Explore more Centres, Projects and Groups
Welcome Banner
A farmer stands in an indoor chicken coop, surrounded by chickens and reaching into a wooden hatchery

Selecting Efficient Farm-level Antimicrobial Stewardship Interventions from a One Health perspective (SEFASI)

Using participatory stakeholder elicitation, health economic modelling, and multi-criterion decision analysis to rank antimicrobial stewardship interventions from a One Health perspective in England, Senegal and Denmark.

Bottom Content

We are a consortium of JPIAMR funded researchers aiming to provide a ranking of farm-level interventions to control antibiotic resistance from a One Health perspective, taking into account human and animal health impacts and costs. 


The SEFASI project aims to create a range of outputs which can be used by future researchers and policymakers in the interest of open-source science.

About SEFASI 2 columns
About SEFASI 2 columns left paragraph

Selecting Efficient Farm-level Antimicrobial Stewardship Interventions from a one health perspective (SEFASI)

We are a consortium of JPIAMR funded researchers aiming to provide a ranking of farm-level interventions to control antibiotic resistance from a One Health perspective, taking into account human and animal health impacts and costs. 

We aim to do this by applying and expanding our Agriculture-Human-Health-MicroEconomic (AHHME) model, with its epidemiology and economic modules to three key settings: England, Senegal and Denmark. 

In January 2022, the “Selecting Efficient Farm-level Antimicrobial Stewardship Interventions from a One Health perspective” (SEFASI) project was started; with the London School of Hygiene (LSHTM), the International Livestock Research Institute (ILRI), the University of Copenhagen and the UK Health Security Agency (UKHSA) working together to achieve the overall aim of:

To provide a ranking of farm-level AMU interventions, across a One Health perspective, for different AMU and AMR scenarios, utilising evidence on the epidemiological and economic impacts of interventions at the farm, healthcare system and wider economic levels to evaluate the cost-effectiveness of interventions. This is within the context of Denmark, England and Senegal.

Our concept

We have developed an approach to analysing AMR interventions from a One Health perspective, which informs all SEFASI research.

This involves first considering the range of AMR-related interventions available, across human, animal and environmental health. The approach then considers which stakeholders are present across these sectors, as well as all of their objectives and interests. Then, we model the impact of each of these interventions on stakeholder objectives over time. Finally, these outcomes are weighted based on their importance to the various stakeholders, and the available interventions are ranked holistically using the weighted average of the outcomes.

View our infographic on economic evaluations of One-Health interventions.

Our work packages

SEFASI work packages diagram

The SEFASI project will be carried out across five stages, or Work Packages (WP). In the first (WP1), we will bring together experts and stakeholders from our three country settings, as well as globally, to form the SEFASI Knowledge Hubs. WP2 involves organising workshops with these stakeholders, and eliciting their opinions to establish evaluation criteria and to find parameter values for future analysis. In work packages 3-5, we will model the impact of AMR interventions on farms, model the transmission of AMR and the impact of interventions on human health, and finally use these outputs to model the holistic cost-effectiveness of the interventions considered, across the our three country settings.

Our team


Our Consortium Co-Directors (Dr Gwen Knight (PI) & Dr Nichola Naylor (Deputy PI)), and Institutional Principal Investigators (Dr Dagim Belay, Dr Michel Dione & Dr Javier Guitian) are working with colleagues across institutions and regions to deliver the goals of SEFASI. 

Also on the team are Eve Emes at the LSHTM and Ross Booton at Imperial College London. Joshua Aboah and Pape Faye at ILRI, and Yechale Endalew from University of Copenhagen, are also joining the SEFASI consortium to collaborate on the modelling efforts.

Outputs SEFASI 2 columns
Outputs SEFASI 2 columns left paragraph

The SEFASI project aims to create a range of outputs which can be used by future researchers and policymakers in the interest of open-source science. Aside from the creation of the SEFASI Knowledge Hubs, we aim to form a library of up-to-date literature on agriculture and AMR, and to make available the data used in our research so that it can be repurposed for future research. Among our most important outputs are the reusable models that we create as part of the project, and the journal articles that will be published as part of our research. 


How does antibiotic consumption in humans and animals link to AMR at the national level? And how best to estimate this relationship? - an opinion piece with recommendations for future research

This paper argues that we do not know enough about the relationship between antibiotic use and antibiotic resistance, at the population level, to inform health-economic decision making. We emphasise the need to estimate an actual number of resistant infections averted from a given change in national antibiotic use, and that this concern has been given insufficient attention in the literature. We highlight the potential of regression models using population-level data to bridge this gap, and outline a robust methodology for doing so in future research. 

