Public health risks and control measures associated with live animal and animal product trade in traditional food markets in low- and middle-income countries: a scoping review

Abstract

Introduction

Traditional food markets (TFMs) support food security and livelihoods in low- and middle-income countries (LMICs), but the trade of live animals and products of animal origin may facilitate zoonotic and foodborne disease transmission. Evidence on associated risks and control measures remains fragmented across pathogens, market types and interventions. This scoping review maps the literature on zoonotic and foodborne public health risks, risk factors and control measures related to the trade of live animals and animal products in TFMs in LMICs.

Methods

Web of Science, PubMed and Scopus were searched on 24 April 2026 for primary, English-language studies published from 1 January 2017 to 31 December 2025. Eligible studies reported public health outcomes, risk factors or control measures associated with TFMs in LMICs. Reviewers screened titles/abstracts and full texts against predefined criteria, resolving disagreements by consensus. Data were charted and checked using a standardised pro forma; no methodological quality appraisal was conducted.

Results

87 studies were included; 61 (70.1%) reported public health risks and 35 (40.2%) examined control measures, with 9 (10.3%) addressing both domains. Most studies were observational and focused on poultry markets and avian influenza viruses (AIVs), particularly at the human–market interface. Reported risks included occupational exposure among market workers, exposure among community members and contamination of animal products and market environments. Common risk factors included poor hygiene, weak biosecurity, on-site slaughter, species mixing, inadequate waste management, prolonged animal holding and unsafe handling. Control measures included market closure, rest days, species separation, cleaning and disinfection, infrastructure improvements and policy-based interventions such as China’s ‘1110’ policy. Across distinct studies, several measures were associated with lower pathogen detection or modelled infection estimates, whereas other findings were non-significant or inconsistent. Implementation was constrained by economic losses, limited infrastructure, weak enforcement, low awareness and poor adherence to protective practices.

Conclusions

TFMs are important human–animal–environment interfaces where zoonotic and foodborne risks arise from interacting biological, structural and behavioural factors. The findings from this scoping review provide a rationale for context-specific, multilevel One Health approaches combining infrastructure improvements, biosecurity, surveillance, regulation, compensation and vendor engagement. However, because most evidence was observational and concentrated on AIV-focused poultry markets, the effectiveness and generalisability of these approaches require further evaluation across diverse TFM settings.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• This scoping review maps 87 studies and uses a One Health framework to integrate evidence on public health risks, exposure pathways, human health outcomes and control-measure implementation in traditional food markets. It shows that risk is produced by interacting market infrastructure, hygiene, biosecurity, animal management and human practices, rather than by live animals or animal-origin products alone.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The findings suggest potential value in adaptive, context-specific One Health approaches rather than isolated technical controls. Policy and research should evaluate combinations of infrastructure, water, sanitation and hygiene, biosecurity, surveillance, vendor engagement, market regulation and livelihood-sensitive incentives, while expanding evaluation beyond avian influenza virus-focused poultry systems.

Introduction

Traditional food markets (TFMs) are central to food security and livelihoods in many low- and middle-income countries (LMICs), but often operate with limited infrastructure, sanitation and regulatory oversight, increasing risks of food contamination and zoonotic transmission.^1–3 Following WHO terminology,⁴ TFMs include settings such as live bird and wildlife markets where animal and non-animal products are sold, including markets where live animals are housed, traded or slaughtered on-site. Since the COVID-19 pandemic, TFMs have received renewed attention as potential sites for zoonotic spillover and emergence of infectious diseases, including SARS-CoV, SARS-CoV-2, avian influenza viruses (AIVs) and novel pathogens.^4–12

The mixing of multiple animal species, close human–animal contact and inadequate hygiene practices at TFMs facilitate conditions for pathogen amplification, persistence and spillover, posing significant threats to global health security.^13,14 In response to these concerns, the WHO, the World Organisation for Animal Health and the United Nations Environment Programme have advocated for strengthened risk assessment and risk management of live wild animal trade in TFMs.⁴ However, disease control and risk reduction initiatives have often had limited success because of weak infrastructure and enforcement, vendor economic constraints and cultural practices that affect implementation.^15,16 Effective risk mitigation therefore requires a One Health perspective that recognises interconnected human, animal and environmental health and supports integrated approaches to zoonotic risk management in complex market systems.¹⁷

Research on TFMs spans multiple disciplines and provides insights into transmission dynamics and control strategies.^5,6,13,18,19 However, the evidence remains fragmented, often focusing on specific hazards, market types or interventions, such as AIVs in live bird markets (LBMs), wildlife meat trade, market closures, rest days, zoning or cleaning and disinfection protocols. Previous evidence syntheses have addressed public health risks^20–23 and selected control measures^18,19,24; however, few have integrated recent evidence into a broader understanding of the multidimensional challenges facing TFMs in LMICs. Such integrated analytical approaches are needed to connect hazard characterisation with public health risk assessment and the practical complexities of implementing control measures across diverse market settings.^6,13,17

In response to the World Health Assembly Resolution WHA75(23), which highlights the need to clarify the public health impacts of TFMs in the post-COVID-19 era,²⁵ this scoping review maps and characterises existing literature on public health risk and control measures associated with the trade of live animals and animal products in TFMs in LMICs. Specifically, the review examines two dimensions: public health risks, encompassing exposure scenarios, market-related and behavioural risk factors, and human health outcomes; and control measures, including their characteristics, reported outcomes and the barriers and facilitators influencing their implementation. This work adopts a One Health conceptual framework and an integrated analytical framework to synthesise and interpret this evidence. By organising findings across these dimensions, this review aims to describe how these topics have been studied in the literature and to identify key research patterns and knowledge gaps related to zoonoses and TFMs. Using the PCC framework, the review considered populations exposed through TFM-related interfaces as ‘the Population’, public health risks and control measures associated with live animals and animal products as ‘the Concept’, and TFMs in LMICs as ‘the Context’.

Methods

Study design

This scoping review followed the Joanna Briggs Institute Manual for Evidence Synthesis and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR).²⁶ The PRISMA-ScR checklist is provided in online supplemental material 1: online supplemental table S1. No prepublished protocol was used.²⁷

Information sources and search strategy

Exploratory questions were defined to guide the scope and eligibility criteria: which aspects of public health risks associated with the trade of live animals and animal products in TFMs in LMICs have been studied, and which aspects of control measure implementation to mitigate these risks have been examined. Two search strategies were developed using terms related to public health risks, animals, TFMs and control measures. Web of Science, PubMed and Scopus were searched on 24 April 2026 for publications from 1 January 2017 to 31 December 2025. The 2017 start date was selected to focus on contemporary primary evidence published after the 2016 systematic review of interventions in live poultry markets (LPMs),¹⁸ while 2025 was the most recent complete calendar year available at the time of the final search. Database-specific search strategies are provided in online supplemental material 1: online supplemental table S2.

Screening and selection of studies

Retrieved records from all databases were imported into a shared EndNote library. Potential duplicate records were identified using EndNote’s Find Duplicates function and were subsequently reviewed manually before removal. When duplicate records differed in bibliographic completeness, the most complete record was retained. Records were organised and reported according to the PRISMA-ScR flow diagram throughout the screening process. Three reviewers (ASF, AIVH and FES) independently screened titles and abstracts against predefined eligibility criteria, followed by full-text review for final inclusion decisions (AIVH, CZ and FES). Any discrepancies arising during title and abstract screening or full-text assessment were resolved through discussion among the reviewers until consensus was reached. English-language studies of any design conducted in LMICs were included if they reported: (1) health effects or zoonotic risks to humans from the trade of live animals or their products in TFMs; and/or (2) control measures implemented within TFMs to mitigate public health risks. The English-language restriction was applied because resources were not available for standardised translation and independent verification of non-English full texts during screening and data charting. Studies were excluded if they lacked accessible full texts; focused on products from industrial processing facilities or household settings; examined only animal-to-animal transmission without human health implications; or evaluated control measures implemented outside TFMs (eg, premarket animal vaccination programmes). Conference abstracts, editorials, commentaries and reviews without primary data were also excluded.

