J. Dairy Sci. 89:2459-2464
© American Dairy Science Association, 2006.
An Investigation of the Use of the MicroFoss as an Indicator of the Shelf Life of Pasteurized Fluid Milk1
C. H. White2,
J. Wilson and
M. W. Schilling
Department of Food Science, Nutrition, and Health Promotion, Box 9805, Mississippi State University, Mississippi State 39762
2 Corresponding author: chwhite{at}ra.msstate.edu
 |
ABSTRACT
|
|---|
The MicroFoss method was evaluated for its effectiveness as an indicator of fluid milk shelf life. Half-gallon, 2% fat fluid milk samples (n = 90) were obtained from a milk processing plant on 3 occasions postpasteurization and evaluated for shelf life. Sensory evaluation was performed by 3 judges experienced in the use of the American Dairy Science Association scorecard for milk. A score of 5 or less was considered to represent the end of the shelf life of the product. MicroFoss coupled with preliminary incubation (PI) was utilized to estimate the total viable (TVC) and gram-negative counts (GN) in the milk. The MicroFoss functions by using a pH indicator or CO2 production to detect changes in light reflection to estimate bacterial populations. Simple and multiple linear regression analyses were utilized to determine the relationship between MicroFoss (PI-GN and PI-TVC detection times) and product shelf life. It was concluded that using both PI-GN and PI-TVC in a combined algorithm is the optimal way of using MicroFoss as a shelf-life indicator. When PI-TVC was selected in the algorithm, a correlation coefficient of 0.89 existed between PI-TVC and shelf life; PI-GN was used in the algorithm in the place of PI-TVC when its detection time was within 6 h of the detection time of PI-TVC vials. The PI-GN detection times correlated well (r = 0.80) with shelf life, but more importantly, all but one PI-GN sample (n = 50) selected in the algorithm had a shelf life of less than 10 d. This indicates that the PI-GN measurement can be utilized along with PI-TVC detection time to indicate potential shelf-life problems.
Key Words: MicroFoss • shelf-life
|
INTRODUCTION
|
|---|
With longer distribution systems and the continued threat of temperature abuse, it is imperative that the potential shelf life of fluid milk be known before the product is shipped. Although microbiological estimates have been used as a standard for shelf life, the most accurate and meaningful way to determine actual product shelf life of milk is by sensory evaluation. This actual shelf life can then be correlated with any test designed to indicate the shelf life of milk or milk products.
Some of the methods designed to estimate shelf life include bioluminescence (Murphy et al., 1998), impedance microbiology (Bishop et al., 1984), limulus amoebocyte lysate (Byrne and Bishop, 1990), direct reflectance colorimetry, Virginia Tech procedure (Byrne et al., 1989; White, 1993), and the Moseley Keeping-Quality test (Elliker et al., 1964; Richardson, 1985). Because psychrotrophic bacteria cause the majority of milk spoilage that results in an end of shelf life (Law, 1979), it is critical to be able to estimate psychrotrophic numbers in fluid milk. The standard method for estimating psychrotrophic bacteria requires incubation at 7°C for 10 d (Bishop, 1989; Wehr and Frank, 2004). This amount of time is too long to be of any immediate assistance to the dairy processor. Hence, most shelf-life tests have a preliminary incubation (PI) step as an integral part of the process to allow time for any bacteria that may be present to grow to countable numbers (Byrne et al., 1989). White (1991) suggests that there are 4 key points that need to be considered in the selection of a shelf-life test: 1) know the actual, potential shelf life of the product as measured at 7°C; 2) select the shelf life-indicating test or tests that best fit the quality assurance program of a specific dairy processor; 3) routinely run the test (s) and develop a history categorizing the results; and 4) ensure top-management commitment to define a course of action in the event of product failure as projected by the test(s).
MicroFoss (Foss North America, Inc., Eden Prairie, MN) is a computerized instrument designed to rapidly detect microbial contamination in dairy samples. It measures microbial growth by monitoring changes in pH or other biochemical reactions that result in a color change in the broth as the microorganisms grow and metabolize. Samples are inoculated in ready-for-use vials that contain specific broth. Color changes in the agar at the bottom of the vial mirror the color change in the broth. Light from emitting diodes passes through the agar and the color change is read by a photo diode on the other side of the vial as microbial growth occurs. A measurement is taken every 6 min. As soon as a color change is detected, the time of detection is recorded. Detection times (DT) are inversely related to the number of organisms in the sample.
