Food Science & Technology · Packaging, Safety, Quality & Regulation · also Risk Analysis, Microbial Risk Assessment, Quantitative Risk Assessment, Food Safety Risk Assessment
The Codex risk-analysis framework that underlies modern food standards: hazard identification and characterisation, exposure assessment, quantitative microbial and chemical risk assessment, food safety objectives and performance objectives, and risk communication.
In the exams: ICAR AICE 35.b Unit 2 (concept of risk analysis); PG and regulatory-science roles
Risk analysis is the structured way in which food safety authorities decide what is safe enough. It has three parts: risk assessment, the scientific estimate of how likely and how severe harm is under given conditions; risk management, the choice of measures to reduce the risk to an acceptable level; and risk communication, the exchange of information and views among everyone concerned. Codex Alimentarius, the World Trade Organization's SPS Agreement and modern national food laws all require food safety measures to rest on this process, which is why nearly every maximum level, microbiological criterion and process requirement in food law can be traced back to a risk assessment.
The subject matters because resources are always limited and zero risk is never achievable. A regulator must decide which hazards deserve attention first, how strict a limit should be, whether to require a process step or simply advise consumers, and how to explain the decision to the public when science is uncertain. A food company uses the same tools when it sets internal specifications, judges whether a process change is safe, or responds to a new scientific finding about one of its ingredients.
The subject is named in the ICAR AICE-JRF/SRF 35.b syllabus (concept of risk analysis) and underlies the work of regulatory scientists at FSSAI, EFSA, FDA, FSANZ and their counterparts. Examiners ask for the components of the framework and their order, definitions of hazard and risk, dose-response models, exposure calculations, the margin of exposure, the relationship between food safety objectives, performance objectives and performance criteria, and the principles of risk communication. The toxicology behind chemical guidance values is taught in food-toxicology-contaminants-allergens, microbiological criteria and sampling in food-microbiology, and the legal frameworks that apply risk analysis in food-law-regulatory-affairs.
A hazard is a biological, chemical or physical agent in food, or a condition of food, with the potential to cause an adverse health effect. Risk is a function of the probability of an adverse health effect and the severity of that effect, consequential to a hazard or hazards in food. The distinction is fundamental: a hazard is a property of an agent, risk depends on exposure. Aflatoxin is a hazard in every peanut; the risk depends on how much aflatoxin is present and how many peanuts are eaten.
Codex defines risk analysis as a process consisting of three components. Risk assessment is a scientifically based process of hazard identification, hazard characterisation, exposure assessment and risk characterisation. Risk management is the process, distinct from risk assessment, of weighing policy alternatives in consultation with all interested parties, considering risk assessment and other factors relevant for the health protection of consumers and for the promotion of fair trade practices, and, if needed, selecting appropriate prevention and control options. Risk communication is the interactive exchange of information and opinions throughout the risk analysis process concerning risk, risk-related factors and risk perceptions, among risk assessors, risk managers, consumers, industry, the academic community and other interested parties, including the explanation of risk assessment findings and the basis of risk management decisions.
The framework was set out by FAO/WHO expert consultations in the 1990s and adopted in the Codex Working Principles for Risk Analysis for Application in the Framework of the Codex Alimentarius, published in the Codex Procedural Manual, and in separate principles for governments. It is applied to food safety hazards; nutrition and quality issues increasingly use similar approaches.
Several principles run through the Codex texts. Risk analysis should be applied consistently, be open, transparent and documented, and be evaluated and reviewed as new information appears. There should be a functional separation between risk assessment and risk management, to ensure the scientific integrity of the assessment and avoid confusion over their roles, while recognising that interaction between assessors and managers is essential: managers frame the questions, and assessors must understand what decision the assessment will inform. In practice the separation can be institutional, as between EFSA and the European Commission, or functional within one agency.
