This E. coli O157:H7 outbreak cost a school district $6,100,000.
On Thursday, October 15 the Benton-Franklin Health Department (BFHD) was notified of two hospitalized children at Kennewick General Hospital with bloody diarrhea, one of whom was culture confirmed with infection of E. coli O157:H7. On the same day a clinician in Kennewick, Washington notified BFHD of two additional cases of bloody diarrhea in children. BFHD staff interviewed parents of the four children and determined that they all attended the Finley Elementary School (FES). As the average incubation period for E. coli O157:H7 ranges from 3 to 8 days, exposure to the agent most likely occurred sometime during the school week of October 5-9.
On October 16 Dr. Dan Jernigan at the Washington State Department of Health (WDOH) was notified by the Benton-Franlin Health District ~FHD) of the cluster of cases. Dr. Lauren Ball and Patti Waller, WDOH epidemiologists, were assigned to the BFHD outbreak investigation team. This report summarizes the findings.
Finley is located in a predominately rural area in east Benton County. Family farms/pastures, irrigation systems, and residential water supplied by private wells are common in the community. The Finley School District includes a high school (grades 9-12), a middle school (grades 6-8) and an elementary school (pre-school classes and grades K-S). The elementary school serves 466 students and employs 55 staff.
Both breakfast and lunch are served at the schools. Typically meals are partially prepared in a central kitchen located at the high school and delivered to each of the schools in the morning for additional cooking and preparation for serving. Water is supplied to the elementary school via a private well system. The BFHD inspects the system regularly. Sewage disposal is via a private drain field.
Surveillance and Case Definition
On Friday October 16, 1998 the Finley School District telephoned parents/guardians of all children who had been absent from school to inform them of illness occurring among Finley students. Parents of absent children were advised to seek medical advice if their children were experiencing a diarrheal illness or other flu-like illness. Informational letters from the school district were sent to all parents of children attending Finley Elementary notifying them that the communicable illness E. coli O157:H7 had been diagnosed in some FES children. These letters included educational information describing possible sources of infection, clinical symptoms, and the possibility of person to person transmission. Information on preventing E. coli infections was also disseminated widely by BFHD through local media, local physicians, and through the school district. In addition, the school district sent a letter to all residents of the Finley community to inform the public of the outbreak.
A confirmed case was defined as a Finley resident or visitor with culture confirmed E. coli O157:H7 diarrhea occurring after October 1, 1998 and epidemiologic linkage to FES.
A probable case was defined as a Finley resident or visitor with bloody diarrhea of unknown origin occurring after October 1, 1998, and/or complications of hemolytic uremic syndrome with epidemiologic linkage to FES.
Initial interviews with parents of patients at an area clinician’s office and at the hospital revealed that all patients attended a Finley area school. BFHD staff met with key personnel trom the school and the Finley School District to discuss the cluster and possible causes. As noted, exposure to E. coli O157:H7 was thought to have taken place at the school during the week of October 5-9. As some of the ill children attend school only on Tuesdays and Thursdays, events that occurred on Tuesday, October 6 and Thursday, October 8, were of particular interest. A ground beef taco lunch was served on Tuesday, October 6. A ham and cheese sandwich was served on Thursday, October 8.
A case-control study was conducted to compare children with confirmed or probable E. coh O157:H7 (cases) to a pair of randomly selected, classroom-matched classmates who did not develop a diarrheal illness (controls). Most cases were interviewed in-person while controls were interviewed by telephone using a structured questionnaire. A section of the questionnaire focused on the menu items served by the school district as reported by FES officials to BFHD. Factors evaluated included age, grade, classroom and exposure to such risk factors as foods eaten in and out of school, contact with livestock (particularly bovine), potable water, other water exposures, contact with diapered children, travel, and other possible exposures.
Cohort Study of Student Meals Purchased
To help ascertain if a particular meal or a particular lunch period was associated with illness, a cohort study was conducted using the FES daily meal payment records maintained on each student as an indicator of meal consumption. Data regarding the order in which classrooms ate lunch were obtained.
Cohort Study of Staff
Finley Elementary School staff and adult volunteers were given a self-administered questionnaire which contained questions about illness during the first three weeks of October, treatment, and consumption of school prepared meals.
Interview data for the epidemiologic studies was collected on standardized forms, entered into a computer and analyzed using Epi Info Version 6.04 software. For the case-control study, both matched and unmatched analyses were performed. Odds ratios, relative risks, and 95% confidence intervals (CI) were calculated where possible.
Results of Epidemiological Investigation
After public notification of illness occurring at Finley Elementary School, children from the Finley area with complaints of illness or diarrhea were seen at area hospital emergency rooms, urgent care facilities and by local private providers. Active surveillance resulted in 9 confirmed cases E. coli O157:H7 and 2 probable cases being identified. All were students at FES. Epidemiologic curves are shown in Appendix C. There were cases in all grades, except grade 4. Cases were distributed among eight of eleven K-S homerooms. The mean case age was 8, median case age was 9, and the mode was 10 years old Thirty-three percent (n=4) of the cases occurred among children in the fifth grade. One fifth grade class room had three ill children.
Matched and unmatched analyses were performed for food items consumed. There was no statistically significant association in either analysis between illness and any particular school meal, any particular school food item, and any other risk factors. Selected unmatched odds ratios and confidence intervals are presented in Table 1.
Table 1. Odds Ratio of Illness Associated with Foods Served at FES, October 5-8, 1998.
|Food Item||# CasesEating Item||# ControlsEating Item||Odds Ratio,(95% CI)||p-Value*|
|Dipping Sauce||5||9||1.81(0.29, 11.61)||0.69|
|Taco, Cheese/Lettuce||6||10||1.50(0.25, 9.33)||0.70|
|Buttered Corn||7||9||2.59(0.40, 18.43)||0.43|
|Hot Dog||9||13||2.77(0.38, 24.80)||0.43|
|Tomato Soup||5||4||3.54(0.54, 24.86)||0.21|
|Veggie Sticks||5||6||1.94(0.33, 11.96)||0.45|
|Ham Sandwich||7||11||1.91(0.29, 13.25||0.69|
|Celery Sticks||2||4||1.21(0.12, 11.30)||0.76|
* Fishers exact test, 2-tailed. A p-value <0.05 indicates statistical significance.
Cohort Study of Student Meals Purchased
There was no statistically significant association between eating any particular school meal served October 5- October 9 and illness. There was no statistically significant association found with a particular lunch period and illness. Selected relative risks and 95% confidence intervals for the meals are presented in Table 2.
Table 2. Relative Risk of Illness Associated with Meals Served at FES, October 5-8, 1998.
|Meal||# IllEating Meal||# Not IllEating Meal||Relative Risk,(95% CI)||p-Value*|
|Lnch 10/5||8||262||1.22 (0.33, 4.54)||1.00|
|Bkfst 10/6||1||89||0.34 (0.04, 2.60)||0.46|
|Lnch 10/6||9||286||1.51 (0.33, 6.87)||0.73|
|Lnch 10/7||8||260||1.25 (0.34, 4.65)||1.00|
|Bkfst 10/8||1||71||0.45 (0.06, 3.44)||0.69|
|Lnch 10/8||9||294||1.35 (0.30, 6.14)||1.00|
* Fishers exact test, 2-tailed. Ap-value <0.05 indicates statistical significance.
Cohort Study of Staff
Sixty-two percent (34/55) of the staff at FES responded to a self-administered questionnaire regarding adult exposures during the same time period. The respondents can be described as 13 of 2l homeroom teachers, 3 kitchen staff; and 18 other staff (administration, other teachers, and staff). Five of the 34 respondents reported diarrhea, however only three of the five reported illness in the appropriate time frame. Of the three that were suspect for illness, only one person reported eating from the student serving line. This respondent ate both the taco meal on Tuesday October 6 and the ham sandwich meal on Thursday October 8. Two other adult staff members reported eating the taco meal. They did not report diarrhea, nor did they report obtaining their food from the student serving line.
Officials from BFHD conducted facility inspections, obtained food and water samples, and interviewed district, elementary school, and food service staff Appendix D contains the report of the environmental investigation conducted by BFHD. Briefly, common food stock is served to all schools in the district. Typically, meals served in the Finley School District are partially prepared in a central kitchen located at Finley High School and transported to the middle school and elementary school via vans which may or may not be equipped for hot holding of foods. Additional cooking andlor preparation may take place at each school depending on the food item. Different hot holding techniques are used at the various Finley schools when the hot meal arrives from the central kitchen. There is variability in the way meals are served. Some food items are served by food service personnel, some items are served by student helpers, and some items are self serve.
In regards to the ground beef tacos served on Tuesday October 6 the following points were noted:
1. The ground beef used for the tacos was from frozen stock, dated October 1997.
2. The ground beef was prepared in one large batch in a steam jacketed kettle at the central kitchen.
3. On this particular day, there was a slight variation in the method of ground beef preparation in that the meat was not rinsed after cooking. The meat was not rinsed because it was reportedly lean and excessive fat was not a problem.
4. After cooking, the taco beef was portioned out into serving pans and delivered to the various schools.
5. The meat delivered to the elementary school was transported in a district van which does not have hot holding capacity.
6. On Tuesday October 6, the ground beef at the elementary school was stored in warrring bins prior to service. At the other schools the trays of meat were placed on the stove top and the burners were lit to keep the food warm.
The Health District also noted that at the elementary school the ham and cheese sandwiches served on Thursday, October 8 were prepared by heating them in the warming bins only. At the other district schools, hot sandwiches are prepared in regular ovens. Also, at the time of the outbreak, the elementary school had been conducting a milk promotion/contest. Winners would select a cookie out of a common repository. There were no records of the Lucky Milk promotion winners.
Water is supplied to the elementary school via a private well system. Sewage disposal is via a private drain field. No water quality violations prior to or during the outbreak were noted.
Samples of leftovers and other food items were cultured for E. coli O157:H7. Samples of cooked taco meat left over from the meal served on Tuesday October 6, 1998 tested negative for E. coli O157:H7. Gross examination of the meat, however, revealed “golf ball” sized chunks with pink, undercooked centers. Other food items tested include samples from one 9 pound chub of raw ground beef from the lot used to prepare taco meal, lettuce, cheese, tomatoes, ham, and chocolate milk. All food items tested negative for E. coli O157:H7.
Potable water samples were collected by the Finley School District and analyzed by the BFHD laboratory immediately after the outbreak was detected. These samples were negative for fecal coliforms.
USDA conducted sampling and analysis on the frozen/raw ground beef left over from the same lot served at FES on October 6, 1998. Results of these tests were negative for E. coli O157:H7.
Pulsed-field gel electrophoresis (PFGE) was performed on E. coli O157:H7 isolates from confirmed case patients. Comparisons were made between the isolates from samples obtained from patients in this cluster. Seven out of nine confirmed cases had isolates of E. coli O157:H7 submitted for analysis. All seven confirmed cases were infected with identical strains of E. coli O157:H7. The FES subtype was then compared to those subtypes from previous outbreaks contained in databases maintained by the WDOH and the federal Centers for Disease Control and Prevention (CDC). The subtype identified in the FES cluster did not match any subtypes from recent cases in the Washington State database nor the CDC database.
One probable case and a single suspect secondary case (see Addendum) had frozen sera available for immunologic testing. These samples were forwarded to the CDC for analysis. Both samples were positive for acute and convalescent antibodies to E. coil O157:H7.
Prevention measures and other actions in response to this outbreak included:
1. Immediate community notification and follow-up. This was accomplished by phone calls, letters, media reporting, and BFHD representation at meetings with the community (October 29, 1998) and the school district.
