with a much lower CSF, the corresponding upper

bound cancer risk estimate would be five million times fewer

cases (i.e. 0.000000000002 cases).

From this perspective we argued that even the most

precautionary view of safety should acknowledge that safety

can be achieved at a risk that is numerically greater than zero.

3.2. Regulated chlorination DBPs as carcinogens

The changing fortunes of chloroform over the years illustrate

some of the problems in risk management for DBPs in the

presence of uncertainty and incomplete evidence. The chlo-

roform experience shows the challenges inherent in revising

entrenched regulatory measures while trying to keep pace

with improved scientific knowledge.

The initial extremely high dose bioassay results on chlo-

roform (NCI, 1976) provided the case that was widely cited

throughout the late 1970s and early 1980s that chloroform and

by extension THMs were carcinogenic. The NCI (1976) results

were obtained by a method (high dose of chloroform in corn

oil) that was later found to be much more toxic than the

equivalent dosing of chloroform in water (Bull et al., 1986). The

comparison of corn oil vs. water as a vehicle was undertaken

to explain the results from a study providing high concen-

trations of chloroform (up to 1,800,000 mg/L) dissolved in

drinking water (Jorgenson et al., 1985) that produced no

significant carcinogenic response.

The impact of extremely high doses to the liver of chloro-

form in corn oil was first noted as evidence of cytotoxicity on

liver cells. Larson et al. (1994, 1995) demonstrated by direct

experimentation that the corn oil gavage delivery of chloro-

form induced cytotoxicity and cell proliferation in liver for

mice and kidney and liver for rats. The mouse experiments

found this effect for the corn oil gavage, but not for direct

delivery of similar daily doses orally by drinking water. These

findings of a plausible mechanism for chloroform carcinoge-

nicity were supported by extensive evidence showing virtually

no genotoxic activity for chloroform (Golden et al., 1997). The

earlier noted distinction in mechanism of tumor formation

from cytotoxicity rather than genotoxicity justifies a threshold

approach to risk assessment rather than a no-threshold

approach for THMs (Fawell, 2000).

According to the U.S. SDWA, the Maximum Contaminant

Level Goal (MCLG) is the maximum level of a contaminant in

drinking water at which no known or anticipated adverse

health effects would occur, and which allows an adequate

margin of safety. U.S. EPA policy for carcinogens in drinking

water had required a MCLG to be zero, apparently ignoring the

possibility of a non-genotoxic carcinogen having a threshold

or the possibility that a non-zero exposure to a genotoxic

carcinogen can pose such an infinitesimally small risk that

a reasonable person would deem it to be safe despite techni-

cally being a non-zero risk (Hrudey and Krewski, 1995).

The toxicological evidence about chloroform led a U.S. EPA

expert review panel to recommend abandoning the MCLG of

zero in favour of a limit based on an estimated threshold. The

USEPA (1998a) proposed to raise the MCLG to 300 mg/L. Because

many intervenors protested this precedent-setting measure,

the USEPA (1998b) withdrew the proposal to change the MCLG

for chloroform from 0 to 300 mg/L (Pontius, 2000).

The Chlorine Chemistry Council sought a court review of

the U.S. EPA decision. The U.S. District Court ruled that the

U.S. EPA had violated the SDWA by failing to use the best

available science. The court found that the EPA action to be

‘‘arbitrary and capricious’’ and in excess of statutory authority.

The U.S. EPA withdrew the zero MCLG, subsequently replacing

it with a MCLG of 70 mg/L.

Meanwhile, WHO had changed its drinking water guideline

for chloroform from 30 mg/L in its first edition of Guidelines for

Drinking-water Quality (WHO, 1984) to 200 mg/L in the second

edition (WHO, 1993). The rationale that chloroform exhibited

a threshold for acting as a carcinogen was invoked to justify

this change.

3.3. Epidemiologic evidence for chlorinationDBPs and cancer

A very brief explanation of epidemiology study types,

measures of association and causal inference will be provided

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22080

here with more information provided for readers unfamiliar

with the discipline of epidemiology in the Supplementary

Data.

Direct evidence of human health risk gathered by epide-

miologic methods involves studying human populations as

opposed to the majority of risk assessment estimates that are

extrapolated from animal studies.

Epidemiology determines any meaningful associations

between exposure to hypothesized causal agents and adverse

health outcomes (disease). This may be simplified as a 2� 2

table where exposures and outcomes can be dichotomized as

shown in Fig. 3. Evidence of a positive association between

exposure and disease arises when individuals in the study

population are found more commonly in boxes a and

d combined than in boxes b and c combined.

The underlying premise of seeking an association between

exposure and outcome is fundamental to the epidemiologic

method. Inadequate data for individual study subjects in

either of the disease or exposure categories severely under-

mines the capability of an epidemiologic analysis to draw

meaningful conclusions.

There are several types of epidemiology study designs with

varying complexity, rigor, cost and most important for our

discussion, ability to test causality. This account will be

limited to two types of analytical, observational studies: case-

control and cohort. In the former, cases (with disease) are

recruited, along with suitably matched controls (free of

disease) and their exposure histories are reconstructed. In

cohort studies, the cohort is identified and followed to deter-

mine whether or not they get a disease, as well as whether or

not they are exposed. Case-control studies are inherently

retrospective, whereas cohort studies can be done prospec-

tively into the future.

The measure of association for a case-control study is the

odds ratio (OR) which is the ratio of the odds of cases being

exposed to the odds of controls being exposed. For a cohort

study, the measure of association is the rate ratio or relative

risk (RR), which is the ratio of the disease incidence rate in the

exposed group over the disease incidence rate in the unex-

posed group. For a relatively rare disease, the OR will tend

towards the RR and for our purposes, we will discuss the OR as

if it was a measure of RR. For either OR or RR, a value of 1.0

means there is no effect. Because there are inevitably errors

associated with these measures, ranging from random

sampling to biases or confounding errors, the estimated

measure of association must be judged for its statistical

significance, i.e. the 95% confidence interval (CI) of the OR or

RR must exclude 1.0.

The initial NCI (1976) carcinogenic finding on chloroform,

taken together with the background expectation that

substantial numbers of human cancers could be explained by

environmental contamination, resulted in more than 65

epidemiology studies (see Supplementary Data) of varying

quality from 1977 to 2008, seeking to determine if some

Disease (+) No Disease (-)

Exposed (+) a bNot Exposed (-) c d

Fig. 3 – Basic 2 3 2 table for epidemiologic analysis.

measure of chlorination DBPs was associated with an increase

in one or more types of cancer. The epidemiologic evidence

regarding cancer has been reviewed at various times (IARC,

1991, 2004; Mills et al., 1998; ICPS, 2000). Overall, the epide-

miologic evidence has generally been found to be insufficient

to declare chlorination DBPs to be carcinogenic in humans.

The evidence for colon and rectal cancer has been suggestive

of a causal association while the evidence for urinary bladder

cancer has been the most consistent, providing the greatest

likelihood of being causally associated with chlorination DBPs

(Mills et al., 1998).

To date, at least 33 epidemiologic studies have studied

urinary bladder cancer risk in relation to some measure of

chlorination DBPs (Supplementary Data). Of that total, 12

studies (10 case-control and 2 cohort) are of sufficient

quality to offer some possibility of providing meaningful

evidence towards causal inference. These are summarized

in Fig. 4.

Overall, the epidemiology related to cancer outcomes has

proven to be an interesting story over the past 30 years. Early

studies were launched with an expectation that cancer

outcomes should be found because chloroform/THMs had

been found to be carcinogenic in rodents. Only in the past 15

years has our improved understanding of the importance of

the mode of caction observed in rodent bioassay studies

tempered expectations of increased cancer outcomes from

exposure to chlorinated drinking water.

At present, a causal link between urinary bladder cancer

and chlorine-disinfected drinking water remains a working

hypothesis with various elements of support, primarily from

several epidemiology studies. Overall, the consistency of

findings on urinary bladder cancer is notable, but the speci-

ficity and plausibility regarding a causal agent are weak to

negative and the strength of association is generally low

enough to be susceptible to even minor confounding.

Regarding the merits of further regulatory action for THMs

or HAAs on the basis of the epidemiology findings Bull et al.

(2001) noted: ‘‘there is no evidence that decreasing THM and HAA

concentrations in drinking water will reduce the risk from bladder

cancer. There are no data to indicate that any of these compounds

can contribute to bladder cancer by any mechanism. More focused

attention on identifying the cause of bladder cancer would directly

resolve the question of whether drinking water disinfection inevi-

tably leads to unacceptable risk or whether those risks can be

rationally mitigated.’’

Finding a resolution of our understanding of whether

chlorination DBPs cause urinary bladder cancer is an impor-

tant public health issue. The extensive population exposure to

chlorine-disinfected drinking water means that the relatively

small increase in risk that has been estimated could translate

into a substantive public health issue. Based on some of these

studies (Cantor et al., 1987, 1998; Freedman et al., 1997;

King and Marrett, 1996; McGeehin et al., 1993), the U.S.