[upcoming] Creating a better framework for understanding how antimicrobial use affects society across all sectors

This paper (in preprint) creates a theoretical framework which can be used to estimate the health-economic effect of antimicrobial use on all sectors of society. This includes all of the effects considered in AHHME (with some tweaks and improvements), as well as the indirect effects on population health via food security, economic security, and constrained healthcare resources. This framework proposes a mixture of statistical and mathematical models with much lower data requirements than previous models. It can be applied to a range of AMR issues outside of agriculture, and we hope to work with collaborators in Zambia and Malawi in the coming months to estimate the national impact of AMR in these countries using this framework. 

[upcoming] Cross-Sectoral Cost-Effectiveness of Interventions to Control Antimicrobial Resistance in Livestock Production - A One Health Modelling Exercise Using AHHME

This paper (in preprint) performs a range of modelling exercises using our Agricultural Human-Health Micro Economic (AHHME) model (see Models for information and resources on the model itself). We explore the hypothetical impact of reducing antibiotic use in livestock production in hypothetical countries of various income levels, exploring uncertainty in our parameters and assumptions. We also apply the model to the case of present-day Vietnam, simulating national-level rollout of an intervention that was trialled in 2020. Our results showed that national rollout of antimicrobial stewardship interventions in farms can provide a societal benefit equivalent to millions or billions of $USD per year, and highlighted that farm productivity and macroeconomic outcomes can be just as significant as the direct effect on human health.

[upcoming] country case study: drug-resistant Campylobacter infections and antibiotic use in English poultry farms

Campylobacter is a bacterium which infects humans and can cause debilitating gastrointestinal infections. Infections with Campylobacter have grown more prevalent over time in England, with surges during the summer months. It is thought that the bug reaches humans primarily through the consumption of domestically-produced poultry, with antibiotic-resistant strains being particularly dangerous. We plan to use recent changes in the antibiotic use patterns in English poultry farms as a natural experiment to see if these antibiotics contribute to the spread of such 'superbugs'. 

[upcoming] country case study: using Danish data to tease apart the relationship between agricultural antibiotic use and AMR in humans 

Denmark is a world leader in antibiotic stewardship, with programmes like the ‘yellow card initiative’ discouraging farmers from overusing and misusing antibiotics in livestock production. The wealth of data collected on antibiotics used in farms across Denmark, as well as on the incidence of antibiotic-resistant infections in humans, will allow us to investigate in detail how antibiotic use and antibiotic resistance relate to each other. Beyond simply establishing the existence of a connection between antibiotic use and AMR, we will use regression methods to look at the size and shape of this relationship. This will help us to understand the health-economic implications of each additional dose of antibiotics, and will inform cost-effectiveness analysis of future policies concerning antibiotic use in agriculture. 

[upcoming] country case study: what drives antibiotic use in poultry production? The case of Dakar and Thiès' peri-urban chicken farms

In order to manage or reduce antibiotic use in farms, we first must understand why farmers use antibiotics in the first place, and if there are any strategies which can reduce reliance on them. Better vaccination, better biosecurity, and better awareness of AMR issues have all been proposed as ways of reducing the need for antibiotics in farms. Using survey data from semi-intensive poultry farms in Dakar and Thiès, we investigate the extent to which these factors drive antibiotic use, and if they can help to mitigate any productivity loss from reducing antibiotic use. 

    Conference and workshop outputs

    SEFASI consortium outputs are regularly shared with researchers, policymakers and stakeholders, and engaging this participatory process is one of our main objectives. So far, this has meant that:

    • The framework presented in this preprint was presented and discussed at the JPIAMR New Perspectives on Bacterial Drug Resistance workshop on 9 June 2022
    • From 26-27 September 2022, the SEFASI consortium hosted a workshop in Dakar where we presented our ongoing work to a range of researchers, policymakers and stakeholders from across the One Health - AMR spectrum, starting a two-way conversation with stakeholders about priority-setting in AMR research and policy

    The SEFASI GitHub Page

    The SEFASI GitHub page is currently under construction, and when completed will contain all coding resources from the numerous projects which make up SEFASI.

