Risk assessments for Campylobacter in broiler meat: what we learned and what we need. Maarten Nauta

Risk assessments for Campylobacter in broiler meat: what we learned and what we need Maarten Nauta

Campylobacter in the broiler meat chain Campylobacter enter the food chain Carcasses get contaminated by leaking faeces Last point for official control Consumer food handling gives cross-contamination

Controlling Campylobacter in chicken meat What is the best point in the chain to control it? a) farm b) processing c) retail d) consumer

Risk assessment: a scientific food chain approach What is happening to the pathogen? What can be done to eliminate or reduce it? What is the most efficient way of doing that? where in the food chain? what intervention?

This lecture Example of a Campylobacter risk assessment The Netherlands 2005 Assessment of the effect of control measures: Farm (flock prevalence) Slaughter and processing (concentrations) Consumer stage Microbial criteria Outlook

A food chain risk model for Campylobacter in chicken meat (2005) Campylobacter on chicken breast fillets produced in the Netherlands, from animals reared in the Netherlands - current risk - potentials for interventions Exposure via cross contamination to a salad prepared after cutting the meat. all Campylobacters are killed of with cooking the meat (?)

Farm Farm to fork Processing scalding defeathering evisceration washing chilling Cutting breast cap double filet filet Storage Consumer preparation Ingestion / Dose response

Farm A Farm to Fork Microbial Risk model Processing Cutting Storage scalding defeathering evisceration washing chilling breast cap double filet filet Model transmission of microbial hazard through the food chain (prevalence and numbers) Assess consumer exposure and combine with dose response models to assess risk Consumer preparation Ingestion / Dose response Compare alternative scenarios due to interventions and estimate risk reductions achieved

source: Nauta et al, Risk Analysis, 2007

Exposure and illness, baseline scenario Cross-contamination from fillet to salad; the Netherlands, 2000 0.8% of salads is contaminated; yearly 12000 cases of GE 60.0% 50.0% exposures cases frequency 40.0% 30.0% 20.0% 10.0% 0.0% 1 2 3-10 11-100 >100 cfu ingested

Interventions reduce prevalence and/ or concentrations somewhere in the food chain Farm Monospecies farms Improved hygiene No thinning Phage therapy Logistic slaughter Channeling Processing Reduction of faecal leakage Decontamination of scalding tank Treatment of carcasses: decontamination, crust freezing, irradiation Freezing of products Channeling to meat processing Consumer Freezing at home Improved kitchen hygiene Information campaign

Risk reduction by interventions thinning farm hygiene phage therapy red. faecal leakage decont. scald tank decont. carcass dip decont. dip and spray crust freezing irradiation product freezing prepared meat kitchen hygiene home freezing -20% 0% 20% 40% 60% 80% 100%

More Campylobacter risk assessments

Generic results Reducing the concentration is more effective than reducing the prevalence. The true prevalence of contaminated carcasses after slaughter is proportional to the risk Logistic slaughter has little effect For consumers, cross-contamination has more impact than undercooking The variability in concentrations determine the risk, not so much the mean concentrations

What does risk assessment learn us about options for control Interventions at the farm Interventions at the slaughterhouse Improving consumers hygiene Microbiological criteria

Options for control Interventions at the farm Interventions at the slaughterhouse Improving consumers hygiene Microbiological criteria

Interventions at the farm CamCon project (2010 2015) Risk factor study at the farm How are farm management practices related to Campylobacter flock prevalence? Compare different EU countries, ES, NL, NO, PL, UK Risk assessment Effectivity of intervention on public health risk Cost effectiveness study

Effects of interventions at the farm Risk factors identified Large studies and NO 20 farms studies for 2 years in all countries

Risk factors for Camp. colonization in broilers Results All countries Variable Effect on prevalence Country Campylobacter positive flocks descending order: PL, ES, UK, NL, and NO Temperature Increasing temperature increasing number of positive flocks Age of house Newest house lower prevalence Biosecurity (anteroom/barrier) Biosecurity (designated tools) Ante-room/barrier Designated tools lower prevalence lower prevalence Downtime Low downtime (< 10 days) lower prevalence Drinkers (Bells, nipples with cups, nipples without cups) Nipples without cups lower prevalence Sommer et al. CamCon Annual Meeting 2015

Effects of interventions at the farm Identified risk factors translated to interventions Flock prevalence reduction assessed Additionally considered Ban on thinning (EFSA) Reduce slaughter age to 35 days (EFSA) Apply fly screens (Denmark) Sommer, et al 2016 Microbial Risk Analysis

