For our Clinic of Pediatric Hematology and Oncology this was the first actively managed RSV-outbreak. In the previous two winter seasons in total only 3 RSV-positive patients were detected on the affected ward.
We studied the epidemiologic and molecular background of this outbreak.
Considering bed and room occupancy on the ward during the outbreak, direct patient to patient transmission (e.g. via droplets or contaminated surfaces) in cases 1 and 2 as well as 3 and 4 seemed epidemiologically possible as each pair was accommodated in the same room before samples were tested positive for RSV. Cases 5 and 7 acquired RSV at day 6 and 5 of their stay on the ward, respectively, suggesting nosocomial RSV acquisition. These two patients did not share rooms with other infected patients but were on ward during the outbreak. Cases 6 and 8 were tested positive for RSV on admission. Nosocomial acquisition was considered possible, as case 6 and 8 had been discharged 8 and 7 days, respectively, from the affected ward prior to re-admission. During this previous stay patients with symptoms of RTI and a positive RSV test were already on the ward. Nonetheless, community-onset still was an option for case 6 and 8.
Based on this epidemiologic background, our main hypothesis was that direct and indirect patient to patient transmission (the latter for example via the HCWs’ hands) caused the outbreak. However, at this point transmission by an infected visitor or HCWs acting as a point source could not be excluded. Moreover, taking all epidemiological data into account, a random introduction of several different community-acquired strains seemed unlikely to us. We suspected ongoing transmission of a single RSV variant and sequencing was used retrospectively to test this hypothesis (see below).
The standard, pre-outbreak infection control measures regarding RSV were mainly in line with previously made recommendations for hospitalized patients with hemato-oncologic disease[19, 28]. The additionally implemented measures, in particular single room accommodation for contact patients (quarantine), suspension of all social activities, and surgical masks for all HCWs and visitors at any time, addressed the postulated RSV transmission pathways during this outbreak. These postulated pathways were direct patient to patient transmission (e.g. roommate to roommate), but also transmission via HCWs and visitors.
Direct patient to patient transmission as the most probable route of infection has been shown by Lehners et al. in a large RSV outbreak in a German hematology and transplant unit . Jensen et al. described direct patient to patient transmission, mixed with introduction of strains from outside, in an outbreak affecting immunocompromised adults . We therefore focused on patient to patient transmission early during the outbreak by strict isolation precautions for RSV infected patients and contacts. Isolation for infected patients was also a key measure in a multimodal control bundle described by Inkster et al. . Contact patients were isolated for 8 days and repeatedly tested in order to disrupt infection chains as described in literature . This so called quarantine concerned 2 patients in our outbreak. One of them (case 4) was eventually tested RSV-positive at day 8 of quarantine while being negative at day 2 and 5. This underlines the value of the measure. Finally, we re-emphasized in training sessions the need for preemptive isolation of patients with respiratory symptoms. As all these measures required more isolation capacity on the ward, we restricted elective admissions and located all patients in single rooms.
As another measure we reduced direct patient to patient contacts on the ward by suspending community events, as active social behavior can be a risk factor for nosocomial RSV acquisition . Even so this noticeably restricted the social life for the patients and their families during the outbreak, we enforced this measure. We further restricted social contacts by temporally limiting visits of infants to the ward, as (especially young) infants are known to be the main reservoir for RSV and as our outbreak was approximately concurrent (slightly delayed) to the RSV community peak. Only parents were allowed to the ward, which is in line with an intervention done by Kelley et al. . A restrictive visiting policy is as well described by Singh et al. in a pediatric RSV outbreak .
The use of surgical masks for everyone on the ward is an important measure to prevent droplet associated nosocomial RSV transmissions. This is even more rational as RSV may be transmitted via symptomless or oligosymptomatic persons (e.g. HCWs or visitors) and the infectious period can in fact already begin 1 to 2 days before actual onset of symptoms. A literature review by French et al. concluded that personal protective equipment might be advantageous for reducing nosocomial RSV transmission . Kelly et al. showed that five HCWs showing only mild symptoms were involved in a RSV outbreak on an adult stem cell transplant unit . This underlines the necessity that HCWs with respiratory symptoms should not participate in direct patient care activities, at least in a high risk patient care setting. We re-emphasized this issue in training sessions for the HCWs. Although staff screening is described in literature , we were able to terminate this outbreak without staff screening. A cohort of HCWs to take care of solely RSV-positive patients as reported before  had also not been established but would have been another option in case of an ongoing outbreak.
Temporal survival of respiratory viruses in general  and specifically RSV  on inanimate surfaces is described, thus contact transmission via the hands of staff was conceivable for nosocomial acquisition. This is especially of importance as cough etiquette and compliance to basic hygienic principles may be reduced for obvious reasons in pediatric patients, so a higher environmental RSV burden is probable. Nonetheless we did not implement changes in the well established cleaning and disinfection procedures on the ward.
