Materials and Methods

Samples—All milk samples submitted from June 2006 through April 2007 by members of the Bovine Ambulatory Clinic of the Université de Montréal Faculté de Médecine Vétérinaire to the Bacteriology Laboratory at the Université de Montréal for standard bacteriologic testing were used for the study. Additional milk samples from cows with CM were collected and submitted by members of the Ormstown Veterinary Hospital, Ormstown, QC, Canada. Milk samples also were collected during a study conducted to evaluate the incidence of CM after parturition in cows treated with or without an internal teat sealant^a during the nonlactating period. Submitted milk samples were obtained aseptically by a veterinarian or an animal health technician during monthly veterinary herd health visits or from cows with CM. Most samples were cultured on the day of collection. Milk samples that could not be submitted to the laboratory within 24 hours after collection were frozen. Milk samples were submitted for standard bacteriologic testing and simultaneously tested by use of microbiological media plates.^b,c

Bacteriologic procedures—Bacteriologic analysis was performed by personnel at the clinical bacteriology laboratory at the Université de Montréal Faculté de Médecine Vétérinaire. The procedures were consistent with those described by the National Mastitis Council.¹¹ On arrival at the laboratory, fresh milk samples were cultured immediately and frozen samples were allowed to thaw at 20°C. Samples were vortexed and then streaked by use of 0.01-mL disposable plastic loops onto trypticase soy agar plates enriched with 5% sheep's blood.^d Plates were then incubated at 35°C for 24 hours in accordance with the standard method. After incubation, plates were examined and the colonies were tentatively identified on the basis of morphological features, pattern of hemolysis, Gram stain reaction, and results of a catalase test and then enumerated. MacConkey agar was used only after gram-negative bacilli were identified in colonies obtained after incubation for 24 hours. A second evaluation was made at 48 hours.

Gram-positive and catalase-positive cocci were subjected to a coagulase test or, if necessary, to a DNase test to distinguish between S aureus and CNS. Catalase-negative gram-positive cocci presumptively identified as streptococci were subjected to streptococcus identification tests (CAMP reaction, esculine hydrolysis, hippurate hydrolysis, inulin test, and raffinose fermentation).¹² Identification by use of a bacterial identification system^e was performed when confirmation was necessary. Gram-positive bacilli were classified on the basis of their microscopic morphology and results of the catalase test. Gram-negative bacilli were reinoculated onto MacConkey agar^f to assess lactose fermentation and identified on the basis of results of several tests (oxidase, triple sugar iron, urea, citrate, indole, and motility). Other bacteria or yeast were identified on the basis of their morphology by use of Gram staining.

A milk sample was considered to have a positive result for a bacteria via standard bacteriologic testing when the bacteria were isolated from the primary sample. All bacterial growth was identified and recorded. A sample was considered contaminated when ≥ 3 types of bacteria were identified.

Procedures for microbiological media plates— Immediately after inoculation onto trypticase soy agar plates, each mastitic milk sample was also inoculated on 2 sets of RCC,^b AC,^b and SRSC^c plates. The first set comprised nondiluted milk samples (1 mL), and the second set comprised diluted milk samples (1:10 dilution [0.1 mL of milk and 0.9 mL of Butterfield PBS solution^g]). Microbiological media plates were used by laboratory technicians who were trained and had obtained certification from the manufacturer on microbiological media plating and interpretation techniques. Evaluation and interpretation of the SRSC plates was performed (mean ± SD) 24 ± 2 hours after inoculation. When colonies were detected on an SRSC plate, an SRSC disk^c was used. The top film of the SRSC plate was lifted, the SRSC disk was placed in the well of the plate, the top film was lowered back into position, and gentle pressure was applied by firmly sliding a finger across the disk area to ensure uniform contact between the disk and gel. The SRSC plates were incubated for an additional 1 to 3 hours and then reevaluated. When results of an SRSC plate were positive for S aureus, the colony was recultured by use of standard methods and verified as S aureus.

