INTRODUCTION
Cryptosporidiosis is a gastrointestinal illness caused by the parasite Cryptosporidium [Reference Chen1]. Outbreaks of human cryptosporidiosis have been associated with exposure to both recreational water and drinking water systems [Reference Hlavsa2–Reference Goldstein5]. The first reported cryptosporidiosis outbreak related to a public water system occurred during 1987 [Reference Hayes6]. During 1993, the largest known Cryptosporidium community drinking water outbreak occurred in Milwaukee, Wisconsin, sickening an estimated 403 000 persons with 58 confirmed cryptosporidiosis-associated deaths [Reference Hoxie7, Reference MacKenzie8].
In the general population, typical symptoms of cryptosporidiosis include diarrhoea (which can last 1–2 weeks) and abdominal cramps, although one study reported 39% of infections were asymptomatic [Reference DuPont9]. Cryptosporidium infections are more common and more severe in immunocompromised persons [Reference Chen1, Reference Guerrant10]. Two species, Cryptosporidium hominis, for which humans serve as predominant host, and Cryptosporidium parvum, which is commonly associated with cattle, cause the majority of human cryptosporidiosis [Reference Xiao11]. Ingestion of oocysts expelled in faeces cause human infection [Reference Bouzid12–Reference Yoder and Beach15].
Because Cryptosporidium tolerates the levels of chlorine typically used in water treatment, Cryptosporidium-specific treatment is typically accomplished through removal by filtration or inactivation by ultraviolet (UV) disinfection. After Milwaukee's outbreak, in 1993, the U.S. Environmental Protection Agency (EPA) issued enhanced regulations to control microbial contamination of surface drinking water sources, concentrating efforts on contamination by Cryptosporidium [16]. These regulations require surface water-supplied community drinking water systems that have an exemption to filtration to implement Cryptosporidium-specific treatment in addition to chemical disinfection (e.g. chlorination) due to Cryptosporidium's chlorine tolerance. EPA regulations have been implemented in stages on the basis of water system size. Certain water systems (those supplying fewer than 10 000 persons) were not required to implement Cryptosporidium treatment until 1 October 2014, with an allowance for an extension when capital improvements were necessary.
Cryptosporidium-specific municipal water testing was performed in Baker City, Oregon during April 2010–March 2011; two water samples were tested monthly for Cryptosporidium, and three oocysts were detected in the 24, 10-l samples. On the basis of population, test results, and need for capital improvements, Baker City received an extension for Cryptosporidium-specific treatment from Oregon's Drinking Water Services until October 2016 [17, 18].
On 29 July 2013, the Baker County Health Department (BCHD) received notification of a positive Cryptosporidium laboratory report from a faecal specimen submitted by an ill patient treated at the local medical centre. Two additional Cryptosporidium laboratory reports were received on 30 July, after which BCHD notified Oregon Health Authority (OHA) of a cryptosporidiosis outbreak. Laboratory-confirmed cryptosporidiosis case patients or (if aged <18 years) their family members were interviewed by BCHD about exposures using a general case-investigation form to identify potential sources of infection, vehicles of transmission, and the possibility of ongoing transmission. The case-investigation form asked about exposure to risk factors commonly associated with Cryptosporidium infection including travel, recreational water exposure, consumption of unpasteurized milk or juice, contact with daycare attendees, sick people, or sick animals, occupational animal exposure and source of drinking water.
Initial interviews of confirmed case patients identified no common exposures, and no travel outside of the county was reported. BCHD staff also reported hearing many anecdotal reports of cryptosporidiosis-compatible illnesses from Baker City residents distributed throughout the city. The lack of other common exposures suggested drinking water was the likely source. On 31 July, a boil-water advisory was issued, and public health officials initiated an investigation.
During the first week of August, in order to better understand the geography of the outbreak, OHA epidemiologists conducted rapid assessment telephone surveys among residents of both Baker City and a neighboring city, LaGrande, which uses a separate community water supply. Respondents were interviewed using a standardized questionnaire to ask about recent diarrhoeal illness in their households. In Baker City, 36 (20%) of 179 respondents reported ⩾3 days of diarrhoea during the previous week, compared with two (1%) of 176 respondents in LaGrande, suggesting the outbreak was limited to Baker City. A convenience sample door-to-door survey of Baker City households revealed a 26% crude attack rate. On the basis of these preliminary findings, public health personnel performed additional investigations, the objectives of which were to confirm drinking water as the outbreak source, assess outbreak extent, and evaluate adequacy of control measures.
