United States
EnvironmentalProtection
Agency
EPA-600/1-84-004
Aug. 1984
Research and Development
vvEPA
Health Effects
Criteria for Fresh
Recreational Waters
Wofe: This electronic version was prepared from the original document using
optical character recognition software.
-------
EPA-600 / 1-84-004
August 1984
Health Effects Criteria
for
Fresh Recreational Waters
by
Alfred P. Dufour
Toxicology and Microbiology Division
U.S. Environmental Protection Agency
Cincinnati, OH 45268
HEALTH EFFECTS RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 277711
-------
NOTICE
This document has been subject to the U.S. Environmental Protection Agency's
peer and administrative review and approved for publication. Mention of trade
names or commercial products does not constitute endorsement or recommendation
for use.
-------
FOREWORD
The many benefits of our modern, developing, industrial society are accompanied
by certain hazards. Careful assessment of the relative risk of existing and new man-
made environmental hazards is necessary for the establishment of sound regulatory
policy. These regulations serve to enhance the quality of our environment in order to
promote the public health and welfare and the productive capacity of our Nation's
population.
The complexities of environmental problems originate in the deep interdependent
relationships between the various physical and biological segments of man's natural
and social world. Solutions to these environmental problems require an integrated
program of research and development using input from a number of disciplines. The
Health Effects Research Laboratory, Research Triangle Park, NC and Cincinnati, OH
conducts a coordinated environmental health research program in toxicology,
epidemiology and clinical studies using human volunteer subjects. Wide ranges of
pollutants known or suspected to cause health problems are studied. The research
focuses on air pollutants, water pollutants, toxic substances, hazardous wastes,
pesticides and nonionizing radiation. The laboratory participates in the
developmentand revision of air and water quality criteria and health assessment
documents on pollutants for which regulatory actions are being considered. Direct
support to the regulatory function of the Agency is provided in the form of expert
testimony and preparation of affidavits as well as expert advice to the Administrator
to assure the adequacy of environmental regulatory decisions involving the protection
of thehealth and welfare of all U.S. inhabitants.
This report provides an assessment of the relationship between microbiological
indicators of water quality and illness that may have resulted from swimming. The
data base resulted from a series of in-house and extramural epidemiological-
microbiological research projects designed to develop the criterion for fresh waters.
The development and periodic reevaluation of such criteria is mandated by Section
304(a)l of Public Law 92-500: Federal Water Pollution Control Act Amendments of
1972; Clean Water Act of 1977.
F. Gordon Hueter, Ph.D.
Director
Health Effects Research Laboratory
in
-------
ABSTRACT
A criterion for the quality of the bathing water, based upon swimming-associated
gastrointestinal illness, was developed from data obtained during a multi-year
freshwater epidemiological-microbiological research program conducted at bathing
beaches near Erie. Pennsylvania and Tulsa, Oklahoma. Three bacterial indicators of
fecal pollution were used to measure the water quality. E. coli, enterococci and fecal
coliforms. A good correlation was observed between swimming-associated gastro-
intestinal symptoms and either E. coli or enterococci densities in the water. Fecal
coliform densities showed little or no correlation to gastrointestinal illness rates in
swimmers. In general, high gastrointestinal illness rates were associated with high
densities of fecal indicator bacteria. A comparison of the freshwater results with the
results obtained from studies at marine bathing beaches indicated that a separate
criterion should be used with each type of bathing water.
IV
-------
CONTENTS
Number Page
Forward iii
Abstract iv
Figures vi
Tables vii
Acknowledgments viii
1. Introduction 1
2. Conclusions 2
3. Background 3
Swimming-Associated Outbreaks of Disease 3
Retrospective Epidemiological Studies 5
Prospective Epidemiological Studies 6
Water Quality Standards for Bathing Beaches 1924-1980 7
4. Freshwater Studies 10
Experimental Design 10
Lake Erie Study 12
Keystone Lake 13
5. Development of a Criterion 18
Criteria for Freshwater Bathing Areas 21
6. Marine Versus Freshwater Criteria 27
References 31
-------
FIGURES
Number Page
1 Estimated Regression Lines for Highly Credible and Total
Gastrointestinal Symptom Rates on E. coli Densities 22
2 Estimated Regression Lines for Highly Credible and Total
Gastrointestinal Symptom Rates on Enterococci Densities 22
3 Estimated Regression Lines for Highly Credible and Total
Gastrointestinal Symptom Rates on Fecal Coliform Densities 23
4 Criterion for Estimating Swimming-Associated Gastrointestinal
Illness Rate from the Geometric Mean Density of E. coli per
100 ml in Freshwater Samples 23
5 Criterion for Estimating Swimming-Associated Gastrointestinal
Illness Rate from the Geometric Mean Density of Enterococci per
100 ml in Freshwater Samples 24
6 Criterion for Estimating the Geometric Mean E. coli Density
per 100 ml from an Acceptable Risk Level of Swimming-Associated
Gastrointestinal Illness 25
7 Criterion for Estimating the Geometric Mean Enterococci Density
per 100 ml from an Acceptable Risk Level of Swimming-Associated
Gastrointestinal Illness 26
8 Data Summary of Highly Credible Gastrointestinal Symptom
Rates and Indicator Densities from Marine and Freshwater Studies 28
9 Marine and Freshwater Criteria for Swimming-Associated
Gastrointestinal Illness and Water Quality Using Enterococci to
Measure the Water Quality 29
10 Marine and Freshwater Criteria for Swimming-Associated
Gastrointestinal Illness and Water Quality Using E. coli to
Measure the Water Quality 30
VI
-------
TABLES
Number Page
1 Outbreaks of Disease Associated with Swimming in Natural Waters 3
2 Follow-up Success Rate for Beach Contacts at Lake Erie,
Pennsylvania, 1979, 1980 and 1982 12
3 Symptom Rates by Category for Swimmers and Nonswimmers at
Lake Erie Beaches,1979-1982 14
4 Bacterial Indicator Densities at Lake Erie, Pennsylvania
Bathing Beaches, 1979, 1980 and 1982 15
5 Follow-up Success Rate for Beach Contacts at Keystone Lake,
Oklahoma, 1979-1980 15
6 Symptom Rates by Category for Swimmers and Nonswimmers at
Keystone Lake Beaches, 1979-1980 16
7 Indicator Densities at Keystone Lake Bathing Beaches, 1979-1980 17
8 Summary of Microbiological and Epidemiological Results from
Lake Erie and Keystone Lake Bathing Beach Studies 19
9 Summary of Regression Statistics Related to Swimming-
Associated Illness and Water Quality Indicators 20
10 Summary of Statistics Related to Marine and Freshwater
Criteria for Highly Credible Swimming-Associated Illness
and Water Quality Indicators, E. coli and Enterococci 29
vn
-------
ACKNOWLEDGMENTS
The author gratefully acknowledges the work and cooperation of Prof. Stanley
Zagorski, Dr. Richard Gammon and Dr. Gerald Kraus of Gannon University, Erie,
Pennsylvania, and Drs. James Robertson, Donald Parker, Garry McKee and David
Shadid of the University of Oklahoma, Oklahoma City, Oklahoma.
The author also is indebted to Mr. Leland McCabe, Dr. Morris Levin, Dr. William
Watkins, Ms. Cynthia Thomas and the staff of the HERL Marine Field Station,
Kingston, Rhode Island for their many contributions.
A special acknowledgment is reserved for Dr. Victor Cabelli who directed the EPA
Recreational Water Quality Program until his retirement from government service in
1978. His dedicated work and steadfast guidance were, in large part, responsible for
the program reaching its goal, the development of a water quality criteria for
recreational waters.
Vlll
-------
SECTION 1
INTRODUCTION
The current EPA recommended criteria for bathing waters is that given in Quality Criteria
for Water (I) which states:
"Based on a minimum of five samples taken over a 30-day period, the fecal coliform
bacterial level should not exceed a log mean of 200 per 100 ml, nor should more than
10 percent of the total samples taken during any 30-day period exceed 400 per 100 ml"
This criterion, which is used by 95% of the states and territories of the United States (2),
was first proposed by the National Technical Advisory Committee (NTAC) to the Federal
Water Pollution Control Administration in 1968 (3). The NTAC used epidemiological data
collected by the United States Public Health Service (USPHS) from 1948-1950(4) to develop
the criteria for recreational bathing waters. The criterion was closely examined in 1972 by
a committee of the National Academy of Sciences - National Academy of Engineers (NAS-
NAE) and they concluded that the epidemiological data base used to develop the NTAC
criterion was too limited to be scientifically defendable (5). The NAS-NAE committee
decided not to recommend a criterion for recreational bathing waters based on the paucity
of epidemiological information available.
hi 1972, the Environmental Protection Agency (EPA) initiated a long-term recreational
water quality research program that was to examine the relationship between water quality
and swimming-associated acute infectious disease. The first phase of the program, from 1972
to 1978, was conducted at multiple marine bathing beaches in New York, Louisiana and
Massachusetts. The result of these studies was a marine recreational water criterion which
described a direct linear relationship between swimming-associated gastroenteritis and water
quality which was indexed by the density of enterococci in the water (6).
From 1978 to 1982, the EPA recreational water quality research program was directed at
freshwater bathing areas. This report will describe and summarize the results of freshwater
beach studies conducted in Pennsylvania and Oklahoma, and will present two fresh
recreational water criteria, which relate swimming-associated gastroenteritis to water quality,
characterized with either one of two bacterial indicators, enterococci or E. coli. It will be
shown that the model developed for the marine criterion, i.e., a direct linear relationship, has
been validated by the freshwater studies and, lastly, the results of the marine studies will be
compared to those of the freshwater results to show that a single criterion cannot be used for
marine and fresh bathing waters.
The material presented in this report is a natural extension of the information given in
"Health Effects Water Quality Criteria for Marine Recreational Water" (6). Many references
will be made to that report, since the rationale for the marine studies and the study design
have been used in the freshwater studies. Whenever possible the data presentations in this
report will be in such a manner that the results can be compared directly to those of the
marine studies. Although most of the information pertinent to developing a water quality
criterion are included in the organization of this report, there are certain elements that have
been omitted, which can be found in the marine water quality report, such as the two
sections which relate to water quality indicators and the limitations associated with the use
of criteria developed with bacterial water quality indicators of enteric origin.
-------
SECTION 2
CONCLUSIONS
The results of the freshwater bathing beach studies conducted at two sites over a
three-year period lead to the following conclusions.
