Sarah Cahill, Ph.D., is Nutrition Officer (Food Microbiology) in the Food Quality and
Standards Service, FAO Food and Nutrition Division. She is a member of the Secretariat
for the joint FAO/WHO activities on microbiological risk assessment (JEMRA). This paper
summarizes one of the risk assessments undertaken as part of the joint activities.
Salmonellosis is a leading cause of food-borne illness in many countries. International data summarized by Thorns (2000) provide estimated incidences of salmonellosis per 100 000 people for the year 1997: 14 in the United States of America, 38 in Australia and 73 cases in Japan. In the European Union, the estimates range from 16 cases per 100 000 people in the Netherlands to 120 cases per 100 000 people in parts of Germany. Eggs and poultry have been identified as important vehicles in the transmission of Salmonella (Bryan and Doyle, 1995; Humphrey, 2000; Bean et al., 1997; Humphrey, Mead and Rowe, 1988).
Poultry and egg production and consumption are increasing worldwide (FAO, 2001). Broiler chickens are the main type of chicken consumed as poultry in many countries. Large numbers of broiler chickens are colonized by salmonellae during grow-out and the skin and meat of carcasses are frequently contaminated by the pathogen during slaughter and processing. Salmonella Enteritidis, in particular, has become a leading cause of human infections during the last two decades. Chicken eggs are a principal source of this pathogen (Thorns, 2000). The emergence of Salmonella Enteritidis as the leading cause of human salmonellosis in many countries was attributed to the unusual ability of this serovar to colonize ovarian tissue of hens and be present within intact shell eggs. Poultry and eggs play a major role as vehicles in human cases of salmonellosis in an environment of increasing inter-national trade in poultry and increasing consumption of poultry meat. Therefore, an assessment of the different factors affecting the prevalence, growth and transmission of Salmonella in eggs and on broiler chicken carcasses would be useful to risk managers for identifying the most effective intervention strategies for reducing human infections.
Risk assessment is a tool that can be used in managing the risks posed by food-borne pathogens and in elaborating standards for foods in international trade. The four components of this tool are hazard identification, hazard characterization, exposure assessment and riskcharacterization (Figure 1). However, undertaking a microbiological risk assessment (MRA), particularly quantita tive MRA, is a resource-intensive task requiring a multidisciplinary approach. It is not currently within the capacity of many countries to carry out a complete quantitative MRA. However, new approaches to addressing the problems and impact of food-borne illness are required. Risk assessment can be used to provide increased understanding of how to address this problem and also to justify the introduction of more stringent standards for imported foods. Therefore, a knowledge of MRA is important for trade purposes, and countries need tools for understanding and, if possible, undertaking MRA. These needs, combined with the requests by the Codex Alimentarius Commission (CAC) and the Codex Committee on Food Hygiene (CCFH) for scientific advice based on risk assessment, led FAO and WHO to initiate a programme of activities to address the issue of MRA at the international level. 1
FAO and WHO undertook risk assessments of Salmonella in eggs and in broiler chickens to provide scientifically based expert advice to their member countries and to the CAC, and more specifically the CCFH, to help them address the problems associated with the presence of this pathogen in these foods. The work involved the collection and compilation of all currently available relevant information, the development of a framework and a mathematical model for risk assessment of this pathogen-commodity combination, consideration of how the risk assessment might be used to evaluate the efficacy of specific risk management interventions and provision of a toolbox for countries to use in the elaboration of their own national risk assessments. This paper summarizes the risk-assessment work and its potential application.
