ARCHIVED - SEVERE OUTBREAK OF ESCHERICHIA COLI O157:H7 IN HEALTH CARE INSTITUTIONS IN CHARLOTTETOWN, PRINCE EDWARD ISLAND, FALL, 2002

 

Introduction

Escherichia coli O157:H7 is a verotoxin-producing bacterium that causes an illness characterized by severe abdominal pain and diarrhea, which sometimes leads to hemorrhagic colitis(1). There is little or no fever(2). The incubation period for E. coli O157:H7 is from 2 to 8 days (median 3 to 4 days)(2). Excretion generally takes a week or less in adults but can take up to 3 weeks in one-third of children(1). There is no curative antibiotic treatment for E. coli O157:H7; only supportive treatment is available.

Livestock is the main reservoir. E. coli is transmitted primarily through the ingestion of contaminated food or water(3-5), direct contact with animals(1,4,5), or human-to-human contact(1,4). Outbreaks have been associated with various food sources: undercooked ground beef(1,2,4,5), salami(2,3,5), unpasteurized milk(2,3,5), lettuce(1-3), raw vegetables(6,7), unpasteurized cider(8), etc.

On 6 November, 2002, the public health authorities in Charlottetown, Prince Edward Island (PEI), were informed that a number of cases of diarrhea had occurred in a psychiatric hospital, affecting residents and staff. This article describes the ensuing epidemiologic investigation and its major conclusions.

Methods

The investigation was based on active case surveillance, an environmental assessment, microbiologic analyses, and a case-control study.

Active Case Surveillance

Case definitions for possible, probable and confirmed cases were developed to assist in case finding over the period from 16 October to 22 November, 2002, among individuals living in or working at the hospital and other Charlottetown health care facilities and the symptomatic contacts of the cases reported in these facilities:

  • Possible case: a person who had had at least two soft stools in 24 hours;
  • Probable case: a person who had had at least one loose, bloody stool without laboratory confirmation;
  • Confirmed case: a person who had provided a stool sample in which E. coli O157:H7 or verotoxin was identified.

The cases were documented using clinical records, interviews with health care personnel, administrative records and a questionnaire. Cases were recorded in list form to trace epidemic curves and calculate attack rates.

Environmental Assessment

The investigators visited food service areas and clinical departments, reviewed menus, recipes and preparation procedures, examined patient records, interviewed key individuals, and conducted epidemiologic case follow-up, primarily at the psychiatric hospital. Food preparation, storage, and distribution procedures at the hospital were assessed by PEI government inspectors.

Microbiologic Analyses

All residents and health care personnel at the hospital and the individuals who had had at least one loose stool were asked to provide a stool sample for E. coli O157:H7 identification. The microbiology laboratory at a different hospital carried out the testing. The positive isolates were characterized and subtyped by the Health Canada National Microbiology Laboratory in Winnipeg using phage typing and pulsed field gel electrophoresis (PFGE). The laboratory also tested for fecal verotoxins in the stools of individuals who were affected by hemorrhagic colitis but who tested negative for E. coli O157:H7.

Food and drinking water samples and smears from the hospital's kitchen surfaces were tested for microbiologic contaminants by the microbiology laboratory at the Prince Edward Island Food Technology Centre. The City of Charlottetown conducted microbiologic analyses on its drinking water distribution system.

Case-control Study

A case-control study was conducted among staff at the hospital to test hypotheses on the source of the outbreak. The selected cases included all the probable or confirmed cases of E. coli O157:H7. Three controls were matched randomly with each case on the basis of type of work and simultaneous presence at work for at least 1 day during the 8-day incubation period.

A questionnaire developed in the field was pretested with staff at the hospital, and experienced nurses with the PEI Department of Health and Social Services were trained to administer it. The questionnaire collected demographic information (e.g. sex, date of birth, residence), clinical information (e.g. symptoms, laboratory results, medical history), and exposure information (e.g. food preparation or consumption, family contacts, presence on a farm, drinking water consumption, travel).

