AMR-Intervene: a social–ecological framework to capture the diversity of actions to tackle antimicrobial resistance from a One Health perspective

Abstract

The global threat of antimicrobial resistance (AMR) requires coordinated actions by and across different sectors. Increasing attention at the global and national levels has led to different strategies to tackle the challenge. The diversity of possible actions to address AMR is currently not well understood from a One Health perspective. AMR-Intervene, an interdisciplinary social–ecological framework, describes interventions to tackle AMR in terms of six components: (i) core information about the publication; (ii) social system; (iii) bio-ecological system; (iv) triggers and goals; (v) implementation and governance; and (vi) assessment. AMR-Intervene provides a broadly applicable framework, which can inform the design, implementation, assessment and reporting of interventions to tackle AMR and, in turn, enable faster uptake of successful interventions to build societal resilience to AMR.

Background

The natural phenomenon of antimicrobial resistance (AMR) is amplified by the increasing use of antimicrobials in humans, animals and crops, and by global connectivity through travel and trade. Growing demand for food products and human-caused environmental change (e.g. man-made water constructions that lead to an imbalance of the environment and amplify vectors for pathogens) increases society’s reliance on antimicrobials. Yet our expanding, but still limited, understanding of the impacts of human activities on the biosphere suggests that, like pollutants, there are limits to the level of antimicrobials that we can use while maintaining antimicrobial susceptibility in the microbiome.¹ Addressing a ubiquitous problem of evolutionary complexity, such as AMR, requires innovation in terms of governance, policies and practices at both national and global levels.² Growing understanding of the epidemiology, drivers and impacts of AMR, including the spread of resistant organisms and genes and antimicrobial residues between food animals, humans and ecosystems, shows that AMR requires an integrated perspective of humans and animals in their shared environments (Figure 1).

Figure 1.

Social–ecological system of antimicrobial resistance (AMR) in a One Health context. The figure is organized into two parts: (i) the human system and its governance of AMR and use; and (ii) the ecosystem of AMR. This figure highlights the complexity of the AMR social–ecological system and the diversity of potential AMR interventions. The ecosystem part has been adaped from Chereau et al.⁸⁴ and icons are designed by Arthur Shlain, downloaded from the Noun Project (https://thenounproject.com). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

With efforts ongoing at both the national and international levels,³ there is a need to improve our understanding of which actions work under which conditions with the overall goal of building resilience to AMR.⁴ Resilience broadly refers to the capacity of social–ecological systems to respond and adapt to often surprising change, while continuing to maintain key functions.⁵ From a One Health perspective, resilience to AMR means the capacity of human societies and their subcomponents (e.g. people, hospitals, farms, cities and countries) to provide quick and effective responses for critical healthcare benefits, ensure food security and safeguard the environment. While resilience depends on the state of the system (e.g. typically slow variables such as the strength of institutions, human and financial resources), human action to tackle AMR can enhance or reduce resilience. The capacity to assess resilience then relies on our capacity to understand the ability of human actions to impact the system in which they take place and the drivers of success and failure of these actions.⁶

An intervention corresponds to any coordinated action driven by a social group on a targeted population in a bio-ecological context that interferes with the outcome or course of a difficult situation in order to improve it or prevent it from getting worse.^7–12 Recent years have seen a diversification of strategies and interventions to tackle AMR.¹³ In several areas, efforts have been made to describe and evaluate these interventions.^10,14–19 However, there remain important gaps regarding the design, implementation, assessment and reporting of AMR interventions from a One Health perspective. Based on insights from the health, social, sustainability and implementation sciences, as well as recent efforts to develop integrated approaches for the evaluation of One Health initiatives,^20,21 this article introduces the AMR-Intervene framework as a way to: (i) systematically characterize the broad range of interventions that affect AMR and make sense of their diversity (Figure 2); (ii) describe how they are implemented; and (iii) evaluate success factors, limitations and obstacles related to each intervention type.

Figure 2.

Mindmap of the framework for the evaluation of interventions to tackle antimicrobial resistance (AMR) based on resilience principles. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Building and validation of the AMR-Intervene framework

The AMR-Intervene framework is composed of categorical and numerical variables, grouped into five main components (described in the ‘Main components of the framework on AMR interventions’ section), where the variables are the relevant pieces of information about each intervention. The framework and its list of variables were developed both inductively and deductively by the authors (hereafter called the AMR resilience consortium)—a group of researchers with diverse disciplinary backgrounds including human medicine, animal health, environmental sciences and social sciences. We used reviews, reports and assessments of AMR and One Health interventions in the published literature to develop a starting list of variables,^4,14,21–24 which allowed us to avoid duplication of effort and to cover the broad range of interventions regarding AMR. Details on the scientific publications used to identify variables are given in the ‘Main components of the framework on AMR interventions’ section and Tables 1–5 (which provide definitions of the variables). For the identification of variables related to the drivers of interventions (‘Triggers, goals and challenges of the intervention’ section) and their governance (‘Governance and implementation of the interventions’ section), we performed a search of the (health) policy literature and drew on the Driver, Pressure, State, Impact, Response (DPSIR) framework.²³

Table 1.

Variables of the AMR-Intervene framework in the building block ‘Social system of the intervention’