EPA estimated that the population attributable risk

(PAR) for urinary bladder cancer, i.e. the proportion of all

observed cases of urinary bladder cancer that could be

attributed to chlorinated DBPs, might range between 2 and

17% (Odom et al., 1999).

In Canada, for 2004 (total population 33 million), there were

6370 new cases of urinary bladder cancer (4748 male, 1622

Fig. 4 – Summary of analytical epidemiology evidence on urinary bladder cancer and exposure to chlorination DBPs

(OR [ 1.0 means no association).

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2081

female), suggesting that a PAR between 2 and 17% could

account for between 120 and 1100 new cases of urinary

bladder cancer per year from exposure to chlorination DBPs.

At current survival rates in Canada for this type of cancer, the

number of cancer fatalities could range from 30 to 240 deaths

per year. The USEPA (1998b) cautioned that its level of confi-

dence in these urinary bladder cancer data did not preclude

the real number of cases being caused by chlorination DBPs

from being zero because causation of urinary bladder cancer

by this exposure has not been proven. However, the public

health consequences potentially associated with a urinary

bladder cancer risk from chlorination DBPs cannot be dis-

missed lightly.

More focused attention on causes of urinary bladder cancer

is necessary because a large proportion of the comparisons of

high chlorination DBP exposures with lower chlorination DBP

exposures apparently has involved comparing exposure to

disinfected surface water vs. non-disinfected groundwater (or

groundwater receiving a very low dose of disinfectant). Such

comparisons in North America will inevitably carry some,

even a very large, element of urban vs. rural residence. Such

comparisons involve differences in a wide variety of socio-

economic and cultural factors between populations, some of

which may bear on health outcomes. Any attempt to mathe-

matically adjust for such non-specific factors will be imperfect

because the models of such factors will be neither precise nor

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22082

completely accurate. Particularly where the strength of asso-

ciation (magnitude of the OR) is low (being generally less than

2, with lower confidence intervals often approaching 1.0),

anything less than perfect adjustment for confounding by

non-drinking water quality factors could certainly allow weak

spurious associations to be found.

3.4. Other chlorination DBPs and cancer risk

MX was shown to be a sufficiently potent carcinogen in rats

(Komulainen et al., 1997) that it caused tumors at lower

exposure levels that did not cause obvious cytoxicity. MX was

already well established as being genotoxic. One account

(Melnick et al., 1997) estimated the CSF for MX to be 170 times

greater than chloroform (which we now know should not have

a CSF at all).

Although the estimated carcinogenic potency of MX is

much higher than what had been used (incorrectly) for chlo-

roform, the concentrations of MX in drinking water are

expected to be much lower than chloroform. The THMs occur

in tens of mg/L, while MX has generally been reported a 1000

fold lower at tens of ng/L (McDonald and Komulainen, 2005),

although some exceptionally high values for MX, into the

hundreds of ng/L, were reported in 17% of drinking water

plants surveyed in the U.S. (Krasner et al., 2006). Even at these

higher levels, using upper bound cancer risk estimates, MX

concentrations alone are insufficient to explain possible

cancer risks suggested by epidemiology studies. There are,

however, numerous structural analogues of MX (the haloge-

nated furanones) which have not yet been studied for carci-

nogenicity. There remains a possibility that other compounds

in this class could collectively contribute to a more substantial

cancer risk that might be more consistent with the epidemi-

ological cancer risk estimates.

The NDMA standard-setting exercise in Ontario was an

interesting illustration of uncertainty in cancer risk assess-

ment. The Advisory Committee on Environmental Standards

(ACES, 1992) held hearings to determine the evidence avail-

able for developing a standard for NDMA. ACES received four

submissions for estimating cancer risk from NDMA, all relying

on the same bioassay data (Peto et al., 1991). These four

provided a range of concentrations proposed for the standard

that differed by a factor of 5100 (510 if all were standardized to

the same risk target). The application of judgement in

choosing various assumptions for calculating the standard

showed the industry submission had the most lenient

proposed standard at 0.2056 mg/L. The federal government,

which did not need to worry about being accountable for

implementing the provincial standard, proposed the tightest

standard at 0.00004 mg/L. The number recommended by ACES

was squarely in the middle (near the geometric mean) of this

range of proposals.

California has ultimately adopted a public health goal

(PHG) of 0.003 mg/L for NDMA based on an estimated lifetime

cancer risk of 1 in a million (10�6) and a CSF of 12.8 (mg/kg-

d)�1 (about 3 times higher than the CSF estimated for MX)

derived from an analysis of the same bioassay data used by

all parties contributing to the 1992 ACES hearings (Peto et al.,

1991). Nitrosamines are more plausible as bladder carcino-

gens than any other DBPs identified to date, but as with MX,

the observed concentrations for nitrosamines in treated

drinking waters are usually at least 1000 times lower than the

THMs or HAAs.

4. Chlorination DBPs as a reproductive risk

4.1. Regulated chlorination DBPs as reproductive toxicagents

Chlorination DBPs in general and THMs in particular have

been the subject of a wide range of toxicology studies for

adverse reproductive outcomes. There have been a number of

excellent reviews of possible adverse reproductive effects of

disinfection by-products which have also included an

assessment of toxicological evidence (Nieuwenhuijsen et al.,

2000; Graves et al., 2001; Tardiff et al., 2006).

Chloroform has been extensively studied. Although the

data are described in a number of ways, one common theme

of high chloroform exposure is a reduction in fetal body

weight or survival. The high doses commonly involved have

raised the prospects that maternal toxicity was a factor in

some cases. There is only limited evidence of teratogenic

effects from chloroform exposure and it would be difficult to

use any of the available evidence to justify an expectation that

chloroform at drinking water exposure levels could explain

human birth defects. While some of the evidence is suggestive

of providing toxicological support for adverse reproductive

outcomes that have been studied in epidemiologic studies

(e.g. possible support for spontaneous abortion), the level of

evidentiary support is modest at best.

A key message for chloroform is that the data provide no

indication of potent reproductive toxic effects at realistic

drinking water exposure levels. The observed experimental

animal thresholds for no adverse effect or the lowest adverse

effect level were often more than 1% of the LD50 or the median

lethal concentration (LC50) for chloroform. These studies

finding either apparent thresholds or lowest effect levels within

a factor of 100 of the lethal dose suggests that the dose–response

curve issteep (rising fromminimal effect to lethality overa short

dose range). A steep dose–response curve at high doses suggests

that a no effects threshold should be much higher than the

much lower exposure levels allowed by regulatory standards.

Studies on BDCM provide results that are generally similar

to that for chloroform with one major exception. Bielmeier

et al. (2001, 2004, 2007) and Chen et al. (2003, 2004) have

evolved a focus on a possible mechanistic explanation for fetal

loss related to hormonal function in the placenta. Chen et al.

(2003) found that the lowest level of effect measured using in

vitro culture of human placental cells was within a factor of 35

of the maximum reported human blood BDCM concentration

after a showering exposure to THMs. This line of inquiry will

bear watching, but other than this single case, other experi-

mental dose levels in animals have been extremely high to the

point of being irrelevant to drinking water exposure.

Toxicology research on various haloacetic acids (e.g.

dichloroacetic, trichloroacetic, bromoacetic and dibromo-

acetic acid) generally occur at very high doses, often with

substantial toxicity evident to the mother. These haloacetic

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2083

acids do not appear to be very likely active agents for driving

adverse reproductive outcomes in humans.

To date, toxicology research on individual chlorination

DBPs have not shown evidence of either the necessary

potency or specificity of action that would suggest any of them

as primary candidates to explain the observations of adverse

reproductive outcomes found in some epidemiology studies.

4.2. Epidemiologic evidence for adverse reproductiveoutcomes

The number of possible adverse reproductive outcomes is

large and many possible outcomes have been studied. A total

of 34 studies are summarized (Supplementary Data). There are

many published reviews of the evidence for chlorination DBPs

causing adverse reproductive outcomes. These include: Tar-

diff et al. (2006), Hwang and Jaakkola (2003, 2002), Graves et al.

(2001), Nieuwenhuijsen et al. (2000) and Reif et al. (2000, 1996).

The possibility of chlorination DBPs causing adverse

reproductive outcomes was largely one of academic and

research interest before the publication of a major prospective

cohort study in California (Waller et al., 1998). Numerous

previous studies had found suggestive, but inconsistent and

usually not significant, associations of a variety of adversebirth

outcomes with chlorination DBPs. The large size and compar-

ative strength of the Waller et al. (1998) study drew justifiable

attention to the reported significant association of sponta-

neous abortion (miscarriage; pregnancy loss at� 20 weeks of

gestation) with THM4 and even stronger association with

BDCM exposure. The studies which have addressed the

possible association of spontaneous abortion with exposure to

chlorination DBPs are summarized in the Supplementary Data.

Waller et al. (1998) used personal interview questionnaires

to characterize the level and nature of water consumption

combined with water utility distribution system monitoring to

estimate THM4 exposure levels. They found a significant OR of

1.8 (CI: 1.1–3.0) for an exposure metric of high THM4 exposure

(�75 mg/L THM4 with >5 glasses of cold tap water consump-

tion per day) vs. low THM4 exposure (either <5 glasses of cold

tap water with THM4<75 mg/L or receiving water from a utility

providing �95% groundwater). When exposure was classified

according to high BDCM exposure (�18 mg/L), a higher OR¼ 2.0

(CI: 1.2–3.5) was found.

