Gordon C. Ashton Memorial Lecture System s and Risk Modelling for - - PowerPoint PPT Presentation

Gordon C. Ashton Memorial Lecture

Aamir Fazil Public Health Risk Sciences Division, LFZ, PHAC

System s and Risk Modelling for Food Safety Decision Making

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

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 Su m Pr o ba bi l i t y 2 1/ 36 3 2/ 36 4 3/ 36 5 4/ 36 6 5/ 36 7 6/ 36 8 5/ 36 9 4/ 36 10 3/ 36 11 2/ 36 12 1/ 36

History

Took someone who threw a lot of dice to finally begin understanding probability and risk

Girolamo Cardano (1501–1576), Mathematician & Physician

• ne of first clinical descriptions of

Typhoid Fever

• “Liber de ludo aleae” written in the 1560s
• first systematic treatment of probability, as

well as a section on effective cheating methods

• Probability of rolling various dice combinations

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
• utcomes

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
• Epidemiology

Risk Modelling

• Risk Mgmt Decisions
• Policy
• Link: Research / Data and Decisions

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

System Modelling

System risk modeling

System Experiment with the actual system Experiment with the actual system Experiment with a model of system Experiment with a model of system Physical Model Physical Model Mathematical Model Mathematical Model Analytical Solution Analytical Solution Simulation Solution Simulation Solution

• Food production system
• Health system
• Farm system
• Farm to consumption system
• Food production system
• Health system
• Farm system
• Farm to consumption system

Adapted from 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

(1) Set Targets (1) Set Targets

• Link between contamination and public health
• How much of a public health issue
• Public health impact of potential reductions

(2) Focus Attention (2) Focus Attention

• What parts of the system influence risk most
• What parts of the system do we not know

enough about

(3) Formulate Strategy (3) Formulate Strategy

• What can we do to reduce the risk
• What options do we potentially have to work with

(4) Test Strategy (4) Test Strategy

• How much of an impact could a strategy have
• What potential cautions should we consider

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Risk Modelling Contribution to Decision Making

(1) Set Targets (1) Set Targets

• Link between contamination and public health
• How much of a public health issue
• 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

Reduced contaminated volume

• f chicken

pDR Probability of illness per volume of contaminated chicken in market Future human cases resulting from chicken Cost per case Chicken Contamination Rate Volume of Chicken in Market Volume of Contaminated Chicken in Market Reported Illness Rates Underreporting Rates Actual Overall Illness Rate Attribution

• f Illness to

Chicken Actual Illness Rate Attributable to Chicken Cost of illness Illness Severe Illness Morbidity Reduced contaminated volume

• f chicken

pDR Probability of illness per volume of contaminated chicken in market Future human cases resulting from chicken Cost per case Chicken Contamination Rate Volume of Chicken in Market Volume of Contaminated Chicken in Market Reported Illness Rates Underreporting Rates Actual Overall Illness Rate Attribution

• f Illness to

Chicken Actual Illness Rate Attributable to Chicken Cost of illness Illness Severe Illness Morbidity

Decision Making(1)

Set Targets: Model Overview

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

• Reported Illness Rates
• Contamination Rates
• Underreporting 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

Target of 20% reduction Attribution 0.2 - 0.5 Expected to produce mean cost savings of $17M to $43M Target of 20% reduction Attribution 0.2 - 0.5 Expected to produce mean cost savings of $17M to $43M

Decision Making(1)

Set Targets: Salmonella Cost Contour

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 % Reduction Attribution $450 - $500 $400 - $450 $350 - $400 $300 - $350 $250 - $300 $200 - $250 $150 - $200 $100 - $150 $50 - $100 $- - $50

$43M $17M

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Target of 20% reduction Attribution 0.4 - 0.7 Expected to produce mean cost savings of $66M to $115M Target of 20% reduction Attribution 0.4 - 0.7 Expected to produce mean cost savings of $66M to $115M

Decision Making(1)

Set Targets: Campylobacter Cost Contour

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 % Reduction Attribution $950 - $1,000 $900 - $950 $850 - $900 $800 - $850 $750 - $800 $700 - $750 $650 - $700 $600 - $650 $550 - $600 $500 - $550 $450 - $500 $400 - $450 $350 - $400 $300 - $350 $250 - $300 $200 - $250 $150 - $200 $100 - $150 $50 - $100 $- - $50

$115M $66M

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Risk Modelling Contribution to Decision Making

(1) Set Targets (2) Focus Attention (2) Focus Attention

• What parts of the system influence risk most
• What parts of the system do we not know

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

Prevalence Concentration

Farm & Transport Slaughter & Processing Preparation & Consumption RISK

Dose Response

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

RISK RISK

STORAGE

Refrigeration Freezing

PREPARATION

Cross contamination and Cooking

FARM & TRANSPORT

Flock Prevalence Within Flock Prevalence External Contam.