METHODS
Setting
Baker City, population 9828, located in eastern Oregon, relies on an unfiltered surface water supply. The watershed includes ~10 000 acres (40 km2) of mountainous terrain located northeast of the city. One ridge of the Elk Horn Mountains contains the western watershed boundary. The remaining watershed boundaries are marked by signs and fencing. Cattle grazing allotments border a substantial portion of the watershed. Surface water can enter any of 11 watershed diversions and travels about 10–20 miles (16–32 km) through one of two underground transmission lines to a water treatment plant outside the city where the water is chlorinated before distribution (Fig. 1). A combination of chlorination and disinfectant contact time was provided prior to the first user, achieving a minimum inactivation of 99·9% of Giardia cysts, which also inactivates bacteria and viruses but not Cryptosporidium. When the outbreak occurred, Baker City did not perform Cryptosporidium-specific treatment of the water supply.
Fig. 1. Schematic diagram of Baker City's watershed. A diversion is any transfer of water across watershed boundaries through a man-made pipeline or canal. Installations are man-made structures connected to the water system, Baker City, Oregon, 2013. (Figure created by Bill Keene.)
Environmental and water system investigations
Epidemiologists, environmental health specialists, and water system operators examined Baker City watershed maps and performed watershed inspections on 31 July, 3 August, and 22 August, seeking evidence of water contamination or prohibited entry of livestock on watershed lands. Investigators took photographs of the watershed; however, no survey equipment was used during watershed inspections. Investigators reviewed rainfall data from the National Center for Environmental Information, recorded in inches per 24 h, during the months before the outbreak because meteorological events can be a contributing factor during disease outbreaks. OHA Drinking Water Services (DWS) reviewed Baker City's water testing policies and procedures, and turbidity and coliform count data regarding water samples collected during July 2013.
Laboratory examination of water sampling
Baker City's water department (BCWD) collected untreated and treated water samples for Cryptosporidium-specific testing at a local laboratory using EPA Method 1623 [19], which involves passing water through a filter to concentrate and recover oocysts; eluting filtered materials and separating oocysts from extraneous materials; staining them; and microscopically examining samples. On 31 July, BCWD collected seven, 10-l water samples as follows: one untreated water sample from a suspect surface water diversion point (suspicious because of its close proximity to a herd of goats), one sample from one of two underground transmission lines, and five treated water samples from five distribution sites in Baker City. On 4 August, BCWD collected six untreated water samples from six surface water diversions and four treated samples (one from the city's reservoir and three from Baker City sites). OHA established criteria for lifting the boil-water advisory based on expert recommendation. BCWD continued testing water samples during the outbreak to determine when the advisory could be lifted.
OHA epidemiologists collected two untreated water samples (114 l and 120 l) from the suspect surface water diversion on 3 August, and two treated water samples (71 l and 130 l) on 4 August by dead-end ultrafiltration (DEUF) [Reference Mull and Hill20]. DEUF uses dialysis filters with pore sizes ranging from 0·01 to 0·1 µm to rapidly collect high-volume samples and trap submicron particles in addition to larger particles including Cryptosporidium oocysts, typically 4–6 µm in diameter. The used filters were shipped to the Centers for Disease Control and Prevention (CDC) for Cryptosporidium testing. DEUF samples were filtered, processed, and analysed according to procedures described by Mull and Hill [Reference Mull and Hill20, Reference Hill21].
Laboratory: human and animal samples
During the rapid assessment of the health status of Baker City residents, symptomatic persons were asked to provide a stool specimen for testing at the state laboratory. BCHD also requested stool samples from persons who contacted the health department to report illness. Nine Cryptosporidium-positive human stool samples were sent to CDC for genotyping and subtyping.
During watershed inspections, OHA epidemiologists collected 149 ruminant (cattle, goat, elk) faecal samples for microscopic evaluation and polymerase chain reaction (PCR) testing to determine genotype of any recovered pathogens at CDC.