Risk managers should set a risk assessment policy before the assessment begins: guidelines for the value judgements and policy choices that will be needed at decision points in the assessment, such as which populations to protect, which default assumptions to use when data are missing, and how to treat uncertainty. Risk assessment should be based on the best available science, should state clearly its assumptions, uncertainties and variability, and should be proportionate to the question. Risk management should follow a structured approach, take into account the risk assessment and, where appropriate, other legitimate factors such as feasibility, cost, consumer concerns and trade, and should be proportionate and non-discriminatory. Precaution is an inherent element: where scientific evidence is insufficient, measures may be provisional while more evidence is obtained. Risk communication should involve all interested parties from the start, not only after decisions are made.
Hazard identification is the identification of biological, chemical and physical agents capable of causing adverse health effects that may be present in a particular food or group of foods. For chemicals, it asks what effects a substance can cause, drawing on toxicological studies in animals and cells, human epidemiology, structure-activity relationships and mechanistic data; the classifications of the International Agency for Research on Cancer, which state whether an agent can cause cancer, are an example of hazard identification. For microorganisms, it draws on outbreak investigations, surveillance data, clinical studies and knowledge of the organism's ecology to establish that the organism causes disease and is associated with the food.
Hazard identification is qualitative. A finding that a substance is a hazard does not mean that it poses a risk at the levels found in food; that requires the later steps. Confusion between the two causes much public alarm, as when a hazard classification is reported as if it were a risk estimate. The output of hazard identification is a clear statement of the hazard, the food concerned and the health effects of interest, which defines the scope of the remaining assessment.
Hazard characterisation is the qualitative or quantitative evaluation of the nature of the adverse health effects associated with the hazard, including, where data are available, a dose-response assessment. For chemicals it typically identifies the critical effect, the adverse effect occurring at the lowest dose in the most sensitive relevant species, and a reference point such as the NOAEL or the BMDL, and then derives a health-based guidance value, such as an ADI, TDI or ARfD, by applying uncertainty factors. Where a substance is genotoxic and carcinogenic, no guidance value is set and the reference point is used later to calculate a margin of exposure. These steps are explained in food-toxicology-contaminants-allergens.
For microbiological hazards, characterisation describes the severity and duration of illness, the sequelae such as haemolytic uraemic syndrome or Guillain-Barré syndrome, the populations most at risk, and the dose-response relationship, the probability of infection or illness as a function of the number of organisms ingested. Factors relating to the pathogen (virulence and strain variation), the host (age, immunity, pregnancy, gastric acidity, medication) and the food matrix (fat content and buffering capacity, which can protect organisms during passage through the stomach) all modify the response; this is sometimes called the epidemiological triangle of pathogen, host and matrix. Severity can be summarised in a common measure such as disability-adjusted life years, discussed in chapter 9.
For chemicals with a threshold mechanism, the classical approach identifies the highest dose without an adverse effect (NOAEL) and divides it by uncertainty factors. The benchmark dose approach instead fits mathematical models, such as log-logistic, Weibull, gamma or multistage models, to the dose-response data, and estimates the dose that causes a specified benchmark response, for example a 10 per cent extra risk of a quantal effect (BMD10) or a 5 per cent change in a continuous measure. The lower one-sided 95 per cent confidence limit of the BMD, the BMDL, is used as the reference point. Model averaging across several models is now recommended by EFSA and others to avoid depending on one arbitrary model choice. The BMD approach uses all the data, gives a reference point with a stated level of effect and accounts for the quality of the study through the width of the confidence interval.
For genotoxic carcinogens, low-dose extrapolation of the dose-response curve to estimate risk at human exposure levels has been used, particularly in the United States, but the results depend heavily on the model chosen for the region far below the observed data. The margin of exposure approach avoids that extrapolation by comparing the reference point directly with exposure.
Microbial dose-response models usually assume that each ingested organism acts independently and has some probability r of surviving the host's defences and causing infection, and that a single surviving organism can start an infection. The simplest model is the exponential model: P = 1 − exp(−r × d), where P is the probability of infection or illness and d the mean ingested dose. If r = 0.001 and the dose is 100 organisms, P = 1 − exp(−0.1), about 0.095. At low doses P is approximately r × d, so the curve is linear at low doses and there is no threshold.
