An outbreak of Norwalk-like virus gastroenteritis in a nursing home in Rotterdam

O. Ronveaux, D. Vos, A. Bosman, K. Brandwijk, J. Vinjé, M. Koopmans, R. Reintjes 1 Départment de l’Epidémiologie des Maladies Infectieuses, RIVM, Pays-Bas 2 Participant au Programme Européen de Formation à l’Epidémiologie d’Intervention (EPIET*) 3 Service Municipal de Santé, Rotterdam, Pays-Bas 4 Hôpital de long et moyen séjour de Rustenburg, Rotterdam, Pays-Bas 5 Laboratoire de recherche des Maladies Infectieuses, RIVM, Pays-Bas


Sixty-two per cent of elderly and disabled residents of a Dutch nursing home (74/120) and 32% of staff (33/102) became ill in an outbreak of Norwalk-like viral gastroenteritis. The outbreak spread from person to person was supported by temporal clustering of cases. Being fed by members of staff was associated with higher attack rates.
N orwalk-like viruses (NLVs), previously known as small round structured viruses, are pathogens commonly recovered in outbreaks of gastroenteritis.They were identified in 43% of outbreaks of infectious intestinal diseases reported in England and Wales in 1995 and 1996 (1).In the Netherlands in 1996, NLVs caused 83% of reported outbreaks of gastroenteritis, 59% of which occurred in nursing homes (2).Five per cent of cases of gastroenteritis for which Dutch general practitioners are consulted have been attributed to NLVs (3), indicating circulation in the community.NLVs are transmitted by the faecal oral route, and the virus can spread between people by direct contact or through contaminated food or water.Forty per cent of NLV outbreaks are believed to be transmitted by food (4,5); food may be contaminated from the outset (usually molluscan shellfish) or be contaminated by infected foodhandlers.
The physican responsible for a nursing home in the Rotterdam region notified an acute outbreak of gastroenteritis to the municipal health service (the Gemeeschappelijke GezondheidsDienst, GGD) on 13 January 1999.The nursing home had 150 beds on four floors for elderly and disabled people.New cases were still arising so the GDD visited to assess, advise, and investigate.We describe here the outbreak and identify transmission routes and risk factors associated with NLV infection.

Case definition
A case of gastroenteritis was defined as a person living or working in the nursing home with acute onset of diarrhoea (defined as three or more loose stools / 24 hours) or vomiting between 27 December 1998 and 25 January 1999.

Epidemiological investigation
A retrospective cohort study was undertaken among staff and residents of the nursing home.A written questionnaire was used to investigate risk factors related to place of eating and, for staff, activities performed.Questionnaires were completed by the nursing home staff who also helped most residents in their completion.

Laboratory investigation
Faecal specimens were examined for Shigella, Campylobacter, Salmonella, and Yersinia in Rotterdam public health laboratory.A subset were assayed at the RIVM reference laboratory for the presence of Norwalk-like viruses by reverse-transcriptase polymerase chain reaction (PCR), and for group A rotavirus, adenovirus 40/41, and astroviruses by enzyme linked immunosorbent assay (ELISA).Contaminated food or water was not suspected during the outbreak, therefore no environmental sampling was performed.

Statistical analysis
Data were entered and analysed using Epi Info (Atlanta, GA).The possibility of person to person transmission between roommates was investigated by examining for temporal clustering of cases: illness was examined as a function of exposure to a roommate who had become ill one or two days earlier (6,7).The rate of illness among residents who became ill in the two days after the sick roommate became ill was compared with the rate of illness among residents who became ill at any other time.Denominators were the total number of days contributed by residents to each of the exposed and non-exposed groups, respectively.
Relative risks of becoming ill in association with particular risk factors were calculated, with 95% confidence intervals.Multivariable analysis (logistic regression) was performed with the Egret package (version 0.26.6,Seattle, WA) proving odds ratios (OR) of becoming ill.

Setting
One hundred and twenty-four residents lived in the home in January 1999, and 138 staff members worked there in shifts.One hundred and twenty residents who stayed throughout the outbreak period and 102 staff members completed the questionnaires (response rates 97% and 74%, respectively).All floors had the same number of single (4 per floor) and shared rooms (4 with 2 beds, 2 with 3 beds and 5 with 4 beds), and occupancy levels were similar between floors.Staff were usually assigned to one floor but some of them worked on all floors.
Residents on the first floor were mentally disabled and were more likely to be bedridden.Residents of the three other floors were all of equal dependency, and could go up and downstairs.The average age of the residents (75 years, standard deviation 14) was similar on all floors (range 73 years on floor 1, to 79 on floor 4).Meals were prepared in the nursing home kitchen and eaten in two dining rooms (one large and one small) or in the resident's rooms.Nine per cent of the residents had taken a meal outside the nursing home in the week before the symptoms began.Among staff members who gave information on eating places, 68% (59/88) had eaten (in the past four weeks) in the staff restaurant.

