Butler 2012.

Methods	Study design: cluster‐randomised controlled trial Unit of randomisation: general practices Trial duration: conducted during 2007 and 2008 Recruitment: 212 general practices approached at random from 454 eligible practices in Wales, UK. 102 practices expressed interest to participate; 70 recruited; 68 practices (˜480,000 patients) randomised to intervention or control groups (34 each) Methods of data collection: routine administrative systems (see 'Outcomes') Data collection time points: total numbers of dispensed oral antibiotic items (primary) and hospital admissions for possible RTIs and their complications (secondary): rate per 1000 patients for the year after the intervention practices were exposed to the intervention; re‐consultation for RTIs: (secondary; 7, 14 and 31 days after initial consultation). Cost data not extracted Length of follow‐up: 12 months
Participants	Clinicians (general practitioners (GPs) and nurse practitioners) and all patients registered with and consulting a participating general practice in Wales (practice list)
Interventions	Brief intervention name: Stemming the Tide of Antibiotic Resistance (STAR) educational programme: multifaceted flexible blended learning approach to continuing education for clinicians Recipients: clinicians (GPs and nurse practitioners) Providers: web‐based modules and practice‐based seminar led by a facilitator Health professional components: the programme is a blended learning experience, and based on Social Learning Theory to develop GPs sense of importance about change (the 'why' of change) and confidence in their ability to achieve change (the 'how' of change). The intervention consist of 7 parts (5 online, 1 face‐to‐face and 1 facilitator‐led practice‐based seminar): case‐scenarios and updated summaries of research evidence and guidelines; reflections on clinical judgement on antibiotic prescribing; a facilitator‐led practice‐based seminar presenting regional, local and practice‐level antibiotic prescribing and resistance data; novel communicative consulting skills and information exchange based on motivational interviewing; personal reflections on clinical practice; web‐based forum to share experiences and views; and a booster module completed 6 to 8 months after completion of the initial training to reinforce previously outlined communication skills. GPs had to complete each online learning component before the software would allow them access to the next. The intervention was flexible to allow GPs to access online components and try out new skills with patients at their convenience Patient components: nil Materials: web‐based materials Mode of delivery: interactive web‐based modules (including online videos in addition to a facilitator‐led practice‐based seminar Duration and intensity: not specified Comparator: usual care
Outcomes	Primary: total number of dispensed oral antibiotic items per 1000 registered patients for the year after practices were exposed to the STAR programme (Prescribing Audit Reports and Prescribing Catalogues; www.nhsbsa.nhs.uk/prescriptions) Secondary: hospital admission rates for possible RTIs and their complications per 1000 registered patients for the year after practices were exposed to the STAR programme.(Patient Episode Database for Wales); and practice re‐consultation rates (for patients with RTIs, practice re‐consultation rates were identified using diagnostic READ codes recorded by the general practitioner over 7, 14 and 31 days after an initial consultation) Costs data not extracted
Notes	Funding: yes Conflict of interest: none disclosed Published trial protocol: yes Trial registration: yes Ethics approval: yes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was conducted once all practices were recruited and all participating physicians had provided written consent. Dynamic block allocation was used to achieve balance between groups of practices for the potential confounders of previous rate of antibiotic dispensing (averaged over the past year), practice size (number of whole time equivalent staff at recruitment), and proportion of clinicians in the practice registered for the study. The practices were divided into 3 sets of 24, 22 and 22 practices; within each set we generated all possible allocations into 2 groups and selected the 1000 allocations within each set with the best balance with respect to the specified confounders. The independent statistician on the trial steering committee selected 1 allocation at random for each set and randomly assigned intervention or control to the 2 groups in each set to construct the final allocation
Allocation concealment (selection bias)	Low risk	Clinicians and researchers were blinded to group allocation until after randomisation
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not possible (multifaceted intervention programme)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data on antibiotic dispensing, hospital admissions and re‐consultations were collected through routine administrative systems that were not influenced by the study research process
Incomplete outcome data (attrition bias) All outcomes	Low risk	68 practices (˜480,000 patients) randomised to intervention (34 practices; 137 GPs, 2 nurse practitioners) or control (34 practices; 122 GPS, 2 nurse practitioners) groups. 2 practices (one in each group; including 12 intervention GPs and 7 control GPs) withdrew after randomisation but were included in the ITT analyses
Selective reporting (reporting bias)	Low risk	All indicated results reported. Published trial protocol available
Other bias	Low risk	Sample size (power) calculation: yes ITT or per protocol analysis: ITT analysis for primary outcome