Effectiveness of behaviour change interventions to reduce the risk of faecal contamination in urban irrigated vegetable value chains – applying the COM-B behavioural framework

Most leafy vegetables consumed in cities of low- and middle-income countries (LMICs) are grown locally (Orsini et al. 2013). Urban and peri-urban agriculture enhances urban livelihoods, food sovereignty, recycling of organic waste, greening cities, and limiting floods (FAO 2012). However, the urban water systems most often used to irrigate these vegetables are contaminated with faecal, chemical, and emerging contaminants due to lagging sanitation. It is estimated that urban farmers practise de facto indirect wastewater reuse on 30 million hectares of farmland, supplying fresh vegetables to 800 million people globally (Thebo et al. 2017), and these figures are likely to increase (Raschid-Sally & Jayakody 2008; Figuié 2019).

Exposure to contaminated water and consuming raw contaminated vegetables are the two major pathways to contracting enteric, acute respiratory, and parasitic diseases (Hoffmann et al. 2017; Raj et al. 2020). Faecal pathogens (bacteria, protozoa, and viruses) can provoke diarrhoeal diseases (including cholera), acute respiratory infections (e.g., tuberculosis from breathing sprinklers droplets), as well as other water- and food-borne invasive infectious (e.g., hepatitis, typhoid fever) and parasitic diseases (e.g., helminth infections such as Taenia and Ascaris). Exposure to contaminated water caused about 1.4 million deaths and the loss of 74 million disability-adjusted life years (DALYs) in LMICs in 2019 (Wolf et al. 2023). The consumption of contaminated food caused over 220,000 deaths and the loss of 17 million DALYs in LMICs in 2010, mainly in Africa and Southeast Asia (WHO 2015). Even though these mostly preventable diseases primarily affect children under 5 years, they also impact the well-being and productivity of all age groups and impose a heavy economic burden (Hutton & Haller 2004; WHO 2015).

The World Health Organization (WHO) proposes a multiple-barrier approach adapted to LMICs to reduce health risks from using wastewater (WHO 2006, 2022). It suggests public authorities set appropriate health-based targets, working with primary stakeholders (farmers, vegetable traders, street food vendors, and consumers) to combine safe practices from ‘farm to fork’ to protect workers' and consumers' health. Where possible, sanitation is counted as the first barrier. Amoah et al. (2011) demonstrated the cost-effectiveness of adopting low-cost irrigation practices on farms, safe vegetable handling, and washing methods in markets (including street food restaurants) and households. Increasing uptake of these practices remains challenging (Drechsel et al. 2022) and should be considered a priority (Prain et al. 2022).

To date, there has been no systematic analysis of the factors that result in the success or failure of interventions to change stakeholder hygiene and food safety practices in the urban irrigated vegetable value chains of LMICs. This research responds to that gap and proposes modifications for more successful interventions. The COM-B behavioural model (Capability + Opportunity + Motivation = Behaviour) developed by Michie et al. (2011), explained in the following section, was used to structure a review of the range of approaches used to address the risk of faecal contamination in the value chain. This provides a basis for explaining why safe practices are not more widely adopted.

This review systematically appraises the impact of interventions on stakeholders' adoption of good hygiene and food safety practices in the urban irrigated value chains of LMICs.

The systematic search of the literature ensured comprehensive conceptual coverage by querying the titles, abstracts, and keywords in the Scopus, Web of Science, and ProQuest-Social Science Premium Collection against the Boolean combination of five queries as follows:

• #1 (the presence of human excreta through 16 terms including faeces, faecal sludge or wastewater)

• AND

• #2 (the occurrence of oro-faecal diseases through 25 terms including water-borne, diarrhoea or helminth)

• AND

• [#3 (irrigated vegetable production through 11 terms including farmer, horticulture or

• crop)

• OR

• #4 (vegetable trading through 21 terms including produce, trader and consumer)]

• AND

• #5 (low- and middle-income countries through terms such as Global South or developing countries and a list of LMICs countries).

Query #5 was adapted from Liverpool-Tasie et al. (2020) and is based on the World Bank classification of economies from 2021. The detailed terms of the search are available in Table 3 in the Supplementary Material.

