Hamburger-Associated Escherichia coli - NCBI

Hamburger-Associated Escherichia coli 0157:H7 Infection in Las Vegas: A Hidden Epidemic

Palul R. Cieslak, MD, Sulsan1 J. Noble, DVM, Daniiel J. Maxsoni, RS, Lonnie C. Empex, PA, Otto Rav'enholt, MD, MPH, Gretcheni Legarca,

......

Jessica Tuttle, MD, Michael P Dovle, PhD, Timothv J. Barrett, PhD, Jox G. Wells, MS, Anin Marie McNamara, ScD, anid Patricia M. Griffi"n, MD

Introduction Escherichia coli 0157:H7 is an important cause of nonbloody diarrhea, bloody diarrhea, and hemolytic uremic syndrome.' This serotype of E. coli differs from most in that it ferments sorbitol slowly or not at all.' Clinical laboratories can therefore screen stool specimens for this pathogen by plating them on sorbitolMacConkey agar.3 Confirmation of E. coli 01 57:H7 infections by local laboratories in Washington State enabled public health officials there to identify an outbreak in January 1993 and to implicate hamburgers from a particular fast food chain (chain A). Health officials issued a press release on January 18, and the outbreak received widespread national media attention. ' Ultimately, 501 cases were identified in that state, with 45 cases of hemolytic uremic syndrome and 3 deaths.6 Although E. coli 0157:H7 infection was also reportable in Nevada, no cases were reported there during this period. On January 21, 1993, however, a pediatrician caring for a 4-year-old girl with hemolytic uremic syndrome contacted the Clark County (Las Vegas) Health District. According to her mother. the child had recently eaten at a Las Vegas area chain A restaurant. Clark County health officials learned that chain A was in the process of recalling hamburgers from its Las Vegas franchises. On January 22, the health distnrct issued a press release, naming chain A but requesting notification of all cases of bloody diarrhea. The health district was flooded with calls and began an investigation into the magnitude and cause of the outbreak.

1. A....a J

a of P l H

Methods and Materials

Caise Definiitioni A case was defined as either hemolytic uremic syndrome or bloody diarrhea (defined as three or more loose stools in a 24-hour period with visible blood in the stool), with onset between December 1, 1992, and February 7. 1993, in a person who had been in Las Vegas during this interval. Persons from whom a pathogen other than the outbreak strain of E. coli 0157:H7 was isolated were excluded.

Case Finidinig We contacted infection control nurses and microbiologists from all Las Vegas hospitals and representatives from all nephrology practices in the city. looking At the time this paper was written. Paul R. Cieslak was wvith the National Center for Infectious Diseases. Centers for Disease Control and Prevention. Atlanta. Ga. He is now with the Oregon Health Division, Portland. Ore. Susani J. Noble, Jessica Tuttle. Timothy J. Barrett. Joy G. Wells. and Patricia M. Griffin are with the National Center for Infectious Diseases. Centers for Disease Control and Prevention. Daniel J. Maxson. Lonnie C. Empey. and Otto Ravenholt are with the Clark Counts Health District, Las Vegas. Nev. Gretchen Legarza is with the Bureau of Laboratorx Services. Nevada Division of Health. Reno. Michael P. Dov ie is with the Center for Food Safety and Quality Enhancement. Unixersity of Georgia. Georgia Station. Griffin. Ann Marie McNamara is with the Food Safety Inspection Service. US Departmiient of Agriculture. Washington. DC. Correspondence should be sent to Paul R. Cieslak. MD. Oregon Health Dixision. 800 NE Oregon St, Suite 772. Portland. OR 97232-2 162.to Requests for reprints should be sent Foodborne and Diarrheal Diseases Branch. MS A-38. Centers for Disease Control and Prexention. 1600 Clifton Rd. Atlanta. Ga 311333. This paper xwas accepted May 21. 1996.

