Chronic Shedders as Reservoir for Nosocomial Transmission of Norovirus

Abstract

Norovirus (NoV) infection in immunocompromised patients may lead to prolonged norovirus shedding. Here, we demonstrate the involvement of three chronic shedders in hospital outbreaks. Combined epidemiological and molecular evidence suggests that in one case, NoV transmission occurred at least 17 days after the first diagnosis.

Norovirus (NoV) is a leading cause of acute gastroenteritis affecting people of all age groups (1, 3). Outbreaks with NoV occur often and worldwide (7). In the Netherlands, large numbers of outbreaks are reported each winter, particularly from health care institutions (6). In a systematic evaluation of newly diagnosed patients with norovirus in a large tertiary care hospital, we found that nosocomial norovirus transmission is common and may lead to chronic infection, disease, and shedding in at least 6% of patients (1). The case histories of the chronic patients showed various underlying illnesses resulting in impaired immunity, followed by prolonged NoV shedding, in some cases for periods longer than 1 year (5). The question arose whether these chronic shedders were possible sources for nosocomial infections within the hospital setting, even after they had been infected for a number of weeks. Because NoV cannot be cultured in vitro (2), it has remained impossible to assess whether the viruses shed by such patients are still infectious. The finding that noroviruses evolved within chronic patients suggested that detailed molecular virological data in combination with epidemiological data could be used to track possible routes of NoV transmission within the hospital (9, 10).

Epidemiological records for all NoV-positive patients diagnosed between 2005 and 2007 were retrieved from the Erasmus Medical Center data bank, including admission dates, sampling dates, and departments. Fecal samples associated with these cases had been stored at −80°C. Patient samples were sequenced as previously described (9). Briefly, the P2 domain of ORF2, with a length of approximately 700 nucleotides, was sequenced in both directions using an ABI Prism BigDye terminator version 3.0 ready reaction cycle sequencing kit. Strain sequences from patients with chronic NoV had been described previously (5). Norovirus-positive patients hospitalized in the same period (defined as 6 months before to 6 months after the first sampling of all chronic shedders) were selected, and their stool samples were used for analyses. Only strains that were unique and showed clustering with those of the chronic patients were included. This selection was made to represent the background diversity of norovirus strains circulating within the hospital. The strain sequences circulating in the community were represented by strains diagnosed at day zero after admission. To identify patients who may have been nosocomially infected by chronic shedders (5), strain sequences were obtained from patients that were routinely hospitalized with various disorders, excluding NoV as a cause. The sequences obtained were identified on the basis of 100% identity to sequences previously obtained from chronic shedders over a minimum fragment length of 600 nucleotides. The sequences were subsequently analyzed using TREECON for Windows (8) with the neighbor-joining method (single rooted) followed by bootstrapping. (Fig. 1).

FIG. 1.

Phylogenetic tree (TREECON for Windows) representing the sequences of the P2 domain of the GII.3 strains (left), analyzed with the neighbor-joining method followed by bootstrap analysis using 100 pseudoreplicates. Each strain is labeled with the patient identification code and XX-YYYY/Z, where XX is the year (e.g., 07 is 2007), YYYY is the unique case code, and Z is the time of diagnosis by the number of days after admission. At the right, the transmission scheme of cluster 3 is shown. In this scheme, detailed information describing the admission times and locations of the patients involved in the outbreak with patient 8 are shown. The chronically infected patients included in this tree are patients 1, 5, and 8. The colors (green, blue, and light blue) in the scheme indicate different departments where the patients were localized during their admission; gray indicates that the patient was discharged from the hospital. The black diagonal slashes show the sampling points during admission, while the sharp black arrow with a question mark indicates that no admission and discharge information was available. The blunt black arrow shows the admission date of NT23 (almost 2 months before that of chronic patient 8). The asterisk indicates that there was not enough sample (feces) available to perform norovirus diagnosis, while the patient had diarrhea or symptoms.