The search was limited to articles focusing on LMICs published in English between 1 January 2000 and 1 August 2022. Furthermore, the Context–Intervention–Mechanism–Outcome (CIMO) logic proposed by Denyer & Tranfield (2009) was adopted to develop the inclusion and exclusion criteria as follows:

• Context: the study relates to the urban irrigated vegetable value chain in LMICs.

Excluded: high-income countries, rural settings, health facilities, schools, focus on drinking water or groundwater, and food products other than vegetables or non-faecal contaminants.

• Intervention: the study reports an initiative aiming to manage the risk of faecal contamination rather than assess the risk.

Excluded: risk assessments.

• Mechanisms: the study reports a risk management mechanism other than conventional effluent treatment.

Excluded: conventional onsite or centralised wastewater or sludge treatment.

• Outcomes: the study reports a reduction of contamination or a change of practices.

Excluded: absent results.

Records were screened using Rayyan (https://rayyan.ai/), and those not meeting eligibility and inclusion criteria were excluded. The remaining articles were included for qualitative analysis. Relevant information was systematically identified: authors, publication year, and country of implementation; sector of intervention (farm and/or market and/or household-consumer); mechanism of intervention; and whether the intervention targets stakeholders' opportunities, capabilities, and/or motivations.

We hypothesised that many WASH interventions follow top-down designs that primarily focus on providing stakeholders with an innovation (such as a technology, service, or technique), sometimes supported by education or training. They often, however, fail to account for stakeholders' perceptions and willingness to integrate the innovation into their practices. Assessing interventions against a robust theoretical behavioural framework helps identify which factors promote or hinder the success of the cases reviewed. The COM-B behavioural framework (Michie et al. 2011) was selected for its accessibility and robust theoretical foundation, unifying several theories and models. Numerous studies build on this model to explain behaviours or promote changes in contexts ranging from informing the British government's behaviour change policies (Michie & West 2013) to improving hand and food hygiene behaviours in LMICs (Wodnik et al. 2018).

Deficits in Capability, Opportunity, and/or Motivation can be addressed through nine functions (see Table 1): restricting opportunities for competing behaviours, restructuring the physical or social environment, enabling, educating, training, incentivising, persuading, coercing, and modelling (Michie et al. 2011). One intervention function can address more than one behavioural component and can incorporate functions. Enablement is understood broadly as providing the means to make it easier to enact a behaviour innovation to reduce barriers or improve the opportunity (beyond environmental restructuring) or the capability (beyond education and training). Examples include providing physical or social support and material or financial resources. These interventions are promoted by different types of policy: legislation, service provision, communication, or marketing campaigns. Michie et al. (2011, 2014) recommend diagnosing what people need to adopt recommended behaviour by identifying intervention functions and policy categories adapted to foster change, then selecting the behaviour change technique and mode of delivery that best serves the intervention function.

The results presented and discussed below are structured around the intervention functions described in the records under review and clustered around the related Opportunity, Capability, and Motivation components of the COM-B model (see Table 1).

The COM-B framework suggests that physical (e.g., equipment and infrastructure) or social (e.g. services) opportunity needs are best addressed by restricting the opportunities to perform competing behaviours, restructuring the physical and/or social environment, and enablement interventions.

Regulations on wastewater reuse in LMICs reflect different political priorities. To address water scarcity, Mexico and several countries in the Middle East/North Africa and other arid/semi-arid regions promote formal irrigation schemes using treated wastewater (Shoushtarian & Negahban-Azar 2020). Countries including Ghana, India, Jordan, Mexico, Morocco, Tunisia, and Vietnam, however, have officially banned the use of untreated wastewater with mixed success. Regulations around wastewater reuse for irrigation are inadequate or missing in many LMICs, including Southeast Asia (Winkler et al. 2017). Regulations that aim to protect public health from faecal pathogens in other sectors can also be unclear or inconsistent (see Moya et al. (2019) on the use of excreta-derived fertilisers in Kenya and Okoruwa & Onuigbo-Chatta (2021) on food safety regulations in Nigeria).