Data charting process and synthesis

Four reviewers (ASF, AIVH, CZ and FES) extracted and validated the data using a tabular pro forma.²⁸ The tabular pro forma was piloted on an initial subset of included studies and refined before full data charting. Each included study was charted by one reviewer and independently checked by a second reviewer. Any discrepancies were resolved through discussion among the reviewers until consensus was reached. For each included study, information on title, author, year, country, study aim, results and key findings relevant to the review was recorded, along with topic-specific details. For studies on public health risks, extracted data included study duration, animal or animal-product information, including the animal species or animal-derived product investigated; hazard information, including the pathogen or pathogen group, pathogen-specific taxonomic or typing information where reported, and pathogen category coded as viral, bacterial, protozoan or parasitic; sample source category coded as human, animal, food or environmental; study population, sample size, methods, comparator type, outcome types. For control measures studies, extracted data included the same general study characteristics, together with hazard information, outcome types, sample size, type of measure/intervention and challenges of implementation. Data were summarised using descriptive counts, basic qualitative analyses, tables, a Sankey diagram and narrative synthesis guided by the conceptual and analytical framework. The Sankey diagram was generated using RStudio (V.2025.05.0+496) and the networkD3 package. No formal methodological quality or risk-of-bias appraisal was conducted, consistent with the evidence-mapping purpose of a scoping review.

Conceptual and analytical framework

A One Health conceptual framework was adopted and tailored to this scoping review to organise and interpret the body of evidence and standardise the reported concepts. This framework integrates definitions of key terms from various domains, including food safety, public health, zoonoses and animal health. The definitions used in this review are provided in the online supplemental material 1: online supplemental table S3 and were primarily drawn from the One Health European Joint Programme Glossary, supplemented with selected additional definitions.^29,30

An analytical framework was developed to organise, guide and interpret the mapping and provide a schematic overview (figure 1). This analytical approach comprised two main dimensions: public health risks associated with infectious disease transmission in TFMs and control measures implemented to mitigate these risks. The public health risk dimension distinguishes studies describing risk-exacerbating conditions and exposure scenarios from those reporting human health outcomes. The control measure dimension distinguishes studies describing the characteristics of interventions, their reported outcomes and factors influencing their implementation, including barriers and facilitators. This structure was used to organise the scope of the review and to support consistent interpretation of the evidence across included studies.

Figure 1. Integrated analytical framework for analysing health risks and interventions in TFMs, focusing on infectious disease transmission associated with live animals and animal-derived products. TFM, traditional food market.

Data extraction and synthesis

The conceptual and analytical framework was used for the data extraction and summarisation process. For studies addressing public health risks in humans associated with infectious disease transmission in TFMs in LMICs, extracted data were classified into two analytical groups: (1) risk-exacerbating conditions and exposure scenarios and (2) human health outcomes. For studies addressing control measures in TFMs, extracted data were classified into three analytical groups: (1) intervention characteristics, (2) reported outcomes associated with interventions and (3) implementation factors including barriers and facilitators. Detailed subgroup definitions, rationale, inclusion criteria and extracted variables are presented in online supplemental material 1: online supplemental tables 4 and 5.

Author reflexivity

The BMJ Global Health Author Reflexivity Statement is provided in online supplemental material 2.

Results

Selection of sources of evidence

The search yielded 5974 records. After removal of 2989 duplicate records and three retracted records, 2982 records underwent title and abstract screening. Of these, 441 reports were sought for retrieval; 13 were not retrieved and 428 full-text reports were assessed for eligibility. 87 studies met the inclusion criteria and were included in the review (figure 2).

Figure 2. PRISMA-ScR flow diagram of study selection process. PRISMA-ScR, Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews.

Characteristics of included studies

Among the 87 included studies, 61 reported public health risk outcomes, 35 examined control measures in TFMs and 9 addressed both. The studies represented 23 country or territory settings: Bangladesh (n=19), Burkina Faso (n=1), Cameroon (n=1), China (n=35), Democratic Republic of the Congo (n=1), Egypt (n=1), Ghana (n=1), Hong Kong SAR (China) (n=4), India (n=1), Indonesia (n=3), Kenya (n=2), Lao People’s Democratic Republic (n=1), Libya (n=1), Malaysia (n=3), Mozambique (n=1), Nepal (n=1), Pakistan (n=3), Peru (n=1), Philippines (n=1), South Africa (n=1), Taiwan (n=1), United Arab Emirates (n=1) and Vietnam (n=5). Evidence was concentrated in China (n=35) and Bangladesh (n=19), whereas evidence from Africa was sparse and geographically dispersed, and Latin America was represented by one study from Peru. No studies were identified from Central America or the Caribbean. Figure 3 summarises the geographical, temporal, thematic, market-type, analytical and pathogen distribution of the evidence. The full dataset is available through Zenodo,³¹ and study characteristics are summarised in online supplemental material 1: online supplemental table S6.

Figure 3. Graphical overview of included studies by information category: (a) geographical distribution of studies across 23 country or territory settings, with shading indicating the number of studies per country; (b) temporal distribution of included studies by publication year and type of evidence; (c) distribution of studies by type of TFM; (d) study distribution across analytical categories; (e) heatmap showing the distribution of studied pathogens by study subjects and related animal hosts. TFM, traditional food market.

Public health risks in traditional food markets

General study characteristics

The 61 studies reporting on public health risks predominantly examined domestic poultry and AIVs, with outcomes assessed in humans and at the market interface. Bacterial zoonoses, particularly Campylobacter spp and Salmonella spp, were the second most frequently investigated pathogen groups. Online supplemental material 1: online supplemental figure S1 provides a schematic overview.

Risk-exacerbating conditions and exposure scenarios

Human exposure interfaces in TFM contexts

15 studies described human exposure interfaces in TFMs that were linked to measurable health risks (table 1).^32–46 10 studies focused on AIVs, whereas the remaining five reported on Salmonella spp (n=1), multidrug-resistant Escherichia coli (n=2), shared antimicrobial-resistance profiles (n=1) or hepatitis E virus (n=1). Reported exposure interfaces included direct hand contact with contaminated raw chicken meat, occupational contact with live poultry or contaminated market environments, handling or slaughtering poultry purchased from markets, visiting LPMs and consuming contaminated animal-derived products, including raw or undercooked pig liver and chicken meat. Risk estimates were based on positive human sampling and comparisons across occupational groups. Several studies also identified risk among customers or people exposed to TFM-sourced animal products at home, indicating that TFM-associated risks extend beyond market premises.

Table 1. Human exposure interfaces in TFM contexts by population type.

