A STATEWIDE OUTBREAK OF ESCHERICHIA COLI 0157:H7

AMERICAN JOURNAL OF EPIDEMIOLOGY VOL 132, No. 2Copyright O 1990 by The John* Hopkins University School of Hygiene and Public Health Printed in U.S.A.All rights reserved

A STATEWIDE OUTBREAK OF ESCHERICHIA COLI 0157:H7INFECTIONS IN WASHINGTON STATE

STEPHEN M. OSTROFF,1 PATRICIA M. GRIFFIN,2 ROBERT V. TAUXE,2

LARRY D. SHIPMAN,2 KATHERINE D. GREENE,1 JOY G. WELLS,1

JAY H. LEWIS,' PAUL A. BLAKE,1 AND JOHN M. KOBAYASHI*

Ostroff, S. M. (CDC, Atlanta, GA 30333), P. M. Griffin, R. V. Tauxe, L. D. Shipman,K. D. Greene, J. G. Wells, J. H. Lewis, P. A. Blake, and J. M. Kobayashi. Astatewide outbreak of Eschertchla coll 0157:H7 infections in Washington State.Am J Epidemiol 1990; 132:239-47.

In November 1986, a statewide outbreak of Eschertchla coll 0157:H7 infectionsin Washington State was Identified after a physician in an eastern Washingtoncommunity hospitalized three patients with hemorrhagic colitis which progressedto thrombotic thrombocytopenic purpura. Epidemiologic investigation identified37 cases in this community and linked the illnesses to a local restaurant whichhad served ground beef that was the suspected initial vehicle of transmission.The plasmid profile and toxin production pattern (Shiga-like toxin II alone) of theoutbreak strain provided a unique strain marker. E. coll 0157:H7 Infections causedby this strain were simultaneously seen in other parts of the state among nursinghome residents and in patients with the hemolytic-uremic syndrome, and anincrease in sporadic cases of hemorrhagic colitis was noted at a Seattle healthmaintenance organization. It is suspected that a contaminated product, probablyground beef distributed statewide, was the common source. Tracing of this meatled to farms where rectal swabs from six (1%) of 539 cattle tested yielded £. coll0157:H7, although the plasmids and toxin production patterns of these isolatesdiffered from the human outbreak strain. Introduction of a single strain of E. coll0157:H7 has the potential to cause widespread concurrent outbreaks. Suchoutbreaks are likely to escape recognition until heightened screening and sur-veillance for E. coll 0157:H7 is established.

Escherlchla coll infections; hemolytic-uremic syndrome; purpura, thromboticthrombocytopenic

Escherichia coli 0157:H7, first identified has emerged as a major cause of outbreaksas a pathogen during the investigation of (2-6) and sporadic cases (7-11) of bloodytwo outbreaks in Oregon and Michigan (1), diarrhea in North America. Its clinical

Received for publication August 24, 1989, and in assistance in the investigation; Drs. Richard Simon,final form February 19, 1990. Lawrence Zawatsky, and the other physicians and

1 Division of Field Services, Epidemiology Program hospital staff in Walla Walla, Washington, for iden-Office, Centers for Disease Control, Atlanta, GA. tifying the patients; Dr. Michael Doyle and the staff

2 Enteric Diseases Branch, Center for Infectious of the Food Research Institute, Madison, Wisconsin,Diseases, Centers for Disease Control, Atlanta, GA. for testing food specimens; Drs. Marguerite Neil],

1 Washington State Department of Social and Phillip Tarr, and Paul Ramsey of the University ofHealth Services, Seattle, WA. Washington School of Medicine for input regarding

Reprint requests to Dr. Stephen M. Ostroff, Enteric the investigation; Dr. Nancy Hargrett-Bean for statis-Diseases Branch, Mailstop CO9, Centers for Disease tical assistance; Mary Chadden and Bertha Smith forControl, Atlanta, GA 30333. help with the manuscript; Loraine Good for technical

The authors thank the staff of the Walla Walla editing; and the people of Walla Walla for their as-City-County Health Department and the Columbia sistance in the outbreak investigation.County Health Department, Dayton, Washington, for

239

at Pennsylvania State University on M

ay 17, 2016http://aje.oxfordjournals.org/

Dow

nloaded from

http://aje.oxfordjournals.org/

240 OSTROFF ET AL.

spectrum of illness includes asymptomaticcarriage and nonbloody diarrhea (12), butthe hallmark of infection is a distinctivesyndrome characterized by painful, bloodydiarrhea with little or no fever, termedhemorrhagic colitis (13). The hemolytic-uremic syndrome and thrombotic throm-bocytopenic purpura, two similar condi-tions marked by renal impairment, hemo-lytic anemia, and thrombocytopenia, havealso been linked to infection with E. coli0157:H7 (14-17).