Februarv 1997. Vol. 87. No. 2

E. coli in Las Vegas

for patients who met the case definition. These individuals and other persons who called the health district were questioned about symptoms. People who reported bloody diarrhea were interviewed by telephone with a standard questionnaire about potential exposures, including eating at any of eight specified fast food hamburger chains in Las Vegas. Persons who ate at a chain A restaurant were asked about specific food items eaten, and ill persons were asked to provide names of meal companions.

Cases

[3 Hemolytic uremic syndrome

6

* Culture confirmed

2.

-n- -m, ,n n Dec 20

Dec 27

1' Jan 3

ll11n 11f1 Jan 10

Jan 17

-m.n. ,n Jan 24

Jan 31

Feb 7

Date

FIGURE 1 -Dates of onset for 58 cases of bloody diarrhea: Las Vegas, December 1992 through February 1993.

Case-Control Study Meal companions of persons with bloody diarrhea who ate at chain A were interviewed by telephone about specific food items eaten. Potential control subjects who reported nonbloody diarrhea, nausea, vomiting, abdominal cramps, or fever were excluded from analysis. Food exposures for patients and their symptomfree meal companions were compared, and matched odds ratios (ORs) with 95% confidence intervals (CIs) were calculated by conditional logistic regression. EXACT computer software was used in conducting regression analyses.7 Twotailed Fisher's exact P values of less than .05 were considered significant. The Mann-Whitney U test was used to compare dates of regular and jumbo hamburger consumption; among two clusters of illness for meal companions, only the first case in each cluster was included.

Laboratory Studies Two commercial laboratories tested all stools sent for bacterial culture from Las Vegas area hospitals. We asked microbiologists at these laboratories about standard procedures and about requests for E. coli 0157:H7 culture. We requested stool specimens of all ill persons and meal companions. Specimens were cultured on sorbitol-MacConkey agar, and at least three sorbitolnegative colonies were selected from each plate and tested for 0157 antigen by means of a latex agglutination reagent (Pro-Lab Diagnostics, Ontario, Canada);3 agglutinating isolates were confirmed as E. coli 0157:H7 via standard biochemical and serologic methods.8 We cultured frozen hamburger patties recalled from chain A restaurants in Nevada, Washington, and Southern California using methods previously described.9 Antibiotic susceptibility was tested by disk diffusion.'0 Shiga-like toxin types were determined with oligonucleotide

February 1997, Vol. 87, No. 2

DNA probes." E. coli 0157:H7 isolates were subtyped by pulsed-field gel electrophoresis, as described previously.'2

Restaurant Investigation The health district reviewed lot numbers of unused, frozen chain A hamburger patties in Las Vegas, compared them with culture results of lots recalled from restaurants in Washington and Southern California, and ensured the recall of implicated lots from Las Vegas restaurants.

Chain A officials demonstrated their standard hamburger cooking procedures. Temperature probes were obtained from health district sanitarians. Grill temperatures were measured with a calibrated surface probe (type K thermocouple model 98221-K; Omega Engineering Inc, Stamford, Conn) before hamburgers were cooked. Internal temperatures were measured via a calibrated probe (type K bead-tip thermocouple; Electronic Controls Design Inc, Milwaukie, Ore) in the center and in four quadrants of each hamburger.

Hamburger Meat Trace Back We visited and assessed operations at the plant where all chain A hamburger patties distributed to West Coast restaurants were produced. We observed meat handling practices, examined quality control records, and reviewed plant records to determine the origin and distribution of implicated production lots.