Epidemiological investigation and statistical analysis
We surveyed Baker City residents to assess the community attack rate using a standardized questionnaire. The questionnaires collected data regarding age, sex, symptom profiles, illness onset date and symptom duration, health care-seeking behaviour, self-reported immunodeficiency, and ingestion of Baker City water (measured in glasses of municipal water per day) before and after outbreak identification. Respondents were also asked about awareness of and compliance with the boil-water advisory. We defined presumptive cryptosporidiosis as onset of an acute diarrhoeal illness lasting ⩾3 days after 30 June 2013, in a Baker City resident.
By using Baker City's 2010 census, an estimated 20% incidence of cryptosporidiosis based on the results of the rapid assessment telephone survey and a 0·05 significance level, we determined that a survey sample size of ~240 persons was needed to estimate the attack rate with a 5% margin of error. We used tax-lot data from the Baker County Assessor, to assign a number to each of Baker City's 4222 housing units by using a random number generator and sorted housing units by this random number. Assuming a response rate of ~65%, we attempted to visit the first 380 households randomly assigned. Two-member interview teams administered door-to-door questionnaires in Baker City during August and attempted to contact each selected household two times on separate days and at different times of day. One person from each housing unit visited was selected by using the next birthday method [Reference Salmon and Nichols22]. An adult proxy provided answers if the selected person was aged <12 years. Responses were collected using paper questionnaires and then transferred to a FileMaker Pro electronic database.
The weighted attack rate, weighted percentages, and confidence intervals were calculated, taking into account the two-stage sampling design by using SAS v. 9·3 (SAS Institute Inc., USA). We investigated associations between presumptive cryptosporidiosis and age, sex, and amount of Baker City tap water consumed daily by logistic regression by using survey procedures.
RESULTS
Environmental and water system investigations
During watershed inspections on 31 July, 3 August, and 22 August, OHA epidemiologists observed downed and damaged fencing surrounding the watershed documented through photographs. They reported goat, elk, and cattle faeces within watershed boundaries; cattle were grazing about 150–250 yards (137–229 m) from watershed limits.
Rainfall in Baker City was ~39% below average during January–June 2013 [23, 24]. On 19 June, 1·3 inches of rainfall was recorded, marking the highest single-day rainfall total in Baker City since 1984 [24, Reference Jacoby25].
Oregon DWS reported two inadequacies during review of the city's water sampling techniques. First, rather than testing water combined from both transmission lines, the sample point only captured water from one surface water transmission line. Second, rather than testing for thermotolerant coliform (commonly referred to as faecal coliform in the United States) counts as required by United States federal and Oregon state regulations [maximum contaminant level: 20 colony-forming units (c.f.u.)/100 ml], the city was testing raw water samples for Escherichia coli counts before and during the outbreak.
No spikes in turbidity or E.coli counts were noted before the outbreak. The highest E.coli count during July 2013 was 4 c.f.u./100 ml (no maximum contaminant level has been established for E. coli), and the highest turbidity level from continuous readings was 0·236 nephelometric turbidity units (NTU) (maximum contaminant level: 5 NTU).
Laboratory: water sampling
Untreated water samples from 4/7 surface water diversions and 6/9 treated water samples from Baker City distribution locations were positive for Cryptosporidium. One of four positive surface water samples was reported to have 913 oocysts/10 l, prompting Baker City to stop using water from this diversion. Of four DEUF samples, one treated water sample was positive for Cryptosporidium by fluorescence microscopy (four oocysts observed), but negative by PCR.
Laboratory: human and animal samples
A total of 23 laboratory-confirmed cryptosporidiosis cases were identified during the outbreak. Molecular results indicated C. parvum IIaA15G2R1 subtype in all nine human isolates that were sent to CDC.
One of 81 goat faecal samples tested was positive for Cryptosporidium ubiquitum. One of 64 elk faecal samples was positive for Cryptosporidium with species unknown, but negative for C. parvum and C. hominis by PCR. Four cattle faecal samples were negative.
