Environmental antimicrobial resistance and human infectious disease outcomes: a global one health systematic review and meta-analysis of observational studies

Abstract

Environmental antimicrobial resistance (AMR) is increasingly recognized as a key component of the global AMR burden and may be associated with human infectious disease outcomes through complex ecological exposure pathways. However, the overall magnitude and consistency of this relationship across environmental compartments and regions remain insufficiently quantified. This study systematically synthesized observational evidence on the association between environmental AMR contamination and human infectious disease outcomes within a global One Health framework. A systematic review and meta-analysis were conducted following PRISMA 2020 guidelines. Major databases, including PubMed/MEDLINE, Scopus, Web of Science, and Embase, were searched from January 2000 to March 2025. Eligible studies included in the review were published between 2005 and 2025. Observational studies reporting quantitative associations between environmental AMR exposure and human infectious disease outcomes were included. In total, 1,842 records were identified, and 32 studies met the inclusion criteria for meta-analysis, covering evidence from Asia, Europe, Africa, and the Americas. The pooled random-effects analysis showed that environmental AMR exposure was significantly associated with higher odds of human infectious disease outcomes (OR = 1.92, 95% CI: 1.55–2.38; p < 0.001). Subgroup analyses indicated variation across environmental compartments, with higher estimates observed in wastewater-related settings. Stronger associations were also observed in low- and middle-income countries compared with high-income countries. Overall, environmental AMR exposure shows a consistent positive association with human infectious disease outcomes across diverse settings. However, these findings represent observational associations and do not imply direct causal transmission. The results support strengthening environmental AMR surveillance and integrating environmental compartments into One Health antimicrobial resistance control strategies.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12879-026-14138-0.

Introduction

Antimicrobial resistance (AMR) is widely recognized as one of the most serious global public health threats of the twenty-first century, compromising the effectiveness of antimicrobial therapies and undermining the prevention and treatment of infectious diseases worldwide [23, 28]. The rapid emergence and spread of antimicrobial-resistant microorganisms have been driven largely by the misuse and overuse of antimicrobials in human medicine, livestock production, agriculture, and aquaculture, resulting in increased treatment failure, prolonged illness, healthcare costs, and mortality [10, 12, 15].

Although AMR has traditionally been viewed primarily as a clinical problem, growing evidence indicates that the environment plays a critical role in the emergence, persistence, and dissemination of antimicrobial resistance [5, 8]. Environmental compartments such as wastewater, surface water, soil, sediments, agricultural systems, and food production environments can serve as reservoirs of antimicrobial-resistant bacteria (ARB), antimicrobial resistance genes (ARGs), and antimicrobial residues [20]. These environments are continually exposed to selective pressures from hospital effluents, pharmaceutical waste, agricultural runoff, heavy metals, disinfectants, and untreated wastewater discharges, creating favorable conditions for the maintenance and horizontal transfer of resistance determinants [2, 19].

The environmental dimension of AMR is increasingly emphasized within the One Health framework, which recognizes the interconnectedness of human, animal, and ecosystem health [1]. Integrated surveillance approaches that combine human, animal, and environmental monitoring are increasingly recognized as essential for understanding AMR transmission dynamics and supporting coordinated control strategies [4]. Environmental reservoirs may facilitate the transmission of resistant microorganisms and resistance genes to humans through multiple exposure pathways, including contaminated water, food systems, irrigation practices, and recreational water exposure [17, 20]. Consequently, environmental AMR is now regarded as an important interface linking ecological contamination with human infectious disease risks.

Environmental AMR exposure may influence human infectious disease risks through ecological, microbiological, and exposure-related pathways. Higher levels of resistant bacteria, resistance genes, and antimicrobial residues may increase opportunities for human exposure, bacterial persistence, horizontal gene transfer, and colonization. Wastewater systems, particularly hospital and municipal wastewater, represent important exposure hotspots because they receive continuous inputs of resistant organisms, antimicrobial residues, and mobile genetic elements that facilitate resistance dissemination [7, 18]. Other pathways, including contaminated water, agricultural irrigation, food systems, and occupational exposure, may also contribute to human contact with resistant microorganisms. Recent evidence suggests that greater environmental AMR exposure is associated with increased detection of resistant organisms in human populations, although the magnitude of these associations varies with sanitation infrastructure, antimicrobial use patterns, environmental management, and surveillance capacity [7, 17, 20, 31]. These findings support a potential exposure–response relationship, whereby increasing environmental AMR burden may increase opportunities for human exposure and infectious disease risk, while direct causal transmission remains difficult to establish.