Results Sixty-one persons with either bloody diarrhea or hemolytic uremic syndrome were identified. Three of these individuals were excluded because other pathogens were cultured from their stools. Of the remaining 58, 47 (81%) initially contacted the health district themselves, 7 (12%) were reported by medical personnel, 2

(3%) were found because they ate a meal at chain A with another patient, 1 (2%) was referred by a friend, and 1 (2%) was found during a telephone search for healthy community control subjects. The dates of symptom onset are shown in Figure 1. Thirty (52%) of the 58 patients were female. The median age was 31 years (range = 10 months to 83 years). All patients reported bloody diarrhea. Other common symptoms included abdominal cramps (reported by 95%), nausea (82%), subjective fever (81%), headache (72%), and vomiting (67%). Thirty-two patients (55%) sought medical attention, 9 (16%) were hospitalized, and 3 (5%) developed hemolytic uremic syndrome. Overall, 54 (93%) of the 58 patients reported eating at chain A in the week before onset. Each of the 14 Las Vegas area chain A restaurants was named by at least 1 patient. One outlet was named by 9 patients, including 5 of the 9 hospitalized patients, all 3 patients with hemolytic uremic syndrome, and the 1 cultureconfirmed patient.

Case-Control Study Of 43 meal companions named by patients, 35 were located. Of these individuals, 2 (6%) met the case definition and another 18 (51%) were excluded because they had had at least one of the following symptoms since December 1, 1992: nonbloody diarrhea (n = 14), nausea (n = 13), vomiting (n = 6), abdominal cramps (n = 14), or subjective fever (n 8). The 15 remaining control subjects were matched with their 10 ill meal companions (there were 1 to 3 control subjects per case). Only consumption of a regular hamburger was significantly associated with illness (matched OR = 9.0, 95% CI = 1.02, 433.4; P = .046). Of the 52 patients who could recall whether or not they had eaten a chain A regular hamburger, 37 =

American Journal of Public Health 177

Cieslak et al.

Cases

2ases

ao Jumbo Burger Only

2 Dec 20

Dec 13

|

4

......... Dec 27

M. . .P4 Jan 3

Jan 10

Jan 17

..

Jan24

Jan 31

Regular Burger Only

i

2

0; *....._ i

Dec 13

Dec20

Dec27

Jan 3

Jan 10

Jan 17

Jan24

Jan31

Date of Meal Note. For the two meal companion clusters, only the meal date for the first case onset is depicted.

FIGURE 2-Dates of chain A meals and hamburger sizes for patients who ate only a regular or only a jumbo hamburger: Las Vegas, December 1992 through January 1993.

(71%) recalled having done so. After regular hamburger eaters had been excluded, no case-control sets discordant for jumbo hamburger consumption remained for analysis, and jumbo hamburger eaters were therefore not implicated statistically. However, among those who ate at chain A but did not eat a regular hamburger, 12 (80%) of 15 ill persons, but only 2 (17%) of 12 control subjects, ate a jumbo hamburger. Figure 2 shows the date of the implicated meal for those patients who reported eating either only regular hamburgers or only jumbo hamburgers. Patients who ate only jumbo hamburgers tended to have eaten later in the outbreak than those who ate only regular hamburgers (P = .01; MannWhitney U test).

Laboratory Studies Stool specimens were collected from 14 patients by their health care providers. Of the two commercial laboratories in Las Vegas that cultured specimens, one had no sorbitol-MacConkey agar until January 27, when the outbreak was essentially over. The other laboratory tested for E. coli 0157:H7 on physician request only; this laboratory received no such request in December 1992 and only one in January 1993. That specimen was obtained at a Las Vegas hospital 19 days after the patient developed diarrhea, and it was not plated on sorbitol-MacConkey agar at the hospital; instead, one colony was selected from a nonselective medium and sent to the commercial laboratory, where it proved not to be 0157. Hence, none of the specimens from the 14 patients tested at 178 American Journal of Public Health