    The Agriculture Human Health MicroEconomic (AHHME) Model

    The purpose of the model

    Antimicrobial stewardship policies in agriculture affect every sector of society. Reducing antibiotic use in farms may reduce farm productivity and hurt farmers’ incomes, and this effect could be mitigated by improving farming practices in tandem. Implementing these interventions will incur a cost to the public sector. If reducing antibiotic use in agriculture reduces the number of people infected by drug-resistant bacteria, then this will prevent illness. Few illnesses also means fewer long-term sequelae, and fewer deaths. This, in turn, reduces the burden on the healthcare system, and means that fewer people are left unable to work (both in paid and unpaid work). 

    The Agriculture-Human-Health-Micro-Economic (AHHME) model takes all of these outcomes into account, and compares them in like terms. This gives us the cross-sectoral cost-effectiveness of hypothetical antimicrobial stewardship policies in agriculture. Depending on the information available, this can take the form of net monetary benefit of a proposed policy, or a threshold price (the maximum amount that we should be willing to pay to implement that policy).

    Applications of the model

    The AHHME model is flexible to a number of different methodological assumptions, types of intervention, and country characteristics. By customising these parameters, the model can be applied to any antimicrobial stewardship policy in any setting, and with various scopes of analysis.

    In an upcoming paper, we apply the model to a range of hypothetical antimicrobial use interventions in high-, middle- and low-income contexts (see ‘Publications’ for information on this and other SEFASI outputs). 

    On our public GitHub repository, we have included the code used to create and run the AHHME model in R. It also includes an example of how to adapt the model to any context and purpose. Our hope is that researchers and policymakers can use AHHME to model the cost-effectiveness of prospective agricultural antimicrobial stewardship policies in their own countries. Comparing outcomes in like terms across sectors will allow AMR policies to be compared to other, unrelated policy alternatives, allowing AMR to be correctly prioritised on the national policy agenda.

    We have also created an interactive Shiny App which allows users to edit model inputs in a more straightforward way. This is intended to allow researchers to familiarise themselves with the model, and for exploratory analysis of potential interventions.

    Knowledge Hub
    Knowledge Hub SEFASI 2 columns
    Knowledge Hub SEFASI 2 columns left paragraph

    Background and rationale

    SEFASI workshop, Dakar
    September 2022 SEFASI Workshop in Dakar

    Stakeholder participation has been recommended in the evaluation of interventions tackling One Health issues (Naylor, 2020). This includes stakeholders from across the One Health and antimicrobial resistance (AMR) system; from farmers, clinicians, veterinarians, ministers of sectors and/or international organisations, the pharmaceutical industry and the general public. 

    The overarching goal of the SEFASI Knowledge Hub is therefore to bring together a group of experts and stakeholders from across the One Health - agriculture - AMR spectrum, and to elicit their expertise with the aim of informing multi-criterion decision analysis which will allow us to rank and prioritise AMR policies.

    Knowledge Hub aims and research questions

    The aims of the Knowledge Hub are:

    To utilise knowledge hubs for the expert elicitation of ranking of the importance of: 

    1. the interventions, AMR prevalence and agricultural system scenarios of greatest interest, in relation to farm-level AMU, 
    2. the impact outcome measures of most importance, 
    3. stakeholder preferences in terms of the ranking of such outcome measures and 
    4. missing input parameters critical for epidemiological and economic modelling of relevant interventions across One Health systems, within the three country case studies. 

    Knowledge hub research questions are:

    What do stakeholders believe are: 

    1. the interventions and AMR and One Health scenarios of greatest interest? 
    2. the evaluation criteria of most importance? 
    3. their preferences in terms of the weighting of such outcome measures? and 
    4. missing input parameters critical for epidemiological and economic modelling of relevant interventions across One Health systems?

    Our methods

    Multi-criteria decision analysis (MCDA) is a formal process in which stakeholder preferences for the importance of different measures of intervention impact are used to weight impact estimates (e.g. cost-effectiveness of one intervention compared to others against equity impacts of intervention implementation) to allow for a final ranking of interventions. MCDA approaches have been used in zoonotic disease fields and AMR separately. These processes can (i) improve end-user use of research outputs and (ii) utilise the knowledge resource of stakeholders in understanding the importance of evaluation measures and unknown parameters in a formal manner.

    Additionally, expert elicitation for the parameterisation of models where data are missing and/or need more context (for example for microbes where surveillance data are missing or for predicting pandemic-AMR scenarios) can be used in the parameterisation of infectious disease and economic evaluation models. Previously experts have been utilised to understand AMR and its burden, focusing more on AMR-burden data needs and quality. 