Despite growing research interest, the association between environmental AMR and human infectious disease outcomes remains insufficiently quantified. Existing observational studies report variable associations due to differences in study design, environmental matrices, AMR indicators, exposure assessment methods, and outcome definitions [7, 16]. Previous reviews have largely focused on environmental AMR occurrence rather than quantitatively synthesizing its relationship with human health outcomes across diverse environmental settings [24]. Therefore, a comprehensive meta-analysis is needed to clarify the magnitude, consistency, and public health relevance of environmental AMR-associated risks.

Therefore, this study aimed to systematically review and quantitatively synthesize global observational evidence on the association between environmental antimicrobial resistance and human infectious disease outcomes within a One Health framework. By integrating evidence across multiple environmental reservoirs and exposure pathways, this study seeks to clarify the magnitude and consistency of these associations and provide evidence to support integrated environmental AMR surveillance, risk assessment, and public health intervention strategies.

Methods

Study design

This study employed a systematic review and meta-analysis to synthesize global observational evidence on the association between environmental antimicrobial resistance (AMR) and human infectious disease outcomes within a One Health framework. The review specifically focused on the environmental–human interface of AMR transmission across interconnected ecological, environmental, and public health systems. The methodology was designed to integrate evidence from diverse environmental reservoirs and exposure pathways in order to evaluate the consistency and magnitude of reported associations between environmental AMR contamination and human infectious disease outcomes.

The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [13] guidelines to ensure transparency, methodological rigor, and reproducibility throughout the study process [26].

Prior to literature searching and data extraction, a detailed review protocol was developed that predefined the research question, eligibility criteria, search strategy, data extraction procedures, risk-of-bias assessment methods, and statistical analysis plan, including subgroup and sensitivity analyses. This prospective planning helped minimize selection bias and ensured methodological consistency throughout the review.

Although the protocol was not registered in the International Prospective Register of Systematic Reviews (PROSPERO) because of administrative and timing constraints before study initiation, the protocol was finalized prior to the review process and remained unchanged throughout the study. The protocol is available from the corresponding author upon reasonable request.

Research question and conceptual framework

The research question was developed using the Population–Exposure–Comparator–Outcome (PECO) framework [29] to define the scope of the review and guide study selection.

The population of interest included human populations living in, interacting with, or exposed to environments where antimicrobial-resistant microorganisms or antimicrobial resistance genes had been identified. The exposure was defined as environmental antimicrobial resistance occurring within environmental reservoirs such as wastewater systems, surface water, groundwater, soil, sediments, agricultural environments, food production systems, and animal–environment interfaces. Environmental AMR exposure was operationalized as the presence, abundance, or increased detection of antimicrobial-resistant organisms, antimicrobial resistance genes (ARGs), antimicrobial resistance indicators, or related resistance determinants measured within environmental compartments.

The comparator consisted of populations or settings with lower levels or absence of reported environmental AMR contamination, as defined within individual studies. The outcome of interest included human infectious disease outcomes, such as infection incidence, prevalence, colonization, clinically confirmed infections, and laboratory-confirmed resistant infections associated with environmentally relevant pathogens.

This review was conceptually guided by the One Health framework, which recognizes the interconnectedness of human, animal, and environmental health systems [27]. Within this framework, environmental reservoirs were considered potential interfaces for the circulation and dissemination of antimicrobial resistance across ecological boundaries. However, the framework was used primarily as an interpretive and integrative perspective rather than as evidence of direct transmission pathways. Accordingly, the review synthesized reported observational associations without inferring causality between environmental AMR exposure and human infectious disease outcomes.

Data sources and search strategy

A comprehensive literature search was conducted to identify studies investigating environmental antimicrobial resistance and its association with human infectious disease outcomes. Electronic databases, including PubMed/MEDLINE, Scopus, Web of Science, and Embase, were systematically searched for eligible studies published between January 2000 and March 2025. However, the earliest study meeting the predefined eligibility criteria was published in 2005. Additional records were identified through Google Scholar and citation tracking. Search terms related to antimicrobial resistance, environmental exposure, and infectious diseases were combined using Boolean operators. The detailed search strategies used for each electronic database are provided in Supplementary Table S1.

A combination of controlled vocabulary terms, including Medical Subject Headings (MeSH) where applicable, and free-text keywords was used to maximize search sensitivity and specificity [9]. Boolean operators (“AND” and “OR”) were applied to combine terms related to antimicrobial resistance, environmental reservoirs, environmental exposure pathways, and human infectious disease outcomes.

Google Scholar was not used as a primary database because of limitations in indexing consistency and reproducibility. However, it was used for backward and forward citation tracking of eligible studies to identify additional relevant publications not captured in the primary database search.

The core search strategy included the following concepts:

(“antimicrobial resistance” OR AMR OR “antibiotic resistance” OR “resistant bacteria” OR “resistance genes” OR ARGs) AND (“environment*” OR wastewater OR “surface water” OR groundwater OR soil OR sediment OR agriculture OR “food system” OR “environmental reservoir”) AND (“human infection” OR “infectious disease” OR colonization OR “resistant infection” OR “community-acquired infection” OR diarrheal disease OR “urinary tract infection”).

