1

Risk-Based Inspection

Philip Derfler Assistant Administrator, Office of Policy, Program and

Employee Development

&

Don Anderson Office of Program Evaluation, Enforcement & Review

2

Purpose of Inspection

Traditional Approach Designed to find

problems if they occur

Risk-Based Approach Designed to find

problems that occur Designed to anticipate

problems, thereby minimizing risk

3

Deployment of Resources

Traditional Based on what needs

to be done Inspecting carcasses Making inspection

once per plant per shift

Risk-Based Align resources also

with level of risk: Hazards based on type

of species and process Likelihood hazard will

exist in plant Significance of effects Ongoing assessment

of food safety system

4

Work to Be DoneTraditional Perform assigned

procedures Under HACCP, 70%

food safety and 30% other procedures.

Risk-Based Varies based on risk

Basic procedures Others based on

inspectional findings and decision criteria

5

Activities of Inspection PersonnelTraditional Procedures designed

to find non-compliance

Risk-Based Focused on where loss

of control: Is more likely to occur Would have serious

public health consequences

Intensify inspection if plant is losing/has lost control

6

Response to Inspection FindingsTraditional Evidence of compliance

or non-compliance has no effect on intensity of inspection.

Evidence of non-compliance could lead to enforcement action.

Risk-Based Evidence of non-

compliance could lead to enforcement action.

Intensity of inspection based on findings: Good control = less

intense inspection Losing control =

intensified inspection

7

Response to Emerging ProblemTraditional Not designed to have

inspectors make a judgment about risk of non-compliance

Risk-Based Inspection personnel

would have flexibility, data, responsibility and training to be able to focus on emerging problems

8

Factoring in Food DefenseTraditional Food defense

procedures performed with specified frequency

Risk-Based Food defense

procedures performed at frequency appropriate for national security situation and security situation in establishment

9

Attention to Product in Commerce

Traditional Random visits to

facilities that handle meat and poultry products to ensure conditions are sanitary

Risk-Based Scheduled visits to

facilities based on findings and other information

10

Risk Control Components

Important Components of Risk ControlFood Safety System ImplementationFood Safety System DesignPathogen ControlIn-Commerce FindingsEnforcement ActionsOther Components

11

Food Safety System Implementation

System implementation consistencyFSIS documents all regulatory

noncompliances– and will continue to do so under RBI

However, not all NRs are equally indicative of risk control deficiencies

Our goal is to identify, enumerate, and properly weight public health-related NRs

12

Food Safety System Design

Efficacy of the food safety systemFood Safety Assessment Findings

When was the last FSA?What was the outcome?

13

Pathogen Control

Pathogen Control in Ready-to Eat Products, Ground Beef, and Other Raw Products

Lm, Salmonella, and E. coli O157:H7 RTE testing program results

E. coli O157:H7 (raw ground beef) testing program results

Salmonella verification testing program results

14

In-Commerce Findings

Adverse Findings In-CommerceSignificant Consumer Complaints?Class I or II Recalls?Other Considerations?

15

Enforcement Actions

Enforcement ActionsInvolving Food SafetyNot preceded by significant NRsNot initiated as a result of an FSA

16

Other Considerations

Other Serious Public Health ConcernsE. coli O157:H7 Positives Suppliers?USDA/AMS school lunch testing results?Others?

17

Questions for the Committee

Are these all appropriate objectives for measuring risk control? Should any objectives or corresponding features be deleted? Should any be added?

18

Questions for the Committee

Are some components more important—i.e. better indicators of risk control—than others? If yes, should more important components have greater “weight” in our numerical control measure than less important components?

19

Questions for the Committee

Should findings from Food Safety Assessments or other sources that indicate exceptionally effective risk controls be allowed to lower (improve) an establishment’s risk control measure?

20

FSIS RBI Announcement February 22, 2007

The Following Slides Were Prepared by Dr. Joe Blair of the HACCP Consulting Group, L.L.C. based on a Memorandum of this date by:

Olsson, Frank and Weeda, P.C.

Attorneys at Law

Washington, DC

21

FSIS RBI AnnouncementFebruary 22, 2007Begin “piloting” RBI in about 250

processing plants in 30 locationsPhase I

Selected ‘prototype’ locations in April 2007Phase II

Begin in January 2008

22

Ranking Plants – Phase I

Inherent Product RiskPlant’s ControlsFinal RankingInspection Intensity

23

Inherent Product Risk

Use Current Expert Elicitation for Phase I and New Elicitation for Subsequent Phases Each Product Assigned a Score of 1-20 Plants with Multiple Products – Use the Highest

Risk Product

Multiply Risk by Volume 5 Categories for Volume (highest volume = 5)

Plant Will Have a Product Score of Between 1 and 100

24

Plant Controls

Algorithm Criteria Enforcement Actions

Notice of Intended Enforcement (NOIE) = 3 points Suspension = 6 points

Recalls Class I = 3 points (Reasonable Probability of

Adverse Health Risk) Class II = 2 points (Remote Probability of Adverse

Health Risk) Consumer Complaints (verified by FSIS as

accurate) = 1-3 points

25

Plant Controls 2

Algorithm Criteria - Continued FSIS Microbiological Test Results

Positive E. coli O157:H7 = 3 points Positive Listeria monocytogenes = 3 points Maximum of 9 points for this category

RTE Regulatory Alternative (Listeria regulations) Alt. 1 (Post Lethality Treatment) = 0 points Alt. 2 (Post Lethality Suppression) = 2 points Alt. 3 (Sanitation Only) = 3 points

26

Plant Controls 3

Algorithm Criteria - Continued

Salmonella Performance Standards Verification CategoriesCat. 1 -- < 50 % of Standard = 0 pointsCat. 2 -- 51-100% of Standard = 2 pointsCat. 3 -- Exceeds Standard = 3 points

27

Plant Controls 4

Algorithm Criteria - Continued Non-Compliance Records (NRs)

Only Consider NRs having Public Health Significance

Rank Each NR on Severity (1-3) Compare Totals with Other Plants in Phase I Divide Plants into Quintiles (I/5th)

Top 20% = 5 points Bottom 20% = 1 point

28

Plant Controls 5

Algorithm Criteria - Continued

Once a Score is Generated by AlgorithmDetermine % for Plant by Using the Actual

Score Divided by the Maximum Score for any Plant in Phase I that Conducts the Same Type of Operation

29

Final Ranking

The Plant’s Final Score will be the Average of:

The Product Risk Divided by the Volumeand

Plant’s Control Score

30

Phase I Specifics

Ranking -- FSIS will NOT use the Precise Calculated Number -- Rather the Plants will be Grouped into 3 Categories:

Highest Risk Plants - Receive More Inspection Intensity

Middle Risk Plants – Receive about the Current Level of Inspection Intensity

Lowest Risk Plant – Receive Less Inspection Intensity.

31

Inspection Intensity

All Plants will Continue Receiving Daily Inspection Inspectors at Higher Risk Plants will have

Extra Scheduled Tasks to Perform (more time in the plant)

Inspectors Lower Risk Plants will have Fewer Scheduled Tasks to Perform (less time in the plant)

32

Phase II – 2008 ?

The Results from Phase I will be Utilized to Refine and Improve the Program Before Moving to Phase II

Acceptance by Industry, Consumers and Inspection Personnel Unknown at this Point.

Such an Approach Offers Opportunity for More Effective Use of Resources

Time Will Tell!