Drinking Water; National Primary Drinking Water Regulations: Disinfectants and Disinfection Byproducts

SUMMARY: In this document, EPA is proposing maximum residual

disinfectant level goals (MRDLGs) for chlorine, chloramines, and

chlorine dioxide; maximum contaminant level goals (MCLGs) for four

trihalomethanes (chloroform, bromodichloromethane,

dibromochloromethane, and bromoform), two haloacetic acids

(dichloroacetic acid and trichloroacetic acid), chloral hydrate,

bromate, and chlorite; and National Primary Drinking Water Regulations

(NPDWRs) for three disinfectants (chlorine, chloramines, and chlorine

dioxide), two groups of organic disinfection byproducts (total

trihalomethanes (TTHMs)--a sum of the four listed above, and haloacetic

acids (HAA5)--a sum of the two listed above plus monochloroacetic acid

and mono- and dibromoacetic acids), and two inorganic disinfection

byproducts (chlorite and bromate). The NPDWRs consist of maximum

residual disinfectant levels or maximum contaminant levels or treatment

techniques for these disinfectants and their byproducts. The NPDWRs

also include proposed monitoring, reporting, and public notification

requirements for these compounds. This notice proposes the best

available technology (BAT) upon which the MRDLs and MCLs are based and

the BAT for purposes of issuing variances.

DATES: Written comments must be postmarked or hand-delivered by

December 29, 1994. Comments received after this date may not be

considered. Public hearings will be held at the addresses indicated

below under ``ADDRESSES'' on August 29 (and 30, if necessary) in

Denver, CO and on September 12 (and 13, if necessary) in Washington,

DC.

ADDRESSES: Send written comments on the proposed rule to Disinfectant/

Disinfection By-Products Comment Clerk, Drinking Water Docket (MC

4101), Environmental Protection Agency, 401 M Street, S.W., Washington,

D.C. 20460. Commenters are requested to submit three copies of their

comments and at least one copy of any references cited in their written

or oral comments. A copy of the comments and supporting documents are

available for review at the EPA, Drinking Water Docket (4101), 401 M

Street, S.W., Washington, DC 20460. For access to the docket materials,

call (202) 260-3027 between 9:00 a.m. and 3:30 p.m.

The Agency will hold public hearings on the proposal at two

different locations indicated below:

1. Denver Federal Center, 6th and Kipling Streets, Building 25, Lecture

Halls A and B (3d Street), Denver, CO 80225 on August 29 (and 30, if

necessary), 1994.

2. EPA Education Center Auditorium, 401 M Street SW., Washington, D.C.

20460, on September 12 (and 13, if necessary), 1994.

The hearings will begin at 9:30 a.m., with registration at 9:00

a.m. The Hearings will end at 4:00 p.m., unless concluded earlier.

Anyone planning to attend the public hearings (especially those who

plan to make statements) may register in advance by writing the D/DBPR

Public Hearing Officer, Office of Ground Water and Drinking Water

(4603), USEPA, 401 M Street, S.W., Washington, D.C. 20460; or by

calling Tina Mazzocchetti, (703) 931-4600. Meeting dates are tentative

and should be confirmed by calling the Safe Drinking Water Hotline

prior to making travel plans. Oral and written comments may be

submitted at the public hearing. Persons who wish to make oral

presentations are encouraged to have written copies (preferably three)

of their complete comments for inclusion in the official record.

Copies of draft health criteria, analytical methods, and regulatory

impact analysis documents are available at some Regional Offices listed

below and for a fee from the National Technical Information Service

(NTIS), U.S. Department of Commerce, 5285 Port Royal Road, Springfield,

Virginia 22161. The toll-free number is (800) 336-4700 or local at

(703) 487-4650.

FOR FURTHER INFORMATION CONTACT: General information may be obtained

from the Safe Drinking Water Hotline, telephone (800) 426-4791; Stig

Regli, Office of Ground Water and Drinking Water (4603), U.S.

Environmental Protection Agency, 401 M Street, SW., Washington, DC

20460, telephone (202) 260-7379; Tom Grubbs, Office of Ground Water and

Drinking Water (4603), U.S. Environmental Protection Agency, 401 M

Street, SW., Washington, DC 20460, telephone (202) 260-7270; or one of

the EPA Regional Office contacts listed below.

SUPPLEMENTARY INFORMATION:

EPA Regional Offices

I. Robert Mendoza, Chief, Water Supply Section, JFK Federal Bldg.,

Room 203, Boston, MA 02203, (617) 565-3610

II. Robert Williams, Chief, Water Supply Section, 26 Federal Plaza,

Room 824, New York, NY 10278, (212) 264-1800

III. Jeffrey Hass, Chief, Drinking Water Section (3WM41), 841

Chestnut Building, Philadelphia, PA 19107, (215) 597-9873

IV. Phillip Vorsatz, Chief, Water Supply Section, 345 Courtland

Street, Atlanta, GA 30365, (404) 347-2913

V. Charlene Denys, Chief, Water Supply Section, 77 W. Jackson Blvd.,

Chicago, IL 60604, (312) 353-2650

VI. F. Warren Norris, Chief, Water Supply Section, 1445 Ross Avenue,

Dallas, TX 75202, (214) 655-7155

VII. Ralph Flournoy, Chief, Water Supply Section, 726 Minnesota

Ave., Kansas City, KS 66101, (913) 234-2815

VIII. Doris Sanders, Chief, Water Supply Section, One Denver Place,

999 18th Street, Suite 500, Denver, CO 80202-2405, (303) 293-1424

IX. Bill Thurston, Chief, Water Supply Section, 75 Hawthorne Street,

San Francisco, CA 94105, (415) 744-1851

X. William Mullen, Chief, Water Supply Section, 1200 Sixth Avenue,

Seattle, WA 98101, (206) 442-1225.

Abbreviations used in this document.

AECL: Alternate enhanced coagulant level

AOC: Assimilable organic carbon

ASDWA: Association of State Drinking Water Administrators

AWWA: American Water Works Association

AWWARF: AWWA Research Foundation

BAC: Biologically active carbon

BAF: Biologically active filtration

BAT: Best Available Technology

BCAA: Bromochloroacetic acid

BDOC: Biodegradable organic carbon

BTGA: Best Technology Generally Available

CI: Confidence interval

CWS: Community Water System

DBP: Disinfection byproducts

D/DBP: Disinfectants and disinfection byproducts

D/DBPR: Disinfectants and disinfection byproducts rule

DBPP: Disinfection byproduct precursors

DBPRAM: DBP Regulatory Assessment model

DPD: N,N-diethyl-p-phenylenediamine

DWEL: Drinking Water Equivalent Level

EBCT: Empty bed contact time

EMSL: EPA Environmental Monitoring and Support Laboratory

(Cincinnati)

EPA: United States Environmental Protection Agency

ESWTR: Enhanced Surface Water Treatment Rule

FY: Fiscal year

GAC: Granular Activated Carbon

GWDR: Ground Water Disinfection Rule

GWSS: Ground Water Supply Survey

HAA5: Haloacetic acids (five)

HOBr: Hypobromous acid

IC: Ion chromotography

ICR: Information Collection Rule

IOC: Inorganic chemical

LOAEL: Lowest observed adverse effect level

LOQ: Limit of Quantitation

MCL: Maximum Contaminant Level (expressed as mg/l, 1,000 micrograms

(g) = 1 milligram (mg))

MCLG: Maximum Contaminant Level Goal

MDL: Method Detection Limit

MF: Modifying factor

mg/dl: Milligrams per deciliter

mg/l: Milligrams per liter

MGD: Million Gallons per Day

MRDL: Maximum Residual Disinfectant Level (as mg/l)

MRL: Minimum reporting level

MRDLG: Maximum Residual Disinfectant Level Goal

NCI: National Cancer Institute

ND: Not detected

NIPDWR: National Interim Primary Drinking Water Regulation

NOAEL: No observed adverse effect level

NOMS: National Organic Monitoring Survey

NORS: National Organics Reconnaissance Survey for Halogenated

Organics

NPDWR: National Primary Drinking Water Regulation

NTNCWS: Nontransient noncommunity water system

OBr: Hypobromite ion

OR: Odds ratio

PE: Performance evaluation

POE: Point-of-Entry Technologies

POU: Point-of-Use Technologies

ppb: Parts per billion

PQL: Practical Quantitation Level

PTA: Packed Tower Aeration

PWS: Public Water System

RIA: Regulatory Impact Analysis

RMCL: Recommended Maximum Contaminant Level

RNDB: Regulations Negotiation Data Base

RSC: Relative Source Contribution

SDWA: Safe Drinking Water Act, or the ``Act,'' as amended in 1986

SM: Standard Method

SMCL: Secondary Maximum Contaminant Level

SMR: Standardized mortality ratios

SOC: Synthetic Organic Chemical

SWTR: Surface Water Treatment Rule

THMFP: Trihalomethane formation potential

TOC: Total organic carbon

TTHM: Total trihalomethanes

TWG: Technologies Working Group

VOC: Volatile Synthetic Organic Chemical

WIDB: Water Industry Data Base

WS: Water Supply

Table of Contents

I. Summary of Today's Action

A. Applicability.

B. Proposed MRDLGs and MRDLs for disinfectants

C. Proposed MCLGs and MCLs for organic byproducts

D. Treatment technique for DBP precursors

E. Proposed Stage 1 MCLGs and MCLs for inorganic byproducts

F. Proposed BAT for disinfectants

G. Proposed BAT for organic byproducts

H. Proposed BAT for inorganic byproducts

I. Proposed Compliance Monitoring Requirements

J. Analytical Methods

K. Laboratory Certification Criteria

L. Variances and Exemptions

M. State Primacy, Recordkeeping, Reporting Requirements

N. System Reporting Requirements

O. D/DBP Stage 2 Rule requirements

P. Guidance

Q. Triennial Regulation Review

II. Statutory Authority

A. MCLGs, MCLs, and BAT

B. Variances and Exemptions

C. Primacy

D. Monitoring, Quality Control, and Records

E. Public Water Systems

F. Public Notification

III. Overview of Existing Interim Standard for TTHMs

IV. Overview of Preproposal Regulatory Development

A. October 1989 Strawman Rule

B. June 1991 Status Report on D/DBP rule development

C. Initiation of Regulatory Negotiation Process

V. Establishing MCLGs

A. Background

B. Proposed MRDLGs and MCLGs

1. Chlorine, hypochloriteion and hypochlorous acid

2. Chloramines

3. Epidemiology Studies of Chlorinated and Chloraminanted Water

4. Chlorine dioxide, chlorite, and chlorate

5. Chloroform

6. Bromodichloromethane

7. Dibromochloromethane

8. Bromoform

9. Dichloroacetic acid

10. Trichloroacetic acid

11. Chloral hydrate

12. Bromate

VI. Occurrence of TTHMs, HAA5, and other DBPs

A. Relationship of TTHMs, HAA5 to disinfection and source water

quality

B. Chlorination Byproducts

C. Other Disinfection Byproducts

1. Ozonation Byproducts

2. Chlorine Dioxide Byproducts

3. Chloramination Byproducts

VII. General Basis for Criteria of Proposed rule

A. Goals of regulatory negotiation

B. Concern for downside microbial risks and unknown risks from

DBPs of different technologies

C. Ecological concerns

D. Watershed protection

E. Narrowing of regulatory options through reg-neg process

VIII. Summary of the Proposed National Primary Drinking Water

Regulation for Disinfectants and Disinfection Byproducts

A. Schedule and coverage

B. Summary of DBP MCLs, BATs, and monitoring and compliance

requirments

C. Summary of disinfectant MRDLs, BATs, and Monitoring and

compliance requirements

D. Enhanced coagulation and enhanced softening requirements

E. Requirement for systems to use qualified operators

F. Basis for analytical method requirements

G. Public Notice Requirements

H. Variances and Exemptions

I. Reporting and Record Keeping requirements for PWSs

J. State Implementation Requirements

IX. Basis for Key Specific Criteria of Proposed Rule

A. 80/60 TTHM/HAA5 MCLs, enhanced coagulation requirements, and

BAT

1. basis for umbrella concept vs. individual MCLs

2. basis for level of stringency in MCLs, BAT, and concurrent

enhanced coagulation requirements

3. basis for enhanced coagulation and softening criteria

4. basis for GAC definitions

5. basis for monitoring requirements

B. Bromate MCL and BAT

C. Chlorite MCL and BAT

D. Chlorine MRDL and BAT

E. Chloramine MRDL and BAT

F. Chlorine dioxide MRDL and BAT

G. Basis for analytical method requirements

H. Basis for compliance schedule and applicability to different

groups of systems, timing with other regulations

I. Basis for qualified operator requirements and monitoring

plans

J. Basis for Stage 2 proposed MCLs

X. Laboratory Certification and Approval

A. PE-Sample Acceptance Limits for Laboratory Certification

B. Approval Criteria for Disinfectants and Other Parameters

C. Other Laboratory Performance Criteria

XI. Variances and Exemptions

A. Variances

B. Exemptions

XII. State Implementation

A. Special primacy requirements

B. State recordkeeping

C. State reporting

XIII. System Reporting and Recordkeeping Requirements

XIV. Public Notice Requirements

XV. Economic Analysis

A. Executive Order 12866

B. Predicted cost impacts on public water systems

1. Compliance treatment cost forecasts

2. Compliance treatment forecasts

3. DBP exposure estimates

4. System level cost estimates

5. Effect on household costs

6. Monitoring and State implementation costs, labor burden

estimates

C. Concepts of cost analysis

D. Benefits

XVI. Other Requirements

A. Consultation with State, Local, and Tribal Governments

B. Regulatory Flexibility Act

C. Paperwork Reduction Act

D. National Drinking Water Advisory Council and Science Advisory

Board

XVII. Request for Public Comment

XVIII. References and Public Docket

I. Summary of Today's Action

In 1992 EPA initiated a negotiated rulemaking to develop a

disinfectant/disinfection byproduct rule. The Agency decided to use the

negotiated rulemaking process because it believed that the available

occurrence, treatment, and health effects data were inadequate to

address EPA's concerns about the tradeoff between risks from

disinfectants and disinfection byproducts and microbial pathogen risk,

and wanted all stakeholders to participate in the decision-making on

setting proposed standards. The negotiators included State and local

health and regulatory agency staff and elected officials, consumer

groups, environmental groups, and representatives of public water

systems. The group met from November 1992 through June 1993.

Early in the process, the negotiators agreed that large amounts of

information necessary to understand how to optimize the use of

disinfectants to concurrently minimize microbial and disinfectant/

disinfection byproduct risk were unavailable.

Therefore, the group agreed to propose a disinfectant/disinfection

byproduct rule to extend coverage to all community water systems that

use disinfectants, reduce the current total trihalomethane (TTHM)

maximum contaminant level (MCL), regulate additional disinfection

byproducts, set limits for the use of disinfectants, and reduce the

level of compounds that may react with disinfectants to form

byproducts. These requirements were based on available information. The

group further agreed that revisions to the current Surface Water

Treatment Rule might be required at the same time to ensure that

microbial risk is not increased as byproduct rules go into effect.

Finally, the group agreed that additional information on health risk,

occurrence, treatment technologies, and analytical methods needed to be

developed in order to better understand the risk-risk tradeoff, whether

further control was needed, and how to accomplish this overall risk

reduction.

The outcome of the negotiation was three rules: a Disinfectant/

Disinfection Byproduct rule (this notice), an Enhanced Surface Water

Treatment Rule (also proposed today and appearing separately in today's

Federal Register), and an Information Collection Rule (proposed

February 10, 1994, 59 FR 6332). The Information Collection Rule will

provide information necessary to determine whether the Enhanced Surface

Water Treatment Rule needs to be promulgated and, if so, what

requirements it should set. The Information Collection Rule will also

provide information on the need for, and content of, long-term rules.

The schedule to produce these rules has also been negotiated and is

provided elsewhere in this document. A summary of today's rule follows.

A. Applicability. This action applies to all community water

systems and nontransient noncommunity water systems that add a

disinfectant during any part of the treatment process including

addition of a residual disinfectant. In addition, certain provisions

apply to transient noncommunity water systems that use chlorine

dioxide.

B. Proposed MRDLGs and MRDLs for disinfectants. EPA is proposing

the following maximum disinfectant residual level goals and maximum

residual disinfectant levels.