The RCC plates were evaluated 6 to 12 hours after inoculation, and a final evaluation was performed 24 hours after inoculation. When results of the RCC test were positive for a coliform, the colony was cultured by use of standard methods and verified as a coliform. The AC plates were evaluated in a plate reader^h (a machine that counted the colonies on the AC plates) 24 and 48 hours after inoculation.

Milk samples obtained from cows with CM were frozen after analysis and stored at −20°C for 7 days. Milk samples were then thawed and inoculated again on RCC, AC, and SRSC plates. These results were compared with results for the fresh milk samples.

Fresh milk samples from cows that recently calved or that had an elevated SCC during lactation were inoculated only on SRSC plates (diluted and nondiluted) for S aureus screening. Criteria for an elevated SCC during lactation were an SCC > 200,000 cells/mL and a 50% increase in SCC from the value for the preceding month or an SCC > 1,000,000 cells/mL. The interpretation procedures were the same as for mastitic milk samples.

Statistical analysis—Statistical analysis was performed by use of statistical software.ⁱ Sensitivity, specificity, PPV, NPV, and κ of the microbiological media tests (diluted and nondiluted) were compared with results for the standard milk bacteriologic tests by use of χ² analysis.¹³ Results for the SRSC and RCC plates were compared with results of standard bacteriologic testing and a criterion-referenced standard; the criterion-referenced standard had positive results when the bacteria were isolated via standard bacteriologic testing or through the identification of the positive microbiological media isolate as confirmed by the bacteriology laboratory.

When the results for AC plates were positive but results for RCC and SRSC plates were negative, the isolate was considered to be a Streptococcus sp. This result was compared with results for standard bacteriologic testing to calculate the test characteristics of the AC plate.

Contaminated samples (≥ 3 types of bacteria) were removed from the data set. Determination of a true IMI was not performed because it would have required further diagnostic efforts that were not included in the objectives of the study.

Results

A total of 1,208 fresh milk samples from post-partum cows were submitted for analysis. Sixty-three samples were removed (4 samples were removed because colonies that had positive results on the SRSC plates were not recultured for confirmation, and 59 were contaminated samples). Thus, 1,145 fresh milk samples were used in the analysis. Staphylococcus aureus was isolated from 79 of 1,145 (6.9%) and 98 of 1,145 (8.6%) samples via standard bacteriologic testing and the criterion-referenced standard, respectively. Sensitivity of the SRSC plate, compared with standard bacteriologic testing, for identification of S aureus was highest with diluted samples (74/79 [93.7%] samples; Table 1). Agreement between nondiluted and diluted samples for the SRSC plate and standard bacteriologic testing was good at κ = 0.79 and 0.80, respectively. The SRSC plates were considered to have false-positive results when standard bacteriologic testing did not identify S aureus. However, on the basis of the definition of the criterion-referenced standard in the study, if a colony had positive results for the SRSC plates, it was recultured via standard bacteriologic testing and verified as S aureus. When results for nondiluted samples cultured on the SRSC plates were compared with results for standard bacteriologic testing, there were 23 false-positive results (13 S aureus and 10 CNS), as determined by reculturing colonies from SRSC plates via the criterion-referenced standard. Therefore, the SRSC plates had only 10 false-positive results attributable to CNS, whereas standard bacteriologic testing had 13 false-negative results. For diluted samples on the SRSC plates, there were 28 false-positive results (17 S aureus, 9 CNS, and 2 no growth), compared with results for standard bacteriologic testing. In the diluted samples, 17 of 28 (60.7%) false-positive results were attributable to S aureus. When results for the SRSC plates were compared with results for the criterion-referenced standard, the PPV for nondiluted and diluted samples was higher (88.0% and 88.2%, respectively; Table 2).