Outbreak description
A total of 107 cases were reported (attack rate (AR) 48%).The epidemic curve (figure) shows a slow start and a protracted course.The first people to meet the case definition were a resident of the fourth floor on 27 December and a nurse on the second floor on 28 December.The small wave observed from 3 to 6 January was due mainly to cases from the third floor, and the main wave (10-15 January) was due to cases on the first and fourth floors.

Residents
The attack rate among residents was 62% (74/120).Cases were older than non-cases (77 years versus 72, p=0.04).The attack rates differed widely between floors (table 1).The attack rates were 53% among residents of single rooms, 57% in rooms with two beds, 54% in rooms with three beds, and 69% in rooms with four beds (p=0.15).Eating places were not associated with significantly different risks of becoming ill; the lack of association remained after adjusting for the floor.
Among the residents who shared rooms, 33% (21 out of 62 with known dates of onset of illness) became ill one or two days after the illness of a roommate.The relative risk of becoming ill one or two days after a roommate became ill rather than becoming ill at other time was 5.02 [3.0-8.4](20 cases/171 susceptible days one or two days after a roommate became ill compared with 42 cases/1804 other susceptible days).
The single variable analysis suggested that activities that required the closest contact with residents were associated with the highest risk of becoming ill.Entering the room only (room cleaning) and preparing the food (the plate) at floor level (but outside the resident's room) were not significant risk factors (table 2).

Laboratory results
No bacterial pathogen was detected among faecal specimens of 18 residents (14 cases) and 10 kitchen staff (one case) who were sampled.Faecal specimens from the 14 resident cases were assayed for virus detection.NLV-RNA phylogenetically clustered with viruses of the Hawaii genotype was detected in eight of them.

Infection control measures
Gloves and aprons were used from the start of the outbreak.Further infection control measures recommended by the regional GDD were implemented from 13 January onwards: they included increased handwashing discipline among nursing staff and visitors (using liquid soap and paper napkins); increased cleaning of beds, toilets, and bathrooms; disinfecting of all discharges and soiled articles; avoidance of staff and resident transfers between floors; distribution of food at floor level instead of from the central kitchen; and closure of the home to new admissions.

Discussion
This gastroenteritis outbreak was caused by a NLV strain often identified in the Netherlands (2), in January which is also typical of this infection (8).The shape of the epidemic curve and the clustering of cases among residents suggested that person to person transmission was the most likely mode of spread.The source that contaminated the index case was not identified but the virus was probably introduced by a single person and then spread throughout the building.No environmental investigation was performed, so the possibility that infection was introduced to a few people by a contaminated food cannot be excluded.
We obtained evidence that nursing staff contributed to person to person spread: close contact with residents (particularly feeding) exposed them to infection, which they could then pass on, (table 2), as described elsewhere (9,10).We have no objective explanation for the differences in attack rates between floors but the fact that the relationship between attack rates in residents and staff followed similar patterns supports the idea that transmission occurred between the staff and residents.
The attack rates reflected those reported elsewhere (1,2).They were Décembre / December Janvier / January higher for residents than for staff, which is typical (11) (45% and 29% for residents and staff, respectively, in 12 outbreaks in the Netherlands (2)), but not always the case (10).
Attack rates in residents increased with age, which could indicate higher dependency or lack of mobility, requiring more intensive nursing, but might suggest decreasing immunity with age (12).
It is hard to assess the efficacy of infection control measures, as the interventions coincided with the peak of the outbreak, after which cases began to decrease.In this outbreak, stringent measures were not applied in the first two weeks, during which the virus spread throughout the home.Aggressive measures, particularly restricting staff and resident movement, clustering of cases (cohorting) (8) and reinforcing personal hygiene (after close contact with an infected patient), may be needed at the start of a NLV outbreak to prevent extensive spread (13,14).Preventing the introduction of the virus into nursing homes appears to be more difficult (8). s ) of the risk factors was performed on the same data set, for staff who worked at floor level.Variables retained by the final model were helping the resident to eat (OR 8.3 [1.6-44]) and the working floor of the staff: floor 3 (OR 3.0 [0.4-20]), floor 4 (OR 0.9 [0.1-9]), floor 1 (OR 0.5 [0.1-4]), and floor 2 (OR 0.2 [0.1-2]), taking the staff working on all floors as a reference.