The First Reported Outbreak of Diarrheal Illness Associated with Cyclospora in the United States

Methods

Case-Finding and Cohort Studies

The health clinic records of employees and housestaff physicians who reported having had diarrhea in July and August 1990 were reviewed. Open-ended interviews were conducted with six housestaff physicians whose stool specimens contained Cyclospora organisms; these interviews were intended to generate hypotheses about the mode of transmission and to characterize the clinical syndrome.

A case-patient was defined as any housestaff member, any employee working in the physicians' dormitory, or any employee working in the hospital administrative building next to the dormitory who reported onset of diarrhea (three or more loose stools in a 24-hour period) from 5 July through 12 July 1990, provided that the diarrhea had lasted for at least 3 days. It was thought that persons who had had diarrhea for less than 3 days might include a mix of cases and non-cases; thus, they were entirely excluded from the analysis to clean both the case and noncase categories.

We did a retrospective cohort study of housestaff physicians who collected paychecks on 8 and 9 August 1990. A questionnaire asked about any illness after 1 July 1990 (the first day of new clinical rotations) and for demographic and exposure information.

Because it was difficult to separate exposures related to being a housestaff physician from those associated with the physicians' dormitory, we did a second retrospective cohort study on 20 August 1990 of administrative personnel who worked in the physicians' dormitory. The questionnaire given to these personnel was similar to that given to the housestaff physicians, but it asked about illness after 4 July 1990 (a holiday for administrative personnel).

A final cohort study was done on 23 August 1990 of personnel who worked in the administrative building next to the physicians' dormitory. This study was done to help determine whether illness had been limited to the physicians' dormitory.

Laboratory Methods

Stool specimens were obtained from housestaff physicians and administrative employees who reported having had diarrhea. Specimens obtained from 5 July through 7 August 1990 were cultured for Campylobacter, Salmonella, Shigella, Vibrio, and Yersinia organisms, and microscopic examinations of wet-mount and acid-fast stained specimens were done by the hospital laboratory. Hospital laboratory procedures included the collection of stool specimens in 10% formalin and polyvinyl alcohol. The formalin-treated stool was examined directly using the D'Antoni iodine preparation and was concentrated using the formalin-ethylacetate procedure [12]. The sediment was examined using an additional iodine preparation, and smears were made for staining with a modified acid-fast procedure [13]. Permanent smears were prepared from polyvinyl alcohol-preserved material and stained with the Mallory iron-hematoxylin procedure. Fresh stools were cultured on 5% sheep blood agar, MacConkey agar, Hektoen enteric agar, eosin-methylene blue agar, and Campylobacter medium, and in selenite broth. These specimens were then examined at the Centers for Disease Control and Prevention (CDC) by microscopic examination under ultraviolet light for the presence of Cyclospora organisms. Concentration methods were abandoned at the CDC when parasite counts were discovered to be lower in concentrated stool specimens. Specimens obtained from 8 through 27 August were sent directly to the CDC and there examined for the presence of Cyclospora organisms by microscopic examination under ultraviolet light.

Between 6 September and 5 October 1990, stool specimens were collected from seven patients whose stools had shown Cyclospora organisms and from asymptomatic volunteers who worked in the physicians' dormitory or the hospital administrative building. All stool specimens underwent microscopic examination under ultraviolet light at the CDC. The CDC examiner was blinded to the case status of the sources of these specimens.

Surveillance Methods

Laboratory directors at the hospital, at surrounding hospitals, and at the Illinois Department of Public Health were asked to do acid-fast stains on all stool specimens submitted for any reason between 6 August and 27 August 1990. Microbiology records for July 1990 from the hospital laboratory were also reviewed.

Environmental Investigation

On 20 July 1990, the City of Chicago Bureau of Water Services inspected the plumbing and took water samples from the physicians' dormitory. On 16 August 1990, inspectors from the Illinois Environmental Protection Agency and the City of Chicago Bureau of Water Services inspected the plumbing system in the physicians' dormitory. Water samples were collected from the bottom of the south storage tank of the physicians' dormitory and from other selected locations throughout the building (especially from areas that had had little or no use since the outbreak onset). Water specimens (about 50 mL each) from the storage tanks were placed in four sterile cups. The specimens were centrifuged at 3500 rotations per minute for 30 minutes and then treated as routine parasitologic specimens. Water samples were also tested on site for chlorine residuals and were submitted to the Illinois Environmental Protection Agency laboratory for bacterial and plankton analyses and for phycologic examination by light microscopy. Water samples from the storage tanks were also sent to the CDC for microscopic examination under ultraviolet light.

On 29 August 1990, representatives from the United States Environmental Protection Agency, the Illinois Environmental Protection Agency, the Illinois Department of Public Health Division of Environmental Health, and the City of Chicago Bureau of Water Services inspected the physicians' dormitory and reviewed plumbing maintenance logs.

Statistical Analysis

The data were analyzed using Epi Info [14] and True Epistat [15] software packages. Relative risks and Greenland, Robins 95% CIs were calculated to evaluate associations between illness and categorical variables. When expected cell sizes were less than 5, exact confidence limits (calculated using the Thomas and Gart method) are reported. Comparisons of attack rates were done using the Fisher exact test for comparison of proportions.

Results

The Clinical Syndrome

The 21 case-patients identified in the housestaff and physicians' dormitory cohorts were used to characterize the clinical syndrome. All reported having had diarrhea with a median duration of 5 days (range, 3 to 18 days). Sixteen (76%) reported abdominal cramping; 11 (52%), muscle aches; 10 (48%), decreased appetite; 10 (48%), fever; 9 (43%), nausea; 8 (38%), headache; 5 (24%), constipation; 5 (24%), vomiting; and 4 (19%), chills. Six (43%) of the 14 housestaff case-patients reported having sought medical attention. This information was not collected from the 7 employees. In addition, open-ended interviews with 6 Cyclospora-positive housestaff physicians suggested a typical clinical presentation consisting of a 1-day prodrome of malaise and low-grade fever (4 of 6 patients), followed by explosive watery diarrhea (6 of 6 patients) accompanied by anorexia (5 of 6 patients), severe abdominal cramping (4 of 6 patients), nausea (3 of 6 patients) with occasional vomiting (1 of 6 patients), and a bloated feeling (1 of 6 patients). Five of the 6 housestaff physicians also reported having had remission of the diarrhea after a few days, followed by a cycle of relapses and remissions lasting up to several weeks. The patients noted having had continued malaise and anorexia, sometimes accompanied by constipation, during remission periods.

Epidemiologic Studies

Questionnaires were completed by 210 (42%) of the 496 full-time housestaff physicians working at the hospital. Four respondents reported having had diarrhea for less than 3 days between 5 July and 12 July; their questionnaires were excluded from analysis. Fourteen (6.8%) met the case definition (Figure 1). The earliest onset of illness was on 5 July, and the peak of onsets was on 6 July. The median age of the 14 case-patients was 30 years (range, 27 to 34 years), and 86% were men. The median age of the other 196 members of the cohort was also 30 years; 65% of these 196 persons were men.

Drinking tap water in the physicians' dormitory and attending a housestaff association party held on 29 June in the physicians' dormitory were significantly associated with illness (Table 1). Ten of the 12 case-patients who had attended the party reported drinking dormitory tap water, and 2 other case-patients reported drinking dormitory tap water without having attended the party. No other exposures were associated with illness.

Seventy-nine administrative employees working in the physicians' dormitory completed questionnaires. Two questionnaires were excluded from analysis because the employees had not worked between 25 June and 6 July, and 3 were excluded because the employees reported having had diarrhea for less than 3 days between 5 July and 12 July. The 74 remaining questionnaires represented almost all employees (excluding those conducting this study) who had occupied the administrative building during the period under study.

Nine employees had diarrhea for at least 3 days (Figure 1). Seven met the case definition restrictions for date of onset; this yielded an attack rate of 9.4%. Two of these seven case-patients submitted stool specimens, one of which was positive for Cyclospora organisms. The median age of these case-patients was 50 years (range, 22 to 60 years), and three were men. The age and sex of these seven case-patients were similar to those of the study population. All seven reported having had some type of exposure to tap water from the physicians' dormitory (drinking it or using it to wash fruit or vegetables, prepare food, or brush teeth) (Table 2). Only two of the seven case-patients attended the housestaff association party on 29 June. No other exposures were associated with illness.

The third study was conducted in the hospital administrative building, a 16-story office building located next to the physicians' dormitory. Approximately 530 employees have offices in the administrative building; however, many spend little time there. The building has a cafeteria that serves administrative staff and provides free meals to housestaff. One hundred thirty-five employees working in this building completed questionnaires. Four were excluded from analysis because the employees had not worked between 25 June and 6 July. Two (1.5%) of the 131 remaining employees met the case definition. This attack rate was significantly lower than the attack rates of the housestaff physicians (6.8%; P = 0.02) and the employees working in the physicians' dormitory (9.4%; P = 0.01).

