|
|
| |||||||||||||||||||||||||||||||
Background In the summer of 1991, simultaneous outbreaks of bloody diarrhea and hemolytic-uremic syndrome caused by Escherichia coli O157:H7 and of bloody diarrhea caused by Shigella sonnei were traced to a lakeside park near Portland, Oregon.
Methods We identified cases primarily from routine surveillance reports. In case-control studies, the activities of persons with park-associated E. coli O157:H7 or S. sonnei infections were compared independently with those of three sets of controls. We also evaluated environmental conditions at the park and subtyped the bacterial isolates.
Results We identified 21 persons with park-associated E. coli O157:H7 infections (all of them children; median age, six years) and 38 persons with S. sonnei infections (most of them children). These 59 people had visited the park over a 24-day period. Their illnesses were not associated with food or beverage consumption. All the case patients reported swimming, however, and in case-control studies swimming was strongly associated with both types of infection (P = 0.015 or less). The case patients were more likely than the controls to report having swallowed lake water, and they had spent more time in the lake. Numbers of enterococci indicative of substantial fecal contamination (geometric mean, >50 per deciliter) were detected in the swimming area during some but not all of the outbreak period. Park-associated E. coli O157:H7 isolates were identical by pulsed-field gel electrophoresis and were distinguishable from other isolates in the Portland area.
Conclusions Lake water that was fecally contaminated by bathers was the most likely vehicle for the transmission of both the E. coli O157:H7 and the S. sonnei infections. The unusually prolonged outbreak suggests both the survival of these enteric organisms in lake water and a low infectious dose.
In 1987, Washington became the first state to mandate public health reporting of E. coli O157:H7 infections by physicians and diagnostic laboratories. Oregon instituted a similar requirement in August 1990, and all case reports are now followed with in-depth standardized interviews of patients that include questions about possible sources of infection.
In late July 1991, the number of routine surveillance reports of E. coli O157:H7 infection increased in the Portland area. At the same time, case reports of shigellosis, which in previous months had primarily come from one urban neighborhood, began to appear throughout the metropolitan area. Initial interviews revealed that many patients with both infections had recently visited a nearby county park. We conducted an investigation to identify the extent of the outbreak, the source of infection, and means of control.
Methods
Blue Lake Park is a multi-use county facility, lying along the shore of a banana-shaped lake some 1.4 km long. The lake is fed by underwater springs and has no outlet. The park facilities include picnic grounds, ball fields, concession stands, and a swimming beach. In July 1991, hot weather encouraged park use, with as many as 10,000 weekend visitors.
Definitions and Case Finding
We defined case patients with E. coli O157:H7 as persons with positive stool cultures for E. coli O157:H7 or serologic evidence of E. coli O157:H7 infection and either bloody diarrhea or hemolytic-uremic syndrome.
Case patients with shigellosis were defined as persons with a positive stool culture for Shigella sonnei or diarrheal illness and a household contact who was culture-positive for S. sonnei.
Park-associated case patients were defined as subjects whose symptoms began 1 to 10 days (in the case of E. coli O157:H7 infections) or 1 to 4 days (in the case of shigellosis) after visiting the park. Persons whose symptoms began two or more days after another household member's illness were considered possible secondary case patients and were excluded from the analysis.
Case patients were identified from routine surveillance reports or through follow-up of these and other reports to local health departments. Once the outbreak was recognized, we encouraged area physicians and laboratories to consider E. coli O157:H7 infections and to perform appropriate cultures.
We interviewed all residents of the four-county Portland area (Multnomah, Clackamas, and Washington counties in Oregon and Clark county in Washington) with reported E. coli O157:H7 or S. sonnei infections and onset of illness from July 1 through August 20, 1991.
E. coli O157:H7 -- Case-Control Studies
Using a standardized questionnaire, we obtained information from all park-associated case patients (or their parents) about their demographic characteristics, illness, and possible sources of infection at the park (e.g., consumption of foods from concession stands, drinking water, and swimming). To determine risk factors for illness among persons who visited the park, we conducted case-control studies comparing case patients independently with three different sets of control groups, as follows.
Exit Controls
On August 3, before there was any publicity about park-associated illness, we stopped automobiles as they left the park and interviewed one child in each automobile who was under 18 years of age (if one was present).
Friend Controls
We questioned all persons who came to the park in the same automobile as a case patient.
Picnic Controls
From the park administrators, we obtained a list of more than 200 groups that had reserved picnic sites during the period of the outbreak. We contacted a convenience sample of the smallest groups (<100 persons) that had visited the park from July 12 through August 3, and attempted to question all children under the age of 18 in those groups.
