a risk & vulnerability assessment methodology for food systems
TRANSCRIPT
February 19, 2007
A Risk & Vulnerability Assessment Methodology for Food Systems
Ryan Newkirk, MPHNational Center for Food Protection & Defense
Preview
Context & backgroundSystemic risk/vulnerability assessment methodologyModel results
Background
RiskFeasible, detrimental outcome of an activity or action
Characterized by:Severity: magnitude of possible adverse consequencesProbability: likelihood of occurrence of consequence
Background
Food SystemInputs, outputs, and processes occurring along the production-to-consumption continuum of one or more foods
Background
DHS identified food protection & defense among its top priorities
Continuing threat of intentional contaminationRecent, high-profile foodborne disease outbreaksSalmonella typhimurium & peanut butterMelamine & milk
Background
Risk assessments Many are qualitativeCommon for single location / facilityDifficult on dynamic / highly integrated systemsNo consensus on best methodologyPaucity of system-level research
Purpose
Develop a methodology to assess risk / vulnerability on a food systemPilot project
MN fluid milk system
Methodology
2 Main StepsStep 1. Characterize the MN fluid milk system
3 Phases
Step 2. Create model / conduct risk assessment3 Phases
Step 1. Phase 1. System Characterization
Main ActivitiesIdentify system inputs and outputsIdentify system processesQuantify / estimate system variabilities & uncertainties
Step 1. Phase 1. System Characterization
System Diagram Collate and synthesize all information
Trace commodity flowCan include specs
CapacitiesRatesRegulations
Step 1. Phase 2. Parameter Identification
Model parameter identification / specification 2 categories
System characterization parameters / Parameter archetypesModular parameters
Step 1. Phase 2. Parameter Identification
System characterization parametersBased on system characterization diagram / dataKey system nodesParameter Archetypes
EssentialBroadly-aggregated categories Included in majority food systems
Step 1. Phase 2. Parameter Identification
Farm / production site TransportationProcessingDistributionConsumptionAgent characteristicsEvent detection
Economics
Systemic Relationships among Archetypes
Farm
Transportation
Processing Distribution Consumption
Agent Detection
Step 1. Phase 2. Parameter Identification
Modular ParametersThreat module
Probability of attack
Vulnerability moduleProbability of success given an attackDetection and/or destruction of contaminantsCan utilize expert solicitation
Step 1. Phase 2. Parameter Identification
Modular ParametersConsequence module
Estimation of social and economic effectsPopulation susceptibility to contaminantRecall logisticsProduct loss
Step 1. Phase 3. Scenario Development
Select agent / contaminantBotulinum toxin
Select location of contaminationAssumptions
Can be directly tied to threat and vulnerability analysesCan utilize expert solicitation
Step 2. Model Creation / Risk Assessment
3 PhasesDeterministic
Ideal for use in well characterized and controlled systems
ProbabilisticIncorporates variability and uncertainty in systems
SimulationMultiple model runs
Step 2. Phase 1. Model Creation / Risk Assessment
Deterministic PhaseIdeal for well characterized and controlled systemsFarm-to-table system-based milk flow patternsContamination concentration followed through system to consumerIntentional contamination modeled at different locations / times
Step 2. Phase 1. Model Creation / Risk Assessment
Key points / nodes are linkedInter-worksheet connections
Different worksheet for each nodeAll calculations linked
Intra-worksheet parametersInputs, outputs, volumes, cleaning cycles, etc.
Morbidity, mortality, and economic outputsIdentifies candidate model parameters for probabilistic phase
Step 2. Phase 2. Model Creation / Risk Assessment
Probabilistic PhaseIdentifies / assesses risks in complex systems to improve safety and performance
Step 2. Phase 2. Model Creation / Risk Assessment
Probabilistic PhaseBased on system characterization & deterministic model phasesUse distributions instead of point estimates for values that are
Not well characterizedInherently variableImportant degree of uncertainty
Step 2. Phase 3. Model Creation / Risk Assessment
Simulation PhaseMonte Carlo methodology
Using distributions generated in previous phase Randomly selects inputs from distributionPerform deterministic step RepeatAggregate results of individual computations into final result
Model Results
Probability distribution of main outputs
MorbidityMortalityEconomic costs
Sensitivity analyses
Application of Results
Results identify system node(s) that contributes most to risk/vulnerability
Insert mitigation strategiesRerun model
Assess morbidity/mortality estimatesIf estimates are reduced, proceed to next system nodeRepeat as necessary
End result = systemic risk/vulnerability reduction
Challenges
DataIndustry access & cooperationModel assumptions
Summary
Reducing systemic risk/vulnerability requires a systemic approachDifferent system attributes/nodes require different assessment methodologiesAssess risk/vulnerability on a food system
System characterizationModel creation / Risk assessment
Questions?