The objective of this study was to determine the ability of the MicroFoss to indicate the actual product shelf life of fluid milk samples produced at a commercial dairy processing plant by correlating actual shelf life to DT of the instrument.
|
MATERIALS AND METHODS
|
|---|
Sample Collection
The milk utilized in this research was obtained from a commercial fluid milk plant. The milk was a pasteurized, homogenized, reduced-fat (2% fat) product packaged in half-gallon plastic jugs. The milk was pasteurized at 76°C for 28 s, and was homogenized at 126 K/cm2 (1,800 psi), in a single-stage homogenizer. To minimize variation among samples, each jug was obtained from the same filler head of the filling machine. The samples were immediately placed in a 4.4°C cooler before being picked up by the researchers. For transport to the laboratory (100 km), the jugs were packed in ice. Upon arrival, the jugs were placed in a 7°C ± 1°C low temperature incubator for the duration of the study; the temperature of the cooler was verified daily.
MicroFoss Testing
To assess the ability of the MicroFoss system to function as a shelf-life indicator of processed milk, different levels of bacteria in the milk were needed. To conduct the study under "real-life" conditions, samples were not artificially loaded with prescribed numbers of spoilage bacteria or bacterial blends. The samples were tested at different intervals to permit natural spoilage. A new set of 8 jugs was tested on each of d 0, 3, 6, and 10. The jugs remained unopened until the day of testing on the MicroFoss system. The MicroFoss testing involved each sample being placed into 2 different vials. One vial was used for the total viable count (TVC), and contained a nutrient-based medium with dextrose as the carbon source. Acidification of the medium due to glucose use changes the pH, and bromocresol purple was used as the pH indicator. This vial was used to assess the total bacterial population.
The other vial was selective for gram-negative bacteria (GN) and used a vial within a vial. The inner vial contained a nutrient growth medium and inhibitors of gram-positive bacteria; the inner vial was inserted into the outer vial, which contained a dye indicator at the bottom window. Production of CO2 changed the pH dye indicator.
A portion of the sample was preincubated at 21°C for 18 h, placed in the 2 different vials (designated PI-TVC and PI-GN), and loaded into the MicroFoss system. After analyzing the results of the first replication, it was discovered that the preincubated samples showed a higher correlation to actual shelf life than the samples that were not preincubated. From that point, the TVC and the GN tests were discontinued.
Determination of Shelf Life
After sampling, each jug was replaced in the 7°C cooler and held for 7 d for a Moseley Keeping-Quality test. Sensory evaluation was performed by 3 expert judges experienced in use of the ADSA scorecard for milk. Sensory testing of milk in the jugs that had previously been opened began on d 6 after bottling. The d-10 jugs remained unopened until the BioSys tests were performed; then sensory testing was initiated. The milk was scored by 2 different methods. One method was based on the ADSA scorecard in which a flavor descriptor was given as well as a score. Scores were based on a 10-point scale, with 10 referring to a perfect sample of milk with no off-flavors. A drop in the numerical score correlated to a drop in the flavor quality of the sample. A score of 5 or less was considered to indicate the end of the shelf life of the product.
The second method of sensory evaluation was a simple accept/reject test. An acceptable sample was scored as 1 and an unacceptable sample was scored as 2. When the sample score became 2, the product was considered to be at the end of its shelf life. To calculate shelf life, the day that a jug was opened and sampled for the BioSys test was considered d 0. The shelf life was the number of days that the sample scored 1 in the accept/ reject test. The relationship between shelf life and MicroFoss was identical for each method. The benefit of using the ADSA method was to obtain the flavor descriptor at the end of shelf life.
Moseley Keeping-Quality Test
The Moseley Keeping-Quality test was performed for each sample because it has been a standard indicator of keeping quality for the dairy industry for more than 40 yr (Elliker et al., 1964). The samples were stored at 7°C for the duration of the study. On d 0, 3, 6, and 10, a small portion from the previously unopened sample jug was poured off into a sterile container. The container was replaced in the 7°C incubator for 7 d. On d 7, the sample was plated onto standard methods agar (a general growth medium to estimate the total aerobic viable bacteria capable of growing at 32°C), and incubated for 48 h at 32°C.