------------------------------------------------------------------------

Disinfectant Residual MRDLG (mg/l) MRDL (mg/l)

------------------------------------------------------------------------

(1) Chlorine................ 4 (as Cl2).......... 4.0 (as Cl2).

(2) Chloramines............. 4 (as Cl2).......... 4.0 (as Cl2).

(3) Chlorine dioxide........ 0.3 (as ClO2)....... 0.8 (as ClO2).

------------------------------------------------------------------------

C. Proposed MCLGs and MCLs for organic byproducts. EPA is proposing

the following maximum contaminant level goals and maximum contaminant

levels.

------------------------------------------------------------------------

MCLG(mg/ MCL(mg/

l) l)

------------------------------------------------------------------------

Total trihalomethanes (TTHM)....................... \1\N/A 0.080

Haloacetic acids (five) (HAA5)..................... \2\N/A .060

Chloroform......................................... 0 \1\N/A

Bromodichloromethane............................... 0 \1\N/A

Dibromochloromethane............................... 0.06 \1\N/A

Bromoform.......................................... 0 \1\N/A

Dichloroacetic acid................................ 0 \2\N/A

Trichloroacetic acid............................... 0.3 \2\N/A

Chloral hydrate.................................... 0.04 \3\N/A

------------------------------------------------------------------------

\1\Total trihalomethanes are the sum of the concentrations of

bromodichloromethane, dibromochloromethane, bromoform, and chloroform.

\2\Haloacetic acids (five) are the sum of the concentrations of mono-,

di-, and trichloroacetic acids and mono- and dibromoacetic acids.

\3\EPA did not set an MCL for chloral hydrate because the TTHM and HAA5

MCLs and the treatment technique (i.e., enhanced coagulation) for

disinfection byproduct precursor removal will control for chloral

hydrate. (See Section IX.)

D. Treatment Technique for DBP Precursors. EPA is proposing that

water systems that use surface water or ground water under the direct

influence of surface water and use conventional filtration treatment be

required to remove specified amounts of organic materials (measured as

total organic carbon) that may react with disinfectants to form

disinfection byproducts. Removal would be achieved through a treatment

technique (enhanced coagulation or enhanced softening) unless the

system met certain criteria.

E. Proposed Stage 1 MCLGs and MCLs for inorganic by-products. EPA

is proposing the following maximum contaminant level goals and maximum

contaminant levels.

------------------------------------------------------------------------

MCLG(mg/

l) MCL(mg/l)

------------------------------------------------------------------------

Chlorite.......................................... 0.08 1.0

Bromate........................................... 0 0.010

------------------------------------------------------------------------

F. Proposed BAT for disinfectants. EPA is proposing the following

best available technologies for limiting residual disinfectant

concentrations in the distribution system.

Chlorine residual--control of treatment processes to reduce

disinfectant demand and control of disinfection treatment processes to

reduce disinfectant levels

Chloramine residual--control of treatment processes to reduce

disinfectant demand and control of disinfection treatment processes to

reduce disinfectant levels

Chlorine dioxide residual--control of treatment processes to reduce

disinfectant demand and control of disinfection treatment processes to

reduce disinfectant levels.

G. Proposed BAT for organic byproducts. EPA is proposing the

following best available technologies for control of organic

disinfection byproducts in each stage of the rule.

1. Proposed Stage 1 BAT for organic by-products. Total

trihalomethanes--enhanced coagulation or GAC10, with chlorine as the

primary and residual disinfectant. Total haloacetic acids--enhanced

coagulation or GAC10, with chlorine as the primary and residual

disinfectant.

2. Proposed Stage 2 BAT for organic byproducts. Total

trihalomethanes--enhanced coagulation and GAC10, or GAC20; with

chlorine as the primary and residual disinfectant. Total haloacetic

acids--enhanced coagulation and GAC10, or GAC20; with chlorine as the

primary and residual disinfectant.

H. Proposed BAT for inorganic by-products. EPA is proposing the

following best available technologies for control of inorganic

disinfection byproducts.

Chlorite--control of treatment processes to reduce disinfectant demand

and control of disinfection treatment processes to reduce disinfectant

levels.

Bromate--control of ozone treatment process to reduce production of

bromate.

I. Proposed Compliance Monitoring Requirements. Compliance

monitoring requirements are explained in Section IX of the preamble and

were developed during the negotiated rulemaking. EPA has developed

routine and reduced monitoring schemes that address the health effects

of each disinfectant or contaminant in an individually appropriate

manner.

J. Analytical Methods. EPA is proposing to withdraw one method for

measurement of chlorine residual and to approve three new methods for

measurement of chlorine residuals. EPA is proposing to approve one new

method for measurement of trihalomethanes; two new methods for

measurement of haloacetic acids; one new method for measurement of

bromate, chlorite, and bromide; and two new methods for measurement of

total organic carbon.

K. Laboratory Certification Criteria. Consistent with other

drinking water regulations, EPA is proposing that only certified

laboratories be allowed to analyze samples for compliance with the

proposed MCLs and treatment technique requirements. For disinfectants

and other specified parameters in today's rule that the Agency believes

can be adequately measured by other than certified laboratories and for

which there is a good reason to allow analysis at other locations

(e.g., for samples which normally deteriorate before reaching a

certified laboratory, especially when taken at remote locations), EPA

is requiring that such analyses be conducted by a party acceptable to

EPA or the State.

L. Variances and Exemptions. Variances and exemptions will be

permitted.

M. State Primacy, Recordkeeping, Reporting Requirements.

Requirements for States to maintain primacy are listed in Section XII

of the preamble. In addition to routine requirements, EPA has included

special primacy requirements.

N. System Reporting Requirements. System reporting requirements

remain consistent with requirements in previous rules.

O. D/DBP Stage 2 Rule requirements. EPA is proposing a total

trihalomethane MCL of 0.040 mg/l and a haloacetic acid (five) MCL of

0.030 mg/l, to apply only to systems using surface water or ground

water under the direct influence of surface water and serving at least

10,000 persons, as part of a plan to develop new standards which

incorporates the results of additional research conducted under the

Information Collection Rule (59 FR 6332).

P. Guidance. EPA is in the process of developing guidance for both

systems and States for implementation of this rule.

Q. Triennial Regulation Review. Under the provisions of the Safe

Drinking Water Act (SDWA or the Act) (Section 1412(b)(9)), the Agency

is required to review national primary drinking water regulations at

least once every three years. As mentioned previously, today's proposed

rule revises, updates, and (when promulgated) supersedes the

regulations for total trihalomethanes, initially published in 1979.

Since that time, there have been significant changes in technology,

treatment techniques, and other regulatory controls that provide for

greater protection for health of persons. As such, in proposing today's

rule, EPA has analyzed innovations and changes in technology and

treatment techniques that have occurred since promulgation of the

initial TTHM regulations. This analysis, contained primarily in the

cost and technology document supporting this proposal, supports

amendment of the TTHM regulation for the greater protection of persons.

EPA believes that the innovations and changes in technology and

treatment techniques will result in amendments to the TTHM regulations

that are feasible within the meaning of SDWA Section 1412(b)(9).

II. Statutory Authority

Section 1412 of the Safe Drinking Water Act, as amended in 1986

(``SDWA'' or ``the Act''), requires EPA to publish Maximum Contaminant

Level Goals (MCLGs) and promulgate National Primary Drinking Water

Regulations (NPDWRs) for contaminants in drinking water which may cause

any adverse effect on the health of persons and which are known or

anticipated to occur in public water systems. Under Section 1401, the

NPDWRs are to include Maximum Contaminant Levels (MCLs) and ``criteria

and procedures to assure a supply of drinking water which dependably

complies'' with such MCLs. Under Section 1412(b)(7)(A), if it is not

economically or technically feasible to ascertain the level of a

contaminant in drinking water, EPA may require the use of a treatment

technique instead of an MCL.

Under Section 1412(b), EPA was to establish MCLGs and promulgate

NPDWRs for 83 contaminants by June 19, 1989. An additional 25

contaminants are to be regulated every 3 years. To meet this latter

requirement, EPA has developed a list of contaminants (National

Drinking Water Priority List; 53 FR 1892) including pesticides, organic

and inorganic elements or compounds, and disinfectants and disinfection

by-products (D/DBP), plus the protozoan Cryptosporidium. From this

list, EPA is to choose at least 25 contaminants for regulation every

three years. Today's regulatory proposal represents part of the first

group of 25 chemicals to be regulated. Both the general contaminants

(organics, inorganics, and pesticides), and the D/DBPs were considered

for regulation. In today's notice, EPA is proposing to regulate certain

disinfectants and disinfection byproducts; Cryptosporidium is proposed

to be regulated in a separate Notice today.

In October of 1990, EPA entered into a consent order with Citizens

Concerned about Bull Run Inc. regarding a timeframe for proposing the

first group of 25. The consent decree stipulated a June 1993 date for

proposal. That decree was subsequently amended to establish a proposal

date of May 30, 1994, for the Disinfectants/Disinfection Byproducts

Rule and a proposal date of February 28, 1995, for the other

contaminants that comprise the required group of 25.

A. MCLGs, MCLs, and BAT

Under Section 1412 of the Act, EPA is to establish MCLGs at the

level at which no known or anticipated adverse effects on the health of

persons occur and which allow an adequate margin of safety. MCLGs are

nonenforceable health goals based only on health effects and exposure

information.

MCLs are enforceable standards which the Act directs EPA to set as

close to the MCLGs as feasible. ``Feasible'' means feasible with the

use of the best technology, treatment techniques, and other means which

the Administrator finds available (taking cost into consideration)

after examination for efficacy under field conditions and not solely

under laboratory conditions (SDWA, section 1412(b)(5)). Also, the SDWA

requires the Agency to identify the best available technology (BAT)

which is feasible for meeting the MCL for each contaminant.

Also, in this proposal, EPA is introducing several new terms--

``maximum residual disinfectant level goals (MRDLGs)'' and ``maximum

residual disinfectant levels (MRDLs)''--to reflect the fact that these

substances have beneficial disinfection properties. As with MCLGs, EPA

has established MRDLGs at the level at which no known or anticipated

adverse effects on the health of persons occur and which allow an

adequate margin of safety. MRDLGs are nonenforceable health goals based

only on health effects and exposure information and do not reflect the

benefit of the addition of the chemical for control of waterborne

microbial contaminants.

MRDLs are enforceable standards, analogous to MCLs, which recognize

the benefits of adding a disinfectant to water on a continuous basis

and in addressing emergency situations such as distribution system pipe

breaks. As with MCLs, EPA has set the MRDLs as close to the MRDLGs as

feasible. The Agency has also identified the best available technology

(BAT) which is feasible for meeting the MRDL for each disinfectant.

B. Variances and Exemptions

Section 1415 authorizes the State to issue variances from NPDWRs

(the term ``State'' is used in this preamble to mean the State agency

with primary enforcement responsibility for the public water supply

system program or EPA if the State does not have primacy). The State

may issue a variance if it determines that a system cannot comply with

an MCL despite application of the best available technology (BAT).

Under Section 1415, EPA must propose and promulgate its finding of the

best available technology, treatment techniques, or other means

available for each contaminant, for purposes of section 1415 variances,

at the same time that it proposes and promulgates a maximum contaminant

level for such contaminant. EPA's finding of BAT, treatment techniques,

or other means for purposes of issuing variances may vary among

systems, depending upon the number of persons served by the system or

for other physical conditions related to engineering feasibility and

costs of complying with MCLs, as considered appropriate by EPA. The

State may not issue a variance to a system until it determines that an

unreasonable risk to health (URTH) does not exist. When a State grants

a variance, it must at the same time prescribe a schedule for (1)

compliance with the NPDWR and (2) implementation of any additional

control measures.

Under Section 1416(a), the State may exempt a public water system

from any MCL or treatment technique requirement if it finds that (1)

due to compelling factors (which may include economic factors), the

system is unable to comply, (2) the system was in operation on the

effective date of the MCL or treatment technique, or, for a newer

system, that no reasonable alternative source of drinking water is

available to that system, and (3) the exemption will not result in an

unreasonable risk to health. Under section 1416(b), at the same time it

grants an exemption, the State is to prescribe a compliance schedule

and a schedule for implementation of any required interim control

measures. The final date for compliance may not exceed three years

after the initial date of issuance unless the public water system

establishes that: (1) the system cannot meet the standard without

capital improvements which cannot be completed within the period of

such exemption; (2) the system has entered into an agreement to obtain

financial assistance for necessary improvements; or (3) the system has

entered into an enforceable agreement to become part of a regional

public water system. For systems which serve 500 or fewer service

connections and which need financial assistance to come into

compliance, the State may renew the exemption for additional two-year

periods if the system is taking all practicable steps to meet the above

requirements.

For exemptions resulting from a NPDWR promulgated after June 19,

1986, the system's final compliance date must be within 12 months of

issuance of the exemption. However, the State may extend the final

compliance date for up to three years if the public water system shows

that capital improvements to meet the MCL or treatment technique

requirement cannot be completed within the exemption period and, if the

system needs financial assistance for the improvements, it has an

agreement to obtain this assistance or the system has an enforceable

agreement to become part of a regional public water system. For systems

that have 500 or fewer service connections that need financial

assistance to comply with the MCLs, the State may renew the exemption

for additional two-year periods if the system is taking all practicable

steps to comply.

C. Primacy

As indicated above, States, territories, and Indian Tribes may

assume primary enforcement responsibility (primacy) for public water

systems under Section 1413 of the SDWA. To date, 55 States and

territories have primacy. To assume or retain primacy, States,

territories, or Indian Tribes need not adopt the MCLGs but must adopt,

among other things, NPDWRs (i.e., MCLs, monitoring, analytical, and

reporting requirements) that are no less stringent than those EPA

promulgates.

D. Monitoring, Quality Control, and Records

Under Section 1401(1)(D) of the Act, NPDWRs are to contain

``criteria and procedures to assure a supply of drinking water which

dependably complies with such maximum contaminant levels; including

quality control and testing procedures to insure compliance with such

levels * * *.''

E. Public Water Systems

Public water systems are defined in section 1401 of the Act as

those systems which provide piped water for human consumption and have

at least 15 connections or regularly serve at least 25 people. By

regulation EPA has divided public water systems into community,

nontransient noncommunity, and (transient) noncommunity water systems.

Community water systems (CWSs) serve at least 15 service connections

used by year-round residents or regularly serve at least 25 year-round

residents (40 CFR 141.2). Nontransient noncommunity water systems

(NTNCWSs) regularly serve at least 25 of the same people over six

months of the year. Schools and factories which serve water to 25 or

more of the same people for six or more months of the year are examples

of NTNCWSs. Transient noncommunity systems, by definition, are all

other public water systems. Transient noncommunity systems may include,

for example, restaurants, gas stations, campgrounds, and churches.

This rule would apply to all CWSs, all NTNCWSs, and any transient

noncommunity water systems that use chlorine dioxide as a disinfectant

or oxidant.

F. Public Notification

Section 1414(c) of the Act requires the owner or operator of a

public water system which does not comply with an applicable maximum

contaminant level or treatment technique, testing procedure, or Section

1445(a) (unregulated contaminant) monitoring requirement to give notice

to the persons served by the system. Notice must also be given if a

variance or exemption is in effect or the system fails to comply with a

compliance schedule resulting from a variance or exemption. EPA's

public notification regulations are codified at 40 CFR Section 141.32.

Those regulations were amended by EPA on October 28, 1987 (52 FR

41534).

III. Overview of Existing Interim Standard for TTHMS

In 1974, researchers in The Netherlands and the United States

clearly demonstrated that total trihalomethanes (TTHMs) are formed as a

result of drinking water chlorination (Rook, 1974; Bellar et al, 1974).

EPA subsequently conducted surveys confirming widespread occurrence of

TTHMs in chlorinated water supplies (Symons, 1975; USEPA, 1978). During

this time toxicological studies became available which supported the

contention that chloroform, one of the four trihalomethanes, is

carcinogenic in at least one strain of rat and one strain of mouse

(National Academy of Sciences, 1977).

EPA then set an interim maximum contaminant level (MCL) for the

TTHMs of 100 g/l as an annual average in November 1979 (USEPA,

1979). This standard was based on the need to balance the requirement

for continued disinfection of water to reduce exposure to pathogenic

microorganisms while simultaneously lowering exposure to animal

carcinogens like chloroform.