The information was entered into Epidata 2.5 after initial processing in the field with EpiInfo 6.04. The purpose of the preliminary analyses was essentially to identify the source of the epidemic. The final statistical processing was done with SAS 8.0 (Cary, NC) to identify the associations between exposure and illness more formally, and included univariate and multivariate analyses. The exposure variables were processed individually and pooled to increase statistical power. Multivariate logistic regression analyses(9) were performed on variables significant at the 20% level in the univariate analysis. A final multivariate analysis was developed with the stepwise variable selection approach in stages following adjustment for age, at the 10% level.

Results

The hospital in question is a provincial, long-term psychiatric care centre with about 80 beds. At the time of the outbreak, 203 people were working there: 132 in patient care, 38 in the kitchen, and 53 in administration. The majority of residents moved around fairly freely. Their level of autonomy varied, but they could generally dress themselves and eat on their own. Most ate in the cafeteria.

The hospital kitchen was supplied by food wholesalers that primarily serve the Maritime provinces. Three meals were prepared daily according to a 3-week menu. Hot dishes, salads, sandwiches, and desserts were prepared. They were served in the cafeteria and in the patient care units or sold in the hospital vending machines, and some of the food was distributed to the five other health facilities in Charlottetown.

Cases Identified During the Investigation

During the investigation, 81 possible cases, 11 probable cases, and 17 confirmed cases of E. coli O157:H7 were identified among the staff and patients of Charlottetown health care facilities (Table 1). The overall attack rate based on possible, probable, and confirmed cases of E. coli O157:H7 was 18.7% (38/203) among employees of the hospital, including 22% (29/132) among health care personnel and 24% (9/38) among kitchen employees, and 32.5% (26/80) among residents. The attack rates, based on probable and confirmed cases at the hospital, were 6.4% (13/203) among staff and 8.8% (7/80) among residents.

Table 1. Distribution of possible, probable and confirmed cases of E. coli O157:H7 among staff and residents of health care facilities in Charlottetown, PEI, from 16 October to 22 November, 2002

Case

Psychiatric Hospital

Other facilities

Staff (n = 203)

Residents (n = 80)

Staff

Residents

Possible

25

19

29

8

Probable

6

1

4

0

Confirmed

7

6

3

1

Total

38

26

36

9

Attack rate

18.7%

32.5%

N/A

N/A


The outbreak began among staff and residents of the psychiatric hospital. The first case there occurred on 24 October (Figure 1) and involved the person in charge of vegetable preparation in the kitchen. This person had presented with an episode of diarrhea, worsening between 24 and 26 October, but had continued to work throughout the episode. Previously, the person had had two 10-day courses of antibiotics.

Figure 1. Date of onset of symptoms in reported possible, probable and confirmed cases of E. coli 0157:H7, Charlottetown, PEI, October-November, 2002

Figure 1. Date of onset of symptoms in reported possible, probable and confirmed cases of E. coli 0157:H7, Charlottetown, PEI, October-November, 2002

The person responsible for vegetable preparation had almost exclusive use of these work facilities. The vegetables (lettuce, broccoli, celery, onions, peppers, tomatoes) were rinsed in a sink, cut up on a wooden board, and placed in a bowl before being used in salads, sandwiches, and other menu items.

Clinical Presentation

Of the 109 identified cases of E. coli O157:H7, 20 cases of bloody diarrhea were reported: 10 employees and five patients of the hospital and five persons in other health care institutions. Two of the 10 employees who reported blood in their stools were hospitalized, and they subsequently recovered. No cases of hemolytic uremic syndrome were identified. Two cases resulted in death. According to the initial coroner's report, E. coli O157:H7 was responsible for one death and was a contributory factor in the other.

Microbiologic Analyses

A total of 486 stool specimens were taken from 269 individuals for E. coli O157:H7 testing. All the positive results were of phage type 32 with the same PFGE profile (0756). Two cases of hemorrhagic colitis that were negative for E. coli O157:H7 showed a verotoxin.