A total of 303 milk samples from cows with a high SCC were submitted. Twenty-four samples were removed (3 samples were removed because colonies with positive results on the SRSC plates were not recultured for confirmation, and 21 were contaminated samples). Thus, 279 fresh milk samples from cows with a high SCC were used for the analysis. Staphylococcus aureus was isolated from 38 of 279 (13.6%) and 47 of 279 (16.8%) samples by use of the standard bacteriologic testing and the criterion-referenced standard, respectively. Sensitivity and specificity for the SRSC plates for identification of S aureus, compared with results for standard bacteriologic testing, were 76.3% and 95.4%, respectively, in nondiluted samples and 86.8% and 96.7%, respectively, in diluted samples (Table 3). The PPV was 8.0% higher when diluted samples were used. When results for standard bacteriologic testing were compared with results for the criterion-referenced standard for nondiluted and diluted milk samples, there were 11 (8 S aureus, 2 CNS, and 1 no growth) and 8 (6 S aureus, 1 CNS, and 1 no growth) false-positive results, respectively. Agreement between results for the SRSC plates and standard bacteriologic testing for diluted samples was very good (κ = 0.81). When results for the SRSC plates were compared with results for the criterion-referenced standard, the PPV increased by 20.0%, compared with the PPV for the standard bacteriologic testing (Table 4). The highest PPV (95.1%) and NPV (96.6%) were obtained with the diluted samples.

A total of 536 fresh and frozen mastitic milk samples were tested. Thirty-seven samples were removed (19 were removed because colonies with positive results on the SRSC plates were not recultured for confirmation, 3 had incomplete results, and 15 were contaminated samples). Therefore, 499 milk samples (308 fresh and 191 frozen) were used in the analysis. Staphylococcus aureus was isolated from 61 of 499 (12.2%) and 73 of 499 (14.6%) samples via standard bacteriologic testing and the criterion-referenced standard, respectively. Sensitivity of the SRSC plates was the highest for diluted frozen mastitic milk samples (Table 5). However, the highest PPV (82.5%) was achieved by use of fresh diluted samples. For the combined fresh and frozen milk samples, there were 17 (10 S aureus, 4 CNS, 1 Bacillus sp, and 2 no growth) and 12 (9 S aureus, 1 CNS, and 2 no growth) false-positive results for the SRSC plates for nondiluted and diluted samples, respectively. For the nondiluted samples, 10 of 17 (58.8%) of the false-positive results were attributable to increased recovery of S aureus by use of the SRSC plates. When results for the SRSC plates were compared with results for the criterion-referenced standard, sensitivity was highest (74.2%) for diluted frozen milk samples (Table 6). For nondiluted and diluted fresh and frozen samples, the PPVs were 86.2% and 93.1%, respectively.

The RCC plates were evaluated for use in identification of coliforms in samples obtained from cows with CM. A total of 536 fresh and frozen mastitic milk samples were tested. Thirty-one samples were removed (13 samples were removed because colonies with positive results on the RCC plates were not recultured for confirmation, 2 had incomplete results, and 16 were contaminated samples). Therefore, 505 milk samples (307 fresh and 198 frozen) were used in the analysis. Coliforms were isolated via standard bacteriologic testing and the criterion-referenced standard from 92 of 505 (18.2%) and 132 of 505 (26.1%) samples, respectively. Of the 92 samples in which coliforms were detected via standard bacteriologic testing, 73 of 92 (79.3%) were E coli, 13 of 92 (14.1%) were Klebsiella spp, and 6 of 92 (6.5%) were Enterobacter spp. Sensitivity (81.6%) for the RCC plates was the highest for the diluted frozen samples (Table 7). The highest agreement (κ = 0.67) was achieved with diluted fresh samples. There were 34 false-positive results (19 E coli, 5 Klebsiella spp, and 10 coliforms) for the nondiluted samples and 35 false-positive results (19 E coli, 4 Klebsiella spp, and 12 coliforms) for the diluted samples. When results for the RCC plates were compared with results for the criterion-referenced standard, sensitivity (82.9%) was the highest with diluted frozen samples (Table 8). When results for the RCC plates were compared with results for the criterion-referenced standard, the PPV for diluted fresh and frozen samples combined improved to 99.1%.