Variables	Description
Sector – occupational area where the different activities of the intervention take place
What is(are) the sector(s) of implementation of the intervention?	Please specify the sector(s) (i.e. professional area where the different activities of the intervention take place) covered by the different activities (i.e. specific organized function) of the intervention. One intervention can cover several sectors. Animal: all living animals Environment: soil, water and air, all in the natural environment Human: organizations and individuals involved in health products and services Food: production or consumption of food of animal and plant origin (i.e. all steps after slaughter of the animals or all steps after harvest of plants) Plant: culture of plants and crops before harvesting And/or ‘other’ (to specify)
Please specify the subsector(s) of the intervention (further units of demarcation within a sector)	[If Sector = ‘human’] For the ‘human’ sector, several subsectors can be defined and detailed even within the same intervention: General Hospital healthcare delivery Medical and dental practice healthcare delivery Residential healthcare delivery Community care: outside the healthcare facilities (e.g. schools, nursing homes, public space) Healthcare industry: private actors that provide technologies and/or services (e.g. insurance) And/or ‘other’ (to specify)
	[If Sector = ‘animal’] For the ‘animal’ sector, several subsectors can be defined and detailed even within the same intervention: General Livestock: places where terrestrial animals are farmed, including farms, fields, orchards and pastures Aquaculture: farming of freshwater or marine fish, crustaceans, molluscs, aquatic plants, algae and other aquatic organisms Wild animals: animals that have not been domesticated or tamed, and living in a natural environment Pets: animals kept for a person’s company, protection or entertainment And/or ‘other’ (to specify)
	[If Sector = ‘environment’] For the ‘environment’ sector, several subsectors can be defined and detailed even within the same intervention: General Water: NB: it also includes water from sewage treatment, which is the process of removing contaminants from municipal wastewater Air Soil: intervention aiming to reduce AMR in soil, including manure And/or ‘other’ (to specify)
	[If Sector = ‘food’] For the ‘food’ sector, several subsectors can be defined and detailed even within the same intervention: General Meat products Food from vegetal origin Seafood Dairy products And/or ‘other’ (to specify)
	[If Sector = ‘plant’] For the ‘plant’ sector, several subsectors can be defined and detailed even within the same intervention: Fruits Vegetables Crops And/or ‘other’ (to specify)
Social group – group of people that are the primary target of the intervention
Please specify the social group(s) of the intervention	[If Sector = ‘human’] If the sector of the intervention is ‘human’, the social group(s) of the intervention can be defined in several categories: No specific social group Prescribers – Medical doctors: a professional who practises medicine (e.g. physician, medical doctor) Prescribers – Dentist Nurses: profession focusing on the care of individuals, families and communities Dispensers – Pharmacists: preparation, packaging, labelling, record keeping and transfer of a prescription drug Dispensers – ‘other’ (to specify) Patients: a person receiving or registered to receive medical treatment And/or ‘other’ (to specify)
	[If Sector = ‘animal’] If the sector of the intervention is ‘animal’, the social group(s) of the intervention can be defined in several categories: No specific social group Farmers Veterinarians: professional practising veterinary medicine by treating, e.g. diseases in animals Feed producer Slaughterhouse workers Pet owners And/or ‘other’ (to specify)
	[If Sector = ‘food’] If the sector of the intervention is ‘food’, the social group(s) of the intervention can be defined in several categories: Processors: people in charge of preparing food for sale Retailers: people working in places where food products are sold for human consumption (e.g. grocery stores, restaurants) Consumers
Settings – physical places where the intervention takes place
Please specify the setting(s) of the intervention (i.e. physical places where the intervention takes place)	[If Sector = ‘human’] If the sector of the intervention is ‘human’, the setting(s) where the intervention takes place can be defined in several categories: No specific social setting Outpatient care: place where ambulatory care is delivered such as general practice, pharmacy, etc. Hospitals: include both small and larger hospitals such as university hospitals Schools/universities Nursing homes Prison Pharmaceutical companies: private companies responsible for the research and development of new drugs and related technologies Health insurance And/or ‘other’ (to specify)
	[If Sector = ‘animal’] If the sector of the intervention is ‘animal’, the setting(s) where the intervention takes place can be defined in several categories: No specific social setting Farms Slaughterhouse, i.e. place where animals are slaughtered Pharmaceutical industry Zoo And/or ‘other’ (to specify)
	[If Sector = ‘food’] If the sector of the intervention is ‘food’, the setting(s) where the intervention takes place can be defined in several categories: No specific social setting Food factory: place where food is prepared for sale/consumption Supermarkets and retailers: places where food products are sold for humans Restaurants And/or ‘other’ (to specify)
	[If Sector = ‘plants’] If the sector of the intervention is ‘plants’, the setting(s) where the intervention takes place can be defined in several categories: No specific social setting Farms And/or ‘other’ (to specify)
	[If Sector = ‘environment’] If the sector of the intervention is ‘environment’, the setting(s) where the intervention takes place can be defined in several categories: Sewage plants Surface water (e.g. river, lake, wetland, ocean) Open landfill And/or ‘other’ (to specify)
Time dimension
When did the intervention start?	Please provide the year of implementation of the intervention.
When did the intervention end?	Please provide the year when the intervention ended. If the intervention is still ongoing, please mention ‘not applicable’.
Which period of the intervention was described/assessed in the publication?	Please give the period of time that is described and/or assessed in the publication.
At the time of assessment, was the intervention already completed or still ongoing?	Please specify the status of the intervention at the time of assessment of the intervention: final, ongoing or ‘not applicable’.
Spatial dimension
Please specify the country(ies) where the intervention is implemented	Enter the country(ies) involved in the intervention. If the intervention is global/international, please select the option ‘all countries’.
Socio-political space of implementation
Please specify the level of development of the country(ies) involved in the intervention	The categories of level of development of country(ies) are: Low income country: GNI per capita of US$1005 or less in 2016; Lower middle-income country: GNI per capita between US$1006 and US$3955 in 2016; Upper middle-income country: GNI per capita between US$3956 and US$12 235 in 2016; High income country: GNI per capita of US$12 236 or more in 2016
Please specify at which socio-political scale(s) is the intervention implemented	The socio-political scales of implementation are categorized as: Local: district, city or towns Subnational: state or province National: country Regional: several countries in the same region International: countries for different regions or continents

Table 2.

Variables of the AMR-Intervene framework in the building block ‘Bio-ecological system of the intervention’