1. Swimming-associated gastrointestinal illness is related to the quality of the
bathing water. A direct linear relationship was observed between highly credible
gastrointestinal illness and bacterial densities of two indicators of fecal
contamination, enterococci and E. coli.
1. The relationship between the rate of swimming-associated illness and bacterial
indicator density was almost identical for two of the indicators examined, E. coli
and enterococci. Thus, either indicator can be used to measure the potential for
swimming-associated illness in bathing waters. Fecal coliforms showed no
relationship to the rate of swimming-associated gastrointestinal illness.
3. The criterion developed for marine bathing waters is not applicable to fresh
bathing water. At equivalent indicator densities, the swimming-associated illness
rate was approximately three times greater in seawater swimmers relative to that
in freshwater swimmers.
-------
SECTION 3
BACKGROUND
Swimming-Associated Outbreaks of Disease
The history of disease outbreaks and illness associated with poor quality bathing water
has been described in many reviews on this subject (7,8,9). A listing of the most
frequently referenced literature on swimming-associated illness is given in Table I. This
table is instructive in the sense that it shows that some factors assumed to be important
in past considerations of the hazards related to swimming in polluted water may have
little relevance today. For instance, it is obvious that reports of disease in swimmers
caused by Salmonella species began to decline in the 1940's and none were reported after
the Australian outbreak which occurred in 1958. Two factors probably contributed to the
decrease in Salmonella related illness in swimmers. First, there was the steady increase
in the number of sewage treatment plants practicing disinfection, especially in large
population centers, and second, there was a widespread use of newly discovered
antibiotics which greatly aided in limiting the spread of disease and, thereby, the number
of ill individuals in the discharging population. Another obvious point is the lack of
swimming-associated outbreaks caused by poliovirus. Although strong evidence relating this
virus to disease contracted by swimming has never been presented, many of the early studies
Table 1, Outbreaks of Disease Associated with Swimming in Natural Waters
Year
1909
1921
1932
1942
1947
1958
1973
1974
1974
1976
1982
Location
Walmer,
England
Connecticut
New York
California
Beccles,
England
Perth,
Australia
Vermont
Niort, France
S. Carolina
Iowa
Michigan
Etiologic
Agent
Salmonella
Salmonella
Salmonella
Salmonella
Salmonella
Salmonella
Coxsackie B
Coxsackie A16
Hepatitis-A
Shigella
Norwalk Agent
Water
Quality
U*
U*
U*
U*
U*
U*
U
E, col/
50-1000/
100 ml
U*
U
U
No. Cases/
No. at Risk
347 NG
6/NG
51 /NG
NG/NG
9/NG
15/NG
21/33
5/NG
14/30
31/45
126/NG
Refer
ence
7
17
16
30
58
10
11
18
15
12
14
U
- Water quality not measured.
* - Suspected to be grossly polluted.
NG - Not given.
-------
on the effect of polluted water on swimmers were probably stimulated by the anxiety
created by the then incurable disease caused by poliovirus.
After the 1958 outbreak of salmonellosis in Australia (10) there was a long period
when no outbreaks of swimming-associated illness were reported in the literature. This
lull was broken in 1973 when Hawley et al., (11) reported an outbreak of illness,
apparently swimming related, caused by Coxsackie B virus. Most of the reports from
1973 to the present have dealt mainly with viral mediated swimming-related disease.
The exception to the viral etiology of swimming-associated illness was the outbreak
attributed to Shigella sonnet which occurred downstream from Dubuque, Iowa on the
Mississippi River in 1976(12). This pathogen differs from most of the bacterial species
associated with illness in swimmers in that it has a low infective dose. As few as ten
ingested Shigella have caused illness in a significant percentage of volunteers (13).
A notable characteristic of the early outbreaks and those which have occurred more
recently is the lack of good data describing the quality of the water. It is almost
characteristic of outbreaks that they do not occur coincidental to measuring the quality
of the water. This was true in the Michigan (14), Dubuque, Iowa (12), South Carolina
(15) and Perth, Australia (10) outbreaks. In all of these incidents the water quality was
usually examined before or subsequent to, but not during the outbreaks.
The Michigan outbreak, for instance, occurred one week after the water had been
analyzed and one week before the next planned sampling of the water (14). In the
Dubuque outbreak the water quality was not examined until one week after the illnesses
were first observed (12). Bryan noted that coliforms were present in the South Carolina
lake water where the Hepatitis A outbreak occurred, but densities were not reported
(15). The Australian outbreak was similar to other outbreaks in that it was only after the
onset of cases of illness that the water quality was examined and high concentrations
of fecal coliforms were observed.
Investigators in both the Vermont and Beccles, England outbreaks did not examine
the water for water quality indicators. They did, however, isolate the etiologic agent
from water samples obtained from swimming waters where patients had been bathing.
Three outbreaks apparently were associated with swimming in grossly polluted water.
The Walmer, England outbreak report clearly implicated sewage from a nearby outfall
as the source of the typhoid fever among army recruits (7). The often cited somewhat
obscure reports linking typhoid fever with bathing in polluted harbor water in New York
(16) and New Haven (17), on the other hand, did not clearly establish the association
between swimming activity and disease. The New York report indicated an unusual
increase in the number of reported cases of typhoid in the summer of 1932. The cases
were sporadic and did not constitute an outbreak. Neither water nor milk was implicated
as a common source of the etiologic agent. The report stated that, "From all the data at
hand it is very probable that most of the increase (in typhoid fever) may be charged to
bathing in polluted harbor waters condemned by the Department of Health" The report
also states that, "It should be noted that in Brooklyn, up to the age of twenty, the
infections among males are nearly double the number among females, a fact which lends
support for the belief that bathing in polluted waters has played an important part in the
increased prevalence of typhoid fever." The 1921 report by Ciampolini (17) on the incidence
of typhoid in New Haven was more detailed, but similar to the New York report in that an
excess of cases was noted in an area of the city near the harbor. A total of 32 cases was
reported from January to December and none of these could be attributed to drinking water
or food. Many of the cases were due to person-to-person contact. Only nine of the cases
were thought to be due to bathing. The nine cases lived in close proximity to the
harbor, and all had a common history of bathing at some time or other in the harbor
which had been shown some years before to be grossly polluted with sewage.
Thus in both the New York and New Haven typhoid outbreaks, swimming activity was
-------
perceived as being marginally associated with an excess of cases but not the sole cause
of the outbreak attributable to a single source, such as polluted harbor water.
In contrast to the outbreaks discussed above, the Coxsackie A16 outbreak-reported
by Denis et al (18) was quite thorough with respect to its description of the quality of
the water at the time of the incident. Not only was the etiological agent isolated from
the lake water but£. coli, group D streptococci andPseudomonas aeruginosa also were
detected and enumerated.
The nature of disease outbreaks is such that the relationship between an illness and
a common source of an etiological agent is clearly defined because of some coherent
characteristic of the affected group such as a common affiliation or activity. Thus,
disease outbreaks are instructive in the sense that they establish relationships such as
that between swimming-associated illness and water quality which probably would not
have been discovered had the members of the group not had some type of common
characteristic. Most of the above outbreaks fit this description in that the affected
individuals belonged to a group, such as army recruits or boy scouts, or they were taking
part in a common activity such as camping or picnicking. The occurrence of disease in
these groups was instrumental in showing that recreational activity in poor quality water
was a reasonable means by which pathogens could be disseminated from point sources
of pollution to susceptible individuals. However, disease outbreaks are not very useful
for establishing the relationship between the incidence of disease and some measure of
water quality because the water quality is seldom measured at the time of the outbreak.
Thus, the study of outbreaks serves a valuable purpose, but in order to establish water
quality criteria or guidelines for recreational waters, a more purposeful, directed means
of obtaining health effects information must be used. Epidemiological studies provide
a rational means for obtaining the desired information.
Retrospective Epidemiological Studies
Only three studies have been published which attempted to assess the health risks of
swimming by identifying cases of a specific illness and then determining if that illness
was somehow related to swimming activity. The first such study was reported by the
Public Health Laboratory Service (19) in 1959. They described an intensive effort over
a five-year period to identify patients with enteric fever whose illness might be
associated with a history of swimming. Attempts were made to see if the frequency of
enteric illness in coastal areas differed significantly from those in the national
population. The assessment of whether or not a case of paratyphoid or typhoid fever was
related to swimming was made on the basis of(l) the organism causing the disease being
the same type as that found in the bathing water, (2) other common sources of infection
being excluded, (3) the accidental swallowing of a good deal of water, (4) the bathing
waters being highly polluted and (5) a bathing episode prior to the onset of the illness
that was consistent with the time interval of the incubation period. Using these criteria,
no evidence could be found that seaside residents had a higher rate of enteric disease
than the nation as a whole. In all, between 1954 and 1959, only 10 cases of enteric fever
were found whose histories suggested a swimming-associated infection and of these ten
cases only four satisfactorily fit the criteria. These four cases were associated with
beaches known-to be grossly contaminated with untreated sewage. It is interesting to
note that the minimal findings of this study frequently have been interpreted to mean
that swimming in contaminated water does not pose a health risk until the quality
of the water is judged to be unsatisfactory based on aesthetic grounds. The second
retrospective study on the risks associated with swimming also was reported by
Moore (20). In this study, the case-control method was used to determine if the
swimming experiences of children with diagnosed poliomyelitis differed from those
of children the same sex and age who did not have the disease. This was
accomplished by carefully recording the swimming experiences for the three weeks
-------
prior to the onset of illness for each patient and for each control individual at the time they
were selected. One hundred and fifty matched pairs were selected during a two-year period.
The results of the study indicated that the frequency of swimming during the three weeks
preceding the onset of symptoms, in the index case, was no greater in patients than it was
in controls. The conclusion reached was that a history of swimming is not related to
contracting poliomyelitis. This study and the previous one seriously challenged the premise
that swimming in contaminated waters posed a significant health problem.
A third study of this type was conducted by D'Alessio et al. in the Madison, Wisconsin
area in 1977 (21). They examined the swimming histories of 679 well children and 216 ill
children. Enteroviruses were isolated from 119 of the ill children. Statistical analysis of these
data indicated that the risk of enteroviral disease was 3.4 times greater in children who swam
exclusively at beaches than in nonswimmers. The risk was 10.6 times greater in children
who swam exclusively at beaches if they were less than four years old. These positive
findings should be accepted with some skepticism because, as the authors point out,
swimming was not rigidly defined and, therefore, person-to-person contact cannot be entirely
ruled out as a means of transmission.
hi general, when little is known about etiologic factors, retrospective studies are useful
for discovering underlying factors associated with specific disease. This usefulness does not
extend much beyond identifying associations between specific etiologic agents and exposure
factors, which is similar to what is accomplished by examining outbreaks of disease.
