Dr Sam Robson, Reader in Genomics and Bioinformatics and the Bioinformatics Lead at the Centre for Enzyme Innovation (CEI) talks about the work his team did using genome sequencing to track the spread of Covid-19 in local hospitals.
During the COVID-19 pandemic, our team worked closely with Portsmouth Hospitals University NHS Trust to support SARS-CoV-2 genome sequencing and use the data generated to better understand COVID-19 clinical severity and nosocomial spread at the hospital, helping to inform clinical practice and infection control measures.
After generating large amounts of SARS-CoV-2 genomic sequencing data during the first year of the pandemic, we set out to integrate the genome data with healthcare records from the hospital. We wanted to understand if changes in the genome of SARS-CoV-2 over time had a direct impact on patient outcomes. In particular, we wanted to know whether certain mutations in the virus had any correlation with death, ICU admission or the need for mechanical ventilation.
We found that it was unlikely that mutations in the virus had a major impact on differences in patient outcomes. Outcomes were much more likely to be a consequence of host factors, such as age or existing illnesses. We also saw that increased disease severity was often linked to cases that occurred earlier on in the pandemic, when we didn’t have any treatment options or vaccines for COVID-19.
Dr Sam Robson
The approach
As well as retrospectively analysing the data, the team used an artificial intelligence (AI) machine learning approach to see if it was possible to predict patient outcomes by looking at mutations in SARS-CoV-2 samples. With machine learning, you create a model that can be trained on existing data that can potentially predict future outcomes. AI can break down data in a way that humans can’t, and can pick up patterns that we may not see.
Whilst there were some SARS-CoV-2 mutations that were potentially predictive of the disease outcome, we were able to trace these back to other factors. For example, if a version of the virus with a particular mutation was spread around a ward with a vulnerable patient population, such as the ICU, it resulted in an increase in the number of negative outcomes for that particular mutation.
All of this tied into the work that we were doing looking at nosocomial spread. Nosocomial infections are infections that are acquired in hospital that were not present at the time of admission.
What’s important now is that we don’t lose the momentum that has been generated by research efforts like COG-UK and others across the globe. It’s essential that we use this time to fully understand what we don’t already know about SARS-CoV-2, and reflect on the lessons we have learnt from the pandemic.
We analysed how SARS-CoV-2 was being spread around the hospital and used the similarity between virus genomes, along with the length of time from admission to infection, to identify the cases that were most likely to represent transmission chains within the hospital environment. In many cases, those identified purely based on the time from admission to infection were shown not to match any other hospital cases from patients or staff. Instead, they matched infections circulating in the community, suggesting that they may be the result of chance introductions from the community rather than infections actively transmitting throughout the hospital.
We were able to identify whether patients were part of the same transmission chain within a given ward, which helped health care staff to target infection control measures more effectively.
The impact of the work
Overall, all the team’s work had a direct impact on clinical practice at Portsmouth Hospitals University NHS Trust. Results were being turned around and fed back to the hospital rapidly.
We also co-wrote a report to hospital management, along with our hospital research partners, with recommendations for improvements to infection control procedures based upon our analyses during the Alpha wave. Because of the very large increase in prevalence in the general population during the Alpha wave, it became difficult for us to discern nosocomial from community cases. Before Alpha, we knew there were small levels of five different variants circulating in the hospital.
When Alpha came along, we saw a large increase in Alpha cases, but also saw the circulating variants all increase in prevalence significantly. I put this down to difficulties in maintaining infection control procedures due to high numbers of hospital admissions, with the rising prevalence of Alpha cases in the region, which meant that the circulating variants were spreading more than they were before the Alpha wave hit. Recommendations for how to manage these changes all fed into the report that was prepared in collaboration with Portsmouth Hospitals University NHS Trust during the peak of Alpha infections.
Now and then
Looking back, the team did the best they could at that time with the data available. The use of genomic sequencing for infection control alongside epidemiological approaches was new to many at the time. In hindsight, we would likely do some things differently because we have that experience now. So, what’s important now is that we don’t lose the momentum that has been generated by research efforts like COG-UK and others across the globe. It’s essential that we use this time to fully understand what we don’t already know about SARS-CoV-2, and reflect on the lessons we have learnt from the pandemic.
None of this work would have been possible without the hard work of everybody at the University of Portsmouth and Portsmouth Hospitals University NHS Trust, the input from the patients and staff, and the guidance and leadership of COG-UK, particularly the interactions with expertise across the UK and the world that it brought.
I’m proud to have played a small part. It was an amazing example of collaborative science that I hope will act as an inspiration for future cross-partner collaborative scientific projects in the future.
Read the first of two research articles to be published from this work here.