MAJOR ARTICLE

Epidemic Diarrhea due to Enterotoxigenic Escherichia coli

Mark E. Beatty,1,2,a Penny M. Adcock,1,2,a Stephanie W. Smith,3 Kyran Quinlan,1,a Laurie A. Kamimoto,1,a Samantha Y. Rowe,2,a Karen Scott,3 Craig Conover,4 Thomas Varchmin,3 Cheryl A. Bopp,2 Kathy D. Greene,2 Bill Bibb,2 Laurence Slutsker,2,a and Eric D. Mintz2
1

Epidemic Intelligence Service, Division of Applied Public Health Training, Epidemiology Program Ofce, and 2Foodborne and Diarrheal Diseases Branch, Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia; and 3Cook County Department of Public Health and 4Illinois Department of Public Health, Chicago, Illinois

(See the editorial commentary by Daniels on pages 3356)

Downloaded from http://cid.oxfordjournals.org/ by guest on October 6, 2012

Background. In June 1998, we investigated one of the largest foodborne outbreaks of enterotoxigenic Escherichia coli gastroenteritis reported in the United States. Methods. We conducted cohort studies of 11 catered events to determine risk factors for illness. We used stool cultures, polymerase chain reaction, and serologic tests to determine the etiologic agent, and we conducted an environmental inspection to identify predisposing conditions and practices at the implicated establishment. Results. During 57 June, the implicated delicatessen catered 539 events attended by 116,000 people. Our epidemiological study of 11 events included a total of 612 attendees. By applying the median prevalence of illness (20%) among events with ill attendees to the total number of events with any ill attendees, we estimate that at least 3300 persons may have developed gastroenteritis during this outbreak. Multiple food items (potato salad, macaroni salad, egg salad, and watermelon) were associated with illness, all of which required extensive handling during preparation. Enterotoxigenic Escherichia coli serotype O6:H16 producing heat-labile and heat-stable toxins was isolated from the stool specimens from 11 patients. Eight patients with positive stool culture results, 11 (58%) of 19 other symptomatic attendees, and 0 (0%) of 17 control subjects had elevated serum antibody titers to E. coli O6 lipopolysaccharide. The delicatessen had inadequate hand-washing supplies, inadequate protection against back siphonage of wastewater in the potable water system, a poorly draining kitchen sink, and improper food storage and transportation practices. Conclusions. In the United States, where enterotoxigenic Escherichia coli is an emerging cause of foodborne disease, enterotoxigenic Escherichia coli should be suspected in outbreaks of gastroenteritis when common bacterial or viral enteric pathogens are not identied. Enterotoxigenic Escherichia coli (ETEC), a frequent cause of travelers diarrhea, has recently been identied as the cause of an increasing number of outbreaks of foodborne infection in the United States [1, 2]. ETEC also has the potential to cause large outbreaks of gastroenteritis. The largest reported outbreak of foodborne ETEC infection in the United States, resulting in 452 cases, occurred at a restaurant in Wisconsin in 1980 [3]. We describe one of the largest outbreaks of foodborne ETEC infection, which occurred in Cook County, Illinois, in 1998. OUTBREAK On 10 June 1998, the Cook County Department of Public Health (Oak Park, IL) received multiple coma Present afliations: Epidemiology and Prevention Activities, Dengue Branch, Division of Vector-Borne Infectious Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, San Juan, Puerto Rico (M.E.B.); Vaccines/BiologicsClinical Research, Merck Research Laboratories, Blue Bell, Pennsylvania (P.M.A.); Department of Pediatrics, University of Chicago, Illinois (K.Q.); HIV Incidence and Case Surveillance Branch, Division of HIV/AIDS Prevention, National Center for HIV, STD, TB Prevention, Centers for Disease Control and Prevention, Atlanta, Georgia (L.A.K.); Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta (S.Y.R.); and CDC/KEMRI Research Station, Kisumu Kenya (L.S.).

