A creeping threat to human health
While the world has been consumed this year with the immediate, large and fast‑moving threat posed by COVID‑19, another large but creeping health risk has been looming for several years: antimicrobial resistance (AMR).
‘Antimicrobial resistance is one of the biggest threats to human health today. It is our joint responsibility to change the way we use antibiotics so that our next generations can continue to benefit from these important medicines.’ (Prof. Brendan Murphy)
BETA partnered with the Department of Health to see whether a promising finding from the UK Behavioural Insights Team could help reduce the risks of AMR in the Australian context. We released our initial findings in 2018 and recently released a follow‑up report with new and surprising results.
What is AMR?
When we use antibiotics, some bacteria die but resistant bacteria can survive and even multiply. Overuse or inappropriate use of antibiotics makes resistant bacteria more common, accelerating the AMR process. Indeed, the pace of AMR is exceeding our capacity to develop new antimicrobial drugs able to target resistant bacteria.
If not addressed, AMR could take modern medical practice back to the pre-antibiotic era, when simple infections caused significant harm. Resistant infections are more difficult to treat and, in some cases, untreatable. AMR can lead to longer hospital stays, higher medical costs and, in some circumstances, death.
Antibiotic use per person remains high in Australia so it is important we take action to improve stewardship across all sectors where antibiotics are used. While Australia’s National Antimicrobial Resistance Strategy spans human health, animal health and agriculture, GPs have an important role to play as partners in combating AMR.
GPs prescribe more antibiotics than other health professionals due to the large numbers of patients seen in general practice, and the types of illnesses they treat. GPs are sometimes caught between patients demanding antibiotics, time pressures and managing uncertainty in diagnosing an infection. This can result in increased antibiotic prescribing.
Nudge versus superbug
We applied behavioural insights to the design of four different letters sent to high-prescribing GPs (specifically, those in the top 30 per cent of prescribers). The letters aimed to prompt GPs to reflect on whether they could reduce prescribing when appropriate and safe.
One letter simply contained standard information about AMR, and also included two posters from the National Prescribing Service (NPS).
The three other letters all began with peer comparison in the subject line of the letter, for example:
“Dear Doctor Smith, your prescribing rate is higher than 91 per cent of doctors in the Canberra region”.
The letters provided GPs with information on how their prescribing compared to their peers, to help inform future prescribing. Simple peer comparison feedback like this can be powerful because as humans we often look to the behaviour of others to guide our own choices. In particular, the targeting of our letters to individual GPs, and the comparison to GPs in their region provided a proximal and salient reference group.
The second peer‑comparison letter also included an eye-catching graph, illustrating the difference between that GP’s prescribing rate and the average for their region.
The third peer‑comparison letter included additional material on wait-and-see prescribing. For example, the idea was that doctors could place delayed prescribing stickers on a script to encourage patients to monitor their symptoms before deciding whether to fill the script.
While the peer comparison was the critical feature we wanted to examine, the messenger was also important. As these letters came from the then-Chief Medical Officer (CMO), Professor Brendan Murphy, a credible and trusted authority, we wanted to see if receiving the letter from a fellow doctor may have reinforced the impact of the peer comparison.
Finally, we timed the letters to go out in June 2017, around the start of the cold and flu season, when more people get sick with both viral and bacterial infections and more antibiotics are prescribed.
Putting our nudge to the test
We were reasonably confident our letters would reduce unnecessary prescribing. After all, a similar approach had worked in England, and the letters drew on well‑developed theories from psychology. Nonetheless, we weren’t certain—theory isn’t always reflected in practice, and results from one context don’t always translate to another.
To measure the impact of our letters, we randomly assigned high‑prescribing GPs to five groups. Four groups received different versions of the letter while the ‘control’ group received none. We then compared the monthly prescribing rates for GPs in each of the five groups. In conducting a ‘randomised controlled trial’ (RCT), we were seeking to emulate the world of medicine and preventive health by applying the same evaluation method they use to certify pharmaceuticals or vaccines are safe and effective.
The early results surprised us—in a good way. The UK BIT letter reduced prescribing rates by 3.3 per cent over six months, so we were expecting something similar. In fact, our three peer comparison letters drove prescribing rates down by 9 to 12 per cent over a six‑month period.
Persistence and (antimicrobial) resistance
A live issue for behavioural interventions like our letters is whether they only have a temporary effect before wearing off. While our letters still had an effect after six months, it was starting to diminish and we suspected it might continue to do so. Contrary to our expectations, however, the monthly effect plateaued at around 5 to 7 per cent through the summer and autumn of 2018.
