The rise of untreatable infections and superbugs has thrown antimicrobial resistance – the ability of a microorganism such as a bacterium, virus or parasite to resist antibiotics, antiviral and antiparasitic drugs – into the world health spotlight. It’s been described as one of the most serious public health threats of the 21st Century. It’s also an area where a great deal of research has already been conducted.
As such, researchers at the Fogarty International Center – one of the 27 institutes and centers that comprise the National Institutes of Health (NIH) in the US – wanted to know where they could focus their computational modeling efforts without going over old ground. In doing so, they embarked on a research project that would show, through the analysis of published research, what aspects of antimicrobial resistance (AMR) research the center should focus on. The project included a partnership with some of Elsevier’s data analytics experts.
The result is a landscape analysis of mathematical modeling research on AMR published in BMC Medicine. The article includes an analysis of the citation impact of the research, provides a clear view of where antimicrobial research should focus in the future and reveals ways in which the field can better align with WHO priorities.
How it all started – and why
Fogarty funds global studies and has interest in antimicrobial research and infectious disease transmission modeling. Dr. Cecile Viboud, a senior research scientist in the Division of International Epidemiology and Population Studies at Fogarty, wanted to identify opportunities to further strengthen its work in that area. To do this, she needed some data. She explained:
When you move into a new area, you don't really know what's interesting, what are the low hanging fruits or where you should focus. We thought a landscape analysis would be a really good way for us to get into the field, which is very broad and covers a lot of pathogens.
Aa AAAS fellow working with Dr. Viboud at NIH, Dr. Bamini Jayabalasingham started the review; she was instrumental in determining the search terms to use for selecting papers and connecting with the librarian at NIH. The team gathered as many studies as possible dealing with the mathematical modeling of AMR, giving them a pool of about 2,500.
Three months into the project, Bamini found a new position at Elsevier. Dr. Anna Niewiadomska, a Research Fellow for Fogerty, joined the team to finish the research and carried out the manual analysis of the papers. Anna, who specializes in researching emerging diseases, pared the pool down to 273 articles and went through the studies more closely to determine the diseases covered, the trends, the study characteristics and where there were gaps Fogarty could address.
They were particularly interested in the intervention strategies that were modelled – medical and behavioral interventions, such as vaccination strategies. The World Health Organization (WHO) and US Centers for Disease Control and Prevention (CDC) have high priority pathogens for antimicrobial resistance, so they could draw interesting links and investigate the literature to see what had already been done on them – very little, as it turned out. Anna commented:
It's quite a new field, and although mathematical modeling has been used for over a hundred years to study infectious diseases, it has only been applied to the spread of antimicrobial resistance since the 1990s. There are relatively few studies in this area compared to those modeling diseases like HIV and tuberculosis. But with the increase in hospital acquired infections, there is a lot of potential and certainly more to be done in this area, especially in low-income settings.
Anna and the team were busy systematically evaluating the 273 shortlisted studies they identified on population-level transmission models of antimicrobial resistance between 2006 and 2016. They identified trends related to timeframes, location and subject matter and noted the interventions studied as well as main findings. Then Bamini came with an unexpected suggestion.
A new view from Scopus data
When Bamini settled into her new position at Elsevier, she realized she had some powerful tools at her disposal that could add a new dimension to the research.
I reached back to Cecile and said, ‘Do you want to keep working together on this project?’ They were interested in the bibliometrics data and analyses to deepen their understanding of the impact and reach of the studies. So we continued to work on the review and I facilitated a way to understand citation impact.
Field Weighted Citation Impact (FWCI) is a value that indicates whether a paper has been cited as much as would be expected based on the age, type and field of publication; it is the ratio between the number of citations for a specific article and the average number of citations received by similar articles.
I think the FWCI told us that there's definitely room for improvement in terms of getting studies and their results publicized more and ensuring they’re used as data and evidence for making policy decisions. And while we didn’t use the geographic data in the final paper, the Scopus data also gave us a really interesting view of where the researchers are and where the studies are taking place.
On the bibliometrics side, Bamini explained, it was surprising to see there weren’t many data citations in the research:
Since modeling research uses data, we would expect to see data sources cited in the research, so it was a surprise not to see that much. The problem is, there is still no consistent model for referencing data – some datasets have a DOI, others don’t. That’s something that needs to be addressed.
The research itself was also in general not highly cited, which means it’s unlikely to have much of an impact on policy, for example. Making efforts to increase the visibility of research already being done is one way to increase its impact.
Bridging gaps in antimicrobial resistance research
The project highlighted a number of gaps in AMR that Fogarty could focus on filling in the coming years. For example, the diseases being covered, and the top five – HIV, influenza, malaria, tuberculosis and MRSA – don’t match up with the priority diseases listed by the WHO and the CDC for studying emerging threats in drug resistance. The location of the studies also didn’t match the need: most studies focused on higher income countries, whereas the mismanagement and overuse of antibiotics tends is an important and understudied issue in lower- and middle-income countries.
Two gaps particualrly well-suited for Fogarty are antibiotic use in animals and cross-species transmission of antimicrobial resistant bugs, and vaccination as an intervention strategy. They were struck by how many of the intervention studies were reactive, looking at preventing transmission of resistance once resistance had already been established, as opposed to proactive interventions that might prevent the evolution or development of resistance before transmission of resistance is taking a foothold. Cecile said:
For us, the vaccine-antimicrobial resistance interaction is a really good niche because we have expertise on modelling the impact of vaccines, especially for viral infections. And there's an indirect connection between respiratory viral infections and the overuse of antibiotics. We're hoping that through our partners we could identify interesting epidemiological data sets to research antimicrobial resistance in developing country settings and guide interventions.
They also identified a gap in the understanding of how antimicrobial resistance is transmitted between humans and animals; it is known that this takes place but there were only a handful of studies looking at it. As Cecile explained:
We're already working a lot on flu, which infects many different hosts and there is transmission between different species. So it would be a natural extension for us to look at this problem for antimicrobial resistance. But again it comes down to finding interesting epidemiological datasets that will be amenable to modeling and answering questions for policy.
The success of their continued work in these areas will depend on the availability of data – something that’s holding antimicrobial resistance modelling back in general, according to Anna:
Although we know the areas that would be interesting to look at, modelers are also limited by the datasets that are available to them. We hope that this work will serve as a useful guide to inform others on areas where data is most urgently needed, so that modellers can continue their work on AMR.
Cecile is now planning workshops on some of the areas they identified as being interesting to consider, inviting collaborators they know may have data or interesting points of view to see if there’s scope to start any projects.
They will also be monitoring the field for trends. Looking at the research over time, they saw a rise in papers on antimicrobial resistance up to a few years ago, at which point the research stagnated. This, Cecile believes, could be due to a decrease in funding; if this is the case, the work will continue to decline in the coming years at a time when we need more of it rather than less of it. Donor fatigue will be something to change urgently to support the fight against antimicrobial resistance.
Bibliomentric data, like the citation data Bamini contributed to the research, could also help monitor this issue, build a clear picture of what’s happening and help identify ways to continue bridging the gaps in research. “We really enjoyed the collaboration with Bamini and having access to some of the methods created by Elsevier,” Ceclile said.
Read the study
This article was published open access:
Anna Niewiadomska, Bamini Jayabalasingham and Cecile Viboud: “Population-level mathematical modeling of antimicrobial resistance: a systematic review,” BMC Medicine (April 2019)