A beach-associated outbreak of Escherichia coli O157:H7.

Introduction

Each year approximately 20,000 people in the United States become ill from exposure to Escherichia coli O157:H7; some develop hemolytic uremic syndrome (HUS) (1-3). Most outbreaks are associated with eating contaminated food, whereas waterborne outbreaks are rarely reported. In July 1995, the Illinois Department of Public Health (IDPH) investigated an outbreak in which 12 children who had gone swimming at a bathing beach in Rock Cut State Park in northern Illinois became ill with E. coli O157:H7 infections. The beach was closed on July 6, 1995, and, since E. coli has been reported to survive in lake water for weeks or months, the beach remained closed for the remainder of the summer (4).

Methods and Results

Cases

On July 5, 1995, Winnebago County Health Department reported an initial cluster of six E. coli O157:H7 cases, five of which were children. Four of the children were from Rockford, Illinois, and one child was a resident of Wisconsin. The sixth case, an adult, was a resident of Freeport, Illinois. This case was later determined to be unrelated to the lake cases. Interviews with the children's parents revealed no common food source, but all of the children had visited Rock Cut State Park on June 24 or 25. The park contains a lake swimming beach operated by the Illinois Department of Natural Resources (DNR).

A case was defined as a resident of or a visitor to the Rockford area who had at least one of the following in the week from June 25 through July 1:

* bloody diarrhea,

* HUS,

* serologic E. coli O157:H7 confirmation by the Centers for Disease Control and Prevention (CDC), or

* culture-confirmed E. coli O157:H7 and pulse field gel electrophoresis (PFGE) pattern consistent with lake-associated cases.

Additional cases were identified through active surveillance in clinical laboratories, from citizen calls following media coverage, and by asking cases if they knew of any additional ill people. Twelve children (including the initial five) met the case definition. Seven stool isolates of E. coli O157:H7 were sent to CDC to test for shiga-like toxin and for PFGE. All stool isolates that came from cases visiting Rock Cut State Park (n = 6) had identical PFGE patterns, and all six isolates produced Type I and Type II shiga-like toxins. The stool isolate from the Freeport adult had a different PFGE pattern. Serum samples from individuals with no positive stool cultures but with symptoms suggestive of E. coli O157:H7 infection were sent to CDC for antibody titer testing.

Of the 12 cases, seven were culture confirmed as having E. coli O157:H7 infections; one had HUS and was culture-confirmed as having an E. coli O157 infection, but H typing was not performed; three had positive serology; and one had bloody diarrhea without being culture-confirmed. Cases from whom stool cultures were obtained (n = 8) were negative for Salmonella, Shigella, and Campylobacter. Two families each had two children affected with E. coli O157:H7. Eight children were from two to six years of age, and four were from seven to 12 years of age. Seven children were males.

The incubation period for the cases ranged from three hours to seven days with a mean of 3.7 days [ILLUSTRATION FOR FIGURE 1 OMITTED]. One child who swam on Saturday (June 24) experienced symptoms three hours after swimming. The symptoms lasted 36 hours and recurred seven days after swimming. Nine of the children experienced bloody diarrhea. Three of the children (two, four, and five years of age) developed HUS and were hospitalized for a minimum of one month. The attack rate for illness in 4,700 lake swimmers on June 24 and June 25 was 0.25 percent. The identification of cases was enhanced by intense media coverage in the Rockford area. All 12 cases had swum at the lake.

Environmental Investigation

The park contains two connected, manmade lakes. The lake with the bathing beach is fed by a stream; an outflow (spillway) connects it to the second lake, where swimming is not permitted. No cases occurred among people visiting only the second lake. There is a large family campground for tents and recreational vehicles at the second lake. The upper lake and bathing beach were constructed in 1989 on approximately 350 acres of farmland acquired by the Illinois Department of Conservation in 1986. The acquired land is known as the Olson Annex to Rock Cut State Park. The Olson Lake beach opened on Memorial Day weekend in 1990.

The lake, approximately 50 acres in size, is shallow, with a maximum depth of about 10 feet. The main source of water feeding into the lake is Willow Creek, a very small stream with low flows most of the year. A few minor springs were reported during the lake's construction, but they are not considered to be a major source of water for the lake. Approximately 0.5 to 1 inch of water was observed flowing over the spillway during the months of July and August 1995, indicating a low flow of water through the lake.