The beta-Poisson model allows r to vary between organisms and hosts according to a beta distribution, giving the approximation P = 1 − (1 + d/β)^(−α), where α and β are parameters fitted to data; it is often written in terms of the median infectious dose, N50, as P = 1 − [1 + d (2^(1/α) − 1)/N50]^(−α), which gives P = 0.5 when d = N50. The beta-Poisson curve is shallower than the exponential, reflecting heterogeneity. Models are fitted to human volunteer feeding studies, where these exist and are ethical, to outbreak data where doses can be estimated, and sometimes to animal data. Separate models may describe infection and illness given infection. Dose-response models are among the largest sources of uncertainty in microbial risk assessment, because data at the low doses typical of food exposure are rare. Highly infectious organisms such as norovirus, Shigella and Shiga toxin-producing E. coli have high probabilities of illness from small doses, while organisms such as Listeria monocytogenes have low r values for the general population but much higher ones for vulnerable groups.
Exposure assessment is the qualitative or quantitative evaluation of the likely intake of a biological, chemical or physical agent via food, and from other sources if relevant. For chemicals, dietary exposure is the sum over foods of concentration multiplied by consumption, divided by body weight: exposure (mg/kg bw/day) = Σ (Ci × Fi) / bw. For example, if bread contains 0.05 mg/kg of a contaminant and a 60 kg adult eats 0.2 kg of bread a day, exposure from bread is 0.05 × 0.2 / 60 = 0.00017 mg/kg bw/day.
Consumption data come from individual food consumption surveys, which record what individuals eat over one or more days, from household budget surveys, or from food balance sheets, which estimate national availability per person and overestimate actual intake. Global estimates use the WHO GEMS/Food cluster diets, which group countries with similar diets. A key distinction is between per capita consumption, averaged over the whole population including non-consumers, and consumers-only consumption, which is higher and more relevant for a food eaten by only part of the population. Concentration data come from monitoring and surveillance programmes, total diet studies in which foods are prepared as eaten and analysed, maximum permitted levels used as a conservative assumption, or, for additives, maximum use levels or industry use data. Results below the limit of detection must be handled by stated assumptions, such as setting them to zero (lower bound), to the limit (upper bound) or to half the limit.
Exposure assessment usually proceeds in tiers, starting with simple conservative methods and refining only if needed. Screening methods such as the budget method for additives assume that a fixed proportion of the diet contains the additive at its maximum level. For pesticide residues, the theoretical maximum daily intake (TMDI) multiplies each MRL by average consumption and sums; the international estimated daily intake (IEDI) uses median residues from supervised trials and processing factors; and for acute exposure, the international estimate of short-term intake (IESTI) uses high consumption on one day with high residues in a single unit. If a screening estimate is below the guidance value, no further work is needed; if not, refined estimates follow.
Deterministic assessments use point values, such as the mean or a high percentile of consumption with a mean or maximum concentration. Probabilistic assessments use distributions of consumption and concentration and combine them, typically by Monte Carlo simulation, to estimate the distribution of exposure across the population, giving, for example, the 95th percentile of exposure. Probabilistic methods describe variability between people more realistically, but require more data and expertise. Assessments consider vulnerable groups separately: infants and young children eat more per kilogram of body weight, pregnant women may be relevant because of fetal effects, and high consumers of particular foods may be far above the average. Aggregate exposure combines dietary with non-dietary routes, such as drinking water, air or consumer products, and cumulative exposure combines several substances with a common mode of action. For microbial hazards, exposure assessment estimates the probability that a serving is contaminated and the number of organisms it contains at the time of eating, which usually requires modelling of the food chain, as described in chapter 6.
Risk characterisation is the qualitative or quantitative estimation, including attendant uncertainties, of the probability of occurrence and severity of known or potential adverse health effects in a given population, based on hazard identification, hazard characterisation and exposure assessment. It is the step that answers the risk manager's question.
