Use of PetrifilmTM to Evaluate the Microflora of Frozen Dessert Mixes

Ninety vanilla frozen dessert mix samples were analyzed to determine total microbial populations and coliforms present in samples inoculated with a coliform isolated from raw milk. Standard methods agar (SMA, PCA, Difco) and violet red bile agar (VRBA, Difco) as well as PetrifilmTM (PSM, PVRB) were used for plating of samples. Standard VRBA 1:10 method produced significantly higher counts of colony-forming units (CFU) than PVRB 2:3 and VRBA 2:3 methods. VRBA 2:3 colony-forming unit counts were also significantly higher than those on PVRB 2:3, but both methods showed a moderately strong linear relationship. Repeatabilities of all three coliform plating methods (VRBA 1:10, VRBA 2:3, and PVRB 2:3) were acceptably low. Less than 10% of samples plated on SMA and PSM resulted in total aerobic colony-forming units in the countable range, making evaluation of data difficult and resulting in a lack of strong linear relationship between PSM and SMA. An additional 70 local retail store samples containing naturally occurring coliforms were evaluated using PVRB 2:3 and VRBA 1:10 methodology, confirmed in brilliant green lactose bile broth (BGLB, Difco) and compared to standard VRBA 1:10 previously analyzed. All methods were equivalent for mean log coliforms, i.e., 1.38,1.33, and 1.31 for PVRB 2:3, VRBA 2:3, and VRBA 1:10, respectively. PetrifilmTM methods were comparable to standard methods for enumerating coliforms in frozen dairy products, and would be a valid alternative to standard coliform and total plate count methods. Consumption of ice cream products, especially cones, sundaes, and novelties, has increased significantly in the past few years. Concern for the microbial quality of these products has been renewed due to several recent recalls of both ice cream and ice cream novelties. As production demands increase, the potential for mishandling and recontamination of these products also increases. The presence of bacteria in frozen dairy products is an indication of poor sanitation procedures and improper handling practices. The microbial quality of manufactured milk products is routinely assessed by plate count methodology. The standard 'Published as paper No. 16611 of the contribution series of the Minnesota Agricultural Experiment Station based on research conducted under Project 18-56, supported by Hatch Funds. "-Department of Food Science and Nutrition. 1 Biostatistical Specialist. plate count (SPC) is used to estimate microbial populations in dairy products (7). The coliform test using violet red bile agar (VRBA) is used to evaluate the efficacy of production procedures designed to minimize microbial contamination in processed dairy products, and to detect recontamination of milk, cream, and processed dairy products (6). A dry medium culture plate, PetrifilmTM Aerobic Count plate, designated PSM, has been evaluated as an alternative method to the SPC for fresh ground beef (9), raw milk (4), and seafood (3). Nelson et al. (8) reported an evaluation of the PetrifilmTM Coliform Count plate (PVRB) as a comparable method for the enumeration of coliforms. Both PSM and PVRB methods have been collaboratively evaluated (5) and are AOAC Official Final Action Methods for testing raw and pasteurized milk, and AOAC Official First Action methods for testing all dairy products. This study was undertaken to evaluate the efficacy of PSM and PVRB as acceptable methods for the enumeration of total aerobic colony-forming units and coliforms in frozen dessert mixes. Vanilla frozen dessert mixes of varying fat and total solids content were chosen to represent ice cream, shake and soft-serve formulations. The decision to use a vanilla mix was made in order to eliminate the possibility of microbial contamination due to the addition of nuts, chocolate, and flavorings, and to minimize the number of false positives often encountered in the coliform test when fruit ice creams are the tested products (I). MATERIALS AND METHODS Experimental design A total of 90 vanilla frozen dessert mix samples (30/wk), packed in ice chests, were received from a local manufacturer and stored at 4°C immediately upon arrival. Testing was done within 12 h of receipt. The compositions of the frozen dessert mix samples were: icecream (10-14% fat, 40-41% total solids); shake (3.5% fat, 25-27% total solids); and soft-serve (4% fat, 31.5% total solids). Ten samples of each type were received weekly and tested. Total aerobic plate count Eleven grams of sample was aseptically transferred to 99 ml sterile phosphate buffer resulting in a 1:10 dilution. Sample was JOURNAL OF FOOD PROTECTION, VOL. 52, ALGLSL 1989 D ow naded rom hp://m eridianenpress.com /jfp/article-pdf/52/8607/0362-028x-52_8_549.pdf by gest on 29 Feruary 2020 550 SMITH, ZOTTOLA, FOX, AND CHAUSSE further diluted to 1:100, and a 1-ml portion of each dilution was plated, in duplicate, on SMA [Standard Methods Agar (PCA), Difco] and PSM (Petrifilm' SM) according to