Variable	Description
Microorganism(s) targeted by the intervention
Please specify the type of microorganism(s) targeted by the intervention	Microorganisms are microscopic organisms and are separated into the following categories: Viruses Bacteria Parasites Fungi And/or ‘other’ (to specify) No specific organism: interventions that do not target any kind of pathogen in particular
Please specify the subtype(s) of the microorganism(s) targeted by the intervention	[If Microorganism = ‘bacteria’] If the microorganism concerned by the intervention is ‘bacteria’, please specify the type of bacteria according to the following categories: No specific bacteria Gram-negative – Enterobacteriaceae: Enterobacteriaceae are a large family of Gram-negative bacteria including Salmonella, Escherichia coli, Yersinia pestis, Klebsiella, Shigella, Proteus, Enterobacter, Serratia and Citrobacter Gram-negative – Other Gram-positive: Gram-positive bacteria have a thick cell wall made of peptidoglycan Mycobacteria: mycobacteria has over 190 species including frequent pathogens in humans and animals And/or ‘other’
	[If Microorganism = ‘viruses’] If the microorganism concerned by the intervention is ‘viruses’, please specify the type of virus according to the following categories: HIV Influenza And/or ‘other’
	[If Microorganism = ‘parasites’] If the microorganism concerned by the intervention is ‘parasites’, please specify the type of parasite according to the following categories: Malaria And/or ‘other’
	[If Microorganism = ‘fungi’] If the microorganism concerned by the intervention is ‘fungi’, please specify the type of fungi according to the following categories: Candida And/or ‘other’
Please specify the full name of the pathogen(s) targeted by the intervention (if applicable)	The full name of the pathogen(s) should be written using the binomial system such as Staphylococcus aureus, Streptococcus pneumoniae (standard international denomination)
Please specify the pathogen strains in more detail, if available
Resistance of the microorganism targeted by the intervention
Please specify the level of resistance of the targeted microorganism	The present categorization of level of resistance is based on the work of Magiorakos et al.37 Simple resistance: one class Multidrug-resistant (MDR): acquired non-susceptibility to at least one agent in three or more antimicrobial categories Extensively drug-resistant (XDR): non-susceptibility to at least one agent in all but two or fewer antimicrobial categories Pan-drug-resistant (PDR): non-susceptibility to all agents in all antimicrobial categories Not applicable
Please specify the main resistance to antimicrobial class(es) targeted by the intervention	[If Microorganism = ‘bacteria’] If the microorganism targeted by the intervention is ‘bacteria’, please specify the main antimicrobial classes of resistance: β-Lactams: broad-spectrum antibiotics that contain a β-lactam ring in their molecular structures Aminoglycosides: different subclasses inhibiting cell wall biosynthesis, e.g. gentamicin, amikacin, neomycins Glycopeptides: class of microbial origin composed of glycosylated cyclic or polycyclic non-ribosomal peptide Macrolides Oxazolidinone Polymyxins Sulphonamides Tetracyclines Quinolones Amphenicols And/or ‘other classes’
Please specify the antimicrobial subclass(es) targeted by the intervention	[If Antimicrobial class = ‘β-lactam’] If the main resistance to antimicrobial class(es) targeted by the intervention is ‘β-lactam’, please specify the subclass of the antimicrobial: Penicillins: penicillin G/V, methicillin, oxacillin, amoxicillin Cephalosporins C1G: ampicillin, cefadroxil Cephalosporins C2G Cephalosporins C3G: ceftriaxone, ceftazidime Cephalosporins C4G + C5G: cefepime, ceftaroline Carbapenems: imipenem, doripenem, ertapenem Monobactams: aztreonam
	[If Antimicrobial class = ‘polymyxins’] If the main resistance to antimicrobial class(es) targeted by the intervention is ‘polymyxins’, please specify the subclass of the AM: Polymyxin B Colistin
	[If Antimicrobial class = ‘quinolones’] If the main resistance to antimicrobial class(es) targeted by the intervention is ‘quinolones’, please specify the subclass of the AM: First generation: nalidixic acid, pipemidic acid Fluoroquinolones (second, third and fourth generation): ciprofloxacin, levofloxacin, moxifloxacin
	Other (to specify)
Does the intervention target well-known pathogen-drug combination(s)?	Please select the targeted pathogen–drug combination(s) by the intervention in the following list: No specific pathogen–drug combination [If Microorganism= ‘bacteria’] MRSA [If Microorganism = ‘bacteria’] MRSP – methicillin-resistant Staphylococcus pseudointermedius [If Microorganism = ‘bacteria’] VRSA – vancomycin-resistant Staphylococcus aureus [If Microorganism = ‘bacteria’] VISA – vancomycin-intermediate Staphylococcus aureus [If Microorganism=’bacteria’] ESBL – extended-spectrum β-lactamase breaks down penicillins and cephalosporins [If Microorganism = ‘bacteria’] CRE – carbapenem-resistant Enterobacteriaceae [If Microorganism = ‘bacteria’] VRE – vancomycin-resistant Enterococci [If Microorganism = ‘bacteria’] MDR-TB – tuberculosis that does not respond to at least isoniazid and rifampicin [If Microorganism = ‘bacteria’] XDR-TB – MDR-TB with resistance to any fluoroquinolones and at least one injectable second-line drug [If Microorganism = ‘bacteria’] plasmid-mediated AmpC non-chromosomic [If Microorganism = ‘parasite’] artemisinin-resistant Plasmodium falciparum And/or ‘other’ (to specify)
What is the molecular mechanism of resistance of the targeted pathogen(s)? (if available/applicable)	Please describe clearly the molecular mechanism for resistance, e.g. blaKPC2, MCR-1
Please specify the resistance genotype(s) if available/applicable
Does the intervention target a resistance coming from plasmids?	AMR can be transmitted by the transfer of antibiotic resistance genes carried on plasmids. The plasmids can be transferred between bacteria within the same species or between different species via conjugation
Host population or substrate, i.e. biological organisms or food/ecological substrate where the intervention is focusing and where the targeted microorganisms carrying AMR are observed.
Please specify the host population(s) or substrate(s) targeted by the intervention	[If Sector = ‘human’] If the sector of the intervention is ‘human’, the potential host population(s) or substrate(s) targeted by the intervention can be defined in several categories: No specific host population General population: no specific target with the population, for all citizens (e.g. awareness campaign) Children: humans between birth and puberty (less than 18 years old) Newborn Adults Elderly Immunocompromised people Migrants/refugees Ethnic group And/or ‘other’ (to specify)
	[If Sector = ‘animal’] If the sector of the intervention is ‘animal’, the potential host population(s) or substrate(s) targeted by the intervention can be defined in several categories: Poultry Bovine Porcine Equine Ovine Birds Dogs Cats Farmed fishes And/or ‘other’ (to specify)
	[If Sector =’ food’] If the sector of the intervention is ‘food’, please specify the potential host population(s) or substrate(s) targeted by the intervention.
	[If Sector = ‘plant’] If the sector of the intervention is ‘plant’, please specify the potential host population(s) or substrate(s) targeted by the intervention.
	[If Sector = ‘environment’] If the sector of the intervention is ‘environment’, please specify the potential host population(s) or substrate(s) targeted by the intervention.
What is the type of disease(s)/ syndrome(s) targeted by the intervention?	Please select the most relevant disease(s)/syndrome(s) targeted by the intervention if available/applicable: No specific diseases or syndrome Bloodstream infection: sepsis and bacteraemia Bone and joint infections Cardiovascular infections Central nervous system infections Digestive tract infections: foodborne infection, gastroenteritis Eye infections Intra-abdominal infections Respiratory tract infection (upper): pharyngitis, laryngitis, sinusitis, epiglottitis, oral cavity Respiratory tract infection (lower): pneumonia, bronchitis, bronchiolitis, lung abscess Sexually transmitted diseases: e.g. gonorrhoeae, syphilis, HIV-AIDS Skin and soft tissue infections Urinary tract infection: cystitis, bladder infection, pyelonephritis And/or ‘other’ (to specify)
What is the origin of infection targeted by the intervention?	Please select the most appropriate origin of infection among the following list: Community acquired Healthcare-associated infection: catheter-related infections or surgical site infections Not applicable
Ecology of resistance
Between which categories is the AMR targeted by the intervention transmitted?	Please select the most appropriate sequence of transmission concerned by the intervention: Unknown Human to human transmission Animal to human transmission Human to animal transmission Animal to animal transmission Environment to human transmission Environmental to animal transmission
What is(are) the reservoir(s) of the AMR targeted by the intervention?	Please select the most appropriate reservoir(s) of the AMR targeted by the intervention: Animal reservoir Human reservoir Environmental reservoir
What is the climate of the place(s) where the AMR is present/studied and the intervention implemented?85	Please select the climate of the location targeted by the intervention (i.e. where the intervention is implemented) and where the AMR is present: Tropical Arid Temperate Cold Polar

Table 3.