The FAO/WHO risk-assessment work on Salmonella in eggs and broiler chickens followed the stepwise approach outlined in Figure 1. In fact, two risk assessments were undertaken, one on Salmonella Enteritidis in eggs, and the other on Salmonella in broiler chickens. However, a common hazard characterization was used for both risk assessments. For Salmonella Enteritidis in eggs, the risk assessment estimated the probability of human illness due to Salmonella Enteritidis following the ingestion of a single serving of shell eggs internally contaminated with this pathogen. The eggs may have been consumed whole, as part of an egg-based meal, or as ingredients in more complex food (e.g. cake).The work addressed selected aspects of egg production on farms, further processing of eggs into egg products, retail and consumer egg handling and meal-preparation practices. For Salmonella in broiler chickens, the risk assessment estimated the prob-ability of an individual becoming ill over a one-year period from the ingestion of Salmonella on fresh whole unskinned broiler chicken carcasses cooked in a domestic kitchen for immediate consumption. The work commenced at the conclusion of slaughterhouse processing and took home handling and cooking practices into consideration. The effects of pre-slaughter interventions and the slaughter process on the probability of exposure to, and becoming ill from, Salmonella on chicken were not included in this model.
FIGURE 1
Steps in a a microbiological risk assessment
Hazard identification, which aims to identify the hazard capable of causing an adverse health effect, the food products in which it may be present, and the adverse health effects associated with the pathogen, was based on a review of published literature and on unpublished information submitted to FAO and WHO by interested parties in response to a call for data issued by both organizations.
Hazard characterization evaluates what will happen if someone is exposed to the hazard, i.e. Salmonella. If possible it will include a dose-response assessment that aims to determine the quantity of the hazard that will make someone ill and how ill that person will be. The hazard characterization for Salmonella detailed the pathogen characteristics, host characteristics and food-related factors that may affect the survival of Salmonella through the stomach. It also presented a review of the work undertaken by the scientific community, to date, on the derivation of dose-response models describing the mathematical relationship between an ingested dose of Salmonella and the probability of an individual becoming ill.
An extensive review of available outbreak data was also conducted. Using the outbreak data, which included information on the numbers of Salmonella in the implicated food, a new dose-response model was derived. This model was used in the next step of both the Salmonella Enteritidis in eggs and Salmonella in broiler-chicken risk assessments, in preference to the pre-existing models. Finally, an attempt was made to discern whether separate dose-response curves could be justified for different human subpopulations defined on the basis of age and "susceptibility", and whether Salmonella Enteritidis had a dose response different from other strains of Salmonella.
Exposure assessment evaluates the likelihood that a person will be exposed to a hazard - in this case Salmonella - and, if so, the amount of the hazard. This re-quires an examination of all or part of the food chain from production to consumption, and in particular the various stages that may impact both presence and quantity of Salmonella in eggs and chicken when an individual eats these foods.
Poultry and eggs play a major role as vehicles in human cases of salmonellosis in an environment of increasing international trade in poultry meat - FAO/18582/R.Faidutti
The exposure assessment for Salmonella Enteritidis in eggs considered the food chain from production to consumption and comprised a production module, a module for the processing and distribution of shell eggs, processing of egg products and a module for preparation and consumption. The production module predicted the probability of a Salmonella Enteritidis-contaminated egg occurring. The processing and distribution, and preparation and consumption modules predicted the probability of people being exposed to various doses of Salmonella Enteritidis from contaminated eggs. The model that has been developed combines existing models that were elaborated in Canada and the United States.