Received 18 July 2005; accepted 16 September 2005; electronically published 27 December 2005. Presented in part: Annual Meeting of the Infectious Disease Society of America, Denver, Colorado, 1215 November 1998; and Centers for Disease Control and Prevention Annual Epidemic Intelligence Service Conference, Atlanta, Georgia, 20 24 April 1999. Reprints or correspondence: Dr. Mark Beatty, Centers for Disease Control and Prevention, 1600 Clifton Rd., Mail Stop A-38, Atlanta, GA 30333 (mbeatty@ cdc.gov). Clinical Infectious Diseases 2006; 42:32934 2005 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2006/4203-0004$15.00

Massive Outbreak of ETEC Diarrhea CID 2006:42 (1 February) 329

plaints of acute gastroenteritis from attendees of 2 unrelated events on 6 June catered by a single delicatessen (delicatessen X). During an interview with the Cook County Department of Public Health on 10 June, the owner of delicatessen X estimated that, during 57 June, his company catered 1500 events attended by 15,00020,000 people. In addition to catering services, delicatessen X offered restaurant seating and a carry-out service. After receiving numerous additional reports of illness among patrons of all 3 venues, delicatessen operations were suspended on 10 June. We conducted an investigation to assess the extent of the outbreak and to determine risk factors for illness. MATERIALS AND METHODS Case nding and hypothesis generation. We requested information on patients who presented to emergency departments with a chief complaint of gastroenteritis from neighboring health departments and hospitals in the south suburban region of Cook County. During case nding, we dened a case as an infection in a person reporting loose stools or abdominal pain after consumption of food from delicatessen X during 3 10 June. We obtained copies of food orders from delicatessen X for catered events held on 57 June. The order forms contained the host contact information, the date and size of the event, and the menu items served. We surveyed all available hosts regarding gastrointestinal illness among event attendees. To generate hypotheses regarding the cause of the outbreak, we interviewed a convenience sample of ill persons, including attendees of 5 events and a sample of restaurant and carry-out patrons. We then developed a questionnaire regarding food consumption, medical history, symptoms, and treatment. Data presented in this manuscript were collected during an outbreak investigation and did not involve human experimentation. Cohort studies. To determine risk factors for illness, we selected a sample of the 11 largest and unrelated catered events that corresponded to the peak exposure period, 57 June. We tailored the food exposure questionnaire to the catered food items served at each event, and we analyzed each event as a separate cohort. We restricted the case denition for the cohort studies to patients who reported having 3 loose stools in any 24-h period during the week after consuming food at an event catered by delicatessen X. We interviewed participants by telephone. Analysis of event orders. Using the cohort studies results, we analyzed data from the order forms of 405 events catered by the delicatessen X during 57 June, looking for a statistical association between foods implicated and subsequent diarrheal illness. Environmental investigation. We queried all delicatessen X employees regarding recent illness, travel, and foreign visitors, and we reviewed their work schedules for unreported absenteeism. We interviewed delicatessen X management and kitchen
330 CID 2006:42 (1 February) Beatty et al.

