Pathogen Reduction Dialogue Pathogen Reduction Dialogue Panel 3 Panel 3

Microbial Testing for Control Microbial Testing for Control VerificationVerification

Robert L. BuchananU.S. Food and Drug AdministrationCenter for Food Safety and Applied

NutritionMay 7, 2002

Microbiological TestingMicrobiological Testing Microbiological

testing important food safety tool

Technologically-based, statistically-based tool

Right tool for the right job

Microbiological TestingMicrobiological Testing One of the most apparent, but poorly

understood parts of food microbiology Different types of microbiological

testing Safety of “batches” * Process control * Investigational sampling Surveillance

Microbiological TestingMicrobiological Testing Effective use

requires a clear understanding of the goals, assumptions, and characteristics of testing programs

Microbiological TestingMicrobiological Testing“Within batch testing” vs.

“Between batch testing”

Different goals, assumptions, and techniques

Microbiological TestingMicrobiological Testing“Within

Batch Testing”: Demonstrate the safety of a single lot of food

Microbiological TestingMicrobiological Testing Within-batch testing

detailed “snapshot” of an operation assumes no prior knowledge of the

process focus on establishing safety (or

quality) of the batch provides only limited capability to do

trend analysis of performance over time

can be used nationally to acquire “state of industry”

Microbiological TestingMicrobiological Testing Effective within certain ranges

of contamination frequency or levels of contamination Above or below those ranges,

becomes increasingly ineffective When acceptable defect rate is

less than 1%, the number of samples needed becomes a limiting factor

Microbiological TestingMicrobiological Testing “Between Batch Testing”: Demonstrate

that a food safety system/process is continuing to function as intended

Microbiological TestingMicrobiological Testing “Between-batch testing”

not designed to assure batch safety which is assumed to be safe if validated process is “in control”

assumes intimate knowledge of process

requires prior analysis for performance and variation

requires sampling over time can also be used nationally to assess

“state of industry”

Microbiological TestingMicrobiological Testing“Between-batch testing”

Much easier to demonstrate that a process is not functioning within a specification than to prove that something (e.g., a pathogen) is not present

Sampling for Process Sampling for Process VerificationVerification Microbiological contamination

typically flows with a process A microbiological sample taken

within a process provides a measure that integrates all the preceding steps in a process

Sampling for Process Sampling for Process VerificationVerification The status of a multiple step

process is the summation the initial level of contamination and all the steps that increase or decrease the level of contamination

Microbial Status = Ho + Microbial Status = Ho + I + I + RR

Sampling for Process Sampling for Process VerificationVerification Sampling of end products integrates the

effects of the entire food safety system For investigations of cause, benefits to

taking samples at multiple locations

Process Control StatisticsProcess Control Statistics The basis for

process control evaluations is the collection of microbiological data over time

Typically arrayed graphically as a control chart

0.00.20.40.60.81.01.21.41.61.8

0 10 20Sample Number

Log

(cfu

/g)

Process Control StatisticsProcess Control Statistics First step is to conduct a process

control study (baseline study) An “under control” process is run for a

period of time to access its “capabilities” and establish:

Central tendency (mean or median performance)

Variance Also used to establish limits (control

values) Often use 3

Process Control StatisticsProcess Control Statistics Loss of process control is then

assessed by determining if “defect rate” has become greater than what would be expected by “chance alone”

Approach can be used with either “variables” (quantitative data) or “attribute” (+/- or binned quantitative data) sampling plans

Example: Moving Window Example: Moving Window SumSum One of simplest for

attribute sampling plans

+ - + - - - - + - - - - - + - -

Moving Window SumMoving Window Sum

Example: Making red marbles

ReceiveBlue Marbles

Paint MarblesRed*

Package Marbles

*Defect rate = 10%

Moving Window SumMoving Window Sum Based on the

probability of finding more than an expected number of “defective” responses within a specified window of samples

Moving Window SumMoving Window SumIs this process under control?(Sampling rate = 1 per 1000)

Moving Window SumMoving Window SumWhen is control of this process lost?(Sampling rate = 1 per 1000)

Operating Characteristics

0

20

40

60

80

100

0 10 20 30 40 50 60 70

Actual % Positive

Prob

abili

ty o

f Pas

sing

(%)

99%90%80%

Impact of StringencyImpact of Stringency

JuiceJuiceHACCPHACCP

Juice HACCPJuice HACCPKey feature:

5-D (99.999%) performance standard

Restricted to juice after it has been expressed

Verification of this requirement is via process validation and review of process records

Goal: Limit risk to >10-5/year

Juice HACCPJuice HACCP Microbiological

testing not required based on ineffectiveness of

testing at low defect rates

treatment affects all juice

processes are validated and reliable

A pple Juice

Microbiological TestingMicrobiological Testing Simplified example

Assume 1 enteric bacteria per ml 5-D treatment reduces to 1 per 10,000

mL To detect survivor, need

1 10-L sample 10 1-L samples 100 100-mL samples 1000 10-mL samples 10,000 1-mL samples

Juice HACCPJuice HACCP Key exemption: Citrus Juices

Processors of citrus juices may count surface treatments of the fruit as partially or totally fulfilling 5-D treatment

Based on the underlying assumption that the nature of citrus fruit is such that contamination is restricted to the fruit’s surface

Juice HACCPJuice HACCP These citrus juice processors have

an additional HACCP verification requirement of periodic testing for E. coli Two 10-ml juice samples/1000

gal/day Once per week if produce <1000

gal

Juice HACCPJuice HACCP Data evaluated using process

control statistics using a 7-sample window

One positive sample requires process review

Two positive samples require diversion to 5-D after extraction until cause identified

Juice HACCPJuice HACCP Designed to verify that the original

assumption is still valid, i.e. pathogens are restricted to the surface of the fruit Internalized pathogens will not be

treated Potential for growth at least in some

fruit Within range for effective detection Requirement takes into account

potential for chance contamination

Microbiological TestingMicrobiological Testing Microbiological

testing can be an integral part of any integrated program for verifying the effectiveness of food safety control measures