Statistical Analysis
Data analysis was performed on the 3 replications of milk samples using the statistical tools in the XLStat7.5 package (Microsoft Excel, Microsoft, Redmond, WA). Correlation coefficients (r) were determined between PI-TVC and actual shelf life, PI-GN and actual shelf life, and the Moseley Keeping-Quality Test and actual shelf life utilizing simple linear regression. Simple linear regression was also utilized to determine prediction equations for shelf life based on PI-TVC, PI-GN, and Moseley Keeping Quality. Multiple linear regression was used to determine if the combination of PI-TVC and PI-GN did a better job of explaining shelf life than did PI-TVC or PI-GN alone. Detection times for PI-GN within 6 h of detection times for PI-TVC were separated from the original data set to determine the contribution of gram-negative, spoilage bacteria to the end of shelf life.
|
RESULTS AND DISCUSSION
|
|---|
Upon completion of the first replicate of tests, the correlation coefficient was determined between TVC detection time (DT) and shelf life and GN DT and shelf life. The correlation coefficient between TVC DT and actual shelf life was 0.32, and neither the TVC DT nor the GN DT showed a strong relationship to the shelf life of the milk. The data for GN DT revealed that, even though there was a good correlation (r = 0.80) between GN DT and shelf life, these data were misleading because almost all samples with a shelf life between 1 and 8 d had similar GN values and almost all samples with a shelf life greater than 8 d had similar GN values (Figure 1
). In the first replication, PI-TVC showed the highest correlation (r = 0.83) to actual shelf life and PI-GN samples continued to be evaluated because they provide information on psychrotrophic, gram-negative bacteria that cause milk spoilage (Law, 1979). In consequent replications, only preincubated samples (PI-TVC and PI-GN) were evaluated. The flavor defects for the milk at the end of its shelf life were described as "fruity", "bitter", and "unclean", which are typically caused by gram-negative bacteria. The flavor criticisms of "malty" and "high acid" were also present during the course of this study. The fact that these flavors can be caused by gram-positive bacteria could have accounted for the variability of the GN medium in predicting shelf life. Gram-negative bacteria cause off-flavors in milk that can drastically reduce shelf life. Gram-positive bacteria can also cause off-flavors that result in shortened shelf life and would be present in the PI-TVC vials if major psychrotrophic gram-positive growth had occurred (Fromm and Boor, 2004). Gram-positive organisms are often the cause of spoilage of milk that has not been contaminated postpasteurization. Typically, gram-negative psychrotrophic bacteria are most likely to be the source of "dirty" processing equipment such as filling equipment, which is reported to be the main contamination source (Gruetzmacher and Bradley, 1999).
View larger version (8K):
[in this window]
[in a new window]
|
Figure 1. Scattergram showing the relationship between gram-negative (GN) count detection time and actual shelf life for the first replication of data (n = 32) utilizing the MicroFoss Technology.
|
|
Detection time of PI-TVC was a good indicator of shelf life (r = 0.86; Figure 2) for all milk samples. Detection time of PI-GN was added to the regression model but did not provide any additional information (shelf life = –0.338 + 0.627 (PI-TVC) + 0.011 (PI-GN), r = 0.87) pertaining to shelf life. The correlation between shelf life and PI-GN was 0.32, but PI-GN samples with DT within 6 h of PI-TVC DT have been reported to be indicative of samples that include high initial counts of psychrotrophic, gram-negative spoilage microorganisms (S. Beck, Centrus Int., Ann Arbor, MI; and R. Firstenberg-Eden, Microsys, Inc., Ann Arbor, MI; personal communication). Therefore, the data set (n = 160) was separated into 2 subsets of data to evaluate the usability of both PI-TVC and PI-GN in an algorithm in which PI-TVC was utilized when the PI-TVC and PI-GN DT were more than 6 h apart and the PI-GN was utilized individually or with the PI-TVC when DT were within 6 h. The PI-TVC assay yielded a somewhat better correlation coefficient (r = 0.89, n = 110, Figure 3) when separated from the PI-GN samples than when all data points were used in the analysis (PI-TVC, n = 160, r = 0.86). The MicroFoss technology and algorithm provide a 2-pronged approach to indicate both severe and mild shelf-life quality problems. The technology is applicable to dairy processing plants because the PI-GN vial can be utilized to indicate serious shelf life problems due to postpasteurization contamination and the PI-TVC vial can be used to indicate less severe shelf-life concerns due to both gram-positive and gram-negative bacteria.
