This conclusion is based on three key observations of a study of food histories of staff who ate from the hospital kitchen during the period from October 28 to November 3, 2002, and a case-control analysis.

While the analysis did not produce a direct correlation between the bacteria and one salad or sandwich, it does point conclusively to any one of four salads or four sandwiches. None of the foods considered to be the source of the bacteria were available for testing.

Analysis of the food histories indicates that staff who consumed or prepared salads were 5.6 times more likely to become ill from E. coli 0157:H7 than those who did not. Staff who consumed or prepared salads or sandwiches from the vending machines were 7.3 times more likely to become ill. Staff who consumed or prepared sandwiches which contained food items that were also in the salads were 22.9 times more likely to become ill. "All of these outcomes are highly significant. In other words, it is highly likely that these results are true," said Dr. Sweet.

As of November 25th, six hospital employees and four patients have been confirmed as having E. coli 0157:H7. Also diagnosed were one household member of a hospital employee and one resident of Talbot House which receives food from Hillsborough Hospital.

The organism which caused the outbreak was diagnosed at the Health Canada laboratory in Winnipeg as phage type 32, which is unusual in Canada, and different from other types of E. coli confirmed in Prince Edward Island in recent months. "This indicates that the organism involved in the outbreak at Hillsborough Hospital was confined to those who consumed food from the hospital kitchen and is not circulating in the community," said Dr. Sweet.

Cecil Villard, Director of Acute and Mental Health Services with Queens Region Health, advised that although no major deficits were found in the food handling or preparation practices at the hospital, these processes continue to be carefully monitored. Additional infection control measures also continue to be implemented for all patients and staff who tested positive. "This has been a very trying time for patients, families and staff, and we are most appreciative of the tremendous cooperation and support we received from everyone until the source of the bacteria could be isolated," he said.

Testing of original samples from all patients, and those staff who reported symptoms, will continue until their test results are clear. "Because the bacteria isn't always present throughout the sample, it is not unusual to obtain a positive result following one or two negative ones. For this reason, it is quite possible that we may still confirm positive results, although people may not have, or have had symptoms," said Dr. Sweet.

Several recommendations are outlined in Dr. Sweet's report of the investigation to the Minister of Health and Social Services and Queens Region Health. It is recommended that patients who tested positive remain in a separate unit of the hospital until two negative samples taken two weeks apart are obtained; that kitchen staff follow safe food handling and preparation principles and attend regular education sessions on food handling, and that access to the kitchen area be restricted to food handlers and kitchen staff; that kitchen staff remain home when they are ill with appropriate sick leave; that for the next six-week period, samples of all food served in the kitchen be kept refrigerated for 16 days and available for testing; and that additional infection control nursing staff be added to the hospital.

Media Backgrounder E. coli Results

Initial activities

• The presence of the E. coli bacteria was first identified at the Hillsborough Hospital November 7, 2002 when Chief Health Officer Dr. Lamont Sweet received reports from the QEH laboratory of three reports of E. coli. An investigation began immediately.

Public health nurses began patient interviews and case histories to determine possible notable common factors in their histories.

• Infection Control and Occupational Health staff determined that there was a number of ill patients and staff at the hospital, and information was quickly obtained on their symptoms.

• Following contact with the City of Charlottetown and the Department of Environment, it was determined that the water had been tested regularly in the previous weeks and no significant levels of bacteria had been found in the water.

• Contacts were made to all facilities where Hillsborough Hospital provides food including Sherwood Home, Mount Herbert Addictions Facility, Talbot House, Deacon House, Murchison House and Meals-On-Wheels.

• Hillsborough Hospital staff identified 6 patients and 14 staff who had been ill with symptoms over the past 7-10 days. It appeared that most became ill before November 5 and that an outbreak of E. coli in both staff and patients was extremely likely. Several control measures were implemented immediately.

• Cultures were obtained from all patients for testing.

• All patients with symptoms were isolated.

• Staff were asked to conduct careful hand washing and infection control techniques.

• Staff with symptoms were asked to stay at home until their test results were clear.

• Surveillance systems were established to monitor symptoms and record test results.

• Field epidemiologists Dr. Leila Srour and Daniel Bolduc of Health Canada arrived November 8 to assist with the investigation.

Inspection of the Kitchen

• The kitchen facility was inspected by Environmental Health Officers who investigated all hot and cold temperature devices including ovens and refrigeration units, food handling and preparation practices. No problems were found.

• High risk foods for E. coli were held and tested. A total of 43 food samples were tested from November 7 to 13 and all were negative for E. coli 0157:H7.

• A total of 49 environmental swabs taken from the kitchen area were all negative.

• The overall standards of food preparation and handling and the operation of kitchen equipment were determined to be very good. The kitchen remained open with strict supervision.

• By the time the outbreak was detected, any food which may have been contaminated was gone and not available for testing.

Epidemiological Investigation

• A total of 101 persons reported having diarrhea as of November 21, 2002.

• All confirmed cases reported having the onset of symptoms between October 29 and November 3, 2002.

• All persons who reported the onset of symptoms after November 4, except one, did not have positive cultures.

• A food exposure case-control study was conducted by Daniel Boulac, field epidemiologist with Health Canada.

• Public health nurses and a nutritionist conducted interviews with staff November 14 to 16, 2002. Patients could not provide useful food histories.

• Analysis of the food histories indicates that staff who consumed or prepared salads were 5.6 times more likely to become ill from E. coli 0157:H7 than those who did not. Staff who consumed or prepared salads or sandwiches from the vending machines were 7.3 times more likely to become ill. Staff who consumed or prepared sandwiches which contained food items that were also in the salads were 22.9 times more likely to become ill.

• All these outcomes are highly significant. It is highly likely that these results are true.

• Control measures to limit the spread of the bacteria appear to have been successful, with only one new case occurring in the hospital after November 3, 2002.

Results of the investigation

• The organism diagnosed at the Health Canada laboratory showed that the E. coli bacteria is phage type 32, which is unusual in Canada, and different from others confirmed recently in PEI. It appears to have been contained in the hospital.

• Based on statistical evidence, it has been determined that the source of the bacteria was in ingredients used in four types of salads and sandwiches.

• Food items used in these salads and sandwiches include: purple cabbage, purple onion, Romaine lettuce, iceberg lettuce, grated cheese, broccoli, cauliflower, celery, green pepper, tomatoes, bacon bits, croutons, ham, beef, commercial dressing, mayonnaise, egg, tuna and chicken.

• Salads were prepared in a separate room. Vegetables were rinsed and gloves were worn to add the vegetables to the bowl during preparation. Some of the food items not used in the salads were then used to make sandwiches.

• Contamination could have occurred due to contamination from the hands of a food handler. It is also possible that one of the food items was contaminated. The exact source and circumstances of the contamination cannot be proven for certain.