For chemicals with a health-based guidance value, risk characterisation compares estimated exposure with the guidance value, often expressed as a percentage: if exposure is 0.03 mg/kg bw/day and the ADI is 0.1 mg/kg bw/day, exposure is 30 per cent of the ADI and of low concern. Exposure above the guidance value for part of the population calls for attention, but does not mean harm will occur, because guidance values include safety margins and refer to long-term intake. For genotoxic carcinogens, the margin of exposure is calculated, as described in chapter 7. For microbial hazards, risk characterisation combines the exposure distribution with the dose-response model to estimate the probability of illness per serving and, multiplying by the number of servings eaten, the expected number of cases per year in a population. If the probability of illness per serving is 1 in 100,000 and a population eats 100 million servings a year, about 1,000 cases a year are expected. Results may be expressed as risk per serving, per person per year, or for the whole population.
Two different things make a risk estimate imprecise. Variability is real heterogeneity in the population or the system: people differ in consumption, body weight and susceptibility; servings differ in contamination; storage temperatures differ between homes. More data describe variability better but do not reduce it. Uncertainty is lack of knowledge about the true value of a quantity, from limited data, measurement error, model choice or extrapolation; more or better data can reduce it. Good assessments keep the two apart where possible, for example by second-order Monte Carlo simulation, so that managers can see both the spread of risk across the population and the confidence in the estimate.
Sensitivity analysis identifies which inputs most influence the result. It shows where better data would most reduce uncertainty and, in a food-chain model, which steps offer the greatest potential for control: if the risk is most sensitive to the storage temperature in consumers' refrigerators, consumer advice or shelf-life limits may be more effective than further changes at the processing plant. A risk characterisation should state its assumptions, the main uncertainties and their likely direction, and the confidence in its conclusions in plain language, so that risk managers can weigh them. EFSA and other agencies now publish formal guidance on describing uncertainty in scientific assessments.
Quantitative microbial risk assessment (QMRA) estimates the risk of illness from a pathogen in a food by following the pathogen through the food chain to the point of consumption and applying a dose-response model. Because microbial numbers change along the chain, through growth, inactivation, cross-contamination, mixing and partitioning of units, a QMRA usually builds a farm-to-fork or process-to-consumption model. The modular process risk model approach breaks the chain into modules, each described by one of a small number of basic processes: growth and inactivation, which change the number of organisms in a unit, and mixing, partitioning, removal and cross-contamination, which change how organisms are distributed between units. Inputs are distributions, and Monte Carlo simulation propagates them to give the distribution of doses per serving.
Growth and inactivation within the model use the predictive microbiology models taught in food-microbiology: for example, secondary models of growth rate against temperature, combined with distributions of storage temperature and time in retail and domestic refrigerators, to estimate growth of Listeria monocytogenes in a ready-to-eat food during its shelf life. Inactivation during cooking uses D and z values, with distributions for cooking practices. Cross-contamination in the kitchen, from raw chicken to salad via hands, boards and knives, is modelled with transfer rates.
QMRA is used to compare the effect of control options, to set or evaluate food safety objectives, to support microbiological criteria, and to guide research. The Joint FAO/WHO Expert Meetings on Microbiological Risk Assessment (JEMRA) have produced international assessments, including those for Salmonella in eggs and broiler chickens, Listeria monocytogenes in ready-to-eat foods, Campylobacter in broiler chickens, Vibrio species in seafood, and Cronobacter sakazakii and Salmonella in powdered infant formula. The Listeria assessment showed that most cases arise from foods that support growth and are eaten after long storage, and that a small proportion of servings with high numbers accounts for most of the risk, which supported risk-based criteria that focus on growth during shelf life. The broiler chicken assessments compared interventions from farm to kitchen.
Full QMRA is data-hungry and slow. Simpler approaches include risk ranking tools and semi-quantitative models that combine scores or simplified calculations, which are useful for setting priorities. Whatever the approach, the assessment must be transparent: its model structure, data sources, assumptions and uncertainties must be documented so that others can review it. QMRA results are best used to compare options, where uncertainties partly cancel, rather than as precise predictions of the absolute number of cases.
Chemical risk assessment follows the four steps already described, with the type of mechanism deciding the approach. For substances with a threshold, the outcome is a health-based guidance value (ADI for intentionally added substances such as additives and residues; TDI, PTWI or PTMI for contaminants; ARfD for acute effects), and risk is characterised by comparing exposure with it. JECFA performs this work internationally for additives, contaminants and veterinary drug residues, and JMPR for pesticides; regional and national bodies such as EFSA, FSANZ and FSSAI's scientific panels perform their own assessments, which sometimes differ because of different data, methods or exposure patterns.
For substances that are both genotoxic and carcinogenic, a threshold is not assumed and no guidance value is set. The margin of exposure (MOE) is the ratio of a reference point, usually the BMDL10 from animal carcinogenicity data, to the estimated human exposure: MOE = BMDL10 / exposure. EFSA has stated that an MOE of 10,000 or higher, based on an animal BMDL10, would be of low concern from a public health point of view and might be considered a low priority for risk management. If the BMDL10 is 0.4 mg/kg bw/day and exposure is 0.002 mg/kg bw/day, the MOE is 200, which indicates a health concern and a priority for action. The MOE is used to compare and prioritise substances, such as acrylamide, polycyclic aromatic hydrocarbons, aflatoxins and glycidol, rather than to give an absolute risk estimate.
Several further tools support chemical risk assessment. The threshold of toxicological concern (TTC) sets generic exposure thresholds, based on chemical structure using the Cramer classification, below which the probability of harm is very low, for substances with little toxicity data present at very low levels, such as flavouring substances and packaging migrants; a lower TTC applies where a structural alert for genotoxicity exists. Read-across uses data from similar substances to fill gaps. Combined exposure to multiple chemicals is assessed by grouping substances with a common mode of action and using approaches such as the hazard index, the sum of each substance's exposure divided by its guidance value, or relative potency factors, as is done for some pesticides and for dioxin-like compounds through toxic equivalency factors.
Risk-benefit assessment weighs risks against benefits of the same food, such as methylmercury against omega-3 fatty acids in fish, or contaminants against nutrients in whole grains, often by expressing both in a common metric such as disability-adjusted life years. Such assessments inform dietary advice, for example to choose fish species lower in mercury rather than to avoid fish, as discussed in food-toxicology-contaminants-allergens.
Food law has long set process and product requirements without explicitly linking them to a public health goal. The concepts developed by ICMSF and adopted by Codex in its principles for microbiological risk management make that link. The appropriate level of protection (ALOP) is the level of protection deemed appropriate by a country establishing a sanitary measure, often expressed as a number of cases of illness per 100,000 people per year. Because an ALOP cannot be applied directly in a factory, it is translated into targets along the food chain.
A food safety objective (FSO) is the maximum frequency and/or concentration of a hazard in a food at the time of consumption that provides or contributes to the appropriate level of protection. A performance objective (PO) is the maximum frequency and/or concentration of a hazard in a food at a specified step in the food chain before consumption that provides or contributes to an FSO or ALOP. A performance criterion (PC) is the effect in frequency and/or concentration of a hazard in a food that must be achieved by the application of one or more control measures to provide or contribute to a PO or FSO, for example a 6-log reduction of a pathogen by cooking. Process criteria, such as a time and temperature, and product criteria, such as pH or aw, are the parameters that deliver the performance criterion.
ICMSF expressed the relationship in a simple equation, working in log10 units: H0 − ΣR + ΣI ≤ FSO (or PO), where H0 is the initial level of the hazard, ΣR the total reduction and ΣI the total increase through growth or recontamination. Suppose raw material carries 3 log CFU/g of a pathogen (H0 = 3), the process gives a 6-log reduction (ΣR = 6), and growth during distribution may add 1 log (ΣI = 1). The level at consumption is 3 − 6 + 1 = −2 log CFU/g, that is, one organism in 100 g, which meets an FSO of −2 log CFU/g but not a stricter one. Rearranged, the equation tells a manufacturer the reduction needed: to meet an FSO of −2 with H0 = 3 and ΣI = 1, ΣR must be at least 3 + 1 − (−2) = 6 logs.
Because the levels are distributions rather than single numbers, more sophisticated versions account for variability and give the proportion of units that will exceed the FSO. Microbiological criteria can then be designed to verify that a PO or FSO is being met, using sampling plans whose performance can be calculated, as described in food-microbiology. The approach gives industry flexibility, since any combination of control measures that achieves the objective is acceptable, while keeping the link to public health. It is used explicitly in some areas, such as Listeria in ready-to-eat foods, and implicitly in many established processes such as the botulinum cook.
A risk profile is a description of a food safety problem and its context, prepared at the start of risk management to help managers decide what to do next. Codex lists typical contents: a description of the hazard and the food or foods involved; how and where the hazard enters the food supply; the frequency and levels of the hazard in foods; the possible adverse effects and the populations at risk; consumer perceptions; the distribution of risk and benefits; existing control measures and international standards; and gaps in knowledge. The profile helps managers decide whether a full risk assessment is needed, what questions it should answer, whether interim measures are justified, and whether the problem is a priority compared with others. It is usually prepared quickly from existing information and is not itself a risk assessment. For example, a profile on Cronobacter in powdered infant formula would summarise the organism's survival in dry powder, the severe outcomes in newborn infants, reconstitution and holding practices, and existing criteria, and would help managers decide to commission an assessment of preparation practices.
Authorities must choose where to put limited resources, so they rank hazards and hazard-food combinations. Approaches range from simple risk matrices, which score likelihood and severity and combine them, through semi-quantitative scoring tools and multi-criteria decision analysis, which weigh several criteria such as health impact, economic cost, public concern and feasibility of control, to fully quantitative comparisons based on burden of disease.
Burden of disease is usually expressed in disability-adjusted life years (DALYs), which sum years of life lost to premature death and years lived with disability, weighted by the severity of the disability. DALYs allow diseases as different as a short bout of diarrhoea, chronic liver cancer and lifelong disability after congenital toxoplasmosis to be compared on one scale. The WHO Foodborne Disease Burden Epidemiology Reference Group estimated in 2015 that food-borne hazards caused around 600 million illnesses, 420,000 deaths and 33 million DALYs worldwide in 2010, with the greatest burden in children under five and in low-income regions; diarrhoeal disease agents such as norovirus, Campylobacter and non-typhoidal Salmonella caused most illnesses, while invasive infections, some parasites and chemicals such as aflatoxin contributed heavily to deaths and DALYs. Source attribution estimates what proportion of cases of a disease comes from each food or route, using outbreak data, microbial subtyping, case-control studies and expert elicitation, so that burden can be allocated to foods. Ranking results depend strongly on the criteria and weights chosen, so they should be presented transparently and treated as a guide to priorities rather than a precise ordering.
Codex describes risk management as a generic framework with four phases. Preliminary risk management activities identify the food safety issue, develop a risk profile, establish goals, decide whether a risk assessment is needed, set the risk assessment policy, commission the assessment and consider its results. Identification and selection of risk management options then evaluates the available options against the desired level of protection and other factors, and selects one or a combination. Implementation puts the chosen measures into effect, by regulators and by industry. Monitoring and review gather information, for example from surveillance of illness and of hazard levels in foods, to judge whether the measures are working and whether new information requires the decision to be revisited.
Risk management decisions take into account the risk assessment and other legitimate factors: technical feasibility, economic consequences, effects on trade, consumer concerns and ethical considerations. Where a hazard has no threshold, levels are kept as low as reasonably achievable (ALARA). Proportionality, consistency between similar risks and non-discrimination between domestic and imported products are required by the SPS Agreement. Where scientific evidence is insufficient, provisional precautionary measures may be taken pending further information.
Options for managing a food safety risk can act at any point in the chain. Regulatory standards include maximum levels for contaminants, maximum residue limits for pesticides and veterinary drugs, microbiological criteria, and limits on additive use. Process requirements include mandatory treatments such as pasteurisation of milk, the botulinum cook for low-acid canned foods or freezing of fish for raw consumption, and requirements for HACCP-based systems, described in food-safety-management-systems. Controls at primary production include animal vaccination programmes, such as vaccination of laying hens against Salmonella, good agricultural practices and restrictions on harvesting shellfish from contaminated waters. Market controls include import requirements, sampling and testing, recalls and, in extreme cases, bans. Information measures include labelling, such as 'keep refrigerated' or 'cook thoroughly', allergen labelling and consumer advice targeted at vulnerable groups, for example advice to pregnant women to avoid certain soft cheeses and pâtés because of Listeria.
Choosing among options involves weighing effectiveness in reducing risk, cost, feasibility, enforceability, and unintended consequences. A very strict limit on a contaminant may remove a nutritious staple from the diet of poor consumers; a ban on raw milk sales may move them into unregulated channels. Cost-benefit and cost-effectiveness analyses, using burden-of-disease measures, help compare options, and the risk assessment can be used to estimate how much each option reduces risk. Monitoring after implementation, through surveillance of food-borne illness and of hazard levels, closes the loop.
Risk communication is the interactive exchange of information and opinions throughout the risk analysis process. It is not only announcing decisions: it includes involving interested parties when problems are framed and assessment policy is set, explaining assessments and their uncertainties, consulting on management options, and listening to concerns. FAO and WHO published a handbook on risk communication applied to food safety in 2016, and agencies such as EFSA, FSANZ and FSSAI have communication strategies.
People's judgements of risk differ systematically from technical estimates, and these differences are not simply ignorance. Research on risk perception, notably by Paul Slovic and colleagues, found that risks are perceived as greater when they are involuntary rather than chosen, unfamiliar or new, catastrophic or dreaded (such as cancer), uncontrollable by the individual, unfairly distributed, affecting children, or created by human activity rather than nature, and when the institutions responsible are distrusted. This explains why consumers may worry more about pesticide residues and additives, which are involuntary and human-made, than about microbial contamination in their own kitchens, which is familiar and seen as controllable, although the burden of disease suggests the opposite priority. Peter Sandman summarised the idea as risk = hazard + outrage: public reaction depends on outrage factors as well as on technical hazard.
Effective risk communication follows a few principles: start early; know the audiences, which differ in knowledge, concerns and trusted sources; be open about what is known and not known, including uncertainty; explain the basis of decisions and the trade-offs involved; give practical advice that people can act on; use plain language and consistent messages across agencies; and listen, so that communication is two-way. Trust, built through competence, honesty and consistency over time, is the most valuable asset, and it is easily lost by delay, contradiction or the appearance of hiding information.
In a food safety crisis, such as an outbreak or a major contamination incident, communication must be rapid and coordinated: what is known, which products are affected, what people should do, what authorities and companies are doing, and when the next update will come. The Codex principles and guidelines for exchange of information in food safety emergencies, the WHO International Food Safety Authorities Network (INFOSAN) and national systems such as RASFF in the EU support rapid exchange between authorities. Mistakes in past crises, such as reassurance later shown to be wrong during the BSE episode, damaged trust for years and contributed to the reform of food law. Communication about benefits as well as risks, for example of eating fish or fortified foods, needs equal care so that consumers are not driven from healthy choices by poorly explained risks.
International risk assessment for Codex is performed by independent expert bodies convened by FAO and WHO, whose members serve as individual scientists, not as representatives of governments or industry. The Joint FAO/WHO Expert Committee on Food Additives (JECFA), which first met in 1956, evaluates food additives, contaminants, naturally occurring toxicants and residues of veterinary drugs, setting ADIs, tolerable intakes and recommending MRLs for veterinary drugs, and specifications for additives. The Joint FAO/WHO Meeting on Pesticide Residues (JMPR), which first met in 1963, evaluates pesticides, setting ADIs and ARfDs and recommending MRLs on the basis of supervised trial data. The Joint FAO/WHO Expert Meetings on Microbiological Risk Assessment (JEMRA), established in 2000, carry out microbiological risk assessments and provide advice on microbiological hazards. Other ad hoc expert meetings address nutrition (JEMNU) and specific issues such as new technologies.