Exposure interface	Pathogen detail	Risk endpoint	Risk estimation	Ref
Population: market workers (vendors, handlers, traders, poultry workers)
Direct hand contact with raw contaminated chicken meat	Salmonella spp, S. Typhimurium, S. Enteritidis and S. Dublin	Detection of Salmonella.	Hand swabs of retailers showed 9.86% positivity for S. enterica.	32
Human-animal: occupational exposure	Escherichia coli	Shared sequence types between chickens and vendors.	MDR E. coli was found in 38.1% of chicken vendors’ stool samples.	33
		Presence and dissemination of MDR E. coli.	E. coli was detected in 4.60% of poultry sellers’ hand swabs. MDR bacteria were identified in isolates from poultry sellers’ hand swabs, with 100% resistance to ampicillin and high resistance across multiple antibiotic classes (62.5–87.5%).	41
Contact with live poultry	AIVs (H3, H5, H6, H9, H5N1, H9N2, H7N9 and non-subtyped)	Unusual mortality event in crows.	13.9% (21/151) of LBM workers tested positive for influenza A (12.6% of nasal swabs, 4% of throat swabs). Retail LPM poultry workers had the highest H9N2 risk (OR=9.9; 95% CI 2.4 to 40.9), followed by wholesale LPMs (OR=6.4; 95% CI 1.8 to 22.8).	3435
		Touching sick or dead poultry.	Rural cases: 3.70% (Wave 3), 10.53% (Wave 4), 20.97% (Wave 5), p=0.03. Urban cases: 11.11% (Wave 3), 6.67% (Wave 4), 4.00% (Wave 5), p=0.368. Contact with sick or dead poultry was less frequently reported overall but showed a notable increase in rural cases by Wave 5.	36
		Handling different poultry types. Contact with contaminated market environments.	Waterfowl: ~6× higher A(H5) prevalence than chickens; no significant difference among chicken breeds (Broiler, Desi, Sonali). Broilers: 3.8×higher A(H9) prevalence than waterfowl and 1.6×higher than Desi/Sonali. Waterfowl increase A(H5) risk, while broilers increase A(H9) risk. A(H5): No significant difference between slaughter and stall areas. A(H9): Prevalence in slaughter areas ≈2× higher than in stall areas. Slaughter areas pose greater A(H9) contamination risk.	38
		Handling live poultry.	Human influenza A detection was 0.83% (2/240). Both positives were H9N2; HA gene phylogeny showed clustering with poultry closest to A/chicken/Pakistan/425/2019.	42
			Market workers: 14.1% (99/702) had ≥1 respiratory specimen positive for influenza A virus. Among positive workers (n=99): A(H9): 28.3%, A(H5): 2.0%, A(H5)+A(H9): 3.0%, non-subtypeable: 66.7%.	43
			Market workers: influenza A detected in 1.9% (16/828); one sample had a partial avian-like H9N2 PB1 gene.	44
		Persistence of the virus in the market.	Probability of viral persistence increased from 0.05 (n=2000) to 0.7 (n=15 000) with low within-flock transmission—a 1300% rise for a 650% increase in bird population. Under high environmental transmission, persistence rose from ≈0.60 (n=5000) to ≈0.88 (n=10 000). Larger daily bird populations markedly elevate the likelihood of virus persistence in markets. At 10% turnover, persistence probability reached 1 with only 500 birds under baseline parameters, while a 25% reduction in transmissibility required ≈5000 birds to reach >0.99 probability. At 40% turnover, increasing transmissibility, shedding and infection duration by 75% achieved persistence=1 with ≈1000 birds. At 70% turnover, persistence=1 required >10 000 birds. Higher turnover rates shorten exposure duration and lower the probability of viral persistence.	39
Population: community members
Consumption of contaminated pig liver	HEV	Human HEV-4b infection linked to contaminated pig liver from markets.	HEV subtype 4b was found in 13 of 14 human sera; strains showed 97.4–99.6% identity with pig liver strains, supporting zoonotic foodborne transmission.	37
Consumption of contaminated chicken meat	E. coli, Klebsiella pneumoniae, Acinetobacter johnsonii, Citrobacter freundii	Community exposure to shared resistance profiles associated with regular wet market contact.	221 ARG subtypes shared across human faeces, chicken faeces, chicken carcasses and environment. Significant differences in ARG abundance between human and market samples (p<0.05).	45
Visiting LPMs	AIV A(H7N9)	Visiting LPMs.	Rural cases: 59.26% (Wave 3), 52.63% (Wave 4), 74.19% (Wave 5), p=0.14. Urban cases: 61.11% (Wave 3), 66.67% (Wave 4), 68.00% (Wave 5), p=0.891. Visiting LPMs was consistently the most common exposure pathway in both rural and urban cases across all waves.	36
Exposure to live poultry in the market	AIV A(H5N6)	Outbreak investigation.	Patient 1: visited a nearby LPM where there were live chickens and ducks for sale every 2–3 days before the disease onset. Patient 2: he and his wife bought 12 chicks and 15 ducklings from the local LPM and raised them in their backyard; 8 chickens subsequently became ill in June. His wife slaughtered these sick chickens and stored them in the refrigerator. He had cooked and eaten these chickens within 10 days before disease onset. In addition, he visited the local LPM within 10 days before disease onset. There were H5N6 or H5 positive environment samples on the premises where the three patients lived or visited 10 days before their disease onset. Patient 1, the positive percentage of H5N6 was 6.7% (3/45) in the LPMs he visited. Patient 2, the positive percentage of H5N6 was 11.1% (2/18) and 33.3% (1/3) in his family and the LPM he visited, respectively.	40
	AIV A (H10N5)	Outbreak investigation.	Patient 4 had direct exposure to live poultry through slaughtering a duck purchased from market vendors; epidemiological tracing linked the infection to mobile vendors, wholesale market and upstream farm, with H10N5-positive samples detected along the supply chain.	46

AIVs, avian influenza viruses; ARG, antibiotic resistance gene; HEV, hepatitis E virus; LBM, live bird market; LPM, live poultry market; MDR, multidrug-resistant; TFM, traditional food market.

Market-related risk factors

30 studies identified market-related risk factors associated with pathogen transmission (online supplemental material 1: online supplemental table S7).^{32,33,35,37,41,43,47–70} Most studies focused on AIVs in LBMs and LPMs, although bacterial pathogens, including Salmonella spp, Campylobacter spp and multidrug-resistant E. coli, were also reported. Poor hygiene and sanitation were the most frequently reported risk factors and included inadequate cleaning and disinfection, contaminated surfaces, stagnant water, weak biosecurity, limited inspection, poor slaughter hygiene and inadequate storage conditions. These conditions were associated with higher pathogen detection in poultry, carcasses, meat products and market environmental samples.^{31,41,47–50,63,64} Other recurrent risks included on-site slaughter without separation, waste and offal accumulation, contaminated rinse-water reuse, species mixing, lack of animal segregation, poultry from backyard or unknown sources, high animal density, longer market operation, unsafe handling of sick or dead birds, and presence of rodents.^{43,48–53,55,57,59,65,70} Associations with market type and structure were heterogeneous across urban, periurban, rural, retail and wholesale settings.^{35,38,43,47,48,54–57,66} Comparisons with supermarkets, hypermarkets or industrial processing environments generally showed higher bacterial contamination or antimicrobial resistance in TFMs, although not consistently for all pathogens.^{33,37,58,63,64,67–69} Pathogen detection also varied by sampling location, with slaughter areas, floors, cages, pluckers and cutting boards often showing high positivity.^{49,54,56,60,62}

Human behavioural or demographic risk conditions

16 studies identified behavioural or demographic risk factors associated with zoonotic disease transmission in TFMs (online supplemental material 1: online supplemental table S8).^{32,34,43,47,51–54,61,63–65,71–74} Most were behavioural and concentrated among market workers. They included poor hand hygiene, inadequate cleaning, unsafe slaughter or handling, prolonged contact with live animals or contaminated materials, hand defeathering, rinse-bucket use during slaughter, limited freezer storage, poor sanitation during chicken preparation and limited use of protective measures during poultry, meat, live-animal and wild-meat activities.^{32,34,43,47,51–54,61,63–65,71–74}

Human health outcomes

Reported zoonotic infections in humans

13 studies reported human AIV infections associated with TFM exposure: 7 among market workers and 6 among non-market workers. Among workers, infections were linked to contact with live or dead poultry and confirmed through respiratory swabs and epidemiological links to market settings. Several studies also reported asymptomatic AIV infections in workers without influenza-like illness.^62,75–80 Among non-market workers, cases included a child who walked through an LPM, household exposure to poultry purchased from a TFM and clusters with documented LPM exposure; in several cases, market links were supported by environmental sampling.^46,81–85 Additional details are provided in online supplemental material 1: online supplemental table S9.

