AOAC-OMA/MicroVal Harmonized Validation of Peel PlateTM EB (Enterobacteriaceae Bacteria), First Action 2018.05

Abstract Background Peel PlateTM Enterobacteriaceae Bacteria (EB) is dried selective media on a 47 mm plastic plate that produces enzyme substrate colored colonies on rehydration and incubation for 24 h and up to 48 h at 37 ± 1°C. Purpose The method validation compared quantification of EB to reference methods ISO 21528:2017 Parts 1 and 2. Methods Matrixes compared were whole milk, skim powdered milk, vanilla ice cream, butter, infant formulas (soy- and dairy-based), infant cereals ± probiotic, environmental sponge swab of stainless steel surface, and poultry carcass rinse with two different peptone buffers. Results In inclusivity and exclusivity studies, the method detected 54 of 54 EB strains and did not detect 30 of 30 non-EB strains. In matrix studies, the claimed foods were tested at three contamination levels using paired analysis between the reference and Peel Plate EB methods. Colony-forming units per gram or mL [CFU/g (mL)] were log10 transformed for statistical analysis. The candidate method and reference method were shown to be equivalent by the performance requirement of all 95% confidence intervals on mean difference falling between −0.5 and +0.5 log10 CFU/g (mL). An international collaborative study with dried infant formula spiked with Cronobacter sakazakii at log10 CFU/g (mL) 1.05, 2.31, and 3.21 levels, produced method differences −0.16, 0.15, and 0.18 log10 CFU/g (mL) with repeatabilities (r) = 0.33, 0.20, and 0.12 log10 CFU/g (mL) and reproducibilities (R) = 0.45, 0.26, and 0.18 log10 CFU/g (mL). Conclusions Based on these evaluations, the candidate method is considered equivalent to the reference methods at both the 24 h and 48 h incubation periods at 37 ± 1°C. Highlights Ready to use Enterobacteriaceae method equivalent to ISO-21528:2017 Parts 1 and 2; EB test colored colonies at 37°C for 24 h are equivalent at 48 h incubation; Singlet determined CFU/mL are statistically the same as duplicate average results; EB test validated for infant formula and dairy products including with probiotics; EB test for environmental surfaces and poultry carcass rinses using peptone buffers.

The Enterobacteriaceae is a family of Gram-negative, nonsporeforming bacilli bacteria and is one of the most important groups of bacteria known that are found in soil and water, as well as in plants and in animals (both vertebrates and invertebrates). They may be motile or nonmotile, depending on species. They are aerobic or facultatively anaerobic in growth and tend to inhabit the gastrointestinal tract.
Among the most notable foodborne pathogens and spoilage organisms are Escherichia, Salmonella, Enterobacter, Klebsiella, Citrobacter, Cronobacter, Shigella, and Yersinia. Methods for the detection and enumeration of Enterobacteriaceae have changed very little since they were first introduced and many still rely on the growth of the bacterium in selective media along with the use of carbohydrates as an energy source (1). Because Enterobacteriaceae are used so frequently by the food industry, there are needs for simple, low cost, ready-to-use methods for testing. Peel Plate EB is a simple method to detect and quantify Enterobacteriaceae in foods which is studied and validated in this work.
In the study, the target organisms are bacteria in the family Enterobacteriaceae that comprise a broad number of Gram-negative bacteria. Performance testing of heat-processed milk, dairy products, infant formula, cereals, stainless-steel surfaces, and chicken carcass rinses are not statistically different between candidate and reference methods. Statistical difference is determined from CFU/mL results log 10 transformed and all 95% confidence intervals on mean difference between candidate and reference methods falling between À0.5 and þ0.5 log 10 CFU/g (mL) (2)(3)(4). Peel Plate EB is the candidate method and reference methods are ISO 21528-1:2017 Microbiology of the food chain-Horizontal method for detection and enumeration of Enterobacteriaceae-Part 1: Detection (5) and ISO 21528-2:2017 Part 2: Enumeration (6). [Applicable to the enumeration of Enterobacteriaceae from pasteurized whole milk, butter, nonfat dry milk, vanilla ice cream, powdered and liquid infant formula (milk-based) containing probiotic, nonprobiotic liquid infant formula (soy-based), infant cereal with probiotic, infant rice cereal without probiotic, chicken carcass rinse with neutralized buffered peptone water, chicken carcass rinse with buffered peptone water, and stainless-steel surfaces.] Caution: Perform tests with clean, washed, and gloved hands assuming potential pathogenic bacteria. Microbiological cultures and reagents should be collected into biohazardous bags and autoclaved. Dispose according to local, state, and federal regulations.