The sand portion of the beach is approximately 350 feet long and an average of 70 feet wide. Over the five years between the lake's opening and the outbreak, the area roped off for wading and swimming had become smaller because of siltation, the shifting sand bottom, and elimination of the outer swimming area for safety reasons. The buoys and ropes are now approximately 30 feet from the shoreline.

Facilities at the beach consist of a shower and clothes-changing building and a concession trailer. A concrete vault privy is located at each end of the beach, approximately 25 feet from the edge of the sand. Each privy has a separate stool seat for each sex. No handwashing or diaper-changing facilities are available. The shower building and concession trailer are centrally located approximately 140 feet from the shoreline. Normal daily attendance is about 1,000 people, and the maximum number is about 2,500.

The facilities were examined to determine if they could have been a source of the illnesses associated with the bathing beach. Pumping records for the vault privies were checked against content levels in the vaults. To further verify that the vaults were not leaking, they were filled with water to within inches of the top and checked periodically. The test showed that the privy vaults were watertight. The septic tank and drain field for the shower building and concession trailer are located in lower ground southwest of the dam, a location that precludes them from having an effect on Olson Lake. The tank was opened to confirm that it was part of the shower building system. There were no solids or sewage odors in the tank, which was a little more than half full. Of 40 people questioned about their trips to the beach on June 24 or June 25, 28 reported no lines at the toilet. Three individuals did report lines of 10 to 12 people waiting to use the facilities.

A plumbing inspection of the concession trailer was conducted to check for possible cross-connections. The concessionaire was questioned concerning employee illnesses and food handling and preparation procedures. Water and ice samples were taken, and the trailer was inspected. All samples were satisfactory, and the inspection and interview revealed no likely sources for the E. coli O157:H7 infections.

Another element of the environmental investigation was an evaluation of the watershed. Since Olson Lake is located in a state park, the immediate watershed is in a natural state, except for the beach, and there is nothing to negatively affect lake water quality. A United States Geological Survey map was used to find the watershed outside the park that might affect lake water quality. This watershed was checked for a distance of 3 to 4 miles for possible sources of pollution. Immediately east of the park boundary is a recently developed subdivision and the small unincorporated town of Argyle. Both were checked for failing septic systems or other visible pollution sources. County health department records were reviewed for past septic failures. Drainage ways from the subdivision and Argyle to Willow Creek are heavily grassed, which should greatly reduce adverse effects of runoff on water quality of the creek and lake.

To the east beyond the residential development, the watershed is primarily cropland. On a farm approximately 0.75 to 1 mile from the park, eight head of cattle were noted. The small size of the farm made it unlikely that it could support more cattle than those observed. Because of the number of cattle and the nature of the pasture (grass with no discernible slope), this farm would not be expected to produce any runoff that could adversely affect Willow Creek. IDPH records were checked to verify visual observations that there are no dairy farms in the Willow Creek watershed. Three analyses of samples taken near the inlet to Olson Lake after the beach was closed had fecal-coliform counts ranging from fewer than two to six per 100 milliliters (mL). Three inlet samples were tested for E. coli and fecal streptococcus, neither of which was found (fewer than 100 E. coli and fewer than two or fewer than 10 fecal streptococci per 100 mL). Four inlet samples were tested for E. coli O157:H7, and all were negative.

Sampling

The State Swimming Pool and Bathing Beach Code specifies the water quality requirements for licensed bathing beaches in Illinois. Total-coliform bacteria counts of 1,000 per 100 mL or fecal-coliform counts of 100 per 100 mL in any two consecutive samples constitute sufficient grounds for additional investigation, survey, special analysis, and correction of any problems determined to be causing the high counts. Total-coliform counts exceeding 5,000 per 100 mL or fecal coliform counts exceeding 500 per 100 mL in any two consecutive samples constitute sufficient grounds to close the beach. Research has confirmed that 92 to 95 percent of the fecal-coliform colonies isolated in beach water are E. coli (5). The E. coli or fecal-coliform count is a direct measure of fecal contamination and is more applicable for evaluation of beach water quality than is a total-coliform count. Furthermore, IDPH laboratory investigations have discovered that the chromogenic substrate method for coliform analyses will show certain noncoliform organisms. For these reasons, IDPH discontinued use of the total-coliform standard for regulatory purposes. Rule changes adopted in 1996 have also deleted the total-coliform standard for bathing beaches.