Since 1982, outbreaks have been recog-nized in Canada (3, 4), the United States(2, 5, 6), and Great Britain (18). Most haveoccurred in the institutional settings ofday-care centers (5), nursing homes (2, 3),and schools (4); community outbreaks of E.coli 0157:H7 have only rarely been de-scribed (1, 18). We recently investigated astatewide outbreak of E. coli 0157:H7-associated illnesses in Washington State.The largest number of cases occurred inWalla Walla, where ground beef served ata local fast food restaurant was implicatedin transmission. Cases were also detectedin nursing home patients in Dayton, a com-munity 50 miles east of Walla Walla, andin patients in the Puget Sound region ofwestern Washington. The simultaneous oc-currence of cases in various parts of thestate suggests that the implicated meat hadbeen widely distributed. Despite its wide-spread nature, the outbreak was detectedonly when a physician in Walla Wallatransferred three women hospitalized withthrombotic thrombocytopenic purpura to atertiary care center. This investigationdemonstrates the difficulty in recognizingoutbreaks of E. coli 0157:H7, even in a smallcommunity and in spite of the dramaticclinical presentations, and highlights theneed for improved surveillance of thispathogen in the United States.

MATERIALS AND METHODS

Walla Walla investigation

A primary case was defined as illness ina Walla Walla resident with onset between

October 29 and November 19, 1986, withE. coli 0157:H7 isolated from a stool speci-men or 1 or more days of visibly bloodydiarrhea of unknown cause during the out-break period. A secondary case was definedas culture-confirmed E. coli 0157:H7 illnessin a household contact of a primary casewhich began at least 48 hours after theonset of illness in the primary case.Hemolytic-uremic syndrome was defined asthe triad of acute microangiopathic hemo-lytic anemia (hematocrit less than 30 per-cent and evidence of red cell fragmentationon a peripheral smear), thrombocytopenia(platelets less than 150,000 mm3), and im-paired renal function (creatinine more than2.0 mg/dl). Patients diagnosed with throm-botic thrombocytopenic purpura had thehemolytic-uremic syndrome triad, fever,and a newly diagnosed neurologic deficit.

The investigation began on November 14after the Washington State Health Depart-ment was contacted about the transfer ofthe three women with thrombotic throm-bocytopenic purpura from Walla Walla toSeattle. To identify other cases, we askedlocal physicians and hospitals about pa-tients evaluated for bloody diarrhea,thrombotic thrombocytopenic purpura, orhemolytic-uremic syndrome since mid-October, examined the stool culture logs ofthe four clinical microbiology laboratoriesin Walla Walla and the emergency roomlogs of the two local hospitals, and attempt-ed to contact all patients who had submit-ted a stool specimen in which no pathogenwas isolated or who had visited an emer-gency room for diarrhea or abdominal painafter October 15. Announcements in thelocal media asked persons with recentbloody diarrhea to contact the health de-partment.

For determination of the source for theoutbreak, an age- and sex-matched case-control investigation (case-control investi-gation 1) involved the first 18 identifiedprimary cases. Cases or their parents wereinterviewed about dates of onset of illness,signs and symptoms, and exposures duringthe 10 days before the illness began. Con-



Dow

nloaded from


ESCHERICHIA COLI 0157:H7 IN WASHINGTON STATE 241

trols were patients who had visited thearea's largest outpatient health center onthe date the case became ill. Potential con-trols were excluded if they had had diarrheaor abdominal pain during the outbreakperiod or had had an underlying illnesswhich precluded potential exposures. Con-trols were interviewed about the 10-dayperiod before the clinic visit.