staff regarding food preparation and storage, delivery equipment, and sanitizing procedures. We inspected delicatessen X to determine whether it was in compliance with the local health code. This included a detailed examination of the plumbing system and a dye test of the sink drains, oor drains, and toilets to determine whether backow of waste was occurring in the potable water system. Laboratory investigation. We retrieved stool specimens from patients who called to report illness and asked patients who were still ill at the time of the interview to provide a stool sample. We cultured stool samples for enteric bacterial pathogens, including Salmonella species, Shigella, Yersinia, Campylobacter species, E. coli O157:H7, Aeromonas species, Vibrio species, Bacillus cereus, and Staphylococcus aureus. To identify ETEC, we plated patient specimens onto MacConkey agar, and we tested growth by PCR for heat-labile and heat-stable enterotoxins at the Centers for Disease Control and Prevention (CDC; Atlanta, GA) [4]. We used standard procedures to serotype the O and H antigens of enterotoxin-producing isolates [4]. For 5 isolates, we used the disk diffusion method to determine antimicrobial susceptibility to ampicillin, amoxicillinclavulanate, ceftriaxone, chloramphenicol, gentamicin, kanamycin, nalidixic acid, streptomycin, sulsoxazole, tetracycline, trimethoprim-sulfamethoxazole, and ciprooxacin [5]. Nine isolates were compared by PFGE using the restriction enzyme XbaI [6]. We collected 5 mL of serum from a convenience sample of patients. The CDC developed an ELISA for IgG and IgM antibodies to E. coli lipopolysaccharide (LPS) using puried O6 LPS. As controls, we used stored serum samples from patients with stool cultureconrmed infection with E. coli O157:H7 or Salmonella serotype Enteritidis who were not involved in this outbreak. On 10 June, we collected stool samples from all delicatessen X food handlers for culturing for routine enteric pathogens. On 22 June, we collected serum samples and a second stool sample for ETEC testing. On 1017 June, we collected samples of potato salad, macaroni salad, coleslaw, kidney bean salad, hard-boiled eggs and the water in which they were stored, raw potatoes, peeled and unpeeled boiled potatoes, uncut celery, precut celery from an opened bag, precut celery from an unopened bag, and onions from an unopened bag from delicatessen X. We also collected samples of potato salad, macaroni salad, coleslaw, mostaccioli, chicken, and beef from the refrigerators of ill persons. Testing included coliform counts, fecal coliform counts, E. coli counts, and ETEC identication using the methods described above [4]. On 1718 June, we collected samples or swabs for culture from sink faucets, antisiphon ball cock valves of toilets, the garden hose used to rinse the pressure cookers, food storage trays, residue from the ice machine, and grease pit residue in delicatessen X.

Downloaded from http://cid.oxfordjournals.org/ by guest on October 6, 2012

Other samples collected 36 weeks after the outbreak included swabs of the inside of dead-end pipes found within the potable water system and swabs from oor drains and storm drains. The Illinois Department of Public Health laboratory tested the water samples and the environmental swabs from the ice machines for coliforms, fecal coliforms, and E. coli. The CDC tested all other environmental samples for ETEC. Statistical analysis. Statistical analyses were conducted using EpiInfo, version 6.04 (CDC) [7], and SAS software, version 7.0 (SAS Institute) [8]. We conducted univariate analyses, in which risk ratios (RRs) with 95% CIs that excluded 1.0 and P values .05 (determined by the x2 test or Fishers exact test) were considered to be statistically signicant. We constructed multivariate models for each cohort with 11 food item that was associated with illness in univariate analysis at a level of P .1. Using manual stepwise logistic regression, we reduced each model maintaining all exposure variables with a P value of .05. RESULTS Case nding. The Cook County Department of Public Health received telephone reports of acute gastroenteritis from 127 households of persons who had purchased restaurant or carryout food from delicatessen X during 210 June. To determine the peak exposure period, we plotted the date of meal purchase by household (gure 1). Many households reported 11 ill household member; 7 households reported that a family member had been hospitalized, but no deaths were reported. Delicatessen X catered 539 events during 57 June. Among 405 event hosts who were aware of the health status of their guests after the event, 263 (65%) reported that 1 guest had become ill (gure 2). A total of 16,691 persons had attended these 263 events. Cohort studies. Of 612 persons who attended the 11 events studied, 591 (97%) were interviewed. One hundred twenty attendees (20%) subsequently reported a diarrheal illness that met the cohort study case denition. The median age of patients was 39 years (range, 174 years), and 46% were male. The median duration of diarrhea was 5 days (range, 119 days). The median incubation period was 50 h (range, 1168 h). There were no hospitalizations or deaths among the cohorts studied. The results of the cohort studies are summarized in tables 1 and 2. Between 12% and 49% of interviewed attendees at 9 events reported illness (median proportion ill, 20%); attendees at 2 other events reported no illness. In 6 cohorts, univariate analysis identied at least 1 food item that was statistically associated with illness. In 3 cohorts, 11 food item was statistically associated with illness; however, multivariate analysis implicated a single food item in 2 of these cohorts. No single implicated food item was common to all cohorts reporting illness, but prepared salads (e.g., potato, macaroni, and egg

