Ana Allende

Food safety is often framed around the implicit expectation that food should be as free from risk as possible. Advances in hygiene, processing and preservation technologies, genome sequencing, high-resolution mass spectrometry, analytical performance, regulation, and surveillance have reduced the burden of foodborne disease worldwide. Yet paradoxically, these same advances have intensified societal and regulatory pressure toward the pursuit of zero risk (1). Scientific reality tells a different story. All foods inherently carry a residual risk (2).Increasingly sensitive sampling schemes and analytical methods now detect hazards at ever lower levels, but the ability to detect a hazard does not necessarily correspond to a meaningful public health risk (3). It may also generate unintended consequences, including food waste, supply-chain disruptions, and environmental burdens.In their Frontiers in Science Lead Article, Wiedmann et al. (4) argue that zero-risk thinking is neither achievable nor desirable. Instead, food safety governance must move toward proportionate, risk-based decision-making that explicitly acknowledges trade-offs. This argument arrives at a pivotal moment. Food safety is now firmly embedded within broader sustainability, food security, and One Health agendas. Safety can no longer be pursued in isolation from environmental, social, and economic objectives. Acknowledging the limits of zero-risk thinking is not a weakening of protection; it reflects a transition toward more transparent, science-based, and balanced decision-making.Predictive modeling and quantitative risk assessment have transformed food safety over recent decades. These tools support hazard identification, exposure assessment, risk characterization, and risk-based management decisions (5). As described by Wiedmann et al., they have transformed our capacity to anticipate and mitigate foodborne risks. However, as food safety challenges become more intertwined with environmental sustainability, resource constraints, climate variability, and global trade, the limitations of existing predictive frameworks become more apparent. Many tools remain hazard-centric or focused on single-risk outcomes. While the One Health concept has gained widespread recognition, its practical implementation often remains fragmented. Human, animal, plant and environmental dimensions are frequently assessed separately rather than in an integrated manner (6). As a result, current predictive modelling approaches may struggle to capture system-level trade-offs. Actions taken to reduce risk may inadvertently amplify others, for example, increasing energy use, food waste, or inequities in access to safe food. In a rapidly changing global landscape characterized by climate disruptions and supply chain instability, siloed decision-making risks becoming increasingly ineffective. Addressing contemporary food safety challenges therefore requires not only more advanced models and data repositories, but also multidisciplinary frameworks capable of integrating diverse data streams, values, and objectives. Without explicit consideration of trade-offs and co-benefits, One Health risks remaining a conceptual aspiration rather than a fully operational approach to governance.As food safety becomes embedded within broader goals of food security, sustainability, and social stability, risk management can no longer be treated as a technical exercise. Decisions about acceptable levels of risk inevitably involve value judgments and prioritization across competing objectives. Quantitative approaches have already demonstrated that interventions in areas such as cold-chain management may reduce foodborne illness while affecting energy consumption or food waste (7). Such analyses illustrate that trade-offs are inherent, not exceptional.Within this context, risk negotiation emerges as a necessary evolution of food safety governance. Risk negotiation does not imply compromise without rigor. Rather, it refers to a structured, evidence-based process that makes trade-offs explicit and assessable. By integrating predictive modelling with trade-off risk assessment, the framework proposed by Wiedmann et al. (2026) supports decision-making under complexity and uncertainty.Importantly, this approach does not aim to seek universal solutions. It enables fit-for-purpose decisions that reflect the characteristics of specific hazards, food matrices, production systems, and populations. This shift is consistent with developments in international guidance. Work within the Codex Alimentarius framework, including JEMRA (Joint FAO/WHO Expert Meetings on Microbiological Risk Assessment) deliberations on water use in food production (8,9), reflects a gradual transition from strictly hazard-based criteria toward risk-based and context-dependent approaches. Equivalent levels of public health protection may be achieved through different control strategies, depending on local conditions and constraints. Proportionality and flexibility should not be viewed as weaknesses. They are essential features of effective food safety systems operating within complex global environments.Global food systems operate across highly diverse production, environmental, and socioeconomic contexts. Differences in climate, infrastructure, regulatory capacity, and resource availability shape risk profiles and management options. Yet food safety standards often aspire to uniformity. The rigid application of hazard-based criteria across all contexts may disproportionately affect regions with limited resources, even where safe and sustainable production is feasible. This is not an argument for lowering protection standards. Rather, it recognizes that equivalent levels of public health protection can be achieved through different, context-appropriate pathways.Risk-based and fit-for-purpose approaches allow flexibility while maintaining health-based objectives. Trade-off risk assessment provides a structured...

