Antimicrobial Resistance in Rural Settings in Latin America: A Scoping Review with a One Health Lens

Antimicrobial resistance (AMR) in rural Latin America is not fully understood. The transmission pathways are partially known since research predominantly focuses on the urban hospital setting. The contribution to AMR from environmental factors is usually only mentioned in large-scale animal production. To understand the state of the literature on AMR in rural LA, we carried out a scoping review using the One Health (OH) perspective. OH recognises the concomitant contributions and interconnectedness of humans, animal, and the environment, thus, we used the OH perspective to select those articles adopting a holistic view of the problem. We searched original articles in English, Spanish, and Portuguese in four peer-reviewed databases and included 21 publications in the analysis. We charted data on bibliometrics, design, data collection sources, and instruments. We identified the human, animal, and environmental contributions to AMR in rural locations, and information gaps on AMR transmission routes and AMR drivers. Intensive and non-intensive animal production systems and agricultural practices were the most frequently found human contributions to AMR. Poultry, swine, cattle, and fish were the most frequent livestock mentioned as sources of AMR bacteria. Animal carriage and/or transfer of AMR determinants or bacteria was recognised as the primary contribution of livestock to the problem, while water, soil, and farming were predominant environmental contributions. We found that only 1 article out of 21 considered the OH approach as a framework for their sampling scheme, whereas 5 out 21 discussed all the three OH components. There were hardly any descriptions of humans or human waste as reservoirs for AMR in rural locations, and rural health centres or hospitals and wildlife were not represented. No studies identified mining as an anthropogenic activity driving AMR. More OH-oriented studies, with emphasis on molecular approaches—for identification and comparison of AMR genes—are sorely needed to understand better the existence of a network of interconnected transmission routes in rural Latin America and provide efficient strategies to prevent further AMR emergence.


Introduction
Antimicrobial resistance (AMR) is a global public health issue [1][2][3], which is occurring among a wide range of microorganisms with increasing prevalence [2]. AMR is a natural

Inclusion Criteria
The study period spanned from January 2001 until December 2018. Only original, peer-reviewed articles published in Spanish, Portuguese, and English were considered. A search strategy was designed using terminology associated with the three domains of OH and antimicrobial resistance in rural geographic environments for human populations in Latin American countries. We developed and refined key search terms with online databases prior to the article search (Table A1).

Exclusion Criteria
We excluded grey literature. Reviews and meta-analyses were excluded, but their reference lists were scanned. Articles not acknowledging or discussing all three OH components were excluded.
We consulted the following electronic databases concentrating on peer-reviewed articles only: PubMed (biomedical sciences), Web of Science (multidisciplinary), Scopus (multidisciplinary), and SciELO (multidisciplinary for LA and the Caribbean). We carried out the search from 13 November to 3 December 2018. All articles were uploaded to a Mendeley database [33].
A multi-step process was applied for the analysis of inclusion ( Figure 1). Our initial search produced 7936 articles. After screening for duplicates and evaluating the search criteria (AMR link, rural link, environmental link, animal or agriculture link, and LA country link), we screened 1151 publications. After evaluating the titles and abstracts of the articles in our Mendeley database, we created an initial list of 294 articles meeting the eligibility criteria. These selected articles were then read in full and evaluated for inclusion; 21 publications were finally included for the qualitative synthesis. Five reviewers conducted all stages of the scoping review, from relevance screening to data extraction. The five reviewers individually selected the studies for each phase of elimination. In the process, the reviewers met and discussed each study they had identified, and jointly agreed on including or excluding the study for analysis. If no agreement could be reached, the senior author (SH) decided.