Furthermore, this type of study gives no information about the incidence rates in exposed and
non-exposed individuals, both of which are critical elements for determining the importance
of certain exposure factors. Retrospective studies have been used in spite of these recognized
limitations, mainly because they are much less expensive than studies where an exposed
group is identified along with a demographically similar control group and both are followed
for a period of time to assess the proportions of a response in each group, i.e., prospective
studies.
Prospective Epidemiological Studies
The first attempts to show a relationship between swimming-associated health
effects and water quality using a prospective epidemiological study design were carried
out in the late 1940's and early 1950's by the US PHS (4). The studies were conducted
at two freshwater sites, one on Lake Michigan at Chicago, Illinois and another on the
Ohio River at Dayton, Kentucky, and at two marine sites on Long Island Sound at New
Rochelle and Mamaraneck, New York. Essentially the same experimental design was
used in each of the three studies. At each location, an attempt was made to select two
beach sites, one with a high coliform density and one with a low coliform density. At the
New York and Chicago locations beaches were available that had fairly homogeneous
populations nearby and which fit the water quality requirements of the study design. At
the Ohio River location two beaches were not available and, therefore, the population
which frequented the public swimming pool was used as the study group swimming in
water with low coliform densities. Swimming activity and the occurrence of
gastrointestinal, respiratory or "other" symptoms were recorded on a "calendar" given
to each participant at the beginning of the study. The "calendars" were collected when
the study ended. Statistically significant illness rates were determined by comparing the
observed rates in swimmers to an age-adjusted expected rate calculated using the rates
observed in the total study population. The coliform densities at the study sites were
monitored daily. Swimming-associated gastroenteritis was not observed at the marine
sites or at the Lake Michigan site. However, the Ohio River study showed that
gastrointestinal illness was observed more frequently than expected in river swimmers,
based on the illness experience of all members of the study population, when the median coliform
-------
density was about 2300 per 100 ml. This study established the first experimental link
between gastrointestinal illness in swimmers and bathing water contaminated with
fecal material. The design of the USPHS studies has been criticized on a number of
issues (22.23). The most frequent criticisms address the poor definition of swimming
activity and the fact that swimming days were not related directly to water quality
measurements. The use of the "calendar" system was also faulted because it allowed
for possible memory lapses between the swimming episode and the collection of data.
In 1972 the United States Environmental Protection Agency (EPA) began a series
epidemiological-microbiological studies at marine bathing beaches that were designed
to eliminate some of the deficiencies of the earlier USPHS studies. The objectives of
the EPA study were to determine if there was a health risk associated with swimming
in polluted marine waters and what measure of water quality best relates to
swimming-associated illness, and to develop a criterion for swimming- associated
health effects and some measure of water quality if such a relationship existed. The
EPA epidemiological-microbiological studies were conducted in New York City, New
York, Lake Pontchartrain, Louisiana and Boston, Massachusetts (6). The results of the
marine water studies indicated an excess of gastrointestinal illness occurred in
swimmers relative to a nonswimming control group in water contaminated with fecal
material and that the swimming-associated gastroenteritis was linearly related to the
water quality, as measured with a bacterial indicator. Several indicators of fecal
pollution were examined to determine which one best described the relationship
between the quality of the water and the swimming-associated health effect.
Enterococci were showri to have the strongest relationship to swimming-associated
gastroenteritis. Fecal coliforms, the currently recommended bacterial indicator of
water quality, showed no correlation to the incidence of swimming-associated
gastroenteritis. The final report of the marine recreational water quality study
concluded that water quality standards or guidelines for marine bathing beaches be
based on a criterion which describes a direct linear relationship between bathing
water quality, measured with enterococci, and the incidence of swimming-associated
gastroenteritis. The utility of the criterion was that it could be used at any level of
government to set standards or guidelines. The local regulatory body could determine
an acceptable level of risk, based on community perceptions, desires and needs, and
translate that risk level to a water quality standard or guideline indexed by the
enterococcus group. The marine studies had, in fact, established a valid
epidemiological criterion or model which had been unavailable to groups of standard
setters in the previous sixty years.
Water Quality Standards for Bathing Beaches 1924-1980
The evolution of national guidelines and standards for bathing places in the United
States began with the formation of the American Public Health Association Committee
on Bathing Places (24). One of the committee's first actions was to survey physicians
and public health officials across the United States to determine if bathing places
might be considered as important for the transmission of infections. A majority of the
replies expressed the opinion that disease and even epidemics could be attributed to
bathing beach activity. The committee report for 1924, tentatively adopted a
Bacterium coli standard for swimming pools, but did not extend the standard to
natural bathing waters (25). In 1933, the committee considered natural bathing waters
in great detail but did not adopt a bacterial standard because they did not want to
propose arbitrary standards or measures that might promote public hysteria about the
dangers of outdoor bathing places (26). The reluctance to propose bacterial standards
for outdoor bathing places was again evident in the 1936 and 1940 reports of the
committee wherein classification schemes were discussed but actions were not
taken despite pressures to do so from various quarters (27,28). The basis for this
general reluctance of the committee to propose standards for outdoor bathing
places was the paucity of epidemiological evidence linking illness to bathing in
-------
polluted water. The 1936 report noted "the committee is unconvinced that bathing places are
a major public health problem," and the 1940 report reiterated this position by stating,
"Epidemiological evidence does not appear to warrant the conclusions that bathing places
constitute a major public health problem."
The committee did consider the means of classifying natural bathing waters. The 1933
report of the committee noted that California had proposed a standard of 10 B. coli per c.c.
and that New York City had a standard of 30 B. coli per c.c. (26). The State of Connecticut
also had proposed a system of classification based on B. coli densities per 100 ml. Waters
were classified A, B, C and D. Class A. 50 B. coli per 100 ml or less, was considered very
good and Class D, more than 1000 B. coli per 100 ml, was very poor. The 1936 report stated
it was reasonable to conclude that water having a B. coli density less than 1000 per 100 ml
is probably acceptable (27). As late as 1951 this observation of the APHA Committee on
Bathing Places appeared to be still reasonable (28). Streeter (29) summarized the bathing and
recreational standards of 11 states and regions, hi 9 of the 11 standards, the limiting coliform
density was 1000 per 100 ml, either as an average or a maximum. Discher (8) listed the
current standards for all of the states in 1963 and 70% of them considered waters containing
less than 1000 coliforms per 100 ml acceptable.
The 1000 coliform per 100 ml standard first used by many of the States was not derived
from a single line of evidence. Regulatory groups and some states independently established
their standards based on available state-of-the-art information. The California standard, for
instance, was arbitrarily set by the California Bureau of Sanitary Engineering over forty years
ago (30). The standard was not based on epidemiological evidence, but rather on the
perception that it related well with the drinking water standard of that time, that there was
no epidemiological evidence of health effects within the standard, that the 10 coliforms per
ml level could easily be attained and. lastly, that any less stringent standard might result in
waters that would be aesthetically unacceptable. Connecticut, on the other hand, did not want
to set up a classification scheme that would be too arbitrary and, thus, they used a relative
scheme (31). They used coliform bacteria to index four classes of water. Class A, B, C and
D ranged from 0-50, 51-500, 501-1000 and over 1000, respectively. An extensive survey of
the Connecticut shoreline indicated that 92.8% of the samples contained less than 1000
coliforms per 100 ml (32,33,34). This classification agreed well with a sanitary survey
classification which showed that only 6.9% of the shoreline was designated as poor. The high
correlation led to the acceptance of waters having less than 1000 coliforms per 100 ml. Thus,
the standard in this case was based more on easy attainment in over 90% of the shoreline
rather than epidemiological data. Streeter (29) adopted a more analytical approach in arriving
at the 1000 coliform per 100 ml criteria. He used the coliform-Salmonella ratio developed by
Kehr and Butterfield (35), the number of bathers exposed, the approximate volume of water
ingested daily per bather and the average coliform density per ml of bathing water to develop
a bather risk factor. Streeter speculated that in water containing 1000 coliforms per 100 ml
there would be no great hazard for individual bathers, at least from Salmonella typhosa. It is
interesting to note that in spite of the use of different means for obtaining a standard measure
for water quality, either arbitrarily, practically, or analytically, the final results were
approximately the same.
The coliform index was the bacteriologic standard of choice until 1968 when the NTAC to
the Federal Water Pollution Control Administration recommended that fecal coliforms, a
subgroup of the coliform group (now designated total coliforms), be used as the bacterial
indicator of water quality (3). The recommendations of the NTAC committee were based on
prospective epidemiological studies conducted by the USPHS in 1948, 1949 and 1950 (4).
These studies had indicated that gastrointestinal illness in swimmers was significantly higher
than in a control population when coliform densities averaged 2400 per 100 ml (median) on
the Ohio River and that multiple symptomatic illness (respiratory, gastrointestinal and
-------
"other") was significantly higher in swimmers than in nonswimmers when the geometric
mean coliform density was 2300 per 100 ml at a Chicago beach. The NTAC committee used
fecal coliform and total coliform density data collected on the Ohio River in the mid-1960's
to determine that the fecal coliform subgroup was approximately 18% of the total coliform
group. The committee reasoned that if a detectable health effect was observed at a coliform
density of 2300-2400/100 ml then the recommended water quality standard should include
a factor of safety. Eighteen percent of one-half of the coliform density at which a detectable
effect occurred was arbitrarily chosen as the appropriate level and, therefore, 200 fecal
coliforms per 100 mL became the recommended standard.
The recommended fecal coliform standard has been adopted by many states and
municipalities in spite of the fact that the 1972 NAS-NAE report on Water Quality Criteria
did not recommend guidelines for recreational water because of a paucity of valid
epidemiological data (5). The NAS-NAE committee was not alone in questioning the validity
of the USPHS studies which had been summarized by Stevenson in 1953 (4). Henderson (9)
and Moore (23) have discussed the inadequacy of the epidemiological studies used to support
the NIAC recommendation.
hi 1972, the EPA recommended a recreational water quality standard similar to that
proposed by the NTAC (1). Although the Stevenson report (4) is referenced in the rationale
for the criterion, the relationship between the US PHS studies and the 200 fecal coliform per
100 ml standard was not described. Rather, the relationship between the frequency of
occurrence of Salmonella and density of fecal coliforms was emphasized. The rationale
indicated that the frequency of occurrence of Salmonella falls between 60 and 100% when
the fecal coliform density was greater than 200 per 100 ml. This recommended criterion is the
one most widely used in the United States today.