instructions provided by the manufacturer. All plates were incubated at 30-32°C for 48 h. Coliform count Preliminary work in our laboratory indicated that the frozen dessert mix samples which were being analyzed did not contain countable coliforms; therefore, each sample was inoculated with a "naturally-occurring coliform" which had been isolated from raw milk. A 50-g portion of frozen dessert mix was aseptically weighed into a 24-oz. Whirl-Pak bag (Nasco, Fort Atkinson, WI53538), 5 ml of a 10 dilution of the inoculum was added, and the inoculated sample was stomached for 30 sec. Inoculum A col iform isolated from raw milk was used as the inoculum for the frozen dessert mix. Raw milk from the dairy herd, St. Paul campus of the University of Minnesota, was collected after the 6:00 a.m. milking and was immediately refrigerated at 4°C. A sample portion was diluted and a 1-ml aliquot of 10', 10, and 10 dilutions was plated on VRBA and PVRB to determine the presence of coliforms in the raw milk sample. A typical gas-producing colony from the PVRB plate and a typical coliform colony from the VRBA plate were transferred to brilliant green bile broth 2% (BGLB, Difco) and incubated for 48 h. Both growth and gas formation were observed and confirmed the presence of a coliform as outlined in SMEDP. A loopful of the positive BGLB tube was streaked on SMA slants, incubated at 30-32°C for 24 h, and refrigerated at 4°C. The inoculum for each testing was therefore taken from a fresh, refrigerated slant, transferred to TSB, and incubated at 30-32"C for 18-24 h. Population was assumed to be ca. 10 CFU/ml. A fresh 10 dilution of the TSB culture was made and held on ice for every 6 samples tested to eliminate loss of viability due to exposure to phosphate dilution buffer over an extended period of time (2). Procedure The following two procedures were used to evaluate the coliform population in the inoculated sample: (1) Standard VRBA 1:10 Procedure. An 11-g portion of the inoculated sample was aseptically added to 99 ml sterile phosphate buffer. The procedure for plating samples of high solids content and viscous products was followed (4). (2) PVRB 2:3 Recommended Procedure. A 2-ml portion of the inoculated sample was aseptically transferred to a sterile 16 X 150 mm capped test tube. One ml of sterile buffer was added, resulting in a 2:3 dilution. A 0.5-ml aliquot of this dilution was plated, in duplicate, on VRBA and PVRB which had been prehydrated as recommended by the manufacturer specifically for frozen dessert mix products. All plates were incubated at 30-32°C for 24 h. After appropriate incubation periods, colony-forming units were counted using a Gallenkamp Colony Counter, and results reported as outlined in SMEDP (7). Representative coliform colonies on VRBA and PVRB were picked from every sixth sample and transferred to BGLB for coliform confirmation. Statistical analysis The statistical analysis included 1) comparing mean coliform counts/ml between PVRB 2:3, VRBA 2:3, and standard VRBA 1:10 methods: 2) comparing mean total counts/ml between PSM and SMA; and 3) determining repeatabilities for all methods. All analyses were carried out in the log|(J scale to more nearly meet the assumption of normally distributed data. RESULTS AND DISCUSSION The mean log difference between SMA (2.004) and PSM (1.969) was 0.034. This is less than the difference of .036 given in SMEDP as acceptable for an alternate method. Only 8 of 90 samples on SMA, and 5 of 90 samples on PSM, resulted in actual numbers of colonies/plate in the countable range of 25-250. This fact made it difficult to evaluate the total aerobic count data and may explain the lack of a strong linear relationship between PSM and SMA methodologies. All three coliform methods were compared statistically. Standard VRBA 1:10 resulted in significantly higher colony counts (log mean = 2.119) than either VRBA 2:3 (log mean = 2.066) or PVRB 2:3 (log mean = 2.002). VRBA 2:3 was significantly higher than PVRB 2:3. All of the mean log differences (-0.065, -0.117, -0.053, P<0.01 in all cases) were greater (in absolute value) than the "gold standard" of 0.036 given in SMEDP. The highest correlation (0.94) was between PVRB 2:3 and VRBA 2:3, which also produced the best slope (0.85) and intercept (0.25) compared to correlations of 0.87 and 0.88, slopes of 0.64 and 0.71, and intercepts of 0.65 and 0.55 for PVRB 2:3 vs. standard VRBA l:10and VRBA 2:3 vs. standard VRBA 1:10, respectively (Fig. 1-3). Repeatability variances (between-duplicates variances) were 0.011, 0.009, and 0.013 for VRBA 2:3, PVRB 2:3 and standard VRBA 1:10, respectively. These variances were not statistically different. In this study, the standard VRBA 1:10 method produced significantly higher counts than the PVRB 2:3 and VRBA 2:3 methodologies. The VRBA 2:3 resulted in significantly higher colony counts than PVRB 2:3, although these two methods had a moderately strong linear relationship. Repeatabilities of all three coliform methods were low. Due to the