Article Selection
We defined relevant publications as any original peer-reviewed article published between 1 January 2001 to 3 December 2018; in English, Spanish, and Portuguese language that presented data from LA countries only (Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, Dominic Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela); referred to a rural human setting and to agricultural and/or animal-based food production activities, linked to environmental aspects, and focused on the topic of AMR. To be included, an article needed to refer to any aspects of AMR and to partially or totally match the concept of AMR as defined by the WHO: "Antimicrobial resistance (AMR) is the ability of a microorganism (like bacteria, viruses, and some parasites) to stop an antimicrobial (such as antibiotics, antivirals, and antimalarials) from working against it. As a result, standard treatments become ineffective; infections persist and may spread to others" [34].
Considering an OH perspective, and using it as a screening tool, we searched for articles that linked to human populations in rural settings (e.g., rural hospitals or health services, rural communities, farms), had a connection with agricultural and/or animal production activities (e.g., cattle, fish, poultry, swine farming or animal keeping, transportation, slaughtering, meat-processing), and described a strong connection with AMR and the environment. All the inclusion criteria needed to be met, and thus, the criteria needed to be explicitly stated and/or discussed in the article. Finally, we excluded studies on AMR that focused only on topics that were remotely linked to agricultural and animal-based-food production activities, such as articles describing urban hospital settings or industrial food production settings; peripherally related to environmental aspects such as publications dealing with economic impacts; loosely linked to AMR discussing molecular/kinetics analyses of enzymes responsible for AMR, and describing research in parks and zoos within urban spaces, since they would not qualify as "rural". services, rural communities, farms), had a connection with agricultural and/or animal production activities (e.g., cattle, fish, poultry, swine farming or animal keeping, transportation, slaughtering, meat-processing), and described a strong connection with AMR and the environment. All the inclusion criteria needed to be met, and thus, the criteria needed to be explicitly stated and/or discussed in the article. Finally, we excluded studies on AMR that focused only on topics that were remotely linked to agricultural and animal-based-food production activities, such as articles describing urban hospital settings or industrial food production settings; peripherally related to environmental aspects such as publications dealing with economic impacts; loosely linked to AMR discussing molecular/kinetics analyses of enzymes responsible for AMR, and describing research in parks and zoos within urban spaces, since they would not qualify as "rural".

Data Management and Characterisation/Charting
We tabulated data extracted from the selected articles, including authors, year of publication, title, research objectives, DOI, URL, location of the study, language, and summary of the findings. We used a charting spreadsheet established a priori as a guide, which was established through team discussions when reviewing the preliminary results. If investigators from individual studies were contacted, their clarifications were included.

Analysing, Summarising, and Reporting the Results
The analysis and synthesis of literature included quantitative analysis (i.e., descriptive statistics) and qualitative analysis (i.e., content analysis). For the qualitative

Data Management and Characterisation/Charting
We tabulated data extracted from the selected articles, including authors, year of publication, title, research objectives, DOI, URL, location of the study, language, and summary of the findings. We used a charting spreadsheet established a priori as a guide, which was established through team discussions when reviewing the preliminary results. If investigators from individual studies were contacted, their clarifications were included.

Analysing, Summarising, and Reporting the Results
The analysis and synthesis of literature included quantitative analysis (i.e., descriptive statistics) and qualitative analysis (i.e., content analysis). For the qualitative analysis, reviewers extracted common themes that emerged from the findings, and the team discussed the results. Each article was analysed to identify the approach to study AMR and findings regarding each theme.

Literature Profile
A total of 19 articles were included in the analysis, and 2 articles were added after browsing their references ( Figure 1).
The 21 studies included in the analysis originated from 8 Latin American countries: Brazil, Ecuador, Colombia, Argentina, Chile, Mexico, El Salvador, and Peru, with Brazil providing the most articles (9 of 21). All articles but one were published in 2010 and onwards, peaking with five publications in 2014. All of them were written in the English language. Nineteen studies were funded by research funders, public agencies; two did not declare their funding source. One study was partially funded by a microbiological laboratory which supplied Salmonella spp. strains, and three studies were partially funded by private LA universities.
All studies were quantitative and had a cross-sectional design except for one, which was qualitative. Two articles addressed AMR only through molecular methods, six combined phenotypic profiling and molecular techniques (PCR), and eight only through microbiological methods. Three studies analysed microbiological and epidemiological data. Only one article included a chemical identification analysis of antibiotics in samples in addition to molecular genetic analysis. All eligible research works addressed AMR in a rural site, but four studies also took samples in urban or peri-urban sites for comparison. In one study, we assumed the location was rural (poultry production sites), based on current poultry production practices, but it was not explicit in the article. Table 1 presents the 21 publications included in our review and their characteristics are summarised in Table 2. Table 3 outlines the methods used in each of the research works.