Laboratory Investigations

Stool specimens were obtained from 17 of the ill house-staff physicians and 5 ill administrative employees working in the physicians' dormitory between 5 July and 7 August. Cultures and examination for ova and parasites all produced negative results. Modified acid-fast staining of stool specimens from 9 of the 17 housestaff physicians and 1 of the 5 employees showed many spherical organisms approximately 8 to 10 µ m in diameter. The CDC confirmed that these organisms were indistinguishable from those identified as Cyclospora in previous case reports. The ultraviolet illumination method at the CDC detected all specimens that were positive by acid-fast staining done in Chicago.

A follow-up stool specimen was collected from each of 15 recently symptomatic housestaff physicians between 8 August and 27 August. These included specimens from 7 physicians who had previously been positive for Cyclospora. One of these specimens did not show Cyclospora organisms; this specimen was collected approximately 5 weeks after the onset of the patient's clinical symptoms. Specimens from the other 6 physicians were Cyclospora positive. One of the 8 patients whose specimens had previously been negative for Cyclospora organisms had a follow-up specimen that was Cyclospora positive.

Thirty-four stool specimens were collected between 6 September and 5 October for additional follow-up. These included 15 specimens from 7 housestaff physicians who had previously been Cyclospora positive; 10 specimens from 10 physicians' dormitory employees who had reported drinking water from the dormitory but who had not become ill; and 9 specimens from 8 employees working in the physicians' dormitory and in the administrative building who had not been ill and who denied drinking water from the dormitory. Two stool samples from 2 housestaff physicians who had previously been positive for Cyclospora showed few Cyclospora organisms (approximately three per specimen). These specimens were collected approximately 8 weeks after the onset and more than 4 weeks after the resolution of illness. No other stool specimens, including additional specimens from the same 2 housestaff physicians obtained 9 weeks after the onsets of their illnesses, showed Cyclospora organisms.

Surveillance

Laboratories at the study hospital, seven other Chicago-area hospitals, and the Illinois Department of Public Health conducted surveillance for Cyclospora between 6 August and 27 August 1990. A laboratory at a hospital across the street from the study hospital detected one person with a Cyclospora-positive stool specimen; this has been previously reported [5]. The patient was a 35-year-old male construction worker who noted onset of illness on 8 July 1990. His symptoms included severe abdominal cramping, anorexia, and watery diarrhea with remission and relapse. His illness lasted more than 1 month. The patient's wife reportedly had similar symptoms during the same time period, but she never submitted a stool specimen. Both the patient and his wife denied having had any contact with the physicians' dormitory or the study hospital. The patient did report having cleaned out a saltwater aquarium by oral siphoning a few days before the onset of symptoms. The laboratory at the study hospital also detected Cyclospora organisms in the stool specimen of a patient who presented to the hospital outpatient clinic. He had no other known connection to the hospital or to the physicians' dormitory. This patient could not be located for follow-up. No other laboratories detected the organisms.

Environmental Investigation

The physicians' dormitory is a 17-story building next to the hospital. It was built in 1949 and has 236 rooms assigned to housestaff physicians and 66 on-call rooms. Approximately 165 administrative staff members work in several offices. The water supply for the physicians' dormitory comes from the Chicago municipal system into the basement and is pumped up to two storage tanks (each has a capacity of 4000 to 5000 gallons) located in a penthouse area on the 17th floor. Water from the two tanks flows into a common pipe that distributes it throughout the building by gravity.

The penthouse area was not sealed from the outside and had several broken windows without screens. Abundant sunlight was present in the penthouse area. The tops of the water storage tanks were not sealed and were covered with canvas. Bird feces were noted on the storage tank brim, on pipes located above the tanks, and on the canvas cover. The tank interiors had large amounts of rust. The outflow pipes were approximately 8 inches above the bottom of the storage tanks; below this level, water remained undisturbed during normal operations. No direct cross-connections were found between water and sewage lines.

Residents reported that tap water had not been available in the dormitory during the day on 4 July. Review of plumbing maintenance logs showed an interruption in water distribution on the evening of 4 July. A pump failure had been noted at 12:55 a.m. on 5 July. The pump had been repaired by 8:30 a.m. on 5 July.