The implications for chlorination DBP regulation arising

from the findings from Waller et al. (1998) clearly invited some

follow-up study to seek validation of those findings. Improved

exposure assessment was a major need. This was addressed

by locating three communities with differing THM exposures

and implementing greatly expanded exposure assessment in

a prospective cohort study (Savitz et al., 2005, 2006).

The follow-up study found essentially a null result (OR¼ 1.1,

CI: 0.7–1.7) for high personal THM exposure approximately

equivalent to the Waller et al. (1998) study. The Savitz study was

performed on systems using chloramination rather than chlo-

rination, so it was not an exact replication of the Waller et al.

(1998) study. An invited commentary (Howards and Hertz-Pic-

ciotto, 2006) on the Savitz et al. (2006) paper when it was pub-

lished concluded: ‘‘Although the investigation by Savitz et al. does

not preclude effects of DBPs on pregnancy, the study provides some

confidence that exposure toTHMs through most routes is not a threat to

fetal viability during the first 20 weeks of gestation. Considering the

public health value of controlling waterborne pathogens economically

through chlorination, future studies of spontaneous abortion and

THMs are probably not warranted, although studies of swimming may

beuseful.’’ The comment onswimming refers to the possibility of

higher DBP exposures during swimming.

The loss of a pregnancy to a miscarriage provides

substantial scope for emotional distress for the parents, but

the birth of a baby with a serious birth defect is likely every

prospective parent’s worst outcome. A number of studies

have addressed the issue of various birth defects. Because

specific birth defects are relatively rare, a prospective cohort

study design is not feasible. Even case – control studies are

challenging to acquire enough cases that can be traced and

interviewed for detailed exposure assessment. As a result, the

published studies (Supplementary Data) are all retrospective

cohort studies using birth certificates or birth defect registries

(except for Klotz and Pyrch, 1999).

A summary of epidemiology studies reporting all birth

defects combined is provided in Fig. 5. For the six studies

summarized, only Bove et al. (1995) and Chisholm et al. (2008),

both birth registry-based studies, showed a significant rela-

tionship with chlorination DBP exposure. In the case of Bove

et al. (1995), the OR reported only a 90% confidence interval so

the marginal significance for all birth defects combined is

likely not meaningful. Chisholm et al. (2008) is only marginally

significant (OR¼ 1.22, CI: 1.01–1.48). Hwang et al. (2008) found

a significant relationship (OR¼ 1.21, CI: 1.07–1.36) for

a comparison of the low THM4 (5–9 mg/L) exposure group

compared with the lowest THM4 (0–4 mg/L) exposure group,

but little meaning can be attached to this finding given the ORs

of 0.97 and 1.00 found for the medium THM4 (10–19 mg/L) and

high THM4 (�20 mg/L) exposure groups compared with the

lowest THM4 exposure level.

Studies addressing various specific birth defects (cardio-

vascular, cleft, central nervous system, urinary tract and

respiratory defects) are summarized in the Supplementary

Data. Overall, the results of epidemiology studies for birth

defects either in total or as major specific types are not

supportive of a causal linkage between exposure to chlorina-

tion DBPs and any birth defects.

5. Risk management and public health

5.1. Strength of evidence and risk tradeoffs

For the case of chlorination DBPs in drinking water, there is

clearly evidence that large numbers (>600) of chemicals can

be produced (Richardson et al., 2007). Many, if not most, of

these can produce harmful effects through a variety of

toxicological test procedures. The challenge for most, if not

all, of these chlorination DBPs is that they produce

measurable toxic effects in experimental animals at dose

levels much higher (typically more than 100 fold up to more

than 10,000 fold) than any plausible exposure levels in

a reasonable quality disinfected drinking water source.

Unless these substances can plausibly act through a non-

threshold mechanism such that much lower exposure levels

can be reasonably inferred to cause an unacceptably high

Fig. 5 – Summary of epidemiology evidence on all birth defects and exposure to chlorination DBPs (OR [ 1.0 means no

association).

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22084

risk (e.g. cancer), it is difficult to make a case for expecting

harm to human health on the basis of the toxicological

evidence available to date. This seems true, even allowing

for assumptions about the effects of multiple low level

contaminants combining to cause a cumulative effect

equivalent to a much higher individual contaminant

concentration. There is, so far, limited evidence of any

serious synergistic action among identified chlorination

DBPs sufficient to cause a multiplicative effect. Even additive

accumulation of effects can only be expected for DBPs acting

by a similar mode of toxic action.

In contrast, the human evidence from epidemiology

studies has been suggestive of measureable increases (OR

from 1.2 to 2 suggesting expected urinary bladder cancer case

increases of 20–100%) of some adverse effects that have been

associated with human exposure to chlorination DBPs. These

potential outcome levels in an exposed population are

certainly a concern for public health, particularly for any

outcome that occurs commonly in the population. Even for

outcomes that are rare in the population and, therefore, less of

a public health priority, a true doubling of individual risk

(RR¼ 2) would be judged by most people as unacceptable.

The key issue for judging the evidence and deciding on an

appropriate risk management response is how strong is the

epidemiological evidence for supporting a causal association

rather than merely a chance association between exposure

and effect (discussion of causal inference in epidemiology is

provided in the Supplementary Data).

An illustration of an unexplained, statistically significant

association with factors other than chlorination DBPs is

illustrated in a retrospective cohort study by Hwang et al.

(2002) who found an apparently very high OR¼ 2.60 (1.30–5.26)

for neural tube defects. This was found for mothers exposed

to water with high color and no chlorination compared with

mothers exposed to water with low color and no chlorination.

In contrast, the chlorination, high color exposure yielded an

apparently protective OR¼ 0.68 (0.24–1.95).

What distinguishes the risk management of exposure to

DBPs in drinking water from other drinking water quality

issues is that we know from overwhelming direct and relevant

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2085

experience and evidence that a failure to disinfect drinking

water will make consumers ill from microbial disease (Hrudey

and Hrudey, 2004). This is not a matter of if, but rather when,

which pathogen and how many individuals will become ill.

The reason we can generally be so certain about the inevita-

bility of risk from undisinfected drinking water is that the

source of pathogens is so ubiquitous (pathogens which can

infect humans via water ingestion are found in human fecal

waste, pets, livestock and wildlife), making the opportunities

for drinking water contamination by pathogens both preva-

lent and of relatively high probability. These realities create an

inevitable risk management tradeoff between the high confi-

dence that is justified in disinfecting drinking water to reduce

the risks of illness caused by pathogens (rarely fatal, some

with important chronic consequences, most being self-

limiting for healthy individuals) with a vastly lower confi-

dence of chlorination DBPs causing potentially more serious

health risks (e.g. urinary bladder cancer).

5.2. Risk management options

The challenge to various risk management responses for

dealing with chlorination DBPs posed by the risk tradeoff

discussed above has been to seek alternative disinfection

processes or alterations to chlorination practice to control

commonly occurring (typically regulated) DBPs, such as THMs.

There are serious difficulties with simply substituting other

disinfection processes for chlorination.

The regulated DBPs (THMs, HAAs) are now recognized by

most who have studied these issues in detail to serve

primarily as surrogates or indicators for other DBPs rather

than being a likely causal agent for the adverse health

outcomes suggested by some of the epidemiology studies.

Implicit in changing a disinfection process to reduce a regu-

lated DBP is an expectation that controlling the regulated DBPs

will reduce other DBPs as well. Unfortunately, this expectation

is contrary to some important examples. The Krasner et al.

(2006) survey of emerging DBPs found that some of them

increased in alternate disinfection processes that reduced

THMs. These included the cases of iodo-THMs and iodo-acids

showing highest levels with chloramination; halonitro-

methanes and haloaldehydes being enhanced by pre-ozona-

tion; highly mutagenic MX compounds being increased with

chlorine dioxide, and strongly carcinogenic nitrosamines

being increased with chloramination. All of these cases are

problematic because the emerging DBPs measured and found

to be increasing with the alternate disinfection process are

substantially more toxic than the THMs that are being reduced

by switching to the alternative disinfection process.

Because we have yet to identify a plausible causal agent for

adverse human health outcomes potentially identified as

being caused by chlorination DBPs in epidemiology studies,

we are left trying to judge disinfection process alternatives

only in terms of their effect in reducing the surrogate, regu-

lated DBPs, like THMs, which we now can be reasonably

confident are largely unrelated to public health risk.

Perhaps the only risk management alternative which

avoids the major uncertainty about which DBPs we should be

reducing is to take steps to reduce the precursors to DBP

formation, most commonly NOM, in the water source.

Reduction of precursors, unless achieved by a process that

adds another problematic precursor to increase formation of

non-regulated DBPs (e.g. addition of coagulant chemical that

can act as a precursor), should have the effect of reducing

other conceivable DBP formation and consequently should

not create an alternative DBP risk.