The potential controls were interviewed about their exposures in the park and (except for the exit controls) any subsequent illness; those reporting diarrhea or vomiting within 10 days after their park visit were excluded from the study.
Shigellosis -- Case-Control Studies
Data on demographic variables and brief swimming histories were collected for all case patients with park-associated shigellosis. For a random sample of the households in which there was a case patient, the first person to become ill was interviewed with the same questionnaire used for the patients with E. coli infections. The shigellosis case patients were compared with the exit and picnic controls described above and with analogous friend controls.
Environmental Investigations
Relevant park operations were reviewed, including the food-handling procedures used by concessionaires. Drinking water was tested for evidence of fecal contamination. Sanitarians inspected the sewer connections from toilets at the swimming beach to assess their integrity and also tested them with fluorescein dye. Before the outbreak, water samples from the swimming area were routinely collected every two weeks and were assayed for enterococci by standard methods18. After July 15, sampling was more frequent. Lake water collected on August 8 was tested for E. coli O157:H7 and shigella by inoculation into trypticase soy broth and overnight incubation, before plating on standard sorbitol-MacConkey and salmonella-shigella agar plates.
Laboratory Studies
Microbiologic Identification
Isolates of E. coli O157:H7 and S. sonnei obtained from area laboratories were identified by standard methods19. The persons doing the isolate subtyping and antibody assays were unaware of the illness and exposure histories of the subjects.
Serologic Testing
Serum specimens obtained by venipuncture were assayed for antibodies to E. coli O157:H7 lipopolysaccharide antigens as described elsewhere20. Titers of 80 or higher were considered positive.
Subtyping of E. coli O157:H7
Isolates of E. coli O157:H7 were characterized with regard to the production of shiga-like toxins as described elsewhere21. Antibiotic-resistance patterns were determined by standard Kirby-Bauer disc-diffusion methods. A sample of park-associated isolates was compared with other isolates obtained from Portland-area patients between July and September 1991 by pulsed-field gel electrophoresis (PFGE) of genomic DNA, essentially as described elsewhere22. DNA was digested with XbaI and separated on a CHEF DR II apparatus. Electrophoresis through 1 percent PFGE agarose was carried out for 22 hours at 12 °C in 0.5x TBE buffer (44.5 mM TRIS, 44.5 mM boric acid, and 1 mM EDTA) at 200 V, with a linearly ramped switching interval of 5 to 50 seconds.
Subtyping of S. sonnei
A sample of park-associated isolates was compared with other Portland-area isolates obtained between June and August. Isolates were typed by whole-plasmid analysis, analysis of plasmid DNA digested by HindIII, and restriction-enzyme (SpeI) analysis of genomic DNA with the use of PFGE as described elsewhere23,24.
Statistical Analysis
In univariate analyses, we compared observed with expected distributions by the chi-square test or, when expected cell sizes were less than five, using a two-tailed Fisher's exact test. Median swimming times were compared with use of the Mann-Whitney U test25. P values less than 0.05 were considered to indicate statistical significance.
Results
E. coli O157:H7 -- Descriptive Epidemiology
We identified 21 case patients with park-associated E. coli O157:H7 infections from 19 households; of these infections, 18 were confirmed by stool culture and 3 by serologic testing. All the case patients were children (median age, 6 years; range, 1 to 16) who had visited the park between July 11 and August 3 (Figure 1). The median incubation period was 4 days (range, 1 to 10). All the case patients were symptomatic and were treated by a physician. Seven children were hospitalized, including three with hemolytic-uremic syndrome. We identified seven additional cases, presumably the result of secondary transmission, in four of these households.
|
Diarrheal illness developed within six days of a visit to the park in 10 of 100 potential picnic controls (10 percent) who had gone swimming, as compared with 3 of 157 nonswimmers (2 percent) (P = 0.004). These persons and four others who became ill 7 to 10 days after visiting the park were excluded from the analysis. In addition, 17 potential friend controls were excluded because of illness.
E. coli O157:H7 -- Case-Control Studies
The 21 case patients were compared with the 32 exit, 52 friend, and 240 picnic controls. All 21 case patients had been swimming, and swimming was strongly associated with illness (Table 1). No other type of exposure was consistently associated with illness. In comparison with the picnic (but not the exit or the friend) controls, illness was also associated with the consumption of shaved ice. Only eight case patients (38 percent) reported consuming shaved ice, however.
|
Shigellosis -- Descriptive Epidemiology
A total of 38 case patients with park-associated primary shigellosis (confirmed in 28 by stool culture) from 30 households were identified. They visited the park between July 14 and August 4 (Figure 1) and became ill a median of two days later. Most case patients were children (median age, 6 years; range, 1 to 32). Three patients were hospitalized. At least 36 secondary cases were also identified. Brief interviews were conducted with all case patients; 13 were interviewed in depth.