The exposure assessment for Salmonella in broiler chickens tracked the movement of Salmonella-contaminated chickens through the food chain commencing at the end of the slaughter process. Contrary to the model for eggs, here it was not possible to take the effect of prior production and processing steps into consideration, because adequate quantitative data concerning the prev-alence, numbers and fate of Salmonella in chicken during these stages are not currently available. Therefore, in this model a chicken carcass was randomly allocated a "contaminated" or "not contaminated" status after processing. Those carcasses tagged as "contaminated" were randomly assigned a number of Salmonella organisms, to indicate the level to which each chicken was contaminated. These assumptions about contami-nation and extent of occurrence were based on information derived from empirical data gathered in Canada and in the United States. From the post-processing stage until consumption, changes in the size of the Salmonella population on each contaminated chicken were modelled using equations that predict growth and death. The growth of Salmonella was predicted using random inputs for storage time at retail stores, transport time, storage time in homes and the temperatures to which the carcass was exposed during each of these periods. Here, as elsewhere in the model, random inputs were selected from a distribution of possible inputs generated from a database of available empirical information. Death of Salmonella during cooking was predicted using random inputs describing the probability that a carcass was not adequately cooked; the proportion of Salmonella organisms attached to areas of the carcass that were protected from heat; and the temperature to and time period for which these "protected" bacteria were exposed. The number of Salmonella consumed was then derived using a random input defining the weight of chicken meat consumed per serving and the number of Salmonella organisms in the meat, as determined from the various growth and death processes.
Finally, in the risk-characterization step, the probability of illness is derived by combining the number of organisms ingested (from the exposure assessment) with information on the dose-response relationship (hazard characterization). For the assessment of Salmonella Enteritidis in eggs, the output of the risk characterization was the probability of human illness per serving of an egg-containing meal. For that of Salmonella on broiler chickens, the risk characterization estimated the probability per year of an individual becoming ill from Salmonella after eating a serving of chicken.
The inputs for the exposure assessment and hazard characterization steps were obtained from a variety of sources. These included published literature, national reports and unpublished data submitted to FAO and WHO by various interested parties.
In the course of this work it was determined that existing dose-response models for Salmonella, based on experimental volunteer studies in young, healthy subjects, inadequately characterized the apparent dose-response relationships observed in the available outbreak data. A new dose-response model was developed from outbreak data and was considered, by an FAO/WHO expert group, to be the most appropriate estimate for the prob-ability of illness upon ingestion of a defined number of Salmonella cells. This new model was based on observed real-world data, and was not subject to some of the flaws inherent in using purely experimental data. Nevertheless, the current outbreak data also included uncer-tainties, and some of the outbreak data points required assumptions to be made about the numbers of people exposed, the number of people who became ill and the number of Salmonella cells ingested. It should be noted that the outbreak data were supplied by a number of developed countries and may not be applicable to other regions where a difference in host immunity may influence the dose-response relationship. This represents a gap in the knowledge and database that should be investigated in the future.
From the outbreak data used to examine the dose-response relationship, it was not possible to conclude that Salmonella Enteritidis had a different likelihood of producing illness than other Salmonella serovars. In addition, comparing the attack rates of Salmonella for children under five years of age against those for the rest of the identified population in the outbreak database did not reveal an increased risk for this subpopulation. Some indication for a difference in attack rates for the two populations was noted in two of the outbreaks examined. However, the database of outbreak information lacked the capacity to reveal the presence of any true differences that might exist. A key aspect is that severity of illness, once the patient is exposed and becomes ill, could potentially be influenced by age, immune status, Salmonella serovar and pathogen dose. However, this aspect was not evaluated because the current database of information was insufficient to derive a quanti-tative estimate of such factors.
This risk assessment for Salmonella Enteritidis in eggs was intentionally developed so as not to represent any specific country or region. However, some of the model inputs were based on evidence or assumptions derived from specific national situations. Therefore, caution is required when extrapolating from this model to other countries.
The exposure assessment included consideration of yolk-contaminated eggs and growth of Salmonella Enteritidis in eggs prior to processing for egg products. These issues had not been previously addressed by exposure assessments of Salmonella Enteritidis in eggs. Yolk-contaminated eggs may allow for more rapid growth of Salmonella Enteritidis when compared to eggs that are not yolk-contaminated.
