Studying Factors Affecting Success of Antimicrobial Resistance Interventions through the Lens of Experience: A Thematic Analysis

Antimicrobial resistance (AMR) affects the environment, and animal and human health. Institutions worldwide have applied various measures, some of which have reduced antimicrobial use and AMR. However, little is known about factors influencing the success of AMR interventions. To address this gap, we engaged health professionals, designers, and implementers of AMR interventions in an exploratory study to learn about their experience and factors that challenged or facilitated interventions and the context in which interventions were implemented. Based on participant input, our thematic analysis identified behaviour; institutional governance and management; and sharing and enhancing information as key factors influencing success. Important sub-themes included: correct behaviour reinforcement, financial resources, training, assessment, and awareness of AMR. Overall, interventions were located in high-income countries, the human sector, and were publicly funded and implemented. In these contexts, behaviour patterns strongly influenced success, yet are often underrated or overlooked when designing AMR interventions. Improving our understanding of what contributes to successful interventions would allow for better designs of policies that are tailored to specific contexts. Exploratory approaches can provide encouraging results in complex challenges, as made evident in our study. Remaining challenges include more engagement in this type of study by professionals and characterisation of themes that influence intervention outcomes by context.


Introduction
Antimicrobial effectiveness has been diminishing due to microorganisms acquiring resistance [1]. Lagging efforts to invest in development of new antimicrobials [2] are further

Results
Twenty-one interventions were analysed and their general contexts are shown in Table 1 (Supplementary Materials Table S3  ). The majority of interventions were perceived to have positive outcomes in reducing AMU and AMR levels and, thus, were in line with the main triggers of the interventions. Unintended consequences (i.e., outcomes that were not foreseen previously) involved higher consumption of other drugs, such as drugs for suppressing symptoms or narrow-spectrum or non-targeted antimicrobials, the latter reported also as a negative outcome because it contributed to increased resistance for them afterwards. However, the increased use of specific antibiotics was not enough to consider the intervention unsuccessful if overall antimicrobial use decreased. Table 1.
Component [15] Variables [15 Based on thematic analysis of responses, we found eight main success factor themes ( Figure 1). Key sub-themes reported in at least one third of the interventions are detailed in Table 2. From the most frequent to the least, the first theme was behaviour of individuals or institutions towards the intervention. This theme encompasses the reinforcement of correct behaviour, trust/support, collaborative behaviour and flexibility, adaptability, and willingness to change towards the intervention or its implementation. Second, the theme institution characteristics covered the management with training, collaboration, and co-ordination; and the governance, including communication, partnership, and engagement/commitment. Third, information-available for, or resulting from, the intervention-with awareness, regulations/guidelines, outcomes, and data provision/collection was considered important. Fourth, intervention characteristics comprised factors such as multifaceted/multisector interventions, sustained in time (ongoing interventions or without an end date), scalability (possibility of implementing it top-down/bottom-up), and assessment of the intervention. Fifth, the capacity of the system where the intervention takes place was considered important. Within this theme, financial resources, including funding and financed training sub-themes, knowledge/skills of main actors, adequate premises and equipment, and novelty/innovation, were important sub-themes. The sixth theme, implementation characteristics of the intervention, included factors such as well-designed or well-planned, good, and detailed design and implementation, easy to implement, consultation or guidance during the implementation, and implementation supported by main actors. The seventh theme was AMU with the accessibility category, including sub-themes such as prescription, controlled, or monitored use or high cost; plus other sub-themes, including use reduction and use improvement. The last theme was infection control related to or resulting from the intervention and comprised surveillance, decrease in infection incidence, and prevention-hand hygiene/sanitation/vaccination. The 12 identified key sub-themes, which are those mentioned in at least one third of the 21 analysed interventions, were qualitatively mapped in Figure 2 to initially grasp what factors contributing to positive outcomes could be promoted together.   Table 2. Meaning, assumptions, implications, and supporting quotes of the 12 key sub-themes, which were reported in at least one third of the interventions, resulting from the thematic analysis. The main theme where they belong and the total and partial frequency of each are also specified. AMR = antimicrobial resistance, AMU = antimicrobial use; FREQ = frequency; GPs = General Practitioners (medical doctors); IC = infection control; OB = obstacle; PREV = prevalence; SF = success factor. * Category (sub-themes: financed training and funding counted together as they are extremely related). New routines or ideas must be reinforced to ensure their continuity. Habituation needs time and going back to old routines due to inertia is usual.