Shigellosis -- Case-Control Studies
The results of the case-control studies involving case patients with shigellosis were similar to those described for case patients with E. coli O157:H7 infections. All 38 patients with shigellosis had been swimming, and swimming was again strongly associated with illness (Table 1). No other type of exposure in the park was associated with illness. Among bathers, the case patients were more likely than the controls to report swallowing lake water.
Environmental Investigations and Control Measures
No irregularities were identified in the operations of the park or concession stands. The bathroom facilities at the swimming area were well maintained and of adequate capacity. No sewage leaks were revealed by dye testing. No irregularities in sewage disposal had been identified during a sanitation survey of houses on the far side of the lake -- coincidentally, conducted a month before the outbreak. No cattle or other livestock are kept near the park.
Although testing was infrequent, before July 22 enterococcus counts measured at the swimming area were generally within the limits recommended for recreational waters by the Environmental Protection Agency (Table 2)26. These limits were exceeded on July 22 and 29, but not on August 1 or later. Concentrations of enterococci were consistently lower in the deeper portion of the swimming area (data not shown) and were negligible near the center of the lake.
|
Bacterial Subtyping
E. coli O157:H7
All 21 park-associated isolates of E. coli O157:H7 had identical patterns of antibiotic resistance and produced shiga-like toxins 1 and 2. Sixteen isolates of E. coli O157:H7, including six that were park-associated, were subtyped by PFGE. Eight patterns were distinguished. The six park-associated isolates were identical to each other, but were distinct from all other isolates tested (Figure 2).
|
Six park-associated primary or secondary isolates were compared with 10 other isolates from the Portland area. Of the 16 isolates tested, all but 1 (not park-associated) were identical by PFGE of chromosomal DNA. With the same exception, only statistically insignificant differences in banding (coefficient of similarity, >0.9) were seen in plasmid digests.
Discussion
This outbreak was caused by the ingestion of fecally contaminated lake water. Among park visitors, exposure to lake water in the swimming area was strongly associated with both E. coli O157:H7-caused illness and shigellosis. All those affected reported swimming. Furthermore, among bathers, a history of swallowing lake water was associated with illness, and the case patients tended to spend more time in the water than the controls.
No other plausible source of infection at the park was consistently associated with illness. The association of consumption of shaved ice with illness, found in one of the six case-control comparisons, probably reflects the fact that children attending organized picnics were less likely to buy items at concession stands. Only a few of the case patients ate shaved ice.
Because there were no records of visitors to the park, it was not possible to select a random sample of visitors as controls. We selected a number of sets of controls from independent populations in order to minimize the likelihood of being misled by bias in any one group. Each group has its limitations. The friend controls, for example, tend to be overly matched with regard to potential exposures. Recall bias may well have inflated the association of illness with swallowing lake water or with time spent in the water, but the effects of such bias on the overall association with swimming would be minimal. The strength and consistency of our findings, together with the biologic plausibility of the association, argue that the observed relation of exposure to lake water with both infections was causal. No additional cases were reported among persons who visited the park after swimming was banned on August 5.
The most likely source of the two pathogens in lake water was fecal pollution by bathers, among whom were many toddlers not yet toilet trained. The finding of elevated numbers of enterococci in the swimming area, but not elsewhere in the lake, supports this conclusion. No other potential source of these organisms was identified.
Water in recreational areas has not been widely recognized as a potential source of E. coli O157:H7. Indeed, we could find only a single report in which this organism was isolated from a natural body of water27. Shigellae, on the other hand, though rarely detected in water samples, have caused a number of swimming-associated outbreaks, generally in small, relatively stagnant bodies of water similar to Blue Lake28,29,30,31,32,33. In virtually all these outbreaks, the source was believed to be fecal contamination by other bathers.