Ecuador
Higher levels of AMR in bacteria from production versus household birds. Prevalence of AMR in production birds declined with bird age.

Argent-ina
Neither phenotypically ESBL 10 nor CB 11 -producing bacteria detected.  3.2. Antimicrobial Resistance through the One Health Lens Figure 2 shows AMR contributions from animal, environmental, and human domains, applying the OH perspective; it describes how various specific human activities, animalrelated factors, and environmental factors are connected based on the information extracted from the selected articles. extracted from the selected articles.
As shown in Table 1, in most articles, only two OH components were considered, either in the discussion or the description of the sampling procedures. Only Braykov et al. [37], Brisola et al. [38], dos Vieira et al. [47], Miranda et al. [48], and Palhares et al. [49] discussed all three components, but their study designs did not include sampling for all of them. Pehrsson et al. [50] took samples of human, animal, and environmental origin but did not discuss the relevance or impact of their results on animal health.

Human Contribution
Anthropogenic drivers of AMR included intensive and non-intensive (smallscale/extensive) animal production systems [35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51] and agricultural practices [49,50,[52][53][54], such as the use of recycled or composted animal or human manure as fertiliser. Resende et al. [53] showed microbiological evidence of survival of AMR bacteria after biodigestion treatment of cattle manure, which could contaminate soils when applied as fertiliser. Camotti et al. [52] found that each type of manure used as fertiliser in agricultural soils had a unique concentration of antibiotic residues and AMR genes, specific to the particular animal production system it originated from. Perhsson et al. [50] and Kalter et al. [42] identified inadequate human excreta management as an important human AMR-promoting factor in rural sites. Only three articles [35,42,50] identified unrestricted, unregulated, or recent use of antimicrobials in humans as a human contribution to the problem. Cicuta et al. [55] did not identify any human input but acknowledged the need to have an interdisciplinary approach to implement human and As shown in Table 1, in most articles, only two OH components were considered, either in the discussion or the description of the sampling procedures. Only Braykov et al. [37], Brisola et al. [38], dos Vieira et al. [47], Miranda et al. [48], and Palhares et al. [49] discussed all three components, but their study designs did not include sampling for all of them. Pehrsson et al. [50] took samples of human, animal, and environmental origin but did not discuss the relevance or impact of their results on animal health.

