Gordon C. Ashton Memorial Lecture System s and Risk Modelling for Food Safety Decision Making Aamir Fazil Public Health Risk Sciences Division, LFZ, PHAC
Risk • Risk is a function of both the probability and impact of an event • Dependent on who is assessing the risk Aamir Fazil aamir.fazil@phac-aspc.gc.ca
History • The mastery of risk has been suggested as the step man took in moving into modern times “Understanding that the future is more than fate and the whim of the gods” Bernstein, 1996 Aamir Fazil aamir.fazil@phac-aspc.gc.ca
History Took someone who threw a lot of dice to finally begin understanding probability and risk Girolamo Cardano (1501–1576), Mathematician & Physician Su m Pr o ba bi l i t y one of first clinical descriptions of � 2 1/ 36 Typhoid Fever 3 2/ 36 4 3/ 36 5 4/ 36 6 5/ 36 “Liber de ludo aleae” written in the 1560s � 7 6/ 36 � first systematic treatment of probability, as 8 5/ 36 well as a section on effective cheating methods 9 4/ 36 10 3/ 36 Probability of rolling various dice combinations � 11 2/ 36 12 1/ 36 Aamir Fazil aamir.fazil@phac-aspc.gc.ca
History • Blaise Pascal (1623–1662) and Pierre de Fermat (1601 ‐ 1665) • Established the foundations of Probability • Determined the ability to forecast uncertain future outcomes Aamir Fazil aamir.fazil@phac-aspc.gc.ca
History • 1700’s – Mathematicians devising tables of life expectancies – English government financing itself through sale of life annuities • 1750’s – Marine insurance flourishing as a business Aamir Fazil aamir.fazil@phac-aspc.gc.ca
History • 1750 ‐ 1770 – Smallpox in France – To vaccinate or not ? – Daniel Bernoulli calculated risks of taking live vaccine vs. taking your chances – Odds • 1 in 7 dying of smallpox • 1 in 200 dying of vaccine – Pretty bad either way, better to take it Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Risk Modelling • Why do we do risk modelling – Estimate the risk • Regulation • Acceptability – Gain an understanding of the system • Appropriate mitigation • Research direction Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Risk Modelling • Research • Risk Mgmt Decisions Risk Modelling • Epidemiology • Policy • Link: Research / Data and Decisions Aamir Fazil aamir.fazil@phac-aspc.gc.ca
System risk modeling System • Food production system • Food production system • Health system • Health system • Farm system • Farm system • Farm to consumption system • Farm to consumption system Experiment with the Experiment with the Experiment with a Experiment with a actual system actual system model of system model of system Physical Model Physical Model Mathematical Model Mathematical Model Analytical Solution Analytical Solution Simulation Solution Simulation Solution Adapted from System Modelling Law & Kelton Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Modeling Approaches “All models are wrong, some are useful” • Essence of why we model a system: – Not to create a perfect representation – To create a tool that will provide insight into the system Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Risk Modelling Contribution to Decision Making • Link between contamination and public health • How much of a public health issue (1) Set Targets (1) Set Targets • Public health impact of potential reductions • What parts of the system influence risk most • What parts of the system do we not know (2) Focus Attention (2) Focus Attention enough about • What can we do to reduce the risk • What options do we potentially have to work with (3) Formulate Strategy (3) Formulate Strategy • How much of an impact could a strategy have • What potential cautions should we consider (4) Test Strategy (4) Test Strategy Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Risk Modelling Contribution to Decision Making • Link between contamination and public health • How much of a public health issue (1) Set Targets (1) Set Targets • Public health impact of potential reductions (2) Focus Attention (3) Formulate Strategy (4) Test Strategy Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(1) Set Targets: Model Introduction • Salmonella & Campylobacter human health impact model • Objective: Construct human health impact model – Translate pathogen prevalence levels and reductions to health impact – Translate pathogen prevalence levels and reductions to cost and savings Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(1) Set Targets: Model Overview Chicken Chicken Contamination Contamination Rate Rate Reduced