Food safety risk management: Principles and practice

1
Food safety risk management Principles and
practice

• Steve HathawayNew Zealand Food Safety Authority

2
Imported food
3
Domestic food
Domestic food production
4
Exported food
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
What is safe food?

• Codex all conditions and measures applied
  during the production process that are necessary
  to achieve safety and suitability

9
What is safe food?
10
What is suitable food?
11
Risk analysis

• An outcome-based approach to food safety
• Anchors the concept of safe food

12
Positioning risk analysis in food control

• GHP in the 1980s
• HACCP in the early 1990s
• Influence of WTO SPS Agreement in mid-1990s
• Risk analysis in the late 1990s
• Structural change in Competent Authorities in the
  2000s
• New - source attribution and setting risk-based
  public health goals

13
Keeping the questions simple

• What can go wrong?
• How likely is it to go wrong?
• How serious would it be if it went wrong?
• What can be done to reduce the likelihood and/or
  seriousness of it going wrong?

14
Keeping the response simple

• Manage the risks rather than manage the
  hazards
• Hazard - an agent with the potential to cause
  harm
• Risk - adverse health effect on the consumer

15
Hazard-based regulatory activities

• Decisions, standards and actions that are based
  on objective and verifiable information on
  relevant hazards
• Aimed at eliminating/reducing exposure to
  hazards, with the expectation that there will be
  a reduction in risk
• Will continue to be an essential output of
  Competent Authorities

16
Risk-based regulatory activities

• Decisions, standards and actions that are based
  on specific knowledge of risks
• Aimed at achieving an established level of health
  protection and should be able to be validated in
  these terms
• Should be an increasing output of Competent
  Authorities

17
Taking a lead from WTO

• Applying science and risk assessment to the
  extent possible
• Risk management decisions should be based on
  robust science and in proportion to the likely
  benefits in terms of public health
• Standards should ensure fair practices in the
  trade in food

18
Risk analysis
19
Operationalising risk analysis

• Principles of risk analysis now well embedded in
  national legislation
• Operationalisation not as well advanced
• Application of a Risk Management Framework (RMF)
  provides a systematic and consistent process for
  food control
• Codex guidelines provide a good reference

20
Risk Management Framework

21
Risk profiling Scientific evaluation Risk
assessment
Scientific evaluation Decision support tools
Database design Scientific evaluation
Pre-eminent role of science
Validation Verification
22
Risk Management Framework (1)

23
Scientific evaluation

• Risk profiles
• Empirical scientific evaluation
• Safety evaluation (chemical hazards)
• Risk assessment
• Food source attribution
• Ranking tools

24
Risk profiles commissioned by NZFSA

• Campylobacter
• Poultry
• Offals
• Red meat

• Listeria monocytogenes
• RTE Meats
• Ice cream
• Soft Cheese
• Low Moisture Cheese
• RTE Salads

• STEC
• Red meat
• UCFM
• Leafy vegetables
• Raw milk

• Salmonella
• Poultry
• Eggs

• Vibrio parahaemolyticus
• Seafood

• Toxoplasma gondii
• Red meat

• Bacillus spp.
• Rice

• Yersinia enterocolitica
• Pork

• Clostridium
• botulinum
• Seafood
• Honey

• Mycobacterium bovis
• Milk
• Red meat

• Ciguatoxins
• Seafood

• Norovirus
• Shellfish

25
Scientific evaluation

• Many risk management decisions will continue to
  rely on scientific evaluation rather than risk
  assessment
• The norm in emergency or precautionary situations
  
• Supported by applied research on hazard control,
  exposure assessments, monitoring and surveillance
  registers
• Recent NZ examples STEC/VTEC E.coli in UCFMs,
  hazard database for animal feeds, A1 and A2
  beta-casein in milk, tutin toxin in honey, Total
  Diet Survey

26
Risk assessment

• Uses whole of food chain and dose/response to
  estimate risk (population basis or edible portion
  basis)
• Stochastic modelling uses probability
  distributions to describe variability,
  uncertainty
• Primary value - demonstrate impact of different
  interventions throughout the food chain on the
  risk estimate
• Chemical hazards - safety evaluation?

27
NZFSA examples

• Campylobacter in broiler chickens (domestic)
• Cysticercus bovis in beef (export)
• Salmonella in broiler chicken (import)
• S. Brandenberg in sheep meat (domestic and
  export)
• Pasteurisation of milk (domestic)
• Cheeses made from unpasteurised milk (domestic
  and import)
• Mercury in fish (import, domestic and export)

28
Process flow diagram
29
Taenia saginata exposure pathways in beef
produced in New Zealand
30
Taenia saginata Risk estimates

• Mean number of infections in export and domestic
  markets is 0.5 and 1.1 per year respectively
• Probability of infection per edible portion of
  meat is 4.910-10 and 510-9 respectively

31
S. Brandenburg Pattern 14 in humans in NZ
32
Risk analysis at peak of epidemic

• Foodborne risks to human health estimated at 6
  -10 cases per year
• Validation
• - 0 foodborne cases based on case control study
• - 0 foodborne cases on regional trace-back
  investigations
• - 500 notified human cases (all routes)
• No changes made to process control or inspection