Variables of the AMR-Intervene framework in the building block ‘Triggers and goals of the intervention’

Variable	Description
Triggers of the intervention
What is(are) the main trigger(s) of the intervention?	Please select the trigger(s) of the intervention in the following list:23 Driver: e.g. an increase in AMU in animal production Pressure: e.g. an increase in antimicrobial use State: e.g. increase in the prevalence of AMR Impact: e.g. an increase of mortality or morbidity
If necessary, please provide a comment on the selected trigger(s) of the intervention	This free text cell aims to capture information about the trigger(s) in more detail, e.g. ‘on farm disease levels in calves’ or if the intervention is triggered by public reaction/fear. A list of key drivers can be pre-populated from systems map from Majowicz et al.86
Please specify if the intervention implemented preventive or reactive activities regarding its trigger(s)	Please select the most appropriate description of the intervention regarding its activities: Preventive: the intervention is set up to anticipate a problem that has not manifested itself in the setting but is expected to happen Reactive: the intervention is set up to cope with something that has already happened
Goals of the intervention
What is the purpose of the intervention regarding the context of the intervention and AMR issue?	Please select the most appropriate description of the goals of the intervention regarding the AMR issue and its own context: Initiate: a new action to tackle an issue Maintain: a follow-up action to maintain the outcome of a successful intervention Improve: add a new component/action, change the intervention when unexpected or unsuccessful result, solve practical implementation challenge Not applicable
What is the type of leverage point targeted by the intervention activities?	A leverage point is considered as a place in the whole AMR system where a change due to the intervention may lead to changes in the system as a whole. Leverage points might be separated in different categories depending on the potential impact they might have on the AMR system.55 We proposed a simplified version of the categorization: Low leverage point: it is targeted by standard interventions that focus on ‘the relatively mechanistic characteristics [of AMR] typically targeted by policymakers’ including ‘parameters, the size of buffer stocks, relative to the structure of material stocks and flows, and the length of delays’, e.g. reduce the use a specific antimicrobials. Also, a low leverage point focuses on ‘the interaction between elements within a system of interest that drive internal dynamics’. e.g. all interventions that aim to reduce antimicrobial use High leverage point: it is a point in the AMR chain targeted by deeper intervention that aims to fundamentally change how the system works (goal, worldviews, values). For example, ‘The social structures and institutions that manage feedbacks and parameters’ and ‘The underpinning values, goals and world views of actors that shape the emergent direction to which a system is oriented’, e.g. reduce meat consumption or change meat production system
What is(are) the main goal(s) of the intervention, i.e. its main target to fight AMR?	Please specify, among the following list, the main goal(s) of the intervention, i.e. at which level does the intervention want to fight against AMR: Drivers: one driver of AMR is targeted Pressure: reduce or improve antimicrobial use State: decrease AMR prevalence Impact: decrease the impact of AMR on human or animal health
If necessary, please provide a comment on the selected goal(s) of the intervention	This free text cell aims to capture information about the main goal in more detail.
Challenge(s) of collective action of the intervention, i.e. targeted question of collective action covered by the intervention
What is(are) the challenge(s) of collective of the intervention?	Please specify the type of challenges conducted by the intervention: Surveillance: increase the information that we have about the system Conservation: conserve the effectiveness of antimicrobials through reducing use and/or misuse Infection prevention/control: intervention that aims at preventing infectious diseases in the first place Access: intervention that strengthen access to quality antimicrobials for all Innovation: focus on fostering research and development of new antibiotics, other compounds as well as diagnosis Containment: intervention that aims at limiting the spread of resistant infections And/or ‘other’ (to specify)
Please specify the challenge(s) of collective action of the intervention by selecting subchallenges	[if Challenge = ‘infection prevention’] If one challenge of collective action of the intervention is the prevention and control of infections, please specify its topic: Sanitation: all interventions related to water, sanitation and hygiene (WASH) Immunization: intervention using vaccines to prevent infection Hand hygiene Food hygiene Safe invasive procedure Biosecurity: all measures to prevent infection in farm and animal processing And/or ‘other’ (to specify)
	[if Challenge = ‘surveillance’] If one challenge of the intervention is the surveillance, please specify its topic: Use: measuring antibiotic use Epidemiology: measuring the prevalence/incidence of AMR Outcome/syndrome: typically, measuring number of death and adverse outcomes. For example, surveillance with indicators based on the detection of cases
	[if Challenge = ‘conservation’] If one challenge of the intervention is the conservation of antimicrobial, please specify its topic: Antimicrobial use: aim to improve the use of antimicrobials both quantitatively (e.g. use antimicrobial in the right dosage) and qualitatively (e.g. use the appropriate antimicrobial regarding the targeted pathogen) Quality antimicrobials production: guarantee the production of quality drugs in contrast to substandard and falsified product Drug promotion: limit how pharmaceutical producers can promote the use of antimicrobials either to prescribers or to users Antimicrobial residue: limit environmental pollution by antimicrobial residues (e.g. farms, hospitals, pharmaceutical industry)
	[if Subchallenge = ‘antimicrobial use’] If one challenge is the conservation of antimicrobials and, more precisely, the use of antimicrobials, please specify the sub-subchallenge: Awareness campaign: improve information and awareness about antimicrobial use Prescription only use Stewardship programme: aims to reduce the quantity of antimicrobials used or focus on improving correct use of antimicrobials following the responsible use principle: ‘right drug, right time, right dose, right duration’87 Growth promotion: aim to limit the use of antimicrobials as growth promoters Critically important antimicrobials: interventions that aim to separate the use of antimicrobials in general or in one setting or sector (e.g. list of critically important antimicrobials kept for human use only), or limiting non-prescription use And/or ‘other’ (to specify)
	[if Challenge = ‘containment’] If one challenge of the intervention is containment, please specify the transmission pathways targeted by the intervention:84 Human to human – Healthcare: reduce the human to human transmission of resistant infection in healthcare setting Human to human – Community: reduce the human to human transmission of resistant infection in the community Human to animal: reduce the transmission of resistant infection from humans to animals Animal to human – Food: reduce the transmission of foodborne pathogen Animal to human – Direct contact: aims to reduce the transmission of AMR from animal to human through direct contact Animal to animal: aims to reduce the transmission of resistant infection between animals Environment to human: intervention that aims to reduce the transmission of resistant infection from the environment to human (e.g. ingestion of contaminated water) Environment to animal: aim to reduce the transmission of resistant infection from the environment to the animal (e.g. ingestion of contaminated water)
	[if Challenge = ‘containment’] If one challenge of the intervention is containment, please specify the level of implementation of the intervention: Local: intervention that aims to reduce the local transmission of resistant pathogens National: intervention that aims to reduce the national transmission of resistant pathogens International: intervention that aims to reduce the international transmission of resistant pathogens
	[if Challenge = ‘innovation’] If one challenge of the intervention is the innovation, please specify its topic: Antimicrobials Diagnosis Alternatives to antimicrobials Vaccine And/or ‘other’ (to specify)
	[if Challenge = ‘access’] If one challenge of the intervention is the access to antimicrobial, please specify its topic: Subsidies/disbursement: interventions that aim to fund access to medicine Compulsory licensing: intervention that use TRIPS provision to make antimicrobials available Essential list of medicines

Table 4.

