2.4 MRA approaches
2.4.1 Overview
Risk assessment (RA) is the science-based component of risk analysis, which also in- cludes risk management and risk communication as further steps (FAO/WHO 2002) (Figure 2.1). It provides a framework for organizing data and information in a struc- tured manner and thereby allows us to understand better the interaction between haz- ards, foods and human illness (Cahill 2005). Microbiological risk assessment (MRA) is commonly concerned with food safety risks, and involves estimation of the magnitude of microbial exposure at various stages in the food production chain to estimate the risk of food borne infections (Thrusfield 2005). It has become a key feature of food safety management and is regularly used to guide policy (Kelly, Hartnett et al. 2003). MRA is evolving within both food safety regulatory agencies and academia in an ever increasing number of countries and is recognised as a resource-intensive task requiring a multidisciplinary approach (Cahill 2005).
The FAO/WHO Codex Alimentarius Commission (CAC) defines risk assessment as a scientifically process based on four steps (Codex Alimentarius Commission (CAC) 1999):
• Hazardidentification, which is the identification of the biological agent that may be present in a particular food or group of foods and capable of causing adverse health effects.
Figure 2.1: The Risk Analysis Framework (World Health Organisation 2007)
• Hazard characterization, which is the qualitative or quantitative, or both, evaluation of the nature of the adverse health effects associated with the biological agents that may be present in food, and in such cases a dose-response assessment should be performed if the data are obtainable.
• Exposure assessment, which is the qualitative or quantitative, or both, evalu- ation of the likely intake of the biological agent through food, as well as through exposure from other sources, if relevant.
• Risk characterisation, namely the qualitative or quantitative, or both, esti- mation, 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 characterization and exposure assessment. Conceptually the risk assessment approach starts from the dynamics of the hazard in the food chain and uses predictive models to estimate the outcome in terms of public health. It thereby can provide valuable information on the complex dynamics of pathogens during food processing (Havelaar, Braeunig et al. 2007).
MRAs are increasingly used to understand how pathogens are propagated along the food chain and to improve the quality and safety of food. With clear documentation
MRA provides several advantages: scientific justification for actions, a means of demon- strating equivalence as well as an effective communication tool (Cahill 2005). This approach is particularly useful as by describing the system in mathematical models the changes in model output when certain interventions are implemented can be examined (Havelaar, Braeunig et al. 2007). In addition the increased insight in the propagation of hazardous microorganism in the food chain helps to identify more targeted control strategies. Finally, MRAs help to identify data gaps, and thereby facilitate the iden- tification of research needs, the establishment of research priorities and the design of commissioned studies (Cahill 2005). If exposure estimates and dose-response functions are sufficiently accurate, risk assessments may provide excellent estimates of the true impact of illness (Batz, Doyle et al. 2005).
To date risk assessments on several food pathogen combinations have been success- fully completed and support decision making in food safety. However while MRA is becoming an important tool for assessing risk from food borne pathogens, often it is not within the capacity of individual countries to complete a quantitative MRA as it is a resource-intensive and multidisciplinary task (FAO/WHO 2002). Because of this MRAs have been undertaken for only a limited number of pathogen-food combinations to date (Batz, Doyle et al. 2005). However MRAs are increasingly used for animal food products in particular as a result of the adoption of Sanitary and Phytosanitary (SPS) agreement by WTO member states in 1995. This agreement requires all member states to base all food safety regulatory measures on sound scientific risk assessment (Kelly, Hartnett et al. 2003).
Most quantitative RA models deal with one pathogen occurring in a single food commodity (Havelaar, Braeunig et al. 2007), for exampleCampylobacter spp in chicken (Hartnett, Kelly et al. 2001; Rosenquist, Nielsen et al. 2003), and are often aimed to identify options for prevention, intervention and control (Havelaar, Braeunig et al. 2007). Amongst others the risk of infection from Listeria in soft cheese (Bemrah, Sanaa et al. 1998), Salmonella in eggs and broiler chicken (FAO/WHO 2002) and E. coli in steak tartar (Nauta, Evers et al. 2001) has been analysed. A number of different model- based approaches have evolved and been applied in microbial food safety. Factors such as scope, available time, data resources, and the skill and expertise of the risk assessor will all have an impact on the approach taken (Cahill 2005). Table 2.1 gives a schematic overview of different approaches to MRA.
Source attribution has recently evolved as a novel approach to microbial risk assess- ment and a variety of tools and methods are becoming available. Although included in this overview of approaches to MRA these techniques are in more detail described and discussed in section 2.4.4 of this chapter.
Common to all risk assessment is the need for representative data. Therefore in general most risk assessment approaches are resource-intensive and in the face of large
data gaps may not be practical or possible at all. In particular source attribution methods, that rely on pathogen subtyping results, are dependent on the availability of good quality data on a high level of resolution. As in all epidemiological studies, bias is of major concern for the validity of results (Rothman and Greenland 1998) and should carefully be considered.