And then we spotted something remarkable. Winter is the peak antibiotic period as more people present with cold and flu symptoms and GPs have to assess whether the symptoms reflect a virus (in which case antibiotics are unnecessary) or a bacterial infection. During the winter of 2018, a year after we sent the letters, the effect of the peer comparison letters increased somewhat, to around 6 to 8 per cent. While we don’t know the exact steps individual GPs took in response to the letters, it is possible that some of them were motivated to review their prescribing procedures and, as a result, establish new, ongoing protocols for themselves.
Why so effective?
Why was the inclusion of a peer comparison so effective? To explore this question, we draw on the theory underpinning the design of the letters. There are several theories of how peer comparisons may influence behaviour, including that they can give people information about ‘social norms’. People may adopt social norms because they are perceived to be a useful source of information (particularly in ambiguous situations), because they want to fit in, or because the norm represents the behaviour of a group that is important to them, that is, part of their ‘social identity’ (Reynolds, Subasic and Tindall 2014; Reynolds 2019).
In our case, the peer comparison could have been regarded as valuable information, in effect distilling the professional judgement of others who had similar training and routinely faced the same decision.
In addition, it was hypothesised that sending letters to GPs’ workplaces, comparing their prescribing behaviour to their peers, from the Chief Medical Officer would make salient their professional identity as physicians. Once aware their prescribing behaviour was high relative to their peers, this professional identity could motivate them to adjust their behaviour to be closer to what is ‘normal’ for their peer group.
This may be why the biggest change in behaviour was from doctors who were furthest from the norm. In the first month after the letters went out, the effect was about five percentage points higher for GPs in the top 15 per cent of prescribers than for the second 15 per cent. There was a similar gap in subsequent months.
Finally, part of the behaviour change may be due to a perception that prescribing behaviour was being monitored.
Next steps
AMR is a significant issue, and GPs have an important role to play as partners in combating AMR. While GPs understand the threat of AMR, they don’t necessarily know how their prescribing rates compare to other GPs. This makes it more difficult to reflect on whether their prescribing behaviours are appropriate and comparable to their peers. Prescribing data and peer comparison feedback assists GPs to review their prescribing practices, and has also resulted in lower rates of inappropriate antibiotic prescribing in England, the USA and Ireland.
Given the widespread effectiveness of peer comparisons in AMR studies, it would be reasonable to look for other opportunities for peer comparisons amongst professionals. Additionally, one pleasing aspect of this trial was how persistent the effect was. In future projects, it would be beneficial to learn more about what specific changes individual GPs made in response to the letter to understand why the effect was so persistent.
References
BETA 2020 ‘Nudge vs superbugs: 12 months on’ Working Paper, Department of Prime Minister and Cabinet https://behaviouraleconomics.pmc.gov.au/projects/nudge-vs-superbugs-behavioural-economics-trial-reduce-overprescribing-antibiotics
BETA 2018 ‘Nudge vs superbugs: a behavioural economics trial to reduce the overprescribing of antibiotics’ Working Paper, Department of Prime Minister and Cabinet https://behaviouraleconomics.pmc.gov.au/projects/nudge-vs-superbugs-behavioural-economics-trial-reduce-overprescribing-antibiotics
Bradley, D, Allen, S, Quinn, H, Bradley, B and Dolan, M 2019 ‘Social norm feedback reduces primary care antibiotic prescribing in a regression discontinuity study’ Journal of Antimicrobial Chemotherapy Vol 74 Issue 9 pp2797‑2802 https://doi.org/10.1093/jac/dkz222
Hallsworth, M, Chadborn, T, Sallis, A, et al. 2016 ‘Provision of social norm feedback to high prescribers of antibiotics in general practice: a pragmatic national randomised controlled trial’ The Lancet Vol 387 Issue 10029 pp1743‑1752 https://doi.org/10.1016/S0140-6736(16)00215-4
Meeker D, Linder J, Fox C, et al. 2016 ‘Effect of behavioral interventions on inappropriate antibiotic prescribing among primary care practices: a randomized clinical trial’ JAMA Vol 315 Issue 6 pp562‑570 https://jamanetwork.com/journals/jama/fullarticle/2488307
Reynolds, K 2019 ‘Social norms and how they impact behaviour’ Nature Human Behaviour Vol 3 pp14‑15 https://www.nature.com/articles/s41562-018-0498-x
Reynolds, K, Subašić, E and Tindall, K 2015 ‘The Problem of Behaviour Change: From Social Norms to an Ingroup Focus’ Social and Personality Psychology Compass Vol 9 Issue 1 pp45‑56 https://doi.org/10.1111/spc3.12155