Foodborne Disease and Public Health: Summary of an Iranian-American Workshop (2008)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Applying Risk Assessment Methods to Food Microbiology Moderator: Karl R. Matthews I n deciding how to deal with problems relating to foodborne illness, two useful sets of tools are risk-assessment methods and methods for weighing trade-offs between competing risks. The workshop had pre- sentations on each, with Robert L. Buchanan discussing risk assessment methods and Richard A. Forshee describing the evaluation of risk–risk trade-offs. This chapter covers both their presentations and then ends with highlights of the subsequent discussion. RISK ASSESSMENT METHODS Presenter: Robert Buchanan Risk Assessment Overview Risk assessment is a method for applying pertinent scientific data to make risk-management decisions. High-quality risk assessments identify clearly what is and what is not known; they characterize how well the data are known, taking into account the variability and uncertainty in the data; and they are sufficiently transparent to reveal possible biases or errors in reasoning. Such risk assessments provide a powerful tool for estimating the effects of potential options and food safety standards on the risk that is being addressed. Risk assessments can be grouped into four general classes based on 37 

38 FOODBORNE DISEASE AND PUBLIC HEALTH the extent to which the treatment of the data is quantitative. These four classes are 1. formal expert elicitations; 2. qualitative assessments that use non-numerical descriptors; 3. semi-quantitative assessments that combine quantitative data and non-numerical descriptors; and 4. quantitative assessments in which mathematical modeling uses a large amount of scientific data. Although semi-quantitative assessments are the most widely used of these classes, Buchanan focused on quantitative methods. Quantitative Microbial Risk Assessment A quantitative microbial risk assessment produces a mathematical statement that is based on the cumulative probabilities of certain adverse events happening following an exposure to a hazardous agent. The result of such a risk assessment includes not only an estimate of the risk but also an estimate of the attendant uncertainties. In addition to the quantitative factors, the assessment will also generally consider qualitative factors in its discussion of risk and uncertainty. Quantitative microbial risk assessments are generally of two basic types: deterministic and probabilistic. Deterministic assessments use point estimates to describe risk at a certain level of exposure (for example, the mean risk or the 95th or 99th percentile of risk). A disadvantage of this approach is that it may overestimate risk if one tries to combine factors. Probabilistic assessments use entire distributions and require advanced modeling techniques, such as those made possible with Monte Carlo simulation software. Using a  Monte Carlo simulation, for example, one could determine the impact of changing the refrigerator temperature on the growth rate of a pathogenic microorganism and, therefore, on the final risk posed by an organism in a food. The advantages of probabilistic models include more accurate results, the ability to modify the model easily to incorporate new data, the abil- ity to produce “what-if” scenarios, and the ability to evaluate the effects of potential actions to mitigate risk. In a “what-if” scenario, one can make substitutions in the model that make it possible to examine how making a specific change would affect an outcome. One disadvantage of probabilistic assessments is that they may be difficult for risk managers to interpret. Microbial risk assessment could be used in a variety of ways by the U.S. Food and Drug Administration (FDA). Some possibilities are 