-------
SECTION 4
FRESHWATER STUDIES
The establishment of a sound relationship between swimming-associated illness and
marine water quality still left unanswered the question of whether or not this criterion could
be used in fresh water environments. The USPHS studies on the Ohio River indicated that
there was an excess of gastrointestinal illnesses among swimmers when compared to control
populations, which could properly be characterized as barely detectable (4). Furthermore,
the USPHS studies could not find a swimming-associated gastrointestinal illness effect at
Chicago beaches or at marine beaches on Long Island Sound. Since the EPA studies did
show an effect at marine bathing beaches, the expectation was that, not only would a
swimming-associated effect be found, but that the freshwater swimming-associated illness
rates might be significantly higher than the marine rates.
This report examines the data collected during studies carried out by the University of
Oklahoma, Oklahoma City, Oklahoma and Gannon University, Erie, Pennsylvania under the
auspices of the EPA. The objectives of this report were to: (1) determine if the swimming-
associated health effect/water quality criterion model established in the marine studies could
be confirmed at freshwater bathing beaches, (2) determine which indicator of water quality
shows the strongest relationship to swimming-associated health effects, if such a relationship
exists and (3) determine if the marine water quality criterion is applicable to freshwater
bathing areas.
Experimental Design
The design of the freshwater studies followed, whenever possible, the plan used in all of
the marine studies (6). The highlights will be reviewed here for the convenience of the
reader. The beach surveys or trials were conducted only on weekends to take advantage of
the large populations using the bathing beaches and to permit more intensive monitoring of
water quality during the time of swimming activity.
Swimming activity was rigidly defined as having all upper body orifices exposed to the
water. Interviewers were instructed to observe the individuals they were interviewing for
signs of complete body immersion, such as wet hair. This was not always possible and
reliance was then placed in the responses to questions about swimming activity. The
nonswimming control group was selected from beachgoers who did not meet the definition
of a swimmer.
The beach interviews were conducted in two phases, hi the first phase, trained
interviewers approached beachgoers who were about to leave the beach area and solicited
their cooperation in the study. Whenever possible, family units were sought because
information on multiple individuals could be obtained from one person, usually an adult
member of a family. During this initial contact, the following information was obtained on
each participant: sex, age, race and ethnicity, if the person swam and got their head and face
wet, length of time and time of day in the water, the illness symptoms they may have had
in the previous week, and for those who did not swim, the reason for not going into the
water. An address and telephone number was requested so that follow-up information could
be obtained. If an individual had gone swimming in the previous five days, they were not
asked to participate in the study. Telephone interviews were conducted 8 to 10 days after
the swimming experience. The eligibility of each participant was confirmed, i.e., they had
not swam in the week following the initial contact, before they were queried about the
10
-------
onset of any symptoms of illness that might have occurred during the time interval
between the swimming experience and the follow-up telephone call.
The sites for the freshwater bathing beach studies were located at Keystone Lake,
which is about 15 miles from Tulsa, Oklahoma, and on Lake Erie at Erie, Pennsylvania.
Two sites were used on Keystone Lake, one set of beaches was less than three miles
from the point of discharge of a sewage treatment facility (Beach W), and the other was
located about five miles from the outfall (Beach E). In 1979 the sewage treatment
system was two "full retention" lagoons, which discharged an average of 120,000
gallons per day of unchlorinated sewage. The following year the practice of releasing
non-disinfected sewage into the lake was discontinued. After April of 1980,
approximately 60,000 gallons per day of sewage was passed through one of the lagoons,
then through an aeration basin after which it was adequately treated with chlorine
before being discharged. Two sites also were used in the Lake Erie studies. Both sites
were located at a State Park which is situated on a peninsula just north of the City of
Erie. One beach is approximately three-quarters of a mile northwest of the outfall which
discharges the treated sewage of a large urban population (Beach B). An activated
sludge process is used to treat an average of 45 million gallons per day of sewage. The
secondary treatment effluent was chlorinated before being discharged into the lake. The
second beach is located on the opposite side of the peninsula from the effluent outfall
(Beach A). This site does not receive pollutants from a point source and the quality of
the water is usually good.
The key bacterial indicators of water quality which showed the strongest relationship
to swimming-associated illness in the marine bathing beach studies were E. coli and
enterococci. These two indicators were monitored in all phases of the freshwater
studies. Fecal coliforms, the currently accepted bacterial indicator of water quality, were
monitored in both years of the Keystone Lake Study and in two years of the Lake Erie
Study. The enterococci, an indicator group which includes two species, Streptococcus
faecalis and Streptococcus faecium, were enumerated with the method of Levin et al.
(36). E. coli was enumerated by the method of Dufour et al. (37) and fecal coliforms
were quantified according to the procedures outlined in Standard Methods for the
Examination of Water and Wastewater (38).
The data from the freshwater bathing beach studies were analyzed with respect to the
objectives of the recreational water quality research program. One of the goals of the
program was to determine whether swimming in freshwater contaminated with sewage
effluents results in a higher rate of gastrointestinal illness in swimmers relative to the
rate observed in a beach-going, nonswimming reference group. This latter group had a
tendency to be quite small at one of the study sites. The small number of nonswimmers
is a phenomenon of freshwater beaches. Unlike marine beaches, where wading and
sunning are more popular than swimming, the beach goers at freshwater beaches have
a tendency to go into the water for extended periods and to immerse their bodies totally
in the water. This greater water activity results in a much smaller nonswimming
population from which a control group can be chosen. In order to overcome this
limitation of the freshwater studies, it was necessary to pool the nonswimming control
groups from each beach within a single swimming season to form a single control
population. The homogeneity of the nonswimming control groups at the beaches of each
study location with regard to age, sex, race, and socioeconomic status lent itself to this
adjustment. The pooling of nonswimming control groups for each year increased the
probability of detecting a difference in the incidence of illness between swimmers and
nonswimmers if it does exist. The variables used to examine this relationship were the
differences in symptomatic illness rates between swimmers and nonswimmers, and the
density of bacterial indicators in the water at the time of swimming activity. Age was shown
to be a confounding risk factor in the marine bathing beach studies (6) and, therefore, this
factor was controlled in the analysis of the data. The Mantel-Haensel Chi Square
test was used to determine if something other than random processes might account for
11
-------
the observed differences in illness rates between swimmers and nonswimmers, i.e.,
exposure to contaminated bathing water (39).
A second goal was to determine if there is a direct relationship between swimming
associated gastrointestinal illness and water quality as observed in the marine bathing
beach studies. Regression analysis was used to determine if a direct relationship exists
between these variables (40).
Another goal of the study was to determine which bacterial indicator of water quality
showed the strongest relationship to swimming-associated illness. Correlation analysis
was used to measure the degree of association between gastroenteritis and the various
indicators examined. Three statistics, the correlation coefficient, the regression
coefficient and the standard error of the estimate, were used to characterize the strength
of the association.
Finally, the relationship between health effects and water quality observed at
freshwater bathing beaches was compared to the results obtained at marine bathing
beaches. This latter comparison was used to determine if the criterion developed for
marine bathing beaches is applicable to freshwater environments.
Lake Erie Study
The Lake Erie studies were conducted in 1979, 1980 and 1982 at beaches in a State
Park near Erie, Pennsylvania, in 1979 and 1980 two beaches were used, one with good
water quality and the other of excellent water quality, while in 1982 only the good
quality beach was used. Both beaches met local and state standards for recreational
waters.
The demographic characteristics of the study participants have been given elsewhere
(41). In general, the sex ratio among swimmers was about 1:1 and among nonswimmers
there was approximately twice as many females as males. These ratios were rather
constant over the three-year study period. The age distribution at the beaches also was
rather constant during the course of the studies. Among swimmers, the age group
between 1 and 19 years old made up between 43 and 55% of the population, whereas
in the nonswimmers that age group comprised approximately 23% of the population.
The racial distribution of swimmers and nonswimmers, and the socioeconomic status
of these two groups, as measured by a crowding index, was remarkably similar.
The high success rate for follow-up contacts was the result of repeated telephone calls
until the participant was reached (Table 2). The average overall success rate was 92%
during the three-year course of the study.
Table 2, Follow-up Success Rate for Beach Contacts at Lake Erie,
Pennsylvania, 1979, 1980, and 1982
1979
1980
1982
Beach A
Beach B
Beach A
Beach B
Beach B
Total Contacts 2877* 2196 3126
Follow-up inter- 2650 1858 3087
views completed
No response 227 338 39
Success rate (%) 92 85 99
Total number of 3020 2056 2907
swimmers
Total number of 1310 1039 1436
nonswimmers
2517
2493
24
99
2427
1558
8493
8211
282
97
4374
1650
"Indicates number of group contacts.
12
-------
A detailed analysis of the individual symptom rates is presented elsewhere (41). The rate
of symptoms grouped by category is given in Table 3. Gastrointestinal (GI) symptoms
include a positive response for any of the following individual symptoms vomiting, diarrhea
stomachache or nausea. The individual symptoms in the respiratory category were sore
throat, bad cough or a chest cold, and in the "other" category they were fever (greater than
100°F), headache for more than three hours, and backache, Disabling gastrointestinal
symptoms (DGI) were defined as any one gastrointestinal symptoms plus any one of the
following characteristics: stayed home due to symptoms, stayed in bed due to symptoms or
sought medical help due to symptoms. The highly credible gastrointestinal (HCGI)
symptoms are a combination of unmistakably recognized individual symptoms used to
establish the credibility of the gastrointestinal illness. HCGI symptoms are defined as any
one of the following: (1) vomiting, (2) diarrhea with a fever or disabling condition
(remained home, remained in bed or sought medical advice due to symptoms) and (3)
stomachache or nausea accompanied by a fever.
hi general, the symptom rates for swimmers were higher than those for nonswimmers, in
all the categories. However, most of the symptom rates, especially those unrelated to enteric
illness, were not statistically significant (p<0.05). This finding was similar to that observed
in the early USPHS studies (4) conducted in the 1950's and in the marine recreational water
studies conducted by the USEPA in the 1970's (6). Most of the statistically significant
differences between swimmer and nonswimmer illness rates, with one exception, occurred
in those symptomatic illness categories associated with enteric disease. Differences which
occurred in the "other" category were usually due to a fever with a temperature greater than
100° F. The significant swimming related illness rates also had a tendency to occur at the
beach with poorer quality water, Beach B. These data clearly show that there is a
swimming- associated health effect and that the effect appears to be related to the
microbiological quality of the bathing water. The illness rates by age showed a pattern
similar to that observed in the marine bathing beach studies (6), wherein the highest rates
for gastrointestinal illness occurred in children under 10 years old.