Figure 1. Number of households that reported acute gastroenteritis, by date of meal purchase or consumption at delicatessen X, Cook County, Illinois, 1998 (n p 127).

Downloaded from http://cid.oxfordjournals.org/ by guest on October 6, 2012

salads) mixed by hand at delicatessen X were implicated in 5 cohorts. Watermelon cut by hand at delicatessen X was statistically associated with illness in the remaining cohort. A prepared salad was served to all 3 cohorts in which no food item was implicated and to 1 of the 2 cohorts in which no illness was reported. Analysis of event orders. The only characteristic of the 405 events that had a statistically signicant association with illness was the method of food delivery. Events on 6 June with ill attendees were more likely to have had food delivered by refrigerated truck (52% vs. 31%; RR, 1.4; 95% CI, 1.21.6; P ! .01). Although no single food item was served at all events with at least 1 ill attendee, potato salad was served at 242 (92%) of these 263 events; however, this association was not statistically signicant. By applying the median prevalence of illness (20%) among the 9 events with ill attendees to the 16,691 attendees of the 263 events with known illness during 57 June, we estimated that 3338 attendees developed a diarrheal illness that met the cohort study case denition. This estimate does not include those who may have become ill after attending one of the 134 events for which no information on illness status of attendees was obtained, nor does it include cases of illness following consumption of dine-in or take-out food or among persons who attended events before or after the 57 June study period. Environmental investigation. The ingredients common to the potato, macaroni, and egg salads were mayonnaise, celery, and eggs. We contacted local food service facilities that had received these and other salad ingredients from the same distributor that supplied delicatessen X during the exposure period. None reported subsequent illness among patrons. All food handlers denied recent symptoms of a gastrointestinal illness, travel, and foreign visitors. The inspection of delicatessen X revealed that (1) the quantity of food prepared during the week of 37 June was nearly double the normal
Massive Outbreak of ETEC Diarrhea CID 2006:42 (1 February) 331

Figure 2. Attack rates for enterotoxigenic Escherichia coli, by date of catered event, Cook County, Illinois, 1998
Downloaded from http://cid.oxfordjournals.org/ by guest on October 6, 2012

volume (e.g., 2889 kg [6420 lbs] of potato salad), (2) ingredients were cooked and then stored in large 22.5-kg (50-lb) containers, (3) salad ingredient preparation and mixing was done by hand, and (4) the hand-washing station in the main kitchen had no soap or paper towels. In addition, we learned that (5) on 5 June, there was inadequate space in the walk-in coolers for food storage (i.e., some foods were stored overnight in a refrigerated truck that had no internal thermometer and no secure way to ensure the plug for the trucks refrigeration unit remained in the electrical outlet); and (6) it was not unusual for food to be delivered to events in a vehicle without refrigeration or portable coolers. Finally, (7) often for the delivery to smaller events, multiple salads were packed into a single container for delivery with dividers that permitted some mixing of contents.