The Codex committees act as risk managers: for example, the Codex Committee on Contaminants in Foods sets maximum levels after considering JECFA's assessment, and the Codex Committee on Pesticide Residues proposes MRLs on JMPR's advice. National authorities may use these assessments directly or perform their own. Countries with limited capacity rely heavily on them, which is one reason why participation in Codex and support for the expert bodies matter for developing countries.
Risk analysis depends on surveillance, the systematic collection, analysis and interpretation of data on food-borne disease and on hazards in food, to identify problems, provide data for assessments and evaluate the effect of controls. Food-borne disease surveillance includes notification of diseases by clinicians and laboratories, laboratory-based surveillance of pathogens with molecular subtyping, outbreak detection and investigation, and population-based studies that correct for under-reporting, since most cases of food-borne illness are never reported. Networks that share subtyping data, now increasingly based on whole-genome sequencing, detect dispersed outbreaks across regions and countries. In India, the Integrated Disease Surveillance Programme collects reports of outbreaks including food-borne disease, and FSSAI conducts national food surveillance surveys.
Surveillance of hazards in food includes official control sampling, national residue monitoring plans, total diet studies and targeted surveys. Internationally, the WHO GEMS/Food programme compiles data on contaminants in food and consumption, the International Food Safety Authorities Network (INFOSAN), managed by FAO and WHO since 2004, links national authorities for rapid exchange of information during food safety emergencies, and the WHO Foodborne Disease Burden Epidemiology Reference Group estimates the global burden. Surveillance closes the risk analysis loop: it identifies emerging hazards, supplies the data for exposure and hazard characterisation, and shows whether risk management measures have reduced illness.
A student uses this subject to learn a vocabulary and a way of thinking: hazard versus risk, the four steps of risk assessment in order, and how exposure, dose-response and guidance values combine into a judgement. The calculations recur: exposure from concentration and consumption, percentage of an ADI, margin of exposure, probability of illness from the exponential model, and the ICMSF equation. Practising each with real numbers makes the concepts concrete.
Five years into a regulatory, quality or research role, a technologist prepares risk profiles for new issues, estimates exposure to a contaminant or additive in the company's products, reads JECFA, EFSA or FSSAI opinions and works out what they mean for specifications, uses predictive models and simple QMRA to judge shelf life or a process change, and drafts consumer advice or responses to customer questions in plain language.
A senior regulator or head of scientific affairs with twenty-five years' experience commissions and reviews risk assessments, sets risk assessment policy, chooses between management options with incomplete evidence, ranks priorities for surveillance and inspection, represents the country or company in Codex and expert meetings, and leads communication during crises. Such a person keeps the line between science and policy clear while making decisions that must weigh both.
A researcher at this level develops the methods: dose-response models for vulnerable groups, probabilistic exposure assessment with better treatment of uncertainty, source attribution using genomic data, burden-of-disease and risk-benefit metrics, new approach methods replacing animal tests in hazard characterisation, and the evaluation of risk communication. Such work is judged by transparency, reproducibility and its usefulness to decisions actually being made.
This manual explains concepts and simple calculations; it does not teach probabilistic modelling software, second-order Monte Carlo methods, Bayesian dose-response fitting or the statistics of benchmark dose modelling in detail. Specific ALOPs, FSOs and national risk-ranking outputs are not given because few are published and they differ by country. Burden-of-disease figures are the 2015 WHO estimates for 2010 and will be superseded. Toxicological guidance values are explained in food-toxicology-contaminants-allergens, microbiological criteria in food-microbiology and legal frameworks in food-law-regulatory-affairs.
Original practice questions written to the syllabus and pattern of the exams named in each question's examTags. They are not past papers and do not reproduce any board's questions.
These notes are free to read here, with no account.
What the Laisha app adds:
Free for ten days, then a subscription. We would rather write that here than have you find it later.