A second, unmatched case-control inves-tigation (case-control investigation 2) wasperformed to determine the vehicles for E.coli 0157:H7 transmission in the implicatedrestaurant. Cases included all 27 personswho met the initial case definition whoreported eating food from the restaurant inthe 10 days before they became ill. Twocontrol groups were chosen. The first (mealcompanion controls) consisted of well per-sons who had eaten at the restaurant withthe cases during the outbreak period. Thesecond (check controls) consisted of wellpersons who paid for food at the restaurantby bank check during the outbreak period.

All cases were asked to submit stool spec-imens. Swabs from these specimens wereplaced in Cary-Blair transport medium(Medical Media Lab, Boring, OR) andshipped refrigerated to the WashingtonState Public Health Laboratory or shippedfrozen to the Centers for Disease Control.Specimens sent to the Washington StatePublic Health Laboratory were testedfor Salmonella, Shigella, and Campylo-bacter, and were inoculated onto sorbitol-MacConkey agar (19, 20). Colonies that didnot ferment sorbitol after 24 hours of in-cubation were serotyped by tube aggluti-nation for the 0157 and H7 antigens usingCenters for Disease Control reagents (21).The specimens sent to the Centers for Dis-ease Control were tested, serotyped, andanalyzed for plasmid DNA and Shiga-liketoxin (also known as verotoxin) activityusing methods described previously (12).

Restaurant employees were interviewedabout diarrheal illness during the outbreakperiod, and rectal swabs were taken, placedin Cary-Blair medium, and transported fro-zen to the Centers for Disease Control for

testing. Employee work schedules werecompared with dates on which cases hadvisited the restaurant. Food-handling pro-cedures were reviewed, and local sanitari-ans inspected the restaurant.

Beginning on November 14, samples ofraw and cooked foods were obtained fromthe restaurant. All samples were culturedfor E. coli 0157:H7 by Dr. M. Doyle at theFood Research Institute in Madison, Wis-consin, using a previously described meth-od (22). E. coli 0157:H7 food isolates werereferred to the Centers for Disease Controlfor further characterization.

Dayton investigation

In Dayton, a case was defined as a nurs-ing home resident who had had E. coli0157:H7 cultured from his or her stool dur-ing the outbreak period. The three casesresided in two nursing homes which sharedthe same kitchen. Nursing home staff wereinterviewed about recent illness among pa-tients and staff and about foods servedduring the outbreak period. Between No-vember 23 and 30, rectal swabs were ob-tained from nursing staff with diarrhealsymptoms, all kitchen staff, and all resi-dents of the smaller of the two nursinghomes. These were tested for E. coli0157:H7 at the Washington State PublicHealth Laboratory using the methods de-scribed above. Plasmid and Shiga-like toxinanalysis on E. coli 0157:H7 isolates fromDayton cases were performed at the Cen-ters for Disease Control.

Food tracing and animal culturing

Records at the beef and dairy slaughter-houses which processed ground beef usedat both the restaurant and nursing homeswere reviewed to determine farm sources.In January 1987, stool or rectal swabs fromrandomly selected milking cows, heifers,and calves on farms in southwest Washing-ton and adjacent Oregon identified by theslaughterhouse record review were placedin Cary-Blair medium and tested at theCenters for Disease Control by using theprocedures described above.



Dow

nloaded from


242 OSTROFF ET AL.

Other Washington State cases

Data on E. coli 0157:H7 infections inWashington were available on patientswith hemolytic-uremic syndrome andthrombotic thrombocytopenic purpura hos-pitalized in Seattle during 1985 to 1987, onE. coli 0157:H7 isolates found at a Seattle-based health maintenance organizationwhich routinely screened all submittedstools for this organism, and on isolatessent to the Washington State PublicHealth Laboratory for identification orconfirmation. Available E. coli 0157:H7 iso-lates from these sources were tested at theCenters for Disease Control for plasmidand Shiga-like toxin profiles.

Statistical methods

The case-control investigations used theMantel-Haenszel chi-square test for oddsratios with exact confidence limits calcu-lated by the method of Thomas (23). TheFisher's exact test was used when appro-priate. All p values are two-tailed.

