2. Innovative Modeling Approaches Applicable to Risk Assessments Thomas P. Oscar, PhD USDA, ARS Princess Anne, MD, USA

3. Risk Assessment  Hazard Identification  Hazard Characterization  Exposure Assessment  Risk Characterization Predictive Microbiology Food Safety Information

4. What came first?  Risk Assessment Model  Predictive Model I say, I say, son… its easier to fit round pegs into round holes

5. Hazards  Chemical  Physical  Microbial

6. Hazard Events  Rare  Random  Variable  Uncertain growth death survival removal contamination

7. Rare Events Modeling Iteration 1 2 3 : 100 Discrete 1 0 : 0 Pert (0,1,4) 1.8 1.2 0.2 : 2.2 Power 63.1 0 : 0 Round 63 0 : 0 =RiskDiscrete({90,10},{0,1}) =RiskPert(0,1,4) =Power(10,Pert) =Round(IF(Discrete=0,0,Pert),0)

8. 20% J. Food Safety (1998) 18:371-381 Unit Operation Hazard Event IncidentExtentPackaging Initial Contamination 20% 1 (0 – 3) log/bird DistributionGrowth20% 0.5 (0.1-3.0) logs RAREEVENTSMODELINGRAREEVENTSMODELING RISKASSESSMENTRISKASSESSMENT

9. 0.9% J. Food Safety (1998) 18:371-381 Unit Operation Hazard Event IncidentExtentCookingSurvival20% -1.5 (-2 to -1) logs

10. 7.0% J. Food Safety (1998) 18:371-381 Unit Operation Hazard Event IncidentExtentServingCross-contamination25% 2 (1 to 5)% transfer No correlation!

11. J. Food Safety (1998) 18:371-381 Normal Risk High Risk Unit Operation Hazard Event IncidentExtentConsumption Normal Risk 80% 750 (500-1000) cells High Risk 20% 200 (50 to 350) cells Healthy Old Anti-microbials High Fat Probiotic Clinical isolate

12. Relative risk of salmonellosis = (Dose Consumed ÷ Infection Dose) * 100 J. Food Safety (1998) 18:371-381 Higher risk!

13. Hazard Identification  Cornerstone  Expensive  Number and Subtype  Packaging

14. Microbial Ecology  Minority  Unattached  Attached  Entrapped

15. Standard incubation conditions Rare Events Model (Initial Contamination) Detection limit = 10 2 cells/ml Target pathogen (< 1/ml) Detection Time J. Food Prot. (2004) 67(6):1201-1208

16. Sample Control 02468101224Time, h Qualicon BAX™ PCR System for Salmonella spp. PCR Score = 12 None = 0 Faint = 1 <Full = 2 Full = 3 J. Food Prot. (2004) 67(6):1201-1208 10 2 10 3 10 4 >10 4 <10 2

17. Final Standard Curve 95% Prediction Interval J. Food Prot. (2004) 67(6):1201-1208

18. Predictive Model J. Food Prot. (2004) 67(6):1201-1208 Rare Events Model

19. Exposure Assessment Develop predictive models for hazard events from hazard identification to consumption GrowthSurvival Cross-contamination Physical Removal

20. Real Food + Microbial Competition + Low Dose  MPN  CFU J. Food Prot. (2006) 69(9):2048-2057. J. Food Prot. (2008) 71(6):1135-1144. J. Food Prot. (2009) 72(2):304-314. J. Food Prot. (2009) In press

21. General Regression Neural Network (GRNN) Model Rare Events Model J. Food Prot. (2009) in press

22. Hazard and Risk Characterization Severity of Illness Infected Mild Illness Doctor Severe Illness Hospital Chronic Disability Death

23. Hazard Characterization Uniform Pathogen Food Host Human feeding trials are no longer ethical!

24. Risk Anal. (2004) 24(1):41-49. Rare Events Model

25. Risk Anal. (2004) 24(1):41-49.

26. Disease Triangle Modeling Hazard Host Food -1 log -2 log -0.5 log Very young Very old Cancer Diabetes HIV Pregnant : Top clinical isolate Acid resistant : High fat Anti-acid : Oscar, book chapter, in press High Risk

27. Disease Triangle Model Rare Events Model

28. Relative versus Absolute Risk 0% Absolute 100% Absolute Human feeding trials are not ethical! 100% Relative 0% Relative

29. Scenario Analysis  Plant A  Plant B Oscar, book chapter, in press

30. Risk Pathway Packaging (Contamination) Distribution (Growth) Washing (Removal) Cooking (Survival) Serving (Contamination) Consumption (Dose-response) I see only one risk pathway

31. Module A90%10%Plant B Oscar, book chapter, in press Rare Events Model

32. Module B Oscar, book chapter, in press Rare Events Model

33. Risk Assessment Results Your fired! n = 200 replicate simulations per scenario

34. I see two risk pathways I see data gaps! Hazard strain Time & Temp Predictive Models Consumer Surveys Packaging (Contamination) Distribution (Growth) Washing (Removal) Cooking (Survival) Serving (Contamination) Consumption (Dose-response)

35. Research Results Plant A Plant B Initial Contamination 25%10% Temperature Abuse 20%40% Washing15%30% Proper Cooking 90%90% Cross-contamination15%30% High Risk Food 10%10% High Risk Pathogen 20%60% High Risk Host 20%30%

36. Filtered Results

37. Exposure Assessment Oscar, book chapter, in press

38. Hazard Characterization Oscar, book chapter, in press

39. Risk Characterization  Single Risk Pathway  Multiple Risk Pathways

40. Unsafe Safe Single Risk Pathway

41. Multiple Risk Pathways Unsafe Safe

42. The End Thank you for your attention! I will be glad to answer your questions