The interim TTHM standard only applies to systems serving at least

10,000 people that add a disinfectant (oxidant) to the drinking water

during any part of the treatment process. At their discretion, States

are allowed to extend coverage to smaller size systems. About 80

percent of the smallest systems are served by groundwater systems that

are mostly low in THM precursor content (USEPA, 1979).

The proportion of these small groundwater systems that use chlorine

is less than that of large systems; currently, less than half of these

systems disinfect. Also, the shorter hydraulic detention and chlorine

contact times in the small system distribution systems results in lower

TTHM concentrations. Therefore, drinking water systems serving less

than ten thousand people are less likely to have high concentrations of

TTHMs.

Moreover, these small systems are most likely to have greater risks

of significant microbiological contamination, especially if they reduce

or eliminate chlorination. In 1979, the majority of outbreaks

attributable to inadequate disinfection occurred in small systems.

Further, small systems have limited or no access to the financial

resources and technical expertise needed for TTHM control. Therefore,

EPA concluded that small system resources would best be spent on

maintaining and improving microbiological quality and safety. The

revised drinking water regulations now under consideration will extend

to these small systems as required by the Safe Drinking Water Act

Amendments of 1986 (P.L. 99-339, 1986). EPA will also be considering

disinfection as a treatment technique requirement and maximum

contaminant levels (MCLs) for the residual disinfectants. The impacts

these requirements will have on small systems is an important component

of the regulation development process.

Technology Basis for the Interim TTHM Standard

When an MCL is established for TTHMs or any other contaminant that

can be measured, EPA is not required to specify any particular method

for achieving that standard. Instead, the requirement for the interim

regulations was to set an MCL which could be achieved using technology

generally available in 1974. Three general control alternatives were

available:

(1) use of a disinfectant (oxidant) that does not generate (or produces

less) THMs in water;

(2) treatment to lower precursor concentrations prior to chlorination;

and

(3) treatment to remove THMs after their formation.

There are many possible choices among these broad options and in

some cases a combination of approaches might be necessary. The ultimate

choice was left up to the water supplier based on its individual

circumstances.

EPA's evaluation led to the following conclusions concerning

generally available technologies for setting the TTHM MCL:

(1) alternate oxidants like ozone, chloramines, and chlorine dioxide

are available;

(2) precursor removal strategies like changing the point of

disinfection, off-line raw water storage, and improved coagulation are

available; and,

(3) precursor removal using granular activated carbon (GAC) as a

replacement for existing filter media with a regeneration frequency of

one year is feasible as well as biologically activated carbon (ozone

plus GAC) with a regeneration frequency of every two years.

Three conditions concerning modifications of disinfection processes

were also proposed by EPA:

(1) the total quantity of chlorine dioxide added during the treatment

process should not exceed 1 mg/l;

(2) chloramines should not be utilized as a primary disinfectant; and

(3) monitoring for heterotrophic plate count bacteria (HPC) should be

conducted as determined by the State, but at least every day for a

minimum of one month prior to and six months subsequent to the

modifications.

These recommendations concerning disinfection, although useful,

were deleted from the final regulation to allow States greater

discretion. The basis for the MCL became alternate oxidants and

precursor removal.

Technology Basis for Variances

Later, in 1983, EPA promulgated regulations specifying best

technology generally available for obtaining variances (USEPA, 1983). A

variance is granted by the State when a system has installed the best

technology generally available as specified in the regulation and still

cannot meet the MCL. The best technologies generally available for

variances to the TTHM MCL are:

(1) Use chloramines as an alternate or supplemental disinfectant or

oxidant.

(2) Use chlorine dioxide as an alternate or supplemental disinfectant

or oxidant.

(3) Improve existing clarification for THM precursor reduction.

(4) Move the point of chlorination to reduce TTHM formation and, where

necessary, substituting for the use of chlorine as a pre-oxidant

chloramines, chlorine dioxide, or potassium permanganate.

(5) Use of powdered activated carbon for THM precursor or TTHM

reduction seasonally or intermittently at dosages not to exceed 10 mg/l

on an annual average basis.

EPA also identified Group II technologies, which are not

``generally available,'' but may be available to some systems:

(1) Introduction of off-line water storage for THM precursor reduction.

(2) Aeration for TTHM reduction, where geographically and

environmentally appropriate.

(3) Introduction of clarification where not currently practiced.

(4) Consideration of alternative sources of raw water.

(5) Use of ozone as an alternate or supplemental disinfectant or

oxidant.

Note that GAC and BAC are not mentioned as either Group I or Group

II technologies even though they were discussed as technologies for

standard setting purposes (USEPA, 1979). EPA concluded in its cost and

technologies document for the removal of trihalomethanes from drinking

water that (USEPA, 1981):

(1) GAC in the sand replacement mode of operation is often

inappropriate due to the short performance life and high frequency of

regeneration required to achieve substantial TTHM or THM-formation

potential reduction;

(2) the finding took into consideration costs, but primarily was made

due to the complexities of the modifications to prior unit operations,

i.e., disinfection, and in the logistics of the carbon replacement;

(3) greater operating, maintenance, and monitoring than for other

treatments; and

(4) on-site regeneration had only been demonstrated at one U.S. site.

Thus, EPA decided to defer the decision to include GAC and BAC as

best generally available technology for granting variances under the

Safe Drinking Water Act Amendments of 1974.

EPA also did not list ozonation as being ``generally available''

because:

(1) lack of experience in the U.S.;

(2) mixed results in experimental studies; and

(3) most States require a residual in the distribution system which is

not obtainable with ozone.

Thus, EPA decided to defer the decision to include ozone as best

generally available technology for granting variances under the Safe

Drinking Water Act Amendments of 1974.

Economic Impacts of the Interim Standard

Currently, there are 2,700 community water supply systems serving

at least 10,000 people required to comply with the interim TTHM

regulation. In 1988, a survey of large systems found that, on average,

the MCL of 0.10 mg/l had reduced the concentration of TTHMs by 40 to 50

percent (McGuire and Meadows, 1988). Of these, 33 were in violation of

the standard in FY88 (average=115 g/l, range=108-180

g/l). However, by FY92, only nine systems (a decrease of 73

percent) violated the requirements, for a total of 14 violations. Seven

of the nine violating systems and 12 of the 14 violations occurred in

systems serving 10,000 to 50,000 people. This indicates that even when

systems violate, they are able to return to compliance after one or two

violations of the running annual average.

In 1979, approximately 500 systems were estimated to exceed 100

g/l TTHMs. Most of these were able to come into compliance

with minor modifications of chlorination practices. A smaller portion

used alternate oxidants like chlorine dioxide and chloramines. No

system installed ozone or GAC to meet the interim TTHM regulations.

Compliance with the interim TTHM standard involved estimated capital

expenditures of between $31 million and $102 million and yearly

operating and maintenance costs of between $8 million and $29 million

for systems required to comply with the TTHM MCL (i.e., community water

systems serving a population of at least 10,000 people) (McGuire and

Meadows, 1988).

IV. Overview of Preproposal Regulatory Development

A. October 1989 Strawman Rule

1. Purpose. EPA was required to develop rules for additional

contaminants under the 1986 Amendments to the Act. In order to solicit

public comment in developing a rule, EPA released a strawman rule

(preproposal draft) in October 1989. A strawman was used because of the

complexity of the problem, the large amount of (occasionally

contradictory) information, and the ability to reorient the rule

approach based on public comment or new data. In this strawman, EPA

included a lead option of setting MCLGs and MCLs for TTHMs, haloacetic

acids, chlorine, chloramines, chlorine dioxide, chlorite, and chlorate.

The Agency also identified potential add-on compounds: chloropicrin,

cyanogen chloride, hydrogen peroxide, bromate, iodate, and

formaldehyde. Some of these compounds could also conceivably be used as

surrogate monitoring compounds for the compounds identified in Table

IV-1 below.

Additional Candidate Byproduct Compounds

------------------------------------------------------------------------

Chlorination byproducts Ozonation byproducts

------------------------------------------------------------------------

--Individual THMs: chloroform, --Aldehydes: acetaldehyde,

bromodichloromethane, hexanal, heptanal.

dibromochloromethane, bromoform.

--Individual haloacetic acids: mono-, --Organic acids.

di-, and trichloroacetic acids; mono- --Ketones.

and dibromoacetic acids. --Epoxides.

--Individual haloacetonitriles: di- --Peroxides.

and trichloroacetonitrile; --Nitrosamines.

bromochloroacetonitrile,

dibromoacetonitrile.

--Haloketones: 1,1 di- and 1,1,1-

trichloropropanone.

--Chlorophenols: 2-; 2,4-di; and --N-oxy compounds.

2,4,6-trichlorophenol. --Quinones.

--Others: chloral hydrate, N- --Bromine substituted compounds.

organochloroamines.

------------------------------------------------------------------------

In addition, the strawman provided that EPA would set treatment

technique requirements or provide guidance for control of the

following: MX, as a surrogate for mutagenicity; total oxidizing

substances, as a surrogate for organic peroxides and epoxides; and

assimilable organic carbon, as a surrogate for microbiological quality

of oxidized waters. Monitoring parameters based on the particular

disinfection process were also identified.

As BAT, EPA included precursor removal (conventional treatment

modifications, GAC of up to 30 minute duration and three months

regeneration), alternate oxidants (ozone plus chloramines, chlorine

dioxide with chlorite removal plus chloramines), and byproduct removal

(aeration, GAC adsorption, reducing agents, AOC removal). Each of the

options had problems. GAC was not universally applicable to all waters

for either precursor removal or DBP removal. Membranes were not

included as BAT because of lack of full-scale experience.

As lead options, EPA included a TTHM MCL of 25 or 50 g/l

and other MCLs based on feasibility analyses similar to those that

would be used to develop the TTHM MCL.

2. Summary of Public Comments. Several commentors expressed a

desire for EPA to look at coordination of requirements with those for

other regulations, including issues such as requirements for

maintenance of distribution system disinfectant residuals and system

optimization for multiple contaminants. Many commentors were concerned

about the lack of health data and the interpretation of existing data.

Many system operators were also concerned about the effects of

modifying their treatment processes to meet DBP MCLs. These concerns

included lowered microbiological protection, creation of conditions

that favored distribution system microbiological growth (e.g., use of

ozone would create biodegradable organics and use of chloramines would

create a nitrogen source), and creation of other environmental problems

when changing treatment (e.g., residual handling with precursor removal

and GAC regeneration). While commentors expressed concern about use of

alternate disinfectants, several offered to provide data and others

recommended epidemiological studies in systems with long histories of

alternative disinfectant use.

B. June 1991 Status Report on D/DBP Rule Development

1. Purpose and transition from Strawman Rule. EPA published a

status report on the development of D/DBPR in June 1991 that was

designed to indicate the Agency's thinking on rule criteria. The status

report indicated that EPA was considering extending coverage under the

rule to all nontransient systems (instead of just those serving at

least 10,000 people, as under the 1979 TTHM rule) and proposing a

shorter list of compounds for regulation than were included in the 1989

strawman. The 1991 list included disinfectants (chlorine, chloramines,

and chlorine dioxide), THMs, haloacetic acids, chloral hydrate,

bromate, chlorite, and chlorate). For both THMs and haloacetic acids,

three options were included: MCLs for individual compounds, a single

MCL for the total, and a combination of the two. Individual MCLs were

considered because health risks for compounds differed, in some cases

significantly. The total MCL was considered because of the precedent

established in the 1979 TTHM rule and to act as a surrogate to limit

other DBPs for which the Agency lacked adequate health effects and/or

occurrence data.

The list of compounds was shorter than that in the 1989 strawman

for several reasons. Several compounds were deleted because they did

not appear to pose significant health effects at levels present in

drinking water (e.g., haloacetonitriles, chloropicrin). Others were

deleted because the health risks were not expected to be adequately

characterized in time for rule proposal (e.g., certain aldehydes and

organic peroxides), although it was noted that these compounds might be

regulated in the future when more data became available.

2. Major issues. In the status report, EPA identified several major

issues that needed to be considered as the D/DBP rule was developed.

The first was that of trade-offs with microbial and DBP risks. The goal

was to ensure that the water remained microbiologically safe at the

level that disinfectant and DBP MCLs were set. The discussion raised

questions regarding uncertainties in defining microbial and DBP risks,

levels of risks that would be considered acceptable and at what cost,

and defining practical (implementable) criteria to demonstrate that an

achievable risk had been reached.

The second issue was the use of alternate disinfectants to limit

chlorination byproducts. The Agency recognized that while alternate

disinfection schemes (e.g., ozone and chloramines) could greatly reduce

byproducts typical of chlorination, little was known about the

byproducts of the alternate disinfectants and their associated health

risks. EPA did not want to promulgate a standard that encouraged the

shift to alternate disinfectants unless the associated risks (including

both those from byproducts and differential microbial risks from a

change in disinfectants) were adequately understood.

The third issue was integration with the Surface Water Treatment

Rule. Although the rule only mandated 3-log removal or inactivation of

Giardia and 4-log of viruses, EPA guidance recommended higher levels

for poorer quality source waters. EPA was concerned that systems would

reduce microbial protection to levels nearer to the regulatory

requirements by reducing disinfection and possibly greatly increase

microbial risks in an effort to meet DBP MCLs. The Agency wanted to

ensure adequate microbial protection while reducing risk from DBPs.

The last issue was the best available technology. The BAT defined

would determine the levels at which MCLs were set. For example,

allowing alternate disinfectants as BAT would drive the chlorination

byproduct MCLs down, but could result in increased exposure to (not

well characterized) alternate byproducts. EPA believed that it

therefore might be appropriate to define chlorine and a precursor

removal technology as BAT.

To address these issues, EPA suggested two possible regulatory

strategies. One was to define the MCL(s) based on what was possible to

achieve using the most effective DBP precursor removal strategy as BAT

(e.g., GAC or membrane filtration). While installing such precursor

removal technology might minimize health concerns, the costs would be

substantial (without finding out if other less costly technologies,

such as use of alternative disinfectants, provided similar benefits).

Also, since systems are not required to install BAT to meet MCLs, EPA

believed that many systems would attempt to meet the MCLs by lower-cost

alternative disinfectants (ozone, chloramines, chlorine dioxide). Since

health effects for alternative disinfectant byproducts are not

adequately characterized, risks may not be reduced.

The second strategy was a two-phase regulation, with the first

phase designed to address risks using lower cost options during

concurrent efforts to obtain more data on treatment alternatives and

health effects of compounds not currently adequately characterized.

This strategy would prevent major shifts into use of new treatment

technology until the full consequences of such shifts (both costs and

benefits) are better understood.

3. Suggested monitoring scenario. In its fact sheet accompanying

the status report, EPA recommended that routine TTHM and haloacetic

acid monitoring for systems serving at least 10,000 people have the

same monitoring requirements as were in the 1979 TTHM rule. Smaller

systems would have less frequent monitoring requirements, but would

have compliance based on worst-case samples. EPA included provisions

for reduced monitoring (compliance based on worst-case samples or

surrogate monitoring), waiver criteria, and requirements for

disinfectant and other DBP monitoring.

4. Summary of public comments. EPA received comments on the status

report from numerous parties. Many commentors agreed with EPA's

concerns with issues such as alternative disinfectant DBPs and

balancing microbial and DBP risks. Several commentors supported the

two-phase regulatory approach, but expressed concern about timing.

Others recommended that DBP MCLs not be set so low as to force many

systems to install expensive technology or decrease microbial

protection. Several commentors were concerned with the availability of

both analytical methods and certified laboratories for the low levels

that were being considered. One commentor recommended that EPA make it

clear that MCLs set for disinfectants should allow temporary high

levels to address distribution system microbiological problems.

Finally, many commentors supported allowing reduced monitoring wherever

possible.

C. Initiation of the Regulatory Negotiation Process

EPA became interested in pursuing a negotiated rulemaking process

for the development of the D/DBP rule, in large part, because no clear

path for addressing all the major issues identified in the June 1991

Status Report on D/DBP rule was apparent. EPA's most significant

concern was developing regulations for DBPs while also ensuring that

adequate treatment be maintained for controlling microbiological

concerns. A negotiated rule process would help people understand the

complexities of the risk-risk tradeoff issue and, hopefully, reach a

consensus on the most appropriate regulation to address concerns from

both DBPs and microorganisms.