Seroprevalence in exposed populations

10 studies assessed serological evidence of zoonotic infections among workers in TFMs (online supplemental material 1: online supplemental table S10).^{35,43,75,80,86–91} These studies reported seroprevalence of AIV, seroconversion to specific AIV subtypes and occupational risk factors associated with TFMs.^{35,43,75,80,86–88} Other studies documented seropositivity for Rickettsia typhi, Orientia tsutsugamushi, Leptospira spp and MERS-CoV among market workers.^89–91 Significant occupational risks were associated with direct animal contact, including camel handling, medication administration and equipment cleaning; adjusted ORs reached 20.9, with host factors such as diabetes further increasing risk.⁹¹

Epidemiological associations or burden estimates

14 studies provided epidemiological associations between TFMs exposures and zoonotic disease burden in humans (online supplemental material 1: online supplemental table S11). 12 studies examined AIVs, 1 investigated MERS-CoV and individual studies assessed O. tsutsugamushi, Rickettsia spp and Leptospira spp.^{36,43,75,87–89,91,92} Among market workers, studies reported AIV RNA detection, seroprevalence, seroconversion or exposure-associated risk estimates, while the MERS-CoV study reported seropositivity among camel-exposed animal market and slaughterhouse workers.^{43,75,77,87–89,91} Community-focused studies on AIV exposure, human H7N9, H9N2 or H5N6 infections were associated with visits to LPMs, contact with live or freshly slaughtered poultry, and poultry purchased from markets. Several case investigations also reported concurrent environmental detections or phylogenetic relatedness between human and market or poultry samples from epidemiologically linked markets.^{36,62,81–83,85,92}

Control measures and interventions in traditional food markets

Control measure characteristics and levels of implementation

35 studies examined control measures in TFMs, mostly in live bird or poultry markets and mainly targeting AIVs in domestic birds or market environments. Fewer studies addressed other market types, animal groups or bacterial, parasitic and other viral hazards (online supplemental material 1: online supplemental figure S2).

Control measures were classified across macrolevel, mesolevel and microlevel interventions.^{36,49,51,60,61,93–109} Macrolevel measures, applied by authorities or policymakers, included market regulation, market closure, official supervision, infrastructure improvements, the 1110 policy, culling and waste management.^{36,51,61,93–103} Mesolevel measures, enforced at the market level (eg, by managers), were mainly engineering and environmental controls such as clean water supply, smooth and washable surfaces, and rest days.^{49,60,103–107} Microlevel measures, applied by vendors or workers, included animal handling practices and cleaning/disinfection.^{51,53,61,103,104,107–112} This classification is not absolute as some measures span multiple levels. Most studies assessed interventions in LBMs and LPMs, focusing primarily on AIV in birds or the environment, and some examined multiple interventions simultaneously. Online supplemental material 1: online supplemental table S12 summarises the identified interventions, including their definitions, application, targets, hazards and affected populations.

Reported outcomes associated with control measures

31 studies reported outcomes associated with control measures in TFMs. 13 included macrolevel measures such as market closure, 12 included mesolevel measures such as rest days and 13 assessed microlevel measures such as personal protective equipment (PPE) use or separation of sick from healthy animals, with some overlap across categories. Reported outcomes varied across intervention types, study designs, settings and outcome measures. Some studies reported associations with lower pathogen detection, prevalence or modelled human infection estimates, whereas others reported non-significant, inconsistent or unquantified findings. Given this heterogeneity, findings were interpreted as reported associations or suggested effects rather than evidence of causal effectiveness, and interventions were not ranked by comparative effectiveness. Table 2 provides an overview of reported outcomes by implementation level.

Table 2. Reported control measures and associated outcomes in traditional food markets.