Human Contribution
Anthropogenic drivers of AMR included intensive and non-intensive (small-scale/ extensive) animal production systems [35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51] and agricultural practices [49,50,[52][53][54], such as the use of recycled or composted animal or human manure as fertiliser. Resende et al. [53] showed microbiological evidence of survival of AMR bacteria after biodigestion treatment of cattle manure, which could contaminate soils when applied as fertiliser. Camotti et al. [52] found that each type of manure used as fertiliser in agricultural soils had a unique concentration of antibiotic residues and AMR genes, specific to the particular animal production system it originated from. Perhsson et al. [50] and Kalter et al. [42] identified inadequate human excreta management as an important human AMR-promoting factor in rural sites. Only three articles [35,42,50] identified unrestricted, unregulated, or recent use of antimicrobials in humans as a human contribution to the problem. Cicuta et al. [55] did not identify any human input but acknowledged the need to have an interdisciplinary approach to implement human and animal health-oriented research. Only Perhsson et al. [50] gave direct evidence of the human role in the generation of AMR; they proved that humans modify microbiomes and resistomes in rural settings by interacting with animals and the environment by means of horizontal transfer of AMR determinants.
In contrast, Cicuta et al. [55] did not acknowledge any of the above-mentioned contributions of animals to the generation and spread of AMR, but only discussed the phenotypic resistance screening results and their likely cellular resistance mechanisms.
Armas-Freire et al. [35], Brisola et al. [38], Campioni et al. [39], Mattiello et al. [44], and Lopez et al. [51] showed strong evidence of the animal role as reservoirs or carriers of AMR genes obtained by molecular methods-allowing identification and comparison of AMR genes [56]-combined with phenotypic resistance testing-based on viable culturable bacteria [57]. Santamaria et al. [46] and Pehrsson et al. [50] used only molecular methods to study AMR, but the latter applied metagenomics to compare entire resistomes. Camotti et al. [52] used a combined molecular and chemical methodology to identify AMR genes and antimicrobial molecules. Braykov et al. [37], Kalter et al. [42], and Rodriguez et al. [45] included an epidemiological methodology and one of them provided sound evidence regarding the risk and protective factors for AMR presence in humans. One of the main risk factors for AMR were children's or household members' recent antibiotic use. At the same time, AMR was less often described among older children and those living in a community where a greater proportion of homes consumed home-raised chicken [42]. Most of the above-mentioned articles proposed a one-way transmission pathway of AMR genes from animals to the environment. However, Brisola et al. [38] and Pehrsson et al. [50] proposed a more complex scenario of interactions where the dissemination of AMR occurs simultaneously and in two opposite directions linking all reservoirs: human, animal, and environmental, although Brisola et al. [38] pointed at the animal reservoir as the origin.
Brisola et al. [38], Campioni et al. [39], Mattielo et al. [44], Santamaria et al. [46], Pehrsson et al. [50], Lopez et al. [51], and Camotti et al. [52] produced sound evidence regarding the role of the environment in the maintenance and dissemination of AMR. All these articles agreed that the faecally-contaminated environments are a persistent source or reservoir for AMR bacteria from which AMR could easily disseminate. Most of them considered animals as the contamination source but Pehrsson et al. [50] verified the contribution of both animal and human faecal matter in this contamination of the environment. Strong microbiological/epidemiological evidence was provided by Braykov et al. [37] and Kalter et al. [42] for the role of the environment in the spread of AMR.

Information Gaps
As shown in Figure 2, neither of the selected articles investigated nor identified the contribution of effluents of rural hospitals or health services to the environment and their impact on rural populations, animals, and ecosystems. Four articles included human faecal samples in their studies, but human waste collective discharges were not sampled. Likewise, the link between mining and AMR in rural settings was not the focus of any of the eligible studies, despite the role of metals as drivers of AMR [12] and the contribution of mining to metal pollution [58,59]. Moreover, wild animal reservoirs and/or their contribution to the AMR problem in rural locations were not discussed in the selected articles. Even though many studies published in Portuguese or Spanish language were found at the initial steps of the search, none of them met the inclusion criteria, so they were not represented in our selection.