Environmental Assessment

From 7 to 13 November, 2002, 43 food samples and 49 swabs were taken at the hospital. All were negative for E. coli O157:H7. The inspectors found that the equipment was very clean and that the kitchen services were very well organized at this hospital. However, they noted a few irregularities or problems that might have led to cross-contamination or allowed microbes to grow. For example, not all the knives were washed, rinsed, and disinfected between use, and the vegetables used for the salads and sandwiches were sometimes soaked in a sink filled with water instead of being rinsed.

Case-control Study

Initially, 12 cases and 36 control subjects, matched by job, were chosen from hospital personnel to take part in the study. One case and its three controls were rejected. One control proved to be a probable case, and two new controls were matched to it. One of the controls could not be reached within the time limit and four did not meet all the selection criteria. Ultimately, 12 cases and 30 controls took part in the study.

Seven of the 12 cases worked with patients, and five worked in the kitchen (Table 2). Of the controls, 20 worked in patient care and 10 in the kitchen. The average age was 48.2 years for the cases and 45.2 years for the controls; 66.7% of cases and 73.3% of controls were female. Four cases were already taking antibiotics when the symptoms appeared, and three had recently been taking them. Fifty percent (6/12) of the cases presented with an illness or with chronic health problems, compared with 26.7% (8/30) of the controls. For 10 of the cases, gastrointestinal symptoms had required sick leave ranging from 1 to 12 days (median 3 days). Six of the cases had seen a doctor, and two were hospitalized for 2 and 6 days respectively.

Table 2. Demographic description of subjects in the case-control study, Charlottetown, PEI, October, 2002

Demographic variable Cases Control subjects Total
Sex
  Male 4 8 12
Female 8 22 30
Job
  Kitchen 5 10 15
Patient care 7 20 27
Age (number of cases)
  Average 48.2 45.2 46.1
Median 50 44 46
Minimum 40 29 29
Maximum 56 62 62

Among the cases, 75% had eaten or prepared food sold in the hospital's vending machine, compared with 30% of the controls. Hospital staff members were 6.4 times more likely to develop the disease if they had eaten or prepared foods sold in the vending machine (95% confidence interval [CI]: 1.3-31.4). Taken separately, no salad, sandwich, or other food sold in the machine was significantly associated with the disease (Table 3). However, by combining salads and sandwiches containing lettuce or vegetables handled in the kitchen by the designated individual, it is possible to form combined variables associated with the disease. The cases were 10.4 times more likely than the controls to have eaten or prepared the salads (95% CI: 1.2-91.3), and the risk of disease was 5.8 times greater for persons who had prepared or eaten the sandwiches (95% CI: 1.2-29.4).

Table 3. Univariate analysis of food exposure among cases and controls, psychiatric hospital , October to November, 2002

Exposure variable

Cases
(= 12)

Controls
(n = 30)

Matched
odds ratio

95% confidence interval

p value

Vending machine

9

9

6.43

1.31-31.41

0.022

Cafeteria

12

20

   

0.996

Mixed salads*

6

5

10.37

1.18-91.26

0.035

Combination sandwiches**

6

4

5.81

1.15-29.37

0.033

Julienne salad

3

1

   

0.996

Caesar salad

4

2

   

0.997

Green salad

3

1

9.00

0.94-86.52

0.057

Fajita salad

1

1

3.00

0.19-47.96

0.437

Chicken loaf

1

1

2.00

0.13-31.97

0.624

Ham and cheese sandwich

2

2

2.71

0.38-19.40

0.320

Ham and lettuce sandwich

3

0

   

0.996

Bacon, lettuce and tomato sandwich (BLT)

1

1

3.00

0.19-47.96

0.437

Undercooked ground beef

1

0

   

0.996

*Salads containing lettuce and vegetables (Julienne, caesar, green, fajita)