5.3. The public in public health risk management

The motivation for controlling chlorination DBPs in drinking

water is obviously to reduce the possible public health risk that

may be associated with one or more such chemicals. This is

fully consistent with the public health practice foundation of

emphasizing disease prevention. Where substantial uncer-

tainty exists, as in the case of chlorination DBPs, being suitably

precautionary is justified, given the broad public exposure

provided by community drinking water supplies. That said,

there is also a responsibility inherent among public health

professionals to exercise precaution in a responsible manner

that neither undermines the credibility of public health prac-

tice nor causes unwarranted fear among the public who cannot

be expected to understand the nature of uncertainties involved

and what levels of precaution may be warranted.

There have been a number of cases (Hrudey and Hrudey,

2004) where negative attitudes to chlorination (including

taste) or the fear of health effects from chlorination DBPs has

played a role in contributing to allowing a waterborne disease

outbreak to occur (e.g. Creston/Erickson, B.C.; Walkerton,

Ontario; Bramham, England; Asikkala, Finland; Transtrand,

Sweden). A particularly tragic case occurred in Walkerton,

Ontario in 2000 where seven consumers died after a failure to

maintain or monitor chlorination sufficient to prevent an

outbreak of Escherichia coli O157:H7 and Campylobacter (Hrudey

and Hrudey, 2004). Although there were a large number of

failures and contributing factors involved in this outbreak

(Hrudey and Walker, 2005), the operator who was responsible

for maintaining and monitoring chlorine residual was openly

opposed to chlorination without having any understanding of

the role it played in managing the risk of pathogens in this

town’s vulnerable shallow supply well.

In the case of Creston/Erickson, B.C. these two communi-

ties, using a common surface water source with no treatment,

experienced outbreaks of giardiasis in 1985 and 1990 (Hrudey

and Hrudey, 2004). After the second outbreak, in which visi-

tors and newcomers lacking immunity were found to be those

most likely to become ill, the communities were ordered by

public health officials to implement chlorination. Creston

complied, but Erickson resisted, declaring itself a ‘‘chlorine-free

zone’’ and mounting a 55 day blockade to prevent access to the

site where the chlorinator was to be installed. The website

launched by those opposed to chlorination (www.watertalk.

org/wag) provided links (no longer functional) to the Health

Canada website. Viewers were invited to go to these links to

find ‘‘Scientific Proof of the Dangers of Chlorine’’. I found this to be

an insightful commentary on the manner in which docu-

ments prepared by government agencies to interpret public

health evidence end up conveying a frightening message,

often through the use of intimidating jargon, even when the

health evidence being reviewed is highly uncertain. These

cases provide examples where the tradeoff between the

http://www.watertalk.org/wag

http://www.watertalk.org/wag

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22086

certain danger of pathogens and possible dangers of DBPs has

not been balanced effectively.

In recent years, public reaction to the possibility of adverse

health outcomes from chlorination DBPs has been particularly

striking in response to media reports about the epidemiology

studies on adverse reproductive outcomes. After the Waller

et al. (1998) study reported an elevated risk of spontaneous

abortion among women more exposed to THMs compared with

those who consumed lower levels of THMs, media stories about

THM risks to unborn babies were widely reported in the U.S. The

Public Health Department of the municipality of Chesapeake,

Virginia, which was in the midst of changing its water system to

meet the federal THM standard, issued a warning to pregnant

mothers to drink bottled water after the water treatment plant

personnel brought the Waller et al. (1998) study to its attention

(Huslin, 2002). One result of that attempt at informing the public

of the possible health risk was that the municipality became

a defendant in lawsuits from 214 plaintiffs claiming breach of

contract and warranty, battery negligence, nuisance, trespass,

violation of the state Consumer Protection Act and fraud (Anon.,

2005). Ultimately the municipality was found to be immune to

these lawsuits by the Virginia Supreme Court, citing the

municipalities ongoing efforts to comply with the THM4 stan-

dard,but there wasclearly a lot ofexpense and grief experienced

by all the parties involved in litigation that would have likely

been found unwarranted following publication of the Savitz

et al. (2005, 2006) study. It is not difficult to imagine the

emotional toll these circumstances took on affected parties.

That issue was summarized as (Huslin, 2002):

‘‘Today, many are also wondering something else: Might

they have lost their babies simply because they drank tap

water while they were pregnant?’’

A more recent example of media coverage of an epidemi-

ology study exploring adverse reproductive outcomes has

been unfolding in Perth, Western Australia where a study

(Chisholm et al., 2008) was made public and the local news-

paper ran a story under the headline:

‘‘Tapwater ‘increases risk of birth defects’ ’’(Guest, 2008).

The story began: ‘‘Chemical by-products in tap water in some

Perth suburbs are increasing the risk of birth defects and pregnant

women may need to avoid the danger, health researchers led by the

University of WA have warned.’’

It is easy to imagine that any woman who has given birth to

a child with a defect would find this news coverage distressing

because of the implied message that if the mother had

consumed Perth tap water this may have been partially

responsible for her child’s defect. Any public health profes-

sional who has dealt with the media will know that it is

difficult to convey an accurate message within typical media

constraints. However, in this particular case, the evidentiary

basis for raising any risk message whatsoever deserves closer

scrutiny because the results were arguably not practically

significant.

The study relied upon four years of data from a birth

registry whereby birth outcomes were compared with 47

water samples where THMs were analyzed. Sample locations

were linked to maternal residential postal code at time of

birth. Maternal age and socioeconomic status code were also

obtained for each birth. Based on average THM4 analyses on

samples over 6 collection dates, the study area was divided

into 3 regions (low: 54� 16.6 mg/L, medium: 109� 28.9 mg/L,

and high: 137� 24 mg/L). Birth defect data were analyzed for

the sum of all birth defects and 7 individual defect categories.

Results for this retrospective cohort study which lacked indi-

vidualized exposure assessment (no questionnaire data) were

significant for only two categories, all birth defects combined

(OR¼ 1.22, CI: 1.01–1.48) and cardiovascular system defects

(OR¼ 1.62, CI: 1.04–2.51). These two studies are considered

significant because the lower confidence interval exceeds 1.0.

Fig. 5 compared results for six studies reporting all birth

defects combined. Of these, the Chisholm et al. (2008) study is

the first to report any significant association of all birth defects

with chlorination DBPs. Hwang et al. (2002) reported a signifi-

cant association between exposure groups of no chlorination –

high color compared with no chlorination – low color

(OR¼ 1.18, CI: 1.02–1.36), but this comparison clearly has no

causal connection with chlorination DBPs, since neither

exposure group experienced chlorination. The results of the

Chisholm et al. (2008) study are so marginally significant that

taken together with the marginal or negative findings from

the other five studies and the known limitations in study

design combine to provide a very weak basis to make any

credible claims or suggestions about chlorination DBPs being

responsible for birth defects.

6. Where we are and the way forward

6.1. Major lessons

Any issue which has had the influence on practices in

drinking water that chlorination DBPs have had for over 30

years needs to be viewed in its full historical context. Rook’s

discovery in 1974 came at a time when environmental regu-

latory agencies were in their infancy and there was a common

belief that the war on cancer could be won simply by

controlling exposure to environmental contaminants. While

many still ascribe to this belief, there remains scant evidence

amidst our enormous advances in cancer research and our

documented experience with cancer incidence statistics to

support an expectation for reducing cancer by lowering

exposure to environmental chemicals. These realities, by

themselves, have no bearing on whether chlorination DBPs

cause urinary bladder or any other form of human cancer.

That question remains to be answered. However, the collec-

tive beliefs of a generation of water scientists and engineers

about the health risks of chlorination DBPs were inevitably

influenced by the context at the time that chlorination DBPs

were discovered.

Chlorination DBPs were discovered as part of an ongoing

wave of analytical chemistry advances that has provided

increasing sensitivity in detection of individual compounds

and increased capacity to identify individual compounds

where they were not found before. This trend will not end any

time soon and the result will continue to disturb our collective

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2087

peace of mind about the purity of any medium, including

drinking water.

Consider that chloroform currently has a method detection

level of 0.017 mg/L, which means that we might find a sample

(e.g. an ultra-pure bottled spring water) containing 0.016 mg/L

to be a non-detect, which equates with zero for many of us,

consciously or otherwise. With a molecular mass of 119.38 g/

mol and Avogadro’s number at 6.02� 1023 molecules per mole,

a non-detectable concentration of chloroform of 0.016 mg/L

would mean that about 8� 1013 molecules of chloroform per

litre of water could be non-detectable. This number is

certainly not zero and it shows that concepts of purity and

zero are defined by analytical capability rather than any

absolute standard. From a public health perspective there is

no escaping the reality that the only way to judge the impor-

tance of the level of exposure to a chemical substance is to

know its concentration in relation to a concentration that has

been shown by toxicology or epidemiology evidence to cause

specified harm. We will inevitably continue to discover more

DBPs with increasing sensitivity of detection, but the public

health significance of such discoveries can only be judged in

relation to valid knowledge about how much exposure to the

discovered DBP is needed to produce harm.