Routine samples were collected by DNR staff on June 21,1995. Results of those samples were 640 and 990 E. coli per 100 mL. Upon receipt of those results, IDPH advised DNR to collect additional samples. Samples were collected on June 29 and July 5 and analyzed for coliform bacteria by the Winnebago County Health Department. A qualitative result of "present" was obtained on the June 29 sample, and high levels of coliform bacteria were found in the three samples collected on July 5. Since the laboratory was local, it was able to provide a quick turnaround time on those samples, but it did not have the capability of performing fecal-coliform or E. coli analyses.

According to the National Weather Service, the period from June 6 to June 26 was hot and dry in the Rockford area. High temperatures averaged 86 degrees during the period - eight days had temperatures 90 [degrees] or above - with 0.23 inches total rainfall. Therefore, it is unlikely that runoff from any source contributed to high bacteria counts in the beach samples collected on June 21 or to exposure to E coli O157:H7 on June 24 or June 25.

After the beach was closed on July 6, IDPH began an extensive water-sampling program. The bathing beach was divided into three sections: east, center, and west. Water samples were taken from each section and from the lake's inlet and outlet waters. Only two of the 93 analyses after the beach was closed had fecal coliform or E. coli results above 500 per 100 ink. Both of those were obtained from samples collected on July 10, 1995; one from the east and one from the west sampling location. E. coli O157:H7 was not found in any water sample. Nine samples of waterfowl feces were also collected and analyzed for E. coli O157:H7; the findings were negative. A total of 51 analyses were performed for fecal streptococcus, an organism often associated with animal waste. There is no bathing-beach standard for this organism. Only three of the 51 analyses of bathing-beach water found more than 100 fecal-streptococcus colonies per 100 mL: 206 per 100 mL in a sample collected from the center location on July 25, 216 per 100 mL in a sample collected from the west location on August 1, and 150 per 100 mL in a sample collected from the east location on August 15.

Concentrations of potentially harmful bacteria have been reported to be higher in bottom sediments than in the water itself (6). A total of 24 sediment samples were collected from July 18 through August 1, 1995, in two ways. The first was to vigorously agitate the bottom and take a sample of the resulting turbid water. The second was to plunge an inverted sample bottle into the bottom and replace the entrapped air with bottom sand. Fecal-coliform results (six total) were all very low, ranging from six to 30 colonies per 100 mL. Seventeen of 18 E. coli results had fewer than 100 colonies per 100 mL, and the other result was 200 colonies per 100 mL. Sixteen of 18 fecal-streptococcus results had 100 colonies or fewer per 100 mL. The other two results were 176 per 100 mL and over 300 per 100 mL. All 18 sediment samples tested for E. coil O157:H7 were negative.

Epidemiologic Investigation

After the initial investigation by the Winnebago County Health Department and IDPH implicated the lake at Rock Cut State Park, a matched case-control study was conducted to firmly establish whether swimming in the lake was associated with illness. To establish a control group, parents of each case were asked to provide the name of a neighbor or friend with a child who was within one year of the age of the case. The child control was required to have had no symptoms compatible with E. coli O157:H7. A questionnaire about restaurants, foods eaten from the concession stand at the lake, other recreational activities, and park exposure during the week prior to illness was administered to case and control parents. Five of the 12 cases did not participate: Three children who were unrelated to each other were identified later in the investigation, and the parent of two sibling cases refused to be interviewed. In a matched analysis of information from the seven cases and seven controls, only swimming at the park was found to be a risk factor for illness (seven cases swam at the lake, seven controls did not swim at the lake). Common food items, restaurants, and other recreational parks in Rockford were not risk factors. Only one case ate food from the concession stand at the lake.