During the study period, we found three molecular clusters containing sequences of strains from patients who had been recognized as chronic shedders and other hospitalized patients; two clusters consisted of genotype GII.4-2006a strains and one of GII.3 strains (Fig. 1, Table 1). Chronic patient 6 (numbering corresponding with that in reference 5) was admitted to the hospital multiple times, while chronic patients 4 and 8 stayed in the hospital, mainly in the same location, during their norovirus infections. They were sampled and tested for NoV repeatedly during their admissions or visits, and sequences identical to theirs were detected among other admitted patients. Based on molecular information combined with demographic data, the most probable direction of transmission was assumed to be from the chronic patients to other hospitalized patients. The transmissions between chronic patients 4 and 6 and the other patients in the two GII.4 clusters occurred shortly after the chronic patients were first diagnosed (sequence tree not shown). In the GII.3 cluster, transmission was detected both shortly after initial diagnosis of the chronic patient (involving at least five other patients) and also after a longer interval (involving one patient, NT23 [NoV transmission patient 23]) (Fig. 1). The patients who were infected during the first week of this hospital outbreak all shared identical sequences in the genomic region analyzed. The sequence of the NoV strain detected in the second sample drawn on day 17 from chronic patient 8 showed a difference of one nucleotide compared to the sequence of the strain detected on day 0 and was identical to that of patient NT23, whose onset of disease occurred 20 days after the onset of disease of chronic patient 8. This strongly indicates that this patient, a 6-month-old ex-premature child with symptomatic nosocomial infection, was infected by chronic patient 8, at least 17 days after the first time norovirus was detected in patient 8 (day zero). The NT patients were hospitalized concurrently with patient 8, except for NT19, NT20, and NT22, who were admitted 3 to 6 days before chronic patient 8, and NT23, who had been admitted almost 2 months earlier than chronic patient 8, as indicated by the blue bars in Fig. 1. Two other clusters that included chronic shedders remained unresolved with respect to the direction of transmission, because all cases were diagnosed within a few days. However, patient 4 already had chronic diarrhea prior to hospitalization, which was resistant to treatment and coincided with chronic shedding of NoV. Therefore, it is plausible that this is a second example of transmission from a chronic shedder. The evidence was most convincing for patient 8, for whom the second sample (taken at day 17) showed a unique mutation that was identified in another patient hospitalized in the same ward. As this sequence was unique in the entire data set, a link with the chronic shedder is highly likely. However, sources of NoV in the hospital may vary from patients to staff, contaminated environments, and food items, and despite extensive outbreak investigations, the exact modes of transmission often remain unclear. This study, however, shows that chronically infected patients may contribute to the spread of NoV in hospitals. To our knowledge, this is the first study that provides evidence for this hypothesis, and it points to an important aspect of infection control: contrary to earlier beliefs, patients who have had NoV illness may shed the viruses for weeks, and recent data suggest that chronic shedding is relatively common in persons with impaired immune functions who contract the illness. Given the high incidence of NoV infections and the increasing size of the population that is immunocompromised, this problem is likely to increase in the years to come (4). Therefore, as part of infection control policy in the hospital, the possible contribution of such patients to nosocomial spread should be considered.

TABLE 1.

Transmission of NoV from three chronically infected patients

Cluster	No. of patients in clustera	Sources of samplesa	No. of days before NoV diagnosis of chronic patients	Clinical symptoms of chronic patients	Transmission delay (days)b	NoV genotype
1	5	Chr. pat. 4 and 3 NT pat.	3	Admitted with chronic diarrhea	4	GII.4-2006a
2	3	Chr. pat. 6 and 1 NT pat.	13	Admitted with infant regurgitation	4	GII.4-2006a
3	9	Chr. pat. 8 and 6 NT pat.	2	Admitted with diarrhea and vomiting	2, 3, 6, 7, and 17	GII.3

^aEach cluster contains one chronic patient (indicated as Chr. pat. 4, 6, and 8) and other hospitalized patients indicated as NT (norovirus transmission) patients.

^bTransmission delay was defined by the number of days between the date of diagnosis of the chronic patients and the onset of illness of the other patients in the cluster.