Our tools for judging human health risks from chemical

exposures continue to get more sophisticated, but both toxi-

cology and epidemiology face fundamental limitations in

their ability to yield evidence about low level population

exposures to environmental contaminants. The growing

subtlety and complexity of evidence in this regard runs

counter to the simplicity that is generally demanded of

programs to regulate human exposures to environmental

contaminants. This situation calls for a frank acceptance, with

no need for apology, that environmental regulations, such as

those which are invoked for chlorination DBPs, are inherently

precautionary in seeking to protect public health. I believe

that a historical view of regulatory policies on chlorination

DBPs has shown them to have been largely precautionary.

There is an adequate public policy basis to maintain

a reasonable, precautionary approach to regulating a public

exposure as pervasive as drinking water. With that premise in

hand, I do not believe there is any need, value or justification

for overselling inherently uncertain toxicological or epidemi-

ological evidence as being more certain than the evidence

warrants in order to scare consumers, drinking water

providers or politicians into supporting a particular regulatory

standard.

Unfortunately, the more precautionary we seek to be, i.e.

the lower the target risk level for population exposure, the less

that we can hope to have experimental evidence useful in

judging our success (Hrudey and Leiss, 2003). This dilemma is

not unique to the issue of managing public health risks from

chlorination DBPs, but it remains a reality that has to be

recognized.

This account should make clear, partly because it has

necessarily skimmed over many relevant issues and simpli-

fied many complex issues, that it is exceedingly difficult to

gain an overview of an issue such as chlorination DBPs and

public health risk. The number of disciplines involved and the

complexity of evidence that each relies upon tends to intim-

idate anyone who seeks to grasp the entire issue.

6.2. Unresolved issues and future needs

The notable consistency of findings suggesting an association

of urinary bladder cancer with various measures of exposure

to chlorination DBPs at a number of locations by different

investigators in several epidemiology studies should lessen

the chance that all these observed associations are spurious.

However, a substantial concern remains that these studies,

with only one small and substantially qualified exception

(Chevrier et al., 2004), have apparently all relied upon similar

means to achieve the low-end or non-exposures to chlorina-

tion DBPs (i.e. apparent inclusion of un-chlorinated supplies

that are also likely to be generally small or rural) without

necessarily seeking out opportunities to include high end

chlorination DBPs exposures (chlorinated surface waters with

high NOM). Until epidemiology studies are completed with

substantial numbers of participants residing in larger urban

areas who have had low to negligible chlorination DBP expo-

sure because the drinking water supply used alternate disin-

fection and water treatment practices (ozonation, UV or no

disinfection), the possibility will remain of a small systematic

bias sufficient to explain the consistent, but comparatively

weak association (generally OR< 2) of urinary bladder cancer

with chlorination DBPs.

Given that none of the chlorination DBPs so far identified in

drinking water are plausible bladder carcinogens, there is

a need for a coordinated research approach to actively pursue

the identification of plausible bladder carcinogens that might

occur widely enough in chlorinated drinking water to offer an

explanation for the epidemiological associations observed for

urinary bladder cancer. Bull et al. (2006) used quantitative

structure activity relationships and a broad survey of the

toxicological and chemical literature to identify plausible

DBPs that may be from 1000 to 10,000 times more toxic than

the THMs or HAAs. My colleague at the University of Alberta,

Dr. Xing Fang Li, is currently leading a Water Research Foun-

dation-funded international research team (Canada, U.S. and

Australia) to develop reliable and sufficiently sensitive

analytical methods for candidate members of the identified

DBP classes so that surveys could be conducted to determine

which, if any, of these DBP classes actually occur in chlorine or

chloramine disinfected drinking water.

Finally, by far the most noticeable impact among

consumers receiving chlorinated drinking water is the

unmistakable degradation of the aesthetic quality of their

drinking water caused by the taste and odour of chlorine or

some of its by-products. Some improvement in consumer

acceptance can be achieved by maintaining consistent oper-

ations in order to avoid fluctuating chlorine residual concen-

trations that will inevitably increase consumer sensitivity.

There has been a substantial migration of consumers to the

use of bottled water or point-of-use treatment devices among

those who can afford the substantial added cost. This obvious

problem related to chlorine disinfection does not appear to

have received much research attention.

6.3. Closing thoughts

My intuition from countless discussions with water quality

engineers and scientists and front-line water treatment

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22088

personnel is that many, if not most, have a skewed perception

of the evidence concerning public health risks from chlori-

nation DBPs that is not accurately based on the current

evidence. I often perceive a grudging acceptance of the need to

disinfect concurrent with a more fundamental belief that the

presence of hundreds of chlorination DBPs in drinking water

must be causing harm to health. Others whom I encounter in

public health circles are convinced that there must be harm

and suggest that it is simply a matter of time until we prove it.

They may well be proven right over time, but based on the

evidence from 35 years of research, I would not be willing to

bet my pension on that outcome.

No one should interpret my message as advocating aban-

doning any of our current water quality guidelines for chlori-

nation DBPs. One clear exception is Canada’s 2006 guideline of

16 mg/L for BDCM, a level that is too low considering the most

recent evidence (NTP, 2006) that shows BDCM does not pose

a cancer risk. Rook’s discovery of THMs and the subsequent

regulatory responses worldwide have resulted in vast

improvement in our understanding and delivery of drinking

water quality and treatment. My concern is with the misguided

application of our current health effects knowledge as a basis

to blindly pursue technological fixes aimed at further reduc-

tions of THMs. At best we may be wasting investment that

could be better spent; at worst we could be creating other DBPs

that do pose serious health risks.

Water quality engineers and scientists have to be careful

about becoming trapped in a mindset I found all too common

during my years engaged in environmental engineering

research – Don’t ask why we are pursuing a technological fix;

just ask for the regulatory numbers that need to be satisfied!

Clearly, in the interdisciplinary world of environmental

science and public health, we need to work primarily with the

disciplinary training and expertise that we have acquired.

However, the risk assessment and public health risk manage-

ment tradeoffs involved in an issue like chlorination DBPs is

not the exclusive domain of any specific discipline – rather it

depends on the effective input from a number of fields. The

overall process will inevitably rely on default assumptions that

must be based on policy because they are designed to bridge

gaps in our scientific knowledge (e.g. low dose cancer risk

extrapolation). In these areas, water quality engineers and

scientists should be encouraged to be healthy sceptics about

the evidentiary basis for regulatory philosophy and the

resulting numbers so that they can understand and hopefully

test, as far as possible, the rationale for a given risk manage-

ment approach. I would certainly encourage more provocative

discussion about these issues among our colleagues, particu-

larly those engaged in educational programs where our next

generation of practitioners is being trained.

Achieving greater understanding of these complex issues

across a broader cross-section of researchers and practi-

tioners among the water scientific disciplines may well reveal

risk management approaches that achieve a better balance of

precaution in the face of serious but uncertain risks (e.g.

chlorination DBPs and possible urinary bladder cancer) versus

sensible action to avoid potentially less severe but inevitable

dangers (i.e., waterborne pathogens, if we fail to effectively

manage risks by source water protection and treatment). I will

close by again quoting from the wisdom of Sagan (1996) about

the role of science:

‘‘The collective enterprise of creative thinking and scep-

tical thinking, working together, keeps the field on track.’’

Acknowledgements

Background research for this work has been supported by the

National Collaborating Centre on Environmental Health, the

Canadian Water Network and the Natural Sciences and Engi-

neering Research Council.

Considerable assistance in data base searching to establish

the reference base was provided by Jeanette Buckingham.

Technical support and manuscript editorial review were

provided by Elizabeth J. Hrudey and technical assistance was

received from Connie Le, Xing Fang Li, Chuan Wang and Jeff

Charrois.

This field has been enriched because of the research per-

formed by countless investigators with particularly note-

worthy contributions from Richard Bull, Ken Cantor, Linda

Dodds, John Fawell, Will King, Stuart Krasner, Mark Nieu-

wenhuijsen, John Reif, Susan Richardson, Johannes Rook,

David Savitz, James Symons, Phil Singer, Christina Villaneuva,

Clifford Weisel and Kirsten Waller.

Appendix.Supplementary data

Supplementary data associated with this article can be found,

in the online version, at doi:10.1016/j.watres.2009.02.011.

r e f e r e n c e s

ACES, 1992. A Standard for N-Nitrosodimethylamine (NDMA).ACES Report. Advisory Committee on EnvironmentalStandards, Ontario Ministry of Environment, Toronto.

Aschengrau, A., Zierler, S., Cohen, A., 1993. Quality of communitydrinking water and the occurrence of late adverse pregnancyoutcomes. Archives of Environmental Health 48 (2), 105–113.

Albert, R.E., 1994. Carcinogen risk assessment in the U.S.Environmental Protection Agency. Critical Reviews inToxicology 24 (1), 75–85.

Albert, R.E., Train, R.E., Anderson, E., 1977. Rationale developed bythe Environmental Protection Agency for the assessment ofcarcinogenic risks. Journal of the National Cancer Institute 58,1537–1541.

Amoore, J.E., Forrester, L.J., Pelosi, P., 1976. Specific anosmia toisobutyraldehyde: the malty primary odour. Chemical Sensesand Flavour 2, 17–25.