Database-specific modifications were made to accommodate differences in indexing systems and controlled vocabularies. Additionally, the reference lists of included studies and relevant review articles were manually screened to identify potentially eligible studies missed during the database search.

The complete database-specific search strategies are provided in the Supplementary Material to ensure transparency and reproducibility.

Eligibility criteria

Studies were eligible for inclusion if they were original peer-reviewed observational studies, including cross-sectional, cohort, case-control, or ecological study designs, that investigated environmental antimicrobial resistance and reported quantitative associations with human infectious disease outcomes.

Because environmental AMR can be measured using different indicators, eligible exposures included phenotypic resistance measurements (e.g., resistant organisms or multidrug-resistant bacteria), genotypic resistance markers (e.g., antimicrobial resistance genes), or composite environmental AMR indicators reported by individual studies. To ensure consistent classification of environmental exposures, environmental reservoirs were categorized based on the primary exposure compartment assessed in each study. Wastewater systems included municipal, hospital, and industrial wastewater sources; natural water systems included surface water and groundwater environments; soil and sediment systems included terrestrial and aquatic matrices containing AMR determinants; agricultural environments included farming systems, agricultural soils, crops, and irrigation-related exposures; food-associated systems included food products and food-processing environments; and animal–environment interfaces included livestock settings, manure, animal housing, and related environmental exposures. When studies involved multiple interconnected environmental compartments, classification was assigned according to the dominant exposure pathway most closely linked to the reported human health outcome. The detailed definitions and examples of each environmental exposure category are provided in Supplementary Table S7.

To be included in the quantitative synthesis, studies needed to report sufficient statistical information for effect-size extraction or calculation, including odds ratios (OR), risk ratios (RR), prevalence ratios (PR), confidence intervals, or raw data permitting effect-size estimation.

Only English-language studies were included to ensure consistency in data extraction, interpretation, and quality assessment. This restriction was applied for practical reasons and is acknowledged as a potential source of language bias.

Studies were excluded if they were review articles, editorials, commentaries, conference abstracts without complete datasets, methodological papers, or purely laboratory-based experimental investigations lacking epidemiological linkage to human infectious disease outcomes. Mechanistic molecular surveillance studies assessing only environmental resistance occurrence without human health outcome assessment were also excluded.

Study selection process

All identified records were imported into Zotero (version 7.0) for reference management, duplicate identification, and removal before the screening process. Study selection was conducted using a structured two-stage screening process performed independently by two reviewers to minimize selection bias and improve screening consistency.

During the first stage, titles and abstracts were screened according to the predefined eligibility criteria to identify potentially relevant studies. In the second stage, full-text articles were retrieved and assessed for final eligibility.

Any disagreements between reviewers during either stage were resolved through discussion and consensus. When necessary, the original articles were re-evaluated to ensure accurate inclusion decisions.

The complete study selection process was documented and summarized using a PRISMA 2020 flow diagram [13] to ensure transparent reporting of study identification, screening, inclusion, and exclusion.

Data extraction

Data extraction was independently conducted by two reviewers using a standardized extraction form. The extracted information included study characteristics (author, year, country, and study design), environmental exposure characteristics (environmental matrix and exposure pathway), antimicrobial resistance characteristics (AMR detection methods, AMR indicators, resistant organisms, resistance genes, and resistance phenotypes), human health outcomes, effect estimates, confidence intervals, adjusted variables, and other relevant methodological details. When available, information on environmental transmission interfaces and methodological characteristics relevant to risk of bias and quantitative synthesis was also extracted. Disagreements were resolved through discussion and consensus.

Representative examples of the extracted data from studies included in the quantitative meta-analysis are presented in Supplementary Table S3. The table illustrates study characteristics, environmental exposure pathways, AMR detection methods, AMR indicators, resistant organisms and resistance genes, reported health outcomes, effect estimates, and other key variables extracted from eligible studies to improve transparency and facilitate interpretation of the quantitative synthesis. Representative characteristics of selected studies from the qualitative synthesis are presented in Supplementary Table S2 to summarize the range of study designs, environmental matrices, AMR indicators, and reported health outcomes.

Environmental reservoirs were categorized a priori based on the dominant exposure pathway investigated in each study. Detailed definitions and classification criteria for each environmental category are provided in Supplementary Table S7. This approach was used to ensure consistent allocation of studies and improve interpretation of subgroup analyses.

AMR indicators included both phenotypic resistance profiles and genotypic resistance markers, including multidrug-resistant organisms, extended-spectrum β-lactamase (ESBL)-producing bacteria, carbapenem-resistant organisms, methicillin-resistant Staphylococcus aureus (MRSA), and major resistance genes such as blaCTX-M, mecA, and NDM-1.