**Sandwiches containing lettuce and vegetables (submarines; beef; eggs; tuna; chicken and lettuce; chicken loaf; cheese, lettuce and mayonnaise; sliced turkey; ham and cheese; ham, lettuce and tomato; BLT)


In the final multivariate logistic regression model, only the consumption and preparation of salads and sandwiches were associated with the risk of contracting the disease. The risk of disease was 12.8 times higher for staff who had eaten or prepared salads (95% CI: 0.9-189.3; p = 0.06) and 9 times higher for staff who had eaten or prepared sandwiches (95% CI: 0.1-91.0; p = 0.06).

Discussion

This outbreak officially ended on 22 November, 2002. Elsewhere in Canada, an outbreak of E. coli O157 was reported in a psychiatric health care facility in Hamilton in 1990(10), but, in contrast to this episode, members of the public were not affected.

Of the 109 reported possible, probable, and confirmed cases, 20 (18.3%) presented with hemorrhagic colitis, four (3.7%) were hospitalized, and two (1.8%) died (these were residents). This is a relatively small percentage of complications. One Canadian study reported an 8% risk of hemolytic uremic syndrome in infected children during an outbreak(11). Carter et al. reported hemorrhagic colitis in up to 75% of cases, hemolytic uremic syndrome in up to 22% of cases, and death in up to 15% of cases in outbreaks in a health care facility for the elderly(12). The risk factors for development of complications are age (either old or young), being female, diarrheal stools, antibiotic treatment, and use of diarrhea medication(13). In Charlottetown, the cases were adults in generally good health who did not have any particular risk factors for complications.

At 15.6%, the percentage of confirmed cases of E. coli O157:H7 among of all the symptomatic cases was low. Stool samples were taken at an advanced stage of the disease when E. coli O157 H:7 was no longer being excreted, and this would have contributed to the low figure. The fact that the hospital had no surveillance system for infectious diseases delayed case identification and confirmation. As well, some cases could have been erroneously considered possible cases in the study, whereas they were actually related to a gastroenteritis outbreak that had started in a local long-term care facility(14).

The epidemic curve suggests a common contamination curve in the initial phase followed by secondary human-to-human transmission. The case-control study may have included secondary cases. It may have been affected by a selection bias and, because it was a retrospective study, by a recall bias because of the delayed investigation. As well, the size of the confidence intervals for the odds ratios demonstrates the inaccuracy of the results. Combining the consumption and preparation of the foods at risk yields a higher statistical power but represents a potential differential information bias. Food preparers can be the cause of the disease and not be affected by it. As well, the inclusion of persons with chronic diarrhea may represent a potential selection bias.

Outbreaks of E. coli O157:H7 have been documented for over 20 years, and most have been blamed on hamburgers or ground meat(13). Items that are served raw or require no cooking before serving, such as salami, lettuce and cheese, have also been implicated in outbreaks(13) and are possible vehicles of infection. Environmental analyses based on samples taken a few days after the occurrence of the first cluster of cases did not identify any items contaminated by E. coli O157:H7. The fact that foods prepared during the incubation period of these first cases were not available limited the scope of the analysis.

The most likely source of the outbreak was contamination of the salads and sandwiches prepared in the hospital kitchen. The person who prepared the vegetables used in the salads and sandwiches likely contaminated food items in the course of working while symptomatic. Indeed, most of the cases (87.5%) occurred within the E. coli O157:H7 excretion period of this same case. It is not impossible that items contaminated before their arrival at the kitchen were used in the preparation of salads and sandwiches, but the fact that there were no other cases with the same PFGE profile (0756) among the clients of the hospital's food wholesaler makes this hypothesis unlikely. Until this outbreak, profile 0756 had never been identified in Canada before (Dr. Rafiq Amand, National Microbiology Laboratory: personal communication 2002). Livestock from local farms cannot be ruled out as the reservoir.