Anon, 1976. Seventeen principles about cancer, or something.The Lancet 307 (7959), 571–573.

Anon, 2005. Court backs Virginia city immunity to DBP claims.American Water Works Association Journal 97 (2), 18.

Bellar, T.A., Lichtenberg, J.J., Kroner, R.C., 1974. The occurrence oforganohalides in chlorinated drinking waters. Journal of theAmerican Water Works Association 66 (12), 703–706.

http://dx.doi.org/doi:10.1016/j.watres.2009.02.011

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2089

Bielmeier, S.R., Best, D.S., Guidici, D.L., Narotsky, M.G., 2001.Pregnancy loss in the rat caused by bromodichloromethane.Toxicological Sciences 59 (2), 309–315.

Bielmeier, S.R., Best, D.S., Narotsky, M.G., 2004. Serum hormonecharacterization and exogeneous hormone rescue ofbromodichloromethane-induced pregnancy loss in the F344rat. Toxicological Sciences 77 (1), 101–108.

Bielmeier, S.R., Murr, A.S., Best, D.S., Harrison, R.A., Pegram, R.A.,Goldman, J.M., Narotsky, M.G., 2007. Effects ofbromodichloromethane on ex vivo and in vitro luteal functionand bromodichloromethane tissue dosimetry in the pregnantF344 rat. Toxicology In Vitrol 21 (5), 919–928.

Bove, F.J., Fulcomer, M.C., Klotz, J.B., Esmart, J., Dufficy, E.M.,Savrin, J.E., 1995. Public drinking water contamination and birthoutcomes. American Journal of Epidemiology 141 (9), 850–862.

Bove, G.E., Rogerson, P.A., Vena, J.E., 2007. Case-control study of theeffects of trihalomethanes on urinary bladder cancer risk.Archives of Environmental and Occupational Health 62 (1), 39–47.

Brown, C., 1984. High- to low-dose extrapolation in animals. ACSSymposium Series.

Bruchet, A., Costenin, E., Legrand, M.F., Mallevialle, J., 1992.Influence of the chlorination of natural nitrogenous organiccompounds on tastes and odours in finished drinking waters.Water Science and Technology 25 (2), 323–333.

Bull, R.J., Brown, J.M., Meierhenry, E.A., Jorgenson, T.A.,Robinson, M., Stober, J.A., 1986. Enhancement of thehepatotoxicity of chloroform in B6C3F1 mice by corn oil:implications for chloroform carcinogenesis. EnvironmentalHealth Perspectives 69, 49–58.

Bull, R.J., Krasner, S.W., Daniel, P.A., Bull, R.D., 2001. HealthEffects and Occurrence of Disinfection By-products. AwwaResearch Foundation, Denver, 397 pp.

Cantor, K.P., Hoover, R., Hartge, P., Mason, T.J., Silverman, D.,Altman, R., Austin, D.F., Child, M.A., Key, C.R., Marrett, L.D.,Myers, M.H., Narayana, A.S., Levin, L.I., Sullivan, J.W.,Swanson, G.M., Thomas, D.B., West, D.W., 1987. Bladdercancer, drinking water source, and tap water consumption:a case-control study. Journal of the National Cancer Institute79 (6), 1269–1279.

Cantor, K.P., Lynch, C.F., Hildesheim, M.E., Dosemeci, M., Lubin, J.,Alavanja, M., Craun, G., 1998. Drinking water source andchlorination byproducts. I. Risk of bladder cancer.Epidemiology 9 (1), 21–28.

Carson, R., 1962. Silent Spring. Houghton Mifflin, Boston.CDHS, December 22, 2006. A Brief History of NDMA Findings

in Drinking Water Retrieved from: http://www.cdph.ca.gov/certlic/drinkingwater/Pages/NDMAhistory.aspx (accessed18.07.08).

Charrois, J.W.A., Arend, M.W., Froese, K.L., Hrudey, S.E., 2004.Detecting N-nitrosamines in drinking water at nanogram perliter levels using ammonia positive chemical ionization.Environmental Science & Technology 38 (18), 4835–4841.

Charrois, J.W.A., Boyd, J.M., Froese, K.L., Hrudey, S.E., 2007.Occurrence of N-nitrosamines in Alberta public drinking-water distribution systems. Journal of EnvironmentalEngineering and Science 6 (1), 103–114.

Charrois, J.W.A., Hrudey, S.E., 2007. Breakpoint chlorination andfree-chlorine contact time: implications for drinking water N-nitrosodimethylamine concentrations. Water Research 41 (3),674–682.

Chen, J., Douglas, G.C., Thirkill, T.L., Lohstroh, P.N., Bielmeier, S.R.,Narotsky, M.G., Best, D.S., Harrison, R.A., Natarajan, K.,Pegram, R.A., Overstreet, J.W., Lasley, B.L., 2003. Effect ofbromodichloromethane on chorionic gonadotrophin secretionby human placental trophoblast cultures. ToxicologicalSciences 76 (1), 75–82.

Chen, J., Thirkill, T.L., Lohstroh, P.N., Bielmeier, S.R., Narotsky, M.G.,Best, D.S., Harrison, R.A., Natarajan, K., Pegram, R.A.,

Overstreet, J.W., Lasley, B.L., Douglas, G.C., 2004.Bromodichloromethane inhibits human placental trophoblastdifferentiation. Toxicological Sciences 78 (1), 166–174.

Chevrier, C., Junod, B., Cordier, S., 2004. Does ozonation ofdrinking water reduce the risk of bladder cancer?Epidemiology 15 (5), 605–614.

Chisholm, C., Cook, A., Bower, C., Weinstein, P., 2008. Risk of birthdefects in Australian communities with high brominateddisinfection by-product levels. Environmental HealthPerspectives 116 (9), 1267–1273.

Crump, K.S., 1979. Dose response problems in carcinogenesis.Biometrics 35 (1), 157–167.

Crump, K.S., Hoel, D.G., Langley, C.H., Peto, R., 1976. Fundamentalcarcinogenic processes and their implications for low doserisk assessment. Cancer Research 36 (9_Part_1), 2973–2979.

Dakin, H.D., 1916. The oxidation of amino acids to cyanides.Biochemical Journal 10 (2), 319–323.

Doyle, T.J., Zheng, W., Cerhan, J.R., Hong, C.P., Sellers, T.A.,Kushi, L.H., Folsom, A.R., 1997. The association of drinkingwater source and chlorination by-products with cancerincidence among postmenopausal women in Iowa:a prospective cohort study. American Journal of Public Health87 (7), 1168–1176.

Ember, L., 1975. The specter of cancer. Environmental Science &Technology 9 (13), 1116–1121.

EPA-Staff-Report, 1975. Evaluation of the carbon chloroformextract and methylene chloride soluble portion of the carbonalcohol extract. Journal American Water Works Association67, 281–284.

Ehrlich, P.R., 1968. The Population Bomb. Ballantine Books, NewYork.

Epstein, S.S., 1978. The Politics of Cancer. Sierra Club Books, SanFrancisco.

Fawell, J., 2000. Risk assessment case study – chloroform andrelated substances. Food & Chemical Toxicology 38, S91–S95.

Freedman, D.M., Cantor, K.P., Lee, N.L., Chen, L.-S., Lei, H.-H.,Ruhl, C.E., Wang, S.S., 1997. Bladder cancer and drinkingwater: a population-based case-control study in WashingtonCounty, Maryland (United States). Cancer Causes & Control 8(5), 738–744.

Freuze, I., Brosillon, S., Herman, D., LaPlanche, A., Democrate, C.,Cavard, J., 2004. Odorous products of the chlorination ofphenylalanine in water: formation, evolution andquantification. Environmental Science & Technology 38 (15),4134–4139.

Freuze, I., Brosillon, S., LaPlanche, A., Tozza, D., Cavard, J., 2005.Effect of chlorination on the formation of odorous disinfectionby-products. Water Research 39, 2636–2642.

Froese, K.L., Wolanski, A., Hrudey, S.E., 1999. Factors governingodorous aldehyde formation as disinfection by-products indrinking water. Water Research 33 (6), 1355–1364.

Gold, L.S., 1993. The importance of data on mechanism ofcarcinogenesis in efforts to predict low-dose human risk. RiskAnalysis 13 (4), 399–401.

Golden, R.J., Holm, S.E., Robinson, D., Julkunen, P.H., Reese, E.A.,1997. Chloroform mode of action: implications for cancer riskassessment. Regulatory Toxicology & Pharmacology 26, 142–155.

Graves, C.G., Matanoski, G.M., Tardiff, R.G., 2001. Weight ofevidence for an association between adverse reproductive anddevelopmental effects and exposure to disinfection by-products: a critical review. Regulatory Toxicology &Pharmacology 34 (2), 103–124.

Guest, D., 2008. Tap Water ‘Increases Risk of Birth Defects’. TheWest Australian, Perth, WA, Australia, p. 1, 14.