Although dye studies did not reveal back siphonage of sewage water into the potable water system, the plumbing system had several problems that could lead to backow of other wastewater from other sources into the potable water system if a low pressure event occurred, as follows: (1) a backow device in the hot water heating system was nonfunctional, (2) a garden hose used to rinse the pressure cookers did not have proper backow protection, and (3) there were some lengths of pipe in the potable water system that ended blindly, where water could stagnate. Indeed, a low pressure event occurred on 3 June, when the water in the delicatessen was shut off for 30 min during the installation of a new dishwasher. Laboratory investigation. Twenty-one patients who ate food prepared at delicatessen X during 49 June and who reported having diarrhea to the Cook County Department of

Table 1. Patient characteristics and symptom proles for 11 cohorts who attended events catered by delicatessen X, Cook County, Illinois, 57 June 1998.
No. (%) of attendees Cohort 1 2 3 4 5 6 7 8 9 10 11 Total
a b

Date of event 5 5 5 6 6 6 6 7 7 7 7 June June June June June June June June June June June

All 59 48 18 83 65 41 13 89 48 64 84 612

Ill attendees 12 8 0 36 14 11 5 11 0 9 14 120 (20) (17) (0) (43) (22) (27) (38) (12) (0) (14) (17) (20)b

Age, median years 46 40 43 37 41 23 29 39 31 39

Male sex, % 51 63 56 50 64 20 46 33 29 46

Incubation period, median h (range) 59 (1156) 76 (54309) 45 (5309) 49 (15194) 50 (369) 63 (44165) 67 (13214) 28 (5189) 41 (18457) 50 (1457)

Illness duration, median days (range) 5 (119) 9 (117) 5.5 (112) 5 (213) 5 (115) 7 (510) 5 (315) 2 (117) 3 (117) 5 (119)

Symptom, % of ill subjects Abdominal cramps 92 100 83 100 64 100 82 75 71 84 Headache 58 29 49 50 73 80 64 44 21 50 Fever 30 25 35 36 10 80 40 22 0 29

Vomiting 8 38 8 7 0 20 18 11 0 10

Case denition required diarrhea (dened as 3 loose stools in any 24-h period). Median prevalence of illness among cohorts with ill attendees.

332 CID 2006:42 (1 February) Beatty et al.

Table 2. Univariate analysis of exposures of 11 cohorts catered by delicatessen X (deli X), Cook County, Illinois, 57 June 1998.
No. who consumed any deli X food 49 19 18 73 54 39 12 77 45 31 55 Attack rate, % Implicated food Watermelon None Potato saladb Coleslaw 5 6 7 8 63 41 12 85 9 10 11 46 64 76 14 (26) 11 (28) 5 (42) 11 (14) 0 (0) 9 (29) 14 (25) Potato salad Macaroni salad None Potato saladb Tomato slices (relish tray) Macaroni salad None Egg salad sandwichb Chicken salad sandwich Tuna salad sandwich
b b a

Cohort 1 2 3 4

No. of respondents 59 48 18 79

No. of ill attendees (attack rate, %) 12 (24) 8 (42) 0 (0) 36 (49)

Exposed attendees 47 88 87 67 64 29 42 33 58 55 50

Unexposed attendees 14 16 41 5 9 6 13 11 18 21 19
.05.

Risk ratio (95% CI) 3.4 (1.29.7) 5.4 (2.611.4) 2.1 (1.53.1) 13.3 (1.992.6) 7.1 (1.828.1) 4.7 (1.119.8) 3.3 (1.28.9) 3.0 (1.18.6) 3.2 (1.47.2) 2.6 (1.15.9) 2.6 (1.16.0)

P .02

!.01 !.01 !.01 !.01

.02 .04 .05

Downloaded from http://cid.oxfordjournals.org/ by guest on October 6, 2012

.02 .05 .05

a b

Statistically signicantly associated with illness in univariate analysis; 95% CI does not include 1 and P values of Remained statistically signicant (P .05) in multivariate analysis.

