APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUlY 1977, p. 14-17 Copyright © 1977 American Society for Microbiology
Vol. 34, No. 1 Printed in U.S.A.
Rapid Detection of Microbial Contamination in Frozen Vegetables by Automated Impedance Measurements by D. HARDY, S. J. KRAEGER, S. W. DUFOUR, AND P. CADY* Bactomatic, Inc., Palo Alto, California 94303
Received for publication 22 March 1977
Automated impedance measurements can be used to rapidly assess whether a sample of frozen vegetables contains greater or less than 10 organisms per g. Microorganisms growing in pureed food samples cause a change in the impedance of the medium when the organisms reach a threshold concentration of between 10" and 107 organisms per ml. Estimates of the concentration of microorganisms initially present in the food sample can be made by recording the time required for the organisms in the sample to replicate to threshold levels. In this study, the detection times for 357 samples of frozen vegetables were compared with standard plate counts for each sample. The agreement between the two methods in distinguishing samples containing more than 105 organisms per g was 92.6% for 257 assorted frozen vegetables and somewhat higher (93 to 96%) when separate cutoff times were used for each type of vegetable. The time required for analysis was about 5 h, compared to the 48 to 72 h required for standard plate counts.
One of the most widely used methods to assess the microbiological quality of foods is the standard plate count. This procedure, although inexpensive, accurate, and simple to perform, has some shortcomings. It is laborious to perform in high volume; often it is difficult to
old concentration, generally between 106 and 10' organisms per ml. Impedance changes can result from cultures initially of lower concentrations if the organisms are allowed to replicate to threshold values. The time required to reach threshold is the detection time and is a function of the initial concentration, the medium, and the generation time of the organisms in the culture. By measuring the detection time, an estimate of the magnitude of the initial concentration of microorganisms can be made. This paper describes a rapid method for the detection of frozen vegetable samples containing greater than 103 organisms per g by measuring impedance detection times. The method shows promise of being applicable to other types of food as well.
distinguish vegetable particles from colonies; and it is very slow. Frequently, the time required for the organisms to grow sufficiently to form visible colonies is 48 to 72 h. Rapid methods may enable food processors to more readily reject unacceptably contaminated raw foodstuffs prior to processing, to monitor the processing lines for early evidence of microbial problems, and to improve the quality of the final product while reducing costly inventories as well. Several approaches have been made to shorten length of analysis. These methods include a membrane filter technique (4), radio-
MATERIALS AND METHODS Impedance measurements. A BACTOMETER 32 metric measurements (2, 3), and microcalori- Microbial Monitoring System (Bactomatic, Inc.) was metry (2). Recently, changes in electrical used for all impedance monitoring and has been impedance taking place in the medium of grow- described elsewhere (W. K. Hadley, and G. Senyk, ing cultures have been used to detect the pres- Microbiology -1975, p. 12-21, American Society for ence of bacteria (1). These changes can be con- Microbiology, Washington, D. C., 1975). Printed cirtinuously monitored by passing a small electric cuit board modules (impedance measurement samcurrent through the culture medium and meas- ple chambers) were used for the initial studies (Haduring the impedance as the organisms grow ley and Senyk, Microbiology - 1975, p. 12-21). Disposable modules, described previously (1), were used and metabolize. These measurements do not in the study on reproducibility. appear to interfere with the normal growth of Media. Brain heart infusion broth (BBL), 1%, the microorganisms. supplemented with 0.5% glucose (BHIG) or TryptiImpedance changes are noted when the con- case soy broth (BBL) supplemented with 0.1% yeast centration of microorganisms exceeds a thresh- extract (TSBY) (BBL) was used as the growth me14
RAPID DETECTION OF FROZEN FOOD MICROBES
VOL. 34, 1977
dium for impedance measurements. BHIG or 0.1% peptone water was used as a diluent for the frozen vegetables. Dextrose tryptone agar (BBL) was used for control pour plates and was cooled to 48°C before use. Plates were incubated at 32°C. Method. All frozen vegetables were purchased locally and prepared according to one of two methods. For method I, 50 g of frozen vegetables was added to 50 ml of BHIG and blended for 2 min. A 1ml amount of the puree was then transferred directly to the sample chamber of a printed circuit board module. The reference chamber was filled with sterile BHIG. The puree was diluted with sterile water for plate counting. Pour plates of dextrose tryptone agar were incubated for 72 h. For method II, 50 g of frozen vegetables was added to 50 ml of peptone water and blended for 2 min. A 1-ml amount of the puree was then added to the sample chamber of a disposable module, which was previously filled with 0.5 ml of TSBY. The reference chamber was filled with 1 ml of TSBY. The puree was diluted in sterile saline for plate counting. Pour plates of dextrose tryptone agar were incubated for 48 h. In both methods the modules were incubated at 35°C. Detection time was the time required to produce an accelerating impedance change of 0.8%. This amount of impedance change was selected because it represented a full strip chart channel width of change.