Control measure	Intervention/comparator	Outcome assessed	Reported estimate	Ref
Interventions to be applied at macrolevel by government, country or state authorities, policymakers
Market regulation	Market inspections.	Market hygiene. Regular inspection.	Markets without oversight appeared dirtier (magnitude NR). Markets with routine inspection were cleaner and better organised (magnitude NR).	93
	Absence of veterinary inspection.	Odds of Campylobacter spp positivity in birds.	Absence of veterinary inspection was associated with higher odds of being classified in the high-risk category (POR=1.75, 95% CI 0.79 to 3.91; p=0.167); not statistically significant.	51
Market infrastructure	Improved market infrastructure by government policy.	AIV isolation in birds.	OR=0.76 (95% CI 0.51 to 1.14); not statistically significant.	61
	Separate slaughter from selling.	Model-predicted probability of AIV positivity in market stalls.	The model predicted AIV positivity of 51% when slaughtering and selling were not separated and 32% when they were separated.	107
		Worker respiratory-specimen RT-PCR positivity for A(H5), A(H9), or non-subtype influenza A.	Separate slaughtering and selling areas were not significantly associated with worker RT-PCR positivity (aOR=0.71, 95% CI 0.44 to 1.14; p=0.153).	43
Market closure	Periods before vs after COVID-19-related LPM closure and reopening.	AIV prevalence in wholesale and retail markets: AIV prevalence in wholesale. AIV H9 prevalence in wholesale. AIV prevalence in retail markets. AIV H9 prevalence in retail markets.	After reopening, observed AIV positivity. Increase from 25.19% to 45.33%. Increase from 19.63% to 28.89%. Increase from 28.54% to 34.00%. Increase from 22.22% to 32.44%.	100
	Implemented intervention package and modelled temporary, seasonal and permanent closure scenarios over four epidemic seasons.	Modelled human H7N9 infections.	Implemented package vs no intervention: estimated 59.5% reduction (95% CI 48.5% to 70.5%); temporary closure: 67.2%; seasonal closure: 75.6%; permanent closure: 86.6%.	101
	LPM-closure vs non-closure areas across three epidemic waves.	Proportion of human H7N9 cases reported from non-LPM-closure areas.	Proportion of cases from non-LPM closure areas: Wave 3: 80.0%, Wave 4: 70.6%, Wave 5: 89.7%; p=0.034 (statistically significant increase in Wave 5).	36
	Permanent LPM closure vs no closure or the preclosure period.	Human AIV infections or daily human AIV infections.	At one permanent-closure site, the model estimated a 97.0% reduction in daily human A(H7N9) infections (95% CI 94.0% to 100.0%).	97–99
	Long-period LPM closure (>14 days) vs no closure or the preclosure period.		Across long-period-closure sites, modelled reductions in daily human A(H7N9) infections ranged from 48.0% to 98.0%. In Shanghai, closure for 10–13 weeks was associated with RRR=0.98 (RR=0.02, 95% CI 0.008 to 0.068). A separate study recommended closure for at least 3–4 weeks when human cases persisted.
	Short-period LPM closure (<14 days) vs no closure or the preclosure period.		Across short-period-closure sites, modelled reductions in daily human A(H7N9) infections ranged from 73.0% to 96.0%. In a separate ecological study, a 2-week closure was not statistically significant (RR=0.36, 95% CI 0.31 to 1.03; p>0.05).
	Recursive LPM closure, defined as repeated 1–2 day closures, vs no closure or the preclosure period.		Across two recursive-closure sites, the model estimated reductions of 34.0% (95% CI 15.0% to 70.0%) and 71.0% (95% CI 47.0% to 97.0%) in daily human A(H7N9) infections.
	Partial LPM closure, with markets outside the closure area remaining open.		Partial closure was associated with a reported RRR of 0.35. In Shenzhen, partial closure during the first three epidemic waves was not statistically significant (p>0.05).
	7 and 21 days.	Controlling AIV outbreaks and its effect on trading practices and economic factors.	7-day market closure resulted in informal trading outside LBM and reduced traceability, while 21-day closure caused significant economic losses for vendors, who often shifted to other occupations or alternative trades.	94
	10-day and 3-day LPM closure scenarios.	Modelled human H7N9 infections and exposed/latent population. 10 days closure effect on AIV cases. 3 days closure effect on AIV cases.	The 10-day closure scenario reduced the modelled infection rate by 70%. The 3-day closure scenario reduced the modelled exposed/latent population by 40%.	96
	LPMs market closure during an epidemic.	Human A(H7N9) case recurrence after market reopening. Short-term change in human A(H7N9) infection rates after LPM closure. Predicted human A(H7N9) cases with vs without LPM closure.	Recurrence of human AIV cases (magnitude NR). The study reported lower human A(H7N9) infection rates following market closure (magnitude NR). Without market closure, human cases were predicted to be 450 in the 2014 epidemic wave.	95
Culling	Culling of small number of domestic poultry.	AIV prevalence in birds.	The model suggested that culling a small number of domestic poultry had little short-term effect and no sustained long-term effect; magnitude NR.	95
	Early outbreak phase; interventions applied at different time points.	Early vs delayed implementation during outbreak simulations.	Early culling → reduces probability of spillover events. Delayed culling → increases number of human cases. Effect dependent on timing and transmission parameters.	127
Waste management	No management: litter accumulates.	Odds of effect of litter accumulation on Campylobacter spp positivity in birds.	Accumulated litter was associated with higher adjusted odds of the high-risk category (aPOR=19.67, 95% CI 3.01 to 128.52; p=0.002).	51
	Disposal of dead birds in municipal waste or open pits.	Campylobacter spp positivity risk in birds. Municipal waste. Open pits.	Municipal-waste disposal: POR=1.40 (95% CI 0.34 to 6.38; p=0.841); not statistically significant. Open-pit disposal: POR=1.44 (95% CI 0.60 to 3.45; p=0.416); not statistically significant.	51
Interventions to be applied at mesolevel by market authorities, market managers
Rest day	No rest day vs one rest day.	AIV prevalence in stalls.	No rest day was associated with higher adjusted odds of positivity (aOR=1.35, 95% CI 1.12 to 1.63).	104
Biosecurity	FAO-supported biosecurity measures in intervened vs non-intervened LBMs.	Risk of HPAI H5N1 positivity in LBM environments.	The risk of HPAI H5N1 positivity did not differ significantly between FAO-intervened and non-intervened LBMs (RR=1.10, 95% CI 0.44 to 2.76; p=0.413).	105
	FAO intervention combined. (Closed waste containers, vehicle disinfection, wild bird exclusion, regular cleaning and disinfection of floors and stalls, and provision of clean water).	AIV incidence rates in the environment of: FAO intervened markets. Non-FAO intervened markets.	FAO market IR=0.166 (95% CI 0.099 to 0.279). Non-FAO market IR=0.238 (95% CI 0.159 to 0.356). Between-group difference was not statistically significant (p=0.372).	60
	Low hygiene stalls vs high hygiene stalls.	Odds of stall-level AIV positivity in pooled environmental samples.	In the multivariable GEE model, stalls with a hygienic score <3 had higher odds of AIV positivity than stalls with a score ≥3 (OR=3.10, 95% CI 1.70 to 5.60; p<0.001).	103
	Previous training in good market practices vs no training.	Prevalence of ESBL-producing E. coli.	Lower ESBL prevalence among trained vendors (20%) compared with untrained vendors (44%). Lower ESBL prevalence among vendors with≥10 years’ experience (29%) compared with <10 years (71%) (p>0.05; not statistically significant).	113
	Combined interventions (culling/removal of chickens, control of entry, cleaning and disinfection).	Modelled cumulative daily H9N2 prevalence within the LBM.	The model suggested limited effects for individual measures and lower environmental transmission under cleaning and disinfection. Under the combined intervention scenario, the model estimated a reduction in prevalence of >80%.	70
Engineering controls	Partial lid vs no lid on the mechanical defeathering machine.	Modelled aerosol escape and airborne avian 18S rRNA concentration at the worker breathing zone.	CFD simulations estimated reductions in aerosol escape of 66.3% for 5 µm particles, 65.4% for 10 µm particles, 75.0% for 20 µm particles and 99.4% for 50 µm particles. Field testing reported a 57% lower airborne avian 18S rRNA concentration at the worker breathing zone with the partial lid.	106
	Covering defeathering machine.	Worker respiratory-specimen RT-PCR positivity for A(H5), A(H9) or non-subtypeable influenza A.	Use of a covered defeathering machine was associated with lower odds of worker RT-PCR positivity (aOR=0.55, 95% CI 0.32 to 0.96; p=0.035).	43
	Boiling poultry before defeathering.		Boiling poultry before defeathering was associated with lower odds of worker RT-PCR positivity (aOR=0.57, 95% CI 0.33 to 0.96; p=0.035).
	Clean running water available vs no clean running water.	Model-predicted probability of AIV positivity in LBM stalls.	The model predicted AIV positivity of 11% when clean running water was available and 79% when it was unavailable.	107
	Smooth surfaced stalls vs no smooth surfaces.		The model predicted AIV positivity of 24% in smooth-surfaced stalls, 42% in unsurfaced stalls and 60% in partially surfaced stalls.	107