Discussion
This review identified key contributors to AMR in LA considering the OH concept. The following anthropogenic activities were identified as drivers for AMR dissemination in rural Latin American settings: animal husbandry, fish farming, agriculture, and other related practices such as animal waste recycling. The carriage and/or transfer of AMR determinants were the most frequent animal contributions, in addition to the inadequate or unregulated use of veterinary antimicrobials and the role of food of animal origin in the spread of AMR bacteria in the human food chain. Water was the most commonly identified environmental contributor for AMR spread but also soil, farm/bird coops, vectors (flies), and pond sediments were also important contributors mentioned.
Nearly half of the eligible studies showed robust evidence confirming the human, animal, or environmental contributions to the generation or the spread of AMR. However, only one study [50] provided evidence embracing a OH framework to suggest a global scenario in which all the reservoirs-human, animal, and environmental-contribute to the problem, sharing AMR genes through horizontal transfer. Rather than illustrating a mere pathway, this work embraces the numerous interactions between human, animal, and environmental domains portraying an intricate network of AMR spread. Pehrsson et al. [50] took samples of human, animal, and environmental origin simultaneously and compared their resistomes, allowing them to produce strong evidence for the interconnectedness of human, animal, and environmental drivers of AMR. As these AMR drivers converge, the environment might function as both a reservoir and a bridge for antimicrobial determinants giving rise to other potential pathways of AMR transmission to non-contaminated wildlife, humans, and animals [12].
Most of the assessed articles studied the AMR problem from a single viewpoint or emphasised one of the OH components. Five studies [37,38,[47][48][49] considered the importance of animal, human, and environment inputs to the AMR generation and spread, but did not collect samples from all these interconnected sources or omitted discussing the results in an integrated manner. The reasons were not explicit.
One of the most important insights from our study is the scarce research on humans or human waste as sources of AMR determinants in rural locations described in the Latin American literature. Studies focusing on AMR in rural health centres or rural hospitals were not found in the articles eligible for analysis. However, the impacts of rural hospital effluents on the environment and hence, on human and animal microbiomes cannot be ignored [60], mainly because wastewater collection and treatment in rural settings are significantly reduced or absent compared to urban settings in LA [61,62].
We found some information gaps in the selected literature. Mining is an important economic activity in many countries in LA [63] and it has been identified as a contributor to the spread of AMR, however, we did not find any studies on the topic. Mining activities lead to the release of metal-containing effluents into the environment, driving AMR in bacteria due to shared mechanisms of resistance to both metals and antimicrobials [12].
Wildlife is a neglected likely significant contributor to the spread of AMR in rural LA. Although it was not identified as a contributor in the eligible articles, the interaction between humans, wild animals, and farm and domestic animals occurs in rural settings, sharing AMR bacteria [64]. Thus, studying wild animals' AMR gene sources and their genetic similarity in farm animals, human, and environmental reservoirs in LA, should be the focus for future research worth working on. More public health research focusing on wildlife is needed to better understand the impacts of human activities on the environment (habitat fragmentation, land-use change, urbanisation) and the role of wildlife speciesas reservoirs, melting pots, and/or vectors for AMR determinants-in the dissemination of resistance [65]. Our findings underscore the importance of adopting an OH approach as a framework for the design of future studies aimed at understanding the interconnections among its three components to assess AMR more efficiently and propose better strategies to prevent AMR emergence in LA.
Only one article used qualitative methods [43]. Qualitative approaches are useful when trying to elucidate the reasons behind practices, knowledge, attitudes and perceptions, and prove useful in understanding the complexity of AMR transmission pathways. We suggest incorporating a qualitative approach in AMR research since it could be a significant added value to quantitative studies. Mixed-methods approaches allow researchers to identify any contradictions between the quantitative and qualitative findings [66], and could be valuable for identifying deficiencies in sanitation and biosecurity practices in animal production and agricultural systems, knowledge and attitudes regarding these practices, and the structural and economic limitations contributing to AMR dissemination in LA rural settings.
On the other hand, the most robust evidence for the role of animals, humans, and/or the environment in the spread of AMR originated from studies combining molecular methods, phenotypic resistance screening methods, sound study design and sampling, and an integrated OH perspective. A few studies relied upon a combined phenotypic resistance testing and an epidemiological approach, thus identifying risk factors and protective factors for AMR. However, due to the limitations of culturing in assessing AMR, they could not give insight into the specific AMR determinants associated with the AMR phenotypes [56]. Phenotypic resistance profiling enables cultivation of target bacteria. However, assessing AMR through culture-based methods carries an inherent bias since these methods cannot detect cells in a viable but non-culturable state [57]. In contrast, molecular methods provide information regarding the underlying mechanism of resistance, identifying the determinants for that resistance, even if they are not always expressed in the host bacteria [56,67]. In addition, with the use of genomic tools, typing, comparing, and tracing specific allele profiles, it is now possible [68]. Thus, it is necessary to apply molecular methods along with phenotypic profiling methods to have a more detailed and complete picture when assessing AMR [57]. We believe that studies focusing on the total environment using microbiological, epidemiological, and molecular approaches in an integrative way are needed to better understand the existence of a network of interconnected transmission routes. Additionally, given the cross-sectional nature of the eligible studies, they could not demonstrate the directionality of their proposed pathways of transmission of AMR, which-in most cases-pointed at a one-way path only, from animals to the environment.
It is important to note that only one study in Brazil [47], mentioned that local governmental agencies were concerned with the results of the antibiotics' indiscriminate use in aquaculture. However, no other mention was made to the uptake of research findings by local authorities or any other local actors in any sector. This finding may imply the need of a more effective dissemination of scientific findings from academia to government agencies and local actors in LA. Likewise, an OH approach to provide robust evidence on AMR emergence and transmission is key to translate AMR's research results when designing public health and animal production policies.
We did not include grey literature, which we believe would have enriched our findings. There was no systematic way to search for country-level surveillance reports. Since most literature reviews only include publications in the English language, we purposedly looked for articles produced in LA written in Spanish and Portuguese languages, finding a considerable number during the screening process. However, none of them fulfilled the eligible criteria for this review. This limitation may be due to the kind of settings in which these studies have been conducted-urban as opposed to rural-and the AMR perspective applied, which may be one-sided, favouring any of the OH components but not comprising the three domains, as we specified in our inclusion criteria. Since AMR was recognised as a global threat to public health, virtually all countries adopted a national action plan to tackle the problem [69]; however, actions developed in LA may not have had an explicit focus on an integrated OH approach.
Other reasons may explain the scarcity of truly integrated OH research works in LA-as in other low-and middle-income regions. Establishing OH research involves facing barriers such as lack of OH training and expertise, and difficulty in establishing collaboration among multiple and cross-sectoral actors-resulting in scarcity of multidisciplinary training programs-and limited government support and research funding [70,71]. Funding bias could partially explain the absence of articles written in these languages: most comprehensive and well-funded AMR studies adopting an OH approach tended to be published in English.