APPLYING RISK ASSESSMENT METHODS TO FOOD MICROBIOLOGY 39 • setting priorities; • identifying steps that are major contributions to risk; • evaluating the effectiveness of potential control measures, stan- dards, and criteria; • evaluating the contribution of compliance to risk management; • determining subpopulations at increased risk; and • assessing uncertainty and variability. Types of Microbiological Risk Assessments Risk Ranking Risk ranking is a technique used to compare relative risks. It can be used to compare different foods for a single hazard, different hazards for a single food, and, potentially, multiple hazards and multiple foods. In general, the purpose of risk ranking is to help establish priorities, as in developing a budget. Recently, for example, a quantitative assessment was conducted on the relative risk to public health from Listeria monocy- togenes in 23 classes of ready-to-eat foods, such as delicatessen meats and certain cheeses, in the United States (DHHS/USDA, 2003). As part of that assessment, the assessors estimated the risk per serving and the predicted number of cases of listeriosis per year for the total U.S. population. The results are being used to focus inspection activities, surveillance activities, educational strategies, research, and new risk assessments. Product/Pathogen Pathway Analysis Product/pathogen pathway analysis is a technique that can be used to examine factors that contribute to risk over the course of a particular segment of the path from farm to table. It makes it possible to quantify the importance of contamination sources, the effectiveness of interven- tions, the comparative effectiveness of different control measures, the likely effect of performance criteria or standards, and the importance of complying with the criteria. One example of product/pathogen pathway analysis is described in a recent report on Vibrio parahaemolyticus in raw oysters (CFSAN, 2005). The analysis assessed the importance of various possible contributing factors, such as the concentration of the organism in the environment, the length of time the oysters went without refrigera- tion, the number of grams of oysters consumed, the region of the United States, and the season. Figure 4-1 illustrates how varying the standard for the concentration of the organism allowed in oysters would affect the percentage of illnesses prevented and the percentage of the oyster harvest that would be lost. This type of information is valuable to deci- sion makers. 

40 FOODBORNE DISEASE AND PUBLIC HEALTH 100% 100% Percentage of illness prevented Percentage of oyster harvest lost 80% 80% Illness 60% 60% 40% 40% 20% Lost oyster harvest 20% 0% 0% 100,000 10,000 1,000 100 Total Vibrio parahaemolyticus/gram of raw oyster FIGURE 4-1  Effects of the concentration of Vibrio parahaemolyticus in oysters on the percentage of illness prevented and the percentage of the oyster harvest that would be lost. SOURCE: CFSAN, 2005. fig 4-1 Geographical Risk Assessment Geographical risk assessment, sometimes called epidemic risk assess- ment, estimates the effect of introducing a new hazard into a geographical area. This type of risk assessment involves modeling the factors that affect the introduction of a disease agent, the ability of an infection to be self- sustaining in a population, the factors that affect the rate of dissemina- tion of the infection, and the degree to which mitigation activities disrupt the dissemination. An example of such a geographical risk assessment is the Harvard bovine spongiform encephalopathy (BSE) risk assessment (http://www.fsis.usda.gov/Science/Risk_Assessments/index.asp#BSE). It analyzed the potential spread of BSE among cattle if BSE were to be introduced into the United States and the resulting potential for humans to be exposed to contaminated material. Risk–Risk Assessments Risk–risk assessments are performed when reducing the risk of one hazard can be expected to increase the risk of another. One current exam- ple is the FDA’s examination of the risk posed by the methyl mercury in 

APPLYING RISK ASSESSMENT METHODS TO FOOD MICROBIOLOGY 41 fish compared with the risks caused by decreased intake of omega-3 fatty acids if fish consumption is curtailed. To date, risk–risk assessments that involve food microbiology have been qualitative rather than quantitative. Among the challenges of risk–risk assessments are finding a common end point for comparison and redefining a benefit as a risk. Closing Remarks All risk assessments have finite life spans. As new data emerge, new assessments are needed. Many useful resources are available to provide guidance in conducting high-quality microbial risk assessments. A num- ber of them are listed in Box 4-1. BOX 4-1 MICROBIAL RISK ASSESSMENT RESOURCES Publications 1. Brown, M., and M. Stringer. 2002. Microbiological risk assessment in food processing. Cambridge, UK: Woodhead Publishing, Ltd. 2. Center for Food Safety and Applied Nutrition, Food and Drug Administration. 2002. Initiation and conduct of all “major” risk assessments within a risk analysis framework. A Report by the CFSAN Risk Analysis Working Group. http://www. cfsan.fda.gov/~dms/rafw-toc.html (accessed February 5, 2008). 3. Haas, C. N., J. B. Rose, and C. P. Gerba. 1999. Quantitative microbial risk assessment. New York: Wiley & Sons. 4. Schaffner, D. W., ed. 2008. Microbial risk analysis of foods. Washington, DC: ASM Press. 5. Vose, D. 2000. Quantitative Risk Analysis, 2nd edition. New York: Wiley & Sons. Websites 6. Food and Agriculture Organization/World Health Organization Joint Expert Meetings on Microbiological Risk Assessment: http://www.codexalimentarius.net/ web/jemra.jsp 7. Joint Food and Agriculture Organization/World Health Organization Food Standards Program. Codex Alimentarius: http://www.codexalimentarius.net/web/ index_en.jsp 8. Joint Institute for Food Safety and Nutrition • Risk analysis training: http://www.professionalstudies.umd.edu/food_ safety/ • Risk Assessment Clearinghouse: http://www.foodrisk.org/ 

42 FOODBORNE DISEASE AND PUBLIC HEALTH EVALUATION OF RISK–RISK TRADE-OFFS Presenter: Richard A. Forshee A food or nutrient may increase some risks and decrease others. In his presentation, Forshee described an integrated approach for comparing the benefits and risks of consuming a nutrient or food, including a proof of the concept. He also distinguished between risk-based and safety-based approaches. Application of an Integrated Approach A review by Cohen and colleagues provides an example of an inte- grated approach to risk–risk assessment as applied to fish consumption (Cohen et al., 2005). The analysis looked at both the adverse effect that methyl mercury may have on the development of the brain and nervous system and the benefits that n-3 polyunsaturated fatty acids have in reducing the risk of coronary heart disease among older men and women. In doing so, the analysis explicitly recognized that the risks and benefits of fish consumption may vary across subpopulations. As described in the paper, the risk–risk assessment examines expo- sures, dose–response relationships, and both positive and negative health endpoints. After converting those health endpoints to a common scale, Quality-Adjusted Life Years (QALYs), Cohen and coworkers then esti- mated the QALYs gained under five different scenarios. The scenarios differ in the population and fish consumption levels. As shown in Figure 4-2, the most favorable overall results in terms of QALYs gained occur in scenarios 4 and 5, both of which involve a 50 percent increase in total fish consumption. The other scenarios, which produced gains of fewer QALYs, called for women of childbearing age to reduce fish consumption or to switch to fish low in methyl mercury or else for the entire population to reduce fish consumption. This type of analysis can help policy makers evaluate specific risks and examine the effects of different actions by the public on public health. Another analysis of the risks and benefits of fish consumption (Ponce et al., 2000) used a somewhat different method and produced curves showing QALYs as a function of fish consumption measured in grams per day. That analysis found that the only situation in which fish consumption   The Quality-Adjusted Life Year (QALY) is an index that combines the quantity and qual- ity of life. It assigns a value of 1 to one year of “perfect” health-life expectancy and assigns a value of less than 1 to one year of less than perfect life expectancy. Death is assigned a value of 0. 

APPLYING RISK ASSESSMENT METHODS TO FOOD MICROBIOLOGY 43 Annual QALYs gained Scenario FIGURE 4-2  Annual Quality-Adjusted Life Years (QALYs) gained by type and amount of fish intake. Key to scenarios: (1) Women of childbearing age switch to fish low in methyl mercury (≤0.13 μg of methyl mercury/g of fish); (2) women of childbearing age reduce total fish consumption by 17 percent; (3) entire popula- tion reduces total fish consumption by 17 percent; (4) women not of childbearing age and all men increase total fish consumption by 50 percent; and (5) all women and all men increase total fish consumption by 50 percent. SOURCE: Cohen et al., 2006. fig 4-2 "Fixed image" results in a reduction in QALYs is when people eat large amounts of fish that are high in methyl mercury. Proof of Concept Forshee used data and approaches from the article by Cohen et al. (2005) to try to provide a proof of concept concerning coronary heart dis- ease and intelligence quotient (IQ, which is one measure of brain devel- opment) as related to fish consumption measured in servings per week. Using data and approaches taken from the article by Cohen and cowork- ers, he examined the QALY loss for women ages 15 years and older as a function of fish consumption measured in servings per week. (No risk of harm was observed in men.) By combining the lost QALYs from coronary heart disease and from loss of IQ points, Forshee found that the great- est benefit (reduction in QALY loss) occurs when fish consumption is increased from zero to one serving per week. At higher intakes, there is much more uncertainty about the effect of additional fish consumption on QALY loss. Forshee noted that his analysis has a number of limitations, includ- 