	Uneven floors vs even floors.	Odds of AIV prevalence in stalls.	aOR=3.64 (95% CI 2.32 to 5.71).	104
	Wooden display tables.	Odds of H5 AIV prevalence on surface.	OR=0.70 (95% CI 0.50 to 1.00; p=0.047).	49
	Handwashing facility.	Odds of effect of access to hand washing facility on Campylobacter spp positivity in birds.	Access to a handwashing facility was associated with lower adjusted odds of high-risk classification (aPOR=0.32, 95% CI 0.13 to 0.78; p=0.001).	51
Interventions to be applied at micro-level by market vendors, stall holders and workers
PPE	Wearing gloves.	Odds of AIV A virus isolation from birds.	OR=0.70 (95% CI 0.45 to 1.06; p=0.10); not statistically significant.	61
Hand hygiene	Wash hands with soap after handling poultry vs no washing hands.	Bird-level influenza A virus isolation positivity.	In the initial analysis, handwashing was associated with higher odds of bird-level influenza A virus isolation positivity (OR=1.73, 95% CI 1.18 to 2.54; p<0.01); however, questionnaire-derived variables did not remain significant after multilevel adjustment.	61
	Hands sanitiser use.	Microbial contamination on workers’ hands.	On workers’ hands, the highest reductions with calcinated calcium were 3.90 log₁₀ CFU/cm² for total aerobic bacteria, 4.81 for total coliforms, 4.41 for E. coli and 3.72 for Salmonella spp.	114
Stall-cleaning frequency	Daily vs weekly.	Model-predicted probability of AIV positivity in LBM stalls.	The model predicted AIV positivity of 11% with daily stall cleaning and 81% with weekly stall cleaning.	107
	Daily vs none.	Odds of AIV positivity in LBM stalls.	OR=0.41 (95% CI 0.27 to 0.62).	104
	Weekly vs none.		OR=0.37 (95% CI 0.18 to 0.73).
	Monthly vs none.		aOR=0.47 (95% CI 0.28 to 0.80).
	Increasing frequency of stall and cage cleaning.	Influenza A positivity in cage specimens.	Higher cleaning frequency was inversely correlated with influenza A positivity in cage specimens: Stall cleaning: Spearman’s ρ=−0.41, p=0.007. Cage cleaning: Spearman’s ρ=−0.37, p=0.015.	115
Stall disinfection frequency	Weekly vs infrequent.	Model-predicted probability of AIV positivity in stalls.	The model predicted AIV positivity of 84% with infrequent disinfection and 13% with weekly disinfection.	107
Surface and equipment cleaning	Cleaning of cutting boards.	Microbial abundance on boards.	Reduced bacterial load, not specified (magnitude NR).	111
	Scraping as part of cleaning of wooden cutting boards.	Microbial load on cutting boards.	Applying traditional scraping techniques did not remove the microbial load (magnitude NR).	108
	Cleaning in place of wooden cutting boards.	Microbial load on cutting boards.	Cleaning-In-Place treatment achieved lower bacterial loads but did not fully eliminate microorganisms (magnitude NR).
	Use of detergent.	Microbial load on cutting boards.	Reduced bacterial load, not specified (magnitude NR).	111
	Traditional cleaning methods.	Microbial load on cutting boards.	Bacterial load of 6.35 log10 CFU/cm², despite cleaning (precleaning NR).	112
	Cleaning with agitated waterspout.	Bacterial bioaerosols in the market environment. Fungal bioaerosols in the market environment.	Reduced load by 37.7%. Increased load by 2.43-fold.	109
	Cleaning vs not cleaning poultry transport vehicles.	Model-predicted probability of AIV positivity in LBM stalls.	The model predicted AIV positivity of 22% when poultry transport vehicles were cleaned and 63% when they were not cleaned.	107
Surface and equipment disinfection/sanitation	Chemical sanitation.	Bacterial bioaerosols in the market environment. Fungal bioaerosols in the market environment.	Reduced load by 30.8%. Reduced load by 19.2%.	109
	0.01% calcinated-calcium sanitiser vs hot-water washing at 53°C.	Microbial contamination in poultry processing environments.	Reductions with hot water vs 0.01% calcinated calcium were 0.80 vs 1.68 log₁₀ CFU/cm² for aerobic bacteria, 0.79 vs 1.68 for total coliforms, 0.90 vs 2.46 for faecal coliforms, 1.00 vs 1.93 for yeast and mould, 0.96 vs 2.02 for E. coli and 1.38 vs 1.84 for Salmonella spp.	114
Carcass washing	Washing chicken carcasses with water.	Salmonella and Campylobacter prevalence and concentration.	Washing reduced mean Salmonella concentration by 0.60 log₁₀ CFU/g but did not change Campylobacter concentration. Regression coefficients were 0.37 (95% CI 0.03 to 0.71) for Salmonella and 0.32 (95% CI 0.15 to 0.50) for Campylobacter, suggesting cross-contamination. The number of birds washed per water batch was associated with postwash Salmonella concentration (β=0.81, 95% CI 0.08 to 1.55), but not Campylobacter concentration (β=0.32, 95% CI −0.17 to 0.81).	128
Separation of species	Species separation vs mixing.	Model-predicted probability of AIV positivity in LBM stalls.	The model predicted AIV positivity of 14% when chicken breeds were not mixed in the same cage and 76% when breeds were mixed.	107
	Separate chicken from waterfowl.		The model predicted AIV positivity of 23% when chickens and waterfowl were not sold in the same stall and 62% when they were sold together.
	Single-species selling vs multiple-species selling.	AIV prevalence in birds.	Selling one species: 23.29%. Selling multiple species: 40%.	53
	Separate ducks and chickens.	Odds of AIV A positive virus isolation in birds.	OR=1.57 (95% CI 0.84 to 2.75; p=0.13); not statistically significant.	61
	Selling ducks alongside other poultry.	Odds of stall-level AIV positivity in pooled environmental samples.	In the multivariable GEE model, stalls selling ducks alongside other poultry species had higher odds of AIV positivity than stalls selling chickens and non-duck poultry species (OR=2.50, 95% CI 1.50 to 4.10; p<0.001).	103
	Sharing cages, drinking water and feed between bird species.	Influenza A and H9 positivity in poultry drinking-water specimens.	Greater sharing between bird species was positively correlated with influenza A positivity (Spearman’s ρ=0.33, p=0.036) and H9 positivity (ρ=0.38, p=0.014).	115
Removal of sick birds	Remove sick from healthy birds.	Model-predicted probability of AIV positivity in LBM stalls.	The model predicted AIV positivity of 26% when sick birds were separated from healthy birds and 57% when they were not separated.	107
	No separation of sick/healthy birds.	AIV prevalence in birds.	No separation of sick and healthy birds was associated with higher odds of positivity (aOR=1.60, 95% CI 1.04 to 2.45; p=0.030).	53
	Self-medication of sick birds.	Campylobacter spp positivity risk in birds.	POR=1.78 (95% CI 0.44 to 7.47; p=0.542); not statistically significant.	51
	Killing sick birds for disposal.	Campylobacter spp positivity risk in birds.	POR=4.67 (95% CI 0.76 to 49.43; p=0.191); not statistically significant.	51
Handling of unsold birds	Keeping unsold birds.	Odds of AIV prevalence in stalls.	Keeping unsold birds was associated with higher adjusted odds of environmental AIV positivity (aOR=2.35, 95% CI 1.40 to 3.93).	104
	No mixing of unsold birds with newly arrived poultry.	AIV prevalence in stalls.	Mixing unsold and newly arrived birds was associated with higher vendor-level positivity (aOR=3.07, 95% CI 2.01 to 4.70; p<0.01). A separate model predicted positivity of 49% with mixing and 33% without mixing.	53 107
Birds holding areas	Keeping birds in floor pens vs cages.	Odds of stall-level AIV positivity in pooled environmental samples.	In the multivariable GEE model, stalls where birds were kept in floor pens had higher odds of AIV positivity than stalls where birds were kept in cages (OR=1.90, 95% CI 1.10 to 3.40; p=0.03).	103
	Keeping poultry overnight vs not keeping them overnight.	Odds of AIV A positive virus isolation birds.	OR=1.38 (95% CI 0.94 to 2.05; p=0.10); not statistically significant.	61
Slaughter practice	On-site slaughter.	Odds of AIV prevalence in the environment.	aOR=1.87 (95% CI 1.11 to 3.14).	104
		Campylobacter spp positivity risk in birds.	POR=1.22 (95% CI 0.56 to 2.66; p=0.618); not statistically significant.	51
	No slaughter and only carcass.	Odds of H5 AIV prevalence in environment.	No live slaughter/only carcasses was associated with lower odds of H5 detection (OR=0.20, 95% CI 0.10 to 0.50).	49
	Slaughter in separate space and sale of birds outside the market.	Odds of H5 AIV prevalence in environment.	Slaughter in a separate space with sale outside the market was not statistically significant (OR=0.40, 95% CI 0.20 to 1.10).	49
	Slaughter and trade of parent stock vs broilers.	Odds of H5 AIV prevalence in environment.	Trading parent stock rather than broilers was associated with higher odds (OR=5.70, 95% CI 3.60 to 9.20).	49
Sourcing practice	Sourcing directly from farms.	Model-predicted probability of AIV positivity in LBM stalls.	The model predicted AIV positivity of 20% when birds were sourced directly from commercial farms, compared with approximately 46–51% when they were sourced through middlemen, wholesale markets or multiple sources.	107
	Prevent wild bird access to stalls.		The model predicted AIV positivity of 28% when access by crows and other wild birds was prevented and 55% when wild birds had access to the stall.
	Sourcing birds from wholesale markets.	Odds of AIV prevalence in stalls.	Wholesale-market sourcing was associated with higher odds of positivity (aOR=2.96, 95% CI 1.48 to 5.92; p<0.01).	53