The lack of enforcement of regulations to effectively control stakeholders' practices limits the effectiveness of restriction (and coercion) interventions. Examples include Cuenca-Adame et al. (2001) for wastewater irrigation in Mexico, Musa & Akande (2002) for routine medical examination of food vendors in Nigeria, and Salamandane et al. (2020) for food markets in Mozambique. The lack of coordination among public agencies, the resulting poor regulatory coherence, and weak enforcement capability also exacerbate the problem (Okoruwa & Onuigbo-Chatta 2021). As largely informal food value chains, characterised by a multitude of small-scale farmers and mobile vendors, are challenging to control, most administrations focus on the formal/export sectors in capital cities (Antwi-Agyei et al. 2016; Nordhagen et al. 2022). The lack of understanding of microbial risk among urban farmers and vegetable and street food vendors, and their limited access to water, sanitation, or cold storage, make compliance with regulations particularly difficult (Keraita et al. 2014; Salamandane et al. 2020). When non-compliance is detected, sanctions are not always proportionate or dissuasive. Sanctions that are absent or too light entice stakeholders to ignore good practices (Seleman et al. 2020).

Ad hoc regulations and adequate enforcement can lead to more sustainable changes than temporary educational or promotional interventions (Idowu & Rowland 2006; Keraita et al. 2014). Mobilising the political will to develop adequate policies and secure funding sufficient to enforce resulting regulations is challenging, as local data on the prevalence of food-borne diseases are scarce and public authorities rarely prioritise the informal sector (Jaffee et al. 2019). Regulations are more likely to change behaviours if they avoid mimicking international standards they do not have the capability to implement (Mdee & Harrison 2019) and, instead, adapt them into national standards that account for the local context (Keraita & Drechsel 2015).

Certifying vegetable safety to give access to premium channels should motivate urban farmers to adopt safe practices (Keraita & Drechsel 2015). Most domestic food safety management systems, however, consist primarily of controls in the formal sector, overlooking the informal sector despite it serving the largest proportion of the population and accounting for the majority of the food disease burden in LMICs (Figuié 2019; Henson et al. 2023).

Effectively developing standards and enforcing controls seems to be a feature of the export-oriented private sector rather than governments (Moya et al. 2019). Certified farmers adhere to standards developed with little or no participation from their end (González & Nigh 2005), including record-keeping and controlling the quality of the water used for irrigation and washing. Most smallholder farmers, however, are unable to sustain the high certification costs or emulate these practices, and the food safety of domestic markets remains unaffected (Henson et al. 2023).

Third-party certification schemes have recently started to serve domestic markets, adopting requirements more accessible to smallholder farmers. The Ghanaian Ministry of Food and Agriculture and Standards Authority and cooperation partners instigated the Ghana Green Label in 2014 (GhanaGreenLabel.Org 2015). High costs and low domestic demand for certified vegetables limit the supply to a handful of packing houses and high-end outlets in Accra. As of 2022, about 700 farmers are certified nationally, including a few dozen urban and peri-urban farmers. Similarly, since 2016, the supermarket chain SPAR-South Africa has developed Localg.a.p. (for Local Good Agricultural Practices), a programme adapted from the European Farm Assurance Programme GlobalG.A.P. (GLOBALG.A.P./FoodPLUS GmbH 2023), which builds smallholder farmers' capability to comply with good hygiene and safety practices and supply its local supermarkets (SPAR 2021). Both these programmes allow farmers to use a potentially polluted source of irrigation water, provided they supply a water analysis demonstrating they took the necessary steps to ensure the water quality complies with ‘accepted [water] standards’ (Ghana Green Label) or the WHO (2006) Guidelines (SPAR/Localg.a.p.). Unfortunately, the principles of health-based targets and the multi-barrier approach are overlooked in both cases.