Sub-Theme
Training and guidance are essential to make changes in the long term. Follow-up and regular feedback maintain motivation. Sustained efforts and sustained interventions are needed to avoid old habits. Use of new technologies (emoticons or mass media) as reminders.
"change in mentality should also be seen as one of the key success factors of this study" // "There is a human tendency to return to previous practices in the absence of constant motivation and reminders" // "Inertia among prescribers" Financial resources * Capacity 10 5 5 Enough budget and funding to carry out the intervention. Funding for teaching and training the main actors responsible for the intervention.
Good level of funding is crucial for implementing interventions. Costs can be very high including training, personnel, or resources and, without a proper budget, many of them are not going forward.
"Coaching of farmers" // "clinically oriented education through symposia, workshops and focused meetings at the regional and local levels" // "funding" // "budget to begin with" or "very costly to establish" Assessment Intervention 9 9 0 After the intervention has taken place or, for a defined period of the ongoing intervention, checking or measuring outcomes of the actions applied can help to elucidate the usefulness of the intervention or its possible gaps.
Results from assessment can help to maintain motivation and to identify new goals and opportunities to improve outcomes or to promote actions impacting AMR.
"results from the monitoring were used when writing guidelines" // "a decline of 26.5% in the number of antibiotic prescriptions was observed over 5 years" // "significantly increased the usage of hand-rub dispensers in patient rooms, in comparison to the three other tested conditions" Knowledge about AMR and people aware of the problem of untreatable infections enhance positive outcomes. Ignorance of the problem may lead to opposition of public opinion or citizenship (e.g., lack of prescription thought as cutback in health system) Society may behave differently following and finishing prescribed antimicrobial treatments. Prescribers less pressured to prescribe treatments to please patients or farmers. Citizenship engage to preserve antimicrobial effectiveness.
"Patients who insisted on receiving antibiotics" // "Health beliefs by the general populations" // "general reluctance amongst farmers and veterinarians to change their existing antimicrobial treatment practices" // "advisor/coach helps the farmer with explaining what he/she could be improving and what the risk is when certain practices are not performed correctly" Knowledge, skills and education Capacity 9 9 0 Deep and detailed knowledge or education increases system capacity to carry out the intervention.
Contribution and expansion of skills and knowledge to new staff or new performers of interventions. Impede waning of the intervention.
"investment in technical and epidemiological knowledge" // "The programme created a pool of trained technicians who can compensate for transfer and separation of staff and contribute to expansion of programme staff"// "advisor/coach helps the farmer with explaining what he/she could be improving and what the risk is when certain practices are not performed correctly" Trust and support Behaviour 8 4 4 Trust and support of main actors. On the contrary, prejudices and scepticism hamper good outcomes.
They enhance implementation and maintenance of interventions.
"supported by the key doctors of the ICU" // "Sustained efforts and trust of infectious disease pharmacists" // "Perception of the farmer that interventions cost money and time (although often not the case as proven in our studies) Professional training of the actions to improve or initiate in the intervention.
Training empower and increase self-esteem to carry out interventions, especially when actors are not familiar on a daily basis with AMR. Often, this training is funded.