Intervention already OK if intervention involved in intervention (% of all farms) Campylobacter prev. after control ES NL NO PL UK ES NL NO PL UK 19.2 87.7 24.2 3.3 79.2 75.8 anteroom barrier in new houses both physical barrier and anteroom in all houses < 15 yrs build anteroom and barrier (assume not both = none) in all houses < 15 yrs 27.5 30.6 48.8 37.2 39.9 36.8 17.4 84.2 21.0 2.9 77.0 71.5 down time downtime <10 days reduce downtime to < 10 days + rodent control in 30 % of these farms + disinfection (improved cleaning) between all crops at the end of downtime in 3 % of these farms 20.6 89.3 25.0 96.3 84.1 15.4 17.8 79.0 22.5 1.9 72.9 74.3 new houses < 15 years of age build new house if > 15 years + anteroom and barrier+ nipples without cups+ designated tools 42.2 66.3 34.4 30.5 49.0 48.7 14.9 38.1 17.9 2.4 76.0 55.5 drinkers nipples without cups apply nipples without cups in all farms 85.3 81.1 50.8 68.9 67.8 78.6 14.5 80.3 20.9 2.5 75.1 67.0 designated tools per house in new houses each new house ( <15 yrs) has its own tools (basic equipment) have designated tools in each house < 15 yrs 2.0 25.5 32.8 7.9 8.2 36.8 18.9 82.4 20.3 3.1 78.7 67.3 no thinning lower slaughter age fly screens (in new houses) improved cleaning no thinning applied stop thinning 24.8 49.1 42.2 3.3 49.4 63.4 16.5 80.4 18.4 3.2 69.6 60.0 35 days or less slaughter at 35 days 85.0 94.6 85.3 5.1 97.4 80.2 17.2 75.5 19.0 3.3 66.5 66.9 applied for all houses apply fly nets in houses < 15 years 33.0 16.6 Disinfection between all crops start disinfection Not significant, not done combined with downtime rodent control Farmer applies rodent control > 6 times per year start application of rodent control Not significant, not done combined with downtime

Intervention already OK if intervention involved in intervention (% of all farms) Campylobacter prev. after control ES NL NO PL UK ES NL NO PL UK 19.2 87.7 24.2 3.3 79.2 75.8 anteroom barrier in new houses both physical barrier and anteroom in all houses < 15 yrs build anteroom and barrier (assume not both = none) in all houses < 15 yrs 27.5 30.6 48.8 37.2 39.9 36.8 17.4 84.2 21.0 2.9 77.0 71.5 down time new houses downtime <10 days < 15 years of age reduce downtime to < 10 days + rodent control in 30 % of these farms + disinfection (improved cleaning) between all crops at the end of downtime in 3 % of these farms build new house if > 15 years + anteroom and barrier+ nipples without cups+ dedicated tools 20.6 89.3 25.0 96.3 84.1 15.4 17.8 79.0 22.5 1.9 72.9 74.3 42.2 66.3 34.4 30.5 49.0 48.7 14.9 38.1 17.9 2.4 76.0 55.5 drinkers nipples without cups apply nipples without cups in all farms 85.3 81.1 50.8 68.9 67.8 78.6 14.5 80.3 20.9 2.5 75.1 67.0 designated tools per house in new houses no thinning Intervention can be applied, CamCon questionaire each new house ( <15 yrs) has its own tools (basic equipment) Different per intervention and country have dedicated tools in each house < 15 yrs 2.0 25.5 32.8 7.9 8.2 36.8 18.9 82.4 20.3 3.1 78.7 67.3 no thinning applied stop thinning 24.8 49.1 42.2 3.3 49.4 63.4 16.5 80.4 18.4 3.2 69.6 60.0 lower slaughter age fly screens (in new houses) improved cleaning 35 days or less slaughter at 35 days 85.0 94.6 85.3 5.1 97.4 80.2 17.2 75.5 19.0 3.3 66.5 66.9 applied for all houses apply fly nets in houses < 15 years 33.0 16.6 Disinfection between all crops start disinfection Not significant, not done combined with downtime rodent control Farmer applies rodent control > 6 times per year start application of rodent control Not significant, not done combined with downtime