Participatory guarantee systems (PGSs) are an alternative to private third-party certification, consisting of farmers, consumers, and other interested stakeholders agreeing on good practices and working conditions adapted to socio-ecological contexts. Networks of local farmers and consumers assess compliance (Nelson et al. 2010), which can help address food safety issues in value chains where quality management is limited (Moustier et al. 2023) and distrust exists between stakeholders (Antwi-Agyei et al. 2016). More than 1.2 million producers are involved in over 240 PGSs cultivating 900,000 hectares globally, mostly in the Global South (Anselmi & Moura e Castro 2022). PGSs also provide platforms to reduce farmers' costs through collective buying, marketing, and logistics, exchanging good practices and locally adapted seeds (Home et al. 2017). PGSs, however, often suffer from a lack of recognition by institutions and struggle to convince consumers to allocate time and resources to maintain the certification process. Consequently, they are subject to tension among stakeholders (Kaufmann et al. 2023).

Several on-farm techniques can reduce the transmission of pathogens to the crop. Biochar proved ineffective in eliminating pathogens from the soil of farms in northern Ghana (Asirifi et al. 2021). On-farm fly ash/lime filters reduce faecal coliforms by 0.5 log units in domestic greywater, which can then be used for irrigation (Nondlazi et al. 2017). The focus on technical performance, however, often leaves aside cost efficiency, sustainability, and user acceptability, which shape stakeholders' capability and motivation to use them. The project ‘Valorisation of Urine Nutrients in Africa’ included a multi-disciplinary team to better account for economic viability, seek authorities' support, and explore users' attitudes towards the reuse of urine and dehydrated faeces, recognising that adoption is the users' decision (Udert et al. 2016). Sedimentation ponds and filtration techniques (Silverman et al. 2014; Tripathi et al. 2019) can help farmers reduce pathogen concentration in irrigation water, especially helminth eggs. Some Ghanaian urban farmers use small ponds but are not interested in maintaining their sedimentation properties (Silverman et al. 2014). Moreover, the risk of these ponds becoming reservoirs for malaria-transmitting mosquitoes needs to be considered, particularly with the expansion of Anopheles stephensi, which can develop in urban polluted waters (Drechsel et al. 2006; Sinka et al. 2020).

Various irrigation practices impact pathogen exposure differently. The most effective are drip irrigation systems (especially subsurface ones; see Tripathi et al. 2019), but most urban farmers are not willing to accommodate the filtration, pressurisation, and distribution equipment, and reduced cropping density they require (Keraita et al. 2007a). Contrary to the overhead flow and splashing on watering cans, flood and furrow irrigation limit the pathogen transfer to the aerial parts of vegetables but further expose farmers who work barefoot. Sprinklers and spray irrigation produce droplets that transfer pathogens to crops, farmers, passers-by, and farm neighbours (Keraita et al. 2007a). Stopping irrigation a few days before harvest for pathogens to die off naturally is acceptable under temperate climates where crops are only irrigated every few days (Drechsel & Keraita 2014), but not in hot climates such as in Ghana, where it results in wilting and revenue loss (Keraita et al. 2007b). Optimising irrigation to reduce soil moisture during and after the growing season also reduces pathogen concentration (Cuenca-Adame et al. 2001; Tripathi et al. 2019) but is beyond most farmers' capability. Selecting crops that require cooking increases safety but is unattractive due to their lower market value. Switching to safer water sources is often constrained by piped water being too expensive or landlords forbidding farmers to drill boreholes on their land (Amoah et al. 2011).

Sanitary vegetable washing methods can also reduce the level of pathogenic contamination. Washing vegetables in basins, with or without the addition of salt or vinegar, is common in markets and households but less effective at removing pathogens than washing under running water or in water with added chlorine, permanganate potassium, or another disinfectant (Amoah et al. 2007; Googoolee et al. 2020; Traoré et al. 2020). As washing vegetables does not prevent the occupational exposure of farmers and vendors (Woldetsadik et al. 2017), nor is sufficient to protect consumers (Amoah et al. 2007; Samadi et al. 2009; Luchesi et al. 2016; Googoolee et al. 2020), it is essential to address both on-farm and post-harvest contamination.