AIV, avian influenza virus; aOR, adjusted OR; aPOR, adjusted prevalence OR; CFD, computational fluid dynamics; CFU, colony-forming units; E. coli, Escherichia coli; ESBL, extended-spectrum beta-lactamase; FAO, Food and Agriculture Organization; GEE, generalised estimating equation; HPAI, highly pathogenic avian influenza; IR, incidence rate; LBM, live bird market; LPM, live poultry market; NR, not reported; POR, prevalence OR; PPE, personal protective equipment; RR, relative risk; rRNA, ribosomal RNA; RRR, relative risk reduction; RT-PCR, reverse transcription PCR; β, regression coefficient; ρ, Spearman’s rank correlation coefficient.

Control measures adoption and adherence

12 studies reported barriers and facilitators for the adoption and adherence of market control measures.^{61,74,93,94,101,102,105,108,113–116} Reported barriers clustered into six main domains: (1) economic and financial constraints; (2) limited infrastructure and basic services; (3) weak governance, inspection and enforcement; (4) limited awareness, training and knowledge; (5) low acceptability or adherence; and (6) sociocultural or practical constraints affecting hygiene, PPE use and outbreak reporting. Facilitators clustered into five main domains: (1) strict enforcement, (2) financial compensation, (3) training initiatives, (4) improved infrastructure and water access, and (5) positive perceptions or knowledge regarding biosecurity, PPE, cleaning and disinfection and AIV control measures. A comprehensive overview of the identified facilitators and barriers, including the organisational level at which they operate, is reported in online supplemental material 1: online supplemental table S13.

Discussion

This scoping review maps and summarises 87 studies published between 2017 and 2025 on TFMs where live animals and animal-origin products are traded. The findings show that TFMs function as dynamic One Health interfaces in which human, animal and environmental factors interact to shape zoonotic and foodborne disease risk. Interpreted through the conceptual and analytical framework, TFMs should not be understood simply as sites where pathogens are detected, but as market systems in which infrastructure, animal management, hygiene, worker and consumer practices and intervention feasibility jointly influence risk. A clear pattern was the imbalance in the evidence base: most studies focused on risk-exacerbating conditions and exposure pathways, whereas fewer examined human health outcomes or the long-term adoption and sustainability of control measures. The field therefore remains strongly driven by pathogen detection and outbreak-oriented surveillance, particularly for AIVs, with comparatively less emphasis on broader evaluations of how risk is produced and managed in routine market settings.

Public health risks associated with TFMs

Within the public health risk dimension, three connected components help explain how risk is generated. Human exposure studies showed that contact with live animals, contaminated animal products and contaminated market environments creates multiple exposure routes. Market-related factors, including poor hygiene, weak biosecurity, on-site slaughter, inadequate waste management, species mixing and prolonged animal holding, create conditions that may facilitate pathogen persistence and circulation. Behavioural factors, including unsafe handling, inadequate cleaning, poor hand hygiene and limited use of protective measures, can further intensify exposure. Together, these findings indicate that risk in TFMs emerges from interactions between market structure, daily practices and contaminated interfaces, rather than from the presence of live animals or animal-origin products alone.^24,117,118

The findings also reveal that zoonotic transmission and food safety are closely linked in these settings. Although AIVs dominated the literature, several studies also documented Salmonella spp, Campylobacter spp, E. coli and antimicrobial resistance markers in carcasses, meat products, market surfaces and environmental samples. These findings suggest that TFMs are not only potential sites of zoonotic spillover but also environments in which foodborne and environmental contamination can be repeatedly generated and maintained. Therefore, public health risk in TFMs is not limited to outbreaks of viral pathogens.¹⁷

At the level of human health outcomes, the evidence indicates that these exposure pathways can translate into clinically apparent infection and subclinical exposure.^13,32,119 Human infections linked to TFM exposure were reported in 13 studies (14.9% of all studies included), all of which involved AIVs. Genomic comparisons have directly linked AIV subtypes, including A(H5N1), A(H7N9) and A(H9N2), to markets through concordance between patient-derived and market-derived samples in Bangladesh and China^79,120. Serological and epidemiological studies further suggested occupational exposure among market workers and increased risk among people working in or repeatedly visiting LPMs.^35,90,91 The predominance of AIV-focused evidence should therefore be interpreted as a feature of the available literature, rather than as evidence that AIVs are inherently more important than other TFM-associated zoonotic and foodborne hazards. Overall, the literature is therefore stronger at showing how human exposure occurs than at quantifying how much human disease can be directly attributed to TFMs across different pathogens.

Characteristics and reported outcomes of control measures in TFMs

The control measure dimension showed that interventions operate across multiple levels of the market system. Macrolevel measures, such as market closure, regulation and infrastructure improvements, target the wider market environment. Mesolevel measures, including rest days, biosecurity controls and engineering or environmental improvements, target market organisation and contamination pathways. Microlevel measures, such as cleaning, disinfection, species separation, slaughter practices and protective measures, focus on vendor and worker practices. These findings suggest that multilevel intervention packages may address multiple risk pathways more comprehensively than isolated technical controls. However, most intervention studies were conducted in LBMs and LPMs and focused on AIVs in domestic birds or market environments. Extrapolation to other TFM types, animal groups or pathogens should therefore be cautious. Interpretation of intervention impacts also requires caution, as most evidence was observational or modelling-based, few studies directly assessed human health outcomes and intervention designs and outcome measures were heterogeneous across studies. Reported associations should therefore not be interpreted as evidence of causal effectiveness across diverse TFM settings.

The evidence helps explain why technical efficacy does not necessarily translate into sustained public health benefits. Implementation depends on feasibility and acceptability at authority, market and vendor levels. Species segregation may reduce transmission but requires sufficient stall space and infrastructure, which often requires substantial or disruptive market restructuring.¹²¹ Restrictions on the sale of sick poultry may face resistance when they directly reduce vendor income, particularly in resource-constrained LMIC settings.¹²² Therefore, barriers and facilitators should be considered and interventions adapted to local context to support successful adoption and sustainability. For example, one study found that half of market vendors in a LPM were unable to recognise sick birds, indicating that policies requiring their removal would only be feasible if vendors were provided with appropriate training and capacity building.¹²³

Sustained implementation of control measures in TFMs may depend on addressing structural and economic constraints alongside technical feasibility. Improving market conditions often requires substantial investment in infrastructure, including restructuring and upgrading facilities to support hygiene and biosecurity. Carefully structured financial compensation for the removal of sick and dead poultry can support the uptake of control measures; however, resource constraints mean that such approaches may not be feasible in all settings.^61,74,94,116 This variability underscores the need for a combination of complementary interventions and preventive measures tailored to specific market contexts to reduce risks. However, their expected impact, frequency and operational design are likely to vary according to the animal or product traded, pathogen transmission route, environmental persistence, market infrastructure and implementation capacity. Pre-market measures, including vaccination of birds, restrictions on certain species and regulated animal transport, may further reduce pathogen introduction into TFMs.^18,19,101 Collectively, the evidence suggests that technical interventions alone may be insufficient. Sustained implementation may depend on community participation and engagement, economic feasibility and supportive governance. These findings align with the WHO Five Keys for Safer Traditional Food Markets, which translate these priorities into practical guidance on food safety, reduced contact with animals and personal protection.¹²⁴ Reported intervention outcomes may also vary according to timing, outbreak stage, seasonality and geography, supporting the need for adaptive, context-specific implementation.^{19,98,101,125} A combination of risk mitigation measures implemented across multiple levels and tailored to specific markets and pathogens may help address interconnected transmission pathways, although its effectiveness requires further evaluation across diverse TFM settings.

