Emergence of antibiotic resistance (AMR) is a serious concern for Low and Middle Income Countries (LMICs). Unregulated use of antibiotics, a major AMR driver, is highly prevalent in LMICs, with medicine stores as key providers. Physical interactions between One Health compartments increase cross-domain transmission risks, although the relative importance of different reservoirs is uncertain, with community-level dynamics of AMR in LMICs largely unquantified. In two rural health districts in Burkina Faso and DR Congo, a behavioural intervention bundle will be developed, targeting medicine stores and their communities, to optimise antibiotic use and improve hygiene, and hence reduce AMR prevalence and transmission. After a 6-month local co-development phase, the intervention will be implemented over 12 months and evaluated through a comparison between intervention and control clusters, consisting of one or more villages or neighbourhoods largely seeking healthcare with the same provider(s). The primary outcome measure is the change in Watch antibiotic provision from medicine stores (where a formal prescription is not required), assessed via patient exit interviews and simulated client visits. Changes in hygiene practices and AMR pathogen and gene carriage will be assessed in repeated population surveys. Rodents, living in close proximity to humans in much of sub-Sahara Africa, provide a proxy estimate of environmental AMR pathogen and gene exposure. Using modelling and sequencing of selected isolates, impact of AMR transmission by changes in antibiotic use and hygiene practices will be quantified.

Background A major driver of AMR emergence in LMICs is community- and individual-level antibiotic consumption, which have both been associated with an increased risk of acquisition of AMR bacteria for individuals in the general community. As a result of difficult or delayed access to hospitals and formal health centres, the principal sources of antibiotics in many LMIC communities are medicine stores, i.e. community pharmacies or informal medicine vendors. Staff at these outlets are frequently not medically qualified, and informal selling of medicines is common. Two recent systematic reviews estimated the pooled prevalence of reported self-medication with antibiotics in LMICs at 78%, in sSA at 56%, and in West-Africa at 70%. Furthermore, in sSA, over two thirds of visits to community medicine stores were found to result in dispensing prescription-free antibiotics. Appropriately trained and motivated pharmacy staff can be part of a successful AMR control programme, in particular through their role in educating patients, promoting appropriate usage of dispensed antibiotics, and providing guidance to healthcare colleagues on appropriate antibiotic prescribing. In the absence of a clinical or microbiological diagnosis, medicine stores often dispense antibiotics without a clear rationale. Importantly, these antibiotic courses frequently consist of Watch antibiotics. Among the different types of providers listed, and even among those with a good understanding of AMR, there is a need for supporting appropriate antibiotic prescription practices.

A recent study in 6 LMICs showed that context-specific tailored intervention packages are key to improve community antibiotic use. In any stewardship programme targeting unregulated community dispensing of antibiotics, it is therefore crucial to co-develop interventions with medicine stores, and to incorporate the identification of alternative (economic) incentives, as well as targeting communities, to enable sustainable take-up by both medicine stores and their communities. Existing behavioural change interventions have been categorised as persuasive (eg. peer-to-peer feedback on dispensing), enabling (eg. guidelines, training sessions), restrictive (eg. expert approval before dispensing some specific antibiotics) or structural (eg. introducing a clinical algorithm). The effect of individual interventions targeting outpatient dispensing of antibiotics in LMIC has been heterogeneous, with multi-faceted interventions combining educational material with audits and feedback or peer-to-peer comparisons more effective at reducing inappropriate antibiotic use than stand-alone interventions.This project proposes a robust participatory-driven behaviour change intervention to reduce the use of Watch antibiotics from medicine stores, targeting both the demand (community) and supply (medicine store) side, and to reduce emergence and transmission of AMR.

To develop locally acceptable, feasible and relevant interventions, the COM-B model for behaviour change has been found highly suitable. This model forms the centre of the well-known Behaviour Change Wheel which is widely used in designing interventions and has been used by NICE and the UK Department of Health. COM-B identifies three essential conditions for behaviour - capability, opportunity, motivation - which thereby provide intervention opportunities for behaviour change. Capabilities include psychological (e.g. knowledge) and physical (e.g. skills) capabilities. Opportunity includes social and physical opportunities (e.g. social influences and environmental context and resources). Motivation includes reflective and automatic motivation such as beliefs about capabilities and consequences, goals, and ideas about professional role and identity. The intervention bundle in this study therefore aims to address AMR using three intervention components which each address the three conditions of behaviour change of the COM-B model. Target behaviours for the intervention are based upon considerations around impact potential, likelihood for change, potential spill-over effects as well as ease of measurement. The COM-B model would thus be highly suitable to guide ABU interventions addressing highly complex behaviour, and will be used to design the intervention bundles to achieve a joint change in antibiotic demand and supply.

