indicadores de peligros emergentes

8/12/2019 Indicadores de Peligros Emergentes

1/18

Review

Indicators of emerging hazards and risks to food safety

Gijs A. Kleter * , Hans J.P. MarvinRIKILT Institute of Food Safety, Wageningen University and Research Center, Building 123, Bornsesteeg 45, Wageningen, The Netherlands

a r t i c l e i n f o

Article history:Received 27 August 2007Accepted 31 July 2008

Keywords:Food safetyEmerging hazardsIndicatorsEarly warningFood inspection

a b s t r a c t

There is a widely felt need to develop methods for the early identication of emerging hazards to foodsafety with the aim of preventing these hazards from becoming real risks and causing incidents. Thispaper reviews various activities and previous reports that describe methods to select indicators thatcan be used for the purpose of early identication of hazards. These indicators have been divided overthree different environments, including (i) the environment surrounding food production, (ii) the foodproduction chain from farm to fork, and (iii) consumers. Changes in these indicators are signals thatmay require follow-up action. Besides indicators that are linked to specic kinds of hazards, the indica-tors used for vulnerability assessment can help identifying weak spots in the food production system thatare sensitive to a broader range of hazards. Based on the various indicators for emerging hazards thathave thus been identied in literature, a set of generic indicators is provided that can be useful for theearly identication of hazards.

2008 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13. Categorization of methods for the early identification of emerging hazards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10244. Identification of hazards during food production from farm to fork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1024

4.1. Food system vulnerability assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10244.1.1. United States (US) Government approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10244.1.2. Other approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1025

4.2. Situational awareness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10264.3. Good practices and HACCP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1024.4. Prediction of the occurrence of mycotoxin-related hazards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10274.5. Factors influencing the occurrence of chemical and biochemical hazards in food and feed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10274.6. Trend analysis based on data from food law enforcement activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1028

5. Epidemiology and surveillance supporting early response to emerging food safety risks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10285.1. Zoonotic disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105.2. Food-borne human disease. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1029

5.2.1. Surveillance systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10295.2.2. Retrospective studies on trends. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1030

0278-6915/$ - see front matter 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.fct.2008.07.028

Abbreviations: BSE, bovine spongiform encephalopathy; CDC, Centers for Disease Control and Prevention; DON, deoxynivalenol; EC, European Community; ECDC,European Center for Disease Control; EEA, European Environmental Agency; EFSA, European Food Safety Authority; EFTA, European Free Trade Association; ERS, EconomicResearch Service; ESA, EFTA Surveillance Agency; EU, European Union; Eurostat, Statistical Ofce of the European Communities; FAO, Food and Agricultural Organization;FAOSTAT, Statistical Ofce of the FAO; FHB, Fusarium Head Blight; FDA, Food and Drug Administration; FIVIMS, Food Insecurity and Vulnerability Information and MappingSystems; GAP, Good Agricultural Practice; GHP, Good Hygienic Practice; GIEWS, Global Information and Early Warning System; GM, Genetically modied; GMP, GoodManufacturing Practice; GOARN, Global Outbreak Alert and Response Network; GPHIN, Global Public Health Intelligence Network; HACCP, Hazard Analysis Critical ControlPoints; INFOSAN, International Food Safety Authorities Network; IPCS, International Program on Chemical Substances; ITX, isopropyl-thioxanthone; OIE, World AnimalHealth Organization (Ofce International des Epizooties); OTA, ochratoxin A; PFGE, pulsed-eld gel electrophoresis; POP, persistent organic pollutant; RASFF, Rapid AlertSystem for Food and Feed; RFID, radio-frequency identication; SPPA, Strategic Partnership Program Agroterrorism; STEC, Shiga-like-toxin-producing Escherichia coli(synonymous to VTEC); UN, United Nations; US, United States; USDA, United States Department of Agriculture; VTEC, verocytotoxin-producing Escherichia coli (synonymousto STEC); VWA, Dutch National Food and Consumer Product Safety Authority; WHO, World Health Organization.

* Corresponding author.E-mail address: [email protected] (G.A. Kleter).

Food and Chemical Toxicology 47 (2009) 10221039

Contents lists available at ScienceDirect

Food and Chemical Toxicology

j ou rna l homepage : www.e l sev i e r. com/ loca t e / foodchemtox
mailto:[email protected]://www.sciencedirect.com/science/journal/02786915http://www.elsevier.com/locate/foodchemtoxhttp://www.elsevier.com/locate/foodchemtoxhttp://www.sciencedirect.com/science/journal/02786915mailto:[email protected]


2/18

6. Identification of hazards in the surrounding environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10306.1. EMRISK: Holistic-approach-based indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1030

6.1.1. Background of the EMRISK project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036.1.2. EMRISK activities and definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036.1.3. EMRISK project outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106.1.4. EFSA Scientific Committees opinion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036.1.5. Dutch study on indicators for emerging mycotoxin hazards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10336.1.6. Dutch policy support project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

6.2. Exploration of potential future issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10346.2.1. Trends affecting the emergence of zoonotic diseases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10356.2.2. Trends affecting the emergence of food-borne pathogens besides zoonoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035

7. Conclusions and common findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction

Food safety has been the topic of some recent policy changes,increased awareness among the public, and various incidents.These developments indicate that there is a need for a system thatcan identify food safety hazards in an early stage so that these haz-ards can be tackled in time, before developing into real risks. Withregard to food safety hazards that are known as such, measures canbe taken towards the prevention and mitigation of these hazardsbased on their characteristics, behavior and point of entry intothe food chain. For example, good practices for agriculture andmanufacturing (e.g. good agricultural practice (GAP)), as well asthe hazard analysis critical control points (HACCP) approach to as-sess risks and control them, are now commonplace in many juris-dictions. Yet it can be envisioned that for a number of risks, suchmeasures may not be applicable given that these risks are yet un-

known or unanticipated.The overall aim of the SAFE FOODS project, which is primarily

sponsored by the European Commission Directorate for ResearchsSixth Framework Program, is to further develop risk analysis of foods based on inputs from advanced research in both the naturaland social sciences. Among other things, this is also likely to con-tribute to the condence of stakeholders in the European Unionsfood safety governance. SAFE FOODS is composed of various WorkPackages that more or less act as subprojects on different topics,including (i) the use of advanced analytical methodologies to studypotential effects of agricultural practices on crop composition; (ii)emerging risks in food safety; (iii) assessment of consumer expo-sure to food safety hazards by the use of advanced statistical tech-niques; (iv) condence of consumers and other stakeholders in risk

management in food safety; and (v) institutional arrangements forfood safety governance. The ndings of all these Work Packages areintegrated into a new model on risk analysis, which will be renedwith inputs solicited from stakeholders.

The early identication of emerging hazards to food safety isalso a major topic of the activities of SAFE FOODS Work Package2. Previously, Work Package 2 has made a number of accomplish-ments on this topic, including (i) the establishment of an expertdatabase; (ii) description of a framework for timely identicationof emerging hazards; iii) reports describing systems for timelyidentication of emerging hazards to food safety or to hazards of another nature that can be exemplary for food safety as well; iv)an analysis of conspicuous trends in European Union (EU) foodsafety alerts; and (v) reports reviewing the background and charac-teristics of a selected range of hazards to food safety caused bymicroorganisms, mycotoxins, biochemical and chemical agents.

Besides Work Package 2, various parallel projects, such as theEMRISK project funded by the European Food Safety Authority(EFSA), have tackled the issue of early identication of emergingfood safetyhazards. Against thebackgroundof thepreviousoutputsfrom these projects, the current publication by authors from SAFEFOODS Work Package 2 aims at describing the various types of indi-cators and, in particular, the methods for their selection that can beused for the early identication of emerging hazards to food safety.This information, together with the working procedures to be re-viewed in a follow-up study, can then further provide a basis forthe development of a system for emerging hazard identication.

2. Denitions

Various terms with specic meanings are used throughout thisreport, which therefore would merit from further clarication.Where appropriate, the denitions used are in line with interna-tionally accepted denitions published by the Food and AgricultureOrganization (FAO) and the World Health Organization (WHO).

A hazard is an agent that has the potential to exert a negativeeffect on health. An example of a hazard in food is the presenceof Salmonella bacteria that may cause gastroenteritis. The risk is de-ned as the negative effect of the hazard if it really occurs, whichdepends on the likelihood of the occurrence- and severity- of thenegative effect ( FAO, 1995, 1997 ).

The internationally harmonized model for scientic risk assess-ment is composed of four phases, namely (i) hazard identication;(ii) hazard characterization, that is, the characteristics of specicnegative health effects and the doseresponse relationships be-tween hazard and effects; (iii) exposure assessment, in which the

exposure of consumers ingesting the food containing the hazardis estimated; and (iv) risk characterization, in which the outcomesof the three preceding phases are combined into an assessmentand in which also uncertainties are taken into account ( FAO,1995 ). To illustrate this with an example: the hazard characteriza-tion may describe the minimum infectious dose of Salmonella ,while subsequently the exposure assessment may help estimatingthe real dose to which consumers are exposed, so that the riskcharacterization can conclude on the likelihood of gastroenteritiscaused by Salmonella.

Besides risk assessment, which is the scientic process assess-ing risks, also two other activities, that is, risk management andrisk communication, are considered to be part of the internation-ally acknowledged risk analysis model for food safety. During risk

management , policy alternatives are weighed based upon the out-comes of the risk assessment process, and measures to control

G.A. Kleter, H.J.P. Marvin / Food and Chemical Toxicology 47 (2009) 10221039 1023


3/18

and mitigate the risks are subsequently dened. Risk communica-tion is the process of exchange of information and opinion amongthe risk managers and risk assessors, but also between risk profes-sionals, such as assessors, managers, and communicators, andother parties involved. Although preferably these activities arestrictly separated from each other, some overlap cannot be avoidedin practice. For example, risk managers have to formulate the pol-

icy for risk assessment, such as the risks to which considerationwill be given and the issues that have priority (e.g. FAO, 1997 ).With regard to emerging hazards, which are the topic of this re-

port, the emerging characteristics of these hazards may have vari-ous causes. For instance, the hazard can be new and has notoccurred before. For example, certain synthetic man-made sub-stances may not occur in nature and are therefore new. In case suchsubstances are hazardous and also enter the food supply, these sub-stances would turn into emerging food safety hazards. The samemay also hold true for hazards that have only occurred in the non-food area, but that also start entering the food domain. Alterna-tively, hazards that once have disappeared from the food domainmayenter it again, forexample dueto changes in practice or theter-mination of certain risk-eliminating measures. In addition, hazardsthat have previously occurred in food, but that have only recentlybeen discovered, can be regarded emerging hazards as well.

The denition of an indicator is taken from the guide on han-dling indicators and signals that has been published as Annex 4to the report of the EMRISK project, which had been carried out un-der auspices of EFSA. This project has carried out various activitieson emerging food safety risks, including retrospective case studieson food safety incidents, and has recommended a working proce-dure for early identication of emerging food risks ( EFSA, 2006a ).An indicator is considered by EMRISK as an entity that indicatesthe possibility that a risk may occur, due to its direct or indirectrelationship with the risk. A signal is dened as a substantialchange in the indicator. Signals can thus be used to ag the possi-ble occurrence of risks. The EMRISK guide also provides a numberof criteria to select appropriate indicators, which will be discussedin more details in the section on EMRISK below.

3. Categorization of methods for the early identication of emerging hazards

For the purpose of oversight, we propose a categorization of early identication methods based on their points of applicationwithin the food production chain ( Fig. 1). Various methods aim

to identify factors outside the food production chain that contrib-ute to the development of food safety hazards within the foodchain. The consideration of such external factors corresponds tothe larger box in Fig. 1. These external factors may already existand thus be used for measurement, such as in the approach recom-mended by EMRISK (see below), or may occur in future, such as infutures research exploring different plausible scenarios for the

future.Other identication methods focus on hazards as they occur,during food production from farm to fork, which corresponds tothe inserted left hand box of Fig. 1. Food control agencies measureknown hazards and report positive detects through systems suchas the EU rapid alert system for food and feed (RASFF), which canprovide useful inputs for research at this stage of the food chain.In addition, food-producing companies have to carry out theirown risk assessments, which can be both business-managementand legal requirements. Within company settings, identied haz-ards are often measured at critical points in the production processaccording to the HACCP approach. Other relevant methods pertain-ing to this stage of the food chain are the methods of vulnerabilityassessment, which helps identifying weak spots in the productionprocess, and situational awareness, which comprises the real-time measurement of many entities in food production and theappropriate processing of all these measurement data into signalsthat ag potential hazards (see below).

After consumption, identication methods aid the early recog-nition of risks as they occur. Please note that the term risk is usedhere because the hazard has already exerted adverse effects andtherefore is not hypothetical anymore. In this stage, it is desirableto contain the risk, for which early recognition will be useful. Earlyidentication of risks is supported, for example, by food-borne dis-ease outbreak reporting systems, including public health surveil-lance and other epidemiological tools. Recent advances havebeen made for example in molecular epidemiology, in which theuse of molecular detection techniques facilitate the discernmentof food poisonings that are related, thus facilitating the early trac-ing back to the source of the adverse reaction. In addition, volun-tary reporting by consumers, such as through telephone callcenters, can be useful in this respect.

Interestingly, identication of emerging hazards in the sur-rounding environment depicted in Fig. 1 includes the use of pre-dictive early warning systems described by Marvin and co-workers in their review on systems for identication of emerginghazards (2009). The identication of emerging hazards in the foodproduction chain relies upon reactive hazard-focused systems,while hazards during or after consumption do so upon reactiveendpoint-focused systems mentioned by the same review ( Marvinet al., 2009 ).

4. Identication of hazards during food production from farmto fork

This section highlights various examples of initiatives and ap-proaches towards early identication of emerging hazards in thefood chain ranging from agricultural production via retail and theconsumer to the public health system. These approaches take var-ious factors into account, each to a varying degree, such as the sys-tem, the local facility and process, and the hazard itself.

4.1. Food system vulnerability assessment

4.1.1. United States (US) Government approachSince the terrorist attacks of September 11th, 2001, various

measures have been put into place by the US government, includ-ing those that aim to protect the food system against becoming a

Food production chain

Early warning againstemerging hazards

Identification throughHACCP, situational

awareness,vulnerabilityassessment, and RASFF

alert system, among others

Consumers

Rapid response to emerginghazards

Identification through outbreakmonitoring, consumer

complaints, and molecularepidemiology, among others

Surrounding environment

Pre-early warning against emerging hazards

Identification through EMRISK model, researchon hazards and onfutures, among others

Fig. 1. Early identication of emerging hazards at different points within the foodproduction chain, consumers, and the surrounding environment.

1024 G.A. Kleter, H.J.P. Marvin / Food and Chemical Toxicology 47 (2009) 10221039


4/18

vehicle for acts of agroterrorism. The activities against potentialagroterrorism are designated food defense and carried out by theStrategic Partnership Program Agroterrorism (SPPA) in cooperationbetween various departments, including, the Department of Homeland Security, Food and Drug Administration (FDA), USDepartment of Agriculture (USDA), Department of Health and Hu-man Services, Federal Bureau of Investigation, and the private sec-

tor. It was realized that part of the hazards potentially introducedinto food by agroterrorism might be new or at least different fromthe ones that had already been known to occur accidentally. Inter-estingly, this situation is highly similar to the emerging hazardsthat are the topic of this review, although the latter ones are con-sidered to be accidental rather than intentional. The methods andindicators that the US authorities use to identify hazards and theweak spots in the food manufacturing system are therefore alsoof interest to the topical study ( FDA, 2006 ).

As regards potential hazards, the FDA has made a classicationof potential threats by using a list compiled by the Centers for Dis-ease Control and further narrowing it down based on a variety of factors, such as the stability of the hazardous agent in food, theability to distinguish the agent on the basis of color or taste, thetoxicity of the agent, the availability, and the public health impact.In addition, chemical and microbiological agents are classiedaccording to characteristics that are important in this regards, suchthe heat stability of microorganisms and the toxins they produce,or the heat stability and acute toxicity of chemical toxins.

One of the food defense methods employed by the US authori-ties is identication of the parts of the food production system thatappear most vulnerable to acts of agroterrorism and which aretherefore designated critical nodes. Therefore a possible pointof entry of a hazard into the system, the critical node, is predicted,while the exact nature of the hazards may not have to be predicted.

The method initially used for vulnerability assessments was asix-step procedure that is part of Operational Risk Management.The six steps consisted of (i) hazard identication; (ii) risk assess-ment; (iii) analysis of risk control measures; (iv) making controldecisions; (v) implementation of risk controls and (vi) supervisionand review. This procedure had previously been used in the aero-space and nuclear energy sectors. For each scenario involving a cer-tain hazardous agent or activity, the risks were classied based ona matrix in which the severity and likelihood of occurrence are off-set against each other. The outcomes could be high, medium,or low risks. Factors that were considered included the accessi-bility to attackers, impact on health, toxicity or pathogenicity of the hazardous agent, and compatibility of the agent with charac-teristics of the food and its processing.

The tool currently employed for the vulnerability assessment isCARVER + Shock. The acronym CARVER stands for Criticality,Accessibility, Recuperability, Vulnerability, Effect, and Recogniz-ability, and each of these terms can be considered an indicator

in the sense of our denition. The term Shock pertains to theanticipated psychological impact of an attack in the short- andlong- term. For each of the terms of CARVER, a score is assignedto specic phases in a ow diagram representing the food facilityor sub-system for a specic product, ranging from 1 to 10, with10 being the highest risk score. In addition, a prole of the poten-tial attacker is developed, which will be further consideredthroughout the assessment. SPPA carries out CARVER + Shock anal-yses out involving teams of 2030 experts from a wide range of backgrounds. After a time period of preparations and instructions,the team will work together for several days to work out the sce-narios. The outcomes are general for a given food production sectorand do not identify weaknesses in a specic company. It is ex-pected that the information developed and experience gained

with these analyses by SPPA will allow companies to carry outtheir own internal CARVER + Shock analyses.

Criticality means the magnitude of the impact on consumershealth and theeconomy, andis rankedbased on theestimated casu-alties and economic damage. The highest score of 10 corresponds toover 10,000 lives lost andmore than 100million US dollareconomicdamage. Accessibility stands for the ease by which food-producingor handling facilities can be accessed by individuals who havenot been authorized to access the point of potential contamination.

The classication is based on physical barriers, restricted access fornon-staff, observation of the facility, and availability of informationon the facility. Recuperability means the ability of a system to re-cover from damage, with a score of 10 corresponding to more thana year needed for recovery. Vulnerability is characterized by theease by which sufcient quantities of a hazardous component canbe brought into contact with the food as determined both by thecharacteristics of the food and agent and by the structure of envi-ronment, such as the capability to work unobserved. Effect relatesto the impact on the production capacity, in other words the partof the system that will be damaged, with a score of 10 designatingmore than 50% of the system being affected. Recognizability relatesto the ease by which the target can be recognized by the attacker,with the highest score signifying a clear recognizability with littletraining being required to recognize the target.

Shock brings together the data from the CARVER evaluationand assesses the impact that an attack, if it would occur, mighthave on the health, psychological, and economic status of the na-tion. Besides the number of casualties and economic impact, alsofactors like religious or symbolic importance of the target andthe sensitivity of subpopulations are considered. Again, a scoreranging between 1 and 10 is assigned.

Once the vulnerability assessment has been completed, coun-termeasures that can be formulated include the restriction of ac-cess to critical nodes, modication of processing characteristicsin order to warrant destruction of certain hazardous agents, adjust-ment of equipment design, and development of rapid detectionmethods ( FDA, 2006 ).

4.1.2. Other approachesMany recent developments may impact on the agricultural food

production and trading system, including the trend towards glob-alization, climate change, and increased agricultural productivity.In the face of this complexity of the food supply, Fraser et al.(2005) proposed to assess adaptability of food-consuming commu-nities rather than predicting food security in order to map outthose communities or segments of communities that are most vul-nerable to threats.

The model that Fraser and co-workers propose is based onexperience with a similar model, the Panarchy framework, whichis used for landscape ecology. This model discerns three importantfactors that dene a communitys vulnerability towards threats,that is, the wealth of a system, its diversity, and connectivity.

Wealth correlates with biomass in landscape ecology and is alsoconsidered to correlate with loss of diversity accumulated in theinitial phases of biomass accumulation. Examples of dening thewealth for food supply purposes can be found in development eco-nomics and food security studies. On one hand, wealth of the food-consuming community can be regarded to contribute to vulnera-bility because the damage that can be inicted to a system willbe bigger, that is, the capital losses will be higher. On the otherhand, from a social perspective, wealth also implies that resourcesare available that can be invested in order to avert risks. Fraser andco-workers propose to apply entitlements to food as known fromfood security studies. Entitlements may be direct, by self-produc-tion of food, or indirect, by purchase of food from the market. Inaddition, transfer entitlements pertain to food provided as aid or

welfare. A higher number of entitlements correspond to increasedwealth and less vulnerability.



5/18

Connectivity is supposed to contribute to the rapid spread of ahazard. In ecological modeling, the connectivity can be assessedby considering the pathways that can be taken. For environmentalchemicals, this has taken the shape of fugacity representing thepartition and transport of the chemical between different environ-mental compartments, such as the atmosphere. In the food system,this can be translated to the dislocation of hazards through trans-

port media. As an example, these authors mention the contamina-tion of milk with dioxins produced during waste incinerationthrough uptake by cows in the vicinity of the waste incinerator,and subsequent collecting, processing, and sales of the cows milkin urban centers.

Whilstdiversity is considered to contribute to theabilityto with-standshocks in an ecological system, Fraser and co-workers proposeto use a model that nancial investors use to spread risks of invest-ments, theModern PortfolioTheory. In this theory, it is realized thatrisks exists, but the components of a portfolio containing multipleinvestments areso diverse that they arenot equally affected by eco-nomic developments. Although systemic risks that pertain to awhole system, such as a drop of the share market, cannot be elimi-nated, the impact of unsystemic risks, which do not affect all com-ponents, can be reduced. The diversity in a portfolio can beassessed by considering the covariance between components re-sponse to changes, with a high covariance representing similarbehavior and therefore little diversity. For example, the yield of wheat elds in a large monoculture are may show high covariance.A low covariance therefore would signify reduced risk.

Fraser and co-workers conclude that modern agriculture iswealthy, non-diverse, and narrowly connected. Therefore, policiesshould strive towards increasing diversity without causing pricesto rise much. Urban agriculture might help increasing diversityin this respect (e.g . as a complement to rural agriculture, see Mou-geot, 2000 , and references therein). In addition, instead of takingmeasures on a national scale, trade links should be maintained if they are not connected to hazards occurring in the food system,such as in geographically separated areas ( Fraser et al., 2005 ). Sim-ilar to the models used by the US authorities, the model proposedby Fraser and co-workers is not predictive of specic hazards butaims to facilitate the identication of areas of the food productionand trading system that warrant closer examination.

4.2. Situational awareness

Another track followed by the US authorities within their fooddefense activities is the development of situational awareness.This awareness is a combination of measurements of objective val-ues representing the situation and the interpretation of the out-comes of these measurements, as well as the conclusion withregards to the future implications.

Lindberg et al. (2005) proposed to translate this concept of sit-

uational awareness to the domain of food production. Informationabout the situation can be collected in real-time within and outsidethe food production chain. Within food production chains, this in-cludes data generated by implementation of quality managementsystems, such as the HACCP approach and line inspection manage-ment. Such data also can include the tracing of movements of ingredients and food products throughout the chain, from farmto table, as well as waste streams inside food-producing facilities.

Developments in analytical technologies, such as cost-efcientdetection devices at the nano-scale for detection of hazardouschemicals and microorganisms, can further add to the real-timedata collected for this purpose. Modern radio-frequency identica-tion (RFID) devices that are invisible to the eye and added to con-signments of food products, which is currently already being

demanded by a big retail company, could be helpful in this respect,allowing for detection of movements of goods at the door gate.

Data from outside the food chain can include public health reportsand warnings against tampering with foods ( Lindberg et al., 2005 ).

The data collected should then be processed and interpreted,based on insights into the attributes of the specic parameters thatare measured as well as insights into the relations between theseparameters. This can be done automatically using ontologies andinference machines, similar to the search machines used for Inter-

net queries. Therefore the combination of real-time data with theontologies would match this papers denition of an indicator.Food defense is part of the situational awareness activities of

the US Homeland Security Operations Center, which serves as na-tional point for sharing of information and also as national coordi-nation center for incident management. This center also hosts theinter-agency National Bio-Surveillance Integration Center, whichaims to assist decision makers by providing data on naturallyoccurring and articial threats to agriculture and food, among oth-ers. In addition, the US Department of Health and Human Serviceshas modernized its BioSense disease surveillance program throughwhich hospitals can report ndings in real-time, which contributesto situational awareness.

4.3. Good practices and HACCP

As mentioned above, risk assessments are part of the HACCP ap-proach and have to be carried out after the food production processin a given link of the food production chain, such as the factory, hasbeen mapped. Following the risk assessment, critical points have tobe dened at which hazards will be analyzed, of which the resultswill trigger risk-mitigating actions, if necessary. The progress of theHACCP has to be reviewed periodically, and the outcomes of themeasurements and actions undertaken will serve as basis for mod-ications and adjustments of the system.

Brown and McClure (2006) note that emerging pathogens maynot be included in the microbiological risk assessments for goodpracticesand HACCP-basedfoodproduction systemsin an industrialsetting. Yet any known potentialpathogen that mayoccur shouldbeconsidered, alsoif the information is fraught withuncertainties. Riskassessments of emerging pathogens should include the four stagesdened above (Section 2). Uncertainties and variability should beconsidered, and in certain cases, data on other hazards similar tothe one under consideration may be useful as well.

The possible effects that any changes in the food chain have onthe occurrence of risks should be considered, for example in case of changing nature either of raw materials or of their geographic ori-gins. The same also applies to changes in processing and storageconditions. Also the effects of processing per se on the pathogenof interest should be considered. If processing eliminates or re-duces the hazards associated with an identied pathogen, the riskassessment should consider the possible effects of deviations fromthese processing conditions. These authors also recommend that in

case of emergence of pathogens, the effectiveness of measuresagainst this pathogen during food production should be assessedas soon as possible, if need be on the basis of assumptions aboutthe characteristics of the pathogen ( Brown and McClure, 2006 ).

Besides thechanges in thefood chain itself, theauthors also real-ize that other changes may lead to emergence of food pathogens,such as changes in the characteristics and prevalence of the patho-gen itself, in the particular human population consuming or in con-tact with food, and in the environment ( Brown and McClure, 2006 ).

During hazard identication, indicator microbes that are mea-sured to indicate the presence of other microbes should be repre-sentative for the latter, reecting the same distributions andsensitivities to processing conditions. Indicator organisms are lessavailable for viruses and protozoa than for bacteria. Brown and

McClure nevertheless note that there is a need for direct monitor-ing of pathogens.



6/18

As regards the characterization of the hazard, the data availablefor emerging pathogens will frequently be incomplete. As a basisfor estimates on doseresponse relationships, surveillance andclinical data can be used in combination with data on the intakelevels of the implicated foods ( Brown and McClure, 2006 ).

The exposure assessment of an emerging pathogen should con-sider the chain and route of exposure as well as the course of a con-

tamination. The effect of differences in various factors within andoutside the food chain, such as agricultural practices, on the expo-sure should be considered ( Brown and McClure, 2006 ).

The risk characterization, which brings together all the informa-tion from the hazard identication, hazard characterization andexposure assessment, should be able to indicate uncertaintiesand knowledge gaps. Modeling studies can be done based on anal-ogies with other microorganisms that can act as surrogate, includ-ing non-harmful species or pathogens with a long record of knowledge. These modeling studies enable the timely delivery torisk managers of an initial quantitative risk assessment that cannotawait longer-lasting data collection with the specic emergingpathogen. These authors also stress that a comparatively largeamount of information is available on bacterial pathogens thanon other groups of pathogens, such as insight into the mechanismsof horizontal transfer of virulence factors. In addition, they notethat the high share of viruses in the contingent of emerging patho-gens is associated with their ability to mutate relatively rapidly(Brown and McClure, 2006 ).

4.4. Prediction of the occurrence of mycotoxin-related hazards

Mould growth on crops, particularly cereals, has received a lotof interest from the food safety viewpoint because some plant-pathogenic moulds are known to produce compounds namedmycotoxins that are also harmful to the human and animal con-sumers of these crops. Various factors favor mould growth, whichcan differ among the mould species involved, such as the optimaltemperature and humidity. Several predictive models have beendeveloped that aim to predict the growth of moulds on crops andthe formation of mycotoxins by these moulds, particularly on theplant parts that have to be harvested. Such models may help mak-ing decisions on whether fungicides should be used on the crop ornot, or identifying possible contaminations further downstream.Although the current review focuses on generic indicators, it is feltthat these predictive models for mycotoxin formation provide aninteresting example of how environmental data, in this case mete-orological, can give indirect indications of a food safety hazard.

Prandini et al. (2009) provide an overview of current initiativesin the eld of predictive. models for the growth of Fusarium mouldson wheat, causing a plant disease known as Fusarium Head Blight(FHB). These models use meteorological variables, such as tem-perature, precipitation, and humidity. In addition, several models

also predict formation of mycotoxins, such as deoxynivalenol(DON) and zearalenone (ZEA; Prandini et al., 2009 ).

A model used in Italy considers mould behavior and its invasionof wheat, based on temperature, rainfall, crop water content,mould species, and growth stage of the crop. Based on the risk of FHB thus predicted, the risk of production of DON and ZEA can alsobe predicted. In other nations, including US, Canada, Argentina, andBelgium, other predictive models based on meteorological data areused for FHB and mycotoxin in wheat ( Prandini et al., 2009 ).

4.5. Factors inuencing the occurrence of chemical and biochemicalhazards in food and feed

Recent developments have occurred involving the presence of

chemical and biochemical hazards in food besides the mycotoxinsalready mentioned above. A recent study highlights various devel-

opments including toxic contaminations of botanical products,non-authorized antibiotic residues in imported shrimps, naturaltoxins present in pesticides of natural origin, and non-authorizedgenetically modied (GM) crops in crop commodities. The aim of this study has been to formulate common indicators for the haz-ards considered in these cases that can help identifying similarhazards in an early stage of their development, and thereby achiev-

ing the prevention of incidents ( Kleter et al., 2009a ).The case study on non-authorized GM crops includes (i) thepast incidents in the US surrounding Starlink TM maize, which hasbeen authorized for feed use only but nevertheless has been foundto occur in food; and (ii) experimental pharmaceutical crops, forwhich the containment during eld trials has been insufcient,raising the possibility of potential commingling with conventionalcrop harvests. In both cases, regulations have already been inplace but their enforcement could not prevent commingling fromoccurring. The incidents have led to stricter regulations and segre-gation measures that may help preventing such events fromrecurring ( Kleter et al, 2009a ). Recent incidents with rice contain-ing traces of experimental GM varieties that have previously beeneld-tested indicate that such contaminations cannot be com-pletely ruled out yet.

The case on botanicals considers Chinese Herb Nephropathyoccurring in patients on an herbal slimming preparation containingtoxic herbs that have been misidentied as the common herbingredient. In addition, the case of intoxications following the con-sumption of star anise tea has been studied, in which also a misi-dentication of a toxic relative of star anise has occurred.General risks associated with herbal medicines include (i) misi-dentication, which can be prevented by unequivocal designations,such as in Latin, and which may arise from weeds contaminatingcultivated herbs or noxious plants resembling wild-growing herbs;(ii) contamination with harmful contaminants; (iii) interactionwith other medications; (iv) adulteration; and (v) presence of herb-intrinsic natural toxins ( Kleter et al., 2009a ).

The case on pesticides of natural origin considered the reportedproperty of the natural pesticide rotenone that caused Parkinsonsdisease-like symptoms in experimental animals at high doses, andthe possible occurrence of toxic components in neem tree extractdue to novel processing procedures. The potentially new hazardsidentied in this case included new extraction methods for theproduction of natural pesticides, as well as the ability of biologicalpesticides containing living microorganisms to multiply after theirapplication ( Kleter et al., 2009a ).

The case on antibiotic residues in imported shrimp pertains toseveral incidents of traces of chloramphenicol and nitrofuransoccurring in shrimp produced in Asian countries. Because Codexalimentarius has been unable to dene maximum residue limits(MRLs) for these antibiotics based on inconclusive toxicology data,the EU follows a zero tolerance policy against these antibiotics.

Due to intensication of aquaculture of shrimp in third countries,disease pressure may also rise, in response to which producersmay be prone to use antibiotics, depending also on the availabilityof antibiotics, the local regulatory conditions, and education andawareness among producers. Factors have been identied thathave contributed to the incidents, including the lack of interna-tional harmonization and quality assurance; zero tolerance policytowards certain residues with concurrent improvement in analyt-ical sensitivity of their detection; illegal use or easy availabilityof antibiotics; and increased production and trade of aquacultureproducts, indicative of intensication that may possibly lead to in-creased disease pressure conducive to antibiotic use ( Kleter et al.,2009a ).

Based upon these and various other cases considered by these

authors, general recommendations of potential relevance to occur-rence of similar hazards in future have been formulated. First, it is



7/18

recommended to develop databases with data on contaminantsthat potentially occur in food and feed products, particularly con-taminants or products for which importing and exporting nationshave different production methods and regulations. Secondly,pro-active searches should be conducted for the potential presenceof contaminants, among others with the aid of the aforementioneddatabase. Thirdly, in international trade, incidents may be pre-

vented by imposing rigorous requirements for quality systems,regulations, and controls by the exporting nation ( Kleter et al.,2009a ).

4.6. Trend analysis based on data from food law enforcement activities

National and regional authorities have regulations and law-enforcing food control activities in place in order to detect hazardsin food and to ensure that the food supply is safe. Besides the con-trols that food inspection agencies have to carry out on a statutorybasis, also monitoring activities for certain hazards in certain prod-ucts can take place on a selected number of samples in order togain insight in the situation surrounding these specic hazards.The data generated by these activities, such as published in annualreports of some authorities, could provide inputs for trend analysisfor the identication of potentially emerging hazards. However, itis obvious that the hazards measured by the food inspectors musthave previously been identied and investigated. Previously un-known emerging hazard and known or re-merging hazards thatare for some reason not considered at the moment may thereforebe overlooked.

An open source of data coming from food inspection activities inthe EU is the RASFF system, which contains notications of hazardsin food and feed that are led by RASFF members, including na-tional food safety authorities of the EU and the European FreeTrade Association (EFTA), as well as the European Commission.The RASFF publishes weekly overviews of the notications re-ceived from members, as well as annual reviews highlighting con-spicuous trends, on its website. The notications fall into the twocategories of (i) alert notications pertaining to hazards that areof relevance to other members, possibly requiring also additionalmeasures from their side, and (ii) information notications per-taining to hazards that can be contained, for the information of other members only.

An analysis of the hazards notied through RASFF during thefour-year period from July 2003 till June 2007, including 11,403notications, has been carried out by Kleter et al. (2009b) . Thenotications were categorized according to date, product, hazard,specic agent, nation and region of origin, and nation ling thenotication. Hazard categories included microbiological, chemical,biological, mycotoxins, physical, quality, nutrition, fraud, labeling,hygiene, defective packaging, and transport. The developments inthe numbers of notications over time have been considered with-

in these categories in order to verify if trends in the occurrence of hazards can be observed.

A number of conspicuous trends are noted, including an in-crease in the notications pertaining to migration of substancesfrom utensils that may come into contact with food and to fraud-related issues. Also several temporary highs are observed, such asin the occurrence of food contact substances, including the inkcomponent isopropyl-thioxanthone (ITX); the illegal dye ParaRed in spices and condiments, and Sudan dyes in red-colored palmoil; unauthorized GM crops in imported rice and maize products;the unauthorized insecticide isophenfos-methyl in bell peppers;and the sh parasite Anisakis in sh products. For most of thesecases, notications have been followed up by measures taken bythe national authorities and conspicuous events with a given year

have been noted also by RASFF in its annual reports. The authorsconclude that it could be useful to combine the RASFF data with

complementary data, including safety assessments of the particu-lar hazard, risk management measures (e.g. food control inspec-tions) that have been taken before and after the notication, andbackground information on the particular hazard ( Kleter et al.,2009b ).

5. Epidemiology and surveillance supporting early response to

emerging food safety risks

If the occurrence of an emerging hazard has led to food poison-ings and other adverse health effects due to consumption of foodsor feed containing the hazard, it may be more appropriate to speakof emerging risks, with a less hypothetical connotation thanemerging hazards. Early detection of emerging risks in this stagewill still be benecial, such as to prevent diseases from spreadingwidely. International organization, national and regional authori-ties, and the food sector have put various mechanisms in place inorder to carry out monitoring and surveillance of adverse eventsfollowing consumption. Progress has been made in this area, par-ticularly in a number of elds, of which some examples will be fur-ther highlighted below. It should be noted that monitoring of animal disease may be considered an approach applying to farmto fork as well because a disease in food-producing animals mayconstitute a hazard in its own right to health of consumers of edi-ble products produced from the affected animals.

5.1. Zoonotic disease

As previously mentioned above in the summary of the EMRISKproject, the World Animal Health Organization OIE (French: OfceInternational des Epizooties) has set up an electronic noticationsystem for animal diseases. The veterinary authorities of memberstates have to notify the OIE of the occurrence of animal diseasesfrom a list of notiable diseases in the frame of the terrestrial ani-mal health code and the aquatic animal health code. This list also

contains various diseases that are of relevance for human foodsafety, such as bovine spongiform encephalopathy (BSE) in cattle.In addition, emerging diseases that have not been listed, but thatare expected to have major impact on livestock health, should bereported to OIE.

Currently, the OIE, in cooperation with FAO and WHO, is prepar-ing the global framework for the progressive control of transboun-dary animal diseases (GFTAD). The impetus for this initiative hasbeen the strive for a coordinated approach towards interventionsfollowing the recent experiences with the international avian inu-enza epidemics. It is realized that the majority of animal diseaseepidemics take place in developing countries and that appropriateresources and measures can thwart these epidemics before spread-ing to other areas. The GFTAD is to facilitate exchange of informa-

tion between the three organizations (FAO, WHO, OIE) andnational and regional authorities. This will enable early warningsand additional measures to mitigate the threat at its source, suchas the use of appropriate vaccines and healthcare ( Lubroth, 2006 ).

Also national and regional trend analyses of reports of zoonoticdisease incidences can provide useful information on the possibleemergence of such diseases. For example, EFSA publishes regularoverviews of the developments in zoonotic diseases within theEU as one of the European Communitys regulatory tasks as denedunder directive 2003/99/EC. This directive stipulates that the dataon zoonoses collected by national authorities of the EU should bebrought together as part of a coordinated monitoring effort. Thedata to be collected include the prevalence of certain zoonoticpathogens in animals; antimicrobial resistance in some pathogens;

frequencies and details of food-borne outbreaks; and human dis-ease data. Additional human data, including data from Enter-net



8/18


9/18

The WHO Surveillance Program for the Control of Food-borneInfections and Intoxications in Europe involves national food-borne disease outbreak surveillance activities of European andCentral Asian countries. Member states report to this programusing a standardized format into which data have to be added con-cerning, among others, the number of patients, the cause, the foodinvolved, location of food handling, location of purchase, and any

additional factors that contributed to the outbreak occurring.These data facilitate the conduct of outbreak trend analyses, theepidemiology of food-borne diseases, and the identication of sources of outbreaks. The outcomes show, among others, that theimportance of factors contributing to the occurrence of outbreakscan vary from nation to nation. These factors include contamina-tion of food through the use of infected equipment, inadequatefood storage and processing, and contamination by an infected per-son ( OBrien and Fisher, 2006 ).

It is realized though that the methods used by laboratories foridentication of pathogens can vary from one nation to the other.Global Salm-Surv is another initiative of WHO in which most of theEuropean network countries participate. It aims at harmonizingsurveillance data on serotypes and antibiotic resistance of Salmo-nella through training and laboratory quality assurance.

Enter-net, an initiative sponsored by the European Commission,brings together the microbiological and epidemiological servicesfrom its members nations and also collaborates with various inter-national partners, including the CDC mentioned above. The partic-ipants identify the types of Salmonella and STEC bacteria that areinvolved with outbreaks, as well as the antibiotic resistance of Sal-monella isolates. The data thus collected enable the conduct of atrend analysis of Salmonella and STEC outbreaks, while these dataserve as supplement rather than a substitute to national activities.The international collaboration allows for the accumulation of suf-cient data that member nations may not be able to do on theirown, so that an outbreak can be recognized earlier ( OBrien andFisher, 2006 ).

Additional research projects sponsored by the European Com-mission include Food-Borne Viruses in Europe (FBVE) and Diagno-sis, Viability, Networking, and Epidemiology (DIVINENET). Theseprojects focus on new virus detection methods, data exchange, out-break research, and mechanisms of emergence of new viral dis-eases. The Salm-gene project aims at strengthening thesurveillance of Salmonella by standardizing molecular identica-tion methods such as PFGE in Europe ( OBrien and Fisher, 2006 ).

Other European initiatives include the Basic Surveillance Net-work, which has its origin in EU regulations requiring that a Euro-pean-community-wide surveillance network be set up for 40diseases. The network aims to develop standards for minimally re-quired reporting data, which can be collected from national re-ports, so that all member states data can be harmonized. Variousof the 40 diseases mentioned above are food-borne infectious dis-

eases, such as caused by Salmonella , Campylobacter , and Listeria ,among others. For the epidemiological surveillance of variant Cre-utzfeldt-Jakob (vCJD) disease, EUROCJD/NEUROCJD focus on thediagnostics and epidemiology of vCJD. These networks include par-ticipants from all EU member states, as well as some associatednon-EU nations. The coordination of most of the EU initiativesmentioned has been conveyed to the recently established Euro-pean Centers for Disease Control ( OBrien and Fisher, 2006 ).

5.2.2. Retrospective studies on trendsAs noted above for the various initiatives on surveillance, be-

sides rapid identication of the source of an outbreak, surveillancedata may also serve the analysis of trends over longer periods of time in order to reveal possible increases in existent hazards or

appearance of new hazards. This particularly pertains to the sur-veillance of microbiological pathogens, which may reect the

dynamics of such pathogens and the evolutionary changes theythus may undergo. Various studies on the changing incidences of food hazards have been carried out. Such studies may provide in-sight into the feasibility of discerning trends that are of importancefor the study of emerging food-borne diseases.

Gillespie et al. (2006) have carried out a retrospective study onthe incidence of listeriosis, i.e. a disease caused by food-borne Lis-

teria bacteria, in Wales and England in the period of 2001 till 2004.Before this period, in the 90s, listeriosis has occurred at a compar-atively constant level, whereas an increase is observed from 2001till 2004. Possible links of listeriosis with ethnicity, age, socio-eco-nomic status, and residential home versus care center have beenexplored. The authors have not included clusters of outbreaks withmultiple disease reports related to a single common cause, such asconsumption of contaminated sandwiches. Therefore, only spo-radic cases have been considered. The results show that the riskof listeriosis has increased in patients aged over 60 years in the20012004 period compared to that of 19902000, or comparedto younger generations. This increase appears to be real and it isconsidered likely that it is caused by changes in behavior, althoughno change in intake of the major food groups have occurred.

Another study of relevance to emerging hazard identicationconsidered the detectability of cryptosporidiosis, which is the dis-ease caused by the protozoan pathogen Cryptosporidium . A partic-ular property of Cryptosporidium is its long incubation period. Thestudy particularly addressed the question as to whether it couldhave been detected through reporting of nurses reporting diar-rhoea in patients to the UK National Health Service (NHS) DirectCall system. This system has monitoring algorithms in place in or-der to follow particular diseases, such as inuenza. Data from anoutbreak of Cryptosporidium as well as numbers of total calls com-ing in at an NHS center were used to create a simulation. In thissimulation, it was checked what the likelihood was of observingan increase over the background level caused by Cryptosporidium.The data exceeding the threshold level imposed by the detectionalgorithm were considerably more numerous if 90% of the caseshad been reported instead of the 10% initially considered ( Cooperet al., 2006 ). This study also shows that discerning an emergingdisease may be less demanding if the reporting rate is high.

6. Identication of hazards in the surrounding environment

6.1. EMRISK: Holistic-approach-based indicators

6.1.1. Background of the EMRISK project The EMRISK project provides an interesting perspective on the

collection and use of indicators from a holistic perspective. ThisEuropean project was funded by EFSA and was carried out underthe supervision of the Dutch national Food and Consumer Product

Safety Authority (VWA). EMRISK participants included EU memberstate authorities and institutes involved with food safety, as well asinternational organizations. EMRISK extended the work that hadpreviously been done under VWAs supervision by the PeriApt project, which had been supported by the European Commissions re-search funding as a specic supportive action in the frame of theERANET program for linking national research activities with theaim of establishing a research area at the European level. Boththe EMRISK report & annexes, and the opinion of EFSAs ScienticCommittee based on the outcomes of the report, have been postedon the EFSA website ( EFSA, 2006a; VWA, 2006 ).

Within PeriApt, a holistic model has been developed, whichtakes into consideration that hazards within the food productionchain may originate from or be inuenced by events and devel-

opments that take place within inuential sectors outside the foodsector. This model of a host environment analysis builds upon



10/18

previous outcomes of broadly focused OECD activities on futurerisks. OECD has discerned four inuential sectors, within whichcritical factors can be further identied and the indicators linkedto these critical factors. PeriApt has expanded this range to eightinuential sectors based upon the outcomes of three case studiesand consultations with stakeholders. These eight sectors includeagriculture, technology & industry, government & policy, economy,

nature & environment, information, consumer behavior, and cul-ture & demography. Two additional retrospective case studies havealso considered the inuence of human behavior aspects in thedevelopment of past food safety incidents, mad cows disease(BSE) and the formation of the potentially carcinogenic compoundacrylamide in heated products. The PeriApt project recommendsreiterating the exercise of identication of inuential sectors withstakeholders, as well as building European and national networksto harmonize indicators and to collaborate with international orga-nizations like FAO, WHO, and OECD on issues such as surveillance(VWA, 2005 ).

EMRISK has extended upon PeriApt, providing recommenda-tions and a blueprint to EFSA on how to build a pre-early warn-ing system for the timely, anticipatory, and pro-active detection of emerging food safety hazards in an early phase of their develop-ment. Therefore this system is different from the monitoring andsurveillance systems that are currently in place to detect knownhazards.

6.1.2. EMRISK activities and denitionsEMRISK has carried our various activities, including:

(i) a bottom-up review of ve retrospective case studies,leading from specic cases to identication of indicatorsand information sources, was intended to provide insightas to how a pro-active system could have used indicatorsand information sources to prevent these incidents, coveringthe (a) discovery of formation of acrylamide in heated prod-ucts, (b) the avian inuenza epidemic in poultry livestock inHolland in 2003, (c) the BSE incident in the United Kingdom,(d) poisoning with aatoxins in Africa, and (e) the discoveryof semicarbazide as a food contact substance derived fromblowing agents used for the production of jar lids;

(ii) an inventory of activities within EU-sponsored researchprojects pertaining to emerging hazard identication; and

(iii) top-down identication of information sources and theirquality and accessibility, starting from general conclusionsvia generic indicators towards these sources, by stakehold-ers with broad expertise invited to workshops specicallyorganized for this purpose by EMRISK.

In the conclusions to the case studies, it is mentioned that foreach case a combination of factors has been important in the devel-

opment of the case. The use and weighing of multiple indicatorscould therefore provide a useful tool (see the trafc light approachbelow). It is also realized that indicators may not always be quan-titative, and that qualitative indicators should be regarded as indic-ative. Both case-specic and more generic indicators have beenidentied in the case studies, of which the generic indicators havebeen further explored for their usability in the approach to be rec-ommended by EMRISK ( VWA, 2006 ).

EMRISK proposes some revisions to the set of inuential sectorspreviously dened by PeriApt in order to align these better withthe denitions used by other EU and international institutions. EM-RISK also distinguishes between primary and secondary inuentialsectors, with the latter positioned further away from the food sec-tor than the rst. The revised set of primary inuential sectors con-

sists of science & technology, environment & energy, health,agriculture, economy & nance, and industry & trade. The three

secondary inuential sectors thus include government & politics,population & social conditions, and information & communication.

EMRISK also makes a distinction between primary and second-ary information sources that provide information for identicationof hazards. Primary systems are directly linked to food production,such as the monitoring and surveillance systems that have evolvedover time, based on experiences with known hazards. In addition,

more responsibility is currently put on the producers themselves,having to fulll requirements of good and hygienic productionand to perform their own risk assessments as part of the HACCP.National regulatory and food control agencies subsequently verifyif the right checks are in place and carried out according to the reg-ulations. An example of a primary system is an early-warning data-base such as the RASFF system where members report detectedhazards in food and animal feed that are of relevance to othermembers. These members include the European Commission, EUmember state authorities, and the EFTA Surveillance Agency(ESA). Other examples of primary information systems are expertnetworks, stakeholder groupings, scientic panels, public medianetworks, consumer help services, and scientic conferences.These systems bear several limitations including the focus onknown hazards, the non-consideration of factors outside the foodsector, and economic and societal factors that may inuence theresponse towards a certain hazard. EMRISK recognizes the impor-tance of primary systems in providing information across nations,although these data may be focused on one sector only.

Secondary systems, on the other hand, are more broadly fo-cused and not only consider the food sector but also the host envi-ronment of inuential sectors surrounding it. Indicators used bythese systems to predict the likelihood of the emergence of a haz-ard are considered as indirect and to be located within theseinuential sectors.

EMRISK also describes anticipatory systems, which would ex-tend upon the secondary systems by linking all the different kindsof information. It is noted by EMRISK that anticipatory systems arestill within their initial phases of formation. Critical factors as-signed within an anticipatory system may be strongly linked witheach other in a kind of network structure. Assigning importancevalues to the indicators linked with these critical factors may benecessary, based, for example, upon the importance of detectinga signal for the emergence of a hazard ( VWA, 2006 ).

6.1.3. EMRISK project outcomesBased upon the outcomes of the various activities, including the

case studies and workshops, EMRISK has identied of a set of indi-cators within the nine inuential sectors ( VWA, 2006 ). These indi-cators are both quantitative and qualitative by nature and havebeen rened and sized down, based upon a ranking system usingdened criteria. These criteria include (i) measurability, which isthe ability to detect a signal based upon thresholds or ranges for

quantitative indicators or upon afrmative or negative outcomesfor qualitative indicators; (ii) interpretability, which is the abilityto observe trends in the outputs from the indicators; (iii) direct-ness, which is inversely related to the distance between indicatorand emerging hazard, with a high directness signifying a short dis-tance; (iv) potential impact, which relates to the potential harmthat a particular emerging hazard associated with the indicatormay cause to health, as well its degree of dissemination throughsociety; (v) comprehensiveness, with generic indicators being pre-ferred over specic indicators; and (vi) discriminatory nature,which is indicative of the lack of overlap or duplication with otherindicators and which therefore can only be determined in a laterphase of the selection procedure.

A team consisting of ve experts from three organizations par-

ticipating in EMRISK has subsequently used these criteria in orderto make a selection of the indicators collected. Two paths have



11/18

been followed, the rst being a decision-tree approach in whichscores on a scale from 1 (very low) to 5 (very high) were assignedto an indicators compliance with each specic criterion. Each scorefor an indicator has subsequently been transformed into a yes forscores 35 and a no for scores 12. For an indicator to be se-lected, all transformed scores should be yes.

Whilst each criterion has been equally weighed within this

decision-tree approach, the second approach of a ranking systemhas applied unequal weights to the criteria. Furthermore, the rank-ing system has not considered the criteria interpretability andmeasurability at the same time, since they are mutually depen-dent. Therefore, the ranking system has only considered one of both criteria, that is, interpretability. In addition, the criterion dis-criminatory has not been considered either, given the difcultiesin assigning scores for this criterion to indicators. A weight factorof 2 has been assigned to directness and potential impact,whereas a weight factor of 1 has been assigned to the other criteriaof interpretability and comprehensiveness. Therefore the max-imum score that can be assigned to an indicator is 30. The out-comes of the ranking system approach are more diverse thanthose of the decision-tree approach, allowing for the grouping of top-scoring indicators.

Based upon the outcomes of the comparison between the deci-sion-tree and ranking system approaches, EMRISK concludes that aranking system approach employing weighted criteria is preferredfor the selection of indicators of emerging hazards in inuentialsectors. In addition, because of the diversity of the inuential sec-tors that has been considered, EMRISK recommends consulting asufcient number of experts with knowledge of each relevantinuential sector. EMRISK also envisions another role for these ex-perts in further steps downstream of the identication of indica-tors (see below).

Another objective of EMRISK has been to allocate indicatorswithin the inuential sectors it has dened so that each sectorwould be covered well. This objective apparently has beenachieved given that the nal output amounted to a total of 217indicators of which 63% are located within the six primary inuen-tial sectors, in which the number of indicators per sector range be-tween 10 for science & technology and 36 for agriculture &sheries. In annex 5 to the report, which features the list with se-lected indicators, the table provides indicators per class withineach inuential sector. For each indicator, it is indicated what themain features are and which information sources are availablefor data on the indicators. For example, within the inuential sec-tor science & technology, there are two indicators that fall withinthe class of knowledge-based services. These indicators are out-put (results) of risk assessment and number of conferences, forwhich the main features mentioned are assists to distinguish be-tween true and false emerging risks and reects the scienticattention/focus of selected issues; beware of the impact of the con-

ference science (e.g. one big one versus lots of little ones). As keyinformation sources, a variety of national and international scien-tic institutions, risk assessment bodies, expert networks andEurostat, the European Communities statistical ofce, arementioned.

The subsequent steps that EMRISK proposes are to furtherorganize the signals coming from the indicators. For this purpose,time-dependent changes in the relevance of indicators should bemeasured. This can be done by selecting the top-ve indicators ineach inuential sector based on the scores in the ranking system.It should then be checked during each repetition after a certainperiod of time if changes with regard to the ranking have takenplace, that is, if different indicators reach the top-ve with eachrepetition of the ranking procedure. In addition, the signals com-

ing from each indicator can be measured over time and classiedto check for possible changes in their magnitude. In order to deal

with both qualitative and quantitative indicators and to interpretthe meaning of any temporal changes observed, EMRISK proposesto apply the trafc light theory for the interpretation of trendsassociated with changes in signals. With this theory, the directionof trends is indicated by a color similar to trafc lights, withgreen for positive trends, yellow for intermediate results, andred for negative trends. It is also possible to assign different

urgency weighing factors to signals with different trafc lightcolors, reecting the need for immediate action if negative trendsoccur. In addition, one signal may be considered to be moreimportant than another and therefore an additional importanceweighing factor can be assigned to signals. In order to account forthe relationships that signals, including those from different sec-tors, may share with each other, a relationship weighing factorcan be assigned to these signals.

For the timely acquisition of data, EMRISK notes that prelimin-ary scientic data may not be readily available. In addition, alsonon-scientic data can provide useful inputs. Knowledge orga-nized within EU research networks is recognized as a possiblesource of information, where experts would be able to providesignals from selected information sources to which they have ac-cess. In addition, EMRISK provides an outline of how automatedsearches using dedicated search engines can be done for majorchanges in indicators over time and in the indicators space.To this end, indicative questions need being drafted by expertswith a description of the attributes of the indicators under con-sideration. Keywords can subsequently be extracted from thisindicative question by specialized computer programs in orderto enhance the effectiveness of a search on the Internet for dataon changes in the indicators. This process of automated searchesbased on indicative questions and keywords involving both ex-perts and computer specialists may need being reiterated and ad- justed several times in order to further improve the quality of theoutput.

The signals that come out of the system once it has becomeoperational have to be evaluated by scientic experts and informa-tion specialists before a pro-active alert is issued to inform riskmanagers of an emerging hazard. It is also envisioned that thissemi-automatic search system could be self-learning, buildingupon experiences with feedback from the risk management pro-cess following the yield of signals that require further actions bythe management.

Stakeholders, such as consumer organizations, may be involvedat various stages of the recommended procedure, for example asinformation source, for the processing of indicative questions,and after the evaluation of signals.

The EMRISK report also lists a number of interesting initia-tives, however noting that these systems are retrospective. Initia-tives that consider multiple sectors include the completed OECDproject on emerging risks in the 21st century and Eurostats mea-

surement of indicators in various sectors. In addition, the Euro-pean Environmental Agency and the European CommissionsHealth Monitoring Program are examples of initiatives collectinginformation on changes in indicators. Field-oriented, key-word-containing thesauri are compiled by both the European Commis-sion and FAO. Initiatives that provide information sources includetwo systems of the FAO, the Food Insecurity and VulnerabilityInformation and Mapping Systems (FIVIMS) and the Global Infor-mation and Early Warning System (GIEWS). FIVIMS consists of astructure of national and international databases on food insecu-rity. GIEWS monitors agricultural production and prices, amongothers by using satellite data, and also provides inputs to FIVIMS.An early warning system for animal diseases considering bothofcial and unofcial information has been set up by OIE, while

the Canadian authorities have established the Global PublicHealth Intelligence Network (GPHIN) which continually monitors



12/18

information from global public media, including Internet, for cer-tain relevant events. An example of an initiative involving expertnetworks is the ECDC bringing together European networks of sci-entists involved in various elds of relevance, such as researchnetworks on certain pathogenic bacteria ( VWA, 2006 ).

6.1.4. EFSA Scientic Committees opinion

Based upon the outcomes of the EMRISK project, the EFSA Sci-entic Committee, in which the chairs of the various scientic ex-pert panels as well as senior scientic staff of EFSA participate,published an opinion on the early identication of emerging risks(EFSA, 2006a ). The background for this opinion and the EMRISKproject was that the General Food Law, European CommunityRegulation 178/2002/EC, which provided the basis for the estab-lishment of EFSA, also stipulated that EFSA should install moni-toring procedures for the identication of emerging risks withinits eld of work, that is, food safety. EFSA had requested the Sci-entic Committee to give advice on systems and procedures forthe identication of emerging risks, as well as on networks thatcould provide information for this purpose. Interestingly, the Sci-entic Committee provides a denition of emerging risks as is-sues that may pose risks to humans, animals, and environmentin the future. The causes of their emergence are either a signi-cant exposure to hazard that have not been identied earlier ora new or increased exposure to a known hazard. Please note thatthe denition of risk is thus in line with the notion that this isprobability of hazard occurring through exposure. An indicatoris dened as component of risk assessment, which can be com-posed of various qualitative/quantitative parameters, either clo-sely or loosely linked with the food production chain. A signalis dened as a trend in time or space of indicator. Emerging risksmay be indicated by single or multiple signals.

The opinion lists both internal and external sources of informa-tion that could provide signals of emerging problems. Examples of internal sources are EFSA staff, the Scientic Committee, the Advi-sory Forum in which member state authorities representativestake part, the EFSA Scientic Panel experts, and the StakeholderConsultative Platform. External sources include, among others,the European Commissions Scientic Panels, such as the ScienticCommittee on Emerging and Newly Identied Health Risks, as wellas the RASFF system. Another source could be Eurostats Food:from farm to fork initiative, which brings together data on foodsafety monitoring, trade, and public health, with the aim of provid-ing risk managers with a tool. Also the REACH program (Registra-tion, Evaluation, and Authorization of Chemicals) could providedata on chemical substances that may be relevant for food safetyas well. Other contacts mentioned include international organiza-tions and other European institutions, such as for medicine regis-tration and the environment.

Besides listing a number of ongoing, related systems for early

identication of risks, the opinion also recommends a procedurefor the identication of issues and the processing thereof. Theopinion foresees the establishment of links between EFSA andother networks that may provide information on emerging risks,rather than a comprehensive network dedicated to this purpose.Besides the contacts mentioned above, these links also include re-search institutions, national agencies within and outside the EU,external experts, and information sources & search tools. In addi-tion, it is recommended to form an ad hoc working group underthe umbrella of the Stakeholder Consultative Platform with theaim of preparing for the formation of a stakeholder network dedi-cated to information exchange on emerging risks. In addition,transparent and clear procedures on how to process and evaluateinformation regarding emerging risks should be implemented. Fur-

thermore, a database with evaluations of incoming issues shouldbe set up.

With regard to indicators, the Scientic Committee concludesthat consulted experts should make an initial selection of a limitednumber of indicators and, if possible, values triggering follow-upactions. In addition, the extensive list with indicators provided inannex to the EMRISK report is considered to be resource-demand-ing and a possible target in the longer term. For the shorter term,the Scientic Committee recommends a system designated ER Re-

source and procedure based on expert judgment. Within this pro-cedure, information on emerging risks may be provided to EFSA,although not part of a dedicated screening activity. Such informa-tion has to pass an initial ltering by EFSA staff based on relevanceof the information, but also on its quality and possible duplicationof previous information. A quality index was envisioned that couldbe afxed to the information so as to reect, among others, the reli-ability of the source, the multiplicity (>1) of independent sources,and the way in which data had been produced. If selected for closerexamination, a secondary lter will be applied to the incominginformation by evaluating it with experts from EFSA Scientic Pan-els. If information does not pass either one of both lters, it will bestored in the above mentioned database. The selected informationthat has passed the ltering stages is then considered for the ur-gency required to handle it further. Whether urgent or not, theinformation can be assessed either by the existing scientic panels& Scientic Committee or by an ad hoc working group establishedfor this purpose. In case the information on the emerging risk givesrise to concerns, EFSA will send an alert to the relevant authorities.In case no concerns are raised, the outcomes will be recorded, rel-evant parties informed, and the information stored in the abovementioned database. The Scientic Committee acknowledged theimportance of rapid follow-up to signals received within this pro-cedure and also that of trust between the cooperating institutions.Research was regarded a key source of information.

In conclusion, the Scientic Committees opinion highlightedthe preference for the use of a limited set of indicators to be se-lected by experts, but did not specically mention examples of such indicators. In addition, it identies many European institu-tions as potential sources of valuable information on emergingrisks ( EFSA, 2006a ).

6.1.5. Dutch study on indicators for emerging mycotoxin hazardsFrom the indicators recommended by EMRISK, Park and Bos

(2007) have selected a limited number as candidates for the futuredetection of emerging mycotoxin hazards. Selected indicatorslinked with the critical factor climate include drought, temperaturechanges, and rainfall. This is based on the known relationship be-tween growth and mycotoxin production of some moulds, includ-ing Fusarium, on crops at high temperatures and humidity.

Another critical factor is market and consumer trends, giventhat sudden changes in demand may lead to the use of high-yield-ing varieties that are more sensitive towards mould infection or to

longer post-harvest storage periods in which mould growth can oc-cur. Indicators linked with market and consumer trends includemarket prices and crop production, such as reported by Eurostatand FAOSTAT, and trends observed in market research reports.

Similar to the market conditions, the economy is also consid-ered to be a critical factor, for example because the availability of nancial resources inuences agricultural practices and the qualityof storage and transport facilities. Indicators such as gross domes-tic product and ination rate are linked to this critical factor.

The critical factors global trade and transportation relate to theconditions of storage of the product, including the transit time,during transport. Indicators linked with global trade include im-port and export statistics, such as from FAOSTAT, as well as infor-mation on trade barriers. Transportation indicators include strikes

of transporters and registration of transportation companies. An-other critical factor is the introduction of new technologies,



13/18

including crop monocultures, which may facilitate mycotoxin pro-duction and dissemination. The indicators for this comprise scien-tic and non-scientic (newsletter) journal coverage.

Pest control management is considered another critical factorbecause damage to crop by pests may provide points of entry formoulds, besides the mould being a pest in its own right as well,It is realized, though, that measures against pests, such as by spray-

ing fungicides, may not always reduce, but rather stimulate myco-toxin production. Indicators include data on the prevalence of pests, as wel

indicadores de peligros emergentes

Documents