The geometric mean density and range of each of the indicators for each of the years is
given in Table 4. The indicator densities were unexpectedly low at both beaches in 1979.
hi 1980, on the other hand, the indicator densities were high and on one or two occasions,
extremely high as indicated by the range. The levels in 1982 were only moderately high
relative to those observed in 1979. The bacterial indicator densities maintained their
relative position at both beaches during the course of the study. The E. coli density was
always highest and enterococci densities were always lowest. Similarly, Beach B indicator
densities were always greater than those observed at Beach A reflecting on the nearness of
the sewage treatment plant outfall.
Keystone Lake
The Keystone Lake studies were conducted in the summers of 1979 and 1980. The
beaches were selected in a 1978 pilot study which showed the water quality of these two
sites was different based on bacterial indicator densities.
The demographic characteristics of the study participants are given elsewhere
(42, 43). The ratio of females to males was about equal in the swimmer category,
but among nonswimmers about three-fifths of the participants were females.
Similar ratios were observed in both 1979 and 1980. The socioeconomic status of
swimmers and nonswimmers in each year of the study also was very similar. As in
the Lake Erie study population, the age distribution of swimmers and nonswimmers
was nearly constant from one year to the next. Individuals under the age of 20 years
only comprised 45 to 50% of the swimmer population, whereas this age group made up 18
13
-------
r™
0
C
33
m
»*••
(D
c
E
*»-
CM
CO
*7
0>
Is-
OJ
*
m
^
u
ID
CD
.£
yu
m
2
a
CD
CS r-
00 o
O> m
"~ CD
CM
U
O
co
*~ OT
Z
£
Q
ro
(B
CD
w
W
OD
c
1
C
r-^
o
(N
I
c
O5
CO
CN
II
x: I c
u '
m ]
m <°
CD j£
J) CN
03 C
,_.
(/)
z:
~c
o
ro
QJ
CQ
OT
•^f
(N
II
C
O
CN
O
CO
C
E >,
2 o
CX OJ
E a
i. m
CO U
r-.
CD
*J"
00
m
^r
in
*t"
*
^
in
^
m
"*
*
in
in
^
•<*
^J~
CD
r->
<*
CD
t
co
in
00
00
*- m
*
f^ CD
m Tt
•r- CM
co r^
00 ^-'
* *
o> *t
CD CD
^- CN
co r-
00 «-
05 m
03 CD
CN CT)
d Tt
w in
06 oi
CN CD
d •*
m CN
CM K
— 5
O 0
Q I
;
CD CD
CD o
in m
OO
Od
V V
CL CL
CtJ flj
CO Qj
tU O3
CO tf)
to (^
03 0)
j~ c
~ ™
O P
EE|
o £ c
ui E cj?
o J2 m ^
Q^ CC3 ^Z >-
ll'tl
^ CD 73 o
§111!
T— o) *•; >— ^)
^ (D g C
03 C CT ^ H
ao . * ~
"o
-------
Table 4. Bacterial Indicator Densities at Laka Erie, Pennsylvania Bathing
Beaches, 1979, 1980, and 1982 (from Reference 41)
Bacterial Indicator
Enterococci
Year
1979
1980
1982
Beach
A
B
A
B
B
Mean
5'
13
25
71
20
Range
1-29
2-49
3-101
11-192
4-87
£
Mean
23
47
137
236
146
CO//
Range
7-268
16-413
66-536
110-950
23-524
Fecal Coliforms
Mean
37
104
60
Range
1-191
8-279
27-1O7
'Geometric mean density per 100 mL
to 20% of the nonswimmer population. In the nonswimmers, the largest age group was
the 20 to 39 year old portion which ranged from 60 to 70%. The racial characteristics of
the study populations in 1979 and 1980 were similar to each other and to that observed
in the Lake Erie studies. About 96% of all swimmers and nonswimmers over the two
years of the study were Caucasian.
The success rate for follow-up contacts are shown in Table 5. The overall success rate
was about 85% with a range of 83 to 88%. Table 5 also shows the distribution of
swimmers and nonswimmers in the total participating study population for the years 1979
and 1980. The percentage of nonswimmers in the total study population for each of the
two beaches was 15% and 13% at the W beach and 15%and 11% at the E beach for the
respective 1979 and 1980 swimming seasons. The pooling of nonswimming control
groups within years increased the average percentage of nonswimmers in the total study
population from an average of 13.7% to 24.2%.
The detailed health effects data for the Keystone Lake study trials are presented
elsewhere (42,43). The symptom rates grouped by category are shown in Table 6.
The symptoms which make up each category are the same as those defined previously
for the Lake Erie studies. The trend toward higher symptomatic illness rates in
Table 5. Follow-up Success Rate for Beach Contacts at Keystone Lake,
Oklahoma, 1979 and 1980
1979
Total contacts
Follow-up interviews.
Beach W
4242*
3610
Beach E
3457
2859
1980
Beach W
6616
5849
Beach E
4673
3981
completed
No response or uncooperative
Success rate (%)
Total number of swimmers
Total number of
nonswimmers
632
85
3059
551
598
83
2440
419
767
88
5121
774
591
85
3562
437
*Each contact one individual.
15
-------
o
co
en
Ir~
-o
c
a
03
P**
o>
.
in
to
•C
o
«Q
m
jt
3
>
O
« «
fQ
O *%j
>•*
U «
0) U
" 3;
fit |g
£15
|E
C o
C/5 !J1
*0
(O
yj
x:
V)
z
o
0)
00
o
CO
*"~
5*
JC
o
en
til
-C
U
<0
0)
CD
0>
s>
»~
5
o
V
aa
C/)
W)
z
CO
II
c
,_
CM
IT
c
CM
m
II
c
O
cn
II
c
O
•<*
CM
i II
tf)
u>
C
O
O)
II
c
CD
in
o
CT
||
C
E >
2 o
CX O)
E £
>- ro
W U
en CN T— en ai
«— CO CM
r- T- CM
r» •- CM o T-
ro m co •— »—
CM in -r- co
O> CM -r- O) 00
^ «
co r^ o> »— ro
en tt CM «- «—
m m
CM Tt m r^ m
in co in «- •—
CD
r~ o 10 m 10
in r» m — —
u> .n
04 ^t oo r*» to
ID co ur> •— •—
^O CO
T— ^ -f~~ (^2 ^J
~
c 2
-— Ct3
2 t ,_ "
*-• ex tt> *'j fn
in tn .- — l~'
Si -^
y) 03 c:
, — -o
0) O X(/>
16
-------
swimmers, relative to nonswimmers, observed in the Lake Erie studies also was evident
in the Keystone Lake studies. The only exception to the trend occurred at Beach E in
1979 where the respiratory symptom rate in nonswimmers apparently was greater than
that for swimmers although the difference was not statistically significant. In 1979
there was only one symptom category where the difference in illness rates between
swimmers and nonswimmers was shown to be statistically different, and that occurred
in the "other" category. Conversely, in 1980 statistically significant differences in
illness rates between swimmers and nonswimmers were observed in three categories:
GI, respiratory and "other" at Beach W, and in the GI and "other" categories at Beach
E. The failure to find swimmer-nonswimmer differences in the highly credible GI
category, in spite of the fact that statistically significant differences were found in the
GI and "other"' categories, was not unexpected since this was observed on a number of
occasions in the marine studies (6).
The bacterial indicator densities observed during the 1979 swimming season were
consistent between indicators (Table 7). Indicator densities at the beach nearest the
source of the pollution were always higher than those at the more distant beach. The
1980 data, however, do not reflect such constancy. The enterococci and fecal coliform
densities are not different between beaches as would be expected and the E. coli
densities appear to be higher at the beach more distant from the pollution source. These
inconsistent results may have been caused by heavy rains which occurred in the four
days before the start of the beach study trials. In that short four-day period, 8.15 inches
of rain was measured. This, in turn, caused the lake elevation to rise approximately
nine feet above its normal level. The lake elevation did not return to its normal level
until July 18, about a month after the heavy rains. The turbidity of the water also was
increased during this time period. The effect these unusual events might have had on
the swimmer illness rates is unknown.
Table 7. Indicator Densities at Keystone Lake Bathing Beaches,
1979-1980 (from References 42, 43)
Entorococci
Year
1979
1980
Beach Mean1
W
E
W
E
38. 82
6,8
23
20
Range
17-180
2-98
6-64
2-76
E.
Mean
138
19
52
71
coli
Range
30-300
1-44
14-200
12-215
Fecal Coliform
Mean
436
51
230
234
Range
200-920
NG3
58-1300
47-1600
'Geometric mean.
2Density per 100 ml.
3Not given.
17
-------
SECTION 5
DEVELOPMENT OF A CRITERION
The development of a criterion which relates swimming-associated health effects to
some measure of water quality requires that certain basic information must be available
before a valid model can be established. One of the critical elements in such a model is
to show that there is a significant excess of illness in bathers who swim in surface waters
contaminated with domestic wastewater. The USPHS studies on the Ohio River and at
other locations were an attempt to reach this objective (4).
Those studies did show that there was a barely detectable health effect when the
bathing water contained about 2,300 coliforms per 100 ml, an indication that the water
was contaminated with fecal material from humans or warm-blooded animals. The effect
was shown only for symptomatic gastrointestinal illness and not for the respiratory or
"other" categories of symptoms.
The more recent EPA marine bathing beach studies also showed, in unquestionable
terms, that one of the main health effects related to swimming in sewage-polluted water
was gastrointestinal illness (6). Although increased rates of respiratory and
"other" symptomatic illness were related to swimming activity alone, only the rate of
gastrointestinal illness increased significantly as the quality of the bathing water
decreased. The EPA marine studies also accomplished what the early USPHS studies
could not accomplish, that is, show that there is a risk of enteric illness due to swimming
in polluted marine waters.