"Training of GPs"// "The programme created a pool of trained technicians who can compensate for transfer and separation of staff and contribute to expansion of programme staff" // "farmers keep control over the health situation and are less reluctant to change certain AMU treatment procedures" // "pump priming investment to support development of pharmacists" Multifaceted/ Multisector Intervention 8 8 0 Intervention is composed or carried out by different sectors, settings, departments, or professionals Interventions not only affect one type of actors or sectors. Joining efforts from different backgrounds and perspectives may have bigger impacts, reach, and redundancy of interventions. Some tasks can be carried out or complemented by different agents for completion "Involving community pharmacists, care homes, nursing homes staff in this process and using training and care pathways" // "Intensive collaboration between the surveillance team and the medical microbiologists" // "close collaboration between the animal and human sector and between experts and political stakeholders or authorities" Consultation and guidance Implementation 8 7 1 Consultation or guidance for intervention implementation clarify actions and objectives of the intervention. Consultation and guidance available for actors. When lacking, often implies insecurity towards the intervention and actors can go back to old habits This tool during implementation or during the intervention enhances positive outcomes as they can rely on experts or other professionals' criteria when doubts arise. It promotes self-esteem and motivation of executors due to continuous knowledge, feedback and follow-ups "development of practical implementation guidance" // "advice from the Expert Advisory Group on Antimicrobial Resistance ceased in 2004" // "to achieve this (AMU) reduction, it is important to assist and guide farmers in this process" Compromise towards the intervention not only from individuals but also from the institutions designing, implementing, or performing the intervention Ensures effort from the institution to maintain or to carry out the intervention, independently of individual governances.
"Implementation had the support of heads of both departments" or "the veterinarian has often already been the advisor for years resulting in the loss of motivation due to, for example persistent disease problems" Data collection and provision Information 7 5 2 Data collection and provision standardised, available, and shared. In contrast, data collected or provided from different sites, with heterogeneous criteria or not shared hinders availability of knowledge Data of interventions that are shared, with standard reporting, can clarify the exact situation of the epidemiological state; these can be used by different settings or sectors and can clarify and/or quantify assessments.
"This integrated program was made possible because access to all relevant data and samples that were already systematically collected from animals, food, and humans has been shared" // "number of tools to make surveillance findings transparent and accessible to both scientists and non-specialists" // "Diversity in coding in laboratory information systems" Figure 2. Qualitative map between the 12 most frequent sub-themes in our study. Connections between sub-themes were mapped when those were mentioned together in the same response, which illustrates factors that can be promoted together to enhance success or to positively impact AMR.

Discussion
Overall actions to maximise the chance of success in AMR interventions start long before their implementation and, once implemented, interventions still need to be flexible and supported with training, feedback, or advice. Behaviour and acceptance towards the intervention of individuals and institutions are often overlooked, but they seem to have a remarkable impact on success. Influencing behavioural change in combatting AMR is a challenge that involves not only the health sector, but many actors where we lack previous experience or who have not previously been engaged [39]. This is especially challenging when it has been emphasized that increasing AMR awareness does not necessarily transform into behavioural change or into logical changes in AMU practices simply because people in many cases do not choose their behaviour or it is constrained by broader factors [40,41]. This exemplifies the role of political, social, economic, and historical factors in AMU decision making, which are often not considered in our simplistic interpretations of knowledge-driven practices [40].
It seems that causes leading to successful interventions rely on reinforcement of new routines and correct behaviour that can be promoted by initial training and awareness campaigns, regular follow-ups with feedback, and refresher training that enhances continuous collaboration, motivation, and education. Attitudes and feelings were highlighted as key sub-themes (reinforcement of correct behaviour or new routines, trust and support, and collaboration) and these tend to induce motivation, security, and self-esteem of individuals, plus support and engagement from institutions. Confidence and awareness are established through engagement and shared responsibilities, with collaboration and coordination encouraging support of each other, humility, and reflexivity. These factors may lead to building trust and support between individuals, different role positions, and institutions, engaging support and collaboration between sectors and/or professionals. Behavioural change, awareness, and recognition of trade-offs associated with our actions have been highlighted cornerstones in the fight against AMR [39,42,43]. Moreover, well-established behaviours, such as routines and human tendency towards traditions and old habits, are not fully considered when designing or implementing interventions. This aligns Figure 2. Qualitative map between the 12 most frequent sub-themes in our study. Connections between sub-themes were mapped when those were mentioned together in the same response, which illustrates factors that can be promoted together to enhance success or to positively impact AMR.