Public Health provided a stool specimen. All were culture negative for routine enteric pathogens. Eleven of these samples yielded heat-labile and heat-stableproducing ETEC serotype O6:H16, which was susceptible to all antimicrobial agents tested. Nine of these isolates were PFGE typed following digestion with restriction enzyme XbaI. Eight isolates had indistinguishable patterns, whereas the remaining isolate differed by 1 band. We collected serum samples from 27 patients who reported recent illness; 16 of the 27 patients also provided stool specimens. Nineteen of the 27 serum samples had elevated antibody titers to ETEC serogroup O6 LPS. All 8 patients with stool cultures that yielded ETEC had an IgG or IgM anti-O6 LPS titer of 1:160. With a titer 1:160 used to dene seropositivity, 3 of 8 patients who reported a recently resolved illness and who had negative stool culture results had antibody titers 1:160, and 8 of 11 patients who reported a recently resolved illness but did not submit a stool sample for culture also had elevated antibody titers. Anti-O6 LPS titers were negative among stored serum samples (n p 17). Cultures of all stool samples obtained from food handlers (n p 20) were negative for enteric pathogens, including ETEC. Only 2 food handlers consented to have their blood drawn; both tested negative for anti-O6 LPS antibodies. ETEC was not identied in any of the food samples from delicatessen X or from patrons refrigerators; the results for all environmental swabs were negative. Water samples collected from delicatessen X did not yield coliforms.

DISCUSSION We report one of the largest foodborne outbreaks of ETEC gastroenteritis in the United States to date. The outbreak occurred during 210 June 1998. The source was a delicatessen in Cook County, Illinois. We documented 120 illnesses during the investigation but estimate that as many as 3300 area residents may have become ill during this outbreak. We isolated heat-labile and heat-stable toxinproducing ETEC serotype O6:H16 in stool samples obtained from 11 patients. Previously, ETEC serotype O6:H16 was the most common ETEC serotype identied during outbreaks occurring in the United States [1]. However, beginning in 1995, ETEC serotype O169: H41 emerged in the United States, and it subsequently became the predominant outbreak-associated ETEC serotype [2]. During the outbreak, an ELISA to anti-O6 LPS antibodies was developed that enabled the identication of 11 additional cases, which had been asymptomatic at the time of follow-up. This assay may be useful for future outbreak investigations because of its sensitivity after ETEC is no longer shed in stool. This outbreak had many of the typical features of previously reported outbreaks of ETEC infection: stool cultures were negative for routine bacterial pathogens, the ratio of the prevalence of diarrhea to vomiting among patients was 12.5, and the median duration of illness was 160 h [1]. The median incubation period was 50 hjust slightly longer than the characteristic 2448-h range [9]. Our epidemiological study of the outbreak indicated that
Massive Outbreak of ETEC Diarrhea CID 2006:42 (1 February) 333

illness was associated with consumption of multiple salads prepared at delicatessen X, including potato salad, macaroni salad, and egg salad. ETEC could have been introduced into the salads in several ways (e.g., ingredients were contaminated outside the delicatessen, the salad was prepared by an ETEC-shedding food handler, or the ingredients or utensils were contaminated by a malfunctioning plumbing system). Because illness was not associated with other food service establishments that received the same salad ingredients as delicatessen X, it is unlikely that the salad ingredients were contaminated before arriving at delicatessen X. The ingredients in the salads were extensively handled, and there was evidence that hand hygiene was not optimal. An infected food handler could have been the source of the outbreak but remained unidentied, because ETEC is shed in stool for only a few days after infection resolves [10, 11], and stool samples were not collected for ETEC testing until nearly 2 weeks after the exposure period of most patients. Serologic testing for E. coli O6 antibodies could have provided further information about the origin of this outbreak; however, only 2 of 20 food handlers consented to undergo this test. Although there was no direct evidence of sewage contamination of the facilities, several problems with the plumbing system were identied at delicatessen X, including back ow devices that were absent or rendered nonfunctional and could have allowed contaminated water to contact the salads. Although the route by which ETEC was introduced into salads remains unknown, several opportunities for the organism to multiply in the foods were documented. Storage and cooling of ingredients in large containers may have prevented rapid cooling. Illnesses traced to events on 6 June were associated with food that was delivered. The salads delivered on 6 June were stored the night before in a refrigerated truck with no mechanism for temperature monitoring. A malfunction in the refrigeration unit during the night could have occurred and would have gone undetected. In addition, some catered events had food delivered in vehicles without refrigeration or portable coolers. These practices may have provided favorable conditions for signicant multiplication of organisms. This investigation was limited by the small number of interviews conducted relative to the magnitude of the outbreak. However, by selecting multiple cohorts during the peak of the outbreak and surveying hosts of other events, we were able to rapidly determine the source and magnitude of the outbreak. The determination of the etiologic agent that caused the outbreak was similarly based on a relatively small number of stool cultures. The short period of shedding following resolution of