The studies described in this report clearly confirm that the risk of contracting
gastrointestinal illness greatly increases if a person swims in water contaminated with
human domestic wastes. The Lake Erie bathing beach trials showed that almost all of the
statistically significant differences in swimming-associated illness rates occurred only
in those symptom categories related to gastroenteric illness and that there was a greater
preponderance of significant differences at those beaches having the highest degree of
fecal contamination. The results of the Keystone Lake study were not as clear-cut as
those at the Lake Erie beaches, but the observed statistically significant swimming-
associated illness rates were related mainly to gastrointestinal symptomatology and
"other" symptoms such as fever greater than 100°F. Two single exceptions to these
findings were the swimmer-nonswimmer rate differences in the respiratory category
which occurred at Lake Erie and Keystone Lake beaches in 1980.
The second objective of this report was to determine which indicator of fecal
contamination, enterococci, E. coli or fecal coliforms showed the strongest relationship
to swimming-associated illness. Enterococci and£. coli were considered because they
showed the " best" relationship to swimming-associated illness in the marine recreational
water quality studies. Enterococci were judged to be superior toE. coli for use in marine
waters (44). Fecal coliforms also were examined in the freshwater studies because they
are the currently recommended indicator group for monitoring recreational water quality.
Three statistics related to regression and correlation analysis were used to
determine which bacterial indicator had the strongest relationship to swimming-
associated illness in freshwater environments. They are the slope of the regression
equation, the standard error of the estimate and the correlation coefficient. In the
marine bathing beach studies, only the correlation coefficient was used to compare
the "strength of association" of various indicator bacteria to swimming-associated
illness. Since the number of paired data points available in the freshwater studies is
small, it seemed appropriate to use ancillary descriptive information to arrive at a
18
-------
judgment as to which indicator should be used to establish the relationship between
water quality and swimming-associated health effects. The slope was chosen because
it indicates how large a change in a health effect will be associated with a change in
water quality. It also has the advantage of being amenable to statistical testing to
determine if it is different from zero or some other slope. The standard error of the
estimate is useful because it is an average measure of the vertical distances of the
observed points from the regression line. It is defined as the square root of the sum of
the squared vertical distances from the regression line divided by the number of points.
Thus, the smaller this value, the closer the points are to the regression line. The
correlation coefficient is included so that the results of the freshwater data can be
directly compared to the marine data. Since the correlation coefficient can be affected
by the magnitude of the slope and the standard error of the estimate, but gives no
indication of the relative influence of these two components, less weight will be
attached to this statistic relative to the other two.
Table 9 is a summary of the statistics used to describe the relationship of
gastrointestinal illness to enterococci, E. coli and fecal coliforms. The statistics were
generated using the summary data given in Table 8. The first conclusion that can be
drawn from the data in this table is that fecal coliform densities in freshwater are
unrelated to swimming-associated gastroenteritis. The slope of the regression lines of
highly credible and total GI symptomatic illness on fecal coliform densities were not
significantly different from zero. This finding was very similar to that reported for the
relationship of HCGI and total GI symptoms to fecal coliform densities in studies at
marine bathing beaches (6). In those studies, correlation-coefficients for HCGI and
total GI symptoms on fecal coliform densities for data analyzed by summer and by-
beach were respectively -0.01 and 0.01. The implication of these results is quite clear.
Bacteria from sources other than the gastrointestinal tract of man and other warm-
blooded animals, which fit the definition of fecal coliform given in Standard Methods
for the Examination of Water and Wastewater (38), are present at densities high
enough to sufficiently eliminate the usefulness of fecal coliforms as an indicator
Table 8, Summary of Microbiological and Epidemiological Results from
Lake Erie and Keystone Lake Bathing Beach Studies
Geometric Mean
Location
Erie
Erie
Erie
Keystone
Keystone
Year
1979
1980
1982
1979
1980
Beach
A
B
A
B
B
W
E
W
E
Entero-
cocci •
5.22
13
25
71
20
38.8
6.8
23
20
E. coli
23
47
137
236
146
138
19
52
71
Indicator
Density1
Fecal
Coliforms
N.D.
N.D.
37
104
60
436
51
230
234
Symptom Rate
Total GI
9.93
11.7
9.6"
30.0*
11.6
9.0
5,0
17.7*
18.9*
HCGI
2.3
4.6
4.8
14.7*
11*
5.1
0.5
5.2
3.0
'Obtained from trials grouped by beach and year.
indicator density per 100 mL
3Swimmer, nonswimmer illness rate difference per 1000 individuals.
*Swimmers illness rate significantly different from nonswimmer illness rate
at P < 0.05 level.
19
-------
of fecal contamination of surface waters. This hypothesis is supported by the
numerous reports that at least one genus within the fecal coliform group can readily
grow to high densities in the presence of industrial wastewaters. Klebsiella
pneumonia and other Klebsiella species grow to extremely high densities in pulp
mill wastes (45,46), textile processing plant wastes (47) and cotton mill wastes
(48). Industrial wastes are not the only source of thermotolerant Klebsiella. The
proportion of Klebsiella in fecal coliform populations observed in secondary
effluent samples from seven sewage treatment plants has been found to range from
13 to 42% (unpublished data). Furthermore, this genus was shown by Kinney,
Drummond and Hanes (49) to be much more resistant to chlorine than other genera
of the fecal coliform group. This latter observation might account for the finding
that more than one-half of 24 water samples collected over a 15-day period from
Beach B on Lake Erie had fecal coliform populations that were more than 30%
thermotolerant Klebsiella. The percentage of Klebsiella ranged from 17 to 73% of
the fecal coliforms. This ubiquitous organism, many strains of which fit the fecal
coliform definition, may very well be a partial reason why fecal coliform densities
do not show a direct relationship to swimming-associated GI illness.
Eschericia coli densities on the other hand show an excellent relationship to
swimming-associated GI illness. SinceE. coli is by definition a fecal coliform, this
strong association to GI illness can only be attributed to the use of a highly
selective differential enumeration method which effectively eliminates potential
interfering organisms. The slopes of regression lines calculated using illness rates
from the HCGI and total GI categories against E. coli densities both showed a
statistically significant change in illness rates with changes in indicator densities.
The standard error of the estimate associated with HCGI symptom rates was the
smallest of all the estimates, indicating that this indicator had the closest fit of
points to the regression equation. The correlation coefficient for the association
between HCGI symptoms and E. coli densities was 0.804, the largest of all the
correlation coefficients and indicating thatE. coli shows the "best" relationship to
swimming-associated GI illness. This rationale, which was used in the marine
recreational water quality studies to choose the "best" indicator, may not he
applicable to the freshwater studies. This is suggested because of the equally
excellent relationship between GI illness rates and enterococci densities. The
slopes for the two symptom categories generated using enterococci as the
independent variable are very similar to those obtained using E. coli.
If a statistical significance test is performed to test the hypothesis that the slope of
the regression line of HCGI illness rates on enterococci densities equals the slope of the
regression line of HCGI illness rates on E. coli densities, it can be shown that the
Table 9. Summary of Regression Statistics Related to Swimming-Associ-
ated Illness and Water Quality Indicators
Indicator
Enterococci
f, coli
Fecal Coliform
Symptom
Category
Total GI
HCGI
Total Gi
HCGI
Total GI
HCGI
Slope
14.30*
9.40
10.39*
9.40*
5.21
-0.98
Y
Intercept
-4.50
-6.28
-5.56
-11.74
3.81
8.35
Std. Error
of Estimate
5.21
2.97
5.97
2.49
7.53
4.49
Correlation
Coefficient
.673
.744
.528
.804
.249
-.081
Departure
from
Linearity
NS1
NS
NS
NS
IMS
NS
'NS - No significant departure from linearity.
*Slope of regression line is significantly different from zero.
20
-------
slopes are not significantly different (p >0.05). Furthermore, it cannot be shown
that the y-intercepts or the correlation coefficients associated with these two
indicators are significantly different (p >0.05). Since the two indicators show
equally strong associations to swimming-associated gastrointestinal illness, it
seems that an appropriate approach would be to present two criteria and
recommend one of the two based on factors unrelated to the statistics of regression
and correlation analysis.
Criteria for Freshwater Bathing Areas
Cabelli (6) has defined a water quality criterion developed for use with indicator
systems as "a quantifiable relationship between the density of the indicator in the
water and the potential human health risks involved in the water's use." The health
effects-water quality criterion that will be developed in this section fits the above
definition, and its purpose will be to provide a quantifiable relationship which can
be used to set water quality guidelines or standards for fresh bathing waters. The
regression lines which characterize the relationship between highly credible and
total gastrointestinal symptom rates, and bacterial indicator densities were
developed from the regression coefficients given in Table 9. It was assumed that
the logarithm of the bacterial density would graph linearly against the incidence of
disease since this model had previously been shown to be applicable to similar data
in the marine bathing beach studies (6). The validity of the linear relationship was
examined using the run test (50), a crude but simple method for determining
departures from linearity. None of the estimated lines showed statistically
significant departures from linearity (p >0.05). The regression lines for highly
credible and total gastrointestinal symptom rates on indicator densities are shown
for E. coli, enterococci and fecal coliforms in Figures 1, 2 and 3. Each point
represents a pair of variables, the geometric mean indicator density obtained from
water samples collected at a beach over a single bathing season and the
corresponding swimming-associated illness rate. The gastrointestinal illness rate
and fecal coliform density data displayed in Figure 3 serve to emphasize the lack
of association between these two variables. This lack of association was indicated
in Table 9 by the low value of the correlation coefficient and the flatness of the
slope. The regression lines for E. coli and enterococci, on the other hand, show
significant changes in the symptomatic illness rates with changes in indicator
densities. Furthermore, all of the observed data points are in close proximity to the
estimated regression lines, especially those for the highly credible symptoms. The
regression lines for E. coli and enterococci are remarkably similar with respect to
slope, standard error of the estimate and correlation coefficient. The only
differences of any significance are the higher densities of E. coli relative to
enterococci as manifested by the greater y-intercepts associated with the E. coli
regression lines, especially in the case of the HCGI symptoms. The average highly
credible GI symptom rates were about 43% of the average rates observed for total
GI symptoms. The overall mean enterococci density for the nine trials was 18.9 per
100 ml, while that for E. coli was 71.9 per 100 ml. These results were not
unexpected, since enterococci are typically found at densities lower than E. coli,
both in human feces (51) and in sewage effluents (52).
The strength of the relationship between health effects and the various water
quality indicators examined in the marine recreational water quality study showed
clear-cut differences and, therefore, the choice of the "best" indicator was obvious.
The selection of the "best" indicator with respect to the strength of the relationship
between water quality indicator and swimming-associated illness is not obvious in
the results of the freshwater studies. The similarity of the data describing the
relationship of swimming-associated illness toE. coli and enterococci densities is
so great that a criterion will be presented for each bacterial indicator. The health
effects-water quality criteria for E. coli and enterococci are shown in Figures 4 and
5. Each figure shows the estimated lines of best fit and the 95% confidence limits
of the lines.