It also appeared to EPA that the criteria for initiating a

negotiated rule under the Negotiated Rulemaking Act of 1990 for

establishing a negotiated rulemaking could be met. These include:

(1) there is a need for a rule,

(2) there are a limited number of identifiable interests that will

be significantly affected by the rule,

(3) there is a reasonable likelihood that a committee can be

convened with a balanced representation of persons who--

(A) can adequately represent the interests identified under

paragraph (2); and

(B) are willing to negotiate in good faith to reach a consensus on

the proposed rule,

(4) there is a reasonable likelihood that a committee will reach a

consensus on the proposed rule within a fixed period of time,

(5) the negotiated rulemaking procedure will not unreasonably delay

the notice of proposed rulemaking and the issuance of a final rule,

(6) the Agency has adequate resources and is willing to commit such

resources, including technical assistance, to the committee, and

(7) the Agency, to the maximum extent possible consistent with the

legal obligations of the Agency, will use the consensus of the

committee with respect to the proposed rule as the basis for the rule

proposed by the Agency for notice and comment.

In 1992 EPA hired a contractor, Resolve, which added a

subcontractor, Endispute, to assess the feasibility and usefulness of

convening a negotiated rulemaking. Resolve and Endispute conducted more

than forty interviews during the summer of 1992 with representatives of

State and local health and regulatory agencies, water suppliers,

manufacturers of equipment and supplies used in drinking water

treatment, and consumer and environmental organizations. These

interviews revealed that:

(1) The entities interested in or affected by the rulemaking were

readily identifiable and relatively few in number.

(2) The rulemaking required resolution of a limited number of

interdependent issues, about which there appeared to be a sufficiently

well-developed factual base to permit meaningful discussion. Further,

there appeared to be several ways to resolve these issues, providing a

potential basis for productive joint problem-solving.

(3) The parties expressed some common goals, along with an

unusually strong degree of good faith interest in resolving the issue

through negotiation.

(4) The Agency had adequate staff and technical resources and was

willing to commit such resources to the negotiated rulemaking.

Resolve and Endispute recommended to EPA that the negotiated

rulemaking proceed. EPA concurred with this recommendation.

However, it was also noted that reaching consensus on the proposed

rule would be a challenge. The interviews revealed that parties

differed in their perceptions about the nature and magnitude of the

risks associated with DBPs, and many expressed strong doubts about the

adequacy of available scientific and technical information. Moreover,

some parties stated that marginal improvements in disinfection

technology were all that should be done until the relative risks are

better understood, while others said that a fundamentally new approach

focusing on precursor reduction should be considered.

EPA published a notice of intent to proceed with a negotiated

rulemaking on September 15, 1992 (57 FRN 42533), proposing 17 parties

to be Negotiating Committee members. In general, comments indicated

very positive support for the negotiated rulemaking.

As part of the convening process, an organizational meeting was

held September 29-30, 1993. Participants discussed Negotiating

Committee composition and organizational protocols. Between comments

expressed at the meeting and submitted in writing, eleven additional

parties--including water suppliers not substantially represented by the

Committee's original proposed membership, and chemical and equipment

suppliers--asked to be added to the Committee. In addition,

participants discussed the need to develop accurate scientific and

technical information.

On November 13, 1992, EPA published a notice of establishment for

the Negotiating Committee (57 FRN 53866), and an 18th member was added

to the Negotiating Committee.

Based on comments received at the organizational meeting, a

Technical Workshop was organized and conducted on November 4-5, 1992.

Composed of presentations and panel discussion by 23 of the Nation's

leading experts on drinking water treatment, the workshop provided

participants with opportunities to familiarize themselves with the

technical elements in this rulemaking and to explore the range of

scientific opinions about: (1) The nature and magnitude of potential

health effects from exposure to DBPs and microbial contaminants in

drinking water, (2) available information on the cost and efficacy of

precursor removal and drinking water disinfection technologies, and (3)

EPA's efforts to model and compare chemical and microbial risks in

drinking water.

Additional presentations were given throughout the rulemaking

process, as new information became available and more questions were

raised by participants.

At the first formal negotiating session, on November 23-24, 1992,

participants formed a technologies working group (TWG) to develop

reliable and consistent information about the cost and efficacy of

drinking water treatment technologies. This approach provided a forum

for participants to arrive at a shared understanding of complex issues

in the rulemaking, setting a cooperative tone for the rest of their

discussions. The working group, which continued to meet throughout the

rulemaking, also provided a formal opportunity for input from the

chemical and equipment suppliers who had not been named to the

Committee.

In addition, three experts were hired through EPA's contract with

Resolve to provide ongoing scientific advice and technical support to

participants in the Committee and on the technologies working group,

principally for members without access to similar resources within

their own organizations.

Based on scientific data presented and discussed through the

November 23-24 meeting, participants agreed that some type of DBP Rule

was warranted.

The Committee developed and reached agreement on criteria for a

``good'' DBP Rule at the September 29-30 and November 23-24 meetings. A

good rule is one which would be flexible and affordable and would

protect public health from chemical and microbial risks. It was noted

that limiting some DBPs could encourage changes in treatment that might

increase the formation of other DBPs, or compromise protections against

microbial contaminants.

Next, Committee members and other participants were invited to

present regulatory options as a starting point for further discussion.

Sixteen options were introduced at the December 17-18 meeting, and

discussed at the meeting on January 13-14, 1993. These were merged into

three consolidated options at the January 13-14 meeting, and discussion

continued at the meeting on February 9-10. At this point, areas of

disagreement included:

(1) Whether to regulate DBPs through Maximum Contaminant Levels

(MCLs) or through a treatment technique (i.e., by exceeding DBP

``action levels,'' systems would trigger additional steps to minimize

chemical and microbial risks).

(2) Whether to minimize formation of the DBPs about which

relatively little is known by establishing a regulatory limit for their

naturally occurring organic precursors (e.g., Total Organic Carbon, or

TOC) in the water prior to the point of disinfection.

(3) Whether to provide greater protection against microbial

contaminants in drinking water, in conjunction with new DBP limits, by

developing an enhanced Surface Water Treatment Rule (ESWTR).

(4) Whether to develop a second round of DBP controls along with

the first (assuring broad improvements in drinking water quality), or

to wait until better scientific information becomes available.

Concurrently, the TWG modelled systems' potential compliance

choices under several regulatory scenarios, and presented revised

household and national compliance cost estimates at several meetings.

Using a ``strawman'' developed from the consolidated options by EPA

staff as the starting point for negotiation, the Committee worked out

an ``agreement in principle'' on the first round of DBP controls at its

February 24-25 meeting. The ``Stage 1'' agreement set MCLs for

trihalomethanes and haloacetic acids--two principal classes of

chlorination by-products--at levels the Committee deemed protective of

public health, based on current information: 80 and 60 micrograms per

liter, respectively. To limit DBP precursors, the Committee agreed to

develop a series of ``enhanced coagulation'' requirements, to vary

according to systems' influent water quality and treatment plant

configurations. Members also agreed to reconvene in several years to

develop a second stage of DBP regulations, when the results of more

health effects research and water quality monitoring are available. In

addition, members agreed that more expeditious changes to the rules may

be necessary if additional information becomes available on short- term

or acute health effects of DBPs. Members also agreed that, if data on

short-term or acute health effects warrant earlier action, a meeting

shall be convened to review the results and to develop recommendations.

A drafting group was named at the February 24-25 meeting. Assisted

by the TWG, these members drafted an ``agreement in principle'' for

presentation and discussion at the March 18-19 meeting. Using ``straw''

provisions from the facilitators, the Committee devised a regulatory

``backstop'' (i.e., Stage 2 MCLs of 0.040 mg/l for TTHMs and 0.030 mg/l

for HAA5 for surface water systems serving at least 10,000 people) at

this meeting to assure participants favoring further DBP controls that

other members would return for the ``Stage 2'' negotiation. The

Committee also agreed to recommend that EPA propose several ESWTR

options for comment, developed a collaborative process to guide the

health effects research program, and agreed to formulate short-term

water quality and technical data collection provisions within an

Information Collection Rule.

Based on the discussion to this point, one member withdrew from the

Committee at the March 18-19 meeting.

The drafting group presented regulatory language for the DBP Rule,

ESWTR, and ICR at each of the Committee's last two meetings, held May

12-13 and June 22-23, 1993. These texts provided a framework for

further discussion and resolution of remaining issues, including:

limits for residual disinfectants and individual by-products; public

notification and affordability provisions; and timing, applicability,

and conditions under which systems might qualify for exceptions from

various requirements. Committee members agreed to reserve their rights

to comment on the draft preambles.

The drafting group continued working through the summer of 1993,

and revisions to each of the rules and their preambles were mailed to

the Committee for comment on July 8, 1993, September 8, 1993, February

8, 1994, and May 12, 1994. Each member had signed the agreement by June

7, 1994.

Unless otherwise noted, EPA has adopted the recommendations of the

Negotiating Committee and its Technologies Working Group and reflects

those recommendations in the following preamble and proposed

regulations.

V. Establishing MCLGs

A. Background

1. MCLGs and MCLs Must Be Proposed and Promulgated Simultaneously

Congress revised the Safe Drinking Water Act in 1986 to require

that MCLGs and National Primary Drinking Water Regulations (NPDWRs) be

proposed simultaneously and promulgated simultaneously [SDWA section

1412 (a)(3)]. Simultaneous promulgation was intended to streamline the

development of drinking water regulations.

2. How MCLGs Are Developed

MCLGs are set at concentration levels at which no known or

anticipated adverse health effects occur, allowing for an adequate

margin of safety. Establishment of an MCLG for each specific

contaminant depends on the evidence of carcinogenicity from drinking

water exposure or an assessment for adverse noncarcinogenic health

effects.

a. MCLG Three Category Approach. EPA currently follows a three-

category approach in developing MCLGs for drinking water contaminants

(Table V-1).

Table V-1.--EPA'S Three-Category Approach for Establishing MCLGs

------------------------------------------------------------------------

Evidence of

Category carcinogenicity via MCLG approach

drinking water\1\

------------------------------------------------------------------------

I...................... Strong evidence Zero.

considering weight of

evidence,

pharmacokinetics,

potency and exposure.

II..................... Limited evidence RfD approach with

considering weight of added safety margin

evidence, of 1 to 10 or 10-5 to

pharmacokinetics, 10-6 cancer risk

potency and exposure. range.

III.................... Inadequate or no animal RfD approach.

evidence.

------------------------------------------------------------------------

\1\Considering oral exposure data such as drinking water, dietary and

gavage studies.

Each chemical is evaluated for evidence of carcinogenicity from

drinking water. For volatile contaminants, inhalation data are also

considered. EPA takes into consideration the overall weight of evidence

for carcinogenicity, pharmacokinetics, potency and exposure.

EPA's policy is to set MCLGs for Category I contaminants at zero.

The MCLG for Category II contaminants is calculated by using the

Reference dose (RfD) approach (described below) with an added margin of

safety to account for possible cancer effects. If adequate data are not

available to calculate an RfD, then the MCLG is based on a cancer risk

level of 10-5 to 10-6. MCLGs for Category III contaminants

are calculated using the RfD approach.

Category I contaminants are those for which EPA has determined that

there is strong evidence of carcinogenicity from drinking water. The

MCLG for Category I contaminants is set at zero because it is assumed,

in the absence of other data, that there is no threshold dose for

carcinogenicity. In the absence of route specific (e.g., oral) data on

the potential cancer risk from drinking water, chemicals classified as

Group A or B carcinogens (see section c below) are generally placed in

Category I.

Category II contaminants include those contaminants for which EPA

has determined that there is limited evidence of carcinogenicity from

drinking water, considering weight of evidence, pharmacokinetics,

potency, and exposure. In the absence of route specific data, chemicals

classified in Group C (see section c below) are generally placed in

Category II.

For Category II contaminants, one of two options have traditionally

been used to set the MCLG. The first option sets the MCLG based upon

noncarcinogenic endpoints of toxicity (the RfD), then applies an

additional safety factor of 1 to 10 to the MCLG to account for possible

carcinogenicity. An MCLG set by the option 1 approach is compared with

the cancer risk, if quantified. The second option is to set the MCLG

based upon a theoretical lifetime excess cancer risk level of 10-5

to 10-6 using a conservative mathematical extrapolation model. EPA

generally uses the first option; however, the second approach is used

when valid noncarcinogenic data are not available to calculate an RfD

and adequate experimental data are available to quantify the cancer

risk.

Category III contaminants include those contaminants for which

there is inadequate or no evidence of carcinogenicity from drinking

water. If there is no additional information to consider, contaminants

classified in Group D or E (see section c below) are generally placed

in Category III. For these contaminants, the MCLG is established using

the RfD approach.

b. Assessment of Noncancer Health Effects. The risk assessment for

noncancer health effects can be characterized by a Reference Dose

(RfD). The oral RfD (expressed in mg/kg/day) is an estimate, with

uncertainty spanning perhaps an order of magnitude, of a daily exposure

to the human population (including sensitive subgroups) that is likely

to be without an appreciable risk of deleterious health effects during

a lifetime. The RfD is derived from a no- or lowest-observed-adverse-

effect level (called a NOAEL or LOAEL, respectively) that has been

identified from a subchronic or chronic study of humans or animals. The

NOAEL or LOAEL is then divided by an uncertainty factor(s) to derive

the RfD. Although the RfD is represented as a point estimate, it is

actually a range since the RfD is a number with an inherent uncertainty

of an order of magnitude.

Uncertainty factors are used to estimate the comparable ``no-

effect'' level for a large heterogeneous human population. The use of

uncertainty factors accounts for several data gaps including intra- and

inter-species differences in response to toxicity, the small number of

animals tested compared to the size of the population, sensitive

subpopulations and the possibility of synergistic action between

chemicals (see 52 FR 25690 for further discussion on the use of

uncertainty factors).

EPA has established certain guidelines (shown below) to determine

how to apply uncertainty factors when establishing an RfD (USEPA,

1986).

Use a 1- to 10-fold factor when extrapolating from valid

experimental results from studies in average healthy humans. This

factor is intended to account for the variation in sensitivity among

the members of the human population.

Use an additional 10-fold factor when extrapolating from

valid results of long-term studies on experimental animals when results

of studies of human exposure are not available or are inadequate. This

factor is intended to account for the uncertainty in extrapolating

animal data to the case of humans.

Use an additional 10-fold factor when extrapolating from

less than chronic results on experimental animals when there are no

useful long-term human data. This factor is intended to account for the

uncertainty in extrapolating from less than chronic NOAELs to chronic

NOAELs.

Use an additional 10-fold factor when deriving an RfD from

a LOAEL instead of a NOAEL. This factor is intended to account for the

uncertainty in extrapolating from LOAELs to NOAELs.

An additional uncertainty factor may be used according to

scientific judgment when justified.

Use professional judgment to determine another uncertainty

factor (also called a modifying factor, MF) that is greater than zero

and less than or equal to 10. The magnitude of the MF depends upon the

professional assessment of scientific uncertainties of the study and

data base not explicitly treated above, e.g., the completeness of the

overall data base and the number of species tested. The default value

for the MF is 1.

To determine the MCLG, the RfD is adjusted by the body weight of

the protected (or most sensitive) individual (usually a 70 kg adult),

average volume of water consumed daily over a lifetime (2 L/day for an

adult) and exposure to the contaminant from a drinking water source

(relative source contribution or RSC).

Generally, EPA assumes that the RSC from drinking water is 20

percent of the total exposure, unless other exposure data for the

chemical are available [see 54 FR 22069 and 56 FR 3535]. When adequate

data are available and the data indicate that drinking water exposure

contributes between 20 and 80 percent of total exposure, EPA uses the

actual percentage to determine the MCLG, as is indicated by equation

(3), below. When data indicate that contributions from drinking water

are between zero and 20 percent, or between 80 and 100 percent, EPA

utilizes a 20 percent floor and an 80 percent ceiling, respectively.