Reduced Volume of Volume of contaminated volume contaminated volume Contaminated Contaminated of chicken of chicken Chicken in Market Chicken in Market Illness Illness Volume of Volume of Chicken in Chicken in pDR pDR Market Market Future human cases Future human cases Probability of illness per Probability of illness per Severe Severe resulting from resulting from volume of contaminated volume of contaminated Illness Illness chicken chicken Reported Reported chicken in market chicken in market Illness Rates Illness Rates Morbidity Morbidity Actual Illness Rate Actual Illness Rate Actual Overall Actual Overall Attributable to Attributable to Illness Rate Illness Rate Cost of illness Cost of illness Chicken Chicken Underreporting Underreporting Rates Rates Attribution Attribution of Illness to of Illness to Chicken Chicken Cost per case Cost per case Aamir Fazil aamir.fazil@phac-aspc.gc.ca
• Reported Illness Rates Underreporting Rates • Contamination Rates • Chicken in Market • Canadian Population • Cost per Case • Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(1) Set Targets: Model Objective • Primary purpose of model – Explore impact of reduction in contamination rates to: • Public health outcomes • Cost savings – Outcomes are a function of attribution Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(1) Set Targets: Salmonella Cost Contour 1 0.9 Target of 20% Target of 20% 0.8 reduction reduction $450 - $500 0.7 $400 - $450 0.6 $350 - $400 Attribution Attribution Attribution $43M $300 - $350 0.5 $250 - $300 0.2 - 0.5 0.2 - 0.5 0.4 $200 - $250 $150 - $200 0.3 $100 - $150 Expected to Expected to $17M $50 - $100 0.2 $- - $50 produce mean produce mean 0.1 cost savings of cost savings of 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 $17M to $43M $17M to $43M % Reduction Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(1) Set Targets: Campylobacter Cost Contour $950 - $1,000 1 $900 - $950 0.9 $850 - $900 Target of 20% Target of 20% $800 - $850 0.8 reduction reduction $115M $750 - $800 $700 - $750 0.7 $650 - $700 0.6 Attribution $600 - $650 Attribution Attribution $550 - $600 0.5 0.4 - 0.7 0.4 - 0.7 $500 - $550 0.4 $450 - $500 $66M $400 - $450 0.3 Expected to Expected to $350 - $400 $300 - $350 0.2 produce mean produce mean $250 - $300 0.1 $200 - $250 cost savings of cost savings of $150 - $200 0 $66M to $115M $66M to $115M $100 - $150 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 $50 - $100 % Reduction $- - $50 Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Risk Modelling Contribution to Decision Making (1) Set Targets • What parts of the system influence risk most • What parts of the system do we not know (2) Focus Attention (2) Focus Attention enough about (3) Formulate Strategy (4) Test Strategy Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(2) Focus Attention: C.jejuni Process Model Dose Prevalence Response Farm Slaughter Preparation RISK & & & Transport Processing Consumption Concentration Aamir Fazil aamir.fazil@phac-aspc.gc.ca
HAZARD EXPOSURE ASSESSMENT CHARACTERIZATION FARM & TRANSPORT Flock Within External Prevalence Flock Contam. Prevalence PROCESSING Scald De-feather Evisceration Wash CONSUMPTION Chill Amount Consumed STORAGE Refrigeration Freezing RESPONSE PREPARATION Dose-Response (Probability of Infection) Cross contamination Probability of Illness and Cooking RISK RISK Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(2) Focus Attention: Sensitivity Analysis Cooking Temp / Time Load on Birds entering Process Dose Response Process: Defeathering Process: Scalding Transport Refrigerated Storage Process: Chilling -0.80 -0.60 -0.40 -0.20 0.00 0.20 0.40 0.60 Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(2) Focus Attention: Sensitivity Analysis • Cooking Temp/Time – Difficult to control consumer practice – Education is an option, how effective? – Research into survival of C.jejuni in real world cooking scenarios – Research into consumer practices Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Decision Making(2) Focus Attention: Sensitivity Analysis • Load on Birds Entering the Process – Controls to reduce the load entering the process substantial impact on risk. – Research into pathogenicity. • Example: should we be concerned with all strains? – More data to quantify the conc. of pathogenic C.jejuni strains entering process. – Research into ways to reduce contamination pre ‐ processing Aamir Fazil aamir.fazil@phac-aspc.gc.ca
Recommend
More recommend