Discussion
Overall actions to maximise the chance of success in AMR interventions start long before their implementation and, once implemented, interventions still need to be flexible and supported with training, feedback, or advice. Behaviour and acceptance towards the intervention of individuals and institutions are often overlooked, but they seem to have a remarkable impact on success. Influencing behavioural change in combatting AMR is a challenge that involves not only the health sector, but many actors where we lack previous experience or who have not previously been engaged [39]. This is especially challenging when it has been emphasized that increasing AMR awareness does not necessarily transform into behavioural change or into logical changes in AMU practices simply because people in many cases do not choose their behaviour or it is constrained by broader factors [40,41]. This exemplifies the role of political, social, economic, and historical factors in AMU decision making, which are often not considered in our simplistic interpretations of knowledge-driven practices [40].
It seems that causes leading to successful interventions rely on reinforcement of new routines and correct behaviour that can be promoted by initial training and awareness campaigns, regular follow-ups with feedback, and refresher training that enhances continuous collaboration, motivation, and education. Attitudes and feelings were highlighted as key sub-themes (reinforcement of correct behaviour or new routines, trust and support, and collaboration) and these tend to induce motivation, security, and self-esteem of individuals, plus support and engagement from institutions. Confidence and awareness are established through engagement and shared responsibilities, with collaboration and co-ordination encouraging support of each other, humility, and reflexivity. These factors may lead to building trust and support between individuals, different role positions, and institutions, engaging support and collaboration between sectors and/or professionals. Behavioural change, awareness, and recognition of trade-offs associated with our actions have been highlighted cornerstones in the fight against AMR [39,42,43]. Moreover, wellestablished behaviours, such as routines and human tendency towards traditions and old habits, are not fully considered when designing or implementing interventions. This aligns with observations about guidelines, training, or even new and individualised policies being ineffective alone in achieving or maintaining behavioural change or to reform the industry sector [44][45][46].
Antimicrobial stewardship initiatives have been calling for the need to adopt a wider multidisciplinary and multifaceted approach involving experts from social, psychological, and behavioural sciences [45,47]. Assessment of stewardship initiatives in healthcare settings suggest that targets, actions, and locations are well-described but there is a lack of reporting about detailed actors (who is involved) and timing (how often and when), interfering with replicability and long-term behavioural change [48]. Although it is difficult to state how long support is needed for the intervention to be permanent or widely adopted, our findings suggest that impacting behaviour change starts in the design and implementation phases. All the above-mentioned information and the specific context and factors that can influence the behaviour of interest, both from individuals and from the system (socially, economically, and politically), need to be described [11,16] to avoid the tendency to return to or to fall on old routines once support, guidance, or feedback ends. There are internal dynamics to overcome following our and previous findings about addressing human behaviour as a cornerstone for successful intervention outcomes [48,49], which need to consider the whole context in which behaviour is embedded [11,15,44].
Themes related to AMU (AMU) and resistance prevalence (infection control) were the least frequently reported for having successful outcomes and, when they were, it was mainly as indicators (e.g., measures of AMU, AMR surveillance, or infection incidence). Resistance prevalence initiated a quarter of interventions, mainly focused on surveillance, which is still inconsistently implemented globally [50]; improving AMU initiated half of the interventions, mainly via stewardship programs. However, although AMU interventions are well-performed, outcomes may not be the ones expected or wanted if there is not deep knowledge of the complexity of the whole system (e.g., unexpected high resistance indicator after a strict stewardship program suggesting other sources of selective pressure [51]).