Intervention already OK if intervention involved in intervention (% of all farms) Campylobacter prev. after control ES NL NO PL UK ES NL NO PL UK 19.2 87.7 24.2 3.3 79.2 75.8 anteroom barrier in new houses both physical barrier and anteroom in all houses < 15 yrs build anteroom and barrier (assume not both = none) in all houses < 15 yrs 27.5 30.6 48.8 37.2 39.9 36.8 17.4 84.2 21.0 2.9 77.0 71.5 down time downtime <10 days reduce downtime to < 10 days + rodent control in 30 % of these farms + disinfection (improved cleaning) between all crops at the end of downtime in 3 % of these farms 20.6 89.3 25.0 96.3 84.1 15.4 17.8 79.0 22.5 1.9 72.9 74.3 Current prevalence, EFSA baseline 2008 new houses < 15 years of age build new house if > 15 years + anteroom and barrier+ nipples without cups+ dedicated tools 42.2 66.3 34.4 30.5 49.0 48.7 14.9 38.1 17.9 2.4 76.0 55.5 drinkers nipples without cups apply nipples without cups in all farms 85.3 81.1 50.8 68.9 67.8 78.6 14.5 80.3 20.9 2.5 75.1 67.0 designated tools per house in new houses each new house ( <15 yrs) has its own tools (basic equipment) have designated tools in each house < 15 yrs 2.0 25.5 32.8 7.9 8.2 36.8 18.9 82.4 20.3 3.1 78.7 67.3 no thinning lower slaughter age fly screens (in new houses) improved cleaning no thinning applied stop thinning 24.8 49.1 42.2 3.3 49.4 63.4 16.5 80.4 18.4 3.2 69.6 60.0 35 days or less slaughter at 35 days 85.0 94.6 85.3 5.1 97.4 80.2 17.2 75.5 19.0 3.3 66.5 66.9 applied for all houses apply fly nets in houses < 15 years 33.0 16.6 Disinfection between all crops start disinfection Not significant, not done combined with downtime rodent control Farmer applies rodent control > 6 times per year start application of rodent control Not significant, not done combined with downtime

Intervention already OK if intervention involved in intervention (% of all farms) Campylobacter prev. after control ES NL NO PL UK ES NL NO PL UK 19.2 87.7 24.2 3.3 79.2 75.8 anteroom barrier in new houses both physical barrier and anteroom in all houses < 15 yrs build anteroom and barrier (assume not both = none) in all houses < 15 yrs 27.5 30.6 48.8 37.2 39.9 36.8 17.4 84.2 21.0 2.9 77.0 71.5 down time downtime <10 days reduce downtime to < 10 days + rodent control in 30 % of these farms + disinfection (improved cleaning) between all crops at the end of downtime in 3 % of these farms 20.6 89.3 25.0 96.3 84.1 15.4 17.8 79.0 22.5 1.9 72.9 74.3 new houses < 15 years of age build new house if > 15 years + anteroom and barrier+ nipples without cups+ dedicated tools 42.2 66.3 34.4 30.5 49.0 48.7 14.9 38.1 17.9 2.4 76.0 55.5 drinkers nipples without cups apply nipples without cups in all farms 85.3 81.1 50.8 68.9 67.8 78.6 14.5 80.3 20.9 2.5 75.1 67.0 designated tools per house in new houses no thinning lower slaughter age fly screens (in new houses) improved cleaning rodent control each new house ( <15 yrs) has its own tools (basic equipment) have designated tools in each house < 15 yrs 2.0 25.5 32.8 7.9 8.2 36.8 18.9 82.4 20.3 3.1 78.7 67.3 no thinning applied stop thinning 24.8 49.1 42.2 3.3 49.4 63.4 16.5 80.4 18.4 3.2 69.6 60.0 35 days or less slaughter at 35 days 85.0 94.6 85.3 5.1 97.4 80.2 17.2 75.5 19.0 3.3 66.5 66.9 applied for all houses apply fly nets in houses < 15 years 33.0 16.6 Disinfection between all crops Farmer applies rodent control > 6 times per year The best intervention differs per country Reduction in flock prevalence is always limited start disinfection Not significant, not done combined with downtime start application of rodent control Not significant, not done combined with downtime CamCon Annual Meeting 2012

Main conclusions Camcon: interventions at the farm Differences between countries matter No single intervention solves the problem It s all about strict biosecurity