The estimated risk of human illness from Salmonella Enteritidis in eggs varied according to the different input assumptions in the model. The risk of illness per serving increased as the prevalence of Salmonella in flocks (on a countrywide basis) increased, although uncertainty regarding the predicted risk also increased as flock prevalence increased. Reducing the number of Salmonella-positive flocks, i.e. reducing flock prevalence, resulted in a directly proportional reduction in human health risk. For example, reducing flock prevalence from 50 to 25 percent resulted in halving the mean probability of illness per serving, given the manner in which the risk assessment defined the likely processing and consumption of contaminated products. Similarly, reduc-ing the number of Salmonella-positive birds within infected flocks, i.e. the within-flock prevalence, resulted in a directly proportional reduction in human health risk. For example, risk of illness per serving generated from eggs produced by a flock of chickens with a 1 percent within-flock prevalence was one-tenth that of a flock with 10 percent within-flock prevalence.
It was calculated that modifying egg storage temperatures and times from production to consumption greatly affected the predicted risk of human illness. For any defined flock prevalence, the risk decreased when the storage temperature and time were reduced below the baseline time-temperature conditions used in this model. In contrast, the risk increased substantially when the storage time and temperature were elevated above the baseline parameters. The degree of change in human health risk is dependent on the baseline conditions used in the model. However, the simulations carried out in this risk assessment appeared to indicate that changing storage times and temperatures from farm to table resulted in disproportionately large effects on the risk of illness.
The number of Salmonella Enteritidis bacteria in contaminated eggs across the range considered (based on experimental data) at the time of lay appeared to have a small influence on the risk of human illness per serving. For example, whether it was assumed that all contaminated eggs had an initial number of 10 versus 100 Salmonella Enteritidis organisms, the predicted risk of illness per serving was similar. This may be because the effect of Salmonella Enteritidis growth is greater than the initial contamination level in eggs. It should be noted that the available data on which this risk assessment was based were limited. For example, evidence regarding enumeration of the organism within eggs was based on only 63 Salmonella Enteritidis-contaminated eggs. It is difficult to represent uncertainty (an expression of our lack of knowledge) and variability (naturally occurring heterogeneity) with such limited data. In ad-dition, statistical and model uncertainties were not fully explored.
This risk assessment model was defined in terms of a number of parameters that describe the processes of broiler chicken carcass distribution and storage, prepar-ation, cooking and consumption. Some of these parameters can be considered general in that they can be used to describe the situation in many countries. Others are country specific, for example the prevalence of carcasses contaminated with Salmonella at the completion of processing. Predictions of risk for a particular country are best obtained from data relevant to that country.
The Salmonella in risk assessments of broiler chickens did not consider all aspects of the farm-production-to-consumption continuum. This restriction limited the range of control options that could be assessed. This was primarily due to the lack of representative data to analyse how much change in either the prevalence and/or numbers of Salmonella in poultry could be attributable to any specific treatment or action prior to the end of processing. However, the establishment of a baseline model provided a means to compare the effects on human health risk when prevalence of Salmonella and level of contamination (i.e. numbers of Salmonella cells) were changed. The model parameters can be modified to evaluate the efficacy of risk-mitigation strategies that target Salmonella prevalence and numbers. For example, the parameter describing prevalence of Salmonella -contaminated broiler chickens exiting processing can be modified to evaluate the effectiveness of a processing measure such as chlorination of the chilling water to reduce the prevalence of Salmonella -contaminated carcasses. The model permits the effectiveness of such measures to be quantified in terms of reduction of numbers (log reduction) and/or prevalence of Salmonella .
As can be expected, reduction in the prevalence of Salmonella -contaminated chicken was associated with a reduction in the risk of illness. A one-to-one re-lationship was estimated. For a percentage change in prevalence, a reduction in the expected risk by a similar percentage was calculated, if defined distribution, prepar- ation and consumption patterns re- mained the same. For example, a 50 percent reduction in the prevalence of contaminated poultry (e.g. from 20 to 10 percent) produced a 50 percent reduction in the expected risk of illness per serving. Similarly, a larger reduction in prevalence (e.g. from 20 to 0.05 percent) produced a 99.75 percent reduction in the expected risk of illness.