RESULTS Typical impedance changes caused by bacteria in pureed green beans are shown in Fig. 1. In this example there is an initial decrease in impedance which diminishes with time, yielding a roughly constant base line by 1.5 h. This initial decrease is often due to the temperature of the sample coming up to incubator temperature. Starting at around 3.2 h, there is an accelerating decrease in impedance, which reaches the criterion for detection (0.8% impedance change) at 3.7 h. This sample of green beans contained 100,000 organisms per g. The detection times for 257 frozen vegetable samples prepared according to method I, using
15
)0 Z
0
10_-
0
5
10
DETECTION TIME (Mourns)
FIG. 2. BACTOMETER Microbial Monitoring System detection times and standard plate count results for 257 frozen vegetable samples. The horizontal line at 100,000 organisms per g and the vertical line at 4.9 h mark the boundaries between acceptable and excessive contamination for the standard plate count and impedance methods, respectively. The shaded area represents the error inherent in the standard plate count: one standard deviation on each side of the 100,000-organisms per g level.
printed circuit board modules, were plotted against the logarithm of the initial concentration as determined by pour plates (see Fig. 2). The roughly linear relationship between the detection times and the logarithm of the concentration indicates that these samples grew with similar lag phases and generation times from sample to sample. The horizontal line at 100,000 organisms per g indicates the boundary between acceptable and unacceptable levels of microbial content. The vertical line at 4.9 h indicates the detection time, or cutoff time, that best divides all the samples into acceptable and unacceptable categories. The samples in the upper left quadrant (formed by the vertical and horizontal lines) are those that are unacceptable by both impedance and plate count. The samples in the lower right quadrant are those that are acceptable by both methods. Using 4.9 h as the cutoff time, 238 of 257 samples, or 92.6%, were correctly classified by the impedance method. The samples in the upper right quadrant are false negatives; that is, they are unacceptably contaminated by count but acceptable by impedance. Thus, plate FIG. 1. Typical impedance changes observed with 10 of 257 samples (3.9%) are false negatives. frozen vegetable samples. This curve represents the The samples in the lower left quadrant are false impedance changes observed with a sample of green beans with an initial concentration of 100,000 orga- positives; that is, they are acceptable by plate nisms per g. Detection of accelerating impedance count but unacceptable by impedance. Thus, 9 of 257 samples (3.5%) are false positives. The changes occurred at 3.7 h. lU
ul
I
0
5
10
16
APPL. ENVIRON. MICROBIOL.
HARDY ET AL.
percentage of positive samples missed by impedance is 10 samples of 123 positive samples, or 8.1%. Note the shaded region in Fig. 2. This is an approximation of the deviation to be expected in standard plate counts for frozen vegetables. Comparing 100 samples done in duplicate, it was shown that the mean standard deviation was 0.3 log. Some samples deviated by as much as 1.3 log. The shaded area represents all points within one standard deviation from the 100,000organisms per g level. Table 1 shows the results of analyzing the data separately for each type of vegetable. The TABLE 1. Cutoff times and percent agreement for various types of frozen vegetables Frozen vegetables
No. of samples
Cutoff time (h)
Percent agreement
Green beans Corn Peas Mixed Peas & carrots Others
48 48 47 41 29 44
4.9 4.7 4.5 4.7 5.3 5.3
95.8 93.7 95.7 95.1 93.1 88.6
257
4.9
92.6
All
cutoff time that gives the best percent agreement was determined for each class of vegetable. The "other" category includes samples of 10 kinds of frozen vegetables: carrots, cauliflower, broccoli, succotash, zucchini, okra, black-eyed peas, baby lima beans, and butter beans. The optimum cutoff times ranged from 4.5 h for peas to 5.3 h for peas and carrots, and the percent agreement ranged from 88.6% for the "other" category to 95.8% for green beans. To determine the reproducibility of the method, 100 samples of green beans were prepared according to method II and run in disposable modules. The same 100 samples were then again run in disposable modules according to method II and compared for data consistency and reproducibility. Table 2 shows the results of this comparison. The two runs are very similar, with the major differences being two additional false positives in the second run, which produces a corresponding 2% decrease in the overall agreement.
DISCUSSION The data presented here indicate that impedance measurements provide an alternative to the standard plate count for testing microbial
TABLE 2. Analysis of 100 samples of frozen green beans run twice for impedimetric detection of samples containing greater than 105 microorganisms per g Determination
Classification Cutoff time for 105 bacteria per g (h) Percent agreement Samples falsely classified as being under 105 (out of 100 samples) Samples falsely classified as being >105 (out of 100 samples) Samples falsely classified as being 105) Samples falsely classified as being >105 (out of all samples with counts
Vol. 34, No. 1 Printed in U.S.A.
