Prevalence study: detection of NoV on environmental swabs in companies without recently reported outbreaks.
In the period from January 2008 to February 2009, a total of 2,496 swab samples were collected for analyses of the presence of NoV at 832 catering companies without a recently reported outbreak of gastroenteritis. Due to external factors that were beyond the control of the investigators, the goal of 1,000 companies could not be reached, and the number of inspections per months also showed some variation in time. In total 42 (1.7%) out of 2,496 environmental swabs from 35 (4.2%) catering companies tested positive for NoV ( and ), with threshold cycle (CT) values between 15.7 and 32.8 for GI types and CT values between 28.0 and 42.1 for GII types. At five catering companies, multiple samples tested positive. Of the 42 positive swab samples, 26 (61.9%) samples were taken from surfaces in the bathroom and 16 (38.1%) were taken from surfaces in the kitchen. Overall, NoV was detected three times more often on surfaces in the bathroom (26/832; 3.1%) than on surfaces in the kitchen (16/1,664; 0.96%). Assuming that swab testing would be used as a diagnostic method for possible involvement of a food establishment in an outbreak, this indicated a specificity of 95.8% (797/832) overall.
Detection of NoV in environmental samples from catering companies with and without association with recently reported gastroenteritis
Distribution of NoV-positive environmental samples over different types of catering companies
Outbreak study: detection of NoV on environmental swabs collected during an outbreak.
In the period from January 2006 to January 2009, 370 environmental swab samples were collected for analyses of the presence of NoV at 72 sites implicated in a viral gastroenteritis outbreak investigation where food-borne transmission was suspected. The presence of NoV was demonstrated for 44 (61.1%) of the 72 sites under investigation and in 147 (39.7%) of the 370 samples analyzed (). As observed in the prevalence survey, NoV was more frequently detected if the samples had been taken from surfaces in bathrooms (52.9%) than if they had been taken from surfaces in the kitchen (29.4%). Assuming that swab testing would be used as a diagnostic method to predict involvement of a food establishment in an outbreak, this indicated a sensitivity of 61.1% (44/72) overall.
Effect of seasons on the prevalence of NoV in catering industries.
NoV was detected 4 times more frequently on swabs collected from companies inspected in the norovirus season, November to March (28/480; 5.8%), than on swabs collected from companies inspected in the typical norovirus off-season, April to September (7/352; 2.0%) (P = 0.006). In univariate analysis, the risk of NoV presence was found to be significantly higher during the high season (OR, 3.1; 95% CI, 1.3 to 7.1) (). These findings indicate a seasonal difference in the above-mentioned specificity for swabs as a diagnostic method for possible involvement of a food establishment in an outbreak, i.e., a specificity of 94.2% (452/480) in the high season compared to 98% (345/352) in the off-season.
Significant and borderline risk factors for the presence of NoV in catering industries
Effects of characteristics of catering industries on the prevalence of NoV.
During the prevalence study, most of the inspections were performed in restaurants (n = 446), followed by lunchrooms (n = 112). To a lesser extent, other types of companies, for example, food establishments/kitchens for elderly homes or hotels, were also visited for inspection. The prevalence of NoV on surfaces among these different types of catering companies varied significantly (P = 0.015), although the numbers for some of the company types were low ( and ). In univariate analysis, the 9-category variable resulted in an invalid model. When all company types were entered as separate variables in multivariate logistic regression models to enable identification of a potential difference between company types, the elderly home (OR, 8.7; 95% CI, 2.7 to 28.8) and lunch rooms (OR, 2.4; 95% CI, 1.0 to 5.4) showed statistically significant risk of NoV presence. A risk (OR, 6.1; 95% CI, 0.7 to 54.7) was also found for hotels and pensions, showing a borderline significant risk of NoV presence.