Controlled environment agriculture (CEA) offers protection from external contaminants but introduces unique food safety challenges, particularly related to equipment sanitation. This study evaluated the occurrence and persistence of Listeria monocytogenes and related species in a commercial soil-based CEA facility through a 1-year environmental monitoring program (EMP). A total of 169 samples, including baby leaves, soil, water, and swabs from equipment surfaces, were collected across three visits. L. monocytogenes was detected in only one sample (0.59% prevalence), whereas Listeria innocua was isolated from harvesting crates and structural surfaces (3/169). Molecular confirmation and characterization were conducted using PCR ( hly, iap, sigB ), matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS), and whole genome sequencing (WGS). To assess the potential for cross-contamination, L. innocua isolates were used for inoculation studies on reusable plastic crates. Viable cells survived up to 24 h post-inoculation despite water rinsing under 25°C and 60%–70% relative humidity. These findings suggest a low risk of L. monocytogenes contamination in this soil-based CEA system but underscore the importance of targeted sanitation, particularly for harvesting equipment, to prevent harborage and transfer of Listeria spp. IMPORTANCE: Controlled environment agriculture (CEA) is a rapidly expanding approach to producing fresh food year-round with greater resource efficiency. However, it presents unique challenges for managing foodborne pathogens. This study demonstrates that in this soil-based indoor system, contamination risks persist, particularly via inadequately cleaned harvesting equipment. The boot cover finding is interpreted as an indicator of environmental contamination, but it does not constitute evidence of footwear-to-crop transfer risk. Although Listeria monocytogenes was rarely detected, related species like Listeria innocua persisted on reusable plastic crates and structural surfaces, highlighting weaknesses in current sanitation protocols. The use of boot covers as environmental monitoring tools proved valuable for detecting contamination sources. These findings underscore the need for tailored sanitation strategies and comprehensive environmental monitoring programs to enhance food safety in CEA systems and prevent pathogen harborage and spread.

Antimicrobial resistance poses a growing threat to public health, and integrated surveillance strategies across environmental compartments such as treated wastewater and biosolids can substantially improve monitoring efforts. A key challenge is the diversity of available protocols, which complicates comparability for the concentration and detection of antibiotic resistance genes (ARGs), particularly in complex matrices. In this study, we compared two commonly used concentration methods—filtration–centrifugation (FC) and aluminum-based precipitation (AP)—and two detection techniques, quantitative PCR (qPCR) and droplet digital PCR (ddPCR), for the quantification of four clinically relevant ARGs: tet(A), bla_CTX-M group 1, qnrB, and catI. Analyses were performed in both secondary treated wastewater and biosolid samples, including their purified bacteriophage-associated DNA fractions. Results showed that the AP method provided higher ARG concentrations than FC, particularly in wastewater samples. ddPCR demonstrated greater sensitivity than qPCR in wastewater, whereas in biosolids, both methods performed similarly, although ddPCR yielded weaker detection. Importantly, ARGs were detected in the phage fraction of both matrices, with ddPCR generally offering higher detection levels. These results provide comparative insights into established methodologies and highlight the value of selecting appropriate protocols based on matrix characteristics and surveillance objectives.

This study introduces a farm-to-fork quantitative risk assessment (QRA) model for invasive listeriosis from ready-to-eat diced cantaloupe. The modular model comprises seven stages—preharvest (soil and irrigation contamination), harvest (cross-contamination and survival), pre-processing (brushing), processing (flume tank washing, dicing and equipment cross-contamination), lot testing, cold-chain transport and retail growth, and consumer storage/handling. Each stage employs stochastic functions to simulate microbial prevalence and concentration changes (growth, inactivation, removal, partitioning, cross-contamination) using published data. In a reference scenario—good agricultural practices (soil barriers, no preharvest irrigation), hygienic processing and proper cold storage—the model predicts low lot- and pack-level contamination, with few packs >10 CFU/g and most servings below detection; the mean risk per serving is very low. “What-if” analyses highlight critical control points: the absence of soil barriers with preharvest irrigation can increase the risk by 10,000-fold; flume tank water contamination has a greater impact than harvest-stage cross-contamination; and poor consumer storage can raise the risk by up to 500-fold. This flexible QRA framework enables regulators and industry to evaluate and optimize interventions—from improved agricultural measures to targeted sampling plans and consumer guidance—to mitigate listeriosis risk from RTE diced cantaloupe.