RESULTS

Walla Walla investigation

We identified 37 cases among WallaWalla-area residents, including 14 (38 per-cent) with culture-confirmed E. coli0157:H7 infection. No other pathogenswere cultured from the stool specimens.The median age of cases was 36 years(range, 11 months to 78 years), and 30 (81percent) were female. Thirty-five were pri-mary cases; two were secondary cases (themother of a 4-year-old case and the grand-mother of a 1-year-old case). The first casebecame ill on October 29, and primary casescontinued to occur through November 19(figure 1). Seventeen (46 percent) of the 37cases were hospitalized, two underwent ab-dominal surgery, and two died (table 1).

The 14 E. coli 0157:H7 isolates had sim-ilar plasmid profiles with plasmids of ap-proximately 65, 22, and 2 megadaltons; 13isolates produced only Shiga-like toxin II,while one isolate produced neither Shiga-like toxin I nor II. Comparison of this com-

Cases

Secondary

29 31 2 4 6 8 10 12 14 16 18 20 22 24

Oct Onset

FIGURE 1. Cases of Escherichia coli 0157:H7 infec-tion, by onset date, Walla Walla and Dayton, WA,October 29 to November 24, 1986.

TABLE 1

Clinical manifestations of illness reported by the 37patients* with Escherichia coli 0157:H7 infections in

Walla Walla, WA, October-November 1986

Manifestation

DiarrheaAbdominal crampsBloody diarrheaVomitingDocumented fever (>38*C)Thrombotic thrombocytopenic purpuraHemolytic-uremic syndromeAbdominal surgeryHospitalizedDeath

No.

37373620

7312

172

%

100100975629835

466

* The denominator varies for different clinicalmanifestations because cases either could not recall orwere uncertain if the manifestation was present.

bined plasmid profile and Shiga-like toxinpattern with 113 other Washington Stateisolates revealed that this strain had notbeen detected before October 1986 and wasnot seen again during the period of Decem-ber 1986 to December 1987.

In case-control investigation 1, we foundthat eating food from a local Mexican-stylefast food restaurant in the 10 days beforethe onset of illness was significantly asso-ciated with illness. Thirteen (72 percent) of18 cases ate food from this restaurant com-pared with only one (6 percent) of 18 con-trols (matched odds ratio (OR) = 12/0,lower 95 percent confidence limit 3.4, p =0.0002).

Overall, 27 (77 percent) of the 35 primary



Dow

nloaded from


ESCHERICHIA COU 0157:H7 IN WASHINGTON STATE 243

cases ate food from this restaurant duringthe outbreak period, including nine of the12 culture-confirmed primary cases. Forthe 22 cases with a single restaurant visit,the mean interval between this visit andthe onset of illness was 3.1 days (range,1-8 days). In a 1-day survey at the restau-rant, 68 percent of patrons were female, aproportion not significantly different fromthe 81 percent female preponderanceamong cases.

In case-control investigation 2, exposurehistories of the 27 restaurant-associatedcases were compared with those of 26 mealcompanion controls and 158 check controls.Because there were no significant differ-ences in food consumption histories be-tween the two control groups, they weremerged in the analysis.

Restaurant foods consisted of nine baseingredients combined into 22 menu items,as well as condiments and beverages. Ex-posure rates were high for each of the in-gredients and were not significantly differ-ent between cases and controls. All casesand controls ate cheese; ground beef waseaten by 89 percent of cases and 91 percentof controls. Among the menu items, non-dairy ranch dressing was used by 14 (52percent) of 27 cases compared with 53 (29percent) of 184 controls (OR = 2.6, 95percent confidence interval (CI) 1.2-5.8, p= 0.02). Because this item could accountfor no more than 52 percent of cases, weexamined those cases and controls who didnot report use of the ranch dressing; five(38 percent) of the remaining 13 cases atecrisp meat burritos compared with 19 (15percent) of the remaining 131 controls (OR= 3.7, 95 percent CI 1.2-11.8, p = 0.04).These findings were similar when onlyculture-confirmed cases were considered.