    In SEFASI, we are proposing to bring together groups of experts and stakeholders, across the One Health arena, in the form of “Knowledge Hubs”. We will develop membership of these Hubs through (i) local networks within the three participating countries, (ii) mapping of One Health areas and active determination of experts in each needed arena in each geographical area and (iii) approaching international organisations.

    Initiation of Knowledge Hub

    • Round 1: Intervention, scenario and outcome prioritisation
    • Round 2: Missing parameter elicitation
    • Round 3: Multi-criteria decision analysis
    Settings SEFASI 2 columns
    Settings SEFASI 2 columns left paragraph

    The SEFEASI Consortium are conducting our research in three countries: England, Senegal, and Denmark.


    Area lead

    Dagim Belay

    One Health AMR situation

    Denmark has been monitoring both AMU and AMR in animals, food and humans since 1995 (Hammerum, 2007) and has been a pioneer in developing, implementing and assessing the impact of interventions to curb AMU with an early focus on animal populations. Interventions to reduce fluoroquinolone (2002, 2003) and cephalosporin (2010) use in food-producing animals, alongside systems to monitor antibiotic use in pig-production (Yellow Card in 2010), resulted in a marked reduction in the use of antibiotics across pig farms between 2010-2017 (Antunes, 2020). Denmark has combined this low use with being one of the largest pig producers in the world with a production of up to 32 million pigs a year, which is highly intensified and highly productive (DAFC, 2021). We expect that this country will have most agricultural sector evidence available for the development of our models: decades of reliable surveillance data, including antibiotic use at animal level (i.e., DANMAP reports), several research study publications and at least one case study conducted by FAO to explore the effectiveness of different interventions to curb AMU in Danish pork production (FAO, 2019). The focus on “food, food animals and humans” of the Danish action plan on AMR suggests that the environmental aspect may be less well analysed, but the Danish championing of a One Health approach gives confidence that data will be available going forward.

    Upcoming work

    As part of the SEFASI project, we aim to publish a regression analysis which uses Danish surveillance data to characterise the relationship between antibiotic use in different species and antibiotic resistance in humans at the population level. See our Publications section on the Outputs page to find out more.


    Area lead

    Javier Guitian, Royal Veterinary College

    One Health AMR situation

    England has been a pioneer in the use and open-sharing of action plans and data on AMR prevalence within human infections (HMG, 2019), with National Action Plans since 2013. This impetus was reflected in the UK Prime Minister commissioning a review in 2014 of the global state of AMR (O’Neill, 2016). The UK has a well developed dairy and beef cattle sector. It is self-sufficient for beef production with approximately 9.36 million cattle contributing approximately £2.8 billion to the UK economy (NBA, 2021). Following the publication of the O’Neill report on AMR in 2016, a Task Force was created by the Responsible Use of Medicines in Agriculture Alliance (RUMA) to set up targets for reduction of overall and highest-priority Critically Important Antibiotics (HPCIA) use across different livestock sectors (RUMA, 2020). Led by the livestock industry through education and behaviour change interventions, a series of indicators are in place to monitor efficacy of these interventions on levels of AMU, adoption of best practices and impact on animal welfare.

    The use of antibiotics at animal and farm levels are still unavailable for the whole of the national herd but data are available for antibiotic sales. These show that dramatic reductions (>40%) have been observed in the livestock sector between 2015 and 2019 (UK-VARSS, 2020), firstly in poultry, then pig and more recently in cattle. Sales of injectable and topical intramammary HPCIAs antibiotics for use in cattle in the UK has decreased considerably in recent years from 1.1 mg/kg (2015) to 0.26 mg/kg (2019) and from 0.33 mg/kg (2015) to 0.03 mg/kg (2019) representing a reduction of 77% and 91% of sales of these types of formulations, respectively. Further reduction targets have been established for the period of 2021-2024 (UK-VARSS, 2020). On the human side, data on antimicrobial consumption is available at the primary case and secondary care level from 2016/7 onwards (PHE, 2021), with national level data available from 2010 through the English surveillance programme for antimicrobial utilisation and resistance (ESPAUR) reports. As antibiotics are not used on crops, there is little selective pressure in crop agriculture (Haynes, 2020). England now has a framework for understanding environmental antimicrobial resistance led by the Environment Agency (Singer, 2020) which has collated all datasets that could provide an environmental AMR risk assessment for consultation.