Conclusions
This scoping review on AMR in rural settings in LA identified the human, animal, and environmental contributions to AMR using an OH lens, and pinpointed the information gaps on AMR transmission routes and AMR drivers in the literature. Human activities contributed to the spread of AMR through animal husbandry (mainly poultry), fish farming, agriculture, and animal waste recycling (composting). Farm animals contributed by carrying and/or transferring resistant genes or resistant bacteria. Main environmental contributors are faecally-contaminated water, contaminated soil or pond sediments, and farm environments.
Adopting an OH lens proved useful as a framework to determine whether the selected articles considered the impact of AMR on the three aspects of health, animal, human, and environment, and to what extent they did so. However, a small percentage of articles took into account the three OH components in the sampling or in the discussion. Thus, we recommend following the OH approach as a framework for the design of future studies-emphasising on the use of mixed methods and a combination of approachesmolecular-, epidemiological-, and culture-based-to tackle AMR more efficiently and to tailor strategies to prevent AMR emergence in the region, where these efforts are still scant and considerably needed.
Future research efforts should give more attention to the role of mining, wildlife, and rural hospitals' or health services' effluents on the emergence and spread of AMR in rural Latin America, given that these aspects were not identified in the selected literature and were considered information gaps.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board for humans and animal subjects of the Universidad Peruana Cayetano Heredia, protocol code 418-16-18.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available in Table 1. Additional data are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1 shows the full electronic search strategy used in this review. Table A1. Keywords (with synonyms) and syntax used for the literature search. ("antimicrobial drug resistances" OR "antimicrobial drug resistance" OR "antibiotic resistance" OR "drug resistances, microbial") OR" antibiotic resistance, microbial" OR "antimicrobial resistance" AND (rural OR "rural populations" OR "rural settings" OR "rural health services") AND (environment * OR water OR soil OR lixiviation) AND ("animal production" OR animal OR livestock OR agricultur * OR "animal husbandry" OR poultry OR food) AND