Community Health & Antibiotic use Surveillance through Citizen Science

Our solution is a citizen science initiative to collect high-resolution individual-level antibiotic use and resistance data in the community in Thailand, Vietnam, Myanmar, and Cambodia. We will leverage a user-friendly, sustainable mobile platform and integrate collected data into existing analytical pipelines to improve resistance burden estimates.

Cherry Lim, postdoctoral researcher, University of Oxford

Antibiotic consumption is a key driver of antimicrobial resistant (AMR) infections, but we know remarkably little about community-level antibiotic consumption in low- and middle-income countries (LMICs). In countries with limited microbiological diagnostic capacity this problem is compounded by an accompanying lack of AMR data. The lack of reliable data poses challenges for establishing targets, understanding AMR burden and for designing and evaluating interventions.

In settings where antibiotics can easily be purchased from unlicensed sources, commercially available official sales and procurement data are not expected to represent good proxies for household usage and no widely available validated data sources exist. Moreover, such aggregated datasets fail to distinguish between community and hospital usage and do not allow insights into drivers of antibiotic usage. Household surveys provide an alternative source of antibiotic use data, but are hampered by lack of knowledge about which drugs are antibiotics among household members. State-of-the-art approaches, such as the ‘Drug Bag’ method, overcome such challenges, but are labour-intensive and not easily scaled. 

There is an urgent need for an easy-to-use, low cost platform to collect community-level linked antibiotic consumption data for LMICs and for innovative approaches to collect AMR data in the absence of local microbiological diagnostic capacity.

Our solution will serve community members in Southeast Asia, public health policymakers, and the wider research community. The primary needs it will address are lack of reliable information on antibiotic use in the community, the drivers of such use, and the prevalence of antimicrobial resistance in the community. The benefits to these communities will include improved policy decision-making and targeted interventions supported by better data. Citizen scientists participating in the project will also directly benefit from learning opportunities and engagement activities integral to the project. 

To understand the needs of the community we will use the chatbot platform to actively engage with participants, seeking input from them about their needs and priorities, and we will use their feedback to inform decisions about the development of the platform. 

To help understand the needs of public health policymakers, we will leverage an existing collaboration with the Health Intervention and Technology Assessment Program (HITAP), a research unit under Thailand’s Ministry of Public Health who we are working with in a multinational project aimed at setting national antibiotic use targets. Through their international unit, HITAP have extensive experience of working with policymakers throughout the Southeast Asia region.

Our goal is to establish a sustainable, low-cost platform for gathering and monitoring community antibiotic consumption and antibiotic resistance data, initially focused on four countries in Southeast Asia. 

The solution will provide a number of public goods all of which will be globally accessible under fair, reasonable, non-discriminatory terms. 

First it will provide bias-adjusted estimates of community antibiotic consumption and resistance in the four countries, how this varies with space and time, and the drivers of this use. These will be made available in the form of open access publications which we will post on preprint servers and publish in open access peer-reviewed journals.

Second, we will make the analytical pipeline to obtain bias-adjusted estimates of community antibiotic use available by publishing our code under an open access licence as a public repository on GitHub. We will do the same for code to identify antibiotics from uploaded photographs and for the analytical pipeline to estimate antibiotic resistance prevalence using metagenomic data.

Third, we will make aggregated data collected by the platform freely and publicly available in standard formats through widely used data-sharing platforms. We anticipate that such data will be of value to the broader research community.

The primary impact will arise through better data to inform national and international decision-making. This will benefit entire populations of the countries we are working in. National action plans (NAPs) for reducing antimicrobial resistance have been developed by many countries, including Thailand, Vietnam, Myanmar, and Cambodia. Reducing inappropriate antibiotic consumption is a key component of all these NAPs, but the absence of reliable community antibiotic consumption data hinders efforts to determine appropriate reduction targets. This problem is compounded by a lack of microbiological testing data. By generating new data streams for antibiotic consumption and resistance, a tangible impact will be better evidence to inform the setting of and progress towards usage targets. Because we will actively seek to recruit participants in neglected populations, many of whom obtain antibiotics from unlicensed sources and have limited access to healthcare, the data collected will, after statistical adjustment for sampling bias, be more representative of the population as a whole than studies using official sales or hospital data alone. By collecting highly resolved community antibiotic usage and resistance data, the solution will also benefit the AMR research community by providing data to inform predictive models of how changes in antibiotic usage affect resistance.

Year-1

We will launch the chatbot and begin active recruitment. We are part of the Oxford Tropical Network who run a large-scale research programme including mature public engagement initiatives across all four countries; recruitment will leverage this network together with Oxford’s Global Health Network (tghn.org) which has >265,000 registered members and >28 million annual website visits, the majority from LMICs. We will also recruit through social media and online citizen-science platforms, in-person science communication events, and coverage in traditional media. The year 1 target is 5,000 active participants, and >500 in each country.

Year-2

We will promote the platform through community healthcare centres and in-person outreach events. We will introduce new features including badges, leaderboards, and informational content and use A/B testing to determine which increase retention and engagement. The year 2 target is 10,000 participants, and >1000 in each country.

Year-3

We will continue to target platform growth focusing on achieving good representation in marginalised communities including migrant workers and displaced people. The year 3 target is 20,000 participants, and >2000 in each country. 

We will develop plans to roll-out the platform in other countries in Asia and Africa, and expand the chatbot into other platforms to increase inclusiveness.

We plan to monitor impact through the following metrics:

• The number of active users of the platform both in total, and separately for each country.  We will consider any participant who enters information into the platform in a given month to be active.


• The mean number of monthly interactions with the platform

• The number of uploaded photographs of medication taken per month.

We will also measure the quality of the data provided to the platform and medication identification by assessing concordance between self-reported medication through the platform and confirmatory household survey data. This will be assessed comparing the antibiotic consumption rates collected through the platform and through the household survey.


We will also obtain qualitative feedback from participants about the platform to monitor user satisfaction and also actively engage with users for ideas to improve the platform and increase impact.

A  challenge in citizen science projects is building and maintaining high-level engagement. To address this, we will leverage existing networks and public engagement activities established by our colleagues in the Oxford network of research centres in Southeast Asia (https://www.tropicalmedicine.o...) and work closely with public engagement teams at these sites. We will also recruit participants  through Oxford’s widely-used online platform, The Global Health Network (https://tghn.org/), and using traditional and social media.

Political instability is a risk and for this reason it will not be possible to conduct the household survey in Myanmar until conditions stabilise. Working across four countries will mitigate risks to the overall project associated with such instability.

There are technical and cultural barriers to collecting representative antibiotic usage data that our solution is designed to overcome. These include:

• lack of public knowledge about which medications are antibiotics, which we will overcome using image capture and artificial intelligence to identify antibiotics;

• language barriers, which we will overcome by implementing the chatbot in all four languages; 

• Sampling bias, which we will overcome using a statistical approach that provides reliable population-level estimates even from highly non-representative samples.

To address the profound gap concerning  community antibiotic consumption and antibiotic resistance data in LMICs, we believe that highly original solutions are needed that are unlikely to be supported through traditional research funding avenues. Citizen science initiatives have been shown to be capable of generating reliable data in other areas of disease surveillance and we believe they could, with the right support, make a major contribution to filling this data gap for AMR. 

We are therefore applying to The Trinity Challenge because we share their belief in the need for highly original data-driven solutions making use of both novel data streams and analytical approaches. Also, as is evident from our track record of successfully delivering to resource-limited settings high impact initiatives to tackle AMR such as AMASS, we share a commitment to Trinity Challenge’s foundational principles of inclusivity, collaboration and innovation.

Through the partnership with the Trinity Challenge and the mentoring opportunities this would provide, we would aim to replicate our successes in AMR surveillance in hospitals, with an equally impactful initiative in the community. We would particularly value the opportunity to learn from the successes of Trinity Challenge awardees in strategies for growing platform engagement and involving marginalised communities.

We have identified the key sectors and networks who would be ideal partners to help accelerate and scale our solution and increase awareness and inclusiveness. Firstly, we are keen to form new collaborations with different members of the Challenge network, for instance, the Patrick J. McGovern Foundation on development of artificial intelligence engines to support medicine image recognition. We will reach out to the local organisations who are experienced in citizen science projects, for instance the Opendream (https://www.opendream.co.th/en...). Partnering with the local organisation will help to further improve the platform and ensure it is tailored to local needs and chatbot interaction patterns. Also, we will reach out to the ABACUS team (https://abacus-project.org/project/) and explore the potential in combining resources to enrich the medicine image database together. Additionally, we would be very keen to explore the potential for collaboration with The Behavioural Insights Team, a member of the Challenge network, to help accelerate our solution. Furthermore, we would like to reach out to the Strategic and Technical Advisory Group for Antimicrobial resistance (STAG-AMR) of the World Health Organization to ensure our solution synergises with global AMR surveillance initiatives.