The calculations below express the derivation of the MCLG based on

noncancer health effects:

TP29JY94.000

c. Assessment of Carcinogenic Health Effects. For chemicals

suspected of being carcinogenic to humans, the assessment for non-

threshold toxicants consists of the weight of evidence of

carcinogenicity in humans, using bioassays in animals and human

epidemiological studies as well as information that provides indirect

evidence (i.e., mutagenicity and other short-term test results). The

objectives of the assessment are to determine the level or strength of

evidence that the substance is a carcinogen and to provide an

upperbound estimate of the possible risk of human exposure to the

substance in drinking water. A summary of EPA's general carcinogen

classification scheme is (USEPA, 1986):

Group A--Human carcinogen based on sufficient evidence from

epidemiological or other human studies.

Group B--Probable human carcinogen based on limited evidence of

carcinogenicity in humans (Group B1) or based on sufficient evidence in

animals with inadequate or no data in humans (Group B2).

Group C--Possible human carcinogen based on limited evidence of

carcinogenicity in animals in the absence of human data.

Group D--Not classifiable based on lack of data or inadequate

evidence of carcinogenicity from animal data.

Group E--No evidence of carcinogenicity for humans (no evidence for

carcinogenicity in at least two adequate animal tests in different

species or in both epidemiological and animal studies).

d. MRDLGs--appropriateness of a new concept? As stated in section

II.A of this preamble, EPA is proposing a new term, ``maximum residual

disinfectant level goal'' (MRDLG), in lieu of MCLGs for all

disinfectants because disinfectants are intentionally added to drinking

water as a treatment technique to kill disease-causing microorganisms.

The proposal of this concept was agreed to through the negotiated

rulemaking process.

Certain members of the Negotiating Committee were concerned that if

``MCLGs,'' which included the term ``contaminant,'' were set for

disinfectants, water treatment plant operators might be reluctant to

apply disinfectant dosages above the MCLG during short periods of time

to control for microbial risk, even though such exposure to elevated

disinfectant concentration levels would pose little or no risk. For

example, NOAELs for chlorine and chloramines are based upon animal

studies following long term exposure to high levels of the

disinfectants in drinking water. Short-term exposures at elevated

levels would not be a concern (see the following discussion on health

effects for chlorine and chloramines). During emergency situations such

as distribution system pipe breaks or significant fluctuations in

source water quality, systems will on occasion need to apply short term

disinfectant residual concentrations of chlorine or chloramines, well

above the regulatory goal, to protect from waterborne disease.

The MRDLGs are developed in the same way as MCLGs. EPA solicits

comment on the appropriateness of adopting the term ``MRDLG'' in lieu

of MCLGs for disinfectants in the final rule.

B. Proposed MRDLGs and MCLGs

The following includes a summary of the health effects information

available for each disinfectant or by-product. These summaries are

taken from more complete and comprehensive descriptions of the data

given in the cited Health Criteria Documents that have been developed

for each of these chemicals. These documents are available in the water

docket.

1. Chlorine, hypochlorite ion and hypochlorous acid

The following assessment for both chlorine and chloramines includes

a consideration of available animal data, as well as epidemiology

studies which have been conducted on chlorinated or chloraminated

drinking water. The epidemiology data are discussed in section C of

this preamble.

Chlorine (CAS # 7782-50-5) hydrolyses in water to form hypochlorite

(CAS # as sodium salt 7790-92-3) and hypochlorous acid (CAS # 7681-52-

9). Because of their oxidizing characteristic and solubility, chlorine

and hypochlorites are used in water treatment to disinfect drinking

water, sewage and wastewater, swimming pools, and other types of water

reservoirs. They are also used for general sanitation and control of

bacterial odors in the food industry.

Chlorine is a highly reactive and water soluble species. The fate,

transport, and distribution of chlorine in natural waters is not well

understood. Much of the available information comes from the addition

and oxidation reactions with inorganic and organic compounds known to

occur in aqueous solutions. Factors such as reactant concentrations,

pH, temperature, salinity and sunlight influence these reactions.

Occurrence and Human Exposure. For the purpose of setting an MRDLG,

consideration is given to chlorine levels resulting from disinfection

of drinking water. Chlorine exposures from swimming pools and hot tubs

are not evaluated in determining the MRDLG. Persons who swim frequently

or use a hot tub may have greater dermal and possibly inhalation

exposure to chlorine.

Chlorine is added to drinking water as chlorine gas (Cl2) or

as calcium or sodium hypochlorite. In drinking water, the chlorine gas

hydrolyses to hypochlorous acid and hypochlorite ion and can be

measured as the free chlorine residual. Maintenance of a chlorine

residual throughout the distribution system is important for minimizing

bacterial growth and for indicating (by the absence of a residual)

water quality problems in the distribution system. Currently, maximum

chlorine dosage is limited by taste and odor constraints and for

systems needing to comply with the total trihalomethane (TTHM) standard

regularly. Additionally, for systems using chlorination, the surface

water treatment rule (SWTR) requires a minimum residual of 0.2 mg/L

prior to the entry point to the distribution system, and the presence

of a detectable residual throughout the distribution system.

Table V-2 presents occurrence information available for chlorine in

drinking water. Descriptions of these surveys and other data are

detailed in ``Occurrence Assessment for Disinfectants and Disinfection

By-Products (Phase 6a) in Public Drinking Water,'' USEPA 1992a. The

table lists five surveys conducted by Federal, as well as private

agencies. Median concentrations of chlorine in drinking water appear to

range from 50,000 people).

AWWA Disinfection Survey 283 Utilities in the U.S.. Finished Water Entering ........... 0.07-5.0 1.1 ........................

(1991). Distribution System.

--------------------------------------------------------------------------------------------------------------------------------------------------------

\1\Dates indicate period of sample collection.

\2\May not be representative of national occurrence.

\3\Typical dosage used by treatment plants.

SW: Surface Water.

GW: Ground Water.

AMWA: Association of Metropolitan Water Agencies.

AWWA: American Water Works Association.

AWWARF: American Water Works Association Research Foundation.

CDHS: California Department of Health Services.

EPA: Environmental Protection Agency.

WIDB: Water Industry Data Base.

Exposure to chlorine residual varies both between systems and

within systems. Chlorine residual within systems will vary based on

where customers are located within the distribution system and changes

in the system's disinfection needs over time. Using residual

concentrations from the 1989-1991 AWWA Disinfection Survey and WIDB,

exposure to chlorine due to drinking water can be estimated using a

consumption rate of 2 liters per day. Based on the estimated 25th

percentile and 75th percentile chlorine residuals in the 1991 AWWA

Disinfection Survey, exposure was determined to range from 1.5 to 3.8

mg/day and the median would be 2.2 mg/day. Using the WIDB data,

exposures to the average customer from surface and ground water sources

using chlorination, respectively, were determined to be 1.9 mg/day and

1.7 mg/day.

Little information is available concerning the occurrence of

chlorine in food and indoor air in the United States. The Food and Drug

Administration (FDA) does not analyze for chlorine in foods. However,

there are several uses of chlorine in food production, such as the

disinfection of chicken in poultry plants and the superchlorination of

water at soda and beer bottling plants (Borum, 1991). Therefore, the

possibility exists for dietary exposure to chlorine from its use in

food production. However, monitoring data are not available to

characterize adequately the extent of such potential exposures.

Additionally, preliminary discussions with FDA suggest that there are

no approved uses for chlorine in most foods consumed in the typical

diet. Similarly, EPA's Office of Air and Radiation is not currently

conducting any sampling studies for chlorine in indoor air. Data on

levels of chlorine in ambient air are forthcoming.

Considering the limited number of food groups that are believed to

contain chlorine and that no significant levels of chlorine are

expected in ambient or indoor air, it is anticipated that drinking

water is the predominant source of exposure to chlorine. Air and food

are believed to provide only small contributions, although the

magnitude and frequency of these potential exposures are issues

currently under review. EPA, therefore, is considering setting an MRDLG

for chlorine in drinking water using an RSC value of 80%, the current

exposure assessment policy ceiling. EPA requests any additional data on

known concentrations of chlorine in drinking water, food and air.

Health Effects. The health effects information for chlorine is

summarized from the draft Drinking Water Health Criteria Document for

Chlorine, Hypochlorous Acid and Hyperchlorite Ion (USEPA, 1994a). The

studies cited within this section are summarized in the draft criteria

document.

Chlorine and the hypochlorites are very reactive and thus can react

with the constituents of saliva and possibly food and gastric fluid to

yield a variety of reaction by-products (e.g., trihalomethanes). Thus,

the health effects associated with the administration of high levels of

chlorine and/or the hypochlorites in various animal studies may be due

to these reaction by-products and not the disinfectant itself.

Oxidizing species such as chlorine and the hypochlorites are probably

short-lived in biological systems due to both their reactivity and the

large number of organic compounds found in vivo. Scully and White

(1991) noted that reactions of aqueous chlorine with sulfur-containing

amino acids appear to be so fast in saliva that all free available

chlorine is dissipated before the water is swallowed.

Oral studies with radiolabeled (i.e., 36Cl) hypochlorite and

hypochlorous acid indicate that, as measured by the radiolabel, these

compounds may be well absorbed and distributed throughout the body with

the highest levels measured in plasma and bone marrow. However,

considering the reactivity of the hypochlorites, these results may only

reflect the presence of reaction by-products (e.g., chloride). The

major route of excretion appears to be urine and then the feces.

Acute oral LD50 values for calcium and sodium hypochlorite

have been reported at 850 mg/kg in rats and 880 mg/kg in mice,

respectively. Humans have consumed hyperchlorinated water for short

periods of time at levels as high as 50 mg/L (1.4 mg/kg) with no

apparent adverse effects.

Short-term oral studies in animals have indicated decreases in

blood-glutathione levels, hemolysis and biochemical changes in liver in

rodents following a gavage dose of hypochlorite in water. No adverse

effects on reproduction (Druckery, 1968) or development were observed

in rats administered chlorine in drinking water at concentrations of

100 mg/L or less. However, Meier et al. (1985) observed an increase in

sperm-head abnormalities in mice receiving hypochlorite at 200 mg/L,

but not at 100 mg/L or less.

No systemic effects were observed in rodents following oral

exposure to chlorine as hypochlorite in distilled water at levels up to

275 mg/L over a 2 year period (NTP, 1990).

Chlorinated water has been shown to be mutagenic to bacterial

strains and mammalian cells. Investigations with rodents to determine

the potential carcinogenicity of chlorine, or chlorinated water have

been negative. In the most recent study, no apparent carcinogenic

potential was demonstrated following oral exposure to chlorine in

distilled drinking water as hypochlorite, at levels up to 275 mg/L over

a 2 year period (NTP, 1990). However, NTP observed a marginal increase

in the incidence of mononuclear cell leukemia in mid-dose female F344

rats but not in male rats or male and female mice (NTP, 1990).

Mononuclear cell leukemia has a high spontaneous rate of occurrence in

female F344 rats. The levels reported in the NTP study are within the

historical control range of incidence for the sex and strain of rat.

EPA believes that mononuclear cell leukemia can not be solely

attributed to exposures to chlorine in drinking water but rather may

reflect the high background rate of mononuclear cell leukemia in the

test species.

EPA has classified chlorine in Group D, not classifiable as to

human carcinogenicity (IRIS, 1993). This classification stems from the

findings of the NTP (1990) study indicating equivocal evidence in

female rats (increased mononuclear cell leukemia) and no evidence in

male rats or male and female mice. The International Agency for

Research on Cancer (IARC, 1991) also evaluated chlorinated drinking

water and hypochlorite for potential human carcinogenicity. IARC

determined that there was inadequate evidence for carcinogenicity of

chlorinated drinking water and hypochlorite salts in humans and

animals. (See section C for a description of these studies.) IARC

concluded that chlorinated drinking water and hypochlorite salts were

not classifiable as to their carcinogenicity to humans and thus

assigned these chemicals to IARC Group 3. This category is similar to

EPA cancer classification Group D.

Based on the previous discussion, EPA is proposing that chlorine,

hypochlorite and hypochlorous acid be placed in Category III for the

purpose of setting an MRDLG. The study selected for determining an RfD

is the previously mentioned 2 year rodent study that was conducted by

the National Toxicology Program (NTP, 1990). In this study, male and

female F344 rats and B6C3F1 mice were given chlorine in distilled

drinking water at levels of 0, 70, 140 and 275 mg/L for 2 years. Based

on body weight and water consumption values, these concentrations

correspond to doses of approximately 0, 4, 7 and 14 mg/kg/day for male

rats; 0, 4, 8, and 14 mg/kg/day for female rats; 0, 7, 14, and 24 mg/

kg/day for male mice and 0, 8, 14 and 24 mg/kg/day for female mice.

There was a dose related decrease in water consumption for both rats

and mice, presumably due to taste aversion. No effect on body weight or

survival were observed for any of the treated animals. Using a NOAEL of

14 mg/kg/d identified from female rats in the NTP (1990) study an MRDLG

of 4 mg/L, based on lack of toxicity in a chronic study is derived as

follows.

TP29JY94.001

Where 14 mg/kg/d is the NOAEL for female rats in the NTP study, and

100 is the uncertainty factor applied to account for inter and intra-

species differences in accordance with EPA guidelines when a NOAEL from

a chronic animal study is the basis for the RfD. The MRDLG is based on

a 70 kg adult consuming an average of 2 liters water per day over their

lifetime. In addition, an 80% RSC is assumed in the absence of data to

the contrary.

Public comments are requested on the following issues: 1) placing

chlorine in Category III for developing an MRDLG, 2) selection of the

study and NOAEL as the basis for the MRDLG, 3) the 80% RSC, 4) the

appropriateness of the UF of 100, 6) the cancer classification for

chlorine.

2. Chloramines

Inorganic chloramines (CAS Nos. 10599-90-3 and 10025-85-1 for mono-

and trichloramine, respectively) are formed in waters undergoing

chlorination which contain ammonia. Monochloramines, dichloramines and

trichloramines may be formed. Monochloramine is the principal

chloramine formed in chlorinated natural and wastewater at a neutral pH

and is much more persistent in the environment.

Chloramine is used as a disinfectant in drinking water to control

taste and odor problems, limit the formation of chlorinated

disinfection by-products, and maintain a residual in the distribution

system for controlling biofilm growth. At typical pHs of most drinking

waters, the predominant chloramine specie is monochloramine. For

purposes of this regulation, only monochloramine will be considered

since the other chloramines occur at much lower concentrations in

almost all drinking waters. Monochloramine has also been much more

extensively studied.

Monochloramine, the principal chloramine formed in chlorinated

natural and wastewaters at neutral pH, is relatively stable when

discharged to the environment. First-order decay rate constants of 0.03

to 0.075 hr-1 for monochloramine in the laboratory, and higher

rate constants of 0.28 to 0.31 hr-1 outdoors using chlorinated

effluents, have been reported. If discharged into receiving waters

containing bromide, monochloramine will decompose faster, probably

through the formation of NHBrCl and decomposition of the dihalamine.

The rate of monochloramine disappearance is primarily a function of pH

and salinity. For example, at pH 7 and 25 deg.C, the half-life of

monochloramine is 6 hr at 5 parts per thousand (ppt) salinity and 0.75

hr at 35 ppt salinity; at pH 8.5 and 25 deg.C, the half-life is 188 hr

at 5 ppt salinity and 25 hr at 35 ppt salinity. Monochloramine is

expected to decompose in wastewater discharges receiving waters via

chlorine transfer to organic nitrogen-containing compounds.

Occurrence and Human Exposure. Chloramine occurs in drinking water

both as a by-product and intentionally for disinfection. Chloramine is

formed during chlorination when source waters contain ammonia. It is

also used as a primary or secondary disinfectant, usually with

chloramine being generated on site by the addition of ammonia to water

following treatment by chlorination. The use of chloramines has been

shown to reduce the formation of certain by-products, notably

trihalomethanes, relative to the by-products formed with chlorination

alone. Chlorination by-product formation can be minimized when the

ammonia is added prior to or in combination with chlorine by reducing

the chlorine residual of the water being treated. In most plants,

however, ammonia is added some time after the addition of chlorine, to

allow for more effective disinfection since chlorine is a much stronger

disinfectant than chloramines.

Table V-3 presents occurrence information available for chloramine

in drinking water. Descriptions of these surveys and other data are

detailed in ``Occurrence Assessment for Disinfectants and Disinfection

By-Products (Phase 6a) in Public Drinking Water,'' USEPA, 1992a.

Typical dosages of chloramine used as a disinfectant in drinking water

treatment facilities range from 1.5 to 2.7 mg/L. Median concentrations

of chloramine in drinking water were found to range from 1.1 to 1.8 mg/

L.