This outbreak points to the importance of applying personal and food preparation hygiene measures. Government inspection services found no significant problems in the kitchen of this hospital. However, adopting the Hazard Analysis and Critical Control Point System would reduce the number of irregularities that might lead to contamination. The E. coli O157:H7 outbreak affected a large number of people (about 100) despite the diligent application of effective control measures by public health authorities. Active surveillance can identify outbreaks quickly and reduce their impact(15,16).

Acknowledgements

The authors cordially thank hospital staff, PEI and Health Canada laboratory staff, and the PEI health workers and authorities who participated in this epidemiologic investigation.

References

  1. Chin JE. Control of communicable diseases manual, 17th ed. Washington DC: American Public Health Association, 2000.

  2. U.S. Food & Drug Administration - Center for Food Safety & Applied Nutrition. Foodborne pathogenic microorganisms and natural toxins handbook. 25 June, 2002.

  3. European Programme for Intervention Epidemiology Training. An international outbreak of vero cytotoxin-producing Escherichia coli O157 infection amongst tourists; a challenge for the European infectious disease surveillance network. Epidemiol Infect 1999;123(2):217-23.

  4. O'Brien SJ, Adak GK. Escherichia coli O157:H7 - piecing together the jigsaw puzzle. N Engl J Med 2002;347(8):608-9.

  5. Health Canada. Escherichia coli O157:H7 infections associated with ground beef and their control in Canada. CCDR 2000;26(13):111-16.

  6. Hideshi M, Kazuhiro A, Shunsaku M et al. Massive outbreak of Escherichia coli O157:H7 infection in schoolchildren in Sakai City, Japan, associated with consumption of white radish sprouts. Am J Epidemiol 1999;150(8):787-96.

  7. Cieslak PR, Barrett TJ, Griffin PM et al. Escherichia coli O157:H7 infection from a manured garden. Lancet 1993;342(8867):342-67.

  8. Health Canada. An outbreak of Escherichia coli O157:H7 infection associated with non-commercial, custom-pressed apple cider-Ontario, 1998. CCDR 1999;25(13):113-17.

  9. Stokes ME, Davis CS, Koch GG. Categorical data analysis using the SAS system, 2nd ed. Cary, NC: SAS Institute Inc., 2000.

  10. Kohili HS, Chaudhruri AK, Todd WT et al. A severe outbreak of E. coli O157 in two psychogeriatric wards. J Public Health Med 1994;16(1):11-15.

  11. Rowe PC, Orrbine E, Lior H et al. Risk of hemolytic uremic syndrome after sporadic Escherichia coli O157:H7 infection: results of a Canadian collaborative study. J Pediatr 1998;132:777-82.

  12. Carter AO, Borczyk AA, Carlson AK et al. A severe outbreak of Escherichia coli O157:H7- associated hemorrhagic colitis in a nursing home. N Engl J Med 1987;317(24):1496-1500.

  13. Todd ECD. Escherichia coli O157:H7 infections associated with ground beef and their control in Canada. CCDR 2000;26(13):111-16.

  14. The Guardian. Virus closes doors of Prince Edward Home. Charlottetown, PEI, 28 November, 2001:A1.

  15. Roberts CL, Mshar PA, Cartter ML et al. The role of heightened surveillance in an outbreak of Escherichia coli O157:H7. Epidemiol Infect 1995;115 (3):447-54.

  16. Shefer AM, Koo D, Werner SB et al. A cluster of Escherichia coli O157:H7 infections with the haemolytic-uremic syndrome and death in California. A mandate for improved surveillance. West J Medicine 1996;165(1-2):15-9.

Source: D Bolduc, LF Srour, MD, Canadian Field Epidemiology Program, Health Canada; L Sweet, MD, A Neatby, RN, PEI Department of Health and Social Services; E Galanis, MD, Canadian Field Epidemiology Program; S Isaacs, G Lim, Foodborne, Waterborne and Zoonotic Infections Division, Health Canada.

Page details

Date modified: