Within the activities of the project MATRIX, a multiple-choice questionnaire was created for each of the four hazards (Salmonella, Campylobacter, L. monocytogenes, and Hepatitis E virus), to gather the necessary information for the mapping of the existing food chain surveillance activities from national experts in the field. As an adaptation of the approach from ‘farm-to-fork’ to ‘farm-to-patient’, each questionnaire was divided into three different sections: (I) focusing on the animal health aspects (AH), (II) on the food safety aspects (FS), and (III) on public health (PH). In each section, the surveillance was assessed by gathering information on actors, sampling context, collected sample types, laboratory methods for diagnosis, available data sources, and cross-sectoral collaboration in place. To ensure to include all the relevant information, eight experts were consulted during the implementation of the specific questionnaires for each sector.

The draft version was circulated amongst the MATRIX participants for evaluation and implementation. The MATRIX partners were asked to suggest possible contact persons with expertise in the specific field of interest, between project partners and non-partners institutions. The identified experts were individually contacted to verify their interest and availability in taking part in the survey. The final version of the questionnaires was put online on the survey platform Survey Monkey©, for dissemination to the relevant experts previously selected. Given the specificities of the information required, a PDF version of the questionnaires (see Supplementary material, modified with permission from Cito et al., 2022 (15)).

A questionnaire was considered completed when answers from the three involved sectors (AH, FS, PH) were obtained. Upon the reception of the three compiled sections, a preliminary evaluation of the results was carried out. Where missing or unclear information emerged, we requested clarifications by re-sending the questionnaire to the reference expert (or to a different one). For this reason, the questionnaires were open for completion for a period of about six months.

In order to evaluate and display the collected information, a categorisation was put in place: information was classified as part of ‘data’, ‘metadata’, ‘events’, ‘event producing data (EPD)’, and/or ‘identified data source (IDS)’ (15).

The subsequent step was then the identification of the most relevant information, for their graphic representation on a map. Therefore, the information regarding the actors, the sampling context, the collected sample types, the laboratory methods in use in the diagnosis, and the available data sources, for each one of the sections, were highlighted. For the purpose of the task, we designed a template of the mapping and displayed it using MS PowerPoint^© (Figure 1).

One of the main objectives of the MATRIX project was to map the surveillance systems along the food chain. To achieve this objective, we selected a specific food chain to be investigated in detail per each hazard. Combinations that are relevant from the public health point of view were selected, based on a consensus among the MATRIX Consortium on the epidemiological situation in 2020 in Europe.

Concerning Listeria, the selected food chain was dairy products, given the epidemiological relevance of these products for the transmission of L. monocytogenes to humans. The investigated country was Norway, because of the economic and cultural importance of dairy products (23, 37).

Regarding Salmonella, we decided to assess surveillance activities in France in the pork meat food chain to avoid overlapping with the OHEJP project NOVA (38), which investigated the poultry food chain with regard to Salmonella surveillance activities. For this reason, some information was already available, while less information existed for the pork meat food chain and the same pathogen.

In Norway, the national and regional surveillance programmes in place are designed to detect illness cases among humans and animals, and non-compliance to food safety criteria in food, adapted to different production routes (Figure 2).

In France, the Salmonella surveillance is based on a national system composed of approximately fifteen components or networks (36). The system covers the entire food chain and most populations who are more at risk for these pathogens. Surveillance aims at reducing the risk for consumers through earlier detection of contamination by Salmonella in the food chain, limiting the economic impact of these contaminations in the production chains, and advancing knowledge.

The French Public Health Institute, named ‘Santé publique France’ (SpF), defines a foodborne outbreak at the national level as the occurrence of at least two cases of similar symptomatology, generally gastrointestinal, which are attributed to the same food origin. The notification of cases has been mandatory since 1987. A notification can lead to investigations through the whole food chain and within different animal and food production sectors (Figure 3). In the past, the pork food chain has been impacted on several occasions by Salmonella contamination (40, 41).

The methodological approach adopted during the MATRIX project included the use of online questionnaires to collect information about surveillance in place in European countries. Our approach allowed for a substantial set of information to be obtained, in terms of both quality and quantity.

Although in some cases surveillance activities are regulated by the existing European legislation [i.e., control programmes regarding Salmonella (42), official controls under Regulation 2017/625 (43) to verify that food complies with microbiological and process hygiene criteria established by Regulation 2073/2005 (44) or epidemiological surveillance of communicable diseases (45)], in other there is no harmonised surveillance in the European Union. For this reason, the collection of information from the existing European legislation would have represented only a fraction of the overall amount of information gathered by the questionnaires.

The questionnaires mainly asked closed questions with multiple-choice answers and checkboxes. This can potentially lead to biases, defined as a ‘deviation of results or inferences from the truth, or processes leading to such a deviation’ (46). The biases may particularly result from the design of the questions and questionnaires, and/or from their modalities of administration and completion (46). Semi-directive interviews may have allowed for collecting information that is more comprehensive. However, the conduction of interviews would have been more time-consuming and potentially introduced a greater risk of biases, given the interviewer’s subjectivity. Moreover, the use of questionnaires was a good alternative to in-person workshops, which were not feasible during the period of travel restrictions due to the SARS-CoV-2 pandemic. The questionnaire and the subsequent mapping made possible the drawing up of the initial description of the surveillance structure as the starting point for working collectively, and in more detail on each aspect.

When using online questionnaires to collect information, the implementation can be an involved process, and it requires resources with expertise to design, pilot, and put them online. Both compiling and responding to the questionnaires also require deep knowledge of the subject. Therefore, depending on the involved expert in the compilation and response respectively, possible biases may be introduced. In addition, the splitting of the questions according to the three investigated sectors could not be sufficient, because even within the same sector the skills are diversified. As consequence, it could not be expected that each expert had the expertise to cover all aspects included in a single sector questionnaire (i.e., from the surveillance programmes in place, to existing information systems, and to laboratory tests used for diagnosis).

To mitigate these risks, we applied the approach of involving, first, a country expert within the OHEJP MATRIX partner institutes and asking them to share the questionnaires with the appropriate experts, which could belong to different agencies. In this way, we gathered information not only from project partners but also from all three sectors involved in the surveillance of the pathogen under investigation.

The mapping process could be a key step in initiating collaborative work to set up or improve a surveillance system. It seemed essential to clearly identify the actors involved in the monitoring, their role, and their position in the organisation, before considering implementation or possible adaptations and changes as actions, to achieve pre-established consensual objectives (36).

Although some examples of mapping were already available (47), we designed a new template to display the relevant actors and other data regarding HT-specific surveillance. The key aspect of the mapping is the presentation, with a single figure, of the three investigated sectors, and for each sector the implemented surveillance activities. In this way, a clear visualisation and a quick comparison of the information reported is possible and the One Health approach is represented.

The three involved sectors were animal health, food safety, and public health. Besides the food safety area, the OH approach can be applied to many others, covering complex health issues and requiring close collaboration across sectors, stakeholders, and countries (48). Hence, our template can be applied to several different contexts, by simply adjusting the underlying structure. Beyond the purpose of the MATRIX project, in which a method to display/map surveillance activities was developed, the same method could be applied to several other scenarios. As a generic approach, the implementation of this template could facilitate also the description of areas within chemical monitoring, for example, using a preliminary adaptation of the questionnaire. Across further applications, the mapping approach could cover a whole production sector, impacted by several contaminants, or a specific contaminant monitored by multiple production sectors.

In this study, we emphasised the methodology rather than the data collected using the questionnaire. Significantly more data than those shown on the maps were collected. The complete results are enclosed in a specific deliverable of the MATRIX project (15). Here, we presented the application of the mapping of L. monocytogenes in Norway and Salmonella in France, as they were representative of two situations in which such information was thoroughly reported.

The discussion with the experts on the two case studies highlighted how communication between official partners is generally more efficient when colleagues from different sectors know each other. Direct familiarity and trust can be important added values for successful surveillance and outbreak investigations (49).

The mapping clearly showed that surveillance of the animal and food sectors needs to be specifically designed to catch the production, processing, and use of the food products, by covering features such as seasonality, regional differences within a country, and large versus small-scale productions. The mapping method could be particularly useful in the case of a food category with a domestic market and small-scale producers, to follow up with the producers who do not have the size or economy to carry out many analyses. The additional value of using this approach, besides building connections and trust among authorities and producers, is to identify conditions that could lead to outbreaks, rather than detecting outbreaks when they have already started. The approach of having sampling schemes designed for the detection of risk factors within each sector, and combined with suited characterisation analyses and data sharing with other relevant sectors, can result in cost savings and rapid detection of OH challenges, regardless of the original purpose of the surveillance programme.

For the food health segment, the focus has been placed on the consumers. It is possible to arrange different surveillance programmes for various vulnerable groups, but this aspect is already targeted in passive surveillance systems, when consumers go to the doctor if they are ill. The human health surveillance programme operates in a similar manner, regardless of the food segment covered. The contact between animal, food, and the human sector is likely to be easier for domestically produced and consumed food, as the options for signaling are more between people who know each other and work together on a regular basis, than if animal, food, and human health segments need to be alerted with official channels first.

However, it is critical to define the specific situations under which other sectors should be alerted, and what information (in terms of data and metadata) should be shared among the different identified actors. Generally, the implementation of the OH approach is easier under the circumstance of an outbreak, since all the involved actors have the common goal of identifying the source of the infection and implementing control measures. The same thing does not happen during routine surveillance. Therefore, there is a general need for ‘traffic lights’ and checkpoints, about what to share, when, and why. While it is true that trust is important for sharing and respecting the rules agreed upon, active communication between sectors is a prerequisite for building trust. Collaborations are established gradually, based on the adhesion of the partners to a common organisation. A mapping stage could therefore be a prerequisite for establishing a shared and integrative vision of the organisation of surveillance activities, as a ground for further collaborative efforts. As an example, an approach to OH surveillance of listeriosis was suggested already in 2001 from France but was not followed up by other countries (50). The current work in France and Norway to improve the efficiency of food hazard surveillance throughout the food chain is highlighting how long, sensitive, but successful, the process is.

However, these food-borne hazards are not solely present within specific countries but are widespread in Europe and beyond. Because animals, food, and people move between countries, establishing links between specific country hazard maps would be useful. Likewise, efforts towards a OHS should be first made at the national level, and at some point linked internationally.