Medical examinations and deworming (chemotherapy) of street food vendors can limit contamination. The effectiveness of medical examination to prevent transmission of helminth eggs depends on whether the drug used targets all relevant parasites (Idowu & Rowland 2006) and the enforcement is sufficient to ensure results are reported and post-convalescence examinations performed (Musa & Akande 2002). Regular deworming left high levels of contamination in Nigeria as the limited campaign coverage and insufficient access to safe water and toilets did not prevent re-infection (Idowu & Rowland 2006; Knopp et al. 2010). In contrast, Sri Lanka has almost eliminated morbidity due to worm infections thanks to mass deworming (Montresor & Mupfasoni, 2019) and higher access to at least basic urban water and sanitation services (JMP 2022). Education activities should complement deworming campaigns to address low-risk awareness, which is exacerbated by the infection remaining asymptomatic for extended periods (Idowu & Rowland 2006; Mihrshahi et al. 2009; Knopp et al. 2010).

The COM-B framework suggests that a physical, psychological (emotional or cognitive) capability deficit within an individual's realm is best addressed by education, training, and enablement interventions.

Interventions often consider education and training together or even interchangeably. From the perspective of the COM-B framework, however, they serve different purposes and should be differentiated. Education increases knowledge or understanding by providing information, thus improving psychological capability or reflexive motivation. Training imparts skills to increase physical or psychological capabilities (Michie et al. 2011).

Providing opportunities to adopt safe practices needs to be complemented with education and guidance to increase stakeholders' awareness of microbial risks (Luchesi et al. 2016; Reddi et al. 2016; Woldetsadik et al. 2017; Salamandane et al. 2020). Education intervention also aims to change their attitudes (Moya et al. 2019) and motivate them to act (Kragić Kok et al. 2020).

Indeed, throughout Africa, many urban residents do not link food-borne diseases to a lack of hygiene (Salamandane et al. 2020), with some claiming that ‘stomach aches result from consuming food that the worms do not like’ (Idowu & Rowland 2006, p. 163) or that speaking of contaminated vegetables only aims to hurt others' businesses (Abass et al. 2017).

Education is a major health determinant (Mihrshahi et al. 2009; Ngowi et al. 2017). Methods, such as PRECEDE (Predisposing, Reinforcing, and Enabling Constructs in Educational Diagnosis and Evaluation) or PHAST (Participatory Hygiene and Sanitation Transformation), help practitioners structure interventions to more effectively increase participants' knowledge appropriation (Ngowi et al. 2017). However, Information–Education–Communication (IEC) interventions alone offer mixed results, especially if they misunderstand the context in which the behaviour is to be performed or the recipient's perception of the cost and benefits of the behaviour (De Buck et al. 2017). Individuals' choices are not purely rational and based on knowledge (Scott et al. 2007). Rather, they rely on shortcuts and cue associations, such as the credibility of the person conveying the information or the attractiveness of the medium (Glanz et al. 2008).

The profile and legitimacy of the educator, their interpersonal communication skills, and their attention to the gender sensitivity of some behaviours (Salamandane et al. 2020) are as important as their knowledge of the topic (Borghi et al. 2002; Winkler et al. 2017). Equally important is the quality and availability of educational material, as well as the duration and frequency of the interventions (De Buck et al. 2017). Co-designing the content with intended recipients and fieldworkers helps ensure it is appropriate (Amoah et al. 2009). Knowledge can be conveyed through plays, world cafés, and road shows that follow the pathogen pathway from farm to fork (Amoah et al. 2009), radio spots (Borghi et al. 2002), movies, and comic booklets (Ngowi et al. 2017) or educational games (Kragić Kok et al. 2020). Such community-based education involving experiential, empowering exchanges are more effective than one-way approaches, such as sharing health messaging, especially in the long-term (De Buck et al. 2017), but is more challenging to upscale.

Training often completes education interventions (Musa & Akande 2002; Donkor 2009; Salamandane et al. 2020). Training strengthens recipients' self-efficacy, i.e., ‘I know I can do it’ (Ngowi et al. 2017) by allowing them to observe and practise in a controlled environment the behaviour to be performed. Training food handlers is a widely applied strategy to increase food safety (FAO 2003). Urban farmers, however, receive little training to address the health risks of wastewater irrigation (Amoah et al. 2011). In many cases, extension officers can only offer urban farmers training developed for rural farmers, even though the crops and practices differ, reflecting the lack of recognition and marginality of urban agriculture (Keraita & Drechsel 2015). Even when tailored to recipients' needs, training requires refresher interventions (Salamandane et al. 2020) and is challenging to upscale (Grace 2015).