Only one of 22 employees reported hav-ing diarrhea (which was nonbloody) eitherduring the outbreak or in the month beforethe outbreak; this illness occurred in lateOctober. The 22 rectal swabs from thisgroup did not yield E. coli 0157:H7. Exam-ination of work schedules showed no rela-tion between dates and times when cases

visited the restaurant and any single em-ployee or work shift.

Food items were not routinely monitoredfor cooking time or temperature; however,inadequate cooking times were observed byinvestigators. No storage or holding viola-tions were detected. Meat for all menuitems was prepared and used on the sameday except for the meat used in crisp meatburritos, which was cooked, cooled, rolledinto shells, and stored refrigerated for upto 3 days.

One of 29 food samples from the restau-rant yielded E. coli 0157:H7. This samplewas from a 10-pound bag of raw gound beefcollected on November 14. Although nomeat was available from the outbreakperiod, the bag of E. coli 0157:H7-positiveground beef came from the restaurant'susual supplier. The E. coli 0157:H7 meatisolate produced Shiga-like toxin I and hada plasmid pattern different from that of theoutbreak-related human strain.

Dayton investigation

Three (5 percent) of 56 Dayton nursinghome residents had hemorrhagic colitiswhich began between November 17 and 19(figure 1), and all three had the outbreakstrain of E. coli 0157:H7 isolated from astool sample. No other residents of eithernursing home were reported to have a diar-rheal illness during this period. None of thecases had left the nursing home or eatenfood from outside the nursing home in the2 weeks before the onset of illness. Nocontact was identified between staff ornursing home residents in Dayton and theimplicated Walla Walla restaurant or anyof the Walla Walla cases. One staff memberhad nonbloody diarrhea which began onNovember 17, but none of the 26 specimensobtained from staff and other nursing homeresidents yielded E. coli 0157:H7.

Ground beef was served on three occa-sions at both nursing homes during theweek of November 10. Although govern-ment surplus meat was usually served atthe nursing homes, the kitchen supervisorhad obtained ground beef from another



Dow

nloaded from


244 OSTROFF ET AL.

source during October 1986, had kept itfrozen, and did not use it until the week ofNovember 10. It came from the same sup-pliers as ground beef served in the WallaWalla restaurant and was the only fooditem in common between the nursinghomes and restaurant, but none remainedfor testing. All three cases were served thismeat. Review of meat-handling proceduresin the nursing homes' kitchen revealed nodeficiencies.

Meat tracing and animal culturing

By reviewing nursing home and restau-rant purchase records, we traced the groundbeef through grinding plants to dairy andbeef cattle slaughterhouses. Only the dairycattle component of the meat could be ad-equately traced beyond the slaughterhouseto farm sources. In January 1987, 539 rectalswabs or manure samples were obtainedfrom dairy cattle on nine of over 200 farmswhich were possible sources of the nursinghome and restaurant ground beef. E. coli0157:H7 was detected in six (1 percent) ofthe rectal swabs from animals on four (44percent) of the nine farms. None of theplasmid and Shiga-like toxin profiles ofbovine isolates were identical to the out-break strain or the restaurant meat isolate.

Other Washington State cases

Thirteen patients, including the threetransferred from Walla Walla, were hospi-talized in Seattle with hemolytic-uremicsyndrome or thrombotic thrombocytopenicpurpura during the last 3 months of 1986;this represented a large increase over thenumber during the same period in 1985(four cases) and 1987 (no cases). E. coli0157:H7 isolates were available from sevenof the 10 patients with hemolytic-uremicsyndrome or thrombotic thrombocytopenicpurpura in 1986 who were not associatedwith the Walla Walla outbreak. Four of theseven isolates produced only Shiga-like tox-in II and had a plasmid profile indistin-guishable from that of the Walla Wallastrain. Three of the patients with theseisolates became ill between October 21 and

November 24 and one on December 13,while the three patients with nonoutbreakstrains became ill on October 21, November24, and November 29. Three of the patientswith the outbreak strain were from theSeattle area, and one was from easternWashington. All had eaten ground beef inthe 10 days before the onset of illness frommultiple, unrelated sources; none had beento Walla Walla or Dayton.