Evidently, financial resources are vital for success of interventions. Increasing budgets may not be easy but can benefit from public awareness (another key sub-theme), which can lead to institutional pressure to invest more at all levels. Most factors mentioned as obstacles to success (e.g., inertia and old involuntary habits) could be transformed into success factors (e.g., reinforcing new routines with continuous training, follow-ups, and/or feedback). Still, asking about obstacles and success factors separately can help identify features that contribute to success that may be overlooked or unnoticed when they work well or are taken for granted. Only one sub-theme (resources of the system), which belonged to the broader "capacity of the system" theme, was seen only as an obstacle (Supplementary Materials Table S4). Nonetheless, this sub-theme probably is related to shortage of funding, which again emerged as an obstacle for success, and thus in line with other studies that have associated risky behaviours and actions that contribute to the spread of AMR with low economic income and poverty [39,40].
The information extracted highlights the value and importance of participatory studies to better address AMR. Differences between success factors and obstacles depending on context were not observed, probably as a consequence of the homogeneity of interventions analysed and the broad focus of the study. We also found that interventions were triggered reactively, largely focused on humans, and thus action was taken after a concern had arisen. However, AMR affects the "One Health" spectrum, which requires preventive, proactive, and broader interventions applied in multiple sectors to allow development of economies towards the Sustainable Development Goals that have direct or indirect relationships with AMR [52].
Better understanding of factors that contribute to intervention success would benefit from more qualitative research. While there are different qualitative approaches, thematic analysis has been widely accepted as a flexible and consistent framework for capturing perspectives [53]. Research about complex issues that are difficult to assess and to quantify, such as AMR, can advantageously make use of this type of analysis to help guide devel-opment of good practices before translatable information and assessments are available or accessible to other colleagues or when information is time-delayed. Publications often provided limited data on success factors and obstacles to intervention success. Because there is a need for fast publication and for confiding expertise of interventions to other colleagues, the main advantage of this analysis is that it adds value, complements, and improves new insights in interventional science learning from other previous experiences and their socioecological context. Unanticipated insights can enable and facilitate identification and selection of interventions and reduce the theory-practice gap, as well as strengthen the science of intervention against AMR using dissemination and implementation frameworks and making valuable information accessible to other colleagues [54,55].
Limitations of our study are connected to the fairly homogenous context, type, and location where the majority of analysed interventions took place and the broad aim of the study. Interventions are impacted by cultural, political, historical, and societal circumstances and publications do not follow reporting guidelines and, if they do, those may need to be updated for capturing relevant details to AMR. Moreover, types of interventions included were not equally represented nor equally assessed. All these facts can lead to restriction in capturing important themes to intervention success (e.g., identified success factors seem to be applicable in high-income countries but information and important themes affecting success of interventions in low-middle-income countries may still be missing).
Success factors mentioned in reports or publications are often more objective and easier to assess as they are in line with the intended effect of the intervention, while factors retrieved from this qualitative analysis reflected and relied-on subjective responses independent of the level of assessment of the intervention. Engagement of interdisciplinary experts, health professionals, designers, and implementers of interventions worldwide is needed to learn from non-traditional studies in AMR, such as participatory studies, and to capture key features of intervention success. Moreover, engagement of such profiles will help to have a more inclusive view and supporting information to capture or strengthen new themes or sub-themes that this analysis may have missed or may have identified as anecdotes. The unexpected consequences of our actions are difficult to foresee, but preparedness and system thinking approaches, as well as greater and more consistent reporting of experiences will equip interventionists with useful tools and knowledge before implementing new interventions. Broad system integration has been shown to positively affect resilience and tolerance in adverse situations or shocks [56]. Interventions success is also impacted by different baselines that are influenced by cultural, political, historical, and societal circumstances. Thus, better reporting in AMR is needed and initiatives proposed by Léger et al. [15] and Wernli et al. [11] can improve and align information to capture heterogeneity, even if small, of contexts that can affect AMR.