Future research directions and policy implications

The evidence has several implications for policy and practice aimed at reducing public health risks associated with TFMs. The reviewed evidence suggests potential advantages of coordinated, multilayered strategies over isolated actions because risks in TFMs arise through multiple, interdependent transmission pathways. Policies focusing exclusively on market-level controls may be insufficient, as TFMs operate within broader food systems. Multistage, complementary strategies across the supply chain may therefore provide broader coverage of risk pathways, although evidence regarding their comparative effectiveness and long-term sustainability remains limited. The implementation and reported outcomes of infection-control measures may depend on public awareness, vendor engagement, social acceptability, timing, geographical context and resource availability, particularly during outbreak peaks or in resource-limited settings.

Adoption of a One Health approach explicitly recognises that the linkages between humans, animals and the environment are critical to mitigating risks in TFMs. Despite clear evidence of zoonotic risks in these settings, important knowledge gaps remain. Future research should examine why intervention evidence is disproportionately concentrated on AIVs, including whether this reflects outbreak-driven surveillance and research priorities, established AIV-specific monitoring systems or the under-recognition and limited measurement of other zoonotic and foodborne hazards. Intervention studies should also be expanded to other frequently reported pathogens, including Salmonella spp, Campylobacter spp, E. coli and antimicrobial-resistant bacteria, across diverse TFM settings. Prioritisation frameworks, such as the One Health Zoonotic Disease Prioritization process, can support risk ranking and resource allocation by accounting for the complex, context-specific dynamics of TFMs.¹²⁶ More advanced analytical approaches, including longitudinal studies and modelling approaches that integrate genomic, environmental and behavioural data are needed to understand pathogen persistence and transmission. Further examination of how climate variability, market connectivity and trade networks influence pathogen ecology could strengthen predictive models and early warning systems for improved human and animal health.

The evidence indicates that TFM policies should be adaptive and context-specific, accounting for differences in market types, animal–pathogen combinations and socio-economic conditions. Coordination across animal health, food safety and public health sectors, together with market-community engagement and investment in water, sanitation and waste management, may strengthen implementation. These findings provide a rationale for integrated One Health approaches, while their effectiveness and effects on livelihoods require further evaluation across diverse TFM settings. Embedding TFMs within national food-safety and food-system policies can help align livelihood protection with disease prevention, while governance that balances economic, cultural and health priorities is key to long-term sustainability. Across the reviewed studies, sustained investment in market infrastructure and basic services, including water, sanitation and waste management, was repeatedly identified as an important enabling condition for intervention implementation. Such investments may better support sustainable implementation when they also safeguard the health, safety and dignity of market workers and consumers. Ultimately, while TFMs remain indispensable to food systems in LMICs, the evidence provides a rationale for integrated, context-sensitive and well-coordinated One Health approaches to risk mitigation. The effectiveness of these approaches in reducing spillover risk while protecting livelihoods and public health requires further evaluation across diverse TFM settings.

Strengths and limitations

This review consolidates heterogeneous evidence on TFMs through a One Health framework, integrating findings across multiple disciplines using a predefined analytical and conceptual framework. A key strength lies in the structured integration of complex interactions among pathogens, humans, animals and the environment, providing a holistic overview that extends beyond the influenza-focused reviews that have previously dominated the literature.^5,6,13,18,19 By jointly examining public health risks, control measures and the contextual barriers and facilitators influencing their implementation and reported outcomes, the review advances an integrated understanding of TFMs that goes beyond pathogen occurrence alone. This approach supports a system-level characterisation of TFMs as dynamic environments in which biological, behavioural and structural factors collectively determine the magnitude of public health risk. An additional strength is the broad evidentiary scope across study designs and regions. Although AIVs were the most frequently investigated pathogens, findings were generally consistent within the most studied settings, particularly poultry-dominated market systems. While pathogen-specific data for non-AIV hazards were limited, the review identified recurring market interfaces and operational conditions, such as animal handling, slaughter, hygiene and environmental contamination, that warrant investigation across a broader range of zoonotic and foodborne pathogens. However, the present evidence base does not allow these patterns to be assumed to apply uniformly beyond AIV-dominated poultry market systems.

Several limitations should be considered. First, the evidence was unevenly distributed across regions and pathogen types, with a heavy concentration of AIV studies in poultry markets in East and Southeast Asia, particularly China. Differences in regulatory frameworks, enforcement capacity and surveillance resources may further constrain the applicability of these findings to settings with more limited capacity. Evidence from Africa, Latin America, wildlife markets, non-poultry species, seafood and consumption-related foodborne risks remains limited, reducing generalisability beyond AIV-dominated poultry market systems.⁵ Most studies were observational, with few serological studies directly linking market exposure to human infection and few experimental or quasi-experimental evaluations of control measures. Heterogeneity in study design, sampling, diagnostic methods and exposure classification further limits comparability and causal inference.

Additional limitations related to the review process should also be considered. This review was not conducted under a prepublished protocol, which limits external verification of a priori methodological decisions and any subsequent amendments. Consistent with the evidence-mapping purpose of a scoping review, no formal methodological quality or risk-of-bias appraisal was conducted; consequently, reported intervention outcomes were mapped as presented in the included studies and were not assessed or weighted according to study quality. The search was restricted to three bibliographic databases, did not include a dedicated search of grey-literature sources and covered publications from 2017 to 2025. Consequently, potentially eligible dissertations, institutional reports, regional studies or earlier evidence containing primary empirical data may have been missed, reducing the comprehensiveness of the evidence map and potentially introducing selection bias. Findings concerning intervention effectiveness should therefore be interpreted cautiously. Restricting eligibility to English-language articles may also have introduced language and selection biases and resulted in the exclusion of relevant regional evidence from LMIC settings where research is commonly published in French, Spanish, Mandarin or other languages.

A further limitation is the paucity of substantial evidence on markets involving live wild animals, wild meat and other non-poultry species, which constrains understanding of the full diversity of zoonotic hazards within TFMs. Importantly, risk is likely shaped not only by whether animals are classified as ‘wildlife’, but also by sourcing, movement through trade networks, species mixing and handling or slaughter practices. Finally, the analytical framework developed and applied in this review should be regarded as a working model, requiring further testing and validation across diverse market contexts to assess its reproducibility and broader applicability.

Conclusion

This scoping review summarises a substantial body of evidence on public health risks and control measures associated with TFMs trading live animals and animal-origin products in LMICs. The reviewed literature shows that TFMs function as dynamic human–animal–environment interfaces in which zoonotic and foodborne risks emerge from the interaction of biological hazards, market infrastructure, operational practices and human behaviour. Risk is therefore shaped by market systems rather than by single pathogens or practices alone. These findings provide a rationale for context-specific, integrated, multilevel One Health approaches combining infrastructure improvements, biosecurity, behaviour change, supportive governance, surveillance and engagement across the food system from animal production and transport to market operations. However, because the synthesised evidence is dominated by AIVs and LPMs and is largely observational, the effectiveness and generalisability of these approaches across diverse TFM settings, including their effects on livelihoods, require further evaluation. Together, the evidence synthesised and the gaps identified in this review provide a strong foundation for more focused and translational research, strengthened surveillance and evidence-informed, context-responsive policy development for TFMs.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.