This dashboard is designed for policymakers and healthcare professionals who have an interest in how antibiotics are being used in general practice and, in particular, how that usage could be optimised in the face of antimicrobial resistance (AMR). The dashboard showcases some of the critical findings of our research. It is hoped that this information will inform improved healthcare strategies relating to the usage of antibiotics.
This dashboard has many unique features which makes it particularly useful for clinicians, including:
The data used to generate the plots in this analytics platform have been provided by CPRD (Clinical Practice Research Datalink) practices and covers the time period from 2000 -2017. The dataset has been split into two for the analysis:
Dataset covering 2000 – 2015 - comprises over 25 million patient-GP consultation records, and covers around 500 distinct practices in total, from all areas of the UK. It has been used for two reasons:
Dataset covering 2015 – 2017 - covers the previous three complete years and provides the most up-to-date picture of antibiotic usage throughout the country. It contains around 4.5 million patient consultation records, from almost 400 practices. These data have been used to populate the majority of the plots including those showing which antibiotics are being prescribed by each practice, and the appropriateness of prescribing (based on the latest relevant guidelines).
The dataset used in the dashboard covers many years of primary care data for different practices, meaning any insight that is extracted is credible and widely applicable.
As always with any data-driven analysis, the strength of the findings are ultimately reliant on the quality of the data. By using many years of data covering many practices all over the country, we minimise the impact of any outliers/anomalies.
One aspect that affects the data quality is the consistency of coding by practitioners as they record the details of each consultation. This can be hugely variable, but by allowing practitioners to visualise aspects of their prescribing (through the dashboard) there is an incentive to improve coding to produce more accurate results.
The government has set ambitious targets on reducing antibiotic prescribing including an objective of reducing inappropriate prescribing by 50% by 2020/21 (page 15). This clearly highlights that antibiotic usage is suboptimal, and some improvement is needed. The goal of the dashboard is to inform practitioners of the current situation in primary care, and use the data analysis to guide where those improvements can be made.
Prescriptions that deviate from guidelines are judged by examining incidental prescribing of antibiotics. The treatment given to a patient is compared to three UK guidelines (NICE, PHE and GMMMG) for managing infections. If the treatment does not match the first, or second line recommended antibiotic, it is considered to be “a potentially deviation”. This takes into account any allergies that patients may have, but does not currently consider local resistance patterns. It is, however, still a very useful guide to how GPs are prescribing in line with the national and local guidelines.
The main limitation of using electronic healthcare records is that they are a static representation of each interaction between the patient and the healthcare practitioner, and only provide a snapshot of what happened during a consultation. Despite this, the records provide a valuable source of information, and with wider analysis of many records they can yield important insight into the state of prescribing in primary care.