The isolation rate of Escherichia coli Ol 57 : H 7 in Toronto and surrounding communities

Verocytotoxin-producing strains of Escherichia coli, most often serotype 0157:H7, have been associated with both sporadic and epidemic diarrheal disease in Canada. In order to determine the isolation rate of E coli 0157:H7 in outpatients with diarrhea, all stool specimens submitted for culture to Med-Chem Laboratories in Metropolitan Toronto between June 1988 and September 1989 were cultured on MacConkey-Sorbitol agar in addition to standard enteric media. A total of 46 (0.3%) of 16,125 stool specimens yielded E coli 0157:H7 or verotoxin-producing E coli 0157:H(-). These isolates came from 31 patients with diarrhea; only 16 (52%) had a history of hemorrhagic colitis and one patient developed hemolytic uremic syndrome. Although MacConkey-Sorbitol agar was useful as a differential medium for detecting E coli 0157:H7, 14.5% of all specimens yielded nonsorbitol-fermenting isolates. It is not certain whether the routine use of MacConkey-Sorbitol agar is justified when isolation rates of E coli 0157:117 are very low.

exponential increase in the number of isolations of E coli 0157:H7 reported in Canada between 1982 and 1987 (11).This organism accounted for 3.6% of a ll human enteric pathogens referred or reported to the Enteric Bacteriology Division of the L:'lboratory Centre for Disease Control in Ottawa in 1986.
As with other enteric pathogens.the isolation rate of E coli 0 157:H7 may vary geographica lly.Therefore.the purpose of this study was to determine the prevalence of E coli 0 157:H7 as a cause of diarThcal illness in an outpatient-based population submitting stool specimens for cu lture to a diagnostic microbiology laboratory in Metropolitan Toronto and surrounding communities.The study also provided an opportunity to describe the spectrum of illness associated with this infection in symptomatic subjects and to assess the usefulness of routinely screening stool cu ltures with sorbitol-containing MacConkcy agar.

PATIENTS AND METHODS
Med -Chem Laboratories is a privately owned.licensed laboratory offering diagnostic biochemistry.hematology.pathology and microbiology services to family physicians.primary care medical clinics and nursing homes in Metropolitan Toronto and the surrounding communities within a 40 mile radius.Between June J.1988 and September 30.1989 a ll stool specimens submitted to Med-Chem Laboratories for culture were screen ed on MacConkey-Sorbitol agar containing 1% o-sorbitol (PML Microbiologicals.Mississauga, Ontario).incubated at 35°C for 18 to 24 h.Five nonsorbitol-fermenting colonies were picked from each plate for serogroupin g by s lide agglutination with 0157 antiserum (Difco Laboratories.Detroit.Michigan) (14).Positive isolates were identified and antimicrobia l susceptibility tes ting done us ing a commercia lly available microdilulion system.MicroScan (Travenol Laboratories.Mahwah.New Jersey).All E coli 0157 isolates were sent to the Central Public Health Laboratory.Ontario Ministry of Heallh for 0 and H serotype confirmation by tube agglutination and determination of verocytotoxin production.Standard enteric media were used for the isolation of Salmonella.Shigella.Yersinia and Campylobacter species.Examination for ova and parasites or Clostridium dif[tcile culture and cytotoxin assay were done only on request.
The family physicians of patients from whom E coli 0 J 57: H7 or nonmolile.verotox1n -producing E coli 0157 was isolated were contacted by telephone within one month for clinical a nd demographic data.

RESULTS
During the 16 month survey, verotoxin-producing E coli 0157:H7 orE coli 0157:1-I-was isolated from 46 (0.3%: confiden ce interval 0.008%) of 16,125 stool specimens.As shown in Table 1.these were the least frequently isolated bacterial enteric pathogens.For the last two months of the survey.the average number of stool specimens per patient was 1.39; this is probably a reasonable estimate for the entire study period.The E coli 0157 isolates came from 3 1 patients (18 ma les and 13 females) who ranged in age from six months to 70 years.Eighteen patients (58%) were less than lO years of age.None of the cases appeared to be epidemiologically related .The majority (87%) of infections occurred in the s ummer months (Figu re 1); none was acquired during foreign travel.One palienl was coinfected with Salmonella enteritidis.Of a ll patients from whom verotoxin-producing E coli was isolated.74% submitted stool specimens within four days of the onset of symptoms.and 84% s ubmitted within five days.7 ( 23) 16) All patients with E coli 0157 had ,generally watery diarrhea.but only 16 (52%) had c linica l manifestations typical o f hemorrhagic colitisbloody diarrhea.abdom inal pain and litlle or no fever (Table 2) .The mean duration of symptoms was eight days (range one to 21).Nine patients were treated empirically with a ntimi crobial agents (erythromycin.metronidazole.cipron oxacin ot• trimethoprim-sulfamethoxazole).1\vo patients were hospitalized: a two-year-old child with hemolytic uremic syndr ome.and a 25-year-old woman who underwent a laparotomy for presumed ap pendicitis but was found to have severe colitis.
Eighty-four grossly b loody stool specimens were received by the laboratory during the 16 month study.Only two (2.4%) of them yielded E coli 0 l57:H7 despite a his tory of bloody diarrhea from 16 patients with this infection.In contrast.grossly bloody stool specimens yielded Campylobacter jejuni from seven patients.Salmonella species from s ix. and Shigella species from three.

