It may not feel like a game, but a game-theory model reveals the problem with antibiotic prescribing
Next time a patient is in your office with what may or may not be an infection, imagine it’s the whole human population sitting in front of you.
That’s the modest proposal at the heart of a paper that demonstrates what many have casually noted: antibiotic prescribing is a tragedy of the commons, in which a shared resource is over-exploited by individuals acting in their best interests, to the eventual detriment of all.
In “Medical prescribing and antibiotic resistance: A game-theoretic analysis of a potentially catastrophic social dilemma”, published in PLOS One last month, psychology professor Andrew Colman and colleagues, at the University of Leicester, uses game theory and replicator dynamics to show “that rational doctors, motivated to attain the best outcomes for their own patients, will prescribe antibiotics irrespective of the level of antibiotic resistance in the population and the behaviour of other doctors, although they would achieve better long-term outcomes if their prescribing were more restrained”.
This means there is little point in just asking doctors, who are perfectly aware of the resistance threat, to prescribe less. Only by modifying the payoffs of the game – the expected results of each player’s actions – can we avoid catastrophic levels of antibiotic resistance.
The new Interagency Coordination Group on Antimicrobial Resistance’s report to the UN, released last month, says drug-resistant diseases cause at least 700,000 deaths globally a year, which could increase to 10 million deaths by 2050 “under the most alarming scenario if no action is taken”.
The response must include “strengthening surveillance, regulatory frameworks, professional education and oversight of antimicrobial prescription and use” as well as better infection prevention, diagnostics, vaccines and other alternatives.
But the Colman et al study suggests a more radical shift in thinking may be required.
Professor Colman says the tragedy of the commons has been used before metaphorically and as a conjecture to describe the problem, but it has never been mathematically proven. He took on that difficult task while working on another paper about antibiotic resistance with his former student Dr Carolyn Tarrant.
“When we started, the main part of the work was interviewing professional health workers in Britain, South Africa and Sri Lanka, collecting data on their prescribing behaviour,” he tells The Medical Republic.
“But because I’m a game theorist I wondered whether we could put this claim that it’s a social dilemma of the tragedy-of-the-commons type on a rigorous footing. It was just a conjecture. It’s something that everyone ‘knew’ but no-one had ever proved. And actually it was quite complicated to prove that it actually is.”
The maths – which TMR is in no position to explain – led him to conclude that not only was the situation a tragedy of the commons, but that there was “a global attractor at every doctor prescribing maximally”. In other words, from any starting configuration the dynamics of the system always tend towards more prescribing.
“The strategy of prescribing antibiotics to all symptomatic patients may lead to the evolution of antibiotic resistance, ultimately rendering antibiotics useless for treating even the most dangerous infections and also reducing the doctors’ own payoffs,” the paper says.
“Thus, doctors’ decision-making leads inexorably to an outcome that is worse for all of them and their patients than the outcome that would have resulted had they deviated from the game-theoretic rational strategy and exercised restraint.” Interventions to protect the antibiotic “commons” should therefore be informed by theory and evidence from research into social dilemmas, the paper concludes.
The model’s assumptions are not especially unrealistic or over-simplified: that doctors, presented with symptoms consistent with infection, must make prescribing decisions amid uncertainty; and that they are motivated exclusively to minimise morbidity in their patients.
Professor Colman says not only are doctors trained to put their patients’ interests first, they have a quasi-legal obligation to do so.
“So any considerations a doctor might have about the long-term implications of overprescribing have to be put aside,” he says.
“We think that’ll probably have to change.”
The original study led by Dr Tarrant, published last month in Clinical Microbiology and Infection, acts as a companion piece by proposing a raft of behavioural theory-based interventions, including enabling collective decision-making, monitoring, and reputational incentives and sanctions.
Professor Colman says collective decision-making could shield doctors, even GPs, from the risks of not prescribing.
“Prescription decisions are usually taken by single doctors, and it would be better if they were taken more collectively. In a hospital you could have the care team as a whole, and in surgeries with multiple GPs, the decisions could be collectivised to some extent to relieve the pressure on the individual if something goes wrong.”
Monitoring and incentives, more social than financial, are also crucial.
“There’s no point, at least in decision theory, in pleading with people to behave a certain way if the incentive structure leads them another way. If you want to change behaviour, it’s no good to tell people to act against their own interests – it has to be incentive-compatible.”