A second case-control study was performed to identify possible risk factors for illness and to establish if E. coli O157:H7 exposure from swimming at the lake could have caused other illnesses. The parents of 10 cases agreed to cooperate. To obtain a control group, individuals who had camped at the park during the weekend of June 24 and 25 were identified. State park personnel do not record any information about individuals who pay to visit the beach. They do, however, record the names and license plate numbers of groups that camp at Rock Cut State Park. The park permit also lists the number of individuals camping at each site. Of the 286 families that camped at Rock Cut State Park on the weekend of June 24 and 25, 160 had more than two campers.

Permits on which more than two campers were listed were submitted to the Illinois Secretary of State's Division of Motor Vehicles to obtain addresses. Phone numbers were sought through telephone company information operators. Sixty camper families were contacted about their visits to the park. Many lived outside the Rockford area and had not heard of the E. coli 0157:H7 illnesses associated with Rock Cut State Park.

Parents both of cases and of camper children were asked if their children had visited the beach, if they had swum or waded, how long they had been in the water, in what part of the beach they had swum, when they had swum, if they had taken lake water into their mouths or swallowed lake water, if they had drunk from the water fountain or had eaten food purchased from the concession stand, what other activities they had participated in, and if they had shown any symptoms of illness.

This second case control study included 20 camper children controls and 10 case children. No illnesses suggestive of E. coli O157:H7 were reported in any of the camper families, most of whom swam at the beach on Saturday rather than Sunday Most cases of E. coli O157:H7 infection swam on Sunday (n = 8) rather than on Saturday (n = 2). One case family with two children affected did not participate in the study The study was limited to children because all cases were less than 13 years of age. The mean age of cases was six years, and the mean age of controls was 5.3 years. (Among both cases and controls, 70 percent of participants were from one to six years of age and 30 percent were from seven to 12 years of age.)

Activities at the park that were not associated with illness included eating at the concession stand; drinking at the drinking fountain; and hiking, boating, or fishing at either lake. Gender was not a risk factor for infection.

Swimming behaviors that were significant risk factors for illness were taking water into the mouth and swallowing water. Cases were 10 times more likely to have taken water into their mouths than were controls and 12 times more likely to have swallowed water than were controls. Putting the head under water, blowing bubbles, or going past the shallow water rope were not significantly associated with illness.

On Sunday, cases were in the water longer than controls. Cases were also in the water later in the day than controls. To determine where the individuals swam, the swimming area was divided into three parts: the west, the center and the east. One control swam on both days. The two cases who swam only on Saturday swam in the east side of the swimming area, while Saturday controls swam mainly in the center of the swimming area. On Sunday, cases swam in all three parts of the swimming area, and controls swam on the west side of the swimming area and in the center of the swimming area. Inflow is on the east side of the lake, and outflow is on the west side of the lake [ILLUSTRATION FOR FIGURE 2 OMITTED].

Discussion

This is only the third outbreak of swimming-beach-associated E. coli O157:H7 infection reported in the United States since 1991 (7,8). Outbreaks have more commonly been associated with consumption of undercooked hamburger, consumption of improperly pasteurized or unpasteurized apple cider, or contact with or consumption of contaminated water (9-12). In the previous two beach outbreaks, the investigators concluded that a swimmer contaminated the water with E. coli O157:H7 (7,8). In the Rock Cut Park outbreak, the most likely explanation is also that a swimmer was the source of contamination. Epidemiologic analyses did not find an association between illness and consumption of any food items or drinking-fountain water. No other source of contamination for the lake could be identified; there were no leaking toilet facilities and no sewage releases into the lake. On the basis of the watershed investigation, the weather conditions preceding the outbreak, and inlet sample results, it is unlikely that the lake's only tributary (Willow Creek) was a source of contamination. There was no discernible water movement at any of the sample sites. Waterfowl were present at the beach during each sampling episode, and waterfowl feces were present on the beach sand and in the shallow beach water. They may have contributed to E. coli counts that were higher at the beach than in samples from the inlet and outlet, where waterfowl were not found (13,14). Waterfowl feces have not been associated with E. coli O157:H7 outbreaks elsewhere; nonetheless some fresh waterfowl feces were collected and analyzed for E. coli O157:H7. All samples tested negative.

The entire beach area including the grass around the sand beach can accommodate a large number of people. By comparison, the roped-off swimming area is small. On a hot day, attendance increases, as does the number of people actually in the water. Approximately 2,200 people visited the beach on Saturday, June 24, and 2,500 people on Sunday, June 25. If half these individuals were in the water at the same time (approximately 1,000 bathers), there would have been one bather per 10.5 square feet of water surface.

