Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Antibiotic optimization  

Most antibiotics are prescribed in general practices. Prescribing any antibiotic contributes to antibiotic resistance, not only for that type of antibiotic but also for the strongest antibiotics that we have. So we need to keep all antibiotic prescribing as low as possible.  

However, it is hard for healthcare professionals not to prescribe antibiotics in general practice for many reasons. It isn’t usually obvious who does and does not need antibiotics – most infections will get better on their own. Patients are often used to getting antibiotics for infections and explaining why they don’t need them takes time in busy clinics, and patients may not be given advice on other ways to manage their symptoms. Lots of things have been found to work in research studies, but these are usually done in practices with lots of motivation to change and have extra resources like research staff that aren’t available in the real world. 

Instead, we targeted 9 general practice teams who were consistently high antibiotic prescribers. We co-developed, with healthcare professionals and citizens, online resources to give practices three antimicrobial stewardship strategies which might work for them, outside of restrictive research studies. The teams varied in which strategies they chose to use. Several teams liked using new diagnostic tests and some changed the way they worked to use these with their patients. Communication strategies and delayed antibiotic prescriptions were sometimes seen as something teams did already; however, when GPs read the online resources, they found the phrases they could use in discussions with patients extremely helpful. 

To support UKHSA, we were invited to incorporate our online resources into a major redesign of the national UKHSA/RCGP TARGET toolkit to support antimicrobial stewardship. We provided content for “Discussing antibiotics with patients” and “Leaflets to discuss with patients” (available on; averaging ~320 unique users per month). This was launched at a very well attended UKHSA online seminar to share our key messages. In future, we identified that practices need guidance about how to adopt and tailor strategies to fit with existing workflows and resources, rather than just providing the strategies themselves with generic implementation advice 

In a parallel survey study, we found some subgroups were much happier about getting a delayed/back-up prescription, such as women with minor coughs, than others, such as people who had already been ill for a long time. GPs could therefore consider targeting certain subgroups for offering delayed prescribing.


Responding to COVID-19  

When the COVID-19 pandemic started, we knew nothing about how many people were getting infected or how their immune systems were responding to the SARS-CoV-2 virus. Oxford University led several studies which helped UKHSA and the government manage their response. 

First, we were the key university partner with the Office for National Statistics developing the design, implementation, and analysis of the UK’s COVID-19 Infection Survey ( This national surveillance study monitored infection with, and immunity to, SARS-CoV-2 over 3 years, recruiting 535,750 individuals from 267,220 households and conducting 300,000-400,000 swabs and 120,000-150,000 blood tests each month. It played a vital role in informing key policy decisions around national lockdowns and working from home, strategies around vaccination and managing long COVID, and emergence of new variants, which had major economic and healthcare impacts. 

Second, we very quickly started studying >10,000 healthcare workers from Oxfordshire’s hospitals. We found that risks of getting COVID-19 were higher in some clinical areas (like Acute Medicine), and in some professions (like porters) than in places like the intensive care unit, which directly changed how the NHS managed risk. We were then the first worldwide to show that healthcare workers who had high levels of SARS-CoV-2 antibodies had substantially lower risks of getting re-infected, and amongst the first to show the substantial protection from vaccination in healthcare workers.  

Third, we went on to expand these studies to include patients in Oxfordshire hospitals. Using their electronic health records and genetic information from their viruses, we showed that spending time on a ward with patients who had caught SARS-CoV-2 in hospital was associated with substantial infection risks to both healthcare workers and other patients, informing additional measures to stop COVID-19 spreading in hospitals. 

Fourth, we used national NHS Test and Trace contact tracing data to directly estimate how infectious people with COVID-19 were, by following their immediate contacts both inside and outside their household. We found the amount of virus was the main factor in determining whether or not an infected individual was likely to transmit SARS-CoV-2. We then estimated how much vaccination reduced transmission of both the Alpha and Delta variants, including from cases infected despite vaccination. 

Last, we worked out that lateral flow devices (LFDs) could be used to identify and isolate infectious individuals who would be responsible for 83-90% of transmissions, allowing UKHSA to make rational recommendations on how best to use LFDs nationally. 


Modelling healthcare-associated infections  

UKHSA has to make decisions all the time about how best to control healthcare-associated infections which affect nearly 1 million people a year in the UK. However, often there is very little evidence to help decide what to do. We can use mathematical models to test the different things that we could do using a computer simulation and use results to suggest what might work best in the real-world. 

For example, diarrhoea caused by a bacteria called Clostridioides difficile (‘C diff’) has been increasing since 2020 and we don’t know why. People can get C diff from other people in hospital, and some become seriously unwell, but they can also carry the bacteria harmlessly. In the early 2000s, there was a new antibiotic-resistant strain that was much more dangerous. Although this strain is no longer a problem, another dangerous strain could come back again. We have used information from admissions to all hospitals across England to work out which are most “central” in the hospital network; we will use these hospitals like canaries in a mine” as part of a new surveillance system to find out whether there are new strains causing problems as quickly as possible in future.  

Another example is Carbapenemase-producing Enterobacteriales (CPE), which are highly antibiotic-resistant bacteria. More and more CPE are being found across the world. They are a significant threat as CPE infections are associated with poor outcomes for patients and CPE can spread rapidly within healthcare settings. Effective screening, identifying people carrying CPE (even if it isn’t making them sick) when they arrive at hospital, could help control hospital spread. We used models to test lots of different ways we could do this – like who to test and how many samples to take. This is particularly important because the tests are not perfect and different tests take different amounts of time to get results back, in which time people can pass CPE onto to others. We worked out changes to previous recommendations that could double the number of people carrying CPE found by screening and could also reduce the burden on hospital laboratories. These findings informed new national guidelines published by UKHSA, the “Framework of actions to contain carbapenemase-producing Enterobacterales now used by many hospitals. We also found that any future increases in the currently relatively low prevalence of CPE across much of England would have significant clinical and financial implications.