Las Vegas laboratories were processed properly for isolation of E. coli 0157:H7. During the investigation that followed the outbreak, specimens from 32 persons with bloody diarrhea were collected by the health district and cultured at either the state laboratory or the Centers for Disease Control and Prevention; all specimens were tested on sorbitolMacConkey agar for E. coli 0157:H7, and 18 were also tested for Salmonella, Shigella, Yersinia, and Campylobacter. Specimens from 4 persons yielded enteric pathogens: Y enterocolitica (1 person), S. sonnei (1 person), and E. coli 0157:H7 (2 persons). One of the E. coli isolates differed from the Washington State outbreak strain by toxin type and pulsed-field gel electrophoresis pattem. The other isolate was indistinguishable from the Washington strain and from hamburger isolates by antimicrobial susceptibility profile, toxin type, and pulsed-field gel electrophoresis. This specimen was collected by a physician's assistant on February 12, 38 days after the patient became ill and 31 days after his symptoms had resolved. Specimens from the remaining 28 patients yielded no bacterial pathogen. Among the 27 of these patients who ate at chain A and had negative cultures, the first stool specimen was obtained a median of 16.5 days (range = 6 to 57 days) after onset of illness. E. coli 0157:H7 was isolated from frozen hamburger patties from 11 production lots: lots 4 and 9 through 17 of regular patties produced on November 19, 1992, and lot 7 of jumbo patties produced on

November 20, 1992. These isolates of E. coli 0157:H7 produced Shiga-like toxins I and II, and they were indistinguishable from each other and from the Las Vegas and Washington State clinical isolates by antibiogram and pulsed-field gel electrophoresis.

Restaurant Investigation Hamburgers recovered from Las Vegas chain A restaurants during the January 22, 1993, recall included regular hamburgers produced on November 19, 1992 (lots 1, 2, 6, and 12), and jumbo hamburgers produced on November 20, 1992 (lot 6). Before the outbreak was recognized, regular (0.1 lb [0.04 kg]) frozen hamburgers were cooked on a 191PC (375°F) grill for 2 minutes (1 minute per side). As the outbreak unfolded, chain A began cooking regular hamburgers for 2.5 minutes after "searing" (flattening the hamburger onto the grill with a metal press in an attempt to reduce cupping of the hamburger). A final flip of the hamburger was also added to cook any pink juices that might have surfaced. Before the outbreak was recognized, there were two procedures for cooking jumbo (0.25 lb [0.1 kg]) hamburgers: cooking for 2 minutes, covered with a hot metal press, or cooking uncovered for a total of 5 minutes (2.5 minutes per side). During the outbreak, the procedures were changed to 2.5 minutes with the press and 5.5 minutes without the press, both with an initial sear, along with a final flip. Grill temperatures at the Nevada restaurant measured from 186°C to 206°C (367°F to 4020F). After seven hamburgers had been cooked according to the old methods, intemal temperatures ranged from 430C to 87°C (110°F to 1880F), with ranges within a single hamburger of as much as 29°C (53°F) (Figure 3); six of these hamburgers had intemal temperatures below 600C (1400F). When hamburgers were cut open after cooking, areas of pink were seen, corresponding to quadrants with temperatures of less than 600C (1400F). The newer methods with the sear procedure produced hamburgers with temperatures consistently above 680C (1550F), the revised temperature recommended by the US Food and Drug Administration, and they appeared to reduce temperature variation within hamburgers (Figure 3).13

Hamburger Meat Trace Back According to plant records, meat from no single supplier was common to all 11 lots of hamburger patties from February 1997, Vol. 87, No. 2

E. col in Las Vegas

which the outbreak strain of E. coli 0157:H7 was recovered. However, meat from one supplier was used in regular hamburger patty lots 9 through 17 produced on November 19, 1992. Meat left over from the production of those lots was then used the following day, November 20, in several jumbo hamburger lots, including contaminated lot 7. Thus, one supplier could have accounted for 10 of the 11 contaminated lots. Of regular hamburger patties produced on November 19, 1992, with meat from the suspect supplier, 697 boxes (400 patties per box) from suspect lots 9 through 12, 16, and 17 were delivered to the distribution center that served Southern California and Las Vegas. Only 42 (6%) of these boxes were recovered, suggesting that hamburgers in the other 655 boxes had already been served to customers. Of jumbo hamburger patties produced on November 20, 1992, with meat from the suspect supplier, 1267 boxes (168 patties per box) from suspect lots 1 through 4 and 7 through 10 were delivered to the distribution center that served Southern California and Las Vegas. No boxes of jumbo hamburger patties produced on November 20 with meat from the suspect supplier were delivered to the distribution center that served Washington and Idaho.

