AOAC RI ERP Final Action Recommendations

AOAC INTERNATIONAL

470

Study Design

– This study used the Probability of Detection (POD) statistical treatment of

471

data from the current AOAC validation guidelines [5], but the study design was approved

472

using the previous AOAC guidelines [6]. In addition to the POD treatment of data, Chi

473

square analysis was also performed. For unpaired samples the Mantel-Haenszel chi-

474

square test was used, while for paired samples the McNemar chi-square test was applied.

475

A Chi square value of less than 3.84 indicates no significant difference, at the 95%

476

confidence level, between the two methods.

477

478

A) The Mantel-Haenszel chi-square test is as follows for a 2x2 table, where n=a+b+c+d,

479

a = samples positive by the alternative method and culture confirmed, b = samples

480

negative by the alternative method and either positive or negative by the reference

481

method, c = samples positive by the reference method, and d = samples negative by the

482

reference method. χ

= (n-1)(ad-bc)

/[(a+b)(a+c)(b+d)(c+d)]

483

484

B) Chi-square as defined by McNemar – χ

= (|a-b |-1)

/(a+b) where a = samples positive

485

by BAX and negative by the reference method, and b = samples negative by BAX and

486

positive by the reference method.

487

488

489

Results and Discussion

– The results for sample types tested internally and externally

490

are summarized in Tables 2-5 above. For ground beef, lettuce, pet food and poultry

491

carcass rinse samples, the BAX® System returned results identical to the reference

492

culture method. For cream cheese samples tested internally, POD and Chi-square

493

analyses demonstrate that the difference in results between the two methods (2/20

494

positive for the BAX® System method using MP media, and 5/20 positive for the BAX®

495

System using the reference method LB enrichment and 5/20 culture positive samples for

496

the reference method) was not statistically significant.

497

For environmental samples tested internally, the BAX® System returned results identical

498

to the reference method with the exception of one result for samples enriched in BPW

499

Dupont BAX Salmonella PTM Report

PTM Certification No. 081201 (08/07/2012)

For Expert Review Panel Use Only

Do Not Distribute