To ascertain the effect of bather loads on water quality at the beach, attendance records were compared with water sample results. Although the data were limited, no relationship was found between bather load and unsatisfactory water quality at this facility. Bather crowding has, however, previously been implicated in swimming-related disease outbreaks; since the cases in this outbreak swam on the highest bather usage days of the year, reducing bather loads may be beneficial (15).

Water sample results were also compared by month to see if bacterial counts rise as the weather becomes warmer. Although the data were again limited, they did not suggest a difference in bacterial levels between cooler and warmer months. No relationship was found between rainfall amounts and water quality

Several swimming activities increased the likelihood of illness. Cases were 10 times more likely to have taken lake water into their mouths than were controls and 12 times more likely than controls to have swallowed lake water. This information was obtained from parents, however, and is dependent on their recall of their children's behavior. On Saturday, the two cases swam on the east side of the lake, while on Sunday, cases swam in all three lake locations.

The initial date of the E. coli O157:H7 contamination in the water is not known. Although beach water samples collected during the morning of June 21 exceeded IDPH's water quality standards, it is not believed that the lake was contaminated with E. coli O157:H7 on that date. Of the 1,239 bathers at the beach on Wednesday, June 21, the 1,177 on Thursday, June 22, and the 1,362 on Friday, June 23, none are known to have developed illness. Given the attack rate of 0.25 percent on June 24 and June 25, approximately six cases should have been identified among the swimmers on June 21 through 23. Because of the extensive media coverage of this outbreak, it is believed that most ill persons who lived in the Rockford area would have been identified.

Conclusion

The results of the environmental investigation, epidemiological information, and known transmission modes, the most likely cause of the outbreak was ingestion of lake water fecally contaminated by one or more infected bathers.

This outbreak did not appear to be related to the manner in which the beach was operated. A review of current bathing-beach code requirements has, however, identified a number of ways that regulations could be strengthened to prevent future swimming-related disease. Accordingly, the following recommendations were offered to DNR to reduce the possibility of future outbreaks of swimming-related illness at this facility:

1. Require infants to wear tight-fitting rubber or plastic pants.

2. Post signs warning patrons not to swallow the lake water and advising ill persons not to swim.

3. Provide additional toilet facilities to avoid long waiting periods.

4. Either increase the swimming area or restrict the number of bathers in the swimming area to reduce bather crowding.

5. Continue to maintain the beach area free of any debris, including wastes from waterfowl or other wildlife.

Additionally, IDPH has strengthened its bathing beach regulations in several areas. In addition to eliminating the total-coliform standard as mentioned previously, IDPH has added a standard for E. coli of 235 organisms per 100 mL. The standard now allows the option of determining either fecal coliform or E. coli. If the standard is exceeded in both of the samples collected on the same day, the beach is closed. If it is exceeded in one of the two samples, two follow-up samples must be submitted. If one or both of the follow-up samples exceed the standard, the beach is closed.

The warning sign that is required to be posted at the facility has been changed to warn patrons not to drink the beach water or to use the beach facility when ill with diarrhea or vomiting and to require all children who are not toilet trained to wear plastic or rubber pants. This latter requirement was discussed at length. An alternative proposal would have prohibited children who were not toilet trained from being in the water. That measure was thought to be too extreme in view of the overall benefits afforded by use of the beach for recreational purposes by families. Requirement of a separate spray pool or similar facility was also considered but rejected on the basis of cost and potential separation of family members. The requirement [or tight-fitting rubber or plastic pants was ultimately determined to be an appropriate measure.

While the amendments to the bathing-beach code are expected to reduce the likelihood of an illness outbreak, patron cooperation in using the restroom facilities and using plastic pants on infants and toddlers is equally if not more important to averting similar outbreaks in the future.

REFERENCES

1. Boyce, T.G., P.M. Griffin, and D.L. Swerdlow (1995), "Escherichia coli O157:H7 and the Hemolytic Uremic Syndrome," New England Journal of Medicine, 333(6):364-367.