The COM-B framework suggests that stakeholder automatic (e.g., wants, needs, emotions) and reflexive (i.e., conscious planning, beliefs about good and bad) motivation is best stimulated by persuasion, incentivisation, or coercion interventions. In addition, education can stimulate reflexive motivation, while automatic motivation responds to environmental restructuring, modelling, and enablement.

Incentivisation is one of the functions that intervention designers rely on the most to stimulate stakeholders' motivation to improve sanitation services (Moya et al. 2019) or increase the adoption of safe practices among farmers and traders (Amoah et al. 2011). Designers mostly think of market incentives, such as premium prices for safer vegetables, cash transfers, governmental subsidies, access to microcredit, or other forms of monetary incentives (Karg et al. 2010; De Buck et al. 2017).

Implementing viable market incentives by stimulating demand is challenging (Amoah et al. 2011). Studies suggest that stakeholders' returns need to increase by at least 30% for them to adopt a new practice (Karg et al. 2010). This accounts for often underestimated intangible costs, such as cultural resistance, perceived control, opportunity cost, and other psychological costs (Keraita & Drechsel 2015), but not necessarily for quality monitoring costs (Karg et al. 2010). Most customers of informal markets are unaware of the risk (Keraita & Drechsel 2015), and not interested in knowing (Nordhagen et al. 2022) or paying for safer vegetables. The demand from hotel restaurants, supermarkets, and other premium outlets that serve customers willing to pay such premiums is limited. Decision-makers could stimulate the offer of safe vegetables through coercion through fines, fees, taxes, or other disincentives (Karg et al. 2010). Funding the required monitoring of farmers' and food handlers' practices would require policymakers to have a clearer understanding of the return on investment (Keraita & Drechsel 2015).

Other monetary and explicit incentives do not always reach their goal (De Buck et al. 2017), as decision-makers sometimes misunderstand stakeholders' values and preferences. The existence, nature, size, and framing of the incentive convey either positive or negative signals to stakeholders. Too small an incentive may suggest that policymakers think the behaviour is unimportant, whereas if too high it can suggest policymakers consider the behaviour too complicated for stakeholders or that they are only self-interested (Gneezy et al. 2011). Thus, poorly designed incentives can reduce stakeholders' willingness to adopt a safe practice, even once it is removed. Incentives are more effective when framed in a way that preserves stakeholders' feelings of trust and autonomy rather than through control or challenging social norms, image concerns, and other social motivations (Gneezy et al. 2011) as illustrated by Bahraseman et al. (2024).

Non-monetary incentives are also a possibility, including dedicated support from extension officers, ‘best farmer’ awards or compliance certificates, good media publicity, and new marketing channels (Karg et al. 2010; Keraita et al. 2014). The study by Morse et al. (2019) is one of the few to actually report leveraging implicit incentives that mobilise stakeholders' desire for social recognition to align their practices with their personal values and ethics, autonomy, and control. Experts often underestimate farmers' sensitivity to jeopardising consumer health and assume farmers are more interested in explicit incentives, such as being provided with free or subsidised seeds, fertilisers, or other inputs (Keraita et al. 2014).

Interventions that cater to stakeholders' motivations are seldom articulated in the screened literature. Morse et al. (2019) suggests restructuring the social environment by advertising that many households are already performing the recommended hygiene-related behaviours, which are becoming the social norm. Improving farmers' land-tenure security would increase their willingness to adopt safe practices (Amponsah et al. 2016). Others leverage traditional beliefs such as ‘tody’ (similar to Buddhist and Hinduist karma) to convince Malagasy food vendors to avoid making their customers sick (Sarter & Sarter 2012). Role models motivating the uptake of good health practices by those around them have also been highlighted (Yihenew et al. 2014). Opinion leaders or other individuals respected by their community can champion behaviour change (De Buck et al. 2017; Winkler et al. 2017). Modelling through serious games can also stimulate the adoption of good practices (Kragić Kok et al. 2020). Davis et al. (2011) generated personalised information by testing the water quality in homes to persuade households to adopt risk-reducing strategies; Hirai et al. (2016) identified the desire to smell nice as a powerful intrinsic motivation to wash hands with soap, while Kragić Kok et al. (2020) leveraged social learning through games.

