UCL Institute of Health Informatics


Dr Anna Aryee

PhD Student

Institute of Health Informatics
Faculty of Population Health Sciences

Academic Backround

I am a specialist trainee in Infectious Diseases & Microbiology, a previous NIHR Academic Clinical Fellow and a current UCL Wellcome Clinical PhD Fellow. My clinical training and keen interest in developing interventions to combat antibiotic resistance have led me to develop a PhD project which examines the epidemiology of community-onset urinary tract infections and bloodstream infections, and how these relate to antibiotic resistance.


Research summary

My research focuses on "Identifying patients at risk of severe outcomes and antibiotic resistance in relation to community-onset urinary tract infections".

Abstract: Antibiotic resistance is a growing and alarming global problem and there is currently increasing concern around resistant Gram-negative bacteria. There are a number of drivers, including transmission of resistant microbes in healthcare settings and through the food chain, as well as selection pressure from antibiotic use. One of the ways in which we can halt its spread is therefore through prescribing: avoiding the use of antibiotics when they are not warranted and using the narrowest spectrum possible when they are.

Judicious antibiotic use must balance the need for effective antibiotics in certain cases, and the safety of de-escalating, or avoiding them altogether, in others. Fear of treatment failure is a common reason for clinicians to prescribe antibiotics when they may not be warranted.  Antimicrobial stewardship programmes have been shown to be an effective means of controlling and guiding antibiotic use, but our ability to develop effective programmes relies on information about the risk factors, natural history, microbiology and outcomes for the infections that are being treated.

Urinary tract infections (UTIs) are a common indication for antibiotic prescriptions in the community, coming second only to prescriptions for respiratory tract infections, and are mainly caused by Gram-negative bacteria. A potential severe consequence of inappropriately or undertreated community-onset UTIs is bloodstream infection (BSI) or bacteraemia, and the UK has seen an increase in the use of antibiotics of last resort such as carbapenems in hospitals to treat these severe infections. Whilst rates of carbapenem resistance are low in the UK, countries such as Greece and India, which have high levels of carbapenem use, have seen concomitant increases in carbapenem-resistant infections, for which there are few treatment options.

Community studies of respiratory and gastrointestinal infection have shown a very large "clinical iceberg of disease", with most cases of laboratory confirmed infection not consulting in primary care. This suggests that many mild infections can be self-managed without antibiotics. We do not, however, have comparable information on the natural history of community-onset UTIs. Whilst there is no surveillance of UTIs in the community, mandatory surveillance of BSI due to Escherichia coli, the most frequently isolated Gram-negative bacterium, has shown that the majority of these infections are community-acquired, and that the urinary tract is the most common source. Rates of Escherichia coli bacteraemia (ECB) are increasing, and halting this rise is a public health priority. UTIs also provide an opportunity to analyse community microbiology data on a large scale, as urine cultures are commonly sent to laboratories from primary care (albeit from a biased sample of patients).

The usefulness of these urine samples for research has to date been hampered by lack of data on patient’s prior exposure to antibiotics. Similarly, primary care data does not commonly include microbiology results. Our understanding of the relationship between antibiotic exposure and ABR has thus been hampered by the fact that prescribing data and resistance data are largely held in separate datasets, and linkage of these datasets is crucial to this understanding.

In this PhD, I will investigate the epidemiology of community-onset UTI, and explore the potential clinical iceberg of UTI using a series of datasets which provide the opportunity to examine both primary and secondary care settings, and link this to microbiology data. I aim to improve prescribing for community UTIs by identifying those patients most at risk of severe outcomes, and those at most risk of antibiotic resistant infections, through a series of linked studies as outlined below.

1. Bug Watch

The Preserving Antibiotics through Safe Stewardship (PASS) project is funded by the Economic and Social Research Council and aims to inform the development of multifaceted behavioural interventions that will strengthen antibiotic stewardship in the community and health service settings in the UK. This includes Bug Watch, a national community cohort in which 21745 adult participants from the 2014-2016 Health Surveys for England (HSE) will be invited to complete a daily online symptom diary during episodes of common infections. Bug Watch data will be used to investigate when and how individuals in the community self-manage their infections, including UTI, and what triggers the decision to seek medical advice. I will investigate the data collected on patients in the community who report UTI symptoms and estimate the proportion of infection syndromes that lead to primary care consultation and antibiotic prescription, as well as investigating how healthcare seeking and treatment behaviours vary by demographics, infection profiles, and knowledge and attitudes towards antibiotics.

2. Treatment failure and antibiotic resistance in East London Primary Care

The Clinical Effectiveness Group (CEG) East London database provides a unique and rich source which contains linked primary and secondary care data from patients registered at General Practices in the three CCGs City & Hackney, Newham and Tower Hamlets. I will link data from two CCGs (Newham and Tower Hamlets) to microbiology data on urine and blood cultures from Barts Health, which is the centralised laboratory for the region.

This linkage will enable a series of analyses to support better antibiotic prescribing decisions in primary care by examining treatment failure (defined as re-consultation or hospitalisation) and antibiotic resistance. I will use 3 studies to address the following questions: 

  1. Can we predict the risk of treatment failure in patients presenting with UTI syndrome in primary care?

  2. How does a mismatch between initial antibiotic prescribed and urine culture sensitivity affect the risk of treatment failure?

  3. Using data from primary care, can we predict which patients who consult for UTI syndrome will have a urine culture which is resistant to first line therapy?

3. Systematic review of risk factors for Escherichia coli bacteraemia (ECB)

As most ECB is community-acquired, it has been argued that there is limited scope for preventative strategies. Recent studies, however, show that a large proportion of community acquired cases are actually healthcare

associated, with patients having had contact with hospital or outpatient services in the preceding months. This potentially increases the scope for interventions, for example around urinary catheter care, which might reduce their incidence. Given the large and increasing number of bacteraemias, even modest risk factor modification could lead to large reductions in morbidity and mortality. This review aims to provide a systematic synthesis of the available published evidence evaluating risk factors for ECB. The results may inform interventions to reduce ECB, as well as inform antibiotic prescribing policy.

4. Outcomes of patients admitted to University Hospital Birmingham with Gram-negative bacteraemia

A number of studies looking at outcomes in ECB have shown a wide range of case fatality rates, as well as conflicting results in terms of the effect of inappropriate antibiotic therapy on mortality and length of stay. Source of bacteraemia has also been posited as an important factor in patient outcomes, with a number of studies finding lower mortality in patients with urinary source ECB.

Using electronic healthcare records from Queen Elizabeth Hospital Birmingham (QEHB), I will identify a cohort of patients who were admitted with a community-acquired bloodstream infection with Escherichia coli, Klebsiella pneumoniae or Pseudomonas aeruginosa. I will examine current antibiotic prescribing practice by clinical team, including empirical antibiotic choices and de-escalation of therapy. I will then conduct multivariate regression analyses in order to develop risk prediction models for outcomes including in-hospital mortality, ICU admission and total length of stay, looking specifically at the effect of mismatch between empirical antibiotic therapy and antibiotic sensitivity of the blood culture isolate.


Anna is supervised by Professor Andrew Hayward and Dr Laura Shallcross.