Information related to potential human–environment transmission interfaces and environmental exposure pathways was also extracted to support interpretation within the One Health framework.

To enhance the reliability of the extracted data, both reviewers independently verified all entries against the original publications. Any discrepancies were resolved through discussion and, where necessary, re-examination of the source articles until consensus was reached.

Quality assessment and risk of bias

The methodological quality and risk of bias of included studies were assessed using validated appraisal tools appropriate for each observational study design.

Cohort and case-control studies were assessed using the Newcastle–Ottawa Scale (NOS) [21], while cross-sectional studies were evaluated using the Joanna Briggs Institute (JBI) critical appraisal checklist. Where additional detailed assessment was necessary, the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool was considered to evaluate broader domains of bias.

Studies were categorized as having low, moderate, or high risk of bias according to predefined criteria. Risk-of-bias assessment was independently performed by two reviewers, and disagreements were resolved through discussion and consensus.

The results of the quality assessment informed interpretation of the pooled findings and were incorporated into sensitivity analyses examining the influence of methodological quality on overall effect estimates. The detailed risk of bias assessment for individual studies is presented in Supplementary Table S4.

Data synthesis and statistical analysis

A narrative synthesis was initially conducted to summarize the characteristics of included studies, including environmental reservoirs, exposure pathways, AMR indicators, laboratory methods, and reported human infectious disease outcomes.

Where studies were sufficiently comparable, quantitative meta-analysis was performed to estimate pooled associations between environmental AMR exposure and human infectious disease outcomes. Because substantial methodological and contextual heterogeneity across environmental matrices, geographic regions, and observational study designs was anticipated, a random-effects modeling approach was considered most appropriate.

To standardize effect measures across studies, risk ratios (RR) and prevalence ratios (PR) were converted to odds ratios (OR) using the method proposed by Zhang and Yu [32]. Among the 32 studies included in the quantitative synthesis, 24 studies reported ORs directly, whereas 8 studies required conversion from RR or PR to OR. For rare outcomes (< 10% prevalence), RR and OR were considered approximately equivalent under the rare-disease assumption. Among the converted estimates, 3 studies involved outcomes with prevalence greater than 10%; therefore, these estimates were interpreted cautiously because ORs may increasingly overestimate the magnitude of association compared with RR when outcomes are common. The influence of effect-size conversion was evaluated through sensitivity analyses.

All pooled estimates were interpreted as observational associations rather than causal effects because of the inclusion of heterogeneous non-randomized study designs.

Meta-analysis was conducted using a random-effects model with restricted maximum likelihood (REML) estimation [6]. Between-study heterogeneity was quantified using tau-squared (τ²), Cochran’s Q statistic, and the I² statistic, and potential moderators were explored using meta-regression approaches [30].

Subgroup analyses were performed according to environmental matrix, AMR detection method, geographic region, country income category, and study design. Meta-regression analyses were additionally conducted to explore potential sources of heterogeneity associated with study-level variables.

Publication bias and small-study effects were assessed using funnel plots and Egger’s regression test [11] when sufficient studies were available. Where asymmetry was detected, trim-and-fill analysis was applied to evaluate the potential influence of publication bias on pooled estimates. Meta-analysis was conducted using R version 4.4.1 (metafor package) and STATA version 18.

Sensitivity analysis

Sensitivity analyses were performed to evaluate the robustness and stability of the pooled effect estimates under different analytical assumptions.

A leave-one-out analysis was conducted by sequentially excluding individual studies to assess their influence on the pooled estimates. Additional sensitivity analyses were performed by excluding studies with high risk of bias to evaluate the impact of methodological quality on the overall findings.

Separate analyses excluding studies requiring RR- or PR-to-OR conversion were conducted to assess the influence of effect-size transformation assumptions. Sensitivity analyses excluding ecological study designs were also performed to examine whether population-level studies disproportionately influenced the pooled estimates compared with individual-level observational studies.

Ethical considerations

This study was based exclusively on previously published literature and did not involve direct human participants, patient recruitment, or animal experimentation. Therefore, ethical approval and informed consent were not required.

Nevertheless, all stages of the review were conducted in accordance with accepted principles of academic integrity, transparency, and ethical research practice. Appropriate citation and acknowledgment were provided for all original studies included in the review.

Results

Study selection

A total of 1,842 records were identified, including 1,789 records from electronic database searches (PubMed/MEDLINE: 524; Scopus: 488; Web of Science: 421; Embase: 356) and 53 additional records identified through citation tracking and Google Scholar. After removal of 412 duplicates, 1,430 unique records were retained for title and abstract screening. During the initial screening phase, 1,262 records were excluded because they did not meet the predefined eligibility criteria, primarily due to lack of relevance to environmental antimicrobial resistance (AMR), absence of human infectious disease outcomes, or non-observational study designs.