Table V-3.--Summary of Occurrence Data for Chloramines

--------------------------------------------------------------------------------------------------------------------------------------------------------

Occurrence of chloramine in drinking water

---------------------------------------------------------------------------------------------------------------------------------------------------------

Concentration (mg/L)

Survey (year)\1\ Location Sample information (No. of ----------------------------------------------------------------

samples) Range Mean Median Other

--------------------------------------------------------------------------------------------------------------------------------------------------------

AWWARF (1987) McGuire & National Survey........... Finished Water From: ........... ........... ........... Typical dosages:

Meadow, 1988. Lakes.......................... 1.5 mg/L

Flowing Streams................ 2.7 mg/L

EPA/AMWA/CDHS\2\ (1988- 35 Water Utilities Samples from Clearwell 0.9-5.5 2.3 1.8 ........................

1989) Krasner et al., Nationwide. Effluent, 4 Quarters (13).

1989b.

EPA, 1992b\2\ (1987-1991). Disinfection By-Products At the Plant (11).............. 1.2-3.6 2.1 1.5 ........................

Field Studies. Distribution System (8)........ 0.1-3.3 1.4 1.1

--------------------------------------------------------------------------------------------------------------------------------------------------------

\1\Dates indicate period of sample collection.

\2\May not be representative of national occurrence.

AMWA: Association of Metropolitan Water Agencies.

AWWARF: American Water Works Association Research Foundation.

CDHS: California Department of Health Services.

EPA: Environmental Protection Agency.

Based on the residual concentrations given above, a high and low

estimate for exposure to chloramine from drinking water can be

calculated using an assumed consumption of 2 liters per day. Using the

target range of 1.5 to 2 mg/L, the exposure may range from 3 to 4 mg/

day. Some systems may deviate significantly from this range.

No information is available on the occurrence of chloramine in food

or air. Currently, the Food and Drug Administration (FDA) does not

measure for chloramine in foods since the analytical methods have not

been developed. Preliminary discussions with FDA suggest that there are

not approved uses for chloramine in foods consumed in the typical diet.

Similarly, EPA's Office of Air and Radiation is not sampling

chloramines in air (Borum, 1991).

Based on the previous discussion, EPA assumes that drinking water

is the predominant source of exposure to chloramine. Air and food

intakes are believed to provide only small contributions, although the

magnitude and frequency of these potential exposures are issues

currently under review. EPA, therefore, is proposing to establish an

MRDLG for chloramine in drinking water with an RSC value of 80%, the

current exposure assessment policy ceiling. EPA requests any additional

data on known concentrations of chloramine in drinking water, food and

air.

Health Effects. The health effects information in this section is

summarized from the draft Drinking Water Health Criteria Document for

Chloramines (USEPA, 1994b). Studies mentioned in this section are

summarized in the Criteria Document.

Short-term inhalation exposures to high levels (500 ml of 5%

household ammonia mixed with 5% hypochlorite bleach) of chloramines in

humans result in burning in the eyes and throat, dyspnea, coughing,

nausea and vomiting. Inhalation of the chloramine fumes resulted in

pneumonitis but did not result in permanent pulmonary damage.

Short-term exposures to chloramines in drinking water, in which

human subjects were administered single concentrations ranging between

1 and 24 mg/L (1, 8, 18 or 24 mg/L), have not resulted in any adverse

effects in human subjects. Following human exposure, the subject's

physical condition, urinalysis, hematology, and clinical chemistry were

evaluated. No adverse clinical effects were noted in any of the

studies.

In another study, acute hemolytic anemia, characterized by

oxidation of hemoglobin to methemoglobin and denaturation of

hemoglobin, was reported in hemodialysis patients when tap water

disinfected with chloramines was used for dialysis baths. Chloramines

were reported to produce oxidant damage to red blood cells and inhibit

the metabolic pathway used by red blood cells to prevent and repair

such damage. Many dialysis centers have installed reverse osmosis units

coupled with charcoal filtration or the addition of ascorbic acid to

prevent hemolytic anemia.

Animal studies indicate varying sensitivity and conflicting results

among different animal species. Toxic effects noted among rats are

changes in blood glutathione and methemoglobin. Both monkeys and mice

were unaffected during short-term assays with doses up to 200 mg/L

chloramines. Based on studies up to 6 weeks in length, rats appear to

be more sensitive to monochloramine than mice and monkeys.

Toxicokinetic studies of chloramines indicate that the absorption

of chloramines is rapid, peaking within 8 hours of administration. In

the rat, chloramines are metabolized to chloride ion and excreted

mostly through the urine with a small portion excreted through the

feces.

Longer-term oral studies (90 days or longer) showed decreased body

and organ weights in rodents. Some effects to the liver (weight

changes, hypertrophy, and chromatid pattern changes) appear to be

related to overall body weight changes caused by decreased water

consumption due to the unpalatibility of chloramines to the test

animals.

In addition, chloramine may induce immunotoxicity in rats in the

form of increased prostaglandin E2 synthesis, reduced antibody

synthesis, and spleen weight at levels as low as 9 to 19 mg/L

chloramines for 90 days. The significance of these findings for risk

use in risk assessment is compromised by the design flaws of the study

(i.e., animals were exposed to two antigens) and the lack of

corroboration of these findings by a follow-up study.

Two lifetime rodent studies involving oral exposures to rats and

mice via drinking water have been considered by EPA for the derivation

of the MRDLG for monochloramine. Both studies were performed by the

National Toxicology Program (NTP, 1990) and involved 70 animals/sex/

dose exposed to distilled drinking water containing 0, 50, 100 or 200

ppm chloramines.

The first NTP (1990) study was a 2-year study in mice to determine

the potential chronic toxicity or carcinogenic activity of

chloraminated drinking water. B6C3F1 mice were administered

chloramine at doses of 0, 50, 100 and 200 ppm in distilled drinking

water. These doses were calculated based on a time-weighted average to

be 0, 5.0, 8.9 and 15.9 mg/kg/day for male mice and 0, 4.9, 9.0 and

17.2 mg/kg/day for female mice. There was a dose-related decrease in

the amount of water consumed by both sexes; this decrease was noted

during the first week and continued throughout the study. Dosed male

and female mice had similar food consumption as controls except for

females in the 200 ppm dose group that exhibited slightly lower

consumption than controls.

Study results indicated that there was a dose-related decrease in

mean body weights of dosed male and female mice throughout the study.

Mean body weights of high-dose male mice were 10-22% lower than their

control group after week 37 and the body weights of high-dose female

mice were 10-35% lower after week 8. However, the survival of mice

receiving monochloramine in drinking water was not significantly

different than controls. Clinical findings observed were not attributed

to the consumption of chloraminated drinking water. Body weight loss

and systemic toxicity were not considered related to the toxicity of

chloramine, but rather due to decreased water consumption resulting

from the unpalatability of chloramines in drinking water to the test

animals. Therefore, the highest dose tested, 17.2 mg/kg/day, is

considered a NOAEL in mice.

In the second study F344/N rats were administered monochloramine

for 2 years at doses of 0, 50, 100 and 200 ppm in distilled drinking

water. These doses were calculated on the basis of a time-weighted

average to be 0, 2.1, 4.8 and 8.7 mg/kg/day for male rats and 0, 2.8,

5.3 and 9.5 mg/kg/day for female rats. There was a dose-related

decrease in the amount of water consumed by both sexes; this decrease

was noted during the first week and continued throughout the study.

Food consumption of treated rats was the same as the controls with

males consuming more. In addition, mean body weights of 200 ppm dosed

rats (both sexes) were lower than their control groups. However, mean

body weights of rats receiving monochloramine in drinking water (at all

levels) were within 10% of controls until week 97 for females and week

101 for males. Though several clinical changes were noted, no clinical

changes were attributable to chloraminated drinking water. The survival

of rats receiving chloraminated drinking water was not significantly

different than controls except that, for the 50 ppm dose groups,

survival was greater than that of controls. Therefore, EPA considers

the highest dose tested, 9.5 mg/kg/day, as the NOAEL.

Based on two bacterial assays, monochloramine appears to be weakly

mutagenic. One study examining the reproductive effects and another

which examined developmental effects of chloramines concluded that

there are no chemical-related effects due to chloramines.

The NTP evaluation, using the results of the two lifetime NTP

bioassays, concluded that chloramines exhibited equivocal evidence of

carcinogenic activity of chloraminated drinking water in female F344/N

rats. This conclusion results from an increase in mononuclear cell

leukemia. There was no evidence of carcinogenic activity in male rats

or mice of either sex. The findings do not establish a link between

chloramine exposure and carcinogenicity because of the high historical

background occurrence of this type of cancer in test animals. The

incidence of mononuclear cell leukemia in the female control groups

(16%) was substantially less than the incidence reported in untreated

historical controls (25%). Incidence of mononuclear cell leukemia in

test animals reached a high of 32% in the high dose female rats. This

study also discovered incidence of renal tubular cell neoplasms in two

high-dose male mice receiving chloraminated water. Since this type of

tumor is rarely seen in historical controls, there is some concern that

these may be treatment related. However, the overall evidence regarding

the potential carcinogenicity of chloramines in drinking water can be

described as inconclusive since no long-term study has linked any tumor

development to actual chloramination exposure. On this basis, as well

as consideration of those studies described in section C, EPA placed

chloramine in Group D: not classifiable based on inadequate evidence of

carcinogenicity.

EPA selected the lifetime study in rats (NTP, 1990) as the basis

for calculating the MRDLG for chloramines. The NOAEL for the rat (9.5

mg/kg/d) is proposed because the rat was not tested at the higher doses

where mice were tested (17.2 mg/kg/d). Rats appear to be more sensitive

considering observed changes in biochemistry. Following a Category III

approach and using the rat NOAEL of 9.5 mg/L from the NTP study, an

MRDLG of 4 mg/L (measured as total chlorine), based on lack of toxic

effects in a chronic study can be derived for the 70-kg adult consuming

2 liters of water per day applying an uncertainty factor of 100, which

is appropriate for use of a NOAEL derived from an animal study and

assuming an RSC drinking water contribution of 80 percent.

---------------------------------------------------------------------------

*Since chloramines, on a practical basis, will be measured as

total chlorine, it is necessary to present the MRDLG in terms of a

chlorine equivalent concentration. Three mg/L chloramine is

equivalent to 4 mg Cl2/liter, based on the molecular weights of

Cl2 and NH2Cl.

TP29JY94.002

EPA requests comments on the proposed MRDLG for chloramines and the

RSC of 80%, the significance of the findings of immunotoxicity for

setting the RfD instead of the NTP study, the significance of the

finding of mononuclear cell leukemia in female F344 rats, the

significance of the finding of tubular cell neoplasms in high-dose

exposed mice, and whether the adjusted MRDLG, which takes into account

the measurement of monochloramine as total chlorine, is appropriate.

TP29JY94.003

3. Epidemiology Studies of Chlorinated and Chloraminated Water

Several studies have been conducted to evaluate the association of

chlorination or chloramination with the risk of cancer, cardiovascular

disease or adverse reproductive effects in humans. A summary of some of

these studies is given below. This discussion reflects EPA's assessment

of these data and is summarized from the draft Drinking Water Criteria

Documents for Chlorine (USEPA, 1994a) and Chloramines (USEPA, 1994b),

respectively.

Introduction to Epidemiology Studies. Two distinct types of

epidemiology studies have been conducted: ecologic and analytical.

These types of studies differ markedly in what they reveal about the

association between water quality and disease. In an ecological

epidemiology study, information is available on exposure and disease

for groups of people rather than for individuals, and therefore, the

results are difficult to interpret. What is considered to be an

important or relevant group variable may not be important for or may

not pertain to individuals within that group. Theoretical and empirical

analyses have offered no consistent guidelines for the interpretation

of ecological associations, and results from these studies are

appropriate only to suggest hypotheses for further study by analytical

epidemiological methods (Piantiadosi et al., 1988; Connor and Gillings,

1974).

Analytical epidemiology studies provide an estimate of the

magnitude of risk and information which can be used to evaluate

causality. For each individual in the study, information is obtained

about disease status and exposure to various contaminants and other

characteristics. In several of the studies reported here, individual

exposures to disinfected water or specific disinfection by-products

were estimated using group exposure information. All reported

epidemiological associations from analytical studies require an

evaluation of random error (statistical significance) and potential

sources of systematic bias (misclassification, selection, observation,

and confounding biases) so that results can be interpreted properly. It

must be noted that random error or chance can never be completely ruled

out as the explanation for an observed association and that statistical

significance does not necessarily imply biological significance.

Regardless of statistical significance, it is important to consider

potential biological mechanisms. Random error does not address the

possibility of systematic error or bias. Misclassification of exposure

and disease, selection bias, and observation bias must be avoided;

confounding bias, on the other hand, can be prevented both in study

design and during analysis if information is obtained about possible

confounders. It is important to determine for each specific

epidemiology study the validity of the association observed between

exposure and disease before considering possible causality between

exposure and disease or inferring that the results apply to a larger or

target population. Systematic bias can lead to spurious associations;

in some but not all instances, the direction of the bias can be

determined. For example, a random misclassification of exposure usually

biases a study toward not observing an effect or observing a smaller

risk than may actually be present, but nonrandom misclassification can

result in either higher or lower estimates of risk depending upon the

distribution of misclassification.

In addition, because of the observational nature of epidemiology,

the interpretation of epidemiology studies requires a sufficient number

of well designed and well conducted epidemiology studies, and must

include appropriate toxicological and biological information. Judging

causality from epidemiology studies is based largely on guidelines

developed over the years, including sequence of events, strength of

association (relative risk or odds ratio), consistency of results under

different conditions of study, biological plausibility, dose or

exposure response relationship, and specificity of effect. The relative

risk represents a basic measure of an association between exposure and

disease. It is defined as the rate of disease in the exposed population

divided by the rate of disease in an unexposed population.

a. Cancer Studies. Since the early 1970's, numerous epidemiologic

studies have attempted to assess the association between cancer and the

long term consumption of water from various sources with and without

disinfection and of various chemical quality, especially chlorinated

surface waters which supply the majority of the U.S. population.

Ecological, case control, and cohort studies have been conducted. Case

control studies have included incident and decedent cases; in some

studies information about various risk factors has been collected

through interviews, but in others information was obtained primarily

from death certificates.

i. Ecological Studies. The earliest studies were analyses of group

or aggregate data available on drinking water exposures and cancer.

Usually the variables selected for analyses were readily available in

published census, vital statistics, or public records and easily

abstracted and assembled. These analyses, referred to as ecological but

also called aggregate, geographical, or correlational, were designed to

investigate cancer mortality rates, usually on a county or State level.

Areas of different water quality, source, and chlorination status were

compared to identify possible statistical associations for further

study. Drinking water exposures were most often characterized as simple

dichotomous variables which served as indicators of exposure to

differing source water quality, e.g., the drinking water source for the

county or geographic area was categorized as a surface or groundwater

source. In some instances exposure variables included estimates of the

proportion of the area's or county's population that received surface

or groundwater and whether it was chlorinated. Surface water was

assumed to be more contaminated with synthetic organics than

groundwater, but no attempt was made to estimate levels of

contaminants.

In 1974 it was discovered that when surface waters were disinfected

with chlorine, the chlorine reacted with pre-existing organic materials

in the water to create a great number of chemical by-products (Craun,

1988). The major group of disinfection by-products (by weight) was the

trihalomethanes (THMs) which included an animal carcinogen, chloroform.

Chlorinated surface water was evaluated as an exposure variable in

several of the ecological studies, and since almost all surface waters

are chlorinated, the analyses usually compared cancer mortality among

populations receiving chlorinated surface water with those receiving

unchlorinated groundwater. Chlorinated water was assumed to contain

disinfection by-products, and at higher levels than chlorinated

groundwaters. However, the quality of surface and groundwater may also

differ for other contaminants, and this was not considered. Any

observed association might be due to other water quality differences

among surface and groundwaters (e.g., organic contaminants from

nonpoint and point source discharges to surface waters from industrial,

urban, and agricultural sources before pollution control regulations).