symptoms prevented conrmation of additional cases, but we attempted to conrm additional cases through the novel application of LPS serologic tests. Finally, the source of contamination could not be determined. Nevertheless, delicatessen X was the point source of the outbreak, and both employees and environmental factors likely contributed to the contamination and multiplication of the organism. In summary, we describe one of the largest foodborne outbreaks of ETEC infection reported in the United States. This investigation indicates that ETEC has the potential to cause large foodborne outbreaks of acute gastroenteritis. Health care providers and public health practitioners should always consider ETEC testing if the results of bacterial cultures for routine enteric pathogens are negative.
Acknowledgments
We thank Mark Matuck from the Cook County Department of Public Health and Margaret Swartz from the Illinois Department of Public Health for their substantial guidance and technical assistance during this investigation. Potential conicts of interest. All authors: no conicts.

Downloaded from http://cid.oxfordjournals.org/ by guest on October 6, 2012

References
1. Dalton CB, Mintz ED, Wells JG, Bopp CA, Tauxe RV. Outbreaks of enterotoxigenic Escherichia coli infection in American adults: a clinical and epidemiologic prole. Epidemiol Infect 1999; 123:916. 2. Beatty ME, Bopp CA, Wells JG, Greene KD, Puhr ND, Mintz ED. Enterotoxin-producing Escherichia coli O169:H41, United States. Emerg Infect Dis 2004; 10:51821. 3. Taylor WR, Schell WL, Wells JG, Choi K, Kinnunen DE, Heiser PT. A foodborne outbreak of enterotoxigenic Escherichia coli diarrhea. N Engl J Med 1982; 306:10935. 4. Olsvik O, Strockbine NA. PCR detection of heat-stable, heat-labile, and Shiga-like toxin genes in Escherichia coli. In: Persing CH, Smith TF, Tenover FC, White TJ, eds. Diagnostic molecular microbiology. Washington, DC: American Society for Microbiology, 1993:2716. 5. NCCLS. Performance standards for antimicrobial disk susceptibility tests: approved standard. 8th informational supplement. Document M100-S8. Wayne, PA: NCCLS, 1998. 6. Centers for Disease Control and Prevention (CDC). Standardized molecular subtyping of Escherichia coli by pulsed-eld gel. Atlanta: CDC, 1998. 7. Dean AG, Dean JA, Coulombier D, et al. Epi Info, version 6: a word processing, database, and statistical program for public health on IBMcompatible microcomputers. Atlanta: Centers for Disease Control and Prevention, 1994. 8. SAS/STAT users guide, version 8. Cary, NC: SAS Institute, 1999. 9. Centers for Disease Control and Prevention. Norwalk-like viruses. MMWR Morb Mortal Wkly Rep 2001; 50:118. 10. Ryder R, Sack D, Kapikian A. Enterotoxigenic Escherichia coli and reovirus-like agent in rural Bangladesh. Lancet 1976; 7961:65963. 11. Gorbach SL, Kean BH, Evans DG, et al. Travelers diarrhea and toxigenic Eschericia coli. N Engl J Med 1975; 292:9336.

334 CID 2006:42 (1 February) Beatty et al.