Water samples collected on 20 July showed chlorine residuals of 0.2 to 0.4 mg/L throughout the building (acceptable level, more than equals 0.1 mg/L). Water samples collected on 16 August showed chlorine residual levels of 0.0 mg/L in a drinking fountain on the 13th floor and 0.1 mg/L in the bottom of the south storage tank. All water samples had total coliform counts of less than equals 1 colony/100 mL. Water samples collected on 20 July and 16 August had insignificant heterotrophic plate counts. Light microscopy examination of water collected from the bottom of the south storage tank showed that algae, principally diatoms (Cyclotella, Fragilaria, and Tabellaria species), were growing on iron deposit sediments. Cyclospora organisms were not detected in any samples.

Discussion

This is the first reported outbreak of Cyclospora-associated diarrhea in the United States. Cyclospora in this outbreak was identified because the hospital served many patients with the acquired immunodeficiency syndrome and routine modified acid-fast staining of all stool specimens had been implemented in the hospital laboratory to screen for cryptosporidia. It is not certain that Cyclospora caused this outbreak, because the data do not exclude the possibility that these organisms were a marker for the presence of the true pathogen. However, because initial stool studies did not identify a bacterial or parasitic agent and because the clinical course was marked by a distinctive cycle of relapses and remissions, a new or previously unrecognized agent may have been responsible. Both the disappearance of Cyclospora from the stool specimens of affected patients by 9 weeks after onset of clinical symptoms and the absence of Cyclospora in well persons suggest that this agent was involved.

Tap water from the physicians' dormitory was probably the source of the outbreak. Among housestaff physicians, illness was strongly associated with drinking tap water from the physicians' dormitory. Illness among employees who worked in the physicians' dormitory was also associated with exposure to tap water. Although illness among housestaff physicians was also associated with attending the house staff association party held on 29 June, this association was probably due to the correlation between attending the party and being exposed to dormitory tap water. Because the party was held in the physicians' dormitory, those who attended the party were also more likely to be exposed to the dormitory water. Also, only two of the seven ill employees who worked in the dormitory reported having attended the party, although all reported having had exposure to the tap water. The higher attack rates for housestaff physicians and dormitory employees compared with administrative building employees, as well as the higher attack rates for postgraduate year 1, 2, and 3 physicians (who would be expected to spend more time in the dormitory), further support the association between the physicians' dormitory and illness.

The water storage tanks were not sealed from environmental contamination, and the stagnant zone at the bottom of the tanks could have permitted the growth or multiplication of microorganisms. Sporocysts of Cyclospora are not buoyant and sink quickly when suspended in water; thus, they might have collected in the stagnant zone of the storage tanks. After the pump failed and the storage tank emptied, the water at the bottom of the tank would have been stirred up when the pump was fixed and water splashed into the tank. This stagnant water would then have entered the system. Data on the number of persons served by such systems are not available; however, these systems are probably most common in older city buildings.

Cyclospora organisms were not detected in the water tank or plumbing system, and several explanations are possible. Only small volumes of water were collected for examination, and a dilution effect may have required a much larger collected volume to detect the organisms. Also, infection occurred during a brief period, suggesting that contamination may have been an isolated incident. After the stagnant zone of the storage tank was stirred up, the Cyclospora organisms may have been washed out.

More widespread exposure to Cyclospora was suggested by the finding of two Cyclospora-positive persons without known links to the physicians' dormitory or the hospital. These may have been sporadic cases caused by exposure to other sources. It is also possible that the municipal water system had low-level contamination and that the condition of the storage tanks permitted the proliferation or amplification of the etiologic agent. A good filtration system should eliminate Cyclospora organisms, which may explain why infections with these organisms are less common than infections with the smaller Cryptosporidium species. Clusters of illness associated with Cyclospora may indicate problems with water filtration.