At the full-text eligibility assessment stage, 168 articles were reviewed in detail. Of these, 121 studies were excluded. Some studies met more than one exclusion criterion; therefore, exclusion reasons were not mutually exclusive, and the total number of reported exclusion reasons exceeds the number of excluded studies. The recorded reasons for exclusion included lack of measurable environmental AMR exposure (n = 38), absence of relevant human infectious disease outcomes (n = 41), insufficient quantitative data for effect-size estimation (n = 32), non-original publications (n = 12), laboratory-based studies without epidemiological linkage to human health outcomes (n = 8), and duplicate or overlapping datasets (n = 3). Because some full-text articles met more than one exclusion criterion, the reported exclusion categories were not mutually exclusive and therefore their summed frequencies exceed the total number of excluded studies (n = 121). The overall study identification, screening, eligibility assessment, and inclusion process is summarized in the PRISMA 2020 flow diagram (Fig. 1).

Fig. 1.

PRISMA 2020 flow diagram illustrating the identification, screening, eligibility assessment, and inclusion of studies in the systematic review and meta-analysis. Note that exclusion reasons during full-text assessment were not mutually exclusive; therefore, individual studies could contribute to more than one exclusion category

Characteristics of included studies

The included studies were published between 2005 and 2025 and represented a broad global distribution, including Asia, Europe, Africa, and the Americas, reflecting the worldwide nature of environmental AMR research within a One Health context. Of the 47 studies included in the systematic review, 32 provided sufficient quantitative data for meta-analysis, while 15 contributed only to qualitative synthesis because they did not provide comparable effect estimates or sufficient statistical information for quantitative pooling. Across the 47 studies included in the qualitative synthesis, methodological approaches varied substantially, including cross-sectional, case-control, cohort, and ecological study designs, with cross-sectional studies being the most frequently reported. The qualitative synthesis included 47 observational studies published between 2005 and 2025, representing Asia, Europe, Africa, and the Americas. To illustrate the diversity of the evidence base, representative characteristics of selected studies are presented in Supplementary Table S2. Representative examples of extracted data from studies contributing quantitative effect estimates to the meta-analysis are presented in Supplementary Table S3.

Collectively, the included studies represented approximately 148,000 reported data units across different study designs; this figure does not represent the total number of human participants. Rather, it includes a combination of human study populations, environmental samples (such as wastewater, water, soil, and sediment samples), and surveillance-based observations reported by individual studies. Because the included studies used different sampling frameworks and unit definitions, these values were not combined as a single pooled sample size but were used only to describe the overall scope of the evidence base. Environmental AMR exposure was assessed across multiple ecological compartments, including wastewater systems, surface water, soil, sediments, agricultural environments, and food-chain systems. Wastewater environments, particularly hospital effluents and municipal wastewater systems, were the most frequently investigated environmental matrices among the included studies. Several studies reported high levels of antimicrobial-resistant organisms and resistance genes in wastewater systems, particularly ESBL-producing bacteria, carbapenem-resistant organisms, and clinically relevant resistance genes such as blaCTX-M and NDM-type genes ([5, 7, 18]; Supplementary Table S2).

A wide range of microbiological methods was employed across studies. Culture-based phenotypic techniques, including disk diffusion, broth microdilution, and automated susceptibility testing, were commonly reported among included studies [5, 7, 18]. Molecular approaches were widely applied among the included studies, particularly PCR-based detection of resistance genes such as blaCTX-M, blaTEM, mecA, and NDM-1. A smaller number of recent studies used metagenomic sequencing approaches to characterize environmental resistomes and identify broader patterns of resistance gene distribution ([5, 7, 18]; Supplementary Table S2).

Human infectious disease outcomes included gastrointestinal infections, urinary tract infections, wound and soft tissue infections, bloodstream infections, and other community-acquired infections associated with resistant pathogens. Outcome definitions varied across studies, ranging from clinically confirmed diagnoses to laboratory-based and population-level surveillance reports. Overall, the evidence base demonstrated substantial methodological and contextual heterogeneity, reflecting the complexity of studying environmental antimicrobial resistance within interconnected ecological and health systems.

Narrative synthesis of environmental AMR distribution

Antimicrobial-resistant bacteria (ARB) and antimicrobial resistance genes (ARGs) were detected across all environmental compartments assessed in the included studies, although detection frequencies varied considerably depending on environmental conditions, methodological approaches, and levels of anthropogenic pressure. Wastewater systems emerged as the most frequently reported high-burden environmental interface among the included studies, particularly hospital effluents and municipal wastewater systems, where resistant organisms and resistance genes were repeatedly detected ([7, 18, 19]; Supplementary Table S2). These environments represent important reservoirs because they receive continuous inputs of antimicrobial residues, resistant microorganisms, and mobile genetic elements.

