Strategies for tackling drug-resistant bugs put to the test by maths models

To tackle antibiotic resistance, scientists should look at how strains of drug-resistant bugs compete with those susceptible to drugs.

 That is the conclusion of researchers at Imperial College London, who have used mathematical modelling to determine what strategies for the timing and dosage of antibiotics work best to prevent drug-resistant strains of bacteria emerging.

When an infection occurs in the body, it is usually treated with antibiotics. However, some of the infecting bacteria may be resistant to antibiotics and survive the treatment, going on to proliferate and create more resistant bacteria.

There will also be bacteria present that are susceptible to antibiotics and can be killed, but these sometimes acquire drug resistance during the course of infection.


When infections are resistant to drugs, it is known as antimicrobial resistance. The problem means that diseases that are relatively minor today could soon become untreatable and fatal. It is estimated that antimicrobial resistance is already causing an extra 23,000 deaths per year in the United States, with projections reaching up to 10 million deaths by 2050.

The new study, published in the journal eLife, shows that in order to tackle the spread of bacteria that are resistant to antibiotics, scientists need to analyse how bugs that are drug-resistant are interacting with bacteria that are susceptible to drugs. They found that competition between the two populations plays a key role in determining the success of antibiotic treatment strategies.

These population questions are hard to study, but now they are incredibly important for clinical and public health.

– Dr Caroline Colijn

The traditional wisdom has been that infections should be treated ‘aggressively’ with an early, high-dose blast of antibiotics. The idea is that this rapidly reduces the population of drug-susceptible bacteria from which new, drug-resistant bacteria could arise.

Recently, some groups have argued that this strategy creates a greater pressure for the bacteria to change in order to survive – known as ‘selection pressure’ – that accelerates the emergence of resistant mutations from the susceptible bacteria population. Instead, a moderate strategy may be more effective at preventing the rise of resistance.


To test these two strategies, Dr Caroline Colijn from the Department of Mathematics at Imperial College London and her colleague Dr Ted Cohen from Yale University modelled the behaviour of drug-resistant and drug-susceptible bacteria under aggressive and moderate treatment regimes.

They modelled the dynamics under several scenarios, such as whether there were ample or scarce resources for the bacteria to use, the growth capabilities of each population, and what the immune response from the host would be.

The researchers found that both treatment strategies could be effective, but that it depended on the ways in which the drug-resistant and susceptible bacteria were interacting with one another. For example, if the population of susceptible bacteria is large and easily acquires resistance, an aggressive treatment strategy can prevent resistance arising in the first place.

This is because the drug resistant population relies on susceptible bacteria becoming resistant to strengthen their cause. An aggressive treatment quickly wipes out susceptible bacteria, preventing them from becoming resistant.


The team modelled conditions both within a host – one person with an infection – and within a whole population of people. The competition between strains of bacteria was found to have a similar effect at both levels.

However, Dr Colijn says that there is one important difference when looking at individual hosts and whole populations: “Even when aggressive treatment is best for individuals, it can still drive up levels of resistance in whole populations over time.”

In other words, even if each individual patient is unlikely to have resistant strains, once resistance does emerge, aggressive treatments could pave the way for further spread by reducing the competition resistant strains of bacteria face for hosts.

“In that situation, there may be a need to choose what is best for the patient today versus what is best in the long-term for the whole population,” said Dr Colijn.

While it’s not yet possible to determine what kind of interaction is going on in an individual’s body or in a population, these could perhaps be determined in the future with better observational techniques and better monitoring of the population.

“These population questions are hard to study in the lab or in trials as they take a long time to play out, and that’s where mathematical modelling comes in.” said Dr Colijn. “They have been studied for a long time in ecology, but now they are incredibly important for clinical and public health.”


Dr Colijn is part of the Antimicrobial Resistance Collaborative (ARC), a new initiative launched this month at Imperial.

ARC brings together data scientists, pharmacologists, statisticians, evolutionary biologists, electrical engineers, environmental analysts, policy experts and people from business to tackle the rising problem of antimicrobial resistance.