Rapid Detection of Microbial Contamination in Frozen Vegetables by Automated Impedance Measurements by D. HARDY, S. J. KRAEGER, S. W. DUFOUR, AND P. CADY* Bactomatic, Inc., Palo Alto, California 94303
Received for publication 22 March 1977
Automated impedance measurements can be used to rapidly assess whether a sample of frozen vegetables contains greater or less than 10 organisms per g. Microorganisms growing in pureed food samples cause a change in the impedance of the medium when the organisms reach a threshold concentration of between 10" and 107 organisms per ml. Estimates of the concentration of microorganisms initially present in the food sample can be made by recording the time required for the organisms in the sample to replicate to threshold levels. In this study, the detection times for 357 samples of frozen vegetables were compared with standard plate counts for each sample. The agreement between the two methods in distinguishing samples containing more than 105 organisms per g was 92.6% for 257 assorted frozen vegetables and somewhat higher (93 to 96%) when separate cutoff times were used for each type of vegetable. The time required for analysis was about 5 h, compared to the 48 to 72 h required for standard plate counts.
One of the most widely used methods to assess the microbiological quality of foods is the standard plate count. This procedure, although inexpensive, accurate, and simple to perform, has some shortcomings. It is laborious to perform in high volume; often it is difficult to
old concentration, generally between 106 and 10' organisms per ml. Impedance changes can result from cultures initially of lower concentrations if the organisms are allowed to replicate to threshold values. The time required to reach threshold is the detection time and is a function of the initial concentration, the medium, and the generation time of the organisms in the culture. By measuring the detection time, an estimate of the magnitude of the initial concentration of microorganisms can be made. This paper describes a rapid method for the detection of frozen vegetable samples containing greater than 103 organisms per g by measuring impedance detection times. The method shows promise of being applicable to other types of food as well.
distinguish vegetable particles from colonies; and it is very slow. Frequently, the time required for the organisms to grow sufficiently to form visible colonies is 48 to 72 h. Rapid methods may enable food processors to more readily reject unacceptably contaminated raw foodstuffs prior to processing, to monitor the processing lines for early evidence of microbial problems, and to improve the quality of the final product while reducing costly inventories as well. Several approaches have been made to shorten length of analysis. These methods include a membrane filter technique (4), radio-
MATERIALS AND METHODS Impedance measurements. A BACTOMETER 32 metric measurements (2, 3), and microcalori- Microbial Monitoring System (Bactomatic, Inc.) was metry (2). Recently, changes in electrical used for all impedance monitoring and has been impedance taking place in the medium of grow- described elsewhere (W. K. Hadley, and G. Senyk, ing cultures have been used to detect the pres- Microbiology -1975, p. 12-21, American Society for ence of bacteria (1). These changes can be con- Microbiology, Washington, D. C., 1975). Printed cirtinuously monitored by passing a small electric cuit board modules (impedance measurement samcurrent through the culture medium and meas- ple chambers) were used for the initial studies (Haduring the impedance as the organisms grow ley and Senyk, Microbiology - 1975, p. 12-21). Disposable modules, described previously (1), were used and metabolize. These measurements do not in the study on reproducibility. appear to interfere with the normal growth of Media. Brain heart infusion broth (BBL), 1%, the microorganisms. supplemented with 0.5% glucose (BHIG) or TryptiImpedance changes are noted when the con- case soy broth (BBL) supplemented with 0.1% yeast centration of microorganisms exceeds a thresh- extract (TSBY) (BBL) was used as the growth me14
RAPID DETECTION OF FROZEN FOOD MICROBES
VOL. 34, 1977
dium for impedance measurements. BHIG or 0.1% peptone water was used as a diluent for the frozen vegetables. Dextrose tryptone agar (BBL) was used for control pour plates and was cooled to 48°C before use. Plates were incubated at 32°C. Method. All frozen vegetables were purchased locally and prepared according to one of two methods. For method I, 50 g of frozen vegetables was added to 50 ml of BHIG and blended for 2 min. A 1ml amount of the puree was then transferred directly to the sample chamber of a printed circuit board module. The reference chamber was filled with sterile BHIG. The puree was diluted with sterile water for plate counting. Pour plates of dextrose tryptone agar were incubated for 72 h. For method II, 50 g of frozen vegetables was added to 50 ml of peptone water and blended for 2 min. A 1-ml amount of the puree was then added to the sample chamber of a disposable module, which was previously filled with 0.5 ml of TSBY. The reference chamber was filled with 1 ml of TSBY. The puree was diluted in sterile saline for plate counting. Pour plates of dextrose tryptone agar were incubated for 48 h. In both methods the modules were incubated at 35°C. Detection time was the time required to produce an accelerating impedance change of 0.8%. This amount of impedance change was selected because it represented a full strip chart channel width of change.
