We all rely on antibiotics to treat infections, but our supply of antibiotics that work is running out. Some bacteria that cause infections have become highly resistant to antibiotics. This "antibiotic resistance" is much more likely to happen if we use antibiotics too often when we don't really need them. In the NHS, three-quarters of all antibiotics are prescribed by general practitioners (GPs). Often antibiotics are the right treatment for patients but sometimes patients are prescribed antibiotics for viral infections like coughs and colds, where antibiotics don't work. Some patients receive lots of antibiotics, others get them very rarely. Some GPs seem to prescribe antibiotics more often than others. The aim of this study is to use anonymous GP medical records (so individual patients can't be identified) to find out more about when and why antibiotics are prescribed in General Practice. We want to understand why some patients get antibiotics more often than others and when patients really need an antibiotic. We will use our work to develop computer simulations to help GPs decide when to prescribe antibiotics. By reducing the number of times that antibiotics are prescribed unnecessarily we will help to keep our current antibiotics working for longer.
Antibiotic overuse drives antimicrobial resistance. In primary care rates of antibiotic prescribing vary widely, associated with characteristics of the patient and GP prescribing behaviour.
The aim of this study is to investigate whether analyses of patient level characteristics can guide antibiotic prescribing decisions in primary care.
Using data from patients who contributed to CPRD between April 2007 and December 2015, we will describe the diagnosis, antibiotic treatment and clinical outcomes of patients with common infection syndromes. These analyses will be stratified by patient characteristics including: age, gender, social deprivation, selected co-morbidities, obesity and smoking.
In a series of cohort studies of patients with common infection syndromes, we will: a) estimate the rate of adverse outcomes comparing those who were treated with antibiotics to those who were not, using Poisson regression and b) calculate the number needed to treat to avoid one infection related adverse outcome using multi-level logistic regression, taking account of patient's vulnerability through the use of propensity scores. These analyses will be stratified by each of the patient characteristics listed above. This work will be synthesized through models that predict each patient's risk of adverse outcomes, comparing the scenario of antibiotic treatment versus no antibiotic treatment.
Rates of i) GP consultation for infection and ii) antibiotic prescribing
- Description of antibiotic prescribing patterns and the clinical indication for the prescription
- Risk of infection related adverse outcomes in patients with a GP consultation for respiratory tract infection, urinary tract infection, or skin and/or soft tissue infection, comparing patients who were a) prescribed an antibiotic, and b) not prescribed an antibiotic
- Number needed to treat (NNT) with antibiotics to avoid one infection-related adverse outcome
- Risk prediction models to guide clinical decisions about the need for antibiotic treatment for suspected infection in primary care and communication with patients
Laura Shallcross - Chief Investigator - University College London ( UCL )
Patrick Rockenschaub - Corresponding Applicant - University College London ( UCL )
Andrew Hayward - Collaborator - University College London ( UCL )
Anitha George - Collaborator - University College London ( UCL )
Arnoupe Jhass - Collaborator - University College London ( UCL )
David Prieto-Merino - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Logan Manikam - Collaborator - University College London ( UCL )
Peter Dutey-Magni - Collaborator - University College London ( UCL )
Selina Patel - Collaborator - University College London ( UCL )
HES Accident and Emergency;HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation