Microorganisms are the major cause of foodborne diseases worldwide and they continue to remain a major public health concern, despite the stringent food safety system. Food safety has been constantly evolving in the recent past, while also the microorganisms. Microorganisms and their toxic metabolites have known to exist in food since ancient times. The consumption of food or water containing these microbial pathogens and / or toxins can result in foodborne disease. The risk of foodborne disease is due to a combination of several factors, and it is the probability of being exposed to the pathogen, the probability of becoming ill and the severity of illness. An examination of these several factors in a systematic way along the food production chain is necessary to ensure food safety. However, it is not an easy task to perform systematic analyses along the entire food chain. The current globalization trend in food trade makes the food production chain more complex. As the food production chain becomes more complex, the risk involved will also be complex.
Given the current global scenario of food, It is imperative to perform a food safety risk assessment along the food supply chain (Lammerding and Fazil 2000). In general, risk assessment is a part of risk analysis and is aimed at making risk management decisions by taking into account the magnitude of the risk, economic situation and the safety of human health. The concept of risk assessment is multidisciplinary and it has been employed in many fields (e.g. economics; environmental sciences; including food safety). The Codex alimentarius commission is the international regulatory body for food, along with FAO and WHO, a joint expert committee (JEMRA) was formed to set up the framework and guidance documents for microbiological risk assessments applicable to global community.
Risk assessment is a scientific part of risk analysis. 'It has as its objective a characterization of the nature and likelihood of harm resulting from human exposure to agents in food. The characterization of risk typically contains both qualitative and quantitative information and is associated with a certain degree of scientific uncertainty'(WHO 2014). Further, the process of risk assessment can be divided into four parts: hazard identification, hazard characterization, exposure assessment and risk characterization. The predictive microbiology part will be discussed in more detail in this report and it actually falls under the exposure assessment.
Steps in risk assessment
It is the first step in a risk assessment process which is "the identification of biological, chemical, and physical agents capable of causing adverse health effects and which may be present in a particular food or group of foods." (CODEX 2004). The process is qualitative and the organisms of concern are selected according to the information of pathogen, type of food product, status of the host and the host'pathogen'environment (i.e. food here) interface (Lammerding 2001). Microbial hazards can be identified from literatures available online; or from epidemiological outbreak data, which points out the link between food and pathogen; sometimes expert opinion. Further the microbial pathogenicity, their mechanism of action through which it affects the host and duration or severity of disease, status of the population affected by the microorganisms, factors affecting growth and survival must be considered. A useful measure of severity of consequences is the disability adjusted life years (DALY) concept. To conclude, microorganisms and the susceptible population of concern should be taken into account and they will also be analyzed in the subsequent steps.
It is defined as 'The qualitative and/or quantitative evaluation of the nature of the adverse health effects associated with biological, chemical and physical agents which may be present in food. For chemical agents a dose'response assessment should be performed. For biological or physical agents a dose'response assessment should be performed if the data are obtainable' (CODEX 2004). The hazard characterization step involves description of the process of hazard characterization; process initiation; data collection and its evaluation; descriptive characterization; dose-response modelling; and analysis of the results (WHO 2003). Mathematical modelling can be used to determine dose-response in case of limited dataset where the descriptive and epidemiological information are not sufficient. While determining a dose response model, the whole disease process should be taken into account like for example, the exposure, infection, illness, sequelae ' recovery or death. Meanwhile, it is imperative to include the interaction of the food, host and pathogen. The plausibility of biological dose response model depends on three basic factors: threshold vs. non-threshold mechanisms; independent vs. synergistic action; and the particulate nature of inoculum. The threshold response in the literature is usually referred as the 'minimal infectious dose', which is the lowest number of microorganisms required to initiate an infection or disease. On the other hand, the non-threshold model approach uses the mathematical non zero probability of infection and illness. Thus, a single viable pathogenic microorganism can induce the infection (single-hit concept). The dose-response model gives the probability of illness depending on the number of microorganisms ingested. There are several models that are available in the literature which can be used according to the context, flexibility, assumptions and the data available. However, the commonly used dose-response models or hit-theory models are the exponential and the Beta-Poisson models (Ross 2014). The reason for the use of exponential model is the fact that is simple with fewer assumptions, whereas the Beta-Poisson includes the possibility of variability in infectivity. Both the models predict the probability of infection as a simple function of the dose of cells ingested. The probability of illness will not increase after certain upper limiting dose with the models used.
Figure 2: Dose vs. probability of illness relationship. The intercept of the model with the y-axis gives the 'r' value of the exponential model, which is the probability that one cell of the pathogen could cause illness (i.e., 1 in a million in the above figure). The dotted lines illustrate the calculation of the ID50, i.e., approximately 70 000 cells here.
Source: (Ross 2014)
Exposure assessment is 'the qualitative and/or quantitative evaluation of the likely intake of biological, chemical and physical agents via food as well as exposures from other sources if relevant' (CODEX 2004). In a microbiological risk assessment, the exposure assessment process has a certain degree of uncertainty and variability especially in a quantitative evaluation. It determines the probability of occurrence of microorganisms or their toxic metabolites in food at the time of consumption. Predictive microbiology can be very useful for the quantitative estimation of microbial growth, inactivation and survival of microorganisms in the food matrices. The microbial level during processing is highly variable and it is important to consider all sources of hazard entry into food product, to predict the microbial concentration at the consumer level. There are many other interaction factors should be also be considered like competing microbial flora, food chemistry, time, temperature, pressure, inactivation steps etc.. Another important point that should be noted here is the limit of detection. For instance, a test protocol at the end of the process steps could give false negative results, when there are too low levels of microorganisms which bring in uncertainty. To avoid uncertainty, more data and inputs along the farm-to-fork chain will be needed.
Figure 3: Conceptual model for exposure assessment, a Farm-to-fork risk assessment shows the influence of each stage of the process on Prevalence (P) & Concentration (C) of the pathogen in a food product.
Source: (Lammerding and Fazil 2000)
It is the final process which is defined by the (CODEX 2004) as 'the qualitative and/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 characterization and exposure assessment'. It utilizes the data and information from other components of risk assessment and supports the food safety decision making. For instance, by doing a sensitivity analysis, it is critical to picture which steps in a production process influence the concentration of pathogen in the final product. So that the input values can be altered according to the models assumed to get the food safety objective. Further, the uncertainty analysis calculates the total uncertainty of the model outputs and also the relative importance of the inputs. Thus it takes into account the gaps in data and knowledge (Morgan and Small 1992). The estimated risk can be expressed in many ways, for example: risk per serving, risk to an individual or to an entire population and sometimes even on a defined time interval, etc. In essence, risk is a quantitative concept where the ultimate aim of risk assessment is to clearly express the magnitude of risk or relative risk, in order to avoid misinterpretations when the risk is especially expressed in qualitative terms. In those cases, the conceptual model should be expressed in terms of mathematical equations to quantify the magnitude of the risk.
Figure 4: Influence diagram illustrating the 'conceptual model' for assessing risk of microbial foodborne illness from a factory-to-fork food supply chain. Shaded boxes indicate the variables which require data and whereas the variables that are derived from knowledge of other factors, including the final estimate of risk are shown in unshaded box. The relationships depicted by the arrows are defined and quantified by mathematical equations. The model is a form of 'process risk model' and at each stage in the factory-to-fork process the model calculates both the concentration of pathogen (or toxin) in the food and also the likelihood of the occurrence of pathogen in food.
Source: (Motarjemi 2013)
Classification of risk assessment
Qualitative risk assessment
The literature review serves as a good start for qualitative risk assessment, where the factors of risk are described on a non-numerical scale. However, a literature review alone will not be sufficient to make a qualitative risk assessment, a proper analysis of risk impact factors described qualitatively; for example ( more risk, less risk, no risk) risk ranking which should be clearly defined to avoid the confounding effect at measuring the magnitude of risk. Ideally, qualitative risk assessment gives an opening idea about the magnitude of risks and also it helps at deciding whether a detailed analysis will be required or not, thus a better knowledge on risk situation can be concluded. Qualitative risk assessment saves the data collection time and quantity of data collected for performing the subsequent quantitative risk assessment. Conducting a quantitative risk assessment is data and time demanding process.
Quantitative risk assessment
Quantitative risk assessments use mathematical models either deterministic or stochastic to estimate risk as a function of one or more inputs. Deterministic models use point estimates values, which mean the input values are just numerical numbers and the resulting output are again a single numerical value. The problem here in this type of approach is inputs are values with worst-case scenario. The output does not give any information that how likely the risk is about to happen. Thus the output value that is been generated with fixed inputs will be very extreme, but when the average values are calculated based on the mean values, then the corresponding extremes will not be taken into consideration, which actually might represent a group of highly susceptible population or infrequent with severe circumstances (Fazil 2005). An alternative solution to deterministic approach is probabilistic or stochastic quantitative risk assessment, which introduces the variability and uncertainty in the input parameters. The known inputs are described by probability distributions, depending on the available data. However, the output cannot be calculated easily from the probability distributions, thus to generate the output, a Monte Carlo simulation has to be performed using a computer software which will be discussed later in the subsequent chapters.
Data maybe available sometimes and sometimes very fewer data could be available. As the risk assessment is very data demanding, sometimes precise and good quality data will be required for inputs. In reality the data available may be very few and even of poor quality, in those cases, the data constraints should always be clearly acknowledged in the risk assessment report. When the data is not available, additional data should be collected. If it is not possible, an expert opinion on the risk profile can be sought. For poor quality data, the risk assessor should make the data is validated, and it is reliable to be used as inputs in the risk assessment. Moreover in some circumstances, it is better to perform a meta-analysis which will reduce the uncertainty in data through the systematic review process.
Although predictive microbiology has its origin from the beginning of 20th century, its applications have been only recognized in the recent decades. The first use of predictive modelling is the use of log-linear models proposed by (Esty and Meyer 1922) for describing the bacterial death kinetic by heat. This model has a wide application in the canning industry. The field of predictive modelling has advanced with the advent of computer softwares in the early 1980s. At present predictive microbiology has taken its own place in the scientific community, with new publications, scientific conferences, training programs. Even though the field of predictive microbiology is evolving with advanced researches, the basic aim of predictive microbiology is its application in improving food safety and quality. There is a growing potential for predictive microbiology in the food industry but however their application in food industry can be broadly grouped into three main activities: Product innovation, operational support and incidental support (Membr?? and Lambert 2008). Besides, predictive modelling serves to be a valuable tool in risk assessment and also at transferring risk management concepts into practical guidelines (Gorris 2005; Membre, Bassett et al. 2007). There are also some recent applications of predictive modelling to reduce the impact of climate change and seasonal variations on the safety and quality of foods (Janevska, Gospavic et al. 2010).
Models' framework and types
Predictive microbiology is a scientific branch of the food microbiology field intended to quantitatively assess or predict the microbial behavior in food environments by using appropriate mathematical models. Mathematical models use mathematical equations to represent real system in a simplified way based on certain significant parameters.
Source: (P??rez-Rodr??guez and Valero 2013)
In the above equation, Y is the dependent variable or response variable; X1, X2, and X3 are explanatory or independent variables; and ??1, ??2, and ??3 are the regression coefficients obtained from a regression based method and the ?? is the error term which represents the variability.
Traditionally, models have been classified into two types depending on the basis of information used to create the model (McMeekin and Ross 2002). They are mechanistic models and empirical models. Besides, the models describing kinetic process are classified as primary, secondary, and tertiary models. The primary models describe the change in concentration over time (growth or death curve), whereas secondary models incorporate the kinetic parameters derived from the primary models to environmental parameters. The tertiary models are implementations of both primary and secondary model on a computer tool. The tertiary models will be discussed in detail in the subsequent chapter.
Primary models are intended to estimate the growth or inactivation of microorganisms over the function of time. The growth curve of microorganisms shows three distinct phases which are lag phase, exponential phase and a stationary phase whereas the inactivation of microorganisms exhibits a decline phase as shown in the Figure 7.
Figure 7: Representation of microbial 4-phase kinetics over a period of time
Lag phase or the adaptation period is the time required by the microorganisms to adjust themselves to the new environment and initiate exponential growth (Buchanan and Klawitter 1991). Then, microorganisms grow exponentially (exponential phase) until they reach a 'plateau' (stationary phase). When the concentration of nutrients or the physiological state of cells is decreasing, the microbial population starts to decline (decline phase). It is important to note the difference between 'survival' and 'inactivation' models. In survival, there is no intention to kill the microbial population but no growth is allowed; whereas inactivation is specifically implemented to destroy the microbial population by certain log reductions in a specific food. In primary modelling, models are fitted to microbiological data by use of regression and there are varieties of models available that can be chosen to fit the data.
Some of the available softwares that can perform Monte Carlo analysis are: @Risk (Palisade Corp., Ithaca, NY), Crystal Ball (Oracle, Redwood Shores, CA), and Analytica (Lumina Decision Systems, Inc., Los Gatos, CA).
Monte Carlo analysis
Monte Carlo simulation does random sampling of each probability distribution within a model over hundreds or thousands of times, which depends upon the numbers of iteration user wants. It actually produces a new scenario for each iteration. In fact, a new point-estimate will be generated for each parameter within the model at each iteration, and the result is recorded. The process is then repeated until the output is sufficiently stable, thereby satisfying the convergence criteria (Vose 2008).
The simulation of a model using Monte Carlo analysis allows the model to be used for in depth analysis of more highly influencing input variables. For instance, a sensitivity analysis on the model reveals the variable which is highly correlated within the model. Also, Monte Carlo simulation helps with the evaluation of uncertainty and variability in risk assessment. In some advanced statistical software tools, a second order Monte Carlo simulation can be able to differentiate uncertainty and variability. Nevertheless, being a powerful tool, the user should be first aware of Monte Carlo methods in risk assessment (Vose 2008); the basic knowledge of statistical distributions and the appropriate models to use. The Monte Carlo simulation techniques may not be appropriate for all applications (Ferson 1996). But still, Monte Carlo analysis is a reliable technique used in predictive microbiology.
Buchanan, R. and L. Klawitter (1991). "Effect of temperature history on the growth of< i> Listeria monocytogenes</i> Scott A at refrigeration temperatures." International journal of food microbiology 12(2): 235-245.
CODEX (2004). "CODEX ALIMENTARIUS COMMISSION - PROCEDURAL MANUAL Fourteenth edition." Fourteenth Edition. Retrieved June 10, 2014, 2014, from http://www.fao.org/docrep/007/y5817e/y5817e00.htm.
Esty, J. and K. Meyer (1922). "The heat resistance of the spores of B. botulinus and allied anaerobes. XI." Journal of infectious diseases 31(6): 650-663.
Fazil, A. M. (2005). A primer on risk assessment modelling: focus on seafood products, Food & Agriculture Org.
Ferson, S. (1996). "What Monte Carlo methods cannot do." Human and Ecological Risk Assessment: An International Journal 2(4): 990-1007.
Gorris, L. G. (2005). "Food safety objective: an integral part of food chain management." Food Control 16(9): 801-809.
Janevska, D. P., R. Gospavic, et al. (2010). "Application of a HACCP'QMRA approach for managing the impact of climate change on food quality and safety." Food research international 43(7): 1915-1924.
Lammerding, A., Fazil, A. & Paoli (2001). Compendium of Methods for the Microbiological Examination of Foods, American Public Health Association.
Lammerding, A. M. and A. Fazil (2000). "Hazard identification and exposure assessment for microbial food safety risk assessment." International journal of food microbiology 58(3): 147-157.
McMeekin, T. and T. Ross (2002). "Predictive microbiology: providing a knowledge-based framework for change management." International journal of food microbiology 78(1): 133-153.
Membre, J.-M., J. Bassett, et al. (2007). "Applying the food safety objective and related standards to thermal inactivation of Salmonella in poultry meat." Journal of Food Protection?? 70(9): 2036-2044.
Membr??, J.-M. and R. J. W. Lambert (2008). "Application of predictive modelling techniques in industry: From food design up to risk assessment." International journal of food microbiology 128(1): 10-15.
Morgan, M. G. and M. Small (1992). Uncertainty: a guide to dealing with uncertainty in quantitative risk and policy analysis, Cambridge University Press.
Motarjemi, Y. (2013). Encyclopedia of Food Safety, Academic Press.
P??rez-Rodr??guez, F. and A. Valero (2013). Predictive microbiology in foods, Springer.
Ross, T. (2014). Risk Analysis: Risk Assessment: Microbiological Hazards. Encyclopedia of Food Safety. Y. Motarjemi. Waltham, Academic Press: 80-92.
Vose, D. (2008). Risk Analysis: A Quantitative Guide, John Wiley & Sons.
WHO (2003). Hazard characterization for pathogens in food and water: guidelines, Food & Agriculture Org.
WHO (2014). "About microbiological risk assessment (MRA) in food." Retrieved June 6, 2104, 2014, from http://www.who.int/foodsafety/micro/about_mra/en/.
Source: Essay UK - http://www.essay.uk.com/free-essays/science/microbiology-food-industry.php