    Upcoming work

    As part of the SEFASI project, we aim to publish a regression analysis which investigates the link between antibiotic use in English poultry production and the rate of antibiotic resistance in Campylobacter infections at the population level. See our Publications section on the Outputs page to find out more.


    Area lead

    Michel Dione

    One Health AMR situation

    Senegal is a leader in the African continent in terms of its approach to AMR control. It has had a National Program for the Control of Nosocomial Infections (PRONALIN) in place since 2004 (Ndoye, 2009) which has been effective in reducing infections and the transmission of multidrug-resistant bacteria associated with hospital care and a Directorate of Laboratories which set up in 2012 a national system for monitoring antibiotic resistance through laboratories (RNL/SN, 2013). Building on these, Senegal has adopted a wider "One Health" strategy, in support of the global health security agenda, to strengthen integration and intersectoral cooperation in the prevention, detection and response to threats of emerging infectious diseases, including zoonoses and AMR (Senegal Government, 2016). Specifically, a National Multisectoral Antimicrobial Resistance Surveillance and Control Action Plan (2018-2022) was prepared by the Ministry of Health and Social Action (Laboratories Directorate) in 2017 (Coll-Sec, 2020), with technical and financial support from WHO and FAO. Moreover, Senegal has the One Health National Platform composed of all actors involved in the control of zoonotic diseases and AMR including the 3 ministerial pillars of One Health, the private sector, the national and regional research institutions and development organisations. This platform is already supporting an ongoing initiative using a One Health approach to tackle AMR with FAO (FAO, 2020) and the Fleming Fund (2021). However much of the data needs to be aggregated and analysed.

    The poultry population is estimated at 64,541,000 heads in 2016, an increase of 7.7% compared to 2015. This increase is attributable to the 11.0% increase in industrial poultry. In 2019, the turnover of the poultry sector is estimated at at least USD 0.4 billion. The share of industrial poultry has increased considerably to supplant beef in 2019: taking into account family poultry, which is estimated at about 13% of per capita consumption, makes poultry meat the first meat consumed per capita in Senegal, or 44% of the total (Ly, 2020). Most intensive poultry farming happens in peri-urban areas where animals receive a large amount of drugs such as antibiotics to fight diseases and boost growth, and compliance with the biosecurity norms of poultry production remains a challenge (Biagui, 2002). In addition, studies conducted on cattle, sheep and/or poultry meat have revealed the presence of antibiotic residues in their products; but the risk to humans has never been evaluated in such settings (Aambedji, 2004).

    Upcoming work

    As part of the SEFASI project, we aim to publish a regression analysis which investigates the determinants of antibiotic usage in peri-urban semi-intensive poultry farms in Dakar and Thiès. See our Publications section on the Outputs page to find out more.

    Also in the pipeline, consortium member Joshua Aboah is planning to use stakeholder inputs to create a system dynamic model of AMR and poultry production in Senegal, similar to previous work based in Ghana. This modelling project will allow us to understand and model how farmers' behaviour interacts with micro- and macro-level variables, both in terms of production decisions and antimicrobial stewardship.

    In collaboration with Interstate School of Veterinary Medicine of Dakar and The Ministry of Livestock, SEFASI project investigates the Knowledge Attitude and Practices of chicken farmers in peri-urban areas of Dakar and Thiès in Senegal.


    News SEFASI 2 columns
    News SEFASI 2 columns left paragraph
    • 26th -27th September 2022:  SEFASI technical inception workshop for Senegal
    • 25th August 2022: SEFASI update meeting - discussion of data sharing and current status of knowledge hub protocols and website
    • 28th July 2022: SEFASI update meeting - discussion of how to integrate Josh's system dynamics modelling into the AHHME framework. 
    • 22nd June 2022: SEFASI update meeting - lots of discussion after a presentation from Ross Booton of his mathematical modelling across One Health environments
    • May 2022: First SEFASI affiliated publication from Dagim on "Does restricting therapeutic antibiotics use influence efficiency of pig farms? Evidence from Denmark’s Yellow Card Initiative". Congratulations Dagim! 
    • 5th May 2022: SEFASI update meeting - Michel presented data and analysis of the current situation in Senegal. 
    • 21st April 2022: SEFASI update meeting - Dagim presented his analysis and data of work in Denmark. 
    • 10th Feb 2022: SEFASI update meeting - lots of interesting discussion as to current status and ideas for discussion piece and next meeting 
    • 17th Jan 2022: SEFASI kick-off meeting held virtually: AHHME model presentation and discussion
    • 1st Jan 2022: SEFASI officially starts