We detected prolonged RSV persistence (virus shedding), which has been reported in patients with hematological disorders . This finding needs to be considered for efficient outbreak control and favors the practice of repeated testing in immunocompromised patients as we did. Likewise, this is important as pediatric hemato-oncologic patients are often readmitted several times for cancer treatment cycles or fever in neutropenia. When symptoms are no longer present or mild but viruses are still being shed, RSV may be re-introduced to the ward. Thus, for termination of isolation precautions during the outbreak, we required negative results as reported before . In fact, two subsequent negative results at a minimum 2-day interval were necessary. The usefulness of this requirement is supported by the longitudinal course of the samples from patient 5 which were obtained in April and May. This patient produced positive specimens on two occasions, after one specimen had been tested negative (see Fig. 2).
Active RSV-surveillance by screening on admission and twice weekly for all patients on the ward insured rapid detection of RSV-positive patients. This is in line with successful infection control measures reported in literature [9, 12]. We presume that a prophylactic admission and prevalence RSV screening for all patients in the winter season might be helpful as a preventive measure in high risk populations. Therefore, one consequence of this outbreak was the implementation of an active RSV surveillance (admission and prevalence screening once weekly) in our Clinic for Pediatric Hematology and Oncology during the RSV season. The beginning and ending of this seasonal screening period is determined by in-house and regional/national RSV epidemiology . Moreover, pre-RSV-season audits involving clinicians, infection control staff and the Institute of Virology take place to ensure timely beginning of screening procedures and adherence to the existing infection control practices.
Molecular characterization of RSV strains, for instance by whole genome sequencing  or characterization of RSV G-Protein [16, 38], has been used to investigate nosocomial RSV outbreaks. We were able to collect and examine selected outbreak strains by G-Protein gene sequencing. We found that cases 1 to 7 were infected with an RSV A virus with identical G protein coding region. In case of patient C3 one nucleotide difference in the intergenic region of the G gene was observed in one of two samples collected five days apart (Fig. 3a. It is possible that this change was due to natural drift of the infecting virus over time or that this polymorphism is indicative of the presence of two slightly different viruses replicating in parallel and dominating on the one and the other day of sampling, respectively.
Moreover we found that case 8 had a RSV-B infection and that cases 2 and 7 were co-infected by RSV A and RSV B viruses. While sequence analysis of the earlier samples of case 2 and 7 revealed infection by the RSV A virus, the sequence analysis of the later specimen showed infection by an RSV B virus. With the available specimen, we were unable to distinguish if these two patients had a prolonged co-infection between these viruses or if they were sequentially infected by RSV A and RSV B. These findings became available only after the outbreak ended, as routine virological testing during the outbreak did not include molecular differentiation of RSV A and B. In retrospect, these results indicated the decision not to cohort RSV-patients during the outbreak, as we probably might have cohorted RSV-patients with different subtypes. Detailed sequencing analysis suggests that cases 2 and 7 were infected by an almost identical RSV B virus population. We observed three nucleotide differences between these viruses; however, nucleotides of the viruses at these three positions were ambiguous in both cases (G,T,G versus A,A,A residues). Thus, both patients were likely infected by a highly similar RSV B quasispecies which was characterized by two different nucleotide signatures varying in abundance between patients. In contrast, the RSV B virus infecting patient C8 differed in two key criteria. First, it did not show any sequence ambiguity at the three above mentioned residues that was characteristic for the RSV B virus population observed in patients C2 and C7. Second, it displayed three additional polymorphisms in the coding region of the G protein. Taken together, this suggests that patient C8 was infected by another RSV B virus and that there was no transmission from patients C2 and C7 to patient C8.
Outbreak strains of the subtype RSV A were highly similar and different from polyclonal strains from other non-outbreak pediatric patients (Fig. 3c). We therefore conclude that a single RSV A strain was introduced to the ward and then spread within the ward. Interestingly, the RSV B isolate C7_18_3 (1) was identical to the community strain RSV_02_1, however further epidemiologic and clinical information are not accessible for the non-outbreak patient. Taken together, the nucleotide analysis suggests independent introductions of at least 2 different RSV B strains into the ward affecting patient C2, C7 and C8, and transmission of one RSV A strain on the ward between patients C1 to C7.
Looking exclusively at the molecular analysis, it is not possible to disclose the exact transmission pathway of RSV A. RSV A might have been introduced to the ward by an infected patient (index patient) on the ward (maybe case 1) and was then successively transmitted from patient to patient. Alternatively, a point source, such as a RSV-positive HCW, may have caused the outbreak. However, in correlation with the epidemiologic observations such as overlapping patient stays on the ward, stay of case 1 and 2, and case 3 and 4 in a double room, and social activity on the ward in the initial phase of the outbreak, we consider a direct and indirect patient to patient transmission most likely.