The presence of NoV on surfaces appeared not to be significantly related to the number of employees, including serving, dishwashing, and cleaning employees (data not shown). However, the prevalence of NoV was twice as high in catering companies with separate staff bathrooms (5.6%) as in catering companies without separate staff bathrooms (2.8%), resulting in significant univariate risk (OR, 2.1; 95% CI, 1.0 to 4.2) ( and ). This finding was irrespective of the company size or whether hand-washing facilities were present, absent, or only in the near vicinity and irrespective of hands being dried by blowers, towels, or paper.
The distribution of sampling over the five regions was scheduled to be in relation to the number of catering companies in each region (). When the 5 regions were compared separately, no statistical differences were found in the norovirus prevalence. When low- and high-density regions were clustered together, the data suggested that regions with a lower population density had a lower NoV prevalence than regions with a higher population density ( and ), which appeared borderline statistically significant (P < 0.09), showing that this prevalence seems to follow the density of population in these regions.
Typing of NoV RNA detected on environmental surfaces.
In the prevalence study, GI NoV strains were detected in samples collected from two companies, whereas GII NoV strains were detected in samples collected from 33 companies. Sequence analyses were not always successful, as for some GII presumptive positive samples, the signal was too weak. GII strains that were sequenced had been detected by real-time assay at a CT value between 28.0 and 41.2, whereas strains that could not be sequenced had been detected at a CT value between 40.2 and 42.1. For 25 of the 35 positive catering companies, the genotypes of the positive samples could be successfully determined (, top). GII.4 was the most frequently detected genotype, including various GII.4 variants, whereas genotypes GI.2, GI.4, GIIb, and GII.2 were detected only once or twice. For three companies with multiple positive samples, identical sequences were obtained for the two or three samples.
Fig. 2. NoV genotypes detected on environmental swabs. Shown are the different NoV genotypes detected on environmental swabs during the prevalence study (top) and during outbreak investigations (bottom). The number of companies with NoV on swabs per number of (more ...)
During the year-round prevalence study, NoV strains were also detected in samples collected from surfaces in companies with a reported outbreak, including 4 GI NoV strains and 9 GII NoV strains (, bottom). In the 3-year period from January 2006 to January 2009, NoV was detected in 44 (61.1%) out of 72 companies associated with a reported outbreak. In 38 of these outbreaks, sequences were obtained, including 31 GII strains, with GII.4 the most frequently detected genotype, and 7 GI strains. The finding of multiple GI.3 food-borne outbreaks within a 3- to 4-month period (October 2007 to February 2008), all with GI.3-positive swab samples and one with a GI.3-positive food sample (salami) (data not shown), was remarkable.
Comparison of background sequences and sequences from clinical specimens.
Phylogenetic analyses (150 nt) using data obtained from the clinical samples collected showed that the sequences of strains present in catering companies were interspersed with those circulating in Dutch patients. Comparative sequence analysis was performed to identify possible clustering using sequences detected during the prevalence study in catering companies (n = 25) or during food-borne outbreak investigations (n = 13) (either food or swab samples, both in the period from January 2008 to January 2009) and sequences obtained from reported outbreaks of gastroenteritis, i.e., clinical samples collected in the period January 2008 through September 2009 (n = 231).
Thirty-five clusters were identified, all consisting of 100% identical sequences. Most clusters (22 GII.4 and 3 GIIe) consisted of clinical samples only. Four clusters (two GII.4 and two GIIb) consisted of sequences obtained from clinical samples and from environmental samples (e.g., oysters or swabs) collected during food-borne outbreaks without established epidemiological associations. The remaining 6 clusters (5 GII.4 and 1 GIIb) all consisted of at least one sequence detected during the prevalence study and were elaborated to present the divergence in time and location between matching catering companies and outbreak events (). For some clusters (no. 2, 4, and 5 ), the strains first appeared in the catering companies before being reported in outbreaks, whereas in other cases (no. 1 and 3), the reverse was the case. The most striking was the GIIb cluster (no. 1), with a temporal spacing of as little as 12 days and a geographical spacing of only 18 km. No GI clusters were identified.
Clusters of 100% identical sequences through surveillance in food establishments and reported outbreaks of gastroenteritis