In some ecological analyses, the investigators attempted to study

the association between cancer mortality and an estimate of group

exposure to levels of chlorinated by-products based on THM or

chloroform levels was determined from a limited number of water

samples. The exposure information used in ecological studies was

available in only broad geographic units such as census tracts or

counties (Crump and Guess, 1982; Shy, 1985). Although these exposure

variables were statistically associated with mortality rates for all

cancers combined and several site-specific cancers, the interpretation

must necessarily be cautious due to limitations of ecological studies.

In several of these studies, aggregate or group information on several

covariates, e.g., occupation, income, or population density, also was

included in the statistical analysis in an attempt to adjust for

potential confounding factors. In one study the statistical

significance of the observed associations between stomach and rectal

cancer mortality and group exposures to current THM levels disappeared

when migration patterns and ethnic data were included in the regression

model (Tuthill and Moore, 1980). A wide range of cancer sites was found

to be statistically associated with estimates of population group

exposures based on current levels of THM or chloroform including gall

bladder, esophagus, kidney, breast, liver, pancreas, prostate, stomach,

bladder, colon, and rectum. The most frequent associations observed

were for the last four sites; however, these associations were not

consistent when viewed by gender, race, and geographic region.

The ecologic design coupled with the lack of specific exposure

indicators in these studies precludes the inference of a causal

relationship (Morgenstern, 1982). A subcommittee of the National

Academy of Sciences (NAS, 1980) reviewed 12 of the ecological studies

and noted ``Results of these studies demonstrate the problems of

establishing relationships between health statistics and environmental

variables, and lend emphasis to the caution with which they should be

interpreted.'' The NAS further commented that the ecological studies in

which the current THM exposures were estimated were deemed to be more

informative than others and ``suggest that higher concentrations of THM

in drinking water may be associated with an increased frequency of

cancer of the bladder. The results do not establish causality, and the

quantitative estimates of increased or decreased risk are extremely

crude. The effects of certain potentially important confounding

factors, such as cigarette smoking, have not been determined.'' The

studies are useful, however, as an initial step for identification of

potential hazards and they indicated the need for further epidemiologic

studies or analytic studies of individuals with a specific etiologic

hypothesis.

ii. Cohort Studies. A cohort study (or follow-up) study (the study

can be called either retrospective or prospective) is one in which two

or more groups (referred to as `cohorts') of people that differ

according to the extent of exposure to a potential cause of disease are

compared with respect to incidence of the disease of interest in each

of the groups. The essential element of this study type is that

incidence rates are calculable directly for each study group (Rothman,

1986). One advantage of this study type is the ability to study

multiple disease endpoints. One disadvantage of this study type is that

a large study population is needed to detect a relatively small risk.

In addition, because of the latency period for carcinogenicity, a long

follow-up time may be required for the study.

There exists one cancer drinking water cohort study where

individual data were available for a well-defined, fairly homogeneous

area that allowed disease rates to be computed by presumed degree of

exposure to by-products of chlorination, although the population was

relatively small. Wilkins and Comstock (1981) studied the residents of

Washington County, Maryland and ascertained the source of drinking

water at home for each county resident in a private census conducted in

1963. In addition to water source, information was collected on age,

marital status, education, smoking history, number of years lived in

the household, and frequency of church attendance. Death certificates

and cancer registry information was sought for county residents whose

date of death or diagnosis occurred in the 12 year period following the

census. Sex and site-specific cancer rates were constructed for

malignant neoplasms of biliary passages and liver, kidney, and bladder.

Several additional causes of death were analyzed as well for comparison

purposes. The population was stratified into three separate exposure

subgroups: chlorinated surface water, unchlorinated deep wells, and

small municipal systems with a mixture of chlorinated and unchlorinated

water, each reflecting a different history of exposure to by-products

of chlorination. The study group which included individuals who

obtained drinking water from small municipal systems were not included

as a comparison with the other drinking water cohorts because of their

exposure to both chlorinated and unchlorinated water.

Both crude and adjusted incidence rates for liver cancer in males

and females and for cancer of the liver among males were essentially

the same for persons supplied with chlorinated surface water at home

(high THM exposure) and for persons with deep wells (low THM exposure).

The adjusted rates for bladder cancer (RR=1.6; 95% CI=0.54,6.32) and

cancer of the liver (RR=1.8; 95% CI=0.64,6.79) among females were

highest among persons using chlorinated surface water. Given the low

relative risk and broad confidence intervals, the authors indicated

that this finding could be attributed to chance (Wilkins and Comstock,

1981). Confounding bias may also influence the interpretation of a

small relative risk. EPA considers that the results of this study are

inconclusive because the results are based on small numbers of cases,

hence, the reported rates are statistically unstable and subject to

random variation.

iii. Case Control Studies. In a case control study, persons with a

given disease (the cases) and persons without the given disease (the

controls) are selected for study. The proportions of cases and controls

who have certain background characteristics or who have been exposed to

possible risk factors are then determined and compared. Exposure odds

ratios (ORs) are determined. The odds of exposure among cases is

compared with that of controls. For rare diseases, the ORs are

considered good estimates of relative risk. These studies are sometimes

called case-referent or retrospective studies. Because there are many

variations of this study design (e.g., how cases and controls are

selected, how information on exposures, risk factors, and confounding

factors are obtained, and who is interviewed), each case control study

should be evaluated individually to determine if the specific study

design parameters introduce systematic bias (Kelsy et al., 1986). As

previously noted, all epidemiology studies require careful evaluation

of systematic bias. For those studies with major bias, the results are

generally considered inconclusive. Those studies with minor bias may

still provide useful information.

Two types of studies were conducted: (1) Decedent cases without

interviewing survivors for information about residential histories and

risk factors and (2) incident cases with interviews.

Decedent Case-Control Studies. Several case-control studies were

conducted to continue to investigate the possibility that there was a

causal relationship between chlorinated drinking water, including

byproducts such as THMs, and gastrointestinal or urinary tract cancers.

Most of these case-control studies used deceased cases of the specific

cancers of interest, although some continued their investigations in a

relatively nonspecific way by using both total cancer mortality as well

as several of the site- specific cancers studied in the ecologic

studies (Crump and Guess, 1982; Shy, 1985). Controls were noncancer

deaths from the same geographic area and in all but one study matched

for several potentially confounding variables including age, race, sex,

and year of death. As in all studies of this design (i.e., death

certificate studies with no available interviews), control of

confounding factors was restricted to information that is routinely

recorded on death certificates and no information was obtained from

next-of-kin interviews. The exposure variables of interest at this time

included a comparison of surface v ground water sources, or chlorinated

v nonchlorinated ground water sources. The place of residence listed on

the death certificate was linked to public records of water source and

treatment practices in order to classify the drinking water exposure

variable for a particular case or control (Shy, 1985).

Similar to the earlier ecologic studies, the Agency considers the

results from these studies to be inconsistent in their findings. The

calculated ORs, varied by cancer site and sex, as well as in their

magnitude and statistical significance. This variability was found for

all the cancer endpoints studied including those of specific interest,

i.e., bladder, colo-rectal, and/or colon. These endpoints were found to

vary by geographic region. For example, a statistically significant

increased bladder cancer risk was observed in North Carolina for males

and females combined (OR=1.54) and New York for males (OR=2.02), but

not for females; no statistically significant risk was seen in

Louisiana, Wisconsin or Illinois. Increased colon cancer risk was

observed in Wisconsin (OR=1.35) and North Carolina (OR=1.30) for males,

but not females; no increased risk was seen in Louisiana or Illinois.

Increased rectal cancer risk was observed in North Carolina (OR=1.54)

and Louisiana (OR=1.68) for males and females combined, in Illinois for

females (OR=1.35) but not males, and in New York for males (OR=2.33)

but not females; no increased risk was seen in Wisconsin. Although

increased risk was observed for cancer of the liver and kidney

(OR=2.76), esophagus (OR=2.39), and pancreas (OR=2.23) among males in

New York, no increased risk for these cancers was seen among females in

New York, Illinois, Wisconsin or Louisiana.

Although many of the ORs were statistically significant, these

decedent case control studies with extremely limited information on

confounding factors and potential exposures to chlorinated water are of

limited usefulness in assessing whether cancer is associated with

chlorinated drinking water, or judging the causality of such as

association. Although some of the ORs were large enough to cause

concern about an exposure association, the magnitude of the OR was such

that the association could be attributed to incomplete control of

confounding factors and the ORs might represent spurious elevations

(Crump and Guess, 1982).

Although not subject to all the same limitations as ecologic

studies, decedent case-control studies are considered more limited by

some epidemiologists than others as a tool for causal inference because

of a high probability of systematic bias associated with the use of

information obtained only from the death certificate (e.g., inadequate

or no information on residential history, water exposures, and major

potential confounders). The variability seen in these five studies is

likely a combination of several factors, including available sample

size, choice of causes of death included as controls, regional

variability in true composition of the raw and treated drinking waters,

definition of exposure variables, a high probability of exposure

misclassification from imputing a lifetime exposure to a certain water

source or treatment from residence listed on the death certificate, and

uncontrolled confounding (e.g., diet and smoking).

Given the limitations of decedent case control studies without

interviews, the evidence from these studies are considered insufficient

to determine a causal association between any or all the components

which exist in the complex mixture created during the chlorination of

surface waters and any site-specific cancer. The findings provided a

stimulus for a further refined epidemiologic study using incident cases

of bladder and colon cancer and appropriate controls who could be

interviewed for residential history and numerous other covariates.

Case-Control Studies with Interviews. At the time when these more

recent studies were planned, it was still believed that THMs were the

major by-products of chlorinated drinking water that should be

investigated and studies were designed and conducted in areas where a

THM difference might be expected and somehow measurable. Exposure

assessment for individuals remained problematic in the study design.

The best available means of exposure measurement, however, was at best

a surrogate for the true exposure of interest which is the actual level

of THMs or other by-products ingested over a person's lifetime through

consumption of surface water disinfected with chlorine. Only two

studies attempted to estimate long-term exposure to THMs. Most studies

used residence at a location served by chlorinated drinking water. In

all except one of the studies, comparisons of exposure were between

chlorinated surface water and unchlorinated groundwater. As previously

discussed in the section on ecological studies, the water quality for

surface and ground water differ for many other consituents.

Because it was known that disinfection of surface water using

chloramine produced very low levels of THMs and other by-products

compared to the same water disinfected with chlorine, a study was

conducted in Massachusetts to compare the patterns of mortality in

communities which used these different disinfectants (Zierler et al.,

1986). Statewide mortality records for 1969-1983 were analyzed using

standardized mortality ratios (SMRs) and showed little variation by

community. However, mortality odds ratios (MORs) comparing bladder

cancer deaths to all other deaths were considered by the authors to

indicate a slight elevation for last residence in a chlorinated

community compared to a chloraminated community (MOR=1.7; 95% CI=1.3,

2.2). The authors noted that the results were preliminary and ``crude

descriptions of the relationship under study'' (Zierler et al., 1986).

The authors further indicated that the results may have been caused by

unidentified or uncontrolled confounding factors.

Bladder cancer deaths were investigated further using a case-

control design with proxy interviews to determine residential and

smoking histories (Zierler et al., 1988). The association of bladder

cancer was assessed for individuals with lifetime and usual exposure to

chlorinated and chloraminated water depending on the number of years of

residence at a particular water source. Residence in a community using

chlorinated drinking water was used as an index for exposure to

chlorinated by-products, while residence in a community using

chloramine for disinfection was considered an index for no exposure to

chlorinated by-products. An association was observed between bladder

cancer and both lifetime (MOR=1.6; 95% CI=1.2-2.1) and usual (MOR=1.4;

95% CI=1.1-1.8) exposure to chlorinated water. A subgroup of study

participants was noted to have lived their entire lives in an area

served with water supplied by the Massachusetts Water Resources

Authority, disinfected with either chlorine or chloramine (same water

source, different disinfectant, lifetime exposure). Within this group

the bladder cancer mortality risk was 1.6 times higher (MOR=1.6; 95%

CI=1.1, 2.4) when the water had been disinfected with chlorine compared

to chloramine (Zierler et al., 1988).

In addition to analyses using a control group which consisted of

deaths from cardiovascular disease, cerebrovascular disease, chronic

obstructive lung disease, lung cancer, and lymphatic cancer, a separate

analysis was done using only the lymphatic cancer controls. This was

considered necessary by the authors because of the possibility that

some of the other deaths among controls may also be related to the

exposure of interest. If true, then the MOR estimate would be biased

toward the hypothesis of no increased risk. When the analysis

considered only lymphatic cancer controls, the magnitude of the

association with chlorinated water increased for lifetime exposure

(MOR=2.7; 95% CI=1.7-4.3), usual exposure (MOR=2.0; 95% CI=1.4-3.0),

and lifetime exposure in the previously mentioned subgroup (MOR=3.5;

95% CI=1.8-6.7). Sources of misclassification bias that may have been

present were considered to be randomly distributed among the cases and

controls which implies that the observed MOR would be an underestimate

of risk (Zierler et al., 1988). It is also possible that

nondifferential misclassification of the variables used to control

confounding, leading to residual confounding of the summary estimates,

could have caused a systematic spurious elevation in the MORs.

The largest study to date investigating the relationship of

chlorinated water and bladder cancer incidence involved an ancillary

study to the National Cancer Institute's (NCI) 10 area study of bladder

cancer and artificial sweeteners (Cantor et al., 1985, 1987, 1990). The

original study conducted interviews with 2,982 newly diagnosed bladder

cancer cases and 5,782 population controls; lifetime information on

source and treatment of drinking water was collected and analyzed for

only a subset of the original study population (1,244 cases and 2,550

controls). Subgroup analyses of nonsmokers among participants and those

reporting beverage intake necessarily involved even smaller numbers.

Duration of exposure, measured by years of residence at a chlorinated

surface or nonchlorinated ground water source was presumed to be a

surrogate for dose of disinfectant by-products. Overall, there was no

association of duration of exposure with bladder cancer risk (Cantor et

al., 1985, 1987). In nonsmokers who never smoked, a 2-fold increased

risk was reported for those exposed for 60 or more years to chlorinated

surface water (n=46 cases, 77 controls) compared to unchlorinated

ground water (n=61 cases, 268 controls) users (OR=2.3; 95% CI=1.3,

4.2). These data were further analyzed according to beverage intake

level, type of water source and treatment (Cantor et al., 1987, 1990).

It was observed that people who reported drinking the most tap water-

based beverages from any source (>1.96 liters/day) had a bladder cancer

risk about 40% higher (OR=1.43; 95% CI=1.23-1.67, males and females

combined) than people who drank the least. The association between

water ingestion and bladder cancer risk for males was an OR=1.47; 95%

CI=1.2-1.8, and for females an OR=1.29; 95% CI=0.9-1.8.

Evaluation of bladder cancer risk by both duration of exposure and

amount of water consumed showed that the risk increased with higher

water consumption only among those who drank chlorinated surface water

for 40 or more years. Evaluation of risk by smoking status revealed

that most of the duration effect was observed in nonsmokers. Among

nonsmokers who consumed tap water in amounts above the population

median (>1.4 L/day), a risk gradient was apparent only for males.

However, a higher risk was also seen for nonsmoking females who

consumed less than the median level. The increasingly smaller numbers

of cases and controls available for these subgroup analyses produce

statistically unstable OR estimates making it difficult to evaluate the

trend results.

This is the first study of incident bladder cancer cases that

obtained and analyzed fluid consumption patterns in this way. The noted

inconsistencies in the reported data must be more thoroughly explored

and indicate a need for replication before any causal relationship can

be assumed (Devesa et al., 1990). An additional consideration is a more

refined exposure measurement; many of the disinfection by-products are

volatile. Thus exposure may occur through inhalation as well as

ingestion.

Two conflicting studies of colon cancer and presumed THM exposure

have been reported. The first one (Cragle et al., 1985) was a hospital

based case control study that included 200 incident colon cancer cases

from seven hospitals and 407 hospital controls with no history of

cancer who were diagnosed with diseases unrelated to colon cancer. It

should be noted that both colon and rectal cancer cases were included

as cases in the study. Controls were matched to cases on hospital and

admission date, as well as age, race, sex and vital status. Residential

histories were linked with water source and disinfectant information

for the 25 years prior to diagnosis. Logistic regression analysis using

qualitative data groupings for the variables of interest showed a

strong interaction of age and chlorination status (Table V-4). THM

levels were not estimated. Odds ratios computed from the regression

coefficients increased with age, and within age groups. The ORs are

higher for a longer duration of exposure.