DISCUSSION
The association of E coli 0157:117 with the syndrome of h emorrh agic colitis was first recognized in 1983 (12.15).In the same year.Karmali et a l (3) re ported that these organisms and oth er verocytotoxin -producing strains of E coli were associated with cases of hemolytic uremic syndrome.Since then, it has become apparent that verocytoloxin-producing E coli.most com-monly E coli 0157:1 17. a re associated with a spectrum of e nteric illness.including both b loody a nd n on bloody diarrh ea (6.7.16).Recommendations for screen ing of fecal specimens with a Mac-Con key-Sorbitol agar in order to facilitate detection of sorbitol-negative E coli 0 157:H7 have been ma de (1 3. 17 .18).Whether routine screening of a ll s pecimens is warranted has not been establis hed.In 1989.796 isola tes of verotoxin-producing E coli were referred to lhe Enteric Referen ce Laboratory of the Centra l Public Health Laboratory.Ontario Ministry of Health (personal communication) .All but a few of these isolates were E coli 0 157:H7.and they were the third most common bacterial enteric pathogen referred to the laboratmy.after campylobacter (5998 isola tes) a nd salmonella (4868 isolates).However.few studies have determ ined the isola tion rate of E coli 0 157:H7 in stool specimens s ubmitted to diagnostic laboratories.The recovery of verocytotoxinproducing E coli 0157 from only 0.3% of 16.125 specimen s processed by Med-Chem Laboratories in Toronto is comparable to the isolation ra te of 0. 7% reported by regional public h ealth laboratories in Timmins and Peterborough.Ontario (1 9).In the United States.MacDonald et a l ( 16) recovered E coli 0 157: H7 from only 25 (0.4%) of 6485 diarrheal stool s pecimens obtained from members of a h eal th maintenance organization in the state of Washington.The organism was isolated less frequ ently than cam pylobacter or salmonella a nd abou t as often as shigella.E coli 0157:1 17 was recovered from only two (0.08%) of 2552 stools s ubmitted to a hospital clinical laboratory in Chicago ( 17).In contrast.E coli 0 157:H7 was r ecovered from 2.5% of 54 14 pa ti ent s with diarrhea seen at three Calgary hospitals in two years and was.after salmonella.U1e second most common bacterial enteric pathogen identified (7).In British Columbia .E coli 0 157: H7 was second only to campy lobacter as a bacterial cau se of infectious d iarrhea in a tertiary care pediatric hospital (20): it was isolated from 1.9% of 1425 s tools processed over 14 months.The rela tively low isola tion rate of E coli 0157:H7 found in the Metropolitan Tor onto region may be attributable to geograp hic vaiiabilily in prevalence of the orga ni sm .Alternatively. the low ra te may renect culture results from a patient population with a milder spectrum of diarrheal illness.The patients studied were generally outpatients wh o sought medical attention and had s tool cultures requested by the ir physicians.However.these patients did not come to hospitaJ emergency de partments because of their symptoms and were.perh aps.less like ly to have had grossly b loody diarrhea or hemorrh agic colitis.
The results of this study provide further eviden ce that there is variation in the proportion of E coli 0157:117 infections causing nonbloody diarrhea (6).ln a survey of 92 patients with enterocolitis in Newfoundland.Ecoli0157:H7 was recovered from seven of 47 patients (15%) \vith grossly bloody diarrhea but from none of 45 patients with nonbloody dian•hea (8).In a prospective study condu cted at three Calgary hospitals.almost all patients withE coli 0157:H7 infection had bloody diarrhea.and the organism was isolated from 40% of the 137 patients who presented \vith bloody diarrhea (7).In the present study.nearly half (48%) of patients with verotoxinproducing E coli infection did not give a history of bloody diarrhea.and only two provided stool specimens that were grossly bloody at the time of culture.Moreover.both salmonella and campylobacter were recovered more frequently than E coli 0157:H7 from patients whose stools were grossly bloody on receipt in the laboratory.
The present finding that 14.5% of specimens yielded non-sorbitol-fern1enting organisms on MacConkey-Sorbitol agar is similar to the 15% reported by March and Ratnam (13) and the 16% found by Walker et al (18).Organisms that grow on this medium and that may not ferment sorbitol include non-0157 E coli, Proteus species.Morganel• Ia species.and occasionally oth er coliforms or Pseudomonas species (13).Although MacConkey-Sorbitol agar is useful in identifying the majority of stool specimens that are negative for E coli 0 157:H7.a more selcclivc screening test or medium is desirable.Other potentially useful markers for E coli 0157:H7 include lysine and ornithine decarboxylase reactions (21) and raffinose fermentation (22).However. it is important to note that these tests.as well as the use of MacConkey-Sorbitol agar.fail to detect non-0157 serotypes of verocytotoxin-producing E coli.The detection ofvcrotoxin in fecal filtra tes remains the most sensitive and specific method of establishing the diagnosis (23).but is slow and labour intens ive.and requires tissue culture facilities.A simple.rapid and accurate assay for the de tection of fecal verotoxin would therefore be desirable.
The present data indicate that many patients \vilh this infection in the community do not have typical symptoms of hemorrhagic colitis, and that the isolation rate of E coli 0157:H7 in southern Ontario is much lower than that reported elsewhere in Canada (7.8,20).These results also conrrrm that sorbitol-containing MacConkey agar is useful for screening stool specimens for E coli 0157:H7.although even with this medium there was considerable work and expense incurred in identifying nonsorbitol-fermenting isolates.It has been estimated that even with a prevalence of E coli 0157:H7 in diarrheal s tools of 0.9%.the cost of identifying one patient with this infection would be approXimately US$465.00(24).
The authors' estimate of $350.00 per isolate substantiates the high cost of identifying infected patients.Whether routine use ofMacConkey-Sorbitol agar for processing stool specimens is justified when isolation rates of E coli 0157:H7 a re well under 1% is nol certain.Further studies are required to determine whether establishing a specific etiologic diagnosis or early Intervention would ameliorate the risk of subsequ ent com plicalions such as hemolytic uremic syndrome.However.il seems reasonable to use this medium at least for the investigation of outbreaks of diarrheal illness.for culturing bloody stools, and for specimens obtained in summer months or from children.

TABLE 1
Enteric p a thogens isolated from 16,125 feca l spec imens p rocessed b etween June 1988 and September 1989