COG-UK Project Hospital-Onset COVID-19 Infections Study

Hospitals are recognised to be a major risk for the spread of infections despite the availability of protective measures. Under normal circumstances, staff may acquire and transmit infections, but the health impact of nosocomial infection is greatest in vulnerable patients. For COVID-19, like SARS-CoV, MERS-CoV and Ebola virus, the risk of nosocomial spread of infection presents an additional and significant health risk to healthcare workers (HCW). During epidemics, normal infection prevention and control (IPC) practice is further complicated by the difficulties of distinguishing community- and hospital-acquired infections. This can lead to erroneous identification of nosocomial transmission, involving unnecessary IPC efforts, while true nosocomial transmissions are missed thereby putting patients and HCW at increased risk.

There is now good evidence that genome sequencing of epidemic viruses, together with standard IPC, better excludes nosocomial transmissions and, depending on the virus, better identifies routes of transmission, than IPC alone1-3. To date, all studies have been retrospective. However, the development of rapid nanopore sequencing methods enables identification of potentially linked or unlinked viruses within 24-48 hours: this timescale is short enough to inform clinical IPC decisions in near-real-time. Although COVID-19 has a low mutation rate estimated at around 2.5 changes per genome per month, it is generally agreed that sufficient viral diversity now exists to identify where patient and staff infections that are apparently clustered in time and space, are in fact due to different COVID-19 genotypes4. Such information would rapidly exclude nosocomial transmission as the cause of the cluster, reduce the need for IPC intervention and provide reassurance to healthcare workers that IPC measures including personal protective equipment (PPE), had not been breached. However, confirmation of COVID-19 transmission to patients and healthcare workers may be more difficult with a single observed mutation between two genomes feasibly representing anything between one and ten transmissions. Identical genomes will not necessarily provide evidence of a link between two cases. Nonetheless by placing genotypes detected within the framework of all the genotypes detected within the hospital setting, the surrounding community and COG-UK as a whole, it may be possible to postulate nosocomial transmission where comparatively uncommon genotypes are apparently linked or cluster in time and space.

The COG-UK initiative, which aims to sequence as many COVID-19 viruses as possible across the UK thus provides an important and unique opportunity to test whether viral sequence data produced in near-real-time, in addition to providing valuable information for public health planning, could also reduce uncertainties around nosocomial transmission events, better target IPC effort, improve hospital functioning and reduce the role of hospitals as a source of infection to the community.

To address this,the investigators propose an adjunctive study, COG-UK HOCI. COG-UK HOCI will take advantage of the COG-UK design, with its mixed model of smaller sequencing hubs located close to hospitals and a large centralised hub sequencing most viruses, to identify not only whether rapid viral sequencing is useful for patient management but how time-critical this might be ; turnaround times for sequence data from the central hub are likely to be longer (5-7 days) than those from local sequencing hubs (<48 hours).

COG-UK HOCI, by defining and reporting COVID-19 genotype frequencies within its participating hospitals, as compared to those in the wider community, will also have the potential to overcome some of the inherent barriers to identifying the likely sources of HOCI. The data generated will provide as accurate as possible a picture, given the constraints of viral genetic diversity, of numbers of COVID-19 infections being acquired by nosocomial transmission and where these transmissions are occurring. While COG-UK will provide data on the utility of viral genomics for national public health planning, COG-UK HOCI will quantify the utility of the same data for local management of nosocomial infection, whether observed benefits are time dependent and deliver the best estimates of how viral sequence data can be used to quantify HOCI.

The outputs from COG-UK HOCI will further inform decisions about the likely future use of viral genome sequencing for the management of epidemics and pandemics and how it might best be organised, centralised or diversified, to deliver maximal impact.

The overarching aim of this study is to determine the utility of whole-genome sequencing to provide additional insight into hospital-onset COVID-19 infections (HOCI) which, in turn, can optimise IPC measures. In addition, the project aims to provide early data to help quantify HOCI events and where these are occurring. These will contribute to local trust level planning and to understanding of the role of HOCIs in contributing to COVID-19 outcomes and spread.

Specifically, the study will determine the role of real-time availability of COVID-19 sequence data:

• In conjunction with routine IPC data, to identify and characterise HOCI
• To identify and characterise HOCI not previously identified by routine IPC data
• To generate estimated numbers of HOCI and where these are occurring
• To identify linked HOCI and hospital outbreaks
• To identify ways to reduce the incidence of HOCI
• In optimising IPC actions, e.g. by reducing the need for extra cleaning, ward closures etc where a hospital outbreaks are excluded
• In changing workload, e.g. by reducing the need for extra cleaning, ward closures etc where a hospital outbreaks are excluded Augmenting these approaches, the investigators will measure whether the above are influenced by the time to sequence data result.

COG-UK HOCI is a phase III prospective, interventional, cohort, superiority study.

Allocation to either rapid local sequencing (c.24-48h) or lack of rapid local sequencing (i.e. via Wellcome Sanger Institute at >96h) will be dependent on the time of the study (see timelines).

Proposed study duration: 12 months; comprising 6 months of set-up, baseline data collection, interventional data collection) and up to 6 months of data cleaning, data analysis and reporting.