Epidemiological investigation and statistical analysis
Of 380 households visited, 206 (54·2%) completed the survey; no one answered at 84 homes visited (22%), 46 residents refused (12%), 27 homes were vacant (7%), and at 17 homes there was no response due to other reasons (4%). Seven surveys were missing as a result of clerical error; 199 were available for analysis. Three surveys were missing illness duration and thus removed from attack rate calculations.
Of 50 presumptively ill persons (25·1% of surveyed population), 27 (54%) were female; median age was 49 (range 2–89) years; and six (12%) had an immunodeficiency. Symptom onset ranged from 1 July (4 weeks before the outbreak was detected) to 10 August (Fig. 2). Median symptom duration was 6 (range 3–50) days. Ten (20%) persons had sought medical care; three (6%) had sought treatment at an emergency department; and three (6%) had been admitted to a hospital. However, none had laboratory-confirmed cryptosporidiosis.
Fig. 2. Cryptosporidiosis outbreak epidemic curve: laboratory-confirmed cases and presumptive cases from door-to-door survey, by date of illness onset, Baker City, Oregon, 2013.
The weighted attack rate was 28·3% [95% confidence interval (CI) 22·1–33·6]. On the basis of Baker City's 2010 census, ~2780 persons became ill during the outbreak. Presumptive cryptosporidiosis was not significantly associated with sex, age, or daily tap water consumption (Table 1).
Table 1. Presumptive cryptosporidiosis weighted attack rate, by sex, age group, and cups of Baker City water consumed daily for sampled Baker City residents (n = 196), Baker City, Oregon, 2013
OR, odds ratio; CI, confidence interval; Ref., reference.
All household members interviewed reported having heard about the boil-water advisory. The majority (88%) reported hearing about the advisory from an unofficial source (e.g. family, friend, neighbour, or social media), and 67% used bottled water as their primary water source during the advisory (Table 2). Of 77·7% who boiled water ⩾1 time during the advisory, 96% did so for at least the recommended time of 1 min. Certain residents (7·5%, 95% CI 3·5–11·4) continued to drink unfiltered tap water throughout the outbreak.
Table 2. Household survey information about information source, water source, and water treatment, Baker City, Oregon, 2013
CI, Confidence interval.
* Denotes official sources of information.
† Reported estimated percentage of time household boiled water during advisory.
Control measures
The boil-water advisory was put in place on 31 July. After receiving results of water testing from water samples obtained on 31 July and 4 August, OHA established criteria for lifting the boil-water advisory on 8 August; OHA recommended the city perform twice-weekly testing of water samples from each watershed diversion being used, obtained ⩾24 h apart, until test results were negative for Cryptosporidium on two consecutive samples. This criteria was met on 20 August and the boil-water advisory was lifted. The Baker City Council voted to install a Cryptosporidium-specific treatment facility ahead of the October 2016 deadline because of the outbreak. The surface water diversion with the highest oocyst count remained offline until Baker City began UV water disinfection on 25 March 2014. DWS required BCWD to collect and test 10-l, untreated water samples obtained from the combined pipeline for Cryptosporidium-specific testing using EPA Method 1623 twice weekly until the UV water treatment plan began operations. If two consecutive samples detected Cryptosporidium, the boil-water advisory would be reinstated. No additional testing for other waterborne chlorine-resistant pathogens has been performed. As of 1 May 2015, no additional cases of cryptosporidiosis have been reported in Baker County.
DISCUSSION
Baker City's communitywide outbreak is the first reported cryptosporidiosis outbreak detected in a US municipal drinking water system in ~20 years. Cryptosporidium is well-suited to waterborne transmission because it has a low infectious dose, is excreted in substantial amounts for long periods, and has an environmentally hardy form (oocyst) that is immediately infectious upon excretion. Additionally, oocysts’ small size can pass through certain filtration systems and their extreme chlorine tolerance renders treatment with chemical water disinfectants ineffective. Since the last cryptosporidiosis outbreak associated with community surface water-supplied drinking water systems in the United States was reported to CDC in 1994, the majority of reported cryptosporidiosis outbreaks have been associated with treated recreational water venues [Reference Blackburn26–Reference Yoder32].