Harris, R., Page, T., 1974. The Implications of Cancer-causingSubstances in the Mississippi River Waters. EnvironmentalDefence Fund, Washington, D.C.

http://www.cdph.ca.gov/certlic/drinkingwater/Pages/NDMAhistory.aspx

http://www.cdph.ca.gov/certlic/drinkingwater/Pages/NDMAhistory.aspx

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22090

Higginson, J., 1969. Eighth Canadian Cancer Conference.Pergamon Press.

Hill, R.N., Clemens, T.L., Liu, D.K., Vesell, E.S., Johnson, W.D.,1975. Genetic control of chloroform toxicity in mice. Science190 (4210), 159–161.

Holmbom, B., Voss, R.H., Mortimer, R.D., Wong, A., 1984.Fractionation, isolation and characterization of Amesmutagenic compounds in kraft chlorination effluents.Environmental Science & Technology 18 (5), 333–337.

Howards, P.P., Hertz-Picciotto, I., 2006. Invited commentary:disinfection by-products and pregnancy loss–lessons.American Journal of Epidemiology 164 (11), 1052–1055.

Hrudey, S.E., 1995. Major concerns with quantitative cancerrisk assessment for environmental risk management:North American experience. Process Safety andEnvironmental Protection, Transactions of the Institutionof Chemical Engineers Part B 73 (November, Suppl.),S24–S27.

Hrudey, S.E., Daignault, S.A., Gac, A.J., Poole, D., Walker, G.,Birkholz, D., 1989. Swampy or swimming pool odours caused bychlorination and chloramination. In: Seminar Proc. Identificationof Taste and Odor Compounds. 1989 Annual Conference, LosAngeles, CA. American Water Works Association.

Hrudey, S.E., Gac, A., Daignault, S.A., 1988. Potent odor-causingchemicals arising from drinking water disinfection. WaterScience and Technology 20 (8–9), 55–61.

Hrudey, S.E., Hrudey, E.J., 2004. Safe Drinking Water – Lessons fromRecent Outbreaks in Affluent Nations. IWA Publishing, London.

Hrudey, S.E., Krewski, D., 1995. Is there a safe level of exposure toa carcinogen? Environmental Science & Technology 29 (8),370A–375A.

Hrudey, S.E., Leiss, W., 2003. Risk management and precaution –insights on the cautious use of evidence. EnvironmentalHealth Perspectives 111 (13), 1577–1581.

Hrudey, S.E., Walker, R., 2005. Walkerton – 5 years later. Tragedycould have been prevented. Opflow 31 (6) 1,4–7.

Huslin, A., 2002. In a Glass of Water, a Mother’s Worst Fear.Washington Post, Washington, DC, p. C01.

Hwang, B.-F., Jaakkola, J., Guo, H.-R., 2008. Water disinfection by-products and the riskofspecific birth defects:a population-basedcross-sectional study in Taiwan. Environmental Health 7, 23.

Hwang, B.F., Jaakkola, J.J., 2003. Water chlorination and birthdefects: a systematic review and meta-analysis. Archives ofEnvironmental Health 58 (2), 83–91.

Hwang, B.F., Magnus, P., Jaakkola, J.J., 2002. Risk of specific birthdefects in relation to chlorination and the amount of naturalorganic matter in the water supply. American Journal ofEpidemiology 156 (4), 374–382.

IARC, 1991. Chlorinated drinking-water; chlorination by-products; some other halogenated compounds; cobalt andcobalt compounds. In: IARC Monographs on the Evaluation ofCarcinogenic Risks to Humans, vol. 52. International Agencyfor Research on Cancer, Lyon, France, pp. 1–544.

IARC, 2004. Some Drinking-water Disinfectants andContaminants, Including Arsenic. IARC Monographs on theEvaluation of Carcinogenic Risks to Humans. InternationalAgency for Research on Cancer, Lyon, France.

ICPS, 2000. Disinfectants and Disinfectant By-products.Environmental Health Criteria 216. Geneva, InternationalProgram on Chemical Safety. World Health Organization.

Jobb, D.B., Hunsinger, R.B., Meresz, O., Taguchi, V.Y., 1994.Removal of N-Nitrosodimethylamine from the Ohsweken (SixNations) Water Supply. Ontario Ministry of Environment,Toronto.

Jorgenson, T.A., Meierhenry, J.R., Rushbook, C.J., Bull, R.J.,Robinson, M., 1985. Carcinogenicity of chloroform in drinkingwater to male Osborne-Mendel rats and female B6C3F1 mice.Fundamental & Applied Toxicology 5, 760–769.

King, W.D., Marrett, L.D., 1996. Case-control study of bladdercancer and chlorination by-products in treated water (Ontario,Canada). Cancer Causes & Control 7 (6), 596–604.

Klotz, J.B., Pyrch, L., 1999. Neural tube defects and drinking waterdisinfection by-products. Epidemiology 10, 383–390.

Komulainen, H., Kosma, V.M., Vaittinen, S.L., Vartiainen, T.,KalisteKorhonen, E., Lotjonen, S., Tuominen, R.K., Tuomisto, J.,1997. Carcinogenicity of the drinking water mutagen 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone in the rat.Journal of the National Cancer Institute 89 (12), 848–856.

Krasner, S.W., McGuire, M.J., Jacangelo, J.G., Patania, N.L.,Reagan, K.M., Aieta, E.M., 1989. The occurrence of disinfectionby-products in US drinking water. Journal of the AmericanWater Works Association, 41–53.

Krasner, S.W., Weinberg, H.S., Richardson, S.D., Pastor, S.J.,Chinn, R., Sclimenti, M.J., Onstad, G.D., Thruston Jr., A.D.,2006. Occurrence of a new generation of disinfectionbyproducts. Environmental Science & Technology 40 (23),7175–7185.

Krewski, D., Gaylor, D., Soms, A., Szyskowicz, M., 1993.Correlation between carcinogenic potency and the maximumtolerated dose: implications for risk assessment. In: Issues inRisk Assessment. Appendix F. Committee on Risk AssessmentMethodology, National Research Council. National AcademiesPress. Washington, D.C.

Koivusalo, M., Hakulinen, T., Vartiainen, T., Pukkala, E.,Jaakkola, J.K., Tuomisto, J., 1998. Drinking water mutagenicityand urinary tract cancers: A population-based case-controlstudy in Finland. American Journal of Epidemiology 148 (7),704–712.

Kronberg, L., Holmblom, B., Reunanen, M., Tikkanen, L., 1988.Identification and quantification of the Ames mutageniccompound 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H0-furanone and its geometric isomer (E)-2-chloro-3-(dichloromethyl)-4-oxobutenoic acid in chlorine-treatedhumic water and drinking water extracts. EnvironmentalScience & Technology 22 (9), 1097–1103.

Langheld, K., 1909. Uber das verhalten von alpha-aminosaurengegen natriumhypochlorit. Berichte der DeutschenChemischen Gesellschaft 42 (2), 2360–2374.

Larson, J.L., Wolf, D.C., Butterworth, B.E., 1994. Inducedcytotoxicity and cell proliferation in the hepatotoxicity ofchloroform in female B6C3F1 mice: comparison ofadministration by gavage in corn oil vs. ad libitum in drinkingwater. Fundamental & Applied Toxicology 22, 90–102.

Larson, J.L., Wolf, D.C., Morgan, M.K., Butterworth, B.E., 1995.Toxicity and cell proliferation in the liver, kidneys and nasalpassages of female F-344 rats, induced by chloroformadministered by gavage. Food & Chemical Toxicology 33 (6),443–456.

LeCloirec, C., Martin, G., 1985. Evolution of amino acids in watertreatment and the effect of chlorination on amino acids. In:Jolley, R.L. (Ed.), Water Chlorination: Chemistry,Environmental Impact and Health Effects, vol. 5. LewisPublishers, Chelsea, MI, pp. 821–834.

Leiss, W., Hrudey, S.E., 1997. Dioxins or chemical stigmata. In:Powell, D., Leiss, W. (Eds.), Mad Cows and Mother’s Milk: ThePerils of Poor Risk Communication. McGill-Queens UniversityPress, Montreal, pp. 41–76.

Magnus, P., Jaakkola, J.J., Skrondal, A., Alexander, J., Becher, G.,Krogh, T., Dybing, E., 1999. Water chlorination and birthdefects. Epidemiology 10 (5), 513–517.

Marx, J., 1974. Drinking water: Another source of carcinogens?Science 186 (4166), 809–811.

Maugh, T.H., 1979. Cancer and environment: Higginson speaksout. Science 205 (4413), 1363–1364, 1366.

McDonald, T.A., Komulainen, H., 2005. Carcinogenicity of thechlorination disinfection by-product MX. Journal of

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 2 2091

Environmental Science & Health, Part C – EnvironmentalCarcinogenesis & Ecotoxicology Reviews 23 (2), 163–214.

McGeehin, M.A., Reif, J.S., Becher, J.C., Mangione, E.J., 1993. Case-control study of bladder cancer and water disinfectionmethods in Colorado. American Journal of Epidemiology 138(7), 492–501.

Melnick, R.L., Boorman, G.A., Dellarco, V., 1997. Waterchlorination, 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX), and potential cancer risk. Journal of theNational Cancer Institute 89 (12), 832–833.