View larger version (22K):
[in this window]
[in a new window]
|
Figure 2. Relationship between detection time for MicroFoss total viable counts from preliminary incubated samples (DT PI-TVC) and actual shelf life of fluid milk samples (2% fat, n = 160, r = 0.86): Shelf life = –0.19 + 0.64 x DT PI-TVC.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Figure 3. The relationship between detection time for MicroFoss total viable counts from preliminary incubated samples (DT PI-TVC) and actual shelf life of fluid milk samples (2% fat, n = 110, r = 0.89) when preliminary incubation, gram-negative samples with detection times within 6 h of PI-TVC are separated into a new data set: Shelf life = –0.117 + 0.653 x DT PI TVC.
|
|
The PI-GN data were analyzed separately from the PI-TVC data for the 50 points where PI-GN DT were within 6 h of the PI-TVC detection times. Eleven data points were removed because they represented DT between 0 and 5 h. These samples contained very high levels of gram-negative organisms and therefore, did not follow a linear trend with shelf life. However, these samples were still indicative of samples with shelf-life problems. For these data points, PI-GN and PI-TVC were combined in a multiple linear regression model. Detection times for PI-TVC did not add significant information (P > 0.05) to the model for those 39 points for which PI-GN DT were within 6 h of PI-TVC data if PI-GN (P < 0.05) were already in the model (r = 0.81 for the multiple linear regression model, Table 1; r = 0.80 for the simple linear regression model, Figure 4). There were only 39 data points in this set, but PI-GN worked better for this region based on the multiple and simple linear regression models (Table 1 and Figure 4). More data would be helpful in explaining the relationship between PI-GN and actual shelf life, but a large data set would be difficult to obtain in an experimental study because these data were dependent on fluid milk plants having serious shelf life problems that occur but could be difficult to predict.
View this table:
[in this window]
[in a new window]
|
Table 1. Multiple linear regression model1 for 39 points in which preliminary incubation, gram-negative (PI-GN) detection time is within 6 h of preliminary incubation, total viable count (PI-TVC) detection times
|
|
View larger version (19K):
[in this window]
[in a new window]
|
Figure 4. The relationship between detection time for MicroFoss gram-negative counts from preliminary incubated samples and actual shelf life of fluid milk samples (2% fat, n = 39, r = 0.80) when preliminary incubation, total viable count (PI-TVC) samples with detection times within 6 h of preliminary incubation, gram-negative (PI-GN) are separated into a new data set (PI-GN within 6 h of PI-TVC): Shelf life = –4.75 + 0.733 x GN.
|
|
The algorithm wherein PI-GN was used when PI-TVC and PI-GN detection time was within 6 h of each other should be used because PI-GN detection times of less than 5 h had a shelf life lower than 5 d, and 49 out of 50 samples with a PI-GN detection time within 6 h of the PI-TVC sample had a shelf life less than 10 d. This revealed that PI-GN samples indicated shelf life problems due to gram-negative psychrotrophic bacteria when used in conjunction with PI-TVC DT. The second part of the algorithm allowed for use of the PI-TVC samples as the shelf life indicator when PI-TVC and PI-GN samples had detection times that were greater than 6 h apart because they indicated potential shelf-life problems due to both gram-negative and gram-positive microorganisms. Therefore, the following recommendations could be made pertaining to using the MicroFoss: 1) If the difference in DT for PI-TVC and PI-GN was greater than 6 h, only PI-TVC DT should be used to indicate shelf life; 2) if the difference in DT for PI-TVC and PI-GN was less than 6 h, only PI-GN should be used to indicate shelf life; 3) all PI-GN data with DT less than 5 h should be removed from the data because they do not linearly relate to shelf life, but do indicate a shelf life problem. The MicroFoss instrument is able to determine which vial to use based on the algorithm and indicates potential shelf-life problems due to detection times. Once the MicroFoss is calibrated properly, the methodology behind its use is simplistic.
Concerning the Moseley test, the correlation coefficient (r = –0.69) showed a weak to moderate relationship with shelf life (Figure 5). The MicroFoss was superior to the Moseley test in indicating shelf life. The PI-TVC and PI-GN DT correlation to product shelf life of 0.89 and 0.80 was much higher than the correlation between the Moseley Quality Keeping test and product shelf life (r = –0.69). The PI-TVC data also had moderate correlation to the Moseley test (r = –0.67, Figure 6). This was expected because both tests are indicators of product shelf life due to detection of microbial spoilage. The PI-GN samples that were used in the algorithm did not correlate well with the Moseley test (r = –0.35) because almost all PI-GN samples with DT within 6 h of corresponding PI-TVC samples had similar log bacteria counts that were very large (>107 cfu/mL). Correlation coefficients between PI-TVC and PI-GN DT and product shelf life were similar to values for other rapid methods including tests that were researched in other experimental studies with high initial costs such as impedance microbiology (r = 0.87, 0.91), but perform much better than other less expensive rapid methods that have been researched, such as plating procedures with preliminary incubation (r = –0.73, –0.78), respectively (White, 1993).