2. BFHD notified the WDOH and accepted offer of assistance.
3. BFHD disseminated educational information to health care providers, food handlers, and to the general public.
4. Hand washing workshops were conducted by BFHD for all students, teachers, and staff at FES. Hand washing workshops were also conducted for all Finley School District food service employees. These workshops required active participation of attendees.
5. Public education and food safety control measures are being reinforced in the communities served by the BFHD to minimize possible exposures to E. coil 0157:H7 bacteria.
The results of the epidemiologic investigations did not indicate any statistically significant associations between illness and a particular meal, food item, or other risk factor. Microbiologic cultures of food and water samples were negative for E. coli O157:H7. However, the results of this investigation indicate that there was a point source exposure to E. coli O157:H7 that occurred at Finley Elementary School. As no other common school activity was identified other than eating at the school cafeteria, it is reasonable to conclude that a meal served at the school was the likely source of illness. Cattle are the known reservoir of E. coli O157:H7. Thus, it is likely that consuming the ground beef served in the tacos was the vehicle.
Findings supporting this conclusion are:
a. The confirmed and probable cases were restricted to the elementary school.
b. Results of the PFGE analysis of E. coli O157:H7 isolates from confirmed case patients indicate that an identical subtype of E. coil infected the confirmed cases.
c. No confirmed or probable cases were reported from other schools in the district.
d. No teachers or staff reported illness to BFHD or WDOH. Most staff did not eat food supplied from the student serving line.
e. No confirmed or probable cases were reported from the community at large which matched the FES subtype.
f. Results of the environmental investigation suggest that although similar menu items and common food stock were served to all schools in the district, there were differences in food preparation techniques, transportation time, and hot holding techniques between the elementary school and the other schools in the district. In particular, deficiencies in the transport method had been noted in previous inspections.
g. Results of the case-control study suggest that the cases did not have common risky exposures outside of the elementary school.
h. Results of the epidemiologic studies did not reveal any special school activities (birthday parties, field trips, etc.) that could account for possible exposures during this time period.
This investigation has several limitations. The cohort study relied on school lunch payment records and validity of these records was not checked. The cohort was very homogeneous in terms of exposures and demographic characteristics. The case-control study utilized classroom-matched controls. This sampling strategy could have introduced bias secondary to over-matching on age. Recall bias is a major concern in this study and can be contributed to the amount of time between the exposure and the interview and to the young age of both cases and controls. Early in the investigation preliminary evidence indicated that this outbreak may have been related to the taco meal served on October 6. This information was publicized and may have also introduced bias.
Ground beef is a known vehicle for E. coli Ol57:H7 and this investigation noted differences in the preparation, handling, and transport of meat which may have allowed for uneven cooking, uneven cooling, and uneven reheating at the elementary school. This outbreak and the resulting investigation highlight the importance of regular inspections of institutional kitchens and the need for ongoing training of food service workers.
Notification of Suspect Secondary Case
In late October the BFHD was notified of a two year old female taken to the Kennewick General Hospital Emergency Room on October 28, 1998. The patient’s status was critical and she was transferred to Children’s Hospital of Seattle where she was diagnosed with and treated for hemolytic uremic syndrome secondary to hemorrhagic colitis of unknown origin. BFHD interviewed the child’s mother to determine possible epidemiologic linkage to the FES cluster. The results of these interviews revealed a sibling who attended FES. However, it was unclear to BFHD whether this sibling experienced any illness or had contacts with any case patients during the first two weeks of October. The patient was categorized as a suspect secondary case. Serum samples were obtained by the treating physician for antibody testing by the federal Centers for Disease Control and Prevention. The sample was strongly positive for acute and convalescent antibodies to E. coil O157:H7.
This is another it what will be a long – too long – series of outbreak investigations where we have represented consumers in what I hope will be a cautionary tale, and a learning experience, for manufacturers of food.