Michaud, D.S., Kogevinas, M., Cantor, K.P., Villanueva, C.M.,Garcia-Closas, M., Rothman, N., Malats, N., Real, F.X., Serra, C.,Garcia-Closas, R., Tardon, A., Carrato, A., Dosemeci, M.,Silverman, D.T., 2007. Total fluid and water consumption andthe joint effect of exposure to disinfection by-products on riskof bladder cancer. Environmental Health Perspectives 115 (11),1569–1572.

Mills, C.J., Bull, R.J., Cantor, K.P., Reif, J., Hrudey, S.E., Huston, P.,1998. Workshop report. Health risks of drinking waterchlorination by-products: report of an expert working group.Chronic Diseases in Canada 19 (3), 91–102.

Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., Alvarez-Cohen, L., Sedlak, D.L., 2003. N-Nitrosodimethylamine (NDMA)as a drinking water contaminant: a review. EnvironmentalEngineering Science 20 (5), 389–404.

NCI, 1976. Carcinogenesis Bioassay of Chloroform. NationalCancer Institute, Bethesda, MD.

NHMRC, ARMCANZ, 1996. Australian Drinking Water Guidelines.National Health and Medical Research Council & Agriculturaland Resource Management Council of Australia and NewZealand, Canberra, Australia.

Nieuwenhuijsen, M.J., Toledano, M.B., Eaton, N.E., Fawell, J.,Elliott, P., 2000. Chlorination disinfection by-products in waterand their association with adverse reproductive outcomes:a review. Occupational and Environmental Medicine 57 (2),73–85.

NTP, 2006. Toxicology and Carcinogenesis Studies ofBromodichloromethane (CAS No. 75-27-4) in Male F344/N Ratsand Female B6C3F1 Mice (Drinking Water Studies). NationalToxicology Program, Research Triangle Park, NC.

Odom, R., Regli, S., Messner, M., Cromwell, J., Javdan, M., 1999.Benefit-cost analysis of the stage 1 D/DBP rule. JournalAmerican Water Works Association 91 (4), 137–147.

Peto, R., Gray, R., Brantom, P., Grasso, P., 1991. Dose and timerelationships for tumor induction in the liver and esophagusof 4080 inbred rats by chronic ingestion of N-nitrosodiethylamine or N-nitrosodimethylamine. CancerResearch 51 (23_Part_2), 6452–6469.

Pontius, F.W., 2000. Chloroform: science, policy and politics.Journal American Water Works Association 92 (5) 12, 14,16, 18.

Reif, J., Bachand, A., Andersen, M.E., 2000. Reproductive andDevelopmental Effects of Disinfection By-products. HealthCanada, Ottawa, Ontario, p. 83.

Reif, J.S., Hatch, M.C., Bracken, M., Holmes, L.B., Schwetz, B.A.,Singer, P.C., 1996. Reproductive and developmental effects ofdisinfection by-products in drinking water. EnvironmentalHealth Perspectives 104 (10), 1056–1061.

Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R.,DeMarini, D.M., 2007. Occurrence, genotoxicity, andcarcinogenicity of regulated and emerging disinfection by-products in drinking water: a review and roadmap forresearch. Mutation Research – Reviews in Mutation Research636 (1–3), 178–242.

Rodricks, J.V., 2007. 2006 American College of Toxicologydistinguished service award lecture: has quantitative riskassessment been of benefit to the science of toxicology.International Journal of Toxicology 26, 3–12.

Rook, J.J., 1974. Formation of haloforms during chlorination ofnatural waters. Water Treatment and Examination 23,234–243.

Rook, J.J., 1975. Uproar about chlorine. Environmental Science &Technology 9 (1), 92–93.

Sagan, C., 1996. The Demon-Haunted World: Science as a Candlein the Dark. Random House, New York.

Savitz, D.A., Singer, P., Hartmann, K.E., Herring, A., Weinberg, H.,Makarushka, C., Hoffman, C.S., Chan, R., Maclehose, R., 2005.Drinking Water Disinfection By-products and PregnancyOutcome. Awwa Research Foundation, U.S. EnvironmentalProtection Agency, Denver, 209 pp.

Savitz, D.A., Singer, P.C., Herring, A.H., Hartmann, K.E.,Weinberg, H.S., Makarushka, C., 2006. Exposure to drinkingwater disinfection by-products and pregnancy loss. AmericanJournal of Epidemiology 164 (11), 1043–1051.

Singer, P.C. (Ed.), 1999. Formation and Control of Disinfection By-products in Drinking Water. American Water WorksAssocation, Denver, CO.

Symons, J.M., 2001a. The early history of disinfection by-products: A personal chronicle (Part 1). EnvironmentalEngineer. 37 (1), 20–26.

Symons, J.M., 2001b. The early history of disinfection by-products: A personal chronicle (Part 1I). EnvironmentalEngineer. 37 (2), 7–15.

Symons, J.M., 1975. National Organics Reconnaissance Survey.Preliminary Assessment of Suspected Carcinogens in DrinkingWater (Appendices). U.S. Environmental Protection Agency,Washington, D.C., pp. 12–100.

Taguchi, V.Y., Jenkins, S.W., Wang, D.T., Palmentier, J.-P.,Reiner, E.J., 1994. Determination of N-nitrosodimethylamineby isotope dilution, high-resolution mass spectrometry.Canadian Journal of Applied Spectroscopy 39 (4), 87–91.

Tardiff, R.G., Carson, M.L., Ginevan, M.E., 2006. Updated weightof evidence for an association between adverse reproductiveand developmental effects and exposure to disinfectionby-products. Regulatory Toxicology & Pharmacology 45 (2),185–205.

USEPA, 1974. Draft Analytical Report New Orleans Water SupplyStudy (Lower Mississippi River Facility, Slidell, Louisiana). U.S.Environmental Protection Agency, Cincinnati, OH.

USEPA, 1976. Carcinogen assessment guideline. Federal Register41, 21402.

USEPA, 1986. Guidelines for carcinogen risk assessment. FederalRegister 51, 33992–34003.

USEPA, 1998a. National primary drinking water regulations:disinfectants and disinfection byproducts notice of dataavailability: proposed rule. Federal Register 63,15673–15692.

USEPA, 1998b. 40 CFR Parts 9, 141, and 142-disinfectants anddisinfection byproducts; final rule. National primary drinkingwater regulations. Federal Register 63, 69389–69476.

USEPA, 2004. An Examination of EPA Risk Assessment Principlesand Practices. Office of the Science Advisor Staff Paper. U.S.Environmental Protection Agency, Washington, D.C.

USEPA, 2005. Guidelines for Carcinogen Risk Assessment. RiskAssessment Forum. U.S. Environmental Protection Agency,Washington, D.C.

Villanueva, C.M., Cantor, K.P., Grimalt, J.O., Malats, N.,Silverman, D., Tardon, A., Garcia-Closas, R., Serra, C.,Carrato, A., Castano-Vinyals, G., Marcos, R., Rothman, N.,Real, F.X., Dosemeci, M., Kogevinas, M., 2007. Bladder cancerand exposure to water disinfection by-products throughingestion, bathing, showering, and swimming in pools.American Journal of Epidemiology 165 (2), 148–156.

Vinten-Johansen, P., Brody, H., Paneth, N., Rachman, S., Rip, M.,2003. Cholera, Chloroform and the Science of Medicine – a Lifeof John Snow. Oxford University Press, Oxford.

w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 2 0 5 7 – 2 0 9 22092

Waller, K., Swan, S.H., DeLorenze, G., Hopkins, B., 1998.Trihalomethanes in drinking water and spontaneousabortion. Epidemiology 9 (2), 134–140.

Warner, F., 1995. Extrapolation of dose response data for riskassessment. Process Safety and Environmental Protection,Transactions of the IChemE, Part B 73 (B4 Suppl.), S1–S54.

WHO, 1964. Prevention of Cancer. Technical Report Series. WorldHealth Organization, Geneva.

WHO, 1984. Guidelines for Drinking-Water Quality, first ed. WorldHealth Organization, Geneva.

WHO, 1993. Guidelines for Drinking Water Quality, second ed.World Health Organization, Geneva.

Wilkins 3rd, J.R., Comstock, G.W., 1981. Source of drinking waterat home and site-specific cancer incidence in Washington

County, Maryland. American Journal of Epidemiology 114 (2),178–190.

Xie, Y.F., 2003. Disinfection Byproducts in Drinking Water:Formation, Analysis and Control. CRC Press, Boca Raton,FL.

Zhao, Y.-Y., Boyd, J.M., Woodbeck, M., Andrews, R.C., Qin, F.,Hrudey, S.E., Li, X.-F., 2008. Formation of N-nitrosamines fromeleven disinfection treatments of seven different surfacewaters. Environmental Science & Technology 42 (13),4857–4862.

Zhao, Y.Y., Boyd, J., Hrudey, S.E., Li, X.F., 2006. Characterization ofnew nitrosamines in drinking water using liquidchromatography tandem mass spectrometry. EnvironmentalScience & Technology 40 (24), 7636–7641.

Articulo3

Documents

Transcript of Articulo3