Surface water environments, particularly those influenced by wastewater discharge, agricultural runoff, or urban contamination, also showed recurrent detection of AMR determinants, including resistant bacteria and resistance genes ([5, 16, 20]; Supplementary Table S2). Soil, sediment, and agricultural environments functioned as secondary reservoirs where resistant organisms and genes persist over time and may be reintroduced into human and animal systems through environmental cycling processes. Food-associated studies generally reported lower detection frequencies compared with wastewater and aquatic systems; however, resistant organisms were identified in food products and food-production environments, particularly where contamination pathways linked agricultural practices and animal production systems ([1, 27]; Supplementary Table S2).

The detection ranges presented in Table 1 are descriptive summaries extracted from selected representative studies included in the review and are not pooled estimates. These ranges represent the minimum and maximum AMR detection values reported within individual studies for each environmental matrix. Because the included studies differed substantially in sampling strategies, laboratory methods, target organisms, resistance genes, and AMR detection criteria, these values are presented only to illustrate variability in reported environmental AMR occurrence and should not be interpreted as directly comparable prevalence estimates across environmental matrices. Representative references supporting each reported range are provided in Table 1.

Table 1.

Reported AMR detection ranges from selected representative studies across environmental matrices (descriptive summary only)

Environmental matrix	Reported AMR detection range (%)	Sources of reported ranges
Wastewater	60–95	Karkman et al. [18]; Chau et al. [7]; Mekonnen et al. (2021)
Surface water	40–80	Kusi et al. [20]
Soil/agriculture	30–65	Kotwani et al. [19]
Sediments	35–70	Berendonk et al. [5]
Food chain	15–45	Al-Khalaifah et al. [1]

Note: Note: Detection ranges represent descriptive summaries from selected representative studies and are not pooled estimates. Values illustrate variability in reported AMR detection across environmental matrices and should not be interpreted as comparable prevalence estimates

Association between environmental amr and human infectious disease outcomes

A total of 32 observational studies were included in the quantitative meta-analysis, representing diverse geographic regions and environmental exposure settings within a global One Health framework. Of these, 24 studies provided effect estimates directly as ORs, while 8 studies required conversion from RR or PR to OR before pooling. Three converted estimates involved outcomes with prevalence exceeding 10%, and sensitivity analyses excluding studies requiring effect-size conversion demonstrated that the pooled estimate remained stable. Pooled analysis using a random-effects model with restricted maximum likelihood (REML) indicated that environmental AMR exposure, measured using different resistance indicators including resistant organisms, resistance genes, and environmental AMR contamination measures, was significantly associated with increased odds of human infectious disease outcomes (OR = 1.92, 95% CI: 1.55–2.38; p < 0.001).

Substantial heterogeneity was observed across studies (I² = 78%, τ² = 0.118, Cochran’s Q = 141.6, p < 0.001), and the 95% prediction interval ranged from 1.02 to 3.61, suggesting that although the direction of association remained positive, effect sizes varied across contexts and study settings. The forest plot summarizing individual and pooled estimates is presented in Supplementary Figure S1. Table 2 presents the subgroup analysis stratified by study design, showing variation in pooled effect estimates across observational study types.

Table 2.

Subgroup analysis by study design

Study design	No. of studies	OR (95% CI)	I² (%)
Cross-sectional	12	1.85 (1.45–2.35)	72
Case-control	7	2.10 (1.55–2.85)	68
Cohort	6	1.78 (1.30–2.40)	61
Ecological	7	2.05 (1.40–3.20)	80

Across all study designs, the association between environmental AMR exposure and human infectious disease outcomes remained consistently positive, indicating robustness across different epidemiological approaches.

Subgroup analysis by environmental exposure and income level

Subgroup analyses demonstrated variability in effect estimates across environmental exposure pathways and socioeconomic contexts. Table 3 summarizes the subgroup analyses of environmental antimicrobial resistance (AMR) exposure and human infectious disease outcomes across environmental matrices and country income levels. Among environmental matrices, wastewater-associated exposure showed the highest pooled association with human infectious disease outcomes (OR = 2.28, 95% CI: 1.75–2.97; I² = 71%), followed by surface water (OR = 1.74, 95% CI: 1.28–2.36; I² = 66%), food-chain related exposure (OR = 1.61, 95% CI: 1.08–2.40; I² = 54%), and soil/agricultural exposure (OR = 1.52, 95% CI: 1.10–2.10; I² = 59%). However, formal interaction testing indicated no statistically significant differences between environmental matrices (p-interaction = 0.11).

Table 3.

