Waterborne outbreaks reported in the United States

Gunther F. Craun Michael F. Craun (corresponding author) Gunther F. Craun & Associates, 101 West Frederick Street, Suite 207, Staunton, VA 24401, USA E-mail: craunco@cfw.com Rebecca L. Calderon National Health & Environmental Effects Research Laboratory, Office of Research & Development, US Environmental Protection Agency, Research Triangle Park, NC 27711, USA Michael J. Beach Division of Parasitic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, GA 30341, USA Epidemic waterborne risks are discussed in this paper. Although the true incidence of waterborne illness is not reflected in the currently reported outbreak statistics, outbreak surveillance has provided information about the important waterborne pathogens, relative degrees of risk associated with water sources and treatment processes, and adequacy of regulations. Pathogens and water system deficiencies that are identified in outbreaks may also be important causes of endemic waterborne illness. In recent years, investigators have identified a large number of pathogens responsible for outbreaks, and research has focused on their sources, resistance to water disinfection, and removal from drinking water. Outbreaks in surface water systems have decreased in the recent decade, most likely due to recent regulations and improved treatment efficacy. Of increased importance, however, are outbreaks caused by the microbial contamination of water distribution systems. In order to better estimate waterborne risks in the United States, additional information is needed about the contribution of distribution system contaminants to endemic waterborne risks and undetected waterborne outbreaks, especially those associated with distribution system contaminants.


WATERBORNE OUTBREAK SURVEILLANCE SYSTEM
The WBDO surveillance program is conducted to: (1) characterize the epidemiology and etiology of WBDOs and identify important waterborne pathogens and water system deficiencies; (2) improve detection and investigation capabilities; and (3) collaborate with local, state, Federal, and international agencies on initiatives to prevent waterborne disease (Lee et al. 2002;Blackburn et al. 2004).The primary unit of analysis is an outbreak rather than an individual case of illness.State, territorial, and local public health agencies have the primary responsibility for detecting and investigating WBDOs, and these agencies voluntarily report WBDOs to the CDC.When requested, the CDC and EPA assist in outbreak investigations.
A standard reporting form is used to solicit data on the characteristics of the outbreak (e.g.dates of illness onset, duration of illness, and suspected or confirmed etiology), testing of water and patient samples, and contributory issues such as water disinfection practices and environmental factors.Information is also requested about the actual and estimated numbers of cases, hospitalizations, and fatalities.This information is evaluated and reported in WBDO surveillance summaries, which have been published biennially or annually since 1973 (CDC 1973(CDC , 1974(CDC , 1976a(CDC , b, 1977(CDC , 1979(CDC , 1980(CDC , 1981(CDC , 1982a(CDC , b, 1983(CDC , 1984(CDC , 1985;;St. Louis, 1988;Levine & Craun, 1990;Herwaldt et al. 1991;Moore et al. 1993;Kramer et al. 1996;Levy et al. 1998;Barwick et al. 2000;Lee et al. 2002;Blackburn et al. 2004).
Outbreaks associated with drinking water, recreational water, and other types of water exposures are reported.
WBDOs associated with cruise ships are not included in this surveillance system.In this paper, we consider only outbreaks associated with contaminated drinking water.

Outbreaks
For an event to be defined as a WBDO, two or more persons must have experienced a similar illness.This criterion is waived for single cases of laboratory-confirmed primary amebic meningoencephalitis (PAM) and for single cases of chemical poisoning if water quality data indicate contamination by the chemical (Blackburn et al. 2004).
Waterborne pathogens of concern in the United States have multiple transmission routes, including person-to-person contact and ingestion of contaminated food.Thus, epidemiologic evidence must implicate water as the probable source of the illness.
Since 1989, WBDOs have been classified according to the strength of the evidence (Table 1) implicating water (Blackburn et al. 2004).The classification system ensures objectivity in the review of outbreak reports and consistency in the reported statistics as well as encouraging investigators to submit more complete information.Classification is based on epidemiologic and water quality data provided by investigators.Outbreaks without water quality data can be included in the surveillance system, but reports that lack epidemiologic data are not.A classifi-

Limitations of the surveillance data
The information pertains primarily to outbreaks, and the reported statistics do not include endemic or sporadic cases that may be waterborne.In addition, not all WBDOs are recognized and investigated and not all investigated WBDOs are reported.Since not all investigations were optimally conducted, some information (e.g.illness severity) may not be reported.