Variables of the AMR-Intervene framework in the building block ‘Governance of the intervention’

Variable	Description
Participation
Who is(are) the main agent(s) responsible for the intervention? (i.e. main leader group that initiated the intervention)	Please select between the three following categories the profile of the main agent responsible for the intervention, i.e. main organization that initiated the intervention: Government: public sector refers to government-owned organizations and government-provided services. Private organization or individuals: private organizations that are not government owned Academic: intervention is initiated for scientific purposes
Please specify the affiliation of the main agent(s) responsible for the intervention	[If MainAgent = ‘government’] If one agent of the intervention is part of the government, please specify its affiliation: Ministry of Health Ministry of Agriculture Ministry of the Environment Aid agency
	[If MainAgent = ‘private organization’] If one agent of the intervention is part of a private organization, please specify its affiliation: NGO – civil society Private company
	For others, please specify
Please specify the number of responsible agent(s) involved in the intervention64	This variable corresponds to the number of organization(s) involved in the design and/or implementation of the intervention.
Please specify the type of responsible agent(s) involved in the intervention	Choose the most appropriate description of the responsible agent(s) Governmental organizations Non-profit sector (e.g. NGO) Private business Public–private partnerships And/or ‘other’ (to specify)
Please specify the sector(s) of the responsible agent(s) involved in the intervention	Please select among the following choices what is the most appropriate mandate(s) of the responsible agent(s): Organization with human health mandate Organization with animal health mandate Organization with food production mandate Organization with environmental mandate Organization with trade mandate Organization with other mandate (to specify)
Policy instruments of the intervention
What is(are) the type of policy instrument(s) in the intervention?88	Please select among the following choices: Regulation (sticks): interventions that require people (positive obligation) to do something or prevent them from doing something (ban or negative obligation). There are sanctions if you do not do it.66 For example, regulations that ban the use of antimicrobials as growth promoters. Information, awareness or nudging (sermons): interventions that seek to encourage people to act or do something. There are no direct sanctions or reward for doing it. For example, health awareness campaign (promotion) or hospital stewardship. Incentives and deterrents (carrots): interventions that seek to encourage people to act or (not) do something. There are rewards or penalties, but it remains the people’s choice to do it. For example, subsidies and tax that aim to influence target’s behaviour
Is the intervention a positive or a negative intervention?	Choose among the following choices: Positive: do something more or do something right (promotion) Negative: refrain from doing something (restriction)
Please specify the subtype(s) of the intervention according to the type of policy instrument(s)89,90	[If Type = ‘regulation’] If the intervention was classified as a regulation, please specify the subtype: Public regulation: the intervention comes from a regulatory authority Co-regulation: interplay between different organizations (e.g. standards, certification) Self-regulation: the intervention is an attempt at self-regulation. This can be either a firm, a professional organization or a group of people
	[If Type = ‘information’] If the intervention was classified as an information, please specify the subtype: Awareness: ‘efforts to communicate to large numbers of citizens in order to achieve a policy result’, for example through awareness campaign91 Education/training: education focuses on students while training focuses on professionals Stewardship: ‘the job of supervising or taking care of something, such as an organization or property’ Nudge: ‘positive reinforcement and indirect suggestions as ways to influence the behaviour and decision making of groups or individuals’
	[If Type = ‘incentive’] If the intervention was classified as an incentive, please specify the subtype: Push mechanism: push mechanisms reduce a firm’s cost of researching and developing new drugs by distributing the expenditures across multiple parties Pull mechanisms: in contrast, pull mechanisms reward successful development of a drug by increasing or ensuring future revenue. For example, increasing access to research, providing research grants, offering tax incentives and establishing public–private partnerships for sharing research and development outlays
Building blocks of the intervention
To which building block(s) of a One Health system does the intervention belong?	A building block of a One Health system is a core component that structures a system. Choose among the following categories: Service delivery Health workforce Health information systems Access to essential medicines Financing Leadership/governance
Means of the intervention
What is the level of funding of the intervention? (if available)	Knowing the level of funding helps to access the expected costs of running the intervention. Mention the amount in the currency described in the paper.
Who mainly fund the intervention?	Choose the type of investment of the intervention: Government funded: funded through taxpayer money Privately funded: funded by private entities

Table 5.

Variables of the AMR-Intervene framework in the building block ‘Assessment of the intervention’

Variable	Description
Influencing variables
Which is(are) the variable(s) influencing the success/failure of the intervention described in the paper? (if available/applicable)	Please capture any (qualitative, semi-qualitative or quantitative) indication of how a variable influences another variable(s) (e.g. an increase in the number of responsible agents increases the number of meetings), or any information on feedback loops (i.e. positive or negative reinforcing of an action towards the final aim of the intervention). Any indication of magnitude of increase/decrease or magnitude of intervention outcomes or broader impacts is important.
Key results of the intervention
What is(are) the category(ies) of the main outcome(s) of the intervention?	Please select among the following choices: Drivers: reduction in a key driver of AMR (e.g. lack of sanitation) Pressure: reduction in the consumption of antimicrobials in human, animal health or food production State: reduction in the prevalence of AMR Impact: reduction in the impact of AMR such as morbidity and mortality
Please specify the number of measurable key results described in the paper	Only the first five key results can be recorded in the database.
For each of the key results (up to 5), please enter the following information:	Name of the key result: short description of the result measured. For example, prevalence of MRSA in veal calves
	Type of result: Output: ‘The products, capital goods and services which result from a development intervention; may also include changes resulting from the intervention which are relevant to the achievement of outcomes’.73 Outcome: ‘The likely or achieved short-term and medium-term effects of an intervention’s outputs’.73 Do not know
	Change induced consequently to the intervention for the concerned result: Positive change towards AMR: the key result proves an improvement in the fight against AMR before and after the intervention Negative change towards AMR: the key result doesn’t prove an improvement in the fight against AMR before and after the intervention Not applicable
	Percentage of change: this is the actual value of change in the relevant outcome.
	Unit of the key result: description of the unit used to assess the change in the relevant key result. For example DDD/1000/year
	Free comment: this free text cell aims to capture information about outcome in more detail.
In the paper, is the cost- effectiveness of the intervention assessed?	Please select the appropriate answer: Not evaluated Evaluated
Factors influencing the success of the intervention
Please list and explain the potential success factor(s) of the intervention described in the paper	These are the factors that the authors of the study included as important to the success of the intervention
Please list and explain the potential obstacle/failure factor(s) of the intervention described in the paper	These are the factors that the authors of the study included as obstacles or drivers of failure regarding the intervention
Does the paper’s author(s) describe unintended consequences linked to the intervention?	An unintended/unexpected consequence is when one of the main outcomes is different from the planned results due to the intervention implementation. Please answer by yes or no.
Please list and explain the potential unintended consequence(s) of the intervention described in the paper