Current knowledge about Cyclospora is limited. The organisms are 8 to 10 µ m in diameter and do not take up most laboratory stains. However, with the modified acid-fast stain, many Cyclospora organisms stain deep red, some stain pink, and others remain unstained and present as glassy, membranous cysts [9]. In fresh, unpreserved stools, Cyclospora organisms appear as nonrefractile hyaline cysts by light microscopy. Cyclospora organisms autofluoresce strongly under ultraviolet illumination, appearing as neon blue circles. Acid-fast stains work best with fresh stool specimens. In formalin-preserved stools, the organisms become progressively refractory to staining and may not stain at all, so they may be missed if few are present. Ultraviolet illumination, however, does not require any staining and can be used with specimens that are fresh or preserved in formalin or polyvinyl alcohol. Cyclospora can be readily differentiated from the more common Cryptosporidium on the basis of their larger size. The small bowel has been identified as the probable site of infection [6, 17]. Several antibiotic therapies have been tried with varying success, and some data suggest that clinical improvement occurs with trimethoprim-sulfamethoxazole therapy [4, 5, 18-21].

Assuming that Cyclospora caused this outbreak, our investigation provides additional data about the clinical features associated with Cyclospora-related illness. Most notable are the watery diarrhea, the cramping abdominal pain, and the cycle of remissions and exacerbations. The 5-day median duration of diarrhea reported in our cohort studies is much shorter than the 43 ± 24 days of illness reported in a study of immunocompetent patients with Cyclospora infection in Nepal [4]. Our study may underestimate the duration of total illness, because extended interviews with six of the Cyclospora-positive case-patients showed that one reported only the length of his first episode of diarrhea on the questionnaire, although the entire course of his illness (including remissions and exacerbations) as reported in the interview was much longer. Also, we asked about duration of diarrhea only, yet at least one case-patient continued to have illness and gastrointestinal discomfort without diarrhea for more than 2 months. The longer duration of illness in Nepal could have been due to the selection of those with more severe illness, because all patients in that study [4] sought medical attention compared with only 6 (43%) of the 14 housestaff physicians in our study. It is also possible that fewer organisms were ingested in the Chicago outbreak than in other settings. If the water pump failure and subsequent contamination of the dormitory water supply precipitated the outbreak, our findings suggest that Cyclospora organisms may survive in stagnant water and that the incubation period for Cyclospora may be as short as 12 to 24 hours. Follow-up examinations suggest that Cyclospora may persist in stool specimens for as long as 8 weeks after onset of symptoms.

Some limitations of our study should be noted. First, there was much publicity and discussion about the water in the physicians' dormitory before the questionnaires were distributed, and this may have biased the responses. However, because many of the housestaff physicians either spent call nights in or lived in the dormitory, exposure to dormitory tap water would be expected. Second, the cohort studies were done 1 month after the outbreak; thus, recall of exact dates and other details may be inaccurate. However, we tried to relate recall of dates to memorable events (the 4 July holiday and the 1 July change of clinical rotations). Third, because stool specimens from only 12 of the 23 case-patients were examined both for other pathogens and Cyclospora, some of the illnesses may have been caused by other agents.

The only other Cyclospora-related outbreaks reported in the literature have been seasonal outbreaks in Nepal [4, 22, 23]. Investigation of a 1990 outbreak identified Cyclospora organisms on one head of lettuce, from which a patient had eaten 2 days before onset of illness [22]. Investigation of a June 1994 outbreak among British soldiers and dependents stationed in Nepal included the identification of Cyclospora in a 2-L water sample taken from a chlorinated water storage tank and concentrated by filtration through a 5-µ m Millipore filter [23]. This supports our finding that Cyclospora infection may be acquired through contaminated water and suggests that Cyclospora organisms may be resistant to chlorine.

We conclude that Cyclospora probably caused this diarrheal outbreak through a contaminated water supply. Cyclospora may be a new or previously unrecognized human enteric pathogen capable of producing acute and intermittent diarrhea. It should be considered in assessments of patients with unexplained, prolonged diarrheal illness.

Dr. Weber: Centers for Disease Control and Prevention, Mailstop E-46, 1600 Clifton Road NE, Atlanta, GA 30333.

Dr. Sosin: National Center for Injury Prevention and Control, Mailstop F-41, 4770 Buford Highway NE, Atlanta, GA 30341-3724.

Dr. Griffin: Centers for Disease Control and Prevention, Mailstop A-38, 1600 Clifton Road NE, Atlanta, GA 30333.

Dr. Long: Centers for Disease Control and Prevention, Mailstop G-11, 1600 Clifton Road NE, Atlanta, GA 30333.

Dr. Murphy: 55 Sharon Road, Fairview, NC 28730.

Dr. Kocka and Ms. Peters: 627 South Wood Street, Chicago, IL 60612.

Dr. Kallick: 16411 135th Street, Lemont, IL 60439.