A marked increase in the isolation of E.coli 0157:H7 occurred at the Seattle-basedhealth maintenance organization coinci-dent with the Walla Walla outbreak period.Thirteen E. coli 0157:H7 isolates were de-tected in October and 15 in November,yielding an isolation rate during these 2months of 21.5 per 1,000 stools screened.This is the highest bimonthly isolation ratefor E. coli 0157:H7 since screening beganin early 1985. Isolates from the healthmaintenance organization patients werenot available for study, but two isolates sentto the Washington State Public HealthLaboratory from western Washington pa-tients with sporadic hemorrhagic colitisduring the outbreak period were indistin-guishable from the outbreak strain.

DISCUSSION

E. coli 0157:H7 was most likely intro-duced into the restaurant through groundbeef, the vehicle of transmission in otheroutbreaks caused by this pathogen (1, 2, 6,24). Although the outbreak strain of E. coli0157:H7 was not isolated from ground beefsamples or cattle, this conclusion is sup-ported by the association of some cases withcrisp meat burritos made with ground beef,isolation of E. coli O157:H7 from a packageof ground beef taken from the restaurant,and linkage of the common-source groundbeef to the affected restaurant and thenursing homes. Ground beef from a slaugh-terhouse which supplied meat to the restau-rant was distributed throughout the stateduring the outbreak period, which couldexplain the simultaneous isolation of theoutbreak strain throughout Washington.

The high exposure rates of both cases



Dow

nloaded from



and controls to ground beef in the restau-rant made epidemiologic evaluation of thisitem difficult. Furthermore, contaminationor inadequate cooking of the ground beefmay have occurred only intermittently inthe restaurant. Cross-contamination fromthe ground beef to other food items in therestaurant, such as the ranch dressing,could have made multiple vehicles respon-sible for illness. The ranch dressing aloneis unlikely to have caused the outbreak,since laboratory studies demonstrated thatit would not support the growth of E. coli0157:H7 under restaurant conditions (M.Doyle, unpublished observations).

Although we detected strains of E. coli0157:H7 in the ground beef sample and incattle, none were identical with the out-break strain. There are no data about thefrequency of this pathogen in WashingtonState meats and cattle, but data from othersources suggest these findings are unusualin the United States. In Wisconsin, onlyone of the 147 ground beef samples testedat the time of the Washington outbreakcontained E. coli 0157:H7 (22). In the samestudy, five of 17 samples from Calgary,Alberta, Canada, contained the organism.Tracing ground beef back to farm sourceswas difficult; none of the beef cattle meatand only some of the dairy cattle meat couldbe traced, and the large number of possiblycontributing farms made comprehensiveculturing impossible. In addition, animalcultures were taken months after the out-break had occurred. Nonetheless, our find-ings provide further evidence that dairycattle can be a reservoir for E. coli 0157:H7(25, 26).

This report illustrates the difficulties in-volved in detecting outbreaks due to E. coli0157:H7 which are not tightly clusteredeither geographically or temporally. We be-lieve it is unlikely that the outbreak wouldhave been recognized at all had not threepatients developed thrombotic thrombocy-topenic purpura requiring transfer to a ter-tiary care center. It had gone undetectedfor 2 weeks despite the dramatic clinicalpresentations and high number of hospital -

izations in a relatively small community. Itis also unlikely that the three Dayton nurs-ing home cases would have had specimenscultured if not for the investigation of ill-nesses in nearby Walla Walla. This sug-gests that small clusters of cases may havegone unrecognized in many other locationsduring the outbreak period and that thetrue magnitude of the outbreak may havebeen quite large. It also suggests that someof the other reported institutional out-breaks of E. coli 0157:H7 (2-6) might haverepresented only one facet of larger com-munity outbreaks.

It is likely that many infections and out-breaks of E. coli 0157:H7 escape detectionentirely. Many physicians are not yet awareof this pathogen, routine stool cultures donot detect E. coli 0157:H7, and screeningprocedures for isolating this organism arenot widely practiced in this country.Sorbitol-MacConkey culture medium takesadvantage of the fact that E. coli 0157:H7ferments sorbitol slowly, a feature seen inonly 5 percent of E. coli (20). Any E. coliwhich is 9orbitol-negative after 24 hours ofincubation can then be serotyped. In Wash-ington State during 1986, only a few clin-ical laboratories routinely used sorbitol-MacConkey agar, and this is likely to haveresulted in a marked underestimate of thetrue magnitude of the outbreak.