Outbreak reporting
Since WBDO surveillance is passive and reporting is voluntary, the statistics represent only a portion of the waterborne outbreaks that actually occur (Hopkins et al. 1985;Craun 1986;Blackburn et al. 2004).Blackburn et al.Outbreaks most likely to be recognized and investigated are those of (1) acute illness characterized by a short incubation period, (2) serious illness or symptoms requiring medical treatment, and (3) recently recognized etiologies for which laboratory methods have become more sensitive or widely available (Blackburn et al. 2004).Increased reporting often occurs as water system deficiencies and WBDO etiologies become better recognized, often through improved state surveillance activities and laboratory capabilities (Hopkins et al. 1985;Frost et al. 1995Frost et al. , 1996)).
Recommendations for improving WBDO statistics include: (1) enhanced surveillance activities to better detect out-breaks; (2) additional laboratory support for clinical and water analyses during outbreak investigations; and (3) increased attention to potential sources of bias during investigations (Craun et al. 2001;Frost et al. 2003;Hunter et al. 2003).

Illness reporting
The reported cases of illness in WBDOs are primary cases, either actual or estimated.Few investigations have identified secondary cases (i.e.persons infected by contact with primary case-patients).The cases may be defined by signs and symptoms or may be confirmed by laboratory analysis of clinical specimens.Cases may be under-or over-reported in some WBDOs.For example, even though the 1993 Milwaukee cryptosporidiosis outbreak investigation was extensive (MacKenzie et al. 1994;Hoxie et al. 1998;Proctor et al. 1998;Naumova et al. 2003), outbreak-related cases may have been over estimated (Hunter & Syed 2001).
However, a study of Cryptosporidium-specific antibody responses in children by McDonald et al. (2001) also suggests that infection may have been more widespread.
During the investigation it is important to recognize and take steps to control potential biases and assess their affects, especially recall bias.Recall bias may result in the reporting of more illnesses than actually occurred (Craun & Frost 2002;Craun et al. 2001;Cooper 1995;Hunter & Syed 2001).2).

WATERBORNE OUTBREAK STATISTICS
In the remaining six time periods that were evaluated, an average of 4640-9331 cases was reported each year.
WBDOs in community systems ranged from 247 to 5714 illnesses per outbreak, while WBDOs in non-community  1930 1931 to 1940 1941 to 1950 1951 to 1960 1961 to 1970 1971 to 1980 1981 to 1990 1991 to 2002 Time period Average annual number of waterborne outbreaks   2).
Fifty-eight percent of the WBDOs reported since 1971 were relatively small, resulting in 50 or fewer illnesses; only 4% of these WBDOs resulted in more than 1000 illnesses (Figure 3).The six largest WBDOs accounted for the majority (88%) of illnesses during this time period (Figure 3), demonstrating the impact that large WBDOs can have on illness statistics.The largest WBDO, an estimated 403 000 illnesses, occurred in Milwaukee in 1993.

Duration of illness
Information about the duration of illness was available for 40% of the WBDOs reported during 1971-2002.The mean and median of the reported duration of illness for all etiologies was 5.6 and 2.2 days, respectively; the longest reported duration was 74 days.A median duration of 6 days or less was reported in 80% of the WBDOs (Figure 4).
Typically, the shortest duration of illness was found in WBDOs of a chemical or viral etiology.

Mortality
During 1920 to 2002, 1165 deaths were reported, an average of 14 deaths per year.Most deaths occurred before 1940 during WBDOs of typhoid fever (Craun 1986).During the 12-year period 1991 -2002, 73 deaths (an average of 6 deaths per year) were reported (Figure 5).6).During the 20-year period 1971-1990, these deficiencies were implicated in less than 20% of WBDOs.Distribution system-associated WBDOs tend to be small, as contamination usually affects only a portion of the distribution system, limiting the potential exposure.On average during the past 32 years, these WBDOs resulted in 152 cases per outbreak.However, five distribution systemassociated WBDOs resulted in more than 1000 illnesses, with the largest causing 5000 illnesses.Although a chemical etiology is often found (35% of the WBDOs), distributionsystem WBDOs are also caused by bacterial (17%), protozoan (14%), viral (4%), or undetermined (30%) pathogens.