EXPOSURE ASSESSMENT

Scald De-feather Evisceration Wash Chill

PROCESSING RESPONSE

Dose-Response (Probability of Infection) Probability of Illness

HAZARD CHARACTERIZATION

CONSUMPTION

Amount Consumed

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(2)

Focus Attention: Sensitivity Analysis

• 0.80
• 0.60
• 0.40
• 0.20

0.00 0.20 0.40 0.60

Cooking Temp / Time Load on Birds entering Process Dose Response Process: Defeathering Process: Scalding Transport Refrigerated Storage Process: Chilling

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

Risk Modelling Contribution to Decision Making

(1) Set Targets (2) Focus Attention (3) Formulate Strategy (3) Formulate Strategy

• What can we do to reduce the risk
• What options do we potentially have to work with

(4) Test Strategy

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(3)

Formulate Strategy: Reduction Options

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0% 1% 3% 5% 15% 30% 50% 70% 90% 99%

Average Log Concentration Prevalence

Decreasing Risk

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0% 1% 3% 5% 15% 30% 50% 70% 90% 99%

Average Log Concentration Prevalence

Decision Making(3)

Formulate Strategy: Reduction Options

X X

90% Risk Reduction

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0% 1% 3% 5% 15% 30% 50% 70% 90% 99%

Average Log Concentration Prevalence

Decision Making(3)

Formulate Strategy: Reduction Options

X X

90% Risk Reduction

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0% 1% 3% 5% 15% 30% 50% 70% 90% 99%

Average Log Concentration Prevalence

Decision Making(3)

Formulate Strategy: Reduction Options

X X

90% Risk Reduction

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Risk Modelling Contribution to Decision Making

(1) Set Targets (2) Focus Attention (3) Formulate Strategy (4) Test Strategy (4) Test Strategy

• How much of an impact could a strategy have
• What potential cautions should we consider

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(4)

Test Strategy: Model Scenario Analysis

• What effect does changing the internal and surface contamination of chickens

before and through processing have ?

• Four (4) alternative strategies investigated

– Strategy 1: Reduction in surface contamination level after transport – Strategy 2: Reduction in levels contaminating carcasses at evisceration – Strategy 3: Reduction in surface contamination post evisceration – Strategy 4: Reduction in initial internal contamination levels (overall reduction in contamination entering the system)

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(4)

Test Strategy: Scenario Analysis Results

Strategy 1: Reduction in surface contamination post transport Strategy 2: Reduction in amount of contamination deposited at evisceration Strategy 3: Reduction in surface contamination post evisceration Strategy 4: Reduction in overall internal colonization and contamination levels

Baseline 35% reduction 25% reduction 63% reduction 69% reduction

0.0E+00 2.0E-04 4.0E-04 6.0E-04 8.0E-04 1.0E-03 1.2E-03 1.4E-03

Baseline Strategy 1 Strategy 2 Strategy 3 Strategy 4 Estimated Mean Risk

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(4)

Test Strategy: Scenario Analysis Results

• Reducing surface contamination after evisceration can have a

significant impact on reducing the risk

– Reductions of surface contamination prior to this get negated by additional contamination being deposited

• Targeting the internal colonization levels at the farm level has

a significant effect on reducing the risk

– Reducing the overall pool of contamination entering the system

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(4)

Test Strategy: Potential complications

• Freezing chicken as a risk reduction strategy
• What is the difference in risk for refrigerated
• r fresh chicken compared to frozen chicken ?
• Assumptions

– Refrigerated (0 to 9 days) – Frozen ( 1 day to 6 weeks)

Aamir Fazil aamir.fazil@phac-aspc.gc.ca 0.0E+00 1.0E-02 2.0E-02 3.0E-02 4.0E-02

2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 Mean Log Conc. Mean Risk Refrig Risk Freeze Risk(0C)

Decision Making(4)

Test Strategy: Potential complications

Risk for refrigerated chicken Risk for frozen chicken

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(4)

Test Strategy: Potential complications

• Frozen chicken is estimated to result in lower risk
• However, these results can be complicated

– Preparation practices could reverse the effect – Example: it is possible that cooking effectiveness could be diminished for frozen chicken compared to fresh chicken

• Scenario A: Final cooking temperature, 2C cooler in cold spots for

frozen chicken

• Scenario B: Final cooking temperature, 5C cooler in cold spots for

frozen chicken

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Decision Making(4)

Test Strategy: Potential complications

0.0E+00 1.0E-02 2.0E-02 3.0E-02 4.0E-02

2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 Mean Log Conc. Mean Risk Refrig Risk Freeze Risk(0C) Freeze Risk (2C)

Risk for refrigerated chicken Risk for frozen chicken Risk for frozen chicken with 2C cooler cooking

Aamir Fazil aamir.fazil@phac-aspc.gc.ca 0.0E+00 1.0E-02 2.0E-02 3.0E-02 4.0E-02

2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 Mean Log Conc. Mean Risk Refrig Risk Freeze Risk(0C) Freeze Risk (2C) Freeze Risk (5C)

Decision Making(4)

Test Strategy: Potential complications

Risk for refrigerated chicken Risk for frozen chicken Risk for frozen chicken with 2C cooler cooking Risk for frozen chicken with 5C cooler cooking

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Conclusions

• Risk Assessment / Modelling

– Contributes to the understanding of the system – Helps identify critical factors that most significantly influence risk

• Risk mitigation / control implications
• Research direction implications

– Discrimination between information that is unknown and relatively unimportant vs. unknown and important.

– Provides guidance on expected impact of risk mitigation strategies

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Conclusions

• For good decision making, quantitative risk

modelling is a good alternative

– For complex systems – For difficult decisions – For transparent processes – To ensure appropriate allocation of resources

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Lots of Detail in One Area

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Lucky, capture the essence

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Understand the big picture

Aamir Fazil aamir.fazil@phac-aspc.gc.ca

Thank You