Discussion At least 58 persons became ill with bloody diarrhea in Las Vegas, Nev, in December 1992 and January 1993; 3 of these individuals developed hemolytic uremic syndrome. As in the concurrent Washington State outbreak, epidemiologic evidence strongly implicated consumption of regular (0.1 lb) hamburgers at fast food chain A.6 E. coli 0157:H7, indistinguishable from the Washington outbreak strain by pulsed-field gel electrophoresis, was isolated from one patient in Las Vegas, and contaminated hamburger lots had been served in Las Vegas restaurants at the time of the outbreak. Chain A cooking procedures at the time of the outbreak were frequently inadequate to kill E. coli 0157:H7.14 The hamburgers were obviously cooked quite unevenly, as demonstrated by the variability of temperatures recorded within a single hamburger (Figure 3); this indicates the difficulty in cooking hamburgers adequately to kill pathogens. It is likely that our case definition's requirement of grossly bloody diarrhea


100 Regular hamburger

90 O %.

T

80

I

C70

Ef (D I-

i1

t Jumbo hamburger

rew o FDA-recommended temperature: 68C (155 TF) Fonner FDA-recomnmended

a)1 .L'¢>

temperture: 60C (140 °F)

60

50 40

A

B

C

Old Methods

D

F E New Methods

Cooking Method Note. Temperatures were recorded at five points in each hamburger; each vertical line connects the temperature points from one hamburger. Method A: regular hamburger cooked for 2 minutes. Method B: jumbo hamburger covered with metal press and cooked for 2 minutes. Method C: jumbo hamburger cooked uncovered for 5 minutes. Method D: regular hamburger initially seared, cooked for 2.5 minutes, and finally flipped to cook surfacing pink juices. Method E: jumbo hamburger initially seared, then covered with metal press, cooked for 2.5 minutes, and finally flipped to cook surfacing pink juices. Method F: jumbo hamburger initially seared, then cooked uncovered for 5.5 minutes, and finally flipped to cook surfacing pink juices.

FIGURE 3-Internal temperatures of chain A hamburgers, by cooking method.

excluded many infected persons. Of 33 meal companions who did not meet the case definition, 14 (42%) had had nonbloody diarrhea during the outbreak period, and at least some of these individuals may have been infected as well. In a large outbreak of E. coli 0157:H7 infections in 1989, 65% of patients had nonbloody diarrhea.'5

Although, in retrospect, this Las Vegas cluster would rank as one of the largest E. coli 0157:H7 outbreaks investigated to date, it was recognized only in the wake of publicity about the Washington cases. Because local laboratories in Washington cultured for the organism, E. coli 0157:H7 was recovered from the stools of more than 70% of persons with bloody diarrhea during January and February 199316; for the 193 reported case patients in Washington who became ill before the public announcement on January 18, the median interval between onset of symptoms and culture for E. coli 0157:H7 was 3 days (B. P. Bell, unpublished 1993 data). By contrast, in Las Vegas, no local laboratory routinely looked for this pathogen in stool specimens, even when the stools were grossly bloody. Consequently, by the time public health officials could collect specimens for proper culture, a