Dr Jonathan Dartnell, manager of clinical improvement services at Australia’s NPS MedicineWise, which educates professionals and patients, says there is value in the game-theory approach to the problem.
“Generally doctors will prescribe rationally for the benefit of the individual at the expense of the general population,” he says.
“Then what can you do? You need a consensus towards restraints and restrictions.
“So far we do have some restrictions on availability through the PBS, but most of the effort so far has been through education and providing feedback to GPs on their practice in comparison with their peers.”
The Australian Commission on Safety and Quality in Health Care’s newly released third report on antimicrobial use and resistance (AURA) says dispensing under the PBS has declined since 2016, following steady increases in the previous years, but that our rates “remain a serious public health issue” and are high compared with European countries and Canada. In 2017, 41.5% of Australians had at least one systemic antibiotic dispensed under the PBS.
Dr Dartnell says GPs have responded well to education and that more people in the community are concerned about antimicrobial resistance.
When it comes to being told how to do their job of prescribing, GPs’ responses vary widely: “Some think we’re overreaching and others would like us to go further”.
But he believes some GPs would actually welcome it if their discretion to prescribe antibiotics were curtailed.
“If the evidence base is there, they would move with it,” he says. “It also takes away some of the pain for them, because GPs have to spend time with patients arguing why antibiotics won’t be beneficial. If that were more constrained that would assist them in that consultation.”
An updated National AMR Strategy from Australia’s Office of Health Protection is due out soon.
One thing is obvious: we can no longer rely on the market to produce replacement drugs, since restricted sales make it not worth big pharma’s time and money. Sanofi and Novartis sold their antibiotic units last year, and just last month Achaogen Inc filed for bankruptcy after its new drug plazomycin (Zemdri) failed to reach $US1 million in sales in its first six months on the market.
It’s tempting to think that more point-of-care testing for bacterial infections would change the game by removing the uncertainty.
But Professor Mieke van Driel, head of general practice at the University of Queensland, says that’s the wrong way to look at it.
“A lot of bacterial infections are self-limiting – we have immune systems that in most people are very competent and can take care of anything that comes towards it, be it a virus or a bacteria,” she tells The Medical Republic. “So focusing on testing whether this is a bacterial infection or not, and if yes we can treat it with antibiotics, is giving the wrong message because then we’d still be over-treating.”
Tests such as C-reactive protein can actually get in doctors’ way by appearing to confirm that antibiotics are indicated, leading to difficult conversations. And rapid strep tests are problematic because about a quarter of children are asymptomatic carriers, so strep may not be the reason for a sore throat.
“Taking that uncertainty away by identifying the microbe I don’t think is going to solve any of this,” Professor van Driel says.
“All GPs are aware of the world around them and we feel an obligation as clinicians to look after our patients, but also the community and the future of our patients. It is in the best interest of our patient to make sure they don’t develop resistant bacteria – that’s a risk to them, their family and the rest of the community – so it’s not that we’re looking only at the individual.
“We’re aware that healthcare has limited resources and we are custodians. You do the best for your patient, but that’s in context of the world around you. If I’m in a remote Aboriginal community where there’s a high risk of rheumatic heart disease as a consequence of bacterial infections, I’m much more likely to prescribe antibiotics than I would in an urban setting.”
Professor van Driel says there is a large cultural element to the problem. This is illustrated by her own experience training in The Netherlands, one of the lowest-prescribing countries, and moving to neighbouring Belgium, which is one of the highest.
“A colleague of mine interviewed people in both countries about what they thought when they had a cough and were feeling sick. People in The Netherlands said ‘well, I’ll just ride it out’, and in Belgium they said ‘I have bronchitis! I must see my doctor who will prescribe me something’. That shows there’s a lot in the culture of how people look at infections and being sick, and how medicalised we have become, and that has an impact on help-seeking and how doctors respond to that.”
There’s no magic bullet to reduce prescribing, she says – a multi-faceted approach is needed, including campaigns such as NPS MedicineWise runs.
“Restrictions help – Australia has one of the lowest rates of resistance to quinolones because they’re not so accessible. And we know that resistance can go away if you take an antibiotic out of circulation.
“We’d really like to see antibiotic stewardship become part of accreditation in general practice, so it’s a standard measure we acknowledge as being critical in high-quality care, [and] we can preserve this precious resource.”