ETIOLOGY OF WATERBORNE OUTBREAKS
A historical perspective of the etiologies of reported WBDOs is provided in Figure 7 In spite of better laboratory methods and more thorough investigations, WBDOs classified as acute gastroenteritis (AGI) of undetermined etiology continue to be important.
Usually the etiology was not determined because specimens were not collected or laboratory analyses were not available.
However, in some WBDOs, the agent could not be identified even though laboratory analyses were available.During the five time periods that we analyzed, the etiology was determined in 37-73% of reported WBDOs (Figure 7).
During the most recent 12 years, the etiology was determined  1931-40 1941-50 1951-60 1961-70 1971-80 1981-90 1991 to 2002 Time period Average number deaths per year in time period   distribution system, and mostly likely spread by aerosolization of water from the system, usually from hot water taps.
In two WBDOs, Legionella may have entered during a mains break or back-siphonage.
Increasing numbers of waterborne pathogens have been identified as causes of WBDOs in the United States.During 1920 -1940, only four waterborne pathogens were identified; during 1991 -2002, 13 pathogens identified (Table 3).

Among the recently recognized waterborne pathogens is
Cyclospora, which caused a single WBDO in a Chicago building that housed hospital personnel (Herwaldt et al. 1991).Other yet to be identified pathogens may become important.For example, two WBDOs of chronic diarrhea were reported, but no causative agent was identified even after extensive laboratory analyses (Parsonnet et al. 1989).
Before 1970, ten protozoan WBDOs were reported; these were primarily caused by E. histolytica.

DISCUSSION
Although the WBDO surveillance statistics are imperfect for estimating the incidence of epidemic waterborne illness, they can help identify important waterborne pathogens and water system deficiencies.These same pathogens and deficiencies may also be important to consider when assessing endemic risks.Surveillance information can also be used to identify changing sources of contamination and the adequacy of current treatment and regulations.If As the population that is susceptible to severe illness or death (e.g.elderly, organ transplants, HIV infected persons, AIDS patients) becomes larger, future WBDOs may have a greater public health impact.
Since 1991, an increased proportion of WBDOs have been associated with contaminants that have entered the water distribution system.Microbial contaminants have been implicated in two-thirds of the distribution-system-associated WBDOs, and many of these pathogens are not likely to be killed by the relatively low levels of disinfectant residuals maintained in the water distribution system.These WBDOs are also among these that may frequently go unrecognized.
Although these outbreaks have tended to be relatively small, several recent distribution-system-associated WBDOs have resulted in a large number of illnesses.A better understanding is needed of the extent to which these WBDOs are detected and the importance of distribution system contamination for endemic waterborne

ACKNOWLEDGEMENTS
We wish to acknowledge the many epidemiologists, microbiologists, engineers, environment health specialists, and Since 1971, the US Environmental Protection Agency (EPA), Centers for Disease Control and Prevention (CDC) and Council of State and Territorial Epidemiologists have collaborated to collect information about the causes of WBDOs.In this paper, we provide a historical perspective of WBDOs reported in the United States.

( 2004 )
point out that the true incidence of WBDOs is markedly underestimated and studies have not been performed to assess the sensitivity of the surveillance regarding unrecognized or unreported outbreaks.Multiple factors influence whether waterborne outbreaks are recognized and investigated.These factors include public awareness, availability of laboratory testing, requirements for reporting diseases, and resources available to local health departments for surveillance and investigation of probable outbreaks.In addition, changes in the capacity of public health agencies to detect an outbreak might influence the numbers of outbreaks reported in each state relative to other states.Thus, caution is urged in assessing trends in the occurrence of WBDOs.An increase in the number of reported WBDOs could reflect an actual increase or a change in sensitivity of surveillance practices.
systems ranged from 51 to 268 illnesses per outbreak (Table
. During the late 19th and early 20th centuries, cholera and typhoid were frequent causes of WBDOs in the United States.Only three WBDOs of cholera with 131 cases have been reported since 1920.Two occurred in American territories, and one occurred in a non-community system in Texas.Waterborne typhoid fever continued to occur after 1920; 70% of all WBDOs reported during 1920 -1940 were attributed to Salmonella typhi.WBDOs of typhoid fever decreased considerably over the next 30 years to only 22% and 11% of WBDOs reported during 1941 -1960 and 1961 -1970, respectively.An even more dramatic decrease occurred in cases of typhoid associated with WBDOs; 87 675 typhoid cases were reported during 1920 -1941 but only 108 cases occurred from 1961 -1970.Since 1971, five small WBDOs occurred, and only 282 cases of typhoid fever were reported.