Subgroup analysis of environmental AMR exposure and human infectious disease outcomes

Subgroup	No. of studies	OR (95% CI)	I² (%)
Wastewater exposure	11	2.28 (1.75–2.97)	71
Surface water	8	1.74 (1.28–2.36)	66
Soil/agriculture	6	1.52 (1.10–2.10)	59
Food chain	4	1.61 (1.08–2.40)	54
LMICs	17	2.05 (1.65–2.55)	76
HICs	15	1.53 (1.18–1.98)	62

When stratified by country income level, studies conducted in low- and middle-income countries (LMICs) showed stronger associations (OR = 2.05, 95% CI: 1.65–2.55; I² = 76%) compared with high-income countries (HICs) (OR = 1.53, 95% CI: 1.18–1.98; I² = 62%), with a statistically significant difference between subgroups (p-interaction = 0.03).

Environmental categories in the subgroup analysis represent the primary exposure compartment assessed in each study rather than mutually exclusive ecological systems. Where studies involved multiple interconnected compartments, classification was assigned according to the main environmental source linked to the reported human outcome.

Heterogeneity and meta-regression analysis

Significant heterogeneity across studies was further explored using exploratory meta-regression analysis based on predefined study-level covariates. Country income level was associated with variation in effect size, with stronger associations observed in low- and middle-income countries compared with high-income settings (β = 0.31, p = 0.02). Environmental exposure type also showed evidence of contributing to between-study variability (β = 0.27, p = 0.03), particularly for wastewater-related studies, which demonstrated comparatively higher effect estimates. Subgroup findings are summarized in Table 3, while detailed meta-regression results are presented in Supplementary Table S5.

In contrast, microbiological detection method (phenotypic versus molecular approaches) was not significantly associated with differences in effect size (β = 0.09, p = 0.41). Study design showed a non-significant trend toward higher effect estimates in ecological studies (β = 0.12, p = 0.18). Overall, these exploratory findings suggest that contextual environmental and socioeconomic factors may contribute to the observed heterogeneity, although residual variability remained unexplained. The detailed meta-regression results are shown in Supplementary Table S5.

Publication bias

Assessment of publication bias indicated mild funnel plot asymmetry (see Supplementary Figure S2), suggesting potential small-study effects. Because the meta-analysis included 32 studies, Egger’s regression test was considered appropriate for assessing publication bias. The test demonstrated statistically significant asymmetry (p = 0.04). Trim-and-fill analysis suggested that up to three studies may be missing from the analysis. After adjustment, the association remained statistically significant but slightly attenuated (adjusted OR = 1.75, 95% CI: 1.40–2.20). Trim-and-fill analysis suggested that up to three studies may be missing from the analysis. After adjustment, the association remained statistically significant but slightly attenuated (adjusted OR = 1.75, 95% CI: 1.40–2.20).

These findings indicate that although publication bias cannot be excluded, its influence on the overall conclusions is likely limited. Detailed sensitivity analysis results are presented in Supplementary Table S6.

Sensitivity analysis

Sensitivity analyses confirmed the robustness and stability of the pooled effect estimates. The leave-one-out analysis demonstrated that sequential exclusion of individual studies did not materially alter the pooled estimate, which remained stable within a narrow range (OR approximately 1.80–2.05). Exclusion of studies with high risk of bias produced similar results (OR = 1.85, 95% CI: 1.50–2.30), with minimal changes in heterogeneity. Risk of bias classification used for this analysis is presented in Supplementary Table S4.

Additional analyses excluding studies requiring risk ratio or prevalence ratio conversion to odds ratios yielded consistent findings, indicating that effect-size transformation did not significantly influence the results. Similarly, exclusion of ecological studies did not materially affect the overall pooled estimate. Across all sensitivity analyses, the direction of association remained stable, supporting the robustness of the findings. The leave-one-out influence analysis is presented in Supplementary Figure S3. Detailed sensitivity analysis results are presented in Supplementary Table S6.

Discussion

This global systematic review and meta-analysis synthesized observational evidence on the association between environmental antimicrobial resistance (AMR) contamination and human infectious disease outcomes across diverse environmental compartments. However, the environmental AMR exposure definition varied across studies, including phenotypic resistance, resistance genes, and broader contamination indicators, which may have contributed to observed heterogeneity. Despite substantial between-study heterogeneity (I² = 78%), the direction of association remained consistent across study designs, environmental matrices, and geographic regions.

The observed relationship reflects the interconnected nature of environmental contamination and human health within complex ecological systems shaped by antimicrobial use, sanitation infrastructure, and environmental management practices [5, 18]. Rather than indicating direct transmission pathways, these findings likely represent aggregated epidemiological signals arising from shared exposure contexts where antimicrobial resistance determinants circulate across human, animal, and environmental interfaces.