A. Principle
Peel Plate V R EB test is used for the detection and enumeration of Enterobacteriaceae bacteria in food and environmental samples. The method is applicable for the determination of Enterobacteriaceae in samples when incubated at 37 6 1 C for up to 24-48 h. All visible colonies, regardless of color, on the Peel Plate are to be considered an Enterobacteriaceae. The method limit of detection is 1 or greater CFU per milliliter or gram of test sample. The accurate quantitative range for Enterobacteriaceae is 1 to 150 CFU per plate.
The Peel Plate EB test is based on bile salt selective agar, glucose, and multiple colorimetric enzyme substrates to support growth and colormetrically identify the growth of the family of Enterobacteriaceae bacteria. The media also contains gelling and wicking agents which absorb and diffuse the sample.

D. General Preparation
(a) Observe Good Laboratory Practices for microbial testing. Avoid specimen contamination. (b) Test on a level surface, in a clean area, and free of dust and blowing air. (c) Avoid hand contact with test samples and Peel Plate EB medium. (d) Log serially dilute sample into BPW, Butterfield's, or MS water to obtain the countable range 1-150 CFU/plate or test multiple dilutions to attain the countable range.
Sample stainless-steel surfaces by rehydrating a sponge with 25 mL BPW, rinsing aseptically, swabbing a 100 cm 2 surface, adding the sponge to buffer, and stomaching for 1-2 min. (c) Chicken carcasses.-The sampling protocol followed FSIS Directives for chicken carcass (8,9). Add 400 mL BPW or n-BPW to a bag with a chicken carcass, seal and shake bag for 1 min (10

H. Confirmation
The Peel Plate EB method uses selective medium and enzyme substrates to detect Enterobacteriaceae without the need for confirmation steps. Although it is not necessary, it may be desired to confirm colonies on traditional selective medium. The cover may be lifted and colonies picked and streaked onto violet red bile agar with glucose (VRBAG) broth. To confirm Enterobacteriaceae, isolates should be tested for oxidase activity and stabbed into glucose agar containing bromocresol blue and covered with sterile immersion oil. Oxidase negative samples that acidify glucose agar to produce yellow stab are confirmed EB. Enterobacteriaceae confirmation procedures are described in ISO protocols (5, 6). Method developer studies were conducted in at Charm Sciences, Inc. (Lawrence, MA, USA) and included supplemental matrix data for additional claimed matrixes, product consistency and stability studies, and robustness testing.

Precollaborative Validation Study
The independent precollaborative laboratory study was conducted by Q Laboratories, Inc. (Cincinnati, OH, USA) and included the inclusivity/exclusivity studies and matrix studies for the claimed food/and or surface matrixes. Q Laboratories prepared samples and coordinated an international eleven laboratory collaborative study of powdered infant formula containing probiotic. The