We may have also underestimated thesize of this outbreak because of our strin-gent case definition. Other investigationshave demonstrated that E. coli 0157:H7 cancause a spectrum of illnesses which includesnonbloody diarrhea and asymptomatic car-riage (8, 27). While cases of nonbloody diar-rhea occurred in our outbreak, it is unlikelythat the number of such cases was large.The Seattle-based health maintenance or-ganization and the pediatric facility wherethe hemolytic-uremic syndrome cases werehospitalized routinely screened all stoolspecimens for E. coli 0157:H7 during theoutbreak period, and only one patient withnonbloody diarrhea grew this organism. Inaddition, the number of specimens submit-ted for culture from patients with non-



Dow

nloaded from


246 OSTROFF ET AL.

bloody diarrhea did not increase in eitherWalla Walla or the Seattle health mainte-nance organization during the outbreakperiod.

Hemolytic-uremic syndrome and throm-botic thrombocytopenic purpura are nowclearly recognized sequelae of infectionwith E. coli 0157:H7 and were the heraldevents for this outbreak. While this orga-nism is not the only pathogen associatedwith these conditions (28, 29), recent serieshave shown that the majority of hemolytic-uremic syndrome cases in at least two areasof North America can be linked to E. coli0157:H7 (15, 16). Therefore, patients whopresent with hemolytic-uremic syndromeor thrombotic thrombocytopenic purpuraafter bloody diarrhea should be a signal topublic health authorities and medical prac-titioners to be alert for sporadic hemor-rhagic colitis in the community and toensure the availability of appropriatediagnostic tests for E. coli 0157:H7.

In summary, this outbreak demonstratesthe impact of the introduction of a singlestrain of E. coli 0157:H7 into the food sup-ply and the ease with which the outbreakcould have escaped detection. Becausewidespread distribution of food products inthe United States is common, outbreaks onan even wider scale, involving parts of theUnited States where this pathogen is rarelydetected, could occur. The use of plasmidprofiles and Shiga-like toxin productionpatterns was useful in determining theextent of the outbreak, as was routinelaboratory-based surveillance. Improvedclinical recognition of syndromes associ-ated with E. coli 0157:H7 infections, in-creased culturing of stools for the pathogen,and additional studies of its prevalence infoods will further our understanding of thisrapidly emerging pathogen.

REFERENCES

1. Riley LW, Remis RS, Helgerson SD, et al. Hem-orrhagic colitis associated with a rare Escherthiacoli serotype. N Engl J Med 1983;308:681-5.

2. Ryan CA, Tauxe RV, Hosek GW, et al. Esche-richia coli 0157:H7 diarrhea in a nursing home:clinical, epidemiologic, and pathologic findings. J

Infect Dis 1986:154:631-8.3. Carter AO, Borczyk AA, Carlson JAK, et al. A

severe outbreak of Escherichia coli 0157:H7-associated hemorrhagic colitis in a nursing home.N Engl J Med 1987;317:1496-500.

4. Laboratory Centres for Disease Control. Outbreakof gastrointestinal disease—Ontario. Can DisWkly Rep 1987;13:5-8.

5. Spika JS, Parsons JE, Nordenberg D, et al. He-molytic uremic syndrome and diarrhea associatedwith Escherichia coli 0157:H7 in a day care center.J Pediatr 1986;109:287-91.

6. Pavia AT, Nichols CR, Green DP, et al. Hemo-lytic-uremic syndrome during an outbreak ofEscherichia coli 0157:H7 infections in institutionsfor mentally retarded persons: clinical and epide-miologic observations. J Pediatr 1990;116:544-51.

7. Remis RS, MacDonald KL, Riley LW, et al. Spo-radic cases of hemorrhagic colitis associated withEscherichia coli 0157:H7. Ann Intern Med 1984;101:624-6.

8. Pai CH, Gordon R, Sims HV, et al. Sporadic casesof hemorrhagic colitis associated with Escherichiacoli 0157:H7: clinical, epidemiologic, and bacteri-ologic features. Ann Intern Med 1984;101:738-42.

