Managing Risk in a Zero Tolerance World: International Impact of Risk Assessment Robert L. Buchanan Department of Nutrition and Food Science
Presentation Historical Perspective Consideration of Dose-Response Relations Estimating Exposure Importance of Distributions Concluding Remarks
Historical Perspective 1995: WHO/FAO Consultation on "Application of Risk Analysis to Food Standards Issues," o Risk assessment techniques for microbial food safety issues are not likely to be available in the near term o Microbiological food safety concerns were too complex to be amenable to use of formal risk assessment techniques
Historical Perspective 1998-2003: Microbial risk assessments become a tool for evaluating food safety risk management options o Salmonella enteritidis - eggs: USDA, 1998 o Listeria monocytogenes - RTE foods: FDA/USDA, 1999-2003 o Vibrio parahaemolyticus - raw shellfish: FDA, 1999-2003 o Escherichia coli O157:H7 - USDA/FDA, 1999-2004
Historical Perspective Strong international incentives because of the: o Signing of the WTO SPS & TBT Agreements o Drive for more objective determination of Equivalence Microbiological criteria o Codex Alimentarius adopts risk analysis framework
Available Risk Assessments During the past 15 years there have been a series of microbial food safety risk assessments developed by national governments, FAO/WHO- JEMRA, EFSA, and academic institutions Have covered an array of microbiological hazards o Salmonella enterica o Enterohemorrhagic Escherichia coli o Campylobacter spp. o Cronobacter spp. o Vibrio spp. o Listeria monocytogenes o Toxoplasma gondii o Foodborne viruses o Foodborne parasites o Mycotoxigenic fungi o More Have considered an array of foods
Impact of Quantitative Microbial Risk Assessment The emergence of QMRA has moved food safety risk management from largely a qualitative consideration of hazards to a quantitative consideration of risk o More quantitative approach to dose-response relations o Improvement in exposure assessments o Use of scenario analyses for evaluation of prevention and intervention strategies o Formal consideration of uncertainty o Use of sensitivity analysis techniques to quantify the relative importance of risk factors
Microbial Dose-Response Relations: Ending the Search for Minimum Infectious Doses
Zero Tolerance Used as a means of expressing an attitude or level of concern for the importance of safeguarding the public health Concept of a zero tolerance emerged as a result of the inherent inability to establish thresholds for infectious and toxicoinfectious microorganisms Probabilistic nature of infectious processes Concept of independent action
Microbial Dose-Response Relations Toxigenic microorganisms o Acute chemical toxins (C. botulinum toxin, S. aureus enterotoxin) - Threshold models o Carcinogens/ mutagens (e.g., aflatoxin B 1 Nonthreshold models Infectious and toxicoinfectious microorganisms o Most microbial foodborne diseases Non-threshold models
Dose-Response Assessment Probability of Illness 1.0 0.8 0.6 0.4 0.2 0.0 0 4 8 12 2 6 10 Log (Pathogen Cells Ingested) Log(Probability of Illnes 0-2 -4-6 -8-10 0 4 8 12 2 6 10 Log (Pathogen Cells Ingested) A single cell has a definable probability of producing an infection Probability increases as the number of cells ingested increases Use non-threshold models that are linear or log linear in the low dose regions
If develop dose-response model for entire population, the variability in susceptibility represents a large uncertainty o Range from extremely high risk individuals to those who are totally immune % Cases Dose-Response Assessment One way around this is to develop separate doseresponse models for specific subpopulations % Cases 100 90 80 70 60 50 40 30 20 10 0 60 50 40 30 20 10 0 0 0.5 1 Relative Log(Dose) 0 0.5 1 Relative Dose
Log (Risk per serving) Dose-Response Assessment Dose-response curves based on French relative susceptiblity data 0.0-5.0 0 5 10 Transplant AIDS Dialysis Cancer--Pulmonary, etc Bladder -10.0 Gynecological Diabetes, hep. Alcohol Over 65 years -15.0 Log (L. monocytogenes per serving) Less than 65, no other
Dose-Response Implications Death of the concept of minimum infectious dose Encouraging alternative strategies for individuals with increased susceptibility Criticality of informed decisions by consumers based on personal susceptibility Appreciation of the relative infectivity of different pathogens
Exposure Assessment: What Did the Consumer Actually Eat?
Exposure Assessment The data needed for QMRAs is the number of pathogenic microorganisms ingested The data available for risk assessments is almost entirely obtained at some point earlier in the food chain Requires that risk assessors are able to consider the changes in microbial population densities between data points and consumption o These changes can overwhelmingly impact levels to which consumers are exposed, e.g., temperature abuse, cooking o Additionally need to estimate the physiological status of the pathogen, e.g., the case of Vibrio spp.
Exposure Assessment Predicting what is actually being consumed is achieved largely through the use of predictive microbiology modeling o QMRAs would not have been possible if there had not been a 10-year international effort to describe microbial behavior in foods mathematically However, the limiting factor right now is availability of consumer phase models o Consumer behaviors are highly variable and can dramatically influence exposures o Anticipating these behaviors and building in safeguards is a critical component of designing safe foods
DM P SS HFD PM FR RS DFS SRC DS PC HC Total Cases Listeriosis per Serving (log scale) Exposure Assessment The increased use of QMRA has also help quantify more objectively that not all foods pose the same risks, particularly if growth of the pathogen is a primary risk factor -6-7 -8-9 -10-11 -12-13 -14-15 -16
Decreased Risk Per Serving Exposure Assessment Decreased Risk Per Annum A and B C and D E Very High Risk Deli Meats Frankfurters (not reheated) High Risk Pátê and Meat Spreads Unpasteurized Fluid Milk Smoked Seafood Moderate Risk No food categories 1 High Risk High Fat and Other Dairy Products Pasteurized Fluid Milk Soft Unripened Cheese Moderate Risk Cooked RTE Crustaceans Moderate Risk No food categories 2 Moderate Risk No food categories Moderate Risk Deli Salads Dry/Semi-dry Fermented Sausages Frankfurters (reheated) Fresh Soft Cheese Fruits Semi-soft Cheese Soft Ripened Cheese Vegetables Low Risk Preserved Fish Raw Seafood 3 Moderate Risk No food categories Low Risk No food categories Very Low Risk Cultured Milk Products Hard Cheese Ice Cream and Frozen Dairy Products Processed Cheese 4
Importance of Distributions
Distributions Blue Bell Outbreak: The Perfect Storm o Listeria monocytogenes outbreak involving brand of ice cream o The ice cream was used to make milk shakes for hospitalized patients o Ice cream was found to have high frequency of low level contamination (1 to 10 CFU/g in a high % of samples o Re-enforced concern about interpretation of systems failures
Understanding Root Causes When considering how stringent to make a food safety system, need to consider two different types of risks o Risk of Non-compliance: Risk that a proposed standard will not be met Risk that contamination exceeds standard o Ineffective manufacturing o Growth of pathogen Risk that pathogen introduced after manufacturing o Residual Risk: The risk that still exists when food safety system working as intended Varies greatly among different pathogens A zero tolerance assumes that any positive test result is an indication of non-compliance
Understanding Root Causes The Blue Bell outbreak indicated a need to consider a third root cause o Compliance risk: Reliability error Likely associated with outbreak o Residual risk: Residual error while system under control Likely associated with sporadic cases o Systemic GHP risk: Reliability error Likely associated with outbreak Distributions count when developing standards
Considering Risk Type In today s world of genetic fingerprinting, advanced epidemiological methods, and large batch sizes, need to consider all three types of risks Example: Listeria monocytogenes in a readyto-eat food o 10 5 CFU/g in 10% of servings o 10 CFU/g in 1% of servings, diverse genotypes o 50 CFU/g in 90% of servings, single genotype Assume serving size of 100 g and a total number of servings of 100,000,000 and all are consumed by high risk consumers
Probability density Scenarios #3 #2 PO #1 Mean Log(CFU/g)
Considering Risk Type Predicted number of cases per 100,000,000 servings using the dose-response curve for susceptible individuals from the FAO/WHO risk assessment (2004) o Scenario #1: 585 o Scenario #2: 0.00585 o Scenario #3: 2.6 Implies that standard should consider the distribution Listeria in foods to control systemic GHP errors Further implies switching to 3-class sampling plan would be a better testing option
Example of 3-Class Sampling Plan Sampling Plan o o o M: 1.0 Log(CFU/g) m: -0.5 Log(CFU/g) n: 5 samples o C: 1 positive in range > -0.5 and < 1.0 o Standard Deviation: 0.8 o Pr = 95% m 3-Class Plan 0.7 0.6 0.5 0.4 0.3 M Characteristics of Distribution o Median: Log(cfu/g) = -0.2 0.2 0.1 0-2 -1 0 1 2-0.1
Verifying Zero by Testing Verify adherence to a zero tolerance must operationalize zero by specifying a sampling plan and testing protocol to verify compliance This establishes a non-zero value based on sensitivity of the sampling plan One is effectively making a non-transparent risk management decision that establishes a non-zero tolerance
Concluding Remarks Quantitative microbiological risk assessments are: o Forcing a more objective consideration of the science underlying managing microbiological food safety risks o Helping to identify where along the food chain are the greatest vulnerabilities and opportunities for control o Forcing realization that implementation of Zerobased program requires adoption of a non-zero approach o Providing tools to more effectively design and implement food safety risk management programs based on risk o Forcing food safety managers to make hard decisions and find new ways of communicating limitations
Concluding Remarks The emerging scientific consensus of the nonthreshold nature of infectious and toxicoinfectious microorganisms is requiring a reevaluation of management approaches that have been traditionally based on the concept of minimum infectious dose Need to learn lessons acquired from other classes of non-threshold hazards on how to: o Manage risks o Design control strategies o Set standards and establish performance metrics o Communication risks to consumers