This trial aims to reduce unnecessary prescription of antibiotics by general practitioners (GPs) in England. Unnecessary prescriptions are defined as those that do not improve patient health outcomes. The intervention is to send GPs a letter from the Chief Medical Officer (CMO) that gives feedback on their practice's prescribing levels. Specifically, GPs in practices whose prescribing has increase by more than 4% over the past year will receive a letter stating that "The great majority (80%) of practices in England reduced or stabilised their antibiotic prescribing rates in 2016/17. However, your practice is in the minority that have increased their prescribing by more than 4%." The letter will also contain a leaflet to help GPs discuss self-care advice with patients and some advice to use delayed prescriptions. The investigators hypothesize that the antibiotic prescribing rate in will be lower for the treatment group compared to the control group, following the receipt of the letter.

The investigators hypothesize that the antibiotic prescribing rate in will be lower for the treatment group compared to the control group; the statistical analysis will compare prescribing in March, April, and over the summer (treating May-September as a single data point).

Inclusion Criteria:

• GP practices that increased the number of prescriptions in the financial year 2016/17, compared to the baseline of financial year 2015/16, by 4% or more

Exclusion Criteria:

• GP practices are excluded if their current level of prescription (doses per 1000 head of population) is classed as an outlier. The cut off for outliers is made at the 95th percentile of the distribution.
• Practices that have not been open since at least October 2013 will be excluded. This is because they will lack the historical data necessary to apply a control for seasonal effects to the main outcome variable.
• practices whose antibiotic prescribing rate is in the top 20% for their National Health Service (NHS) Local Area, controlling for relevant patient characteristics.

There is no individual participant data (IPD). The outcome measure is practice-level prescribing data, which is already publicly available.