Inclusivity and Exclusivity Studies
The inclusivity and exclusivity evaluations were conducted at Q Laboratories. All test materials required for the Peel Plate EB method were provided by Charm Sciences, Inc.
(a) Methodology.-For the inclusivity evaluation of the Peel Plate EB, 50 Enterobacteriaceae were cultured in BPW (ISO) broth at 37 6 1 C for 24 6 2 h. The 30 exclusivity organisms were cultured in brain heart infusion (BHI) broth at 37 6 1 C for 24 6 2 h. The inclusivity and exclusivity organisms were serially diluted in 0.1% BPW to approximately 100 CFU/mL. All samples were blind-coded and randomized and analyzed by the Peel Plate EB method and ISO 21528-2. One milliliter of each culture was plated in duplicate. All plates were incubated at 37 6 1 C for 24 and 48 h. Colonies were enumerated. (b) Results and Discussion.-Tables 1 and 2 show details of the inclusivity/exclusivity bacterial study strains, respectively. Table 1 demonstrates that of 54 Enterobacteriaceae

Precollaborative Matrix Study
Precollaborative matrix studies were conducted at Q Laboratories and at Charm Sciences, Inc. In these studies, each claimed matrix was evaluated naturally and at three contamination levels. The study outline adhered to Appendix J of the Official Methods of Analysis of AOAC INTERNATIONAL (11). Each food matrix was purchased from a local distributor, and prescreened for natural contamination of the target analyte by the ISO 21528-2:2017 reference method. Following the screening, each matrix tested by the validation laboratory was inoculated with a different strain of Enterobacteriaceae as indicated in Table 3. Additional matrixes were performed by Charm Sciences, Inc.
(a) Methodology.-The precollaborative comparison study consisted of evaluating a total of 20 paired sample replicates for 3.25% pasteurized whole milk, nonfat dry milk powder, infant formula with probiotic, stainless steel, and chicken carcass rinse. In the case of infant cereal with probiotic, the candidate method called for a greater dilution in preparation than the reference method, so unpaired samples were used.
Within each food matrix sample set there was an uninoculated level and three target inoculation ranges: five uninoculated samples (0 CFU/mL), five low-level inoculated samples (10-100 CFU/mL), five medium-level inoculated samples (100-5000 CFU/mL), and five high-level inoculated samples (5000-100 000 CFU/mL). In all matrix studies except chicken rinse, which had natural contamination, Enterobacteriaceae strains shown in Table 3 from cultures were spiked and acclimated in products for 48 to 72 hours before testing. The acclimated material was quantified using the ISO method and then used for creating fortification levels. Each inoculum was prepared by transferring a single colony from trypticase soy agar with 5% sheep blood (SBA) into BHI broth and incubating the culture at 35 6 2 C for 24 6 2 h. Following incubation, the culture was diluted to a target level using BHI as the diluent. For each inoculated food matrix, bulk portions were spiked and blended in large, sterile stainless-steel containers. Sterile spatulas were used to mix the bulk portions to ensure the inoculum was evenly distributed throughout the matrix. The 3.25% pasteurized whole milk was held for 48-72 h at refrigerated temperature (2-8 C) prior to analysis to allow time for the organism to equilibrate within the sample. For nonfat dry milk powder, infant formula with probiotic, and infant cereal with probiotic, a lyophilized inoculum was used to inoculate a bulk lot of each matrix and was then homogenized and held at ambient temperature (20-25 C) for 2 weeks. Prior to inoculation of 3.25% pasteurized whole milk, the broth culture inoculum was heat stressed in a water bath for 10 6 1 min at 50 6 1 C. The degree of injury of each culture was estimated by plating an aliquot of diluted culture onto violet red bile (VRB) agar and tryptic soy agar (TSA). The agars were incubated at 35 6 1 C for 24 6 2 h and the colonies enumerated. The percent of injury was estimated: where n select ¼ number of colonies on selective agar and n nonselect ¼ number of colonies on nonselective agar. Stainless-steel and sealed concrete surfaces were evaluated after artificial contamination. Each test portion area (4 Â 4 in.) was evenly inoculated with 250 mL Salmonella enterica subsp. enterica serovar Typhimurium ATCC (American Type Culture Collection, Manassas, VA, USA) 14028 diluted in BHI and allowed to dry for 16-24 h at ambient temperature (20-25 C). The environmental surface was sampled using horizontal and vertical sweeping motions. Sampling sponges were held for a minimum of 2 h at ambient temperature prior to analysis. To determine the inoculation level for the environmental surface, aliquots of each inoculating organism was plated in duplicate onto TSA and enumerated.
Chicken carcass rinse was positive for natural contamination Enterobacteriaceae. Different lots of the matrix were purchased and screened to identify varying contamination levels. Lots were then mixed to produce three levels of contamination. The chicken carcass rinse was evaluated using naturally occurring Enterobacteriaceae. Within these sample sets, there were five replicates evaluated at a low contamination level targeting 10-100 CFU/g, five replicates evaluated at a medium contamination level targeting 100-1000 CFU/g, and five replicates evaluated at a high contamination level targeting 1000-10 000 CFU/g.
(b) ISO 21528-1:2017 (low levels of contamination; <100 CFU/g or mL).-Using the paired test portions, 25 g test portions were combined with 225 mL BPW (ISO) and homogenized by stomaching for 2 min 6 15 s. From each sample, 10 mL of the initial 1:10 dilution was transferred into three separate test tubes (10 À1 ). A 1 mL transfer of the initial 1:10 dilution was transferred into three test tubes (10 À2 ) containing 9 mL BPW (6). One additional dilution was performed by transferring 1 mL of the 10 À2 dilution into each of the three test tubes (10 À3 ) containing 9 mL BPW (5). All tubes were incubated at 37 6 1 C for 18 6 2 h. Following incubation of the tubes, all tubes were streaked onto violet red bile agar with glucose (VRBAG) agar and incubated at 37 6 1 C for 24 6 2 h. After incubation, all plates were examined for typical Enterobacteriaceae colony morphology. Up to five characteristic colonies were streaked to TSA and incubated at 37 6 1 C for 24 6 2 h. From an isolated colony from each of the TSA plates, a spot oxidase test was performed. For each oxidase negative colony, a stab to Glucose OF Medium and an overlay of sterile mineral oil was added. All Glucose OF Medium tubes were incubated at 37 6 1 C for 24 6 2 h. If a yellow color developed, the reaction was considered  (12). (c) ISO 21528-2:2017 (medium to high levels of contamination; >100 CFU/g or mL).-Using the paired test portions, a 25 g test portion was combined with 225 mL of 0.1% BPW and homogenized by stomaching for 2 min 6 15 s. Further 1:10 serial dilutions were conducted in order to achieve the desired target concentrations. A 1 mL aliquot of each dilution was plated in duplicate and 10 mL of tempered VRBAG agar was added to each plate. After the plates were completely solidified, an overlay of approximately 8-15 mL VRBAG was added to each plate. All plates were incubated at 37 6 1 C for 24 6 2 h. Following incubation, plates containing <150 pink to red and purple CFU were enumerated. The average CFU of the duplicate plates was recorded and multiplied by the dilution factor (reciprocal of dilution) and reported as total Enterobacteriaceae CFU/g or mL. Up to five typical colonies were streaked to TSA and incubated at 37 6 1 C for 24 6 2 h. A spot oxidase test was conducted for each plate, all oxidase negative colonies were stabbed into Glucose OF Medium with an overlay of sterile mineral oil. All Glucose OF Medium tubes were incubated at 37 6 1 C for 24 6 2 h. If a yellow color developed, the reaction was considered positive. (d) Peel Plate V R EB method.-All matrixes were diluted according to the AOAC protocol as described previously in "Method Procedure." After dilution, all test portions were plated following the Peel Plate EB method or in the case of cereal also the Peel Plate EBHV V R (high volume 5 mL) method.
Statistical analysis was conducted for each contamination level for each matrix evaluated comparing the Peel Plate EB method to the ISO reference method (2)(3)(4). Logarithmic transformations of the counts [CFU/g (mL)] were performed, and the difference of means, with 95% confidence intervals, between the candidate method and the reference method was determined for each contamination level. Mean difference and confidence intervals were calculated using the Independent Laboratory Study Workbook for Paired Method Analysis for Micro Testing (Version 1.0) supplied by the AOAC Research Institute (2). A mean difference between methods of <0.5 log 10 CFU/g (mL) with a 95% confidence interval (CI) containing values between [À0.5 log 10 CFU/g (mL), 0.5 log 10 CFU/g (mL)] was used as guidance to determine statistically significant differences between two methods being compared. The repeatability (s r ) of the Peel Plate EB and ISO reference methods were determined for each matrix.
(e) Results and Discussion.-Tables 4-7 are summary tables of evaluated matrixes, showing the spiked bacteria or natural contamination log 10 CFU/g (mL) levels evaluated, and the resulting mean averages and s r from five paired results between the Peel Plate EB and reference methods. The tables include mean differences associated between the candidate and reference with the confidence limits and correlation coefficient, r 2 , of the mean linear regression curve. Tables 6 and 7 present the 48 h Peel Plate EB singlet and duplicate test result compared to the reference. In all analyses, the confidence limits of the candidate method differences with the reference are within 0.5 log 10 CFU/g (mL) and indicate no significant differences with the reference methods. Duplicate analysis compared to singlet analysis produces very little change to the mean differences or the confidence limits. The 24 h analysis statistics are comparable to 48 h analysis showing very little recovery benefit, if any, of the additional 24 h incubation. In all there were 11 matrixes studied with nine different strains of spiked EB and two with natural contamination. In every evaluation the Peel Plate EB method demonstrates equivalence to the reference methods at both the 24 h incubation and 48 h incubation times using either a singlet or duplicate analysis. Table 8 shows cereal data in which the prepared samples were also plated on Peel Plate EBHV, 5 mL volume method. Cereal at a 1:10 dilution preparation is too thick and viscous to test with the Peel Plate method and therefore a 1:50 dilution of cereal is prescribed. This means that 5 mL of the preparation needs to be tested instead of 1 mL to obtain a CFU/0.1 g/plate result.
With the Peel Plate EB method reported in Tables 4-7, five plates were performed and the bacterial colonies on each plate summed for a CFU/0.1 g result. The Peel Plate EBHV plate is designed for 5 mL volume and, therefore, just one plate and sample addition are a preferred option of users. The Peel Plate EBHV method was not significantly different from the reference methods in these cereal evaluations. The recovery of bacteria is improved with the HV method compared to the reference method and the Peel Plate EB 1 mL method. This could reflect the fewer pipetting manipulations and faster time to pipet samples. Table 8 shows the statistical parameters of the EBHV single plate count result at 24 h and the duplicate results at 48 h. There is no significant improvement in the recovery or the confidence limits if a 24 h single plate result is used or the 48 h duplicate result is used; both are statistically the same and equivalent to the reference methods.

Collaborative Validation Study
The inclusivity/exclusivity and matrix studies demonstrated and satisfied validation body requirements that the candidate method accurately enumerated Enterobacteriaceae in select foods and environmental surfaces as claimed by the manufacturer, and that no difference in repeatability was observed between the candidate method and the reference methods. The next requirement of the harmonized MicroVal/AOAC validation is multi-laboratory collaborative study to demonstrate the candidate method can be performed by laboratories routinely doing Enterobacteriaceae analyses and to determine repeatability and reproducibility parameters to be assigned to the method.

Study Design
One matrix, powdered infant formula (milk-based with iron and DHA) containing probiotic (Lactobacillus reuteri), was evaluated in this study. The matrix was obtained from a local retailer and screened for the presence of naturally occurring Enterobacteriaceae by the ISO 21528-1 reference method. No natural contamination was observed so four separate levels of contamination were targeted for the evaluation: uninoculated, 0 CFU/g (mL); low, 10-100 CFU/g (mL); medium, 100-1000 CFU/g (mL); high 1000-10 000 CFU/g (mL). To obtain the required contamination levels, bulk lots of the matrix were artificially contaminated with a lyophilized culture of Cronobacter sakazakii Q Laboratories (QL) isolate 17031.4 (origin-powdered infant formula) at each target contamination level. Two replicate samples from each of the four contamination levels were analyzed by both the candidate and reference methods in a paired study design by each collaborating laboratory.
A detailed collaborative study packet outlining all necessary information related to the study, including media preparation, test portion preparation, and documentation of results, was sent to each collaborating laboratory prior to the initiation of the study. A conference call was conducted prior to the initiation of the study to discuss the collaborative study packet and answer any questions from the participating laboratories.

Preparation of the Inocula and Test Portions
The C. sakazakii isolate used in this evaluation was lyophilized prior to inoculation. The culture was propagated onto SBA from a Q Laboratories frozen stock culture stored at À70 C. To prepare the culture for lyophilization, a single, well-isolated colony from SBA was transferred into BHI broth and incubated at 37 6 2 C for 18-24 h. The culture was diluted in a sterile cryoprotectant, reconstituted in 10% nonfat dry milk, and freeze dried for 48-72 h. A bulk lot of the test matrix inoculated with the culture at a high level was expected to yield all positive results.
An aliquot of the high-level inoculum was further mixed with uninoculated powdered infant formula to produce the low-level inoculum. After inoculation, the matrix was held for a minimum of 2 weeks at ambient temperature (20-25 C). The inoculated test product was packaged into separate 25 g samples in sterile Whirl-Pak V R bags and shipped to the collaborators.

Test Portion Distribution
All samples were labeled with a randomized, blind-coded 3digit number affixed to the sample container. Eleven participants from ten separate locations participated. Test portions were shipped in leak-proof insulated containers via overnight delivery according to the Category B Dangerous Goods shipment regulations set forth by International Air Transport Association. Test portions were shipped at ambient temperatures (20-25 C). Upon receipt, samples were held at ambient temperature until analysis was initiated. In addition to each of the test portions, collaborators also received a test portion for each matrix labeled as 'lactic acid bacteria' (LAB) to determine total background count in the matrix. The LAB samples were prepared from the bulk lot of test matrix, prior to inoculation. Additionally, a temperature probe was included in the shipment. Participants were instructed to submit the data from the temperature probe upon receipt of the shipment.

Test Portion Analysis
Collaborators followed the appropriate preparation and analysis protocol provided to them in the collaborator instructions (Version 3, February 2018). Each collaborator received eight test portions (two high, two medium, two low, and two uninoculated). Sample portions defined by ISO method, 25 g test portion was diluted with 225 mL buffered peptone water (BPW) and homogenized with a paddle blender for 2 m 6 10 s. Ten-fold serial dilutions of each sample were prepared and a 1.0 mL aliquot of each dilution was plated onto a single Peel Plate EB for each dilution. Plates were incubated at 37 6 1 C for 24 h and enumerated. After enumeration, plates were reincubated at 37 6 1 C for an additional 24 h (48 h total). Plates were reenumerated at 48 h. Each spot on the plate represented an EB colony and was enumerated. Plates containing greater than 150 colonies/plate were recorded as too numerous to count. Final CFU/g (mL) results were determined by multiplying the counts by the dilution factor (reciprocal of dilution) for that plate. Each test portion analyzed by the Peel Plate EB method was also analyzed using either the ISO 21528-1 or 21528-2 reference method in a paired study design. The uninoculated and lowlevel test portions were analyzed via the ISO 21528-1 reference method, and the medium-and high-level samples were analyzed via the ISO 21528-2 reference method. For ISO 21528-1, a three-tube MPN was prepared. Positive tubes, those showing turbidity indicating growth, were struck to VRBAG for visual determination of typical colonies (red to purple with or without zones of precipitate). For ISO 21528-2, serial dilutions for each sample were plated in duplicate using VRBAG. Agar plates were incubated for 24 6 2 h at 37 6 1 C. Typical colonies in the countable range (<150 CFU/plate) were enumerated using a standard colony counter. For both ISO 21528 Parts 1 and 2, typical colonies were confirmed positive for Enterobacteriaceae by a spot oxidase test and a glucose agar test.

Statistical Analysis
Each collaborating laboratory recorded the CFU/g (mL) results for the reference methods and the candidate method on the electronic spreadsheet provided. The data sheets were submitted to the study director at the end of the study for analysis. A logarithmic 10 transformation [CFU/g þ 0.1f], where f is the reported CFU/g (mL) corresponding to the smallest reportable result]. A Youden plot was prepared to identify discrepancies between test replicates. Outliers were identified using the Cochran and Grubb's test. The differences of means, including 95% upper and lower confidence limits, were determined for each contamination level (2,13). If the difference of means between the two methods was between À0.5 log 10 CFU/g (mL) and þ0.5 log 10 CFU/g (mL) it was considered that no statistical difference existed between the two methods (3,14). The repeatability (s r ) and reproducibility (s R ) of the methods were also determined (5).

Powdered Infant Formula with Probiotics
(a) Results.-Each collaborating laboratory recorded the CFU/g (mL) results for the reference methods and the candidate method on the electronic spreadsheet provided. The data sheets were submitted to the study director at the end of the study for analysis. The candidate method results at 24 and at 48 h along with the reference method results reported by each laboratory were converted to logarithmic values for statistical analysis and were plotted using a Youden's plot. The log 10 individual laboratory results are presented in Supplemental Appendix Tables 1 and 2. Figures 1 and 2   (mL)] between the 24 and 48 h incubation time points indicating that either time point is acceptable for use. Based on the data presented, the reproducibility values obtained for all contamination levels were generally similar between the candidate and reference methods, indicating that both the between-and within-laboratory variations were consistent between the candidate and reference method. These values indicate that for reproducibility, no meaningful statistical differences [absolute value of <0.50 log 10 CFU/g (mL)] were observed in the data between the candidate and reference methods when test portions were analyzed by different analysts at each laboratory or within each sample set at a given laboratory.

Product Consistency (Lot-to-Lot) and Stability Studies
Peel Plate EB are quality tested after manufacture following the Charm Sciences, Inc. quality control documents which are part of the quality management system have just recently been certified under the ISO 9001 (2015) system. Encompassed in the quality control evaluation are random collection of QC samples throughout the aseptic production, two tests per every 50 manufactured. These are put through a series of evaluations. Sterility checks call for 60 tests per lot, where a lot encompasses a week's production. Tests are rehydrated with 1 mL sterile water and incubated 72 h. There are to be no detected Enterobacteriaceae in any of the tests, and if one or more are detected an additional 200 tests performed with less than 1% containing one Enterobacteriaceae or less.
Detection and recovery evaluation (performance checks) are performed in comparison to the VRBAG reference method using comparing n ¼ 10 test pairs of various Enterobacteriaceae strains. Additionally, naturally EB contaminated chicken samples have been added to verify both exclusion and selection of EB with verification of detected colonies using confirmation methods. Twelve to 25 samples are compared at neat, 10 À1 , and 10 2 dilutions to achieve a countable range 1-150 CFU/plate. Results are compared to reference methods using a statistical population analysis. Peel Plate EB population results should be within 0.2 log mean difference with a population CI greater than P > 0.01. Accelerated stress testing is performed 45 days at 37 C to assure an 18-month refrigerated shelf life and a 1-year shelf life at 0-25 C. Recovery experiments comparing n ¼ 10 test pairs of various Enterobacteriaceae strains are performed to verify no significant difference P > 0.05 from prior production and reference methods. A non-coliform strain, Lactobacillus, is also evaluated to make sure there is no degradation of selection agents in the stressed tests. Real-time storage testing is also performed to verify performance at shelf date.
Additionally, production quality control specifications for the dryness of the plates, <4.5%, are reviewed and additional testing added if manufactured products exceed those specifications.
A summary of these evaluations for several lots of manufactured product are supplied in Table 10. These testing  a Log 10 CFU/g (mL) mean difference between the low and high pairs n ¼ 10 pairs. b 95% Lower confidence limit for difference of means. c 95% Upper confidence limit for difference of means.
parameters are designed to assure the product consistency and stability until 1 year at 0-25 C shelf life.

Robustness Studies
Robustness studies were performed using perturbations of the critical steps of the Peel Plate EB method (13). The steps and perturbations evaluated were pipetting, 1.0 6 0.1 mL; temperature of incubation, low (35) and high (39 C); and time of incubation, low (22) and high (26 h), and 46 and 48 h. The assays were performed in buffer with two ATCC strains, with ten replicate tests under each assay condition. Each perturbation condition was compared to the control condition in a paired t-test analysis.
Results of the robustness analysis are reported in Table 11. Assay temperature showed no significant difference by ttest or paired log-t test confidence levels >0.5. A shorter assay time did not show a significant difference by t-test and there is no significant difference between the shorter (22) and longer (50 h) incubation times. Pipet volume did show a significant difference by t-test as would be expected with a low bias of the 900 compared to the 1100 uL dispense. Despite the measured t-test low bias, using the mean difference and confidence limits >0.5 log as the significance specification, the bias is not considered significant.
The effect of moisture loss from an exposed unsealed test strip and the effect of moisture loss on a test exposed for 15 min in a laminar flow hood were determined. In control experiments with sealed strips, there is less than a 1% loss of weight, while there was a 10-15% weight loss after 15 min open air exposure that would simulate an open environmental air sample taken in a food plant.Moisture loss studies were performed with three lots of EB tests in three sets (n ¼ 10 each) with Butterfield's buffer fortified with either Hafnia alvei ATCC 51815, Pantoea agglomerans ATCC 27155, or Enterobacter aerogenes ATCC 13048. Plates had diluted samples added and were left exposed in a32 C incubator for 15 min to achieve volume loss of 10-15%. These were compared to a control set that were sealed and not exposed to the moisture loss step. Average and standard deviations were calculated and reported in Table 12. Overall, there were not significant differences (62 SD) in the bacterial recovery on the control compared to the air-exposed plates on any of the three lots of tests evaluated.

Conclusions
The precollaborative study demonstrates that the Peel Plate EB method incubated at 37 6 1 C for 24 up to 48 h, without a confirmatory step, selectively detects EB and excludes non-EB. Matrix studies of heat-processed milk and dairy products, infant formula and cereals, environmental surfaces, and chicken rinse samples were not significantly different from the standard method for Enterobacteriaceae, ISO 21528-1:2017 Microbiology of the food chain-Horizontal method for detection and enumeration of Enterobacteriaceae-Part 1: Detection and Part 2: Enumeration. The candidate method was not significantly different from the reference using either a singlet plate result or duplicate plate results at both the 24 and 48 h incubation periods.
International collaborative study by 11 participants from 10 laboratories studying dried infant formula (dairy-based) containing probiotic demonstrated results not significantly different from the reference method with mean differences within À0.2 log 10 CFU/g (mL) of ISO 21528-1 at the low concentration and within 0.2 log 10 CFU/g (mL) of ISO 21528-2 at the medium and high concentrations. Repeatability and reproducibility values at both the 24 and 48 h were comparable to the reference methods.
The study data support that the Peel Plate EB method is equivalent to the ISO 21528 Parts 1 and 2 reference methods within heat-processed and dairy products, infant formula and cereals, stainless surfaces, and chicken carcass rinses studied.

Recommendations
It is recommended that the Peel Plate EB be adopted as Official First Action status for the enumeration of Enterobacteriaceae from pasteurized whole milk, butter, nonfat dry milk, vanilla ice cream, powdered and liquid infant formula (milk-based) containing probiotic, nonprobiotic liquid infant formula (soybased), infant rice cereal (without probiotic), infant cereal with probiotic, chicken carcass rinse with neutralized BPW, chicken carcass rinse with BPW, and stainless-steel surfaces.

Supplemental Information
Supplemental information is available on the J. AOAC Int. website.