Adopting a robust behavioural framework, such as the COM-B framework, helps intervention designers consider all relevant factors and establish a thorough diagnosis and theory of change instead of relying on assumptions and familiar go-to functions. The framework also helps articulate the functions (e.g., sharing information is not training) and objectives of the intervention (e.g., whether education is to strengthen capabilities and/or reinforce reflexive motivation) and align its content and delivery. It is helpful to consider which practices present the best combination of Acceptability, Practicability, Effectiveness, Affordability, Spill-over or Side-effects, and Equity (the APEASE criteria, see Michie et al. 2014) when selecting the ones to promote.

Changing the practices and attitudes of all stakeholders involved in the value chain has coordination and governance implications. Food safety and public health are public goods whose production is one of the sovereign duties of governments. A challenge governments face is to recognise and enable informal businesses to produce and distribute safer vegetables by providing them with information, support, and incentives, rather than just control and sanctions, to comply with minimum food safety standards (Karg et al. 2010; Henson et al. 2023). Henson et al. (2023) also recommended that governments integrate food safety in urban planning and other relevant policies and guide local multisectoral collective action at the municipal level.

This review focuses on LMICs where the multi-barrier approach, as recommended by WHO (2006), remains the accepted procedure. Each country can adopt its own trajectory, especially in relation to enforcing water quality standards and food safety regulations (Drechsel et al. 2023). This review examines the contamination of vegetables primarily through irrigation with water contaminated with faecal matter. Faecal contamination introduced through food handlers' poor personal hygiene and food practices was also accounted for. Other contaminants, such as heavy metals and contaminants of emerging concerns that might require source control or other solutions not mentioned here, might also need to be considered. Experiences of adoption of organic farming practices, which also depend on premiums for ‘safer’ food, have not been included since these topics are extensively covered in the literature.

This review uses the COM-B behavioural framework to synthesise experiences with different approaches adopted to reduce the health risks associated with the use of wastewater contaminated with faecal pathogens in informal vegetable value chains in LMICs. It shows how most interventions focus on enabling the adoption of good practices by providing stakeholders with opportunities or strengthening their capabilities. Fewer interventions aim to stimulate stakeholder motivation to adopt safe practices, and those that do often simply convey their designers' rationales or struggle to identify sustainable monetary or other explicit incentives. Technical innovation, regulations, education, training, and explicit incentives are helpful and may be necessary but are often not sufficient. This review suggests that intervention designers should consider less-used strategies, including implicit incentivisation, promoting good examples to emulate, and persuasion, which can reinforce and complement technical and management controls and increase their impact. It is equally important not to consider behaviour change approaches as a shortcut to transferring responsibilities to primary stakeholders without investing in the required infrastructure or services. To improve their impact, hygiene and food safety interventions should address the opportunity, capability, and motivation needs of both primary stakeholders (farmers, vendors, and/or consumers) and secondary stakeholders, such as supporting institutions. More participatory approaches that include stakeholders early in the design process would likely increase their buy-in and the sustainability of the intervention. This aligns with the WHO recommendation that governments recognise that public agencies, farmers, food business associations, consumers, and community representatives are interdependent (WHO 2018). We suggest that where the multi-barrier approach remains the best option, urban farmers, vegetable traders, street food vendors, and consumer representatives should be encouraged to contribute to existing multistakeholder platforms to develop a shared vision of the value chain that ensures all their opportunities, capabilities, and motivations are catered for.

The authors would also like to thank Kate Medlicott for comments on the manuscript draft.

DG conceptualised the study; carried out the search, analysis, and interpretation; and drafted the review. RES and KVG supervised the study, reviewed the search, and edited the manuscript. PD and BEE reviewed and critiqued the manuscript for clarity and coherence.

This work was supported by the UKRI Engineering and Physical Science Research Council (EPSRC) through a PhD studentship received by the first author as part of the EPSRC Centre for Doctoral Training in Water and Waste Infrastructure and Services Engineered for Resilience (Water-WISER). EPSRC Grant No.: EP/S022066/1.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.