Investigating COVID-19 transmission in a tertiary hospital in Hanoi, Vietnam using social network analysis

Objectives In March 2020, a COVID-19 outbreak in a major referral hospital in Hanoi, Vietnam led to 7664 patients and staff being sent into lockdown for 2 weeks, and more than 52,200 persons across 49 provinces being quarantined. We assessed SARS-CoV-2 transmission patterns during this to-date largest hospital outbreak in Vietnam using social network analysis (SNA). Methods We constructed a directed relational network and calculated network metrics for 'degree', 'betweenness', 'closeness' and 'eigenvector' centrality to understand individual-level transmission patterns. We analysed network components and modularity to identify sub-network structures with disproportionately big effects. Results We detected 68 connections between 46 confirmed cases, of whom 27 (58.7%) were ancillary support staff, 7 (15.2%) caregivers, 6 (13%) patients and 2 (4.4%) nurses. Among the 10 most important cases selected by each SNA network metric, transmission dynamics clustered in 17 cases, of whom 12 (70.6%) cases were ancillary support staff. Ancillary support staff also constituted 71.1% of cases in the dominant sub-network and 68.8% of cases in the three largest sub-communities. Conclusions We identified non-clinical ancillary support staff, who are responsible for room service and food distribution in hospital wards in Vietnam, as a group with disproportionally big impacts on transmission dynamics during this outbreak. Our findings call for a holistic approach to nosocomial outbreak prevention and response that includes both clinical and non-clinical hospital staff. Our work also shows the potential of SNA as a complementary outbreak investigation method to better understand infection patterns in hospitals and similar settings.