Prof Andrew Hayward, Farr Institute London and UCL; Prof Paul Kellam, Imperial College London and Wellcome Trust Sanger Institute
Prof Steven Morris, UCL; Prof Deenan Pillay, UCL; Dr Eleni Nastouli and Dr Bridget Ferns, UCLH; Dr Ruth Blackburn and Dr Ellen Fragaszy, Farr Institute London; Dr Duncan Clark, BartsHealth; Dr Simon Watson, Wellcome Trust Sanger Institute; Dr Zisis Kozlakidis, UCL
In the UK there are an estimated 30,000 hospitalisations for influenza (flu) each year. When patients are admitted to hospital with flu they can spread this to other patients but we don’t know how often this happens. Outbreaks of respiratory infections – such as flu – are a serious concern in hospitals especially for patients whose immune functions are compromised (for example due to chemotherapy for cancer) and are at greater risk of complications including pneumonia and death. Hospital infection control strategies for flu have both human and financial implications. Moving affected patients into separate rooms and closing wards may lead to bed shortages and delays to planned treatment, but failure to isolate patients with flu may lead to further spread of infection. Hospitals currently rely on crude measures of whether or not patients’ caught their flu in hospital. For example, if patients develop flu in different parts of the hospital it is assumed they must be unconnected cases even though they may have been in contact during their stay or could have both been infected by a healthcare worker or visitor. This makes it difficult to identify outbreaks accurately and respond appropriately or to measure the flu impact in hospitals. Accurate and timely identification of outbreaks should allow infection control teams to contain outbreaks faster.
The ICONIC project hosted at The Farr Institute of Health Informatics Research involves UCL, the Wellcome Trust Sanger Institute, five major NHS Trusts and IntelTM. Through a £3.4m award from the Department of Health and the Wellcome Trust, ICONIC researchers use new approaches for investigating infectious diseases, such as flu. The influenza virus genome is constantly changing as viruses reproduce within the body and spread from person to person. We can use next generation sequencing (NGS) to determine the entire genome of the virus causing flu in less than a day, which offers ground-breaking opportunities for informed management of hospital flu outbreaks.
We have combined NGS data with routine hospital data, which describe a patient’s journey through different hospital wards during their stay. This powerful combination can identify how, where and when infected patients might have come into contact with each other. For example the figure below shows the cases of influenza in a hospital over time during two winter flu seasons. Grey dots represent genetically distinct cases. Coloured dots represent genetically highly similar cases. Blue lines represent epidemiological links derived from electronic health records. This illustrates that most cases of flu were acquired in the community and a smaller proportion were part of limited hospital outbreaks. Many of these transmission events would have been missed had we relied only on evidence of direct contact between patients.
Professor Andrew Hayward co-lead of the study said: 'This is an exciting example of how we can use cutting edge genetic technology and detailed electronic health records to investigate a preventable health problem. The approach generates huge amounts of data – through this project we have shown how these data can be automatically processed and influenza outbreak reports generated within a day or two of a suspected outbreak. The work shows spread of flu in hospitals is more common than we previously thought – the tools we have developed can be used routinely to support infection control teams to prevent and control outbreaks and save lives'.
This method allows us to see links between cases that would otherwise have been missed, target infection control processes to specific wards and determine whether sudden surges in case numbers reflect several smaller clusters of infection or a single uncontrolled outbreak. Using these techniques we estimate around one in seven cases of flu admitted to hospital spreads the infection to other patients.
The project was highlighted as one of the major British discoveries, opening the 2015 British Science Festival and gaining extensive coverage in the national press. The ICONIC process has been adopted by three large NHS Trusts. The team has analysed two influenza outbreaks at UCLH hospital and continues to expand the work over the coming flu seasons. The results of the study are currently being published in a journal widely read by doctors and researchers and will be used in medical education courses.