The present analysis included data from general practices that participated in the trial as well as data from non-trial general practices. We aimed to describe the performance of UK general practices with respect to antibiotic prescribing for respiratory illness in young and middle-aged adults. Methods The UK CPRD provided the data source for the study. The CPRD is a database Ruxolitinib clinical trial of prospectively collected electronic medical records

from approximately 7% of UK general practices. It includes records for all prescriptions issued and medical diagnoses recorded.9 The study included all CPRD general practices that were included in the cluster trial,8 as well as sample data for all CPRD general practices that were not included in the trial. All registered patients were included for the trial practices and, in order to provide a manageable data set for analysis, a random sample of registered patients was taken from non-trial practices. The period of study included the 12 months preceding the start date of the cluster trial with the date of random allocation was used as the index date.8 The

practices were allocated in five batches between 26 November 2010 and 26 April 2011. For non-trial practices, the median of the allocation dates, 20 January 2011, was used as the index date. Individual participants were adults aged 18–59 years. This was consistent with the eligibility criteria for the trial,7 which aimed to exclude children and older adults who might be at higher risk of complications. For each participant, we analysed their clinical record for 12 months before the trial index date. General practices were analysed as a single

group as there were no overall baseline differences between trial and non-trial practices with respect to consultation and antibiotic prescribing rates.8 The analysis used 232 general practice Read medical codes (recorded by general practitioners for each patient who consulted with a RTI), including those for ‘colds’ and ‘upper respiratory tract infection’ (URTI); ‘cough’ and ‘bronchitis’; ‘sore Drug_discovery throat’, including pharyngitis, laryngitis, tracheitis, epiglottitis and tonsillitis; ‘otitis-media’ including acute otitis-media and otitis-media; and ‘rhino-sinusitis’ including all forms of sinusitis. These were used to identify consultations for acute RTIs. The source of information for RTI consultations was represented by clinical, referral and test files data. Only first consultations within an episode were included using a 10-day time window. Therapy file data were used to ascertain antibiotic prescribing information. Antibiotic prescriptions were identified using drug codes that map to section 5.1 of the British National Formulary, excluding drugs used to tuberculosis and leprosy.