Cost of interventions on farm Average costs per intervention ( /100 chickens housed/year/farm without measure) Intervention Poland Spain Netherlands UK Denmark Norway Anteroom with hygiene barrier 0.12 0.34 0.16 0.29 0.11 0.76 Dedicated tools 0.26 0.26 0.18 0.13 0.12 0.27 Fly screens 1 - - - - 0.83 - Maximal downtime 10 days 2-21.98 2.08 12.69 14.69 18.39 - Drink nipples without cup 14.93 16.46 18.10 19.04 20.48 25.91 Slaughter at 35 days 2 35.64 118.38 74.13 118.13 61.55 - Stop thinning 28.26 35.56 39.05 34.81 48.41 58.08 New houses 73.10 97.76 57.45 164.69 126.90 174.73 1 only estimated for Denmark (effectiveness only provided for Denmark) 2 not estimated for Norway (average delivery at 33 days; legal maximum number of broilers per year) Big difference in intervention costs between countries High: Norway, Denmark Low: Poland source: Van Wagenberg & Van Horne2016 Microbial Risk Analysis

Cost-effectiveness interventions DALY / case = 0.027 source: Van Wagenberg et al 2016 Microbial Risk Analysis ES PL ES UK ES ES PL ES PL UK NL PL NL UK NL NO PL UK UK NL NL ES NO Expensive ES PL PL UK UK NL NL NO NO NO 55.00

CHEAP Cost-effectiveness interventions ES PL ES UK ES ES PL ES PL UK NL PL NL UK NL NO PL UK UK NL NL ES NO PL ES UK PL NL NL UK NO NO NO 55.00

Cost-effectiveness interventions ES PL ES PL ES ES ES ES PL PL PL ES UK PL UK UK UK UK UK NL NL NL PL NL NL NL UK NL NO Low NO risk reduction NO NO NO 55.00

Cost-effectiveness interventions High risk reduction ES PL ES UK ES ES PL ES PL UK NL PL NL UK NL NO PL UK UK NL NL ES NO PL ES UK PL NL NL UK NO NO NO 55.00

Cost-effectiveness interventions Cheap ES ES PL and PL ES ES PL effective ES UK PL UK UK UK NL PL NL NL UK NL NO NL ES NO PL ES UK PL NL NL UK NO NO NO 55.00

Conclusions Big differences in costs and cost-effectiveness of interventions between countries Interventions with low costs/daly Anteroom with hygiene barrier Designated tools Interventions with high costs/daly Stop thinning New houses Slaughter at 35 days Drink nipples without cup

Options for control Interventions at the farm Interventions at the slaughterhouse Improving consumers hygiene Microbiological criteria

Interventions at the slaughterhouse Defeathering and evisceration are critical stages Complex chaotic process of crosscontamination, removal and survival Decontamination is not allowed But effects can be studied anyway

Relative risk reduction Effect of concentration reduction on human health risk in published risk assessments Product freezing 100.00% Dec. before chil. (TSP) Dec. before chilling (L) 90.00% EO water Dipping Crust freezing Initial load 2 cfu 80.00% 70.00% Crust freezing 60.00% 50.00% Havelaar et al. (2007), Nauta et al. (2005) Lake et al. (2007) Rosenquist et al. (2003) Brynestad et al. (2008) Lactic ac. 40.00% 30.00% Gellynck et al. (2008) Linqvist and Lindblad (2008) WHO-FAO (2009) Tank dec. (TSP) -2.50-2.00-1.50-1.00-0.50 0.00 Log reduction (cfu) Tank dec. (L) Initial load 6 cfu 20.00% 10.00% 0.00% source: EFSA opinion on Campylobacter control 2011 Red lines indicate EFSA model predictions

Intervention by reduction of the Campylobacter concentration Caecal concentration in live birds Vaccination Phage therapy Other methods Faecal contamination at early processing stages (scalding, defeathering, evisceration) Various methods Farm At the end of processing Decontamination Other methods Processing scalding defeathering evisceration washing chilling Cutting breast cap double filet filet

Intervention by reduction of the Campylobacter concentration in caeca 1) Caecal concentration in live birds Vaccination Phage therapy Other methods Numerous studies are done No convincing evidence from field trials Risk assessments: Large variability in results One log reduction in caecal concentration gives less than one log reduction in concentration on the meat No rule of thumb 2 log reduction may give > 60% risk reduction (Nauta et al, Microbial Risk Analysis 2016)

Intervention by reduction of the Campylobacter concentration at early stages 2) Faecal contamination at early processing stages (scalding, defeathering, evisceration) Various methods Risk assessments: Cross-contamination and survival are complex non-linear processes Carcass contamination by faeces at later stages may obstruct the effect of interventions at earlier stages These interventions may be effective, but quantification is difficult

Intervention by reduction of the Campylobacter concentration at late stages 3) At the end of processing Decontamination Other methods Risk assessments: In principle the risk reducing effect would be large: 1 log reduction may give > 80% risk reduction Challenges Decontamination is not allowed Could be used to hide bad practices Consumers don t like it Quality of product may be affected

Options for control Interventions at the farm Interventions at the slaughterhouse Improving consumers hygiene Microbiological criteria

Interventions at the consumer phase Consumers are out of control The fact that people get ill means that something goes wrong during cooking and hygiene What happens at the consumer stage? How does exposure depend on the contamination of the meat? Can we effectively educate consumers?

