EFSA Supporting Publications最新文献

This document provides the conclusions of the pest survey card that was prepared in the context of the EFSA mandate on plant pest surveillance (M-2020-0114) at the request of the European Commission. The full pest survey card for Dendroctonus micans is published and available online in the EFSA Pest Survey Card gallery at the following link and will be updated whenever new information becomes available: https://efsa.europa.eu/plants/planthealth/monitoring/surveillance/dendroctonus-micans

This document provides the conclusions of the pest survey card that was prepared in the context of the EFSA mandate on plant pest surveillance (M-2020-0114) at the request of the European Commission. The full pest survey card for Gilpinia hercyniae is published and available online in the EFSA Pest Survey Card gallery at the following link and will be updated whenever new information becomes available: https://efsa.europa.eu/plants/planthealth/monitoring/surveillance/gilpinia-hercyniae

In 2025, The European Food Safety Authority (EFSA) organised two crisis preparedness trainings: one external crisis exercise simulating a “One Health” crisis response to a vector-borne disease outbreak, and one internal crisis exercise to assess and discuss EFSA's role in crisis communication during an animal health incident under Regulation 178/2002, which escalates into a public health crisis under Regulation 2022/2371 on Serious Cross Border Threats to Health (SCBTH). The exercises were also aiming at testing EFSA's procedures for crisis response. The two-and-half-day external exercise was co-organised with the German Federal Institute for Risk Assessment, and consisted in presentations, facilitated discussions and a crisis simulation exercise based on a fictional, multifaceted scenario combining animal health, public health and food safety aspects. The training mainly focused on how to integrate and implement a national and multi-agency response process to ensure a united crisis response under “One Health” approach. The internal crisis preparedness training focused on i) outlining EFSA's role regarding crisis communication, and the differences whether EFSA's assistance is requested under EU General Food Law (Reg. (EC) 178/2002) or for a serious cross border threat to health (Reg. (EU) 2022/2371), and ii) reviewing the adequacy of EFSA crisis procedures. The objectives of both events were achieved, based on recorded outcomes and feedback provided by participants. This Annual Report provides a concise summary of the external training, which was subject to a dedicated event report (EFSA, 2025), and a more detailed overview of the internal event, including its development, content and conclusions, as well as the evaluation received from the participants.

Highly pathogenic avian influenza (HPAI) is a threat to poultry production. It is desirable to be able to forecast HPAI outbreaks to allow for the implementation of elevated biosecurity measures. The Bird Flu Radar tool is an early warning system for HPAI based on wild bird movement and abundance. Here we develop the wild bird movement component of the Bird Flu Radar model by exploiting abundance data, which have greater spatio-temporal coverage than movement (ring-recovery or tracking) data. We explore two approaches for estimating bird movement from abundance data, building on recent migratory connectivity studies. In the first, week-to-week movement between areas of high abundance was estimated using a graph-theoretic approach, with abundance in the intervening area also informing connectivity between locations. In the second, movement from breeding areas to wintering areas and back was simulated using an individual-based model, the parameter values of which were calibrated for each species using weekly abundance maps. The output pseudo-movements from the individual-based model were easily integrated into the long-distance movement model in the early warning system for HPAI, to update the long-distance movement estimates for all 25 wild bird study species. Overall, we find that there are fundamental shortcomings of abundance data for inferring bird movement. However, when the accuracy of abundance-derived pseudo-movements can be confirmed, then they can complement ring-recovery or tracking data. Spatio-temporal coverage is still sparser for movement data than for abundance data, and so efforts to develop methods to exploit abundance data are likely to be useful in future endeavours estimating bird movement, and in downstream applications such as forecasting HPAI transmission.

In accordance with Article 31 of Regulation (EC) No 178/2002, in conjunction with Regulation (EC) No 396/2005, the European Commission requested EFSA to provide a technical report presenting comparative dietary exposure calculations performed with PRIMo 4 and the currently used version of PRIMo (rev. 3.1). These comparisons shall allow to assess the impact of an implementation of PRIMo 4 on the outcome of the risk assessment performed in the context of setting MRLs under Regulation (EC) No 396/2005 and Regulation (EC) No 1107/2009. EFSA therefore performed detailed analysis to assess the changes on acute and chronic exposure calculations resulting from the use of PRIMo 4 compared to the current version of PRIMo rev. 3.1. The assessment covered both theoretical calculations using default input values and practical examples selected from recently performed risk assessments.