median of more than 2 weeks had elapsed since symptom onset. Because shedding of E. coli 0157:H7 decreases rapidly after illness onset,17 it is not surprising that the organism was recovered from the stool of only one Las Vegas patient. The significant dimensions of this outbreak became known only because patients called the health district in response to a press release; this illustrates the importance of alerting the public to potential outbreaks of infectious diseases. The data strongly suggest that jumbo hamburgers were responsible for some of the illnesses in Las Vegas. Almost a quarter of ill persons reported eating only jumbo hamburgers. The different epidemic curves for illness associated with regular hamburgers and illness associated with jumbo hamburgers argue against a significant recall bias by Las Vegas respondents. These epidemiologic data were corroborated by the finding that some lots of jumbo hamburgers produced on November 20, 1992, contained meat used in the production of implicated regular hamburgers produced on the previous day. Furthermore, the outbreak strain of E. coli 0157:H7 was cultured from one of these jumbo hamburger lots. Finally, the lack of jumbo hamburgerAmerican Journal of Public Health 179

Cieslak et al.

associated illness in Washington State is explained by the finding that 1267 boxes of jumbo hamburgers in suspect lots produced on November 20 had been shipped to the Las Vegas area chain A distributor, whereas none had been shipped to the Washington distributor. Thus, although recall of regular hamburgers alone would have ended the outbreak in Washington, it would have been insufficient to end the epidemic in Las Vegas, demonstrating that there may be important epidemiologic differences among affected sites in multistate common-source outbreaks. Infection with E. coli 0157:H7 may be the most common cause of grossly bloody diarrhea in the United States,'8 and the US Council of State and Territorial Epidemiologists has recommended that it be made reportable in all US states and territories.'9 Although E. coli 0157:H7 infection was already reportable in Nevada, this large outbreak would not have been detected in the absence of publicity from Washington because local laboratories did not screen for the pathogen. The passive surveillance that existed at the outbreak onset was clearly inadequate. In response, Clark County Health District officials have conferred with representatives of the two major laboratories; one has since begun to screen, with sorbitolMacConkey agar, all stools submitted for bacterial culture, and the other screens all grossly bloody stools. Reporting is further stimulated through liaison with hospital infection-control nurses and primary care physicians. Nineteen cases were reported to the health district in 1994, and another 16 were reported in 1995. Mandatory reporting of hemolytic uremic syndrome might have brought the outbreak to light in a more timely fashion. The outbreak was ultimately discovered when a pediatrician reported a case of hemolytic uremic syndrome to the health district. However, the other two cases of hemolytic uremic syndrome (one of which occurred in late December) were voluntarily reported only after the health district's January 22 press release. State officials are currently working to make hemolytic uremic syndrome reportable in Nevada. We recommend that all microbiology laboratories that culture stools for

180 American Journal of Public Health

bacterial pathogens screen stools from persons with bloody diarrhea routinely for E. coli 0157:H7 with sorbitol-MacConkey agar and agglutination with 0157 antisera. While H type is being confirmed at a reference laboratory, the identification of E. coli 0157 should be reported to the clinician and to public health officials. Clinicians should report cases of hemolytic uremic syndrome to their local health departments as sentinel events for E. coli 0157:H7 infection. Restaurants that serve hamburgers should heed the Food and Drug Administration recommendation that all parts of hamburgers be cooked to an internal temperature of at least 68°C (155°F), and consumers should be made aware of the potential hazards of eating undercooked ground beef. D1

Acknowledgments We wish to acknowledge Dr Robert V. Tauxe and Dr Stephen M. Ostroff for their extensive helpful advice; Dr Thomas M. Gomez and Dr David A. Puerto for their work in tracing the contaminated hamburgers; Paul L. Klouse for his work with restaurant investigations; Judith Tawatao, Philippa Pointon, and Kathleen McLaren for interviewing patients and control subjects; and Katherine Greene and James Green for laboratory assistance.