RESULTS Typical impedance changes caused by bacteria in pureed green beans are shown in Fig. 1. In this example there is an initial decrease in impedance which diminishes with time, yielding a roughly constant base line by 1.5 h. This initial decrease is often due to the temperature of the sample coming up to incubator temperature. Starting at around 3.2 h, there is an accelerating decrease in impedance, which reaches the criterion for detection (0.8% impedance change) at 3.7 h. This sample of green beans contained 100,000 organisms per g. The detection times for 257 frozen vegetable samples prepared according to method I, using
15
)0 Z
0
10_-
0
5
10
DETECTION TIME (Mourns)
FIG. 2. BACTOMETER Microbial Monitoring System detection times and standard plate count results for 257 frozen vegetable samples. The horizontal line at 100,000 organisms per g and the vertical line at 4.9 h mark the boundaries between acceptable and excessive contamination for the standard plate count and impedance methods, respectively. The shaded area represents the error inherent in the standard plate count: one standard deviation on each side of the 100,000-organisms per g level.
printed circuit board modules, were plotted against the logarithm of the initial concentration as determined by pour plates (see Fig. 2). The roughly linear relationship between the detection times and the logarithm of the concentration indicates that these samples grew with similar lag phases and generation times from sample to sample. The horizontal line at 100,000 organisms per g indicates the boundary between acceptable and unacceptable levels of microbial content. The vertical line at 4.9 h indicates the detection time, or cutoff time, that best divides all the samples into acceptable and unacceptable categories. The samples in the upper left quadrant (formed by the vertical and horizontal lines) are those that are unacceptable by both impedance and plate count. The samples in the lower right quadrant are those that are acceptable by both methods. Using 4.9 h as the cutoff time, 238 of 257 samples, or 92.6%, were correctly classified by the impedance method. The samples in the upper right quadrant are false negatives; that is, they are unacceptably contaminated by count but acceptable by impedance. Thus, plate FIG. 1. Typical impedance changes observed with 10 of 257 samples (3.9%) are false negatives. frozen vegetable samples. This curve represents the The samples in the lower left quadrant are false impedance changes observed with a sample of green beans with an initial concentration of 100,000 orga- positives; that is, they are acceptable by plate nisms per g. Detection of accelerating impedance count but unacceptable by impedance. Thus, 9 of 257 samples (3.5%) are false positives. The changes occurred at 3.7 h. lU
ul
I
0
5
10
16
APPL. ENVIRON. MICROBIOL.
HARDY ET AL.
percentage of positive samples missed by impedance is 10 samples of 123 positive samples, or 8.1%. Note the shaded region in Fig. 2. This is an approximation of the deviation to be expected in standard plate counts for frozen vegetables. Comparing 100 samples done in duplicate, it was shown that the mean standard deviation was 0.3 log. Some samples deviated by as much as 1.3 log. The shaded area represents all points within one standard deviation from the 100,000organisms per g level. Table 1 shows the results of analyzing the data separately for each type of vegetable. The TABLE 1. Cutoff times and percent agreement for various types of frozen vegetables Frozen vegetables
No. of samples
Cutoff time (h)
Percent agreement
Green beans Corn Peas Mixed Peas & carrots Others
48 48 47 41 29 44
4.9 4.7 4.5 4.7 5.3 5.3
95.8 93.7 95.7 95.1 93.1 88.6
257
4.9
92.6
All
cutoff time that gives the best percent agreement was determined for each class of vegetable. The "other" category includes samples of 10 kinds of frozen vegetables: carrots, cauliflower, broccoli, succotash, zucchini, okra, black-eyed peas, baby lima beans, and butter beans. The optimum cutoff times ranged from 4.5 h for peas to 5.3 h for peas and carrots, and the percent agreement ranged from 88.6% for the "other" category to 95.8% for green beans. To determine the reproducibility of the method, 100 samples of green beans were prepared according to method II and run in disposable modules. The same 100 samples were then again run in disposable modules according to method II and compared for data consistency and reproducibility. Table 2 shows the results of this comparison. The two runs are very similar, with the major differences being two additional false positives in the second run, which produces a corresponding 2% decrease in the overall agreement.
DISCUSSION The data presented here indicate that impedance measurements provide an alternative to the standard plate count for testing microbial
TABLE 2. Analysis of 100 samples of frozen green beans run twice for impedimetric detection of samples containing greater than 105 microorganisms per g Determination
Classification Cutoff time for 105 bacteria per g (h) Percent agreement Samples falsely classified as being under 105 (out of 100 samples) Samples falsely classified as being >105 (out of 100 samples) Samples falsely classified as being 105) Samples falsely classified as being >105 (out of all samples with counts