The challenges Every food chain risk assessment should consider the consumer phase Consumers store, cook and handle their foods in many ways We have few data We cannot control consumers We need it to assess the effect of interventions on the risk

Chapman et al. Microbial Risk Analysis 2-3, 2016

Comparison of Consumer Phase Models for Campylobacter Q ill 30% 25% 20% 15% 10% Lindqvist FAO/WHO Christensen Mylius Nauta Van Asselt Brynestad Calistri 5% 0% -2 0 2 4 6 log C ret (cfu/g) source: Nauta and Christensen, Risk Analysis, 2011

Some conclusions Different performance, but Predicted relative risk reductions consequential to control measures are not that different But they may be if you affect the tail of the distribution of concentrations (!) The Nauta model performs nicely intermediate Nauta et al, Risk Analysis 2008 based on observational data from consumers fair choice as a representative model?

Options for control Interventions at the farm Interventions at the slaughterhouse Improving consumers hygiene Microbiological criteria

Microbiological criteria (MC) A criterion defining the acceptability of a product, a batch of foodstuffs or a process, based on the absence, presence or number of microorganisms, and/or on the quantity of their toxins/metabolites, per unit(s) of mass, volume, area or batch. Legal instrument Action against (heavily) contaminated batches (or their producers) Identifies highest risk products May result in risk reduction 2011: Evaluation of potential impact of MCs (EFSA) 2018: Process hygiene criterion in place in the EU

An instrument to control food safety MC defines a sampling plan for a microbial hazard, for a food product at a point in the food chain with a treshold value for a test result defining acceptability e.g. n=5 samples of which c=1 may contain > m=1000 cfu/g Defines action to be taken when criterion is not met What is the impact of an MC? How many batches are not complying? What is the impact of removal of these batches from the market on the public health risk?

Risk based MCs: Evaluate Microbiological criteria in the food chain by risk assessment Farm scalding MC Processing defeathering evisceration washing chilling MC Cutting breast cap double filet filet MC Microbiological criteria Storage Consumer preparation Risk Assessment Ingestion / Dose response Human health risk

Establish risk-based MCs Define a potential MC n, c, m, sample point, sample type, action Define sampling frequency and action taken Estimate current prevalence and distribution of concentrations e.g. on the basis of the EU baseline study 2008 Estimate % of non-complying batches Estimate relative risk reduction that may be achieved Nauta, Sanaa and Havelaar, 2012 Int. J. Food Micr.

Risk managers can decide upon the best Microbiological criterion

Risk-based Microbiological criteria Risk assessment can be used to inform the impact of setting specific criteria Uncertainty in the assessment Assumptions are needed Other factors play a role as well Costs International politics There is a tool to help you do this tools.food.dtu.dk/trimicri/ See Seliwiorstow et al, 2016 Microbial Risk Analysis

Conclusions and perspectives

What risk assessment has achieved Unique possibility for analyses over the farm to fork continuum A structured integrated approach the assess the potential effects of interventions on public health risk Increased awareness of the complexity of the problem Insight in potentially effective interventions Establishment risk based microbiological criteria is possible

Options for control Interventions at the farm No simple efficient solution, strict biosecurity is key Interventions at the slaughterhouse Concentration reduction may be effective, but hard to achieve Improving consumers hygiene Difficult, education hardly works Microbiological criteria May be a helpful instrument

New developments Update of the dose-response relation (Teunis et al 2018, Epidemics) Modifies risk estimates EU process hygiene criterion Impact? New data? Genome sequencing and big data Source attribution Dynamics of different strains (infection, transfer, survival) Identification of virulent strains?

What we need Campylobacter research Consider the use of new data before you collect them Take advantage of risk assessments done Learn from mistakes of risk assessments Interventions Biosecurity is key Focus on concentration reduction Be innovative

Acknowledgements Many collegues from CARMA project (2005) Consumer project ZonMw (2008) MedVetNet WP24 (2009) EFSA working group Campylobacter (2010) CamCon (2015) You