References 1. Griffin PM, Tauxe RV. The epidemiology of infections caused by Escherichia coli 0157:H7, other enterohemorrhagic E. coli, and the associated hemolytic uremic syndrome. Epidemiol Rev. 1991;13:60-98. 2. Ratnam S, March SB, Ahmed R, Bezanson GS, Kasatiya S. Characterization of Escherichia coli serotype 0157:H7. J Clin Microbiol. 1988;26:2006-2012. 3. March SB, Ratnam S. Latex agglutination test for detection of Escherichia coli serotype 0157. J Clin Microbiol. 1989;27: 1675-1677. 4. Kusumoto K. Sickness linked to Jack in the Box. The Seattle limes. January 19, 1993:B1-B2. 5. Grogan D, Hathorn C. Rare and deadly. People Weekly. February 15, 1993:48-50. 6. Bell BP, Goldoft M, Griffin PM, et al. A multistate outbreak of Escherichia coli 0157:H7-associated bloody diarrhea and hemolytic uremic syndrome from hamburgers: the Washington experience. JAMA.

1994;272:1349-1353. 7. Martin D, Austin H. An efficient program for computing conditional maximum likelihood estimates and exact confidence limits for a common odds ratio. Epidemiology.

1991 ;2:359-362.

8. Ewing WH. Edwards and Ewing's Identification of Enterobactericeae. 4th ed. New York, NY: Elsevier Science Publishing Co Inc; 1986:93-134. 9. Padhye NV, Doyle MP. Rapid procedure for detecting enterohemorrhagic Escherichia coli 0157:H7 in food. Appl Environ Microbiol. 1991 ;57:2693-2698. 10. Barry AL, Thormsberry C. Susceptibility tests: diffusion test procedures. In: Balows A, Hausler WJ Jr, Herrmann KL, Isenberg HD, Shadomy HJ, eds. Manual of Clinical Microbiology. 5th ed. Washington, DC: American Society for Microbiology; 1993: 1117-1125. 11. Strockbine NA, Faruque SM, Kay BA, et al. DNA probe analysis of diarrhoeagenic Escherichia coli: detection of EAFpositive isolates of traditional enteropathogenic E. coli serotypes among Bangladeshi paediatric diarrhoea patients. Mol Cell Probes. 1992;6:93-99. 12. Keene WE, McAnulty JM, Hoesly FC, et al. A swimming-associated outbreak of hemorrhagic colitis caused by Escherichia coli 0157:H7 and Shigella sonnei. N Engl JMed. 1994;331:579-584. 13. US Food and Drug Administration. Food Code. 1993 Recommendations of the United States Public Health Service Food and Drug Administration. Washington, DC: US Dept of Health and Human Services; 1993:50. 14. Tauxe RV, Hughes JM. Food-borne disease. In: Mandell GL, Bennett JE, Dolin R, eds. Mandell, Douglas and Bennett's Principles and Practice of Infectious Diseases. 4th ed. New York, NY: Churchill Livingstone; 1995:1012-1024. 15. Swerdlow DL, Woodruff BA, Brady RC, et al. A waterborne outbreak in Missouri of Escherichia coli 0157:H7 associated with bloody diarrhea and death. Ann Intern Med. 1992;117:812-819. 16. Centers for Disease Control and Prevention. Multistate outbreak of E. coli 0157:H7 infections. MMWR Morb Mortal Wkly Rep. 1993;42:258-263. 17. Tarr PI, Neill MA, Clausen CR, Watkins SL, Christie DL, Hickman RO. Escherichia coli 0157:H7 and the hemolytic uremic syndrome: importance of early cultures in establishing the etiology. J Infect Dis. 1990;162:553-556. 18. Ries A, Griffin P, Greene K, Escherichia coli 0157:H7 Study Group. Escherichia coli 0157:H7 diarrhea in the United States: a 10 center surveillance study. In: Program and abstracts of the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy, October 1993, New Orleans, La. Abstract 1454. 19. CSTE Position Statement #4: National Surveillance of Escherichia coli 0157:H7. Atlanta, Ga: Council of State and Territorial Epidemiologists; 1993.