9. MacDonald KL, O'Leary MJ, Cohen ML, et al.Escherichia coli 0157:H7, an emerging gastrointes-tional pathogen: results of a one-year, prospective,population-based study. JAMA 1988;259:3567-70.

10. Pai CH, Ahmet N, Lior H, et al. Epidemiology ofsporadic diarrhea due to verocytotoxin-producingEscherichia coli: a two-year prospective study. JInfect Dis 1988;157:1054-7.

11. Ostroff SM, Kobayashi JM, Lewis JH. Infectionswith Escherichia coli 0157:H7 in WashingtonState: the first year of statewide disease surveil-lance. JAMA 1989;262:355-9.

12. Griffin PM, Ostroff SM, Tauxe RV, et al. Ilnessesassociated with Escherichia coli 0157:H7 infec-tions. Ann Intern Med 1988;109:705-12.

13. Riley LW. The epidemiologic, clinical, and micro-biologic features of hemorrhagic colitis. Ann RevMicrobiol 1987;41:383-407.

14. Karmali MA, Petric M, Lim C, et al. The associ-ation between idiopathic hemolytic uremic syn-drome and infection by verotoxin-producing£sc/i-erichia coli. J Infect Dis 1985;151:775-82.

15. Neill MA, Tarr PI, Clausen CR, et al. Escherichiacoli 0157:H7 as the predominant pathogen asso-ciated with the hemolytic uremic syndrome: aprospective study in the Pacific Northwest. Pedi-atrics 1987;80:37-40.

16. Laboratory Centres for Disease Control. Entero-hemorrhagic Escherichia coli and the hemolyticuremic syndrome—the Alberta experience. CanDis Wkly Rep 1989;15:9-12.

17. Centers for Disease Control. Thrombotic throm-bocytopenic purpura associated with Escherichiacoli 0157:H7—Washington. MMWR 1986;35:549-51.

18. Morgan GM, Newman C, Palmer SR, et al. Firstrecognized community outbreak of haemorrhagiccolitis due to verotoxin-producing Escherichia coli0167:H7 in the UK. Epidemiol Infect 1988;101:83-91.

19. March SB, Ratnam S. Sorbitol-MacConkey me-



Dow

nloaded from



dium for the detection of Escherichia coli 0157:H7associated with hemorrhagic colitis. J Clin Micro-biol 1986;23:869-72.

20. Farmer JJ III, Davis BR. H7 antiserum-sorbitolfermentation medium: a single tube screening me-dium for detecting Escherichia coli 0157:H7 asso-ciated with hemorrhagic colitis. J Clin Microbiol1985;22:620-5.

21. Ewing WH. Edwards and Ewing"s identificationof Enterobacteriaceae. 4th ed. New York: ElsevierScience Publishing Co., Inc., 1986.

22. Doyle MP, Schoeni JL. Isolation of Escherichiacoli 0157:H7 from retail fresh meats and poultry.Appl Environ Microbiol 1987;53:2394-6.

23. Thomas DG. Exact and asymptotic methods forthe combination of 2 x 2 tables. Comput BiomedRes 1975;8:423-46.

24. Laboratory Centres for Disease Control. An out-

break of Escherichia coli 0157:H7 diarrhea in anursing home—Alberta. Can Dis Wkly Rep 1987;13:206.

25. Martin ML, Shipman LD, Wells JG, et al. Isola-tion of Escherchia coli 0157:H7 from dairy cattleassociated with two cases of haemolytic uremicsyndrome. (Letter). Lancet 1986;2:1043.

26. Borczyk AA, Karmali MA, Lior H, et al. Bovinereservoir for verotosin-producing Escherichia coli0157:H7. (Letter). Lancet 1987;1:98.

27. Karmali MA. Infection by verocytotoxin-producing Escherichia coli. Clin Microbiol Rev1989;2:16-38.

28. Drummond KS. Hemolytic uremic syndrome—then and now. N Engl J Med 1985;312:116-18.

29. Cleary TG. Cytotoxin-producing Escherichia coliand the hemolytic uremic syndrome. Pediatr ClinN Am 1988;35:485-501.



Dow

nloaded from


Documents

A STATEWIDE OUTBREAK OF ESCHERICHIA COLI 0157:H7