At Blue Lake, the transmission of both organisms continued for more than three weeks. Except when there is a source of ongoing contamination, such as a sewage outfall,29 prolonged outbreaks in swimming areas are rarely documented. More commonly, transmission lasts only a few days,32,33,34,35 or two weeks in one case31 -- presumably reflecting the rapid dilution and dying off of potential pathogens. Infection after the incidental ingestion of small amounts of water is most likely if the infectious dose is low, as is well recognized in shigellosis36 and as has been inferred in E. coli O157:H7 infections37. Furthermore, prolonged transmission implies repeated contamination or the survival of the organisms. Shigella and E. coli can survive in lake water for weeks or months38. One outbreak-associated isolate of E. coli O157:H7 remained viable in tap water for more than a month12. The characterization by PFGE of a distinct epidemic strain of E. coli O157:H7 is consistent with the persistence of an initial inoculum, although we cannot rule out the possibility of reintroduction of the outbreak strain by a sick, convalescing, or asymptomatically infected bather. Park-associated isolates of S. sonnei were likewise indistinguishable, but the homogeneity of contemporary Portland-area isolates is consistent with multiple hypotheses about contamination.
No information was available about daily numbers of bathers, so we could not calculate incidence rates. That 10 percent of the potential picnic controls we contacted reported diarrheal illness indicates that the overall attack rate among bathers may have been appreciable. There were hundreds of calls from park visitors about undiagnosed diarrheal illness. The shape of the epidemic curves suggests that transmission was tapering off by the time the lake was closed.
Outbreaks of enteric disease involving multiple pathogens are rarely reported. In a previous swimming-associated outbreak of shigellosis, both S. sonnei (29 isolates) and S. boydii (4 isolates) were cultured from bathers32. The finding of identical modes of transmission for both S. sonnei and E. coli O157:H7 at Blue Lake underscores the fact that these organisms share many characteristics, epidemiologic as well as biologic and clinical.
Routine monitoring of water in recreational areas for evidence of fecal contamination is not required in Oregon, and this experience gives us little reason to modify that policy. During the outbreak, levels of indicator organisms did not consistently exceed the limits recommended by the Environmental Protection Agency. Other studies confirm that the correlation of indicator levels with the risk of enteric disease is far from clear-cut39,40. Moreover, levels may fluctuate considerably during a single day41 or may vary between samples from nearby sites, suggesting that infrequent sampling may yield misleading results. In any event, the inevitable lag between collecting the sample and obtaining the results limits the value of testing for day-to-day risk management. Bacterial sampling may be useful in long-term assessments of the effects of environmental, behavioral, or other interventions at specific sites.
Parents and others may not realize that bodies of water in recreational areas are a possible source of many enteric infections, some of which have potentially grave outcomes. Many such infections may occur sporadically, rather than in recognizable clusters. Primary prevention requires the elimination of fecal contamination at swimming areas -- an effort that may depend more on public cooperation than on governmental regulation. Having adequate, well-maintained toilet facilities does not guarantee that bathers will use them. Children who are not toilet trained pose a danger to others at swimming areas, and they should be provided with alternative opportunities for playing in water.
The swimming area at Blue Lake reopened the following summer, after the installation of a permanent pumping system to increase water circulation. Infants and toddlers are no longer allowed in the lake. A water-spray area was constructed for these youngsters near the beach; it drains to the city sewers. These modifications, coordinated with educational measures designed for older children and parents, should reduce the likelihood of recurrences.
This cluster of E. coli O157:H7 infections was among the first to be identified from the collation of routine surveillance reports, and it shows the value of surveillance for communicable diseases. With cases scattered over four counties, the outbreak would probably have gone unnoticed had the cases not been routinely reported to public health agencies and investigated by them.
We are indebted to the members of the Multnomah, Clackamas, and Washington County Health Departments; the Southwest Washington Health District; the Multnomah County Parks Services Division; and the Center for Disease Prevention and Epidemiology, Oregon Health Division, all of whom contributed greatly to this investigation, and particularly to Steve Mauvais, Robert Sokolow, and others at the Center for Public Health Laboratories of the Oregon Health Division; to Dr. Philip Miller, Dr. Phillip Tarr, Dr. Patricia Griffin, Dr. Robert Tauxe, Joy Wells, and Nancy Strockbine for their useful contributions; and to hundreds of park visitors for their cooperation.
Source Information
From the Center for Disease Prevention and Epidemiology, Oregon Health Division, Portland (W.E.K., J.M.M., F.C.H., L.P.W., K.H., D.W.F.); the Epidemic Intelligence Service (J.M.M.) and the Enteric Diseases Branch (T.J.B.), Centers for Disease Control and Prevention, Atlanta; the Multnomah County Health Department, Portland, Oreg. (G.L.O.); and the Department of Pathology, Oregon Health Sciences University, Portland (M.A.P.).
Address reprint requests to Dr. Keene at the Oregon Health Division, 800 NE Oregon St., Suite 772, Portland, OR 97232.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