To extract further relevant information for AMR-Intervene, we identified AMR interventions targeting human, animal and/or environmental health, which used various intervention strategies such as awareness campaigns, training, education, stewardship, surveillance, communication, policy and legislation. Key papers and interventions were identified from published reviews.^{7–9,14,18,25–27} While reviewing the publications, additional values were added to categorical variables of the AMR-Intervene framework to ensure that all different types of identified interventions could be coded in the database. In addition, several types of interventions were fully coded into the database and reviewed by the team to clarify the meaning of the variables and reduce interpretation bias.

The framework was also inspired and revised from discussions among scientists working on AMR. The AMR resilience consortium developed the initial framework that was reviewed and then validated during two 2-day workshops in June 2018 and September 2019 in Stockholm, Sweden, with AMR experts. During the workshops, we discussed the framework variables and their capacity to cover different types of interventions. Overall, AMR-Intervene is built as a versioning system for the revision and improvements of the variables.

Main components of the framework on AMR interventions

In this section, we describe the different components of the AMR-Intervene framework as detailed in Figure 2.

Core information about the publication and research design data

The first component of the framework aims to extract relevant information about the data source, such as title, abstract, digital object identifier (DOI) number, authorship or year of publication. This component also requests details about the research design used for intervention assessment. While randomized controlled trials provide causal evidence, many AMR interventions are implemented in real-world settings, making experimental control challenging.²⁸ The framework aims to capture all types of interventions (e.g. randomized controlled trials, observational or ecological studies) and assess the quality of the evidence regarding the intervention and its impacts.²⁹ Furthermore, an assessment of the quality of description of the intervention is essential (Table S1, available as Supplementary data at JAC Online). Criteria for assessing the study’s quality, which are typically used in the selection of relevant literature for systematic reviews, are detailed in Table S2. Relevant criteria are the presence of clear descriptions of the objectives of the study, the implementation strategy, the study population, including its size and representativeness, and the presence of a control group.^7,8,25,30,31 Finally, the framework aims to assess whether the main outcomes for the assessment of the intervention are clearly defined and described in the ‘Building and validation of the AMR-Intervene framework’ section (e.g. unit of measure, method for calculation and data sources).

Social system of the intervention

Any intervention designed to tackle AMR is embedded within a social system, which can be understood in terms of six main subcomponents (Table 1 and Figure 2). A first entry point for characterizing the social system of the intervention is to distinguish between interventions taking place in the human, animal and/or environmental sectors. While sectors can be delimited in several different ways, the term usually refers to a part of the economy in which particular activities take place within specific institutions and processes. Sectors can also be subdivided into subsectors (e.g. livestock and aquaculture in animal health), such as those detailed in Table 1.

Sectors are usually associated with social groups, which correspond to groups of people that are the primary target of the intervention. Social groups must be differentiated from the responsible agents, i.e. those funding, designing and evaluating the intervention (the latter is covered in the governance and implementation components of the framework). In addition to interventions targeting users of antimicrobials (e.g. general or specific populations such as children or the elderly), interventions frequently target professions such as prescribers (e.g. medical practitioners, nurses and veterinarians), dispensers (e.g. pharmacists) or producers of antimicrobials (e.g. pharmaceutical manufacturers). In animal health and food production, interventions can also target specific occupations, such as farmers, veterinarians, processors (i.e. where food is prepared for further consumption) and retailers (i.e. places where food is sold). These groups of occupations can also be responsible for implementation of the intervention (e.g. pharmacists providing awareness material to clients). The social group targeted by the intervention has to be distinguished from the bio-ecological host/substrate of AMR genes and residues, which can be a specific group of people (e.g. human patients) or animals (e.g. cats, pigs or fish), and are part of the bio-ecological component described below. However, in certain cases, such as improving antimicrobial use (AMU) in immunocompromised patients, the social group can be equivalent to the bio-ecological host targeted by the intervention if they are directly targeted by the intervention.

The setting(s) of an intervention is(are) the physical (social/institutional) location(s) where an intervention takes place. In human health, the main areas are either healthcare facilities, such as hospitals and nursing homes, or community buildings, such as schools, prisons, supermarkets or restaurants. In human health, most reported interventions have been conducted in hospitals, many of them in tertiary hospitals in high-income countries (HICs). In the animal sector, interventions typically take place on farms, but can also target practices in slaughterhouses, veterinary clinics or other settings where animals are raised (e.g. zoos).

Additionally, interventions can also be described by their temporal dimension, for which important information includes the start and end date of an intervention, as well as its status when the assessment is conducted. For the spatial dimension, relevant parameters include the country(ies) of implementation as well as countries’ capacities to address the problem.²² For example, it is important to be able to differentiate countries by their income (as defined by the World Bank) as limited access to antimicrobials and other technologies can limit the range of intervention options.^22,32 In parallel, interventions can be located within a defined socio-political dimension (i.e. local, subnational, national, regional and international levels in Table 1), which are usually linked to the organization of the social and political system.

Bio-ecological system of the intervention

The bio-ecological system of an intervention can be divided into at least four relevant subcomponents (Figure 2 and Table 2): (i) microorganism(s) targeted by the intervention; (ii) characterization of the resistance(s) of the microorganism(s); (iii) host population or substrate; and (iv) ecology of transmission. AMR has been identified in all main categories of microorganisms including bacteria, viruses, parasites and fungi, with important differences in terms of mechanisms, microbiology and readiness to develop AMR. Within the global resistome, i.e. the interacting global collection of AMR genes that have been identified including intrinsic or acquired resistance,^33,34 the mechanisms underlying the development of AMR are diverse and include de novo mutation and exchange of plasmids and genetic material.^35,36 While increasingly pathogens are showing resistance to several or even all classes of antimicrobials,³⁷ acquired AMR is often expressed as a combination of a pathogen and a class of antimicrobials that represents the most important current problem (e.g. carbapenem-resistant Enterobacteriaceae). The most pressing problems in human health are currently MDR Gram-negative bacteria,^1,38 but AMR is common in other bacteria and other pathogens such as protozoans (e.g. malaria) and viruses (e.g. HIV-AIDS).^39,40

