Skip to content

Reducing antibiotic prescribing: a new BIT study published in The Lancet

19th Feb 2016

The growth of antimicrobial resistance (AMR) is one of the major health challenges of our time. The UK’s Review on Antimicrobial Resistance has forecast that AMR will result in 10 million deaths and $100 trillion in unachieved GDP a year by 2050.

One of the main causes of resistance is the use of antibiotics when they are not needed. This is particularly true in primary care, where some GP practices prescribe antibiotics much more often than others. As recent articles and reports have suggested, this is a situation where behavioural insights could add a lot – not least because all GP prescribing data is now publicly available online.

As a result, BIT, Public Health England, and England’s Chief Medical Officer ran a randomised trial to test whether GPs reduce their prescribing when they are informed that they are prescribing antibiotics at a relatively high rate. We have seen elsewhere that this kind of social norm feedback can be very powerful.

The intervention was simple and cheap. Eight hundred practices were sent a letter from the Chief Medical Officer stating that “the great majority (80%) of practices in [the recipient’s local area] prescribe fewer antibiotics per head than yours”. The letter also contained three simple actions to help ensure prescriptions were appropriate.

Today, The Lancet published the results. Over six months, those who received the letter reduced their antibiotic prescribing rates by 3.3% compared to those who didn’t. This led to 73,406 fewer antibiotic prescriptions, at a cost of 6p per prescription saved. There were direct prescription cost savings of £92,356.

GP antibiotic prescribing rates during study period (2014-2015)


To put this result into perspective: we calculate that if the letter had been sent to all eligible practices (as it was at the end of the trial), England’s overall antibiotic prescribing rate would have fallen by 0.85%. The current Five Year UK strategy has a target to reduce antibiotic prescribing by 4%.

This study reveals another important policy area where social norm information can be effective. These lessons will be incorporated into the way the health system manages antibiotics – and there is the opportunity to transfer them to any area which sees big variations in the way care is delivered.

But there is still much we do not know. For example, we only tested one version of the letter, so we cannot know what elements contributed to its success. And instead of printed letters, there seem to be greater opportunities to give feedback through digital platforms – like Oxford University’s, with whom we will be collaborating in the future. Watch out for further updates.