Figure 5 |
Figure 5 | Deaths associated with reported drinking water outbreaks in the United States 1920-2002.
current treatment is inadequate to remove or inactivate these pathogens and if water system deficiencies that cause outbreaks are not identified and corrected, both endemic and epidemic waterborne risks are increased.Although the number of outbreak-associated illnesses may be relatively small when compared with the possible endemic waterborne risk in the United States, illness estimates should consider the extent to which WBDOs may go unrecognized and the likelihood that one of more large WBDOs may occur in the future.The statistics for 1991-2002 are dominated by the largest WBDO since surveillance began; an estimated 403 000 persons became ill, 4400 persons were hospitalized, and 50 persons died.The concern is whether current treatment technologies, monitoring, and operational practices are adequate to remove or kill a more virulent emerging waterborne pathogen.WBDO etiologies have changed over the years and will likely continue to change.Since 1991, 14 waterborne pathogens have caused WBDOs in the United States.The infectivity and virulence of these pathogens vary as does the host response to infection.The changing nature of waterborne pathogens suggests that other pathogens may well be important in the future.The most frequently identified etiologic agents in the last 12 years have been Giardia and Cryptosporidium, two pathogens characterized by a low infectious dose, good survival in a cold water environment, and resistance to water treatment practices that were once state-of-the art.Pathogens of emerging importance may be resistant to current water treatment practices, which have recently been upgraded to remove or kill Giardia and Cryptosporidium.WBDO surveillance can help identify changing water quality conditions and guide research strategies to ensure that treatment technologies are adequate for newly identified waterborne pathogens.Although the mortality associated with WBDOs has decreased since 1920, an increase has occurred during the last 12 years.This increase is largely due to the 50 deaths during the Milwaukee WBDO.The underlying cause of these deaths was primarily AIDS, but the contributing cause of death was cryptosporidiosis.Cryptosporidium infection may lead to mild or no symptoms in some persons but to an illness of relatively long duration in others.The infection can be severe in persons with a suppressed immune system.

DISCLAIMER
The views expressed in this paper are those of the individual authors and do not necessarily reflect the views and policies of the U.S. Environmental Protection Agency or the Centers for Disease Control and Prevention.The paper has been subject to the Environmental Protection Agency's peer review and approved for publication.

Table 2 |
Average size of waterborne outbreaks in the UnitedStates, 1920States,  -2002 ations were of a bacterial (42%) or protozoan (18%) etiology.Protozoa were responsible for most (91%) cases that required hospitalization.Nine persons were hospitalized during four viral WBDOs, and 46 persons were hospitalized during 15 WBDOs of undetermined etiology.Water system deficienciesSince 1971, each WBDO has been classified into one of five water system deficiency categories.We evaluated the deficiencies associated with WBDOs during 1971-2002 (Figure6).The proportion of WBDOs reported in untreated groundwater systems has remained relatively constant since 1971.The proportion of WBDOs associated with contaminated, untreated surface water has decreased since 1971, and since 1991 no WBDOs have been associated with untreated surface water systems.This is largely due to EPA rules and regulations that require the adequate treatment of public water systems using surface water.Water distribution system deficiencies have now become more important as a cause of WBDOs.These deficiencies were responsible for more than half of all WBDOs reported

Table 3 |
Etiology of waterborne outbreaks reported in the UnitedStates, 1991States,  -2002Without their efforts and cooperation, a summary such as this would not be possible.We also thank our colleagues at the Centers for Disease Control, especially Deborah Levy, Barbara Herwaldt, Sherline Lee, and Brian Blackburn, who have collaborated with us on previous reviews of waterborne outbreaks.Finally, we thank Frantisek Kozisek, National Institute of Public Health, Prague, Czech Republic, for his review and helpful comments in the preparation of this paper.