33
Havelaar et al Campylobacter in poultry
34
Pasteurisation Modelling the food chain
Consump- tion
Primary production
Product processing
Retail sale
Consumer handling
35
Pasteurisation primary production
Milking

• Probability that milk is contaminated
• Probable concentration of hazard

Onfarm storage

• Given that a hazard is present,
• probable growth or inactivation
• Probable concentration of hazard

Transport to the factory

• Given that a hazard is present,
• probable growth or inactivation
• Probable concentration of hazard

• Probability that milk is contaminated at the end
  of Primary Production
• Probable concentration of hazard in milk at the
  end of Primary Production

36
Raw milk survey Total E. coli and S. aureus
37
Raw milk survey - pathogens
38
Pasteurisation product processing
Factory storage

• Given that a hazard is present,
• probable growth or inactivation
• Probable concentration of hazard

Heat treatment

• Given that a hazard is present,
• probable growth or inactivation
• Probable concentration of hazard

• Probability that milk is contaminated at the end
  of Product Processing (milk)
• Probable concentration of hazard in milk at the
  end of Product Processing (milk)

39
Screening pathogenic groups at 60 - 62.5 C for
60 sec
40
Pasteurisation Product processing (cheese)
Factory storage

• Given that a hazard is present,
• probable growth or inactivation
• Probable concentration of hazard

• Given that a hazard is present,
• probable growth or inactivation
• during each step of manufacture
• Probable concentration of hazard
• during each step of manufacture

Cheese manufacture

• Probability that cheese is
• contaminated at the end of
• Product Processing (cheese)
• Probable concentration of hazard in
• cheese at the end of Product
• Processing (cheese)

41
Pasteurisation Risk assessment outputs

• Estimate of risk using a standard process
• Product categorisation based on risk assessment
  outputs
• Category 1 currently permitted for sale
• Category 2 sale permitted
• Category 3 sale not permitted
• Practical tool for risk managers at premises
  level Food Control Plans

42
Epidemiological approach
Risk assessment

Exposure
Illness
Food chain
Epidemiology / food source
attribution
43
Food source attribution

• Risk assessment usually provides information on a
  single hazard/food combination
• Proportion of foodborne disease attributable to
  all hazard/food combinations?
• Analytical epidemiology
• Simulation modelling
• Molecular epidemiology (subtyping / source
  tracking)
• Cannot determine the relative impact of specific
  control measures cf. risk assessment

44
Human campylobacteriosis
45
Human campylobacteriosis Manawatu
water
sheep
chicken
wild bird
cattle
46
Salmonella attribution in NZ (1)
47
Salmonella attribution in NZ (2)
48
Risk Management Framework (2)

49
Availability of control measures

• GHP-based measures traditional, prescriptive,
  vary markedly between countries
• Hazard-based measures measurable at a step,
  less prescriptive, vary between countries
• Risk-based measures outcome focused, flexible,
  international compatibility, opportunity for
  equivalence

50
Decisions on control measures

• Need to anchor concept of safe and suitable
• Risk management decisions on all aspects of food
  control should be proportional to the likely
  reduction in foodborne risks to the consumer
• Decisions on control measures have a significant
  impact on costs to industry and government
• Take suitability (consumer acceptability) and
  consumer perceptions into account

51
Risk management of Taenia saginata

• Mean number of infections in export and domestic
  markets is 0.5 and 1.1 per year respectively /
  probability of infection per edible portion of
  meat is 4.910-10 and 510-9 respectively
• Equivalence case presented to US and EU
• Significantly reduced meat inspection and less
  cross-contamination
• HACCP active surveillance and suspect list

52
Level of consumer protection soft cheese

• Probability of occurrence of severe listeriosis
  per consumed portion of cheese (concentration and
  portion size)
• Probability of occurrence of severe listeriosis
  per person per year (concentration, portion size
  and number of portions)
• Tolerable number of severe listeriosis cases per
  year (average risk per person per year)

Maximum frequency And concentration
Maximum frequency And concentration
Maximum frequency And distribution
of concentration
53
Risk management of chemical hazards
Microbial hazard/food combination
Chemical hazard/food combination
LOP predetermined
Risk assessment
Risk assessment
LOP arrived at
Choice of risk-based control measuresChoice of
risk-based control measures
Implementation)
54
Havelaar et al Campylobacter in poultry
55
Integrated decision making

• Increasing need for an integrated approach e.g.
  health impacts, economic impacts, lifestyle
  impacts and consumer perceptions
• Need risk-based knowledge on all significant
  hazard pathways
• Quantify disease burden e.g. DALYs and
  intervention costs at levels of consumer,
  industry and government
• Multi-objective trade-off analysis needed

56
Risk Management Framework (3)

57
Implementation of measures

• Validation - gaining evidence that a control
  measure is capable of controlling a hazard to a
  specified outcome hazard-based (step) or
  risk-based
• Verification - activities that determine if a
  control measure is or has been operating as
  intended
• Provides maximum flexibility while ensuring
  required outcomes are met
• Validation, verification and monitoring are
  inter-related activities that assure regulatory
  performance

58
Validation
Production step
Process step
Process step
Consumer

• Options for control measures
• based on GHP
• - based on hazards
• - based on risks

No validation Step validation
HACCP
Chain validation
ALOP
Risk model
59
Verification

• Activities that determine if a control measure is
  or has been operating as intended
• Occurs after implementation of control measures
• Regulatory responses to non-compliance at a
  premises level should take into account the
  characteristics of the test system, the likely
  risks to human health and overall compliance

60
Performance-based verification (1)
61
Performance-based verification (2)
62
Data systems NZFSA VAOnLine
Web based reports and files available world-wide
with password access National planning tool and
data repository system Procedures storage
system Regulatory model includes third party
verifiers
63
Microbiological targets as risk-based tools

• Performance objective (Codex) The maximum
  frequency and/or concentration of a hazard in a
  food at a specified step in the food chain .
  that provides or contributes to . ALOP
• Derive risk-based performance criteria, process
  criteria, product criteria, microbiological
  criteria
• Valuable implementation and equivalence tools but
  derivation methods still controversial

64
HACCP as an implementation tool

• CCP A step at which control can be applied and
  is essential to prevent or eliminate a food
  safety hazard or reduce it to an acceptable
  level
• Decisions usually based on ALARA principles i.e.
  hazard-based control measures, and not on ALOP
• Critical limits can only be validated at that
  step
• HACCP plans are usually premises-specific and
  therefore not directly comparable between
  premises / food chains / countries

65
Risk Management Framework (4)

66
Monitoring

• Different systems for different purposes
• Premises or whole country basis
• Compliance-based e.g. hazard levels associated
  with non- compliance with a MRL
• Exposure based e.g. total diet surveys
• Risk-based e.g. Codex performance targets
• Include foodborne disease surveillance
• Include regulatory performance goals

67
One example NZFSA NMD

• National Microbiological Database for all red
  meat species, poultry and ostriches
• Indicators and pathogens
• Monthly reporting of data from all premises
• Premises and national performance targets (80th
  percentile based)
• Ranking of premises on a quarterly basis
  (indicator organisms)

68
Industry performance Bovine NMD
69
Regulatory performance

Intermediate outcomes
Outcomes
Activities/ outputs
Standards Implementation Compliance
Level of hazard control
Level of consumer protection
70
NZFSA public health goals Microbes

• As-low-as-reasonably-achievable (ALARA) rather
  than appropriate level of protection (ALOP)
  approach
• - 50 reduction in foodborne campylobacteriosis
  after 5 years
• - 30 reduction in foodborne salmonellosis after
  five years
• - no increase in foodborne listeriosis after
  five years
• Requires a risk-based approach to food control if
  goals are to be realised

71
Regulatory performance
72
NZFSA public health goals Chemicals

• Hazard-based!
• 70 of school-age children achieve ideal range
  for dietary intake of iodine over five year
  period
• Agricultural compounds measured in the
  five-yearly TDS do not exceed 10 of the ADI
• Exposure to the dithiocarbamate group of
  fungicides, as measured in the TDS, will not
  exceed 20 of the ADI
• Dietary intake of mercury for women 25-years and
  older remains below 20 of the PTWI

73
Performance goal dietary intake of iodine
74
International standards

Codex Alimentarius Commission
75
A Codex opinion

• Codex does not set public health goals,
  implement standards or monitor their performance
  and is likely that Codex standards will
  increasingly be enabling in nature, with
  specific risk-based control measures being the
  domain of national governments

76
Replication of measures A failure of the risk
analysis paradigm?
Kiwifruit
77
BSE as a global issue
Measures not commensurate with risk and driven by
perceived need to eliminate the hazard
Number of Variant-CJD-cases
UK 157 cases, France 13, Ireland 3, 1 each Italy
, Canada , USA , Japan , Saudi-Arabia ,
Netherlands, Spain, Portugal
78
Replication of BSE measures
Cost effectiveness (?)
79
Equivalence

• Key provision of the WTO SPS Agreement
• Mutual recognition and comparability are
  needed as screening processes, taking advantage
  of experience, knowledge, trust and confidence
• Additional criteria are needed to establish
  comparability e.g. risk profiles (NZ/EU
  Equivalency Agreement)
• Objective basis of comparison in terms of risks
  to human health is final arbiter of equivalence

80
Summary (1) Benefits of a RMF process

• Continual improvements in public health
• Ranking of risks and proportional regulatory
  responses
• Cost-effective and efficient food control
• Consistency and transparency of decisions
• Stakeholder participation
• Industry innovation and flexibility (equivalence)
• Compatibility with international standards

81
Summary (2)

• A risk-based approach to food safety brings
  multiple advantages but the benefits can only be
  fully realised with application of a RMF
• The science that provides essential inputs to a
  RMF should be fit-for-purpose
• Translating risk management principles into
  practice at the national level remains a
  significant challenge

82
Thank you