Discussion

In this norovirus cluster on the inpatient HSCT unit, the only significant factor associated with nosocomial norovirus acquisition was room location, congruent with other studies that established the importance of proximity and room contamination in healthcare-associated clusters.^{Reference Doshi, Woodwell, Kelleher, Mangan and Axelrod8,Reference Harris, Lopman, Cooper and O’Brien12} Nenonen et al^{Reference Nenonen, Hannoun and Svensson13} demonstrated that, among hospital wards experiencing an outbreak, 47% of environmental swabs in the affected ward were positive for norovirus, compared with only 7% in the unaffected ward. Positive pressure ventilation present in HSCT units but not standard hospital floors may also play a role in environmental contamination in this population by distributing aerosols outside a patient’s room.^{Reference Alsved, Fraenkel and Bohgard14} These results highlight the importance of early recognition and diagnosis of norovirus in the HSCT population, given the potential for rapid and sustained spread of norovirus long after the index patient has been discharged.

Based on our own findings and earlier studies on outbreaks in HSCT, we implemented a novel infection prevention protocol, which included a strict definition of norovirus infection that permitted earlier and broader implementation of control measures. This protocol successfully terminated the cluster and prevented outbreaks in subsequent norovirus seasons. The strict definition included in the triaging algorithm, which was designed to be extremely sensitive but have low specificity, was an important element of the infection prevention strategy because our data suggested that clinicians often did not consider norovirus as a potential cause of severe GIS in this population and that early identification is critical for preventing clusters (Fig. 3). Other causes, such as C. difficile infection, were more likely to be tested and evaluated. In addition to the highly sensitive case definition, strict screening of staff and patients was undertaken to ensure that cases were identified early in the disease course.

In the years following the implementation of this bundle, no additional healthcare-associated transmissions have been detected on the HSCT unit, and the protocol has been rolled out in a modified form to other units with no additional sustained nosocomial clusters of infection in our facility.

Many of the interventions included in the infection prevention protocol are cost-saving measures, even at very low effectiveness (10%) for preventing additional norovirus transmissions.^{Reference Lee, Wettstein and McGlone15} Healthcare-worker presenteeism is an important factor in nosocomial outbreaks, including on transplant and immunocompromised patient wards; thus our screening protocol was applied to patients and staff.^{Reference Doshi, Woodwell, Kelleher, Mangan and Axelrod8} Healthcare-worker screening and mandated time off if GIS develop is another intervention that may be both clinically effective and cost-saving, even with paid time off and early access to paid sick leave, depending upon the cause of the outbreak.^{Reference Lee, Wettstein and McGlone15} Ward closure is another strategy that has been used to manage and prevent healthcare-associated transmissions; however, modeling studies suggest that this strategy is more expensive than other infection prevention interventions and are only cost-saving if prevention effectiveness exceeds 90%.^{Reference Lee, Wettstein and McGlone15} Beyond cost-effectiveness, due the specialized nature of the HSCT unit, moving these immunocompromised patients to other hospital wards is often not feasible due to facility constraints; thus, ward closure may not be a practical option.

Another important finding was the limited testing for—and identification of—norovirus in the HSCT population.^{Reference Roddie, Paul and Benjamin16} Due to the frequency of GIS in this population and the multitude of infectious and noninfectious cases, differentiating norovirus from other causes of GIS is challenging. Among the 99 patients with severe GIS, only 48 (48%) were tested for norovirus, but most were tested for other types of infections, including C. difficile and Enterobacteriaceae. These results align with those of prior studies suggesting that GIS are often misdiagnosed as other infections, including C. difficile. ^{Reference Koo, Ajami and Jiang17,Reference MacAllister, Stednick, Golob, Huang and Pergam18} Thus, education of staff is of critical importance, especially during norovirus season, to identifying cases and to implementing infection prevention strategies as soon as infections begin to occur. Electronic systems can be leveraged for early case ascertainment and intervention by helping to identify cases early in their course and to encourage testing and isolation before substantial environmental contamination and nosocomial transmissions have occurred. Another important diagnostic intervention is the use of PCR, rather than EIA, to confirm cases. The sensitivity and specificity of EIA are both poor, potentially leading to many missed cases before transmission is identified.

This study was limited by small sample size and retrospective design, the limited number of patients who were tested for norovirus, and the diagnostic test itself (EIA during the study period), which has poor sensitivity and specificity. Thus, it is feasible that the cluster may have been larger and that factors other than environmental contamination may have also played a role. However, limited testing and poor predictive value of the test are likely to bias findings toward the null rather than toward finding an effect; thus, it is unlikely that either of these factors caused a false-positive association between environmental exposure and norovirus acquisition. Another possible source of norovirus was staff presenteeism, particularly for the cases that occurred parts of the HSCT unit that were not in close proximity to the index case room (Supplementary Fig. 1 online). However, we did not find any specific evidence of a healthcare-worker source for this cluster. Because infection prevention interventions were bundled, we were not able to determine which, if any, individual interventions were the most important. However, given the strong association between location and acquisition, improved environmental cleaning coupled with the aggressive screening protocol is likely to have played a major role in ending the outbreak.

In conclusion, norovirus presents similarly to other causes of GIS in the HSCT population, leading to delayed diagnosis and environmental contamination. Use of a standard algorithm for earlier identification of presumptive cases of GIS may help prevent future norovirus outbreaks.