The Baker City outbreak highlights the need for continued vigilance in maintaining multiple barriers to drinking water safety. EPA recommends the multiple-barrier approach [33] to ensure a safe drinking water supply, which consists of the following: (1) risk prevention, including source water protection; (2) risk management; (3) monitoring and compliance; and (4) individual actions. Review of Baker City's outbreak identified deficiencies at each barrier that contributed to outbreak occurrence and serves as a reminder of each barrier's importance.
Although signs and fences mark Baker City's watershed boundaries, they were inadequate to protect the surface water from contamination. Although genotyping data of Cryptosporidium in water samples were not available to link with clinical specimens, molecular studies of human stool samples and the environmental investigation provided circumstantial evidence that implicated cattle as the probable contamination source. All human specimens tested positive for the same Cryptosporidium genotype and subtype, C. parvum IIaA15G2R1, a common subtype found in the United States and the predominant endemic subtype in pre-weaned calves [Reference Xiao11]. Investigators witnessed cattle grazing on land adjacent to the watershed and there is a high likelihood that nursing calves were also present in the herd since spring-born calves nurse for 3–8 months [Reference Filley34]. Dried cattle faeces were observed within watershed boundaries. Risk prevention through reinforcement of watershed barriers to prevent cattle breaches might decrease this source of surface water contamination. Meteorological conditions might have contributed to contamination because drought conditions can alter water flow and concentrate contaminants in water. The substantial amount of precipitation during June might have washed animal faeces into watershed intakes.
Risk management involves water treatment to remove and inactivate contaminants in source water; the Baker City outbreak emphasizes the importance of water regulations that require treatment to control Cryptosporidium. Although Baker City chlorinated its water supply, water filtration or UV treatment to specifically target Cryptosporidium contamination was not required when the outbreak began. Before the most recent EPA rule, 60 water systems serving ~10 million persons in the United States met specific filtration avoidance criteria and distributed unfiltered surface water [16]. Now the majority of systems comparable to Baker City have added Cryptosporidium treatment [16] and only one system has a variance (Portland, Oregon) [35].
Active monitoring of water systems, using easy to measure parameters such as turbidity, allows for adjustments to help maintain a dependable, safe water supply. In Baker City, water turbidity and E. coli counts did not change substantially before the outbreak; thus, officials were not alerted to faecal contamination in treated water. Although the outbreak provided the opportunity to improve water testing practices by ensuring sampling locations included a blend of all surface water sources and adding thermotolerant coliform testing, these improvements might not reliably signal Cryptosporidium contamination.
Effective actions at the individual level rely on timely consumer notification of health risks. In Baker City, delays in outbreak detection by laboratory-based surveillance were apparent, evidenced by the outbreak going unnoticed for ~4 weeks. When laboratory-confirmed cases were reported, public health officials recognized the outbreak promptly and responded quickly with a boil-water advisory, potentially curtailing additional illnesses before water contamination was confirmed. This public health response was consistent with recommendations that during communitywide outbreaks, control measures should be implemented rapidly rather than waiting for identification of a specific transmission source [36]. Although the majority of residents initially were informed about the boil-water advisory through informal channels, the message was disseminated throughout the community and verified by official sources. During the boil-water advisory, 67% of residents reported use of bottled water and 15% reported use of boiled water as their primary water sources demonstrating adherence to public health recommendations.
Outbreaks associated with drinking water can be difficult to detect; in this case, five laboratory-confirmed cases were the total signal through laboratory-based surveillance to public health. However, the community survey indicated 2780 (28% of the population) cases might have occurred. Similarly, during the Milwaukee outbreak, 285 patients had laboratory-confirmed cryptosporidiosis, whereas an estimated 403 000 persons (26% of the population) were affected [Reference MacKenzie8]. Although 70% of the population is exposed to Cryptosporidium during their lifetime [Reference Frost37] and an estimated 748 000 cases occur annually [Reference Scallan38], only 8951 cryptosporidiosis cases were reported in the United States during 2010 [Reference Yoder14]. Underdiagnosis of Cryptosporidium likely contributes to underrecognition of outbreaks.