View larger version (23K):
[in this window]
[in a new window]
|
Figure 5. The relationship between the Moseley Keeping-Quality test [log (cfu)] and actual shelf life of fluid milk samples (2% fat, n = 160, r = –0.69): Shelf life = 13.97 – 1.31 x log (cfu).
|
|
View larger version (23K):
[in this window]
[in a new window]
|
Figure 6. The relationship between the Moseley Keeping-Quality test [log (cfu)] and preliminary incubation, total viable count (PI-TVC) for fluid milk samples (2% fat, n = 160, r = –0.67): detection time PI-TVC = 19.69 – 1.704 x log (cfu).
|
|
Samples preincubated (21°C, 18 h) and tested on the MicroFoss proved to be valid indicators of the shelf life of pasteurized milk. Preincubating the milk enhanced the usability of the TVC media because it allowed the low numbers of bacteria normally present in freshly pasteurized milk to increase sufficiently to be detected in a timely manner by the MicroFoss. The separation of the data into 2 sets in which PI-TVC and PI-GN were utilized under different circumstances demonstrates that both total organisms and gram-negative organisms gave a better indication of shelf life than the PI-TVC value alone. The predictive value of the Moseley test was lower than that of the MicroFoss. In summary, the MicroFoss was able to give a good indication of the shelf life of milk within 38 h compared with the Moseley test that required at least 8 to 9 d to indicate shelf life.
|
ACKNOWLEDGEMENTS
|
|---|
Sincere gratitude is expressed to Ruth Eden for the idea of this shelf-life project.
|
FOOTNOTES
|
|---|
1 Approved for publication as Journal Article No. J-10761 of the Mississippi Agricultural and Forestry Experiment Station, Mississippi State University, CRIS No. 371070. 
Received for publication July 12, 2005.
Accepted for publication January 31, 2006.
|
REFERENCES
|
|---|
Bishop, J. R. 1989. A simple shelf-life estimation method as an integral part of a total dairy quality assurance program. Dairy. Food Environ. Sanit. 9:698–701.
Bishop, J. R., C. H. White, and R. Firstenberg-Eden. 1984. A rapid impedimetric method for determining the potential shelf-life of pasteurized whole milk. J. Food Prot. 47:471–475.
Byrne, R. D., and J. R. Bishop. 1990. The limulus amoebocyte lysate assay and the direct epifluorescent filter technique as estimators of potential shelf-life of pasteurized fluid milk. J. Food Prot. 53:151–153.
Byrne, R. D., J. R. Bishop, and M. L. McGilliard. 1989. Selective preliminary incubation for Gram-negative psychrotrophic bacteria in milk. J. Food Prot. 52:396–398.
Elliker, P. R., E. I. Sing, L. J. Christensen, and W. E. Sandine. 1964. Psychrophilic bacteria and keeping quality of pasteurized dairy products. J. Milk Food Technol. 27:69–75.
Fromm, H. I., and K. J. Boor. 2004. Characterization of pasteurized fluid-milk shelf-life attributes. J. Food Sci. 69:M207–M214.
Gruetzmacher, T. J., and R. L. Bradley. 1999. Identification and control of processing variables that affect the quality and safety of fluid milk. J. Food Prot. 62:625–631.[Medline]
Law, B. A. 1979. Enzymes of psychrotrophic bacteria and their effects on milk and milk products. J. Dairy Res. 46:573–578.
Murphy, S. C., S. M. Koslowski, D. K. Bandler, and K. J. Boor. 1998. Evaluation of adenosine triphosphate-bioluminescence hygiene monitoring for trouble-shooting fluid milk shelf-life problems. J. Dairy Sci. 81:817–820.[Abstract]
Richardson, G. H. Ed. 1985. Standard Methods for the Examination of Dairy Products. 15th edition. Am. Public Health Assoc., Washington, DC.
Wehr, H. M., and J. F. Frank. 2004. Standard Methods for the Examination of Dairy Products. Am. Public Health Assoc. 17th ed. Washington, DC.
White, C. H. 1991. The art and science of predicting shelf-life. Dairy Field April:24–25.
White, C. H. 1993. Rapid methods for estimation and prediction of shelf-life of milk and dairy products. J. Dairy Sci. 76:3126–3132.[Free Full Text]
TOP
ABSTRACT
INTRODUCTION