At the same time, ongoing transmission of newly emerging or existing (drug-resistant) bacteria and the exchange of AMR genes between bacteria harboured by human hosts, animals and their environment, is facilitated by substandard hygiene and sanitation practices. Household transmission was recently found to be a more important mode of bacterial strain sharing than transmission from livestock in urban Nairobi.

The widespread environmental rummaging behaviour of rodents implies that they can serve as a proxy for AMR prevalence in the natural environment . This is of major relevance in settings where sewage surveillance, an alternative measurement to estimate environmental prevalence, may not (yet) be feasible because of the absence of sufficient sewage systems.

Study objectives

Primary

1. Develop, implement, and evaluate the effect of a behavioural intervention bundle targeting medicine stores (including community pharmacies and informal medicine sellers), and the surrounding populations on (Watch) antibiotic use.
2. Develop and pilot environmental AMR surveillance through rodent surveillance.
3. Estimate and model the effect of the intervention bundle on AMR prevalence and transmission, focusing on faecal E. coli and Salmonella carriage.

Secondary

1. Estimate the intervention bundle's effect on hygiene, on case management by medicine stores and on clinical outcomes.
2. Identify pathways and incentives through which educational or peer influence interventions improve quality of care.
3. Compare prevalence of AMR bacterial populations in human (i.e. household members) and rodent reservoirs; compare with AMR prevalence in routine BSI surveillance.
4. Spatial and ecological analysis as well as phylogenetic comparison of AMR bacterial populations and genetic clones identified in human carriers and dwelling rodents.
5. Quantify household transmission of AMR genes or pathogens and estimate the relative importance of an environmental transmission source.

Patient exit surveys

Inclusion Criteria:

• Visitor to a medicine dispenser for his/her own health or that of someone else, regardless of disease or age
• Having just completed a visit to a medicine dispenser in a study cluster who agreed to participate in the study Exclusion Criteria: none

Simulated patient visits (dispensers)

Inclusion criteria:

• Prescribe or dispense medicines in a community pharmacy, medicine store, or other private clinic or outlet within the selected study clusters;
• Consented at the start of the study to participate in the intervention bundle evaluation Exclusion Criteria: none

Faecal Enterobacteriaceae carriage and transmission (members of selected households) and household surveys (household heads) Inclusion criteria

• Member of a household in a study cluster (resident for >/= 3 months);
• Agreement of the household head for all household members (including children) to participate to the collection of four stool samples during the study period, through informed written consent;
• Individual informed consent (plus assent for adolescent participants) for each participating household member;
• be in good physical health (no current infectious disease); Exclusion criteria
• Inhabitant of study clusters who planned to move or be absent during the following year;
• Temporary exclusion: Ongoing infectious disease or ongoing treatment for an infectious disease. Field workers will return to collect stools once the patient has recovered

Access to study data (e.g., pathogen sequencing data, outcome data as antibiotic use or quality scores) will be in accordance with FAIR principles.

Non-confidential data will be fully accessible (such as fully anonymized and aggregated data, pathogen sequencing data, intervention materials, recommendations, etc.). Confidential data (such as pseudonymized participant data, therefore individual but without identifiers) can be made accessible through controlled access procedures (ITM Research data access committee via ITMresearchdataaccess@itg.be). A data sharing agreement will always be in place prior to the transfer of confidential data.

Closure of the (pseudonymized) participant database is provisionally planned by June 2024. Anonymized datasets will be made accessible shortly after. Original study data will be kept for 20 years.

When data is made (openly) accessible, this will be done as much as possible in open, license-free and interoperable formats. We will also make available metadata necessary for further reinterpretation, reanalysis or use of the research results. We will endeavor to use open licenses such as CC-BY or CC-BY-NC as much as possible. Confidential data (such as pseudonymized participant data, therefore individual but without identifiers) can be made accessible through controlled access procedures (ITM Research data access committee via ITMresearchdataaccess@itg.be). A data sharing agreement will always be in place prior to the transfer of confidential data.

The intervention consists of behavioural interventions, therefore masking of participants or care providers is not possible.