The host population or substrate corresponds to the biological organisms or food/ecological substrate (affected by the pathogen) assessed by the intervention, where targeted microorganisms carrying AMR are observed. In humans, the host population can be the general population or more specific groups at greater risk of infection and/or colonization, such as children, patients with chronic conditions or migrants.⁴¹ In animals, a distinction is made between interventions with hosts as farmed animals (livestock or aquaculture) versus pets or wild animals. Efforts have focused on the first category as the absolute consumption of antibiotics is much higher in food animals than in companion or wild animals,^42,43 although other aspects such as exposure and connectivity may not be. In the environment, interventions mainly work to prevent the transmission of antimicrobial residues and genes into waste water (e.g. sewage plant) or surface water (e.g. river, lake or wetland) and into soil (e.g. manure disposal).^44,45 In terms of hosts, the bio-ecological system can finally be detailed via the clinical syndrome caused by pathogens. Infection and colonization of hosts can cause a range of signs and symptoms affecting different physiological systems depending on the pathogen and its location in the body. For example, it is frequent in the literature to find interventions targeting urinary or respiratory tract infections in humans.

The framework also aims to characterize the ecology of resistance (Figure 2 and Table 2). Which populations or substrates act as reservoirs for the transmission of AMR (i.e. population/substrate in which a pathogen lives and multiplies without affecting it, and which remains a source of infection) are also highly relevant to the characterization of the bio-ecological component of the system in which an intervention takes place, but is still not well understood.^46,47

Triggers, goals and challenges of the intervention

The diversity of interventions on AMR can be captured in terms of triggers, goals and challenges of collective action addressed by the interventions (Table 3 and Figure 2). A trigger corresponds to the motivation for the implementation of the intervention. A relevant way to conceptualize triggers is using the DPSIR framework.²³ In the context of AMR, the intervention’s trigger can be an increase of AMU (pressure), a rise in the incidence or prevalence of a resistant pathogen (state) or an increase in mortality associated with AMR (impact). In other cases, the trigger can be more upstream, e.g. a rise in a driver of AMU, such as the burden of infectious diseases or demand for animal protein. Interventions can be undertaken preventively, e.g. when a new, highly resistant strain is occurring in a neighbouring country, or reactively when a disturbance is already happening in the system. A good illustration of a preventive measure is the isolation of internationally transferred patients upon hospital admission in the receiving country.^48,49

In contrast to the trigger, the goal of an intervention corresponds to what an intervention tries to achieve in terms of outcomes. Many interventions ultimately aim to reduce the impact of AMR on human health, but they act on different levels of the long causal chain of AMR from drivers to outcomes.⁵⁰ Sometimes the goal coincides with the trigger, but not necessarily. Using the DPSIR framework, a common intervention goal is to reduce selective pressure by lowering the consumption of antimicrobials. Less frequently, the goal is to prevent or reduce a driver of AMR and AMU. This includes the prevalence of infectious diseases, which in human health depends on variables such as the levels of sanitation and hygiene, immunization and other public health measures, and determinants of the resilience of health systems.⁵¹ In animal health, it can relate to responsible use, vaccines, biosecurity and more.⁵² Finally, an intervention can aim to reduce the prevalence of AMR or mitigate the effect of AMR such as morbidity and mortality or the increased cost associated with treatment of resistant infections through introduction of mitigation measures.^53,54 Interventions can also be classified by their leverage point, i.e. a place in the AMR system where an intervention may lead to changes across the whole system. Leverage points can be categorized depending on the potential impact they might have on the AMR system, e.g. low and high leverage points.⁵⁵ Low leverage points are targeted at a precise parameter of the system (e.g. reducing the use of one specific antimicrobial), while high leverage points aim to change the whole system where AMR is expanding (e.g. change in social norms regarding meat consumption). Another important way to look at an intervention is to capture the main policies used to tackle AMR, which requires coordination between stakeholders, countries and sectors. Drawing on the literature,^56–58 surveillance, infection prevention, conservation, access, innovation and containment were defined as the six main challenges of collective action regarding AMR (Table 3). Slowing down the emergence of AMR includes activities such as infection prevention, which reduces the need for antibiotics in the first place (drivers), but also conservation of antimicrobial effectiveness, which addresses the quantity of antimicrobials used and the way in which they are used (duration, indication, etc.). Conservation also covers behavioural components, representation and norms about AMU. Access to high quality medicines and other health technologies is key for both infection prevention and conservation strategies. Surveillance provides information about the state of the system, which is essential for slowing down the transmission of AMR, as well as continuous monitoring of the situation. Containment of AMR is primarily dedicated to managing the connectivity in the AMR system, which is relevant to many sectors and different geographical scales, with travel and trade identified as major factors contributing to the global risk of spread of resistant strains and other emerging infectious diseases.⁵⁹ Finally, innovation is primarily new technologies to face AMR. For example, new drugs have been one of the main strategies to remain ahead of the problem.⁶⁰

Governance and implementation of the interventions

The factors related to governance, including, for example, responsible agents and partners (i.e. people involved in the design, implementation, evaluation and reporting of interventions), policy instruments and resources (Table 4 and Figure 2), and the implementation strategy of the interventions (i.e. the techniques used to enhance the adoption, implementation and sustainability of an intervention)⁴ are critical to tackle AMR.^61,62 Regarding responsible agents, many interventions are undertaken at small scales (clinicians and/or researchers in hospitals and farms), but an increasing number of interventions emanate from governmental agencies including ministries of health and agriculture,²² as well as other stakeholders, such as non-governmental organizations, international organizations or the private sector. In some cases, people who evaluate an intervention (e.g. researchers) are different from those that implement it (e.g. government). In the social–ecological literature, stakeholder participation is an important determinant of the success of interventions, especially when the problem to be addressed crosses the traditional societal demarcations.^63,Participation can be measured in terms of both the diversity of stakeholders (e.g. government department, non-governmental organization and community organization) and sectors with variable but often overlapping domains of authority (e.g. One Health). Stakeholder participation in the design, implementation, assessment and reporting can range from dissemination of information to much more active forms of involvement in decision making and resource allocation.^64,65

In terms of policy instruments, policy analysis distinguishes between ‘sticks’, ‘sermons’ and ‘carrots’,⁶⁶ detailed in Table 4. Regulations are classically top-down, coming from governmental authority (e.g. prohibition of AMU for growth promotion in farming). It can also be a co-regulation, which results from the interplay between different organizations, or a self-regulation (typically through the adoption of good practice standards by the private sectors). ‘Sermons’ come in different forms, including education and awareness campaigns, and also in the form of positive reinforcement.⁶⁷ Finally, market-based instruments are typically non-coercive approaches that seek to influence economic actors’ behaviour. Incentives to foster innovation of new drugs are typically understood in terms of push and pull mechanisms.¹² While push mechanisms reduce, for example, a firm’s cost of researching and developing new drugs by distributing the expenditures across multiple parties, pull mechanisms reward successful development of a drug by increasing or ensuring future revenue. In an integrated approach to AMR, multicomponent interventions can complement the classical regulatory approaches with incentives and the provision of information. Integrative strategies have been successful for responding to outbreaks, including search and destroy strategies for MRSA in a Dutch tertiary care centre.⁶⁸

Another important way to qualify the governance of an intervention is to identify the targeted building blocks of the health system.^32,69 WHO has proposed six building blocks, which include service delivery, health workforce, health information systems, access to essential medicines, financing and leadership/governance.^70,71 Clinical and public health interventions frequently aim to influence service delivery, while policies developed in national action plans aim to channel funding and improve governance of AMR control. These building blocks provide a common language between stakeholders, from policy makers to global health experts.⁷² Finally, any intervention requires means (e.g. financial, human and knowledge).