To ensure that risk assessment contributes to management decisions that can be successfully implemented, there needs to be communication from the outset with other relevant stakeholders, such as the food industry and consumers
If management strategies affecting contamination levels (number of Salmonella - on chickens) were implemented, then the relationship to the human health risk of illness was estimated to be greater than a one-to-one relationship. For example, reducing the Salmonella cell numbers on broiler chickens exiting the chill tank at the end of processing by 40 percent reduced the expected risk of illness per serving by approximately 65 percent.
A small reduction in the occurrence and extent of undercooking resulted in a marked reduction of the expected risk of illness per serving. An important caveat here is that altering consumers' cooking practices does not address the risk of illness through the cross-contamination pathway. Strategies to influence consumers' cooking practices should be tempered by the fact that cross-contamination may be a more predominant route of exposure to Salmonella - and may be a greater contributor to the risk of illness. The nature of cross-contamination in the home is still a highly uncertain phenomenon; while it is acknowledged that this is an important factor, exactly how, when, how much and how often this occurs is unknown.
Quantitative MRA is intended to answer specific questions of importance to public health. For MRA to deliver benefits, it needs to be purposefully incorporated into the decision-making process. This implies a change in the way national governments approach food safety and public-health decisions. The novelty of MRA is that it quantifies the hazard throughout the food production chain and directly links this to the probability of food-borne disease. The Salmonella risk assessments present a powerful example of the potential in using this approach. However, it must be noted that it would not be necessary in all cases to use such advanced tools to provide an objective basis for food-safety decisions.
Risk assessments are a resource that can be used by many parties, including national authorities. Ensuring the applicability and utility of the risk assessment to all regions and all countries is a priority for future work in FAO. This exercise in conducting risk assessment at the inter-national level has underlined the need to acquire data from all regions and for the development of countries' capacities to conduct risk assessments. The development of these capacities requires an infrastructure for the surveillance of food-borne illness and the monitoring of microbial hazards in foods throughout the food chain. Trained personnel with the technical skills to conduct MRAs are also a fundamental requirement.
The mathematical models developed for this work can be used for risk assessment of these pathogen-commodity combinations at regional or national levels as soon as they become available. The data used should reflect the food item, raw materials, manufacture, retail conditions and consumption habits in the region under consideration. Predictive models for growth, survival or decline of micro-organisms must still be used with parameters fitted to the regional or national conditions of interest.
Not all of the components of the risk assessments may be significant to the end user. The concepts presented here are generic, and may be directly adaptable or considered as stand-alone modules. However, much of the data used is of a more specific nature and will not be representative of situations in many countries. The exposure assessments are reasonably close to the exposure scenarios in some developed countries, but they do not truly represent any one country or situation. Therefore, the exposure assessment component should not be used without careful scrutiny of its applicability to the national situation or without taking into consideration the exposure pathway in the country. The hazard characterization portion of the risk assessment offers the most readily adaptable data and models for users. Nevertheless, the use of the model needs to be assessed in countries other than those from which the data used to generate the dose-response curve were collected. In cases where the information may not be applicable, the required data can probably still be collected. However, within any human population, there are subgroups characterized by different levels of susceptibility to initial infection/illness (acute gastro-enteritis) and/or subsequent severity of the initial illness. The degree of susceptibility and size of these subgroups must be defined carefully.
Microbiological risk assessment series
These Salmonella risk assessments provide information that may be useful in determining the effect of intervention strategies on reducing cases of salmonellosis from contaminated eggs and poultry. This information is of particular interest for Codex Alimentarius in its work on the elaboration of standards, guidelines and related texts. In the risk assessment of Salmonella in broiler chickens, for example, a direct relationship between changing the prevalence of Salmonella on the broiler chickens and a reduction of the risk of illness per serving of chicken has been determined. In that of Salmonella Enteritidis in eggs, reducing the prevalence of Salmonella a Enteritidis in poultry flocks was directly proportional to the reduction in risk to human health. The latter model can also be used to estimate the change in risk of human illness from changing storage times or temperature of eggs.