Table V-4.--Comparison of OR's By Exposure Duration and Age

------------------------------------------------------------------------

OR (95% CI) 1-15 OR (95% CI) > 15

Age (years) years exposure years exposure

------------------------------------------------------------------------

20-29............................. 0.23

(0.11, 0.49) 0.48

(0.23, 1.01)

30-39............................. 0.36

(0.2, 0.66) 0.6

(0.33, 1.09)

40-49............................. 0.57

(0.36, 0.88) 0.75

(0.48, 1.18)

50-59............................. 0.89

(0.83, 1.12) 0.94

(0.69, 1.29)

60-69............................. 1.18

(0.94, 1.47) 1.38

(1.1, 1.72)

70-79............................. 1.47

(1.16, 1.84) 2.15

(1.7, 2.69)

80-89............................. 1.83

(1.32, 2.53) 3.36

(2.41, 4.61)

------------------------------------------------------------------------

From these data, it appears that risk is increased only in those

persons 60 years old and older with greater than 15 years of exposure

to chlorinated water and in those greater than 70 years of age,

regardless of exposure duration.

A second colon cancer study (Young et al., 1987) conducted in

Wisconsin involved 366 incident colon cancer cases, 785 controls

diagnosed with other cancers, and 654 population controls. Extensive

interviews were conducted with all participants to obtain information

on past drinking water sources, drinking water habits and a number of

potentially confounding covariates. This information was combined with

data provided by water companies to construct models to predict

historical levels of THMs to be used as both cross-sectional and

cumulative exposure variables. Simpler methods of defining exposure

were also used, (e.g., surface vs. ground, chlorinated v.

nonchlorinated), and all methods looked at the data by period specific

exposure levels. The results did not indicate any association between

THMs in Wisconsin drinking water and colon cancer risk. Odds ratios for

all exposure variables were uniformly close to 1.0 with few exceptions.

It should be noted, however, that in this study the majority of the

water supplies contained less than 20 g/l of THMs. No excess

risk was observed at these levels, given the limitations of this study

design in detecting a small risk.

The association of THM and colo-rectal cancer was studied in New

York where the THM levels were higher than those in the above Wisconsin

study (Lawrence et al., 1984). A total of 395 colon and rectal cancer

deaths among white female teachers in New York State (excluding New

York City) was compared with an equal number of deaths of teachers from

causes of death other than cancer. All deaths were ascertained using

the defined cohort of the New York State Teachers Retirement System.

Cumulative chloroform exposure was estimated by the application of a

statistical model to operational records from water systems that served

the home and work addresses of the study participants during the 20

years prior to death. The distribution of chloroform exposure was not

significantly different between cases and controls. No effect of

cumulative chloroform exposure was observed in a logistic analysis

controlling for type, population density, marital status, age, and year

of death. No excess risk was associated with exposure to a surface

water source containing THMs (OR=1.07; 90% CI=0.79, 1.43). Although the

data were not presented in the article, the authors reported that no

appreciable differences were seen when the colon and rectal cases were

analyzed separately, compared to the combined analyses reported above.

Although most all the studies reviewed here have looked at colon,

colo-rectal or bladder cancer risk, one recently published work

investigated the risk of pancreatic cancer in relation to presumed

exposure to chlorinated drinking water. Ijsselmuiden et al. (1992)

conducted a population-based case-control study in Washington County,

Maryland, using the same population data that were originally

ascertained during a private population census for an earlier cohort

study (Wilkins and Comstock, 1981). The original cohort study did not

find any association between pancreatic cancer and chlorinated drinking

water (OR=0.80, 95% CI=0.44-1.52).

This case-control study was conducted to reexamine chlorinated

drinking water as a possible independent risk factor for pancreatic

cancer in this population. It is not reported of any of the other

endpoints from the original study also were reexamined, e.g., bladder,

kidney, or liver cancer. Cases were those residents who were reported

to the County cancer registry with a first time pancreatic cancer

diagnosis during the period July, 1975 through December 1989, and who

had been included in the 1975 census (n=101). Controls were randomly

selected by computer from the 1975 census population (n=206). Drinking

water source, as obtained during the 1975 census, was the exposure

variable used. In univariate analyses, municipal water as a source of

drinking water, increasing age, and unemployment were significantly

associated with increased risk of pancreatic cancer. Multivariable

analyses that controlled for confounding variables indicated that the

use of municipal chlorinated water at home was associated with a

significant OR of 2.23 (95% CI=1.24-4.10). The OR adjusted is 2.18 (95%

CI=1.20-3.95); only age and smoking were assessed as potential

confounders.

Interpretation of these findings is hampered by several problems

regarding the assessment of exposure, including the fact that

information obtained in 1975 on type of water and other variables is an

exposure collected at one point in time and may not reflect actual

exposure patterns prior to 1975. In addition, there is no information

on the actual amounts of water consumed. Additionally, different

residential criteria were used for the cases and controls. The cases

had to still be residing in the County at the time of their cancer

diagnosis to be included in the study, but the controls may not have

been current residents. If controls emigrated out of the county

differentially on the basis of exposure, the ORs may be an over- or

underestimate of the risk depending on emigration patterns. Finally, it

can not be ruled out that the exposure variable used for this and other

studies--residence served by a particular water source--is simply a

surrogate for some other unidentified factor associated with nonrural

living. The nonspecific relationship of several different causes of

death and water source at home observed in the earlier cohort study

(Wilkins and Comstock, 1981) lends some support to this possibility.

More valid individual exposure information over a long period of time

for both specific contaminants and the use of chlorinated/unchlorinated

water are needed to assess the results of this and other analytical

epidemiology studies.

Morris et al. (1992) conducted a meta-analysis, evaluating 12

studies and pooling the relative risks from 10 epidemiological studies

of cancer and a presumed exposure to chlorinated water and its

byproducts. Meta-analysis refers to the application of quantitative

methods to combine the published results of a related body of

literature (Dickerson and Berlin, 1992). Morris et al. (1992) reported

a pooled relative risk estimate of 1.21 (95% CI, 1.09-1.34) for bladder

cancer and 1.38 (95% CI, 1.01-1.87) for rectal cancer (i.e., 9% of

bladder cancer cases and 15% of the rectal cancer cases in the U.S. or

approximately 10,000 additional cases of cancer per year could be

attributed to chlorinated water and its by-products). Pooled relative

risk estimates for ten other site specific cancers including colon,

colo-rectal and pancreas were not felt to be significantly elevated nor

were they statistically significant.

If the indications from this analysis are true such that water

chlorination could result in as many as 10,000 cases of cancer a year,

then chlorination could represent a significant cause of rectal and

bladder cancer in the U.S. However, there was disagreement among the

negotiating parties over the appropriateness of this meta-analysis.

Some believed that the use of the meta-analysis may not be appropriate

for these data. Others disagreed, expressing their view that the

analysis was statistically probative and otherwise valuable. Meta-

analysis has been used successfully to combine the results of small

clinical trials and of some epidemiology studies that have similar

experimental design and exposure conditions. Application of meta-

analysis to the water chlorination data requires careful consideration

of exposure variables and systematic bias in each of the studies.

Chlorinated drinking water is a complex mixture of many substances that

vary geographically and seasonally. There is even variability within a

geographic region. In addition, the information on exposure and

potential confounding is much more limited for the four decedent case

control studies used in the meta-analysis. Their study design is

dissimilar to the other studies included, resulting in concerns about

their inclusion in the meta-analysis. Study-specific methodological

problems, systematic bias, and problems of exposure definition and

assessment could not be corrected by this analysis (Murphy, 1993).

Thus, the overall results of the Morris et al. analysis may over- or

underestimate the risk. However, the estimate of risk in regard to

rectal cancer might be particularly affected by the inclusion of these

case control studies. It should also be noted that the results of the

Morris et al. analysis does not provide additional information to

establish causality.

The chlorinated drinking water epidemiology studies have been

reviewed extensively by EPA, the National Academy of Sciences, the

International Agency for Research on Cancer (IARC), and the

International Society for Environmental Epidemiology (ISEE). In 1987,

the National Academy of Sciences Subcommittee on Disinfectants and

Disinfectant By-Products concluded that there was a major health

concern with the chronic ingestion of low levels of disinfection

byproducts (NRC, 1987). The Subcommittee commented that some of the

epidemiology studies reported ``increased rates of bladder cancer

associated with trends of levels of certain contaminants in water

supplies. Interpretation of these studies is hampered by a lack of

control for confounding variables (e.g., age, sex, individual health,

smoking history, other exposures).'' The Subcommittee recommended that

epidemiologists continue to improve protocols and conduct studies on

drinking water and bladder cancer where exposure data can be obtained

from individuals, rather than through estimation from exposure models.

EPA and IARC, along with other individual scientists, have

interpreted the epidemiologic evidence as inadequate. IARC concluded

that ``there is inadequate evidence for carcinogenicity of chlorinated

drinking water in humans.''

The ISEE presented a full spectrum of opinion regarding the

epidemiology data (Neutra and Ostro, 1992). The ISEE reported a

``general consensus that the results of the recent EPA-sponsored

studies of cancer endpoints have strengthened the evidence for linking

bladder cancer with long term exposure to chlorinated drinking water.

The evidence for links with colon cancer are not convincing. * * * Any

risks, if real, are low when compared to the risk of infection from not

disinfecting water.''

In 1992, the International Life Sciences Institute sponsored a

conference with the Pan American Health Organization, EPA, Food and

Drug Administration, World Health Organization, and the American Water

Works Association on the safety of water disinfection. Although they do

not necessarily reflect the views of the sponsoring organizations,

conclusions prepared by the conference's editor and editorial board

(Craun et al., 1993) noted that ``Adverse human health effects may be

associated with the chemical disinfection of drinking water. However,

current scientific evidence is inadequate to conclude that water

chlorination poses a significant risk to humans. Uncertainties about

the available toxicologic evidence limit assessment of human health

risks associated with chlorine, chloramine, chlorine dioxide, and ozone

disinfection. The epidemiologic evidence for increased cancer risks of

chlorinated drinking water is equivocal.''

Some members of the reg-neg committee felt that the epidemiology

data, taken in conjunction with the results from toxicological studies,

provide an ample and sufficient basis to conclude that the usual

exposure to disinfection by-products in drinking water could result in

an increased cancer risk at levels encountered in some public water

supplies.

Because of the spectrum of conclusions concerning these data, the

Agency is pursuing additional research to reduce the uncertainties

associated with these data and better characterize the potential of

cancer risks associated with the consumption of chlorinated drinking

water.

b. Serum Lipids/Cardiovascular Disease. Laboratory studies on

animals, conducted in the early 1980's indicated a possible link

between consumption of chlorinated drinking water and elevated serum

lipid profiles which are indicators of cardiovascular disease (USEPA

1994a). The animal work was followed by a cross-sectional study in

humans (Zeighami et al., 1990) that included 1,520 adult residents,

aged 40 to 70 years, in 46 Wisconsin communities supplied with either

chlorinated or unchlorinated drinking water of varying hardness. The

study was designed to determine whether differences in calcium or

magnesium intake from water and food and chlorination of drinking water

affect serum lipids.

The communities selected for study had the following

characteristics: (1) They were small in population size (300-4,000) and

not suburbs of larger communities; (2) they had not undergone more than

20% change in population between 1970 and 1980; (3) they had been in

existence for at least 50 years; and (4) all obtained water from

groundwater sources with no major changes in water supply

characteristics since 1980 and did not artificially soften water. The

water for the communities contained total hardness of either

80 mg/l CaCO3 (soft water) or 200 mg/l

CaCO3 (hard water); 24 communities used chlorine for disinfection and

22 communities did not disinfect. Eligible residents were identified

through state driver's license tapes and contacted by telephone; an

age-sex stratified sampling technique was used to choose a single

participant from each eligible household. Only persons residing in the

community for at least the previous 10 years were included. A

questionnaire was administered to each participant to obtain data on

occupation, health history, medications, dietary history water use,

water supply and other basic demographic information. Water samples

were collected from a selected subset of homes and analyzed for

chlorine residual, pH, calcium, magnesium, lead, cadmium, and sodium.

Fasting blood specimens were collected from each participant and

analyzed for total cholesterol, triglycerides and high- and low-density

lipoprotein (HDL and LDL, respectively) subfractions.

Among females, adjusted mean total serum cholesterol levels were

statistically significantly higher in the chlorinated communities

compared to the nonchlorinated communities (249 mg/dl and 238 mg/dl,

respectively). These changes are not considered biologically

significant as they reflect background variation. Total serum

cholesterol levels were also higher for males in chlorinated

communities, on the average, but the difference was smaller and not

statistically significant (236 mg/dl vs. 232 mg/dl). LDL mean values

followed a similar pattern to that for total cholesterol, higher in

chlorinated communities for females, but not different for males.

However, for both sexes, HDL cholesterol levels are nearly identical in

chlorinated and nonchlorinated communities and there were no

significant differences found in the HDL/LDL ratios. The implications

of these findings for cardiovascular disease risk are unclear at this

time given the inconsistencies in the data. The possibility exists that

the observed association in females may have resulted from some unknown

or undetermined variable in the chlorinated communities.

The results from a second study, designed to further explore the

findings among female participants in the Wisconsin study (Zieghami et

al., 1990), were presented in 1992 (Riley et al., 1992, manuscript

submitted for publication). Participants were 2,070 white females, aged

65 to 93 years who were enrolled in the Study of Osteoporotic Fractures

(University of Pittsburgh Center) and had completed baseline

questionnaires on various demographic and lifestyle factors. Total

serum cholesterol was determined for all participants. Full lipid

profiles (total cholesterol, triglycerides, LDL, total HDL, HDL-2, HDL-

3, Apo-A-I, and Apo-B) were available from fasting blood samples for a

subset of 821 women. Interviews conducted in 1990 ascertained

residential histories and type of water source used back to 1950 and

all reported public water sources were contacted for verification of

disinfectant practices. Private water sources were presumed to be

nonchlorinated. A total of 1,896 women reported current use of public,

chlorinated water, 201 reported current use of nonchlorinated springs,

cisterns, or wells and 35 reported having mixed sources of water. Most

of the women had been living in the same home with the same water

service for at least 30 years.

Overall, there were no meaningful differences detected in any of

the measured serum lipid levels between women currently exposed to

nonchlorinated water and those exposed to chlorinated water (246 mg/dl

vs. 247 mg/dl, respectively, for total cholesterol). The data were also

stratified by age and person-years of exposure to chlorinated water at

home. There was some suggestion that women with no exposure to chlorine

had lower total cholesterol levels but this finding was inconsistent

and may represent random fluctuation since there was no trend noted

with LDL cholesterol or Apo-B, both of which are known to correlate

with total cholesterol. There was also no association between

increasing duration of exposure to chlorine and HDL cholesterol, Apo-A-

I, or triglycerides.

The only notable differences were that women with chlorinated water

reported significantly more cigarette and alcohol consumption than the

women with nonchlorinated drinking water (Riley et al., 1992). This was

evident in all age groups and across strata of duration of exposure.

This finding lends support to the possibility that the previously

reported association of chlorinated drinking water and elevated total

serum cholesterol (Zeighami et al., 1990) may have arisen due to

incomplete control of lifestyle factors which were differentially

distributed across chlorination exposure groups.

c. Reproductive/Developmental Outcomes. Several recently conducted

epidemiologic studies have examined the relationship between different

reproductive or developmental endpoints and various components of

drinking water. Kramer et al. (1992) conducted a population-based case-

control study to determine whether water supplies containing relatively

high levels of chloroform and other THMs within the state of Iowa are

associated with low birthweight, prematurity, or intrauterine growth

retardation (IUGR). Iowa birth certificate data from January, 1989

through June, 1990 served as the source of both cases and controls.

Definitions for cases and controls were as follows: the low birthweight

group included 159 live singleton infants weighing 2,500 grams from

the same population; the prematurity group included 342 live singleton

infants with gestational ages of 37 weeks; IUGR

analyses included 187 IUGR infants (defined as weighing less than the

5th percentile for a particular gestational age based on California

standards for non-Hispanic whites) and 935 randomly selected controls.

Exposure status was assigned to infants according to reported maternal

residence in a given municipality at the time of birth. The assigned

THM levels came from a water survey conducted in 1987 in th

This text is long and has been trimmed here. Open the source document for the complete record.