Assessment of the intervention

Assessment encompasses both the key results of the intervention and the factors influencing its success or failure (Figure 2 and Table 5). AMR interventions can take several years before translation into measurable impacts.²³ For example, an information campaign might change perception, but not necessarily result in a measurable reduction of AMU or AMR, let alone translate into reductions in mortality or morbidity. The framework first captures outputs, ‘i.e. the products, capital goods and services which result from an intervention’.⁷³ Examples of outputs include guidelines for AMU and annual surveillance reports. Second, outcomes are the ‘likely or achieved short- and medium-term effects of an intervention’s outputs’^73,74 and may encompass behaviour change, implementation of national legislation and creation of an AMR national board. Third, the impact of an intervention is the change of key variables, such as AMU or the incidence/prevalence of AMR pathogens. There are a variety of indicators that can be used to assess outputs, outcomes and impacts of an intervention that depend on the nature of the interventions and the sectors involved.⁷⁵

As implementation science is critical for improving our understanding of interventions, the framework aims to collect information about the evaluation of the implementation strategy and the factors influencing the success or failure of an intervention.^76,77 Narratives of success often rest on the introduction of a new technology or a ‘magic bullet’, but the broader technological, organizational and contextual factors that underpin success and failure of AMR interventions and their implementation are not well understood. Factors for rejection of interventions, e.g. high costs, intensive time demands and lack of customization to the context, have been identified in some areas.⁷⁸ Other typical obstacles in the management of complex systems include unwillingness to change, poor planning and management, and financial hurdles.⁷⁹ Finally, another important topic is to understand under what conditions selective interventions can (or cannot) achieve their goals of strengthening system capacity.⁸⁰ An assessment of interventions should look for both the co-benefits of interventions, especially regarding the Sustainable Development Goals (https://sustainabledevelopment.un.org/sdgs) as well as their unintended consequences,⁸¹ which is when one of the main outcomes is different from the expected results.⁶¹ Unintended consequences are typical of interventions in complex adaptive systems whereby interaction between variables and goals results in effects that are difficult to predict.⁸²

Discussion and recommendations

Traditionally, AMR has been considered as a medical problem that needs to be addressed by the rational use of antibiotics and the development of new technologies.⁵⁰ In contrast, the growing recognition of the complexity of AMR has ushered in the development and application of a variety of strategies to tackle this urgent global health threat. Findings associated with the study of social–ecological systems and their resilience over the past three decades⁸³ inform the development of an appropriate interdisciplinary framework to make sense of the diversity of strategies designed to tackle AMR, to capture relevant elements and intricate dynamics between the ecological and social systems, and to help gain a better understanding of what works and under which conditions.

The framework has some limitations. First, even though we initiated a scoping review of the published literature on AMR interventions to identify the main gaps in terms of sectors and other relevant variables of the framework,⁶ we have not tested the framework against all interventions. Some original and creative interventions might not be captured by the current framework, meaning that some relevant variables might be missing. However, the framework is easily adaptable and can be improved according to new insights. The next step will be to further test and apply the framework to different types of AMR interventions, for example those focusing on Escherichia coli from a One Health perspective. Second, some categories require guidance and learning to be applied correctly. Populating the database is currently done via an online form that provides definitions and some examples to support data completion. In a future development of the database, guidance should be improved with participatory learning tools, e.g. Wiki section or glossary. Third, interventions are only one way to build resilience. As the effectiveness of any intervention depends not only on its design and implementation, but also on the state and capacities of the receiving system, successful evidence-based actions on AMR require the combination of insights from both the implementation of interventions and the context in which they take place. While many of the known/published AMR interventions are currently implemented in HICs (and the success of the interventions in these countries may inspire other countries), cultural, political and ecological differences are important parameters that are likely to modulate the effectiveness of response.

Despite these limitations, to our knowledge, AMR-Intervene is the first framework built to emphasize the characterization of AMR interventions from a social–ecological perspective and the identification of resilience factors of virtually all types of AMR interventions. While few interventions are currently designed to build resilience, the AMR-Intervene framework underpins the creation of a database and learning platform of interventions,⁶ which will be the first known of its kind, acknowledging the significant variety of interventions targeting AMR and the diversity of contexts in which they are implemented worldwide. Ultimately, the application of the framework and associated database will help to improve how interventions are identified, conceptualized, described, reported and assessed.⁶ As is strongly recommended for experimental trials or systematic reviews, the systematic description will also help to improve the quality of interventions and provide a minimum standard for publication of the intervention. Thus, the framework and associated database of interventions will be useful for further research as well as an evidence-based source of information for decision and policy makers.⁶

While learning and experimentation are key determinants of resilience in social–ecological systems, AMR-Intervene aims to become an interface between science and policy that can accelerate the uptake of successful interventions. Providing a repository for information regarding AMR and insight for designing and implementing interventions that are tailored to the context could lead to policy innovation and improved design of new or existing interventions, and help channel research funding. To foster its value and usefulness for all, a learning system about AMR interventions should be open, transparent, collaborative and inclusive of different disciplines.

Acknowledgements

This study is part of the AMResilience (https://amr-resilience.gtglab.net/) project, which aims to provide and validate a comprehensive multimethod assessment of resilience and transformability to limit AMR and antimicrobial use in national and regional One Health systems.

Funding

This work was funded through an operating grant of the Fifth Joint Programming Initiative on Antimicrobial Resistance (JPIAMR 2017). This work was supported by an operating grant from the Canadian Institutes for Health Research (Institute of Infection and Immunity, Institute of Population and Public Health) through the JPIAMR Grant programme (Principal Investigator: S.M., grant number #155210), FORMAS (grant number 2016–00227) and the Swiss National Science Foundation (grant number 40AR40_180189). The funders had no role in the design, analysis or writing of this article.

Transparency declarations

None to declare.

Supplementary data

Tables S1 and S2 are available as Supplementary data at JAC Online.