The risk assessments also provide an example of a format for organizing available information in a readable way, and connecting pathogen-contamination problems in food with human health outcomes. They provide scientific advice and analysis that may be useful for establishing regulatory policies for control of food-borne disease in different countries. In addition, the risk-assessment process has identified important data gaps, and includes recommendations for future research, which can be used to allocate resources to priority areas.
Risk assessment is a tool that can be used to assist decision-making in the risk management process. In undertaking this work a number of lessons were learned with regard to making optimal use of risk assessment as a decision support tool. In order to meet the needs of risk managers, risk assessment must be clearly focused, which can be achieved by adequate planning, good communication and a strong interface between the risk assessors and the risk managers. To ensure that risk assessment contributes to management decisions that can be successfully implemented, there needs to be communication from the outset with other relevant stakeholders, such as the food industry and consumers.
These are the first MRAs to have been undertaken at the international level. During the course of the work it was recognized that MRA is still a developing science. There is a need to develop robust modelling and risk-assessment tools and to provide training in their use, to educate the relevant organizations and personnel on the data needs, and to provide resources for data collection for risk assessment. The final products of this work will be technical documents on the risk assessments of Salmonella Enteritidis in eggs and Salmonella in broiler chickens; an interpretative summary of the work; and modelling tools, which will be made available to the international community. These products should contribute towards meeting the needs identified above.
The technical document is a compilation of a wealth of infor-mation on Salmonella in eggs and broiler chickens, organized in a systematic risk-assessment framework. The organization of data in this format has made it possible to identify key gaps in the information base. The risk-assessment framework and models that have been developed also make it possible to evaluate the relative effect of intervention strategies. The interpretative summaries are aimed more towards risk managers and other potential users of the risk assessments whose work is to develop policy, guidance and/or regulations aimed at reducing the risk to human health associated with Salmonella in poultry. In the longer term, the mathematical modelling tools will be made available in a suitable format to assist those interested in developing their own models. When combined, all of this work provides a valuable and unique resource for those undertaking risk assessments and addressing the problems associated with Salmonella in eggs and broiler chickens.
FAO greatly appreciates the contribution of all those experts who assisted in gene rating these risk assessments and in particular Wayne Anderson, Eric Ebel, Aamir Fazil, Fumiko Kasuga, Louise Kelly, Anna Lammerding, Roberta Morales, Wayne Schlosser, Emma Snary, Andrea Vicari, and Shigeki Yamamoto, whose efforts as members of the risk assessment drafting group made this work possible.
References
Bean, N.H., Goulding, J.S., Daniels, M.T. & Angulo, F.J. 1997. Surveillance for foodborne disease outbreaks - United States, 1988-1992. Journal of Food Protection, 60: 1265-1286.
Bryan, F.L. & Doyle, M.P. 1995. Health risks and consequences of Salmonella and Campylobacter jejuni in raw poultry. Journal of Food Protection, 58: 326-344.
Humphrey, T. 2000. Public health aspects of Salmonella infection. In C. Wray & A. Wray, eds. Salmonella in domestic animals, pp. 245-262. New York, CAB International.
Humphrey, T., Mead, G. & Rowe, B. 1988. Poultry meat as a source of human salmonellosis in England and Wales. Epidemiological Infection, 100: 175-184.
FAO. 2001. Meat and meat products. GIEWS Food Outlook, 3. Rome.
Thorns, C.J. 2000. Bacterial food-borne zoonoses. Revue scientifique et technique, Office International des Epizooties, 19(1): 226-239.
1 The risk assessment document and the interpretative summary can be found on the FAO Web site: http://www.fao.org/es/ESN/food/risk_mra_riskassessment_salmonella_en.stm