21
-------
30-
ra
c «
S g
as E
= I
°§ 5
is
-------
c 2
« c
CD E
I i
e's
S o
E a
Si
U)
30
20
10
HCGI
C
Total Qi
A
A
A
A
Erie 1980
Erie 1982
Keystone 1979
Keystone 1980
10
Figure 3,
100
Mean fecal coliform density per 100 mL
1000
Estimated regression lines for highly credible and total gastrointestinal
symptom rates on fecal coliform densities.
30
>
if
s's
s ™
03 O
0)O
•O
.
s«
20
. o
E
E E
o 10
I I I I |
I IT
Regression equation:
¥=-11.74 + 9.397 (log x)
10
100
Mean £. coli density per 100 mL
1000
Figure 4.
Criterion for estimating swimming-associated gastrointestinal illness rate
from the geometric mean density of E. coli per 100 mL in freshwater
samples.
23
-------
w
o E
c E
25
_
o a
S2
20
15
10
y)
ro
c E
•— 2
E a
E E
•is
CO
Regression equation:
y=-6.278 + 9.40(log x)
10 30
Mean enlerococcus density per 100 ml_
100
Figures. Criterion for estimating swimming-associated gastrointestinal illness rate
from geometric mean density of enterococci per 100 ml_ in freshwater
samples.
Although data for both highly credible and total GI symptomatic illness have been
shown, only the former will be used to develop criteria for fresh recreational bathing
waters. The reason for not considering a criterion using total GI symptomatology is two-
fold. First, the regression and correlation analyses indicate that the strength of the
relationship between the indicators and highly credible symptom rates is much greater
than that with total GI symptom rates, and second, as pointed out by Cabelli (6), highly
credible symptoms should be used "because of the greater credibility of its data base
and because it is more conducive to economic analysis."
The two figures shown are useful for approximately determining the number of
swimming-associated gastrointestinal illnesses that might be expected at a bathing
beach where the density ofE. coli or enterococci falls within the range of the criterion.
However, a relationship of this type most likely will be used to determine, not what the
risk is but what the water quality should be after an acceptable risk level has been
agreed upon by a local or state authority. Since the two characteristics (indicator density
and Illness rate) used to develop the criterion both show variability and only the
variability of the dependent characteristic is accounted for in the regression equation
and its 95% confidence limits, it is necessary to show a second relationship where the
indicator densities play the role of the dependent characteristic. The regression line for
E. coli on swimming-associated illness. The equation of the line and the 95%
confidence limits of the line are shown in Figure 6. A similar equation, line and limits
for enterococci are shown in Figure 7. Either one of the relationships shown in Figures
6 and 7 can be used to establish guidelines or standards based on acceptable risk.
24
-------
10,000
o
o
Regression equation:
log y = 1.464 + 0 0687x
1000
100
10
10
15
20
Swimming-associated gastrointestinal
symptom rate per 1000 swimmers
FigureB.
Criterion for estimating the geometric meanf, co//density per lOOmLfrom
an acceptable risk level of swimming-associated gastrointestinal iltness.
25
-------
1000
Regression equation:
log y = 0.938 + 0.059x
100
c
-------
SECTION 6
MARINE VERSUS FRESHWATER CRITERIA
The recreational water quality studies carried out by the USPHS in the early 1950's (4)
could not detect a swimming-associated illness effect at two marine bathing beaches on Long
Island, New York. Since swimming-associated gastrointestinal health effects were observed
at a freshwater bathing site in the same study series, it was assumed that the results of the
EPA fresh recreational water quality studies would reveal higher swimming-associated
gastrointestinal illness rates than were found in the EPA marine bathing beach studies. This
assumption posed a further question as to whether or not a single criterion could be used for
both fresh and marine bathing beach waters. When the results of the freshwater studies were
compared to those of the marine studies (Figure 8), it was clear that the illness rates in
bathers swimming at marine bathing beaches were significantly higher (p <0.05) than those
in freshwater swimmers when the data were analyzed using the Wilcoxon rank sum test
(53). The mean of the marine highly credible GI illness rate data (15.2 per 1000), grouped
by beach and year, was 2.67 times greater than the mean for the highly credible GI illness
rates in freshwater swimmers (5.7 per 1000). The Wilcoxon rank sum test (53) also was
used to show that the means for the E.coli and enterococci indicator densities in marine
waters were not significantly different from the means of those two indicators in fresh waters
(Figure 8). The similarity in indicator densities in freshwater and seawater can be explained
on the basis of the limitations placed on the selection of study sites and the difference in
gastrointestinal illness rates between marine and freshwater swimmers can possibly be
accounted for by the die-off rates of indicator bacteria and pathogens in marine and
freshwaters.
The constraints of the site selection process stipulated that the water quality at each
location had to meet local standards. All of the sites where studies were conducted used the
200 fecal coliform per 100 ml standard and since the bathing waters were usually in
compliance with the standard, it is not surprising that there is some uniformity of indicator
densities in the marine and fresh beach waters.
The survival of coliforms in seawater and freshwater was examined by Chamberlin and
Mitchell (54). They analyzed 87 seawater studies and 28 freshwater studies on indicator
bacteria die-off The results of their analysis indicated that the median T90 value, i.e., the
time it takes for 90% of the indicator bacteria to die-off, from the seawater studies was 2.2
hours, whereas the mean T90 value from the freshwater studies was 57.6 hours. Hanes and
Fragala (55) have shown that under laboratory conditions enterococci and E. coli behave
much like coliforms do under field conditions. E. coli had a T90of 18 hours in seawater and
110 hours in freshwater, while enterococci had a T90 of 47 hours in seawater and 71 hours
in freshwater. The differential die-off of indicators by itself, however, is not sufficient to
explain the difference in gastrointestinal illness rates between marine and freshwater
swimmers. The linear relationships between gastroenteritis and E. coli or enterococci
in marine and fresh recreational waters shown in Figures 9 and 10 provide additional
information which might be useful for answering the question. The coefficients used to
generate the regression lines in Figures 9 and 10 are given in Table 10. It is
noteworthy, that with the exception of the regression line showing the relationship
between swimming-associated illness in marine waters and£. coli densities, all of the
estimated lines in Figures 9 and 10 intersect the indicator density axis at mean
indicator densities greater than 1 per 100 ml. This suggests that the infectious dose
level of the etiologic agent disappears before the mean indicator densities become
27
-------
30
li
C *s
1 1
c 6
g'|
05
"O
<» *~
5 « 20
1«
« m
85 i_
?|
'c *~*
E E
'5 w
w
10
I i
Arithmetic
mean , ,
o
o
o
8
-
c
•
ft
_
°0 "*i»
»
•
X »
Marine Fresh-
III!
Geometric
mean -*-
0 —
„
0 0
»
O *
8 1
0 § ~
• -*. »
o
"^0 »
•
o —
"*" o i ° «
o -*-*• o •
X X
V X
0 *
0 0 —
o
§ •
•
0
1 1 1 I
3.5
3,0
E
2.5 8
*-
i_
tn
a.
.§•
w
2.0 §
-a
2
£o
u
T)
1.5 .£
?
_J
1.0
0.5
Marine Fresh- Marine Fresh-
water water water water water -water
HCGI Symptoms
Enter ocoeci
E. col/
Figure 8.
Data summary of highly credible gastrointestinal symptom rates and
indicator densities from marine and freshwater studies. (Marine data
obtained from Tables 7 and 8, Reference 6.)
immeasurable in both marine and freshwater environments. The implication of this observation
is that the etiologic agent, which is assumed to be a virus (56), probably dies off at the same rate,
whether in seawater or freshwater, since the swimming-associated effect of the pathogen
infectious dose approaches zero as the indicator density approaches a value of 1 per 100 ml. This
would not be an unreasonable assumption, since Cioglio and Loddo (57) have shown that strains
of Polio, ECHO and Coxsackie virus had similar die-off rates in river and seawater held at 25°C.
Although it is unlikely that these viruses are the cause of swimming-associated gastroenteritis, it
is possible that the unidentified etiologic agent behaves in a similar fashion. The assumed similar
die-off rate of the pathogen in freshwater and seawater, coupled with the greater die-off of
indicator bacteria in seawater than in freshwater, could account for the difference in gastroenteritis
rates between marine and freshwater swimmers, especially when both types of recreational waters
are required to meet the same microbial standard. Thus, an indicator would decay rapidly in
28
-------
0)
E
cfl
O
O
O
I
30
20
10
Figure 9.
Marine Wafer
10 100
Mean enterococcus density per 100 mL
1000
Marine and freshwater criteria for swimming-associated gastrointestinal
illness and water quality using enterococci to measure the water quality.
(Marine data obtained from Table 7, Reference 6.)
Table 10. Summary of Statistics Related to Marine and Freshwater
Criteria for Highly Credible Swimming-Associated Illness
and Water Quality Indicators, f. co//and Enterococci
Type of
Water
Marine3
Fresh
Mean
Swimming-
Associated
Illness Rate
15.2
5.7
Geometric
Mean
Density
EC' ENT2
56 25
72 20
___JS|op_e_______
EC ENT
7.3 11.6
9.4 9.4
Standard
Error Est.
EC ENT
8.5 6.7
2.5 2.8
Correlation
Coefficient
EC ENT
.512 .712
.804 .744
1EC - £. coli.
2ENT - enterococci
3ll!ness rates and bacterial indicator density data obtained from Reference 6.
29
-------
Figure 10.
10 100 1000
Mean E. coli density per 100 mL
Marine and freshwater criteriaforswirnming-associatedgastrointestinal
illness and water quality using E. coli to measure the water quality- (Marine
data obtained from Table 9, Reference 6.)
seawater while the pathogen does not, leaving an excess of pathogen once the standard
is reached. In freshwater, the indicator decays at the same rate or slower than the
pathogen, which results in a low density of pathogen by the time the standard is
attained. At equivalent indicator densities, there will be an excess of pathogen in
marine waters relative to what would be found in freshwaters. and therefore a higher
illness rate will be observed in marine waters. Thus, the difference in marine and
freshwater swimmer illness rates is not only statistically significant, but also is
apparently compatible with many of the known characteristics of indicators and
pathogens associated with the observed phenomenon. The significance of these findings
is that a single water quality criterion for seawater and freshwater has been effectively
eliminated from consideration, and therefore a separate criterion should be used for
each type of bathing water.