Wastewater reuse for agricultural irrigation is increasingly essential, but it carries potential public health risks due to the dissemination of antimicrobial resistance (AMR). This study evaluates the effectiveness of four tertiary wastewater treatment technologies—peracetic acid (PAA), PAA combined with low-intensity ultraviolet-C (PAA/UV Low), high-intensity UV-C (UV High), and ultrafiltration (UF)—in reducing extended-spectrum β-lactamase-producing Escherichia coli (ESBL-E. coli) and antimicrobial resistance genes (ARGs) in reclaimed water used for irrigation. The relative abundance of the genes, normalized to the 16S RNA gene present in the water samples, was then estimated to assess whether there is an amplification of these genes during the reuse process in the wastewater treatment plant (WWTP). The results indicate that while all treatments significantly reduced ESBL-E. coli (≥3 logs cfu/100 mL) and ARGs (≥ 1.5 logs gc/100 mL), complete elimination was not achieved in any WWTP. Among the treatments, UF demonstrated the highest removal efficiency (≈4 log gc ARG/100 mL), against ARGs, followed by UV High (≈3 log gc ARG/100 mL), whereas PAA and PAA/UV Low were less effective (≈2 log gc ARG/100 mL). The study also found that while absolute ARG levels were reduced, their relative abundance remained stable or showed minimal decline, suggesting a persistent environmental reservoir of resistance genes. Among the ARGs analyzed, the most frequently detected were associated with tetracyclines (tetW, tetA), quinolones (qnrB, qnrS), and sulfonamides (sul1, sul2), highlighting potential public health concerns. Moreover, multidrug-resistant (MDR) ESBL-E. coli isolates were present across all WWTPs, exhibiting resistance to β-lactams, quinolones, tetracyclines, and sulfonamides. Nevertheless, notably low levels of resistance to last-resort antibiotics (tigecycline, colistin, and meropenem) were observed. These findings underscore the critical role of tertiary treatments in mitigating antimicrobial resistance (AMR) risks in water reuse systems. However, the persistence of ARGs in effluents suggests that current WWTP processes require further optimization.

A quantitative risk assessment (QRA) model was developed to evaluate the risk of invasive listeriosis from the consumption of non-ready-to-eat (non-RTE) frozen vegetables. On a lot basis, the QRA model simulates Listeria monocytogenes concentration and prevalence in a “Processing module” that comprises blanching, potential recontamination and packaging, any post-packaging inactivation treatment, and within-lot end-product testing and in a subsequent “Consumer’s handling module” that encompasses portioning of frozen vegetables, defrosting, and cooking. Based on available published data, the model was coded in nine sequential R functions designed to assess the effectiveness of blanching, the improvement in processing environment hygiene, the implementation of sampling schemes at the end of processing, and improved consumer instructions on the product’s package. In a reference scenario, the model estimated that 9.4% of 500 g packages of frozen vegetables would be contaminated, although at mean levels lower than 10 CFU/g, and assuming that 20% of the portions of frozen vegetables would be left to thaw at room temperature for 2 h, the lot-level mean risk of listeriosis in the susceptible population would be 2.935 × 10⁻¹⁴ (median 5.446 × 10⁻¹⁵) for uncooked 50 g servings and 2.765 × 10⁻¹⁷ (median 5.184 × 10⁻¹⁸) for cooked 50 g servings. Analysis of selected scenarios suggested that not cooking the non-RTE product contributes to the risk to a greater extent than the level of contamination in the incoming raw vegetables, the latter in turn being more influential than the level of contamination in the processing environment. The QRA model is freely available as an R package with full documentation and can be used as a tool to inform the consideration of strengthened risk management measures in view of the current changes in consumer behavior and new diet trends.