Wastewater-associated environments demonstrated comparatively higher pooled effect estimates. This is consistent with their role as high-load mixing zones where hospital effluents, municipal wastewater, and agricultural discharges converge, creating conditions that favor persistence and amplification of resistant organisms and genes [18, 31]. Advances in wastewater monitoring technologies, including data-driven and artificial intelligence-supported approaches, may provide additional opportunities for improving environmental AMR surveillance and early detection of emerging resistance threats [22]. Other environmental compartments, including surface water, soil, sediments, and food-related systems, similarly contributed to the observed associations, underscoring the widespread environmental distribution of antimicrobial resistance across interconnected matrices [5].

Differences in effect estimates between low- and middle-income countries and high-income countries likely reflect disparities in wastewater treatment infrastructure, antimicrobial stewardship implementation, environmental monitoring systems, and broader public health capacity [25]. These contextual factors influence both exposure intensity and detection probability, thereby shaping observed epidemiological patterns.

From a global health perspective, these findings extend existing qualitative and surveillance-based evidence by providing quantitative synthesis across multiple environmental compartments and observational study designs. This strengthens the empirical foundation supporting environmental AMR as an integral component of infectious disease ecology, while emphasizing that observed associations are context-dependent and mediated through indirect exposure pathways rather than direct causal mechanisms.

Future research should prioritize longitudinal and spatially explicit study designs to better characterize temporal dynamics between environmental AMR exposure and infection outcomes. Standardized protocols for environmental AMR measurement, combined with genomic approaches such as whole-genome sequencing and metagenomic surveillance, are essential to improve comparability and strengthen environmental–clinical linkage resolution [3]. Expanding surveillance coverage in underrepresented regions remains critical to reducing geographic evidence imbalance.

Strengths and limitations

This review has several methodological strengths. It provides one of the most comprehensive global quantitative syntheses of environmental antimicrobial resistance and human infectious disease outcomes across multiple environmental compartments. The study followed PRISMA 2020 guidelines, applied predefined eligibility criteria, and used validated risk-of-bias tools appropriate for observational studies. Robust statistical approaches, including random-effects meta-analysis, subgroup analyses, meta-regression, and sensitivity testing, were applied to account for heterogeneity and assess stability of findings.

Several limitations should be acknowledged. All included evidence was observational, limiting causal inference and leaving potential for residual confounding. Variability in environmental AMR detection methods, including culture-based and molecular techniques, likely introduced measurement heterogeneity and reduced comparability across studies. Inclusion of ecological study designs introduces the possibility of ecological fallacy, where group-level associations may not accurately represent individual-level risk relationships.

Despite extensive analytical exploration, substantial heterogeneity remained, reflecting differences in study design, exposure characterization, outcome definitions, and regional surveillance capacity [14]. The predominance of cross-sectional studies limits temporal interpretation and prevents inference regarding directionality of associations. In addition, evidence from low-income settings remains comparatively limited, despite potentially higher exposure burdens.

Although publication bias was detected, its influence on pooled estimates was modest after adjustment, suggesting limited distortion of the overall findings.

Conclusion and recommendations

This global systematic review and meta-analysis identified a significant association between environmental AMR exposure and human infectious disease outcomes across diverse environmental settings and observational study designs. Environmental AMR exposure was associated with increased odds of infectious disease outcomes (OR = 1.92, 95% CI: 1.55–2.38), although the magnitude of association varied across study contexts.

The findings highlight the importance of environmental compartments, particularly wastewater systems, within the global AMR landscape and the need to incorporate environmental surveillance into One Health strategies. However, because the available evidence is observational, these findings should be interpreted as epidemiological associations rather than evidence of direct causal transmission.

Strengthening environmental AMR surveillance, improving wastewater and sanitation infrastructure, and expanding molecular monitoring capacity are important priorities, particularly in resource-limited settings. Future research should focus on longitudinal designs, standardized exposure measurements, and genomic approaches to better clarify the relationship between environmental AMR reservoirs and human infectious disease outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Gebre Ayanaw Alula and Shegasew Tessema conceptualized the study, developed the protocol, and designed the methodology. The authors performed data extraction, screening, and analysis. The authors also wrote the first draft of the manuscript, interpreted the results, and revised the manuscript for intellectual content. The authors read and approved the final version of the manuscript.

Funding

This study did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

All data used in this study were extracted from publicly available published articles included in this systematic review and meta-analysis. The search strategies, data extraction framework, and supplementary datasets supporting the findings of this study are provided in the Supplementary Materials. Additional data generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study is a systematic review and meta-analysis based exclusively on previously published studies. No primary data collection involving human participants or animals was conducted. Therefore, ethical approval and informed consent were not required.

Consent for publication

Not applicable. This manuscript does not contain any individual person’s data, images, or identifiable information.

Competing interests

The authors declare no competing interests.