Materials and Methods
We examined interventions to gain a preliminary understanding as part of the AM-Resilience project [11,15,16], which compiled interventions identified through scientific databases and extracted the socioecological context of interventions using the AMR-Intervene framework [15], and, also, we added to that framework the respondent profile to identify which type of professionals engaged more in our exploratory study. We invited authors of interventions from different sectors to complete an exploratory questionnaire to identify additional relevant information and to describe facilitators and obstacles to their intervention success. The online questionnaire used open-ended questions to avoid delimiting or biasing their experiences. Then, following the standard for reporting qualitative research [57] (Supplementary Materials Table S1), we performed an inductive thematic analysis to capture themes that relate to positive outcomes from interventions tackling AMR [17,57]. Coding was performed using MAXQDA v.2020, a computer-assisted qualitative data analysis software, without a pre-existing coding frame, and, therefore, was inductive, allowing the data to drive the themes (data-driven). To prevent bias from the main researcher, whose experience involves clinical microbiology and epidemiology, three more co-authors independently coded responses to assess interpersonal consistency. A theme was defined as the main explicit and clear tacit idea behind the participant answer and could be broken into sub-themes, which were detailed specific factors related to the main theme. A category was a (optional) middle classification where sub-themes could be related in a smaller group. Frequency was counted as interventions mentioning a particular theme, category, or sub-theme independently of the number of times repeated on each data item. Redundancies were included to not miss information, but, if present in the same data item, they were counted as one. Factors seen as key components for positive outcomes (either because factors were seen as satisfactory, or the opposite or lack of obstructive) were reorganised and clustered together (total frequency), although each contribution was considered in partial frequency. The method is described elsewhere [17] and, in our study, is also detailed in Supplementary Materials Table S2. Finally, combining sub-themes and not relying on a single one may positively impact AMR, which emphasises the importance of needing to use several approaches to maximise success. Therefore, identified key sub-themes perceived as contributing to positive outcomes that were mentioned in the same response were qualitatively connected with the aim to initially grasp what could be promoted together.

Conclusions
Non-traditional analyses in the AMR field, such as exploratory qualitative studies, can provide us with novel evidence-based insights to alleviate the AMR problem from different perspectives in a proactive manner. Due to gaps in communication, reporting, and differences in contexts, replication of AMR interventions is difficult, with details and experiences often missing or elusive. Lessons from previous experiences can help to identify key factors for such interventions to be successful, and implementation frameworks will enhance interventional science and build resilience to AMR. Themes and sub-themes captured in our exploratory study are meant to encourage more complete appraisal, communication, discussion, and sharing of experiences among different professionals and, ultimately, help to inform context-dependant intervention design and implementation. The themes identified can also be useful for guiding production of reports best suited to helping policy development, which is the focus of many organisations and governments worldwide.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics11050639/s1, Table S1: Standard Reporting for Qualitative research (SRQR) and reference [57] of the main manuscript; Table S2: Thematic analysis iterative process following the methodology described elsewhere, reference number 17 in the main manuscript, and adapted to our study about obstacles and success factors in AMR interventions; Table S3: Published articles included in the thematic analysis with references  in the main text; Table S4: Themes hindering interventions or their implementation not related to mentioned success factors.  PJGH also from FORMAS Inequality and the Biosphere project (2020-00454). PJGH undertook this work as part of the CGIAR Research Programs on Fish Agri-Food Systems (FISH) led by WorldFish and on Climate Change, Agriculture and Food Security (CCAFS) led by CIAT.

Institutional Review Board Statement:
Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research. Respondents' information was removed from the raw dataset and stored separately. Working data are de-identified. This study has been reviewed and received ethics clearance through a University of Waterloo Research Ethics Committee (ORE#40519).

Data Availability Statement:
De-identified data are available on request. Data requests with an expression of interest in pursuing secondary or extended analyses with a specific research question can be made to the corresponding author.