2. Griffin, P.M. (1995), "Escherichia coli O157:H7 and Other Enterohemorrhagic Escherichia coli," In M.J. Blaser, H.B. Greenberg, R.L. Guerrant, J.I. Ravdin, and PD. Smith, Infections of the Gastrointestinal Tract, Philadelphia: Lipincott Ravin, pp. 739-761.

3. Su, C., and L.J. Brandt (1995), "Escherichia coli O157:H7 Infection in Humans," Annals of Internal Medicine, 123(9):698-714.

4. Mitscherlich, E., and E.H. Marth (1984), Microbial Survival in the Environment: Bacteria and Rickettsiae Important in Human and Animal Health, Berlin, Germany: Springer-Verlag.

5. Office of Water Regulations and Standards Criteria and Standards Division (1986), Ambient Water Quality Criteria for Bacteria, EPA 440/5-84-002, NTIS Access Number PB 86-158045, Washington, D.C.: U.S. EPA, p. 7.

6. Seyfried, PL., N.E. Brown, P.F. Ness, and R.S. Tobin (1985), "A PrOspective Study of Swimming-Related Illness, II. Morbidity and the Microbiological Quality of Water," American Journal of Public Health, 75:1071.

7. Keene, W.E., T.J. Barrett, D.W Fleming, K. Hedberg, F.C. Hoesly, J.M. McAnulty, G.L. Oxman, M.A. Pfaller, and P Williams (1994), "A Swimming-Associated Outbreak of Hemorrhagic Colitis Caused by Escherichia coli O157:H7 and Shigella sonnei," New England Journal of Medicine, 331:579-84.

8. Ackman, D., G. Birkhead, M. Caldwell, P. Mack, S. Marks, and T. Root (forthcoming), Swimming-Associated Hemorrhagic Colitis Due to Escherichia coli O157:H7 Infection, Atlanta: CDC.

9. Besser, R.E., T.J. Barrett,, M.P Doyle, P.M. Griffin, S.M. Lett, J.T. Weber, and J.G. Wells (1993), "An Outbreak of Diarrhea and Hemolytic Uremic Syndrome from Escherichia coli O157:H7 in Fresh-Pressed Apple Cider," Journal of the American Medical Association, 269(17):2217-2220.

10. Swerdlow, D.L., N.H. Bean, PA. Blake, R.C. Brady, M. Bright, D. Donnell, E. Geldreich, K.D. Greene, PM. Griffin, A. Meyer, B.J. Payne, S. Tippen, J.G. Wells, and B.A. Woodruff (1992), "A Waterborne Outbreak in Missouri of Escherichia coli O157:H7 Associated with Bloody Diarrhea and Death," Annals of Internal Medicine, 117(10):812-818.

11. Bell, B.P, R. Baron, T.J. Barrett, C.A. Bartleson, M.A. Davis, M. Goldoft, D.C. Gordon, P.M. Griffin, J. Kobayashi, J.H. Lewis, P.I. Tarr, and J.G. Wells (1994), "A Multistate Outbreak of Escherichia coli O157:H7-Associated Bloody Diarrhea and Hemolytic Uremic Syndrome from Hamburgers. The Washington Experience," Journal of the American Medical Association: 272 (17): 1349-1353.

12. Brewster, D.H., J. Bimson, M.I. Brown, G.L. Houghton, D. Robertson, and J.C.M. Sharp (1994), "An Outbreak of Escherichia coli O157 Associated with a Children's Paddling Pool," Epidemiology and Infection, 112:441-447.

13. Fennell, H., D.B. James, and J. Morris (1974), "Pollution of a Storage Reservoir by Roosting Gulls," Water Treatment and Examination, 23 (1):5-23.

14. Gould, D.J., and M. R. Fletcher (1978), "Gull Droppings and their Effects on Water Quality," Water Research, 12:665-672.

15. Calderon, R.L., A. P Dufour, and E.W. Mood (1991), "Health Effects of Swimmers and Nonpoint Source Contaminated Water," International Journal of Environmental Health Research, 1(1):21-31.

Corresponding Author: Clinton C. Mudgett, P.E., M.S., Chief, Division of Environmental Health, Illinois Department of Public Health, 525 West Jefferson St., Springfield, IL 62761-001.