Nosocomial Transmission of COVID-19

When the COVID-19 pandemic was announced by the WHO in 2020, I started a post-doc at Leeds University working on the SARS-CoV-2 Acquisition in Frontline health Care Workers – Evaluation to Inform Response (SAFER) research programme at University College London Hospital (UCLH). The aim of the SAFER project was to follow healthcare workers in UCLH, tracking their acquisition of the disease and investigating the risk factors associated with infection. The ultimate goal being to inform infection prevention and control on how to better protect our front-line healthcare workers.

My role on the project was to quantify the movements and contacts of healthcare workers within the hospital, and to identify how these behaviours relate to the risk of staff testing positive.

Protecting the front-line

During the early stages of the pandemic, front-line healthcare workers were three times more likely to test positive for COVID-19 than the general public, and early results from the SAFER project found 44% of staff at UCLH had evidence of being infected during the first wave. Clearly the risks to hospital staff are huge when a healthcare system is perturbed, in this case due to a novel and emerging disease.

But how do we make our healthcare institutions more resilient to pandemics, and how do we better protect healthcare workers from infections during hospital outbreaks? One step towards this goal is the provision of information and tools that empower decision makers.

Data gaps

Infection prevention and control practitioners are essential components of a hospitals immune system; quickly detecting, responding and eliminating outbreaks of disease. This requires information on the location history of patients and their contacts, which are used for surveillance activities and contact tracing. Data generating efforts have been greatly enhanced in hospitals through the use of electronic medical records; saving time, money and lives.

However, the equivalent information on healthcare workers is often missing, meaning investigations into staff infections are harder to conduct. Filling in these data gaps will boost the immunity of healthcare facilities by providing infection control practitioners with vital evidence to appropriately react to staff infections; which present a risk to patient safety.

My work involved the identification of routinely collected digital data sources relevant to staff behaviour at the hospital. These included rosters, security door access logs and electronic medical records, which can be used to quantify staff contact rates and mobility at an aggregate level. These digital footprints are akin to mobile phone records which helped investigate community level interventions during the pandemic.

Behaviour change

We demonstrated how to derive simple indicators of healthcare worker activity from the routine data, and used these to assess behaviour change. The data provide a passive means of investigating how working patterns react to disturbances within the healthcare system such as outbreaks and changes to occupational practices.

Infection control practitioners can use these behavioural indicators to monitor the success of administrative controls used as interventions to prevent nosocomial transmission; including staff/patient cohorting and restrictions to the flow of staff between wards. Applied in real time, the data can empower policy makers in the hospital with evidence to make informed decisions.

For example, our work showed that throughout the pandemic COVID-19 and non-COVID-19 wards were connected through the movements of staff. We also identified reactive staff cohorting in response to the number of COVID-19 patients, and a spike in indirect patient contacts that, if identified in real time, would have warranted further investigation.

It is worth noting that the behavioural indicators we describe are unable to provide insights into all transmission preventing behaviours, such as compliance with hand washing and PPE policies, but they do provide quantitative makers for spatial and social connectivity that can propagate infections during outbreaks, and that are currently unknown to infection control practitioners.

Rapid risk assessments

We investigated how staff behaviour relates to their risk of COVID-19 infection during the first year of the pandemic. Specifically, we looked at the mobility and contacts of staff two weeks prior to a COVID-19 test.

We found that, during the first wave, staff working more closely with COVID-19 patients were less likely to test positive. This protective effect did not continue beyond the first wave, demonstrating the need for sustained monitoring of risks as circumstances change.

Mobility around the hospital was also important in determining the risk of infection. Staff were more likely to test positive if exposed to a greater number of COVID-19 hotspots and if they moved between hotspots more frequently.

Curiously, not all COVID-19 hotspots in the hospital presented a risk of infection for staff. This was the case for the respiratory and critical care wards which, given their typical patient population, were likely more adequately equipped to deal with COVID-19 patients than other wards.

Activity in non COVID-19 hotspots generally did not result in higher risks of infection, with the exception of A&E. Triage is necessary in the emergency department and staff would have interacted with patients who had not yet been identified as COVID-19 positive or that were asymptomatic, potentially resulting in the higher risk of infection.

Moving forward

Innovations to protect healthcare workers are required if we want our healthcare systems to be resilient to pandemics. In part, this demands data and digital tools that can facilitate rapid risk assessments, thorough epidemiological investigations and ultimately support decision making in real-time. Routinely collected records provide a passive, readily available and cost effective method to fill the deficit of data on staff activity in the workplace. However, several challenges exist in translating this research into tools for routine practice.

Firstly, some caution is required when interpreting the routinely collected data, as biases may exist in the data generating process, and these likely differ between institutions. For example, some staff groups, patients and departments could be underrepresented within the data compared to others. While we have made efforts to conduct validation studies, more work on this is required.

The analysis of risk is relevant to infections other than COVID-19, including the various winter viruses that frequently cause outbreaks in hospitals. However, our research was only possible due to unprecedented levels of testing for COVID-19 during the pandemic. That said, staff screening is no longer routine and is unlikely to be introduced for other pathogens. Costs aside, policies on staff screening may not evolve without a greater emphasis on how protecting staff will contribute towards patient safety. In the absence of routine screening, the behavioural data can still be used to rapidly respond to and manage outbreaks by supporting contact tracing and related investigations.

If these data are integrated into routine surveillance, there must be transparency and safeguards in place that align with GDPR guidelines. Any privacy concerns of staff should be identified and addressed, since the resulting digital tools may be perceived as ‘Big Brother’ surveillance. With that in mind, data on staff activity should only be used for purposes relating to infection control and not for monitoring staff performance.

When healthcare systems are in a state of shock, vital services are at risk. Patient safety is a priority, but this can only be achieved if we bolster the immunity of healthcare facilities with tools to protect staff who are the lifeblood of the system. Simple innovations in the use of readily available data sources could contribute towards greater resilience against future pandemics, but the resulting digital tools need to be developed in consultation with the end user (infection control) to ensure outputs are appropriate and optimized.