44 FOODBORNE DISEASE AND PUBLIC HEALTH ing those of the Cohen article, the fact that only two health endpoints are being considered, the fact that the analysis assumes a constant level of fish consumption over a lifetime, and the large width of the confidence intervals (because he did not use a Monte Carlo approach in estimating them). The approach described above could be useful in examining a variety of risks, such as those related to alcohol consumption, sunshine exposure, and food fortification (which might increase the intake of nutrients above a safe level). Although Forshee believes that the approach holds much promise, he said that it also has some limitations. QALY is far from being accepted as a common metric, for example, and some people do not think it is appropriate to use a common metric for diseases that affect children and those that affect the elderly. Furthermore, the uncertainty that is iden- tified in the calculations can make decision making difficult. Nonetheless, the approach states the uncertainty more explicitly than is typical in the risk-assessment arena, which can be seen as an advantage. One practical limitation is that this type of approach may not be usable in regulatory matters without changes in regulatory law. Also, importantly, risks have attributes that affect how people perceive them. For example, does the risk stem from a natural component of food or from a contaminant? Is it a familiar risk or a new risk? All of these factors can affect risk communica- tion and management. Risk Versus Safety Many regulatory approaches are based on establishing de minimis safety levels for exposure or consumption—intake levels that are believed to carry no risk. For nutrients, these levels include such measures as Toler- able Upper Intake Levels and, for toxic substances, the Reference Dose. These are discrete values with generous safety margins. By contrast, risk-based approaches evaluate the probability and con- sequences across the entire range of exposure and seek the lowest risk for the population being considered. For some substances, there is a generous safe range of consumption. For other substances, as illustrated in Figure 4-3, no level of consumption may be completely safe. Using the informa- tion depicted in Figure 4-3 would lead to a different recommendation than would be obtained using a traditional safety approach. Closing Remarks The consumption of many foods and nutrients is associated with mul- tiple risks and benefits. By applying analyses that integrate multiple risks 

APPLYING RISK ASSESSMENT METHODS TO FOOD MICROBIOLOGY 45 8000 Risk 1 Expected Loss Risk 2 Expected Loss Total Expected Loss Expected Loss 6000 4000 2000 0 0 20 40 60 80 100 Consumption FIGURE 4-3  Hypothetical graph depicting the expected loss from competing risks, individually and combined. This graph could apply to a substance that poses a risk of adverse health effects if intake is low and of toxic effects if intake is high. SOURCE: Richard A. Forshee, Center for Food, Nutrition, and Agriculture Policy, University of Maryland. fig 4-3 and benefits, one can improve approaches to regulatory policy, dietary guidance, and the public health. DISCUSSION Moderator: Karl R. Matthews Any risk assessment involves a challenge in balancing the need for a parsimonious model that can be explained readily against a compre- hensive model that includes all the possible risks and alternatives. It was noted, for example, that changing the intake of one dietary component is likely to change the intake of other components as well; a model that did not take this into account might be simpler but less informative. As another example, focusing solely on V. parahaemolyticus in oysters may be insufficient if there is also a risk of norovirus infection. Dr. Forshee men- tioned Einstein’s view that one’s model needs to be as simple as possible but as complex as necessary to address the question.