Economic and social ramifications of substantial outbreaks are important to consider, but often difficult to quantify. During 2008, a communitywide Salmonella Typhimurium outbreak associated with the public drinking water system in Alamosa, Colorado, resulted in estimated costs to residents of ~$850 000 and to businesses of ~$625 000. One third of businesses surveyed reported closing for a mean length of 8·4 days during the outbreak [Reference Ailes39]. Although investigators did not specifically evaluate economic and social consequences of the Baker City outbreak, anecdotal community experiences illustrate outbreak costs. Grocery stores were forced to dispose of produce that had been sprayed with city water. Certain restaurants closed during the boil-water advisory rather than install expensive water filters or perform additional water treatment. Restaurants that remained open often used disposable flatware and dishes rather than risk contamination during the dishwashing process. Hotels provided bottled water for patrons. One food processing plant was forced to dispose of products made with potentially contaminated water, which resulted in considerable financial losses and subsequent factory layoffs. Although investment costs in water system improvements to comply with more stringent drinking water regulations are often borne by community residents in the form of higher rates, costs of avoiding or delaying implementation of measures designed to protect public health might be considerably higher. An economic analysis evaluating costs attributable to implementing the EPA rule related to Cryptosporidium-specific treatment estimated the annual cost increase/household would be <$7 for >90% of all households served by surface water and ground water under the direct influence of surface water [40].
Limitations of this investigation include a narrow presumptive case definition, which might have underestimated the size of the affected population. However, certain survey-identified cases might not have been Cryptosporidium infections because a background level of diarrhoeal illness was likely present. Residents who experienced illness might have been more likely to respond to the survey, which might result in an overestimate of the attack rate; to minimize bias, we contacted a random sample of households. However, our low survey response rate might have affected validity of results. Finally, the survey was not powered to detect differences in attack rates in groups, limiting our ability to associate illness incidence to amount of water consumed, age, or other individual characteristics or behaviours. Thus, data associating the outbreak to drinking water came from the initial assessments of communitywide diarrhoeal illness and the ensuing laboratory and environmental investigations.
Safe drinking water, among the most basic public health interventions enjoyed by the majority of communities in the United States, has resulted in substantial reductions in mortality in the United States during the 20th century [Reference Cutler and Miller41]. The United States has made considerable progress during >40 years since passage of the 1974 Safe Drinking Water Act with regard to the distribution of high-quality, clean drinking water [Reference Craun42]. After Milwaukee's cryptosporidiosis outbreak during 1993, EPA increased efforts to protect surface water-supplied systems from microbial contamination and prevent Cryptosporidium from causing drinking water-related outbreaks [16, 43, 44]. The possibility exists that these regulatory efforts in the United State are responsible for the ~20 years separating the last reported drinking water-related cryptosporidiosis outbreak in Las Vegas, Nevada (1993–1994) [Reference Goldstein5] and Baker City's outbreak during 2013. Although the majority of public drinking water systems usually supply safe, clean water, public health agencies should be aware that drinking water-associated outbreaks continue to occur, can be difficult to detect, and have considerable health, economic, and social repercussions [Reference MacKenzie8, Reference Ailes39, Reference O'Reilly45]. Baker City's outbreak serves as a reminder of why surface water regulations were implemented during 1989 and strengthened multiple times; surface water is inherently vulnerable to contamination. The outbreak underscores the vital importance of investments in improved water treatment and system upkeep. Proactive investments by community water systems to maintain and improve drinking water quality and continued vigilance by public health agencies are necessary to prevent, identify, and control community drinking water-related disease outbreaks.
ACKNOWLEDGEMENTS
We thank the many individuals who made significant contributions to the successful completion of this investigation, particularly Emilio DeBess, Kara Levri, June Bancroft, Colleen Lysen, and Chandra Schneeberger. We also thank the Baker City Volunteer Fire Department, Lynda Neal, Stephen Christy Alyssa Bostian, Stephanie Busch, Mike Gronostaj, Mark Miller, and Julia Painter for their contributions during the field investigation.
This report is dedicated to the memory of Bill Keene, a pioneering, passionate epidemiologist who worked for the state of Oregon until his unexpected death in December 2013. The Baker City outbreak was one of the last outbreaks in which he participated.
The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
DECLARATION OF INTEREST
None.