30
-------
REFERENCES
1. U.S. Environmental Protection Agency. Quality Criteria for Water. U.S.E.P.A.,
Washington, DC. 1976.
2. Cabelli, V. I, A. P. Dufour, L. J. McCabe, and M. A. Levin. 1983. A Marine
Rereational Water Quality Criterion Consistent with Indicator Concepts and Risk
Analysis. J. Water Poll. Control Fed. 55:1306.
3. National Technical Advisory Committee. 1968. Water Quality Criteria. Federal Water
Poll. Control Adm., Dept. of the Interior, Washington, DC.
4. Stevenson, A. H. 1953. Studies of Bathing Water Quality and Health. Am. J. Public
Hlth. Assoc. 43:529.
5. Committee on Water Quality Criteria. National Academy of Sciences-National Academy
of Engineering. Water Quality Criteria. U.S. Environmental Protection Agency,
EPA-R3-73-033. Washington. DC. 1972.
6. Cabelli, V. J. 1981. Health Effects Criteria for Marine Recreational Waters. U.S.
Environmental Protection Agency, EPA-600/ 1-80-031. Cincinnati. Ohio.
7. Moore. B. 1954. Sewage Contamination of Coastal Bathing Waters. Bull. Hyg. 29:689.
8. Discher, D. M. 1963. The Scientific Basis for Coliform Organisms as an Index of
Bathing Water Safety. In: Coliform Standards for Recreational Waters, Appendix I,
Progress Report, Public Health Activities Committee. J. San. Eng. Div., ASCE. p.89.
9. Henderson, J. M. 1968. Enteric Disease Criteria for Recreational Waters. I. San. Eng.
Div. 94:1253.
10. Anonymous. 1961. Typhoid Traced to Bathing at a Polluted Beach. Public Works.
92:182.
11. Hawley, H. B., D. P. Morin, M. E. Geraghty, J. Tomkow, and C. A. Phillips. 1973.
Coxsackievirus B Epidemic at a Boys' Summer Camp. Isolation of Virus from
Swimming Water. JAMA. 226:33.
12. Rosenberg, M. A., K. K. Hazlet. J. Schaefer, J. G. Wells, and R. C. Pruneda. 1976.
Shigellosis from Swimming. JAMA. 236:1849.
13. Dupont. H. L., and R. B. Hornick. 1973. Clinical Approach to Infectious Diarrheas.
Medicine. 52:265.
14. Koopman,J. S., E. A. Eckert, H. B. Greenberg, B. C. Strohm, R.E. Isaacson,and A. S.
Monto. 1982. Norwalk Virus Enteric Illness Acquired by Swimming Exposure. Am. J.
Epidemiol. 115:173.
15. Bryan, J. A., J. D. Lehmann. I. F. Setiady, and M. H. Hatch.1974. An Outbreak of
Hepatitis-A Associated with Recreational Lake Water. Am. J. Epidemiol. 99:145.
16. Typhoid Fever from Bathing in Polluted Waters. 1932. City of New York Department
of Health Weekly Bulletin. 21:257.
17. Ciampolini, E. 1921. A Study of the Typhoid Fever Incidence in the Health Center
District of New Haven. Report to City of New Haven.
18. Denis. F. A., E. Blanchovin, A. DeLignieres, and P. Flamen. 1974. CoxsackieA16
Infection from Lake Water. JAMA. 228:1370.
19. Committee on Bathing Beach Contamination of the Public Health Laboratory Service.
1959. Sewage Contamination of Coastal Bathing Waters in England and Wales. J.
Hygiene. 57:435.
20. Moore, B. 1959. The Risk of Infection Through Bathing in Sewage-Polluted
Water. In: Proceedings of the First International Conference on Waste
31
-------
Disposal in the Marine Environment. Ed. E. A. Pearson. Pergamon Press, New York.
21. D'Alessio, D. I, T. E. Minor, C. I. Allen, A. A. Tsiatis, and D. B. Nelson. 1981. A Study
of the Proportions of Swimmers Among Well Controls and Children with
Enterovirus-Like Illness Shedding or Not Shedding an Enterovirus. Am. J. Epidemiol.
113:533.
22. Cabelli, V. I, M. A. Levin, A. P. Dufour, andL. J. McCabe.1975. The Development of
Criteria for Recreational Waters. In: International Symposium on Discharge of Sewage
from Sea Outfalls. Ed. A. L. H. Gameson. Pergamon Press, London, p.63.
23. Moore. B. 1975. The Case Against Microbial Standards for Bathing Beaches, hi:
International Symposium on Discharge of Sewage from Sea Outfalls. Ed. A. L. H.
Gameson. Pergamon Press, London, p. 103.
24. Report of the Committee on Bathing Places. 1922. Am. J. Public Hlth. 12:121.
25. Report of the Committee on Bathing Places. 1924. Am. J. Public Hlth. 14:597.
26. Swimming Pools and Bathing Places. 1933. Am. 3. Public Hlth. 23:40.
27. Bathing Places. 1936. Am. 3. of Public Hlth. 26:209.
28. Looking Forward in the Bathing Place Sanitation Field. 1940. Am. J. of Public Hlth.
30:50.
29. Streeter, H. W. 1951. Bacterial-Quality Objectives for the Ohio River. Ohio River Valley
Water Sanitation Commission, Cincinnati, Ohio.
30. Report on a Pollution Survey of Santa Monica Bay Beaches in 1942. 1943. Bureau of
Sanitary Engineering. State of California.
31. Sherman, Leslie K. 1960. Connecticut Studies Its Shore Bathing Waters. Connecticut
Hlth. Bull. 75:129.
32. Scott, W. J. 1932. Survey of Connecticut's Shore Bathing Waters. Public Hlth. Engr.
19:316.
33. Scott, W. J. 1951. Sanitary Study of Shore Bathing Waters. Connecticut Hlth. Bull.
65:74.
34. Scott, W. J. 1953. Recent Sanitary Study of Shore Bathing Waters.Connecticut Hlth.
Bull. 67:139.
35. Kehr, R. W., and C. T. Butterfield. 1943. Notes on the Relation Between Coliforms and
Enteric Pathogens. Public Health Reports. 58:589.
36. Levin, M. A., J. R. Fischer, and V.J. Cabelli. 1975. Membrane Filter Technique for
Enumeration of Enterococci in Marine Waters. Appl. Microbiol. 30:66.
37. Dufour, A. P., E. R. Strickland, and V.J. Cabelli. 1981. Membrane Filter Method for
Enumerating Escherichia coli. Appl. Environ. Microbiol. 41:1152.
38. American Public Health Association. Standard Methods for the Examination of Water
and Wastewater. 14th Ed. Washington, DC. 1975.
39. Kleinbaum, D. G.. L. L. Kupper and H. Morgenstern. 1982. Epidemiologic Research.
Lifetime Learning Publications, Belmont, California.
40. England, J. M. 1975. Medical Research - A Statistical and Epidemiological Approach.
C. Livingstone, New York.
41. Dufour, A., S. Zagorski, R. Gammon, and G. Kraus. Swimming-Associated
Gastroenteritis in Freshwater Swimmers, hi preparation.
42. McKee, G. L. 1980. Development of Health Effects Criteria for Freshwater Bathing
Beaches by Use of Microbial Indicators. Ph.D. Thesis. University of Oklahoma, Norman.
Oklahoma.
43. Shadid, D. L. 1981. Microbial Indices of Recreational Water Quality. Ph.D. Thesis.
University of Oklahoma, Norman, Oklahoma.
44. Cabelli, V. J., A. P. Dufour, L. J. McCabe, and M. A. Levin. 1982.
Swimming-Associated Gastroenteritis and Water Quality. Am. J. Epidemiol. 115:606.
45. Bordner, R. H.. andB. J. Carroll. 1972. Proceedings of the Seminar on the Significance
of Fecal Coliform in Industrial Wastes. U.S. Environmental Protection Agency, Denver,
32
-------
46. Caplenas, N. R., M. S. Kanarek, and A. P. Dufour. 1981. Source and Extent of
Klebsiellapneumonias in the Paper Industry. Appl. Environ. Microbiol.42:779.
47. Dufour, A. P. and V. J. Cabelli. 1976. Characteristics of Kiebsiella from Textile
Finishing Plant Effluents. J. Water Poll. Control Fed. 48:872.
48. Campbell. L. M., G. Michaels. R. D. Klein, and I. L. Roth.1976. Isolation of Klebsiella
pneumoniae from Lake Water. Can: J. Microbiol. 22:1762.
49. Kinney, E. C., P. W. Drummond, and N. B. Hanes. 1978. Effects of Chlorination on
Differentiated Coliform Groups. J. Water Poll. Control Fed. 50:2307.
50. Crow, E. L., F. A. Davis, and M. W. Maxfield. 1960. Statistics Manual. Dover
Publications. Inc.. New York.
51. Slanetz, LW., and C. H. Bartley. 1957. Numbers ofEnterococci in Water. Sewage, and
Feces Determined by the Membrane Filter Technique with an Improved Medium. J.
Bact. 74:591.
52. Miescier, J. J., and V. J. Cabelli, 1982. Enterococci and Other Microbial Indicators in
Municipal Wastewater Effluents. J. Water Poll. Control Fed. 54:1599.
53. Wilcoxon, F., andR. A. Wilcox. 1964. Some Rapid Approximate Statistical Procedures.
Lederle Laboratories. Pearl River, New York.
54. Chamberlin, C. E., and R. Mitchell. 1978. A Decay Model for Enteric Bacteria in
Natural Waters, hi: Water Pollution Microbiology. Ed. Ralph Mitchell, John Wiley &
Sons, New York. Vol. 2. p. 325.
55. Hanes, N. B.,andR. Fragala. 1967. Effect of Seawater Concentration on the Survival of
Indicator Bacteria. J. Water Poll. Control Fed. 39:97.
56. Cabelli, V. J. 1981. Epidemiology of Enteric Viral Infections, hi: Viruses and
Wastewater Treatment. Eds. M. Goddard and M. Butler. Pergamon Press, New York.
p. 291.
57. Cioglia, L., and B. Loddo. 1962. The Process of Self-purification in the Marine
Environment. HI. Resistance of Some Enteroviruses. Nuovi Annali Di Igiene E.
Microbiologia. 13:11.
58. Martin, P. H. 1947. Field Investigation of Paratyphoid Fever with Typing of Salm.
paratyphi by Means of Vi Bacteriophage. Bull. Hyg. 22:754.
33
------- |