This report investigates risk perception and behaviour on social media platforms developing an original and replicable methodology that integrates Natural Language Processing (NLP), complex network analysis, emotion and sentiment analysis, and cluster analysis. The methodology is applied to two case studies focusing on social media discussions about “avian influenza” and “PFAS”, analysing over 41.000 and 24.000 posts respectively from the EU27 and Great Britain between 2023 and 2025. Different analytical breakdowns allow for a comprehensive understanding of the complex dynamics observed on social media. Distinct subtopics were identified, providing insights of public discourse. For avian influenza, discussions were oriented on themes such as “Cross-Species Transmission in Farming” and “Media News, Contested Narratives & Emotional Reactions”. For PFAS, themes included “Health Risks and Scientific Studies on PFAS Toxicity” and “Local Contamination and Community Impact”. Fear emerged as the predominant emotion in both topics, indicating high levels of perceived risk among users, particularly concerning cross-species transmission and distant outbreaks of avian influenza, and health risks associated with PFAS. Most users were classified as “normal users”, characterised by a moderate number of friends and followers. Other relevant groups included the “News and Main Information Networks” and “Public Interest Networks”. Network analysis revealed that users tend to interact predominantly with others within their own country (assortativity), while users of specific types tend to interact with those of different types (disassortativity). Network indicators were computed to assess users’ influence in information dissemination, and communities were identified to highlight groups with stronger internal exchanges than with the rest of the platform. Despite high levels of expressed concern, only limited explicit behavioural declarations were detected in users’ posts, possibly suggesting that users are unsure how to react to situations of concern due to freeze response, or do not disclose their offline behaviour on a social media platform.

The main goal of this report is to assesses the performance of different quantile estimators with respect to estimating both non-extreme and extreme quantiles. An application within the field of exposure assessment is presented, also including the evaluation of different methods for usual intake estimation. The report also compares methods to calculate prediction intervals from historical control data (HCD). For quantile estimation, the standard sample quantiles were found to perform well for estimating non-extreme quantiles. In specific situations listed in the report, the semi-parametric method by Wei, Wang and Hutson is a good alternative. For estimating extreme quantiles, especially the methods based on the generalized extreme value distribution and the generalized Pareto distribution could be recommended as alternatives to the sample quantiles, which would typically require unrealistically high samples to perform well for extreme quantiles. With respect to usual intake and exposure estimation, in addition to the default observed individual mean (OIM) method that is currently applied at EFSA, alternative parametric methods are presented and their performance compared through simulations. It was seen that both the logistic-normal-normal or gamma hurdle model might provide valuable alternatives to the OIM, for which it has been observed that two-day consumption information might not always be sufficient to obtain reliable quantile estimates. For the prediction intervals from HCD, the default method at EFSA is based on mixed model theory. The performance of the default method depends on the ratio of between-to-within-study variability. When the ratio is larger than 1, good coverage was observed when at least 30 studies were included. When the ratio is smaller than 0.5, or when it is between 0.5 and 1, but less than 100 studies were considered, the bootstrap approach from Nagashima and colleagues was found to be a promising alternative. Further recommendations are provided in the report.

Extended-Spectrum β-Lactamases/AmpC β-Lactamases (ESBL/AmpC)-producing E. coli are increasing globally and represent a significant threat to public and animal health. These bacteria belong to the normal microbiota of humans and animals and are widespread in various environmental settings. Understanding their transmission dynamics through a One Health approach is crucial. This study aimed to investigate the transmission of ESBL/AmpC-producing E. coli among animals, food, the environment, and humans using phenotypic methods and Whole Genome Sequencing (WGS). The study included isolates collected and stored between 2014 and 2017, as well as those prospectively collected from 2018 to 2020. Human isolates were obtained from urine and blood cultures using standard methods and tested for antimicrobial susceptibility following EUCAST guidelines. Non-human isolates were cultured according to the EURL-AR protocols and susceptibility tested in line with Commission Implementing Decision 2013/652/EU. WGS was performed using the Illumina MiSeq platform and analysis done according to EURL-AR protocol. During the study period, 903 individuals were identified with ESBL/AmpC-producing E. coli. Every fourth isolate (n=226) was selected for WGS, of which 195 passed quality control. The prevalence of ESBL/AmpC-producing E. coli increased from 2.5% in 2014 to 5.4% in 2020, with the majority associated with ST-131 and CTX-M-type β-lactamase genes. A total of 154 non-human ESBL/AmpC-producing isolates were collected, comprising 14 from food, 108 from livestock, 12 from pets and 20 from the environment. Of these, 131 were successfully recovered and passed WGS quality control. These non-human isolates belonged to 55 different STs, each with low individual prevalence and up-regulated chromosomal AmpC was predominant. Generically related ESBL/AmpC-producing E. coli isolates were identified between animals and food sources, as well as between humans and environmental samples. However, no identical or related isolates were found between humans and animals or humans and food. Due to the low number of non-human ESBL/AmpC isolates, a formal risk assessment of resistance transfer from animals/food to humans could not be conducted.