Use of the RCC plates to detect coliforms as early as 6 to 12 hours after inoculation (ie, start of incubation) was evaluated. Positive results for coliform growth were indicated by a color change and gas formation around the potential colony. Of the 505 samples cultured on RCC plates, 369 were evaluated 6 to 12 hours after inoculation for their use in prediction of coliform growth. Compared with results for the standard bacteriologic testing, sensitivity of the RCC plates was low, ranging from 23.3% to 48.4% for all types of samples. The highest NPV (93.6%) and agreement (κ = 0.49) were obtained with diluted fresh milk samples. Compared with results for the criterion-referenced standard, sensitivity of the RCC plates was slightly improved, ranging from 19.6% to 58.1%. The NPV ranged from 73.1% to 91.1% for all types of samples.

The AC plates were evaluated for their use in detecting streptococci in mastitic milk samples. Samples that had negative results for S aureus on the SRSC plates and negative results for coliforms on the RCC plates but that had growth on the AC plates were assumed to have streptococci. Therefore, samples with positive results for the SRSC and RCC plates were removed to determine the use of the AC plates for the detection of streptococci. For the 287 samples used for the analysis, streptococci were isolated from 57 of 287 (19.9%). Results for the evaluations conducted at 24 and 48 hours after inoculation were summarized (Table 9). Although the sensitivity was good, ranging from 69.2% to 100% at the 48-hour evaluation, the PPVs were extremely low, ranging from 11.3% to 25.0%, and there was poor to no agreement between results for the AC plates and the standard bacteriologic testing.

Test characteristics of the SRSC and RCC plates with frozen mastitic milk samples were evaluated in relation to results for standard bacteriologic testing. Not all of the mastitic samples were frozen; therefore, complete records for 217 samples cultured on SRSC plates and 222 samples cultured on RCC plates were available for analysis. For the SRSC plates, there was an increase of 8.7% in S aureus identification after freezing of nondiluted samples (Table 10). However, when frozen diluted samples were evaluated, S aureus recovery decreased by 2.8%. For the RCC plates, there was no difference in the sensitivity for frozen mastitic milk samples. There was a 3.9% decrease in coliform identification with the use of frozen diluted samples (Table 11).

Discussion

The sample size was larger for fresh milk samples from cows and samples from cows with a high SCC or CM in the study reported here, compared with the sample size in another study.⁶ The SRSC plates had a slightly higher sensitivity, compared with results for standard bacteriologic testing, for the detection of S aureus in fresh milk samples from postpartum cows, compared with that in a similar study.⁶ Sensitivity of the SRSC plates was lower in milk samples from cows with a high SCC or CM. The PPV for the SRSC plates,^c compared with results for standard bacteriologic testing, was good. The implications of a positive result for S aureus depend on the herd protocols and policies regarding udder health. A high PPV can provide high confidence that a positive test result means that a cow is truly infected. A high PPV is important in herds in which cattle with positive results are culled or segregated. The PPVs were lower when the SRSC plates were compared with standard bacteriologic testing versus the criterion-referenced standard. In the evaluation of the SRSC plates versus standard bacteriologic testing for samples from postpartum cows, 13 of 23 (56.5%) false-positive results were attributable to S aureus and 10 of 23 (43.5%) were attributable to CNS. There were 13 more S aureus isolates obtained by use of the SRSC plates than via standard bacteriologic testing. The microbiological media plates, compared with standard bacteriologic testing alone, would appear to have a high rate of false-positive results. The impact of a false-positive result attributable to CNS may be an issue on certain farms, especially if cows are culled on the basis of a positive test result. This is especially true in herds with a low prevalence of S aureus because use of the SRSC plates could lead to a higher percentage of false-positive results.

The SRSC plates have been found to be more sensitive than standard bacteriologic testing; thus, the criterion-referenced standard helps to evaluate the test characteristics by reducing the bias of an imperfect test.⁶ For bacteriologic tests, there can be a diagnostic bias when the criterion-referenced standard used to detect the pathogen is not perfect.¹⁴ This becomes an issue because the more imperfect a test, the greater the underestimation of the difference between the criterion-referenced standard and any other test. If a perfect criterion-referenced standard does not exist, one often can be created by including or combining existing tests.¹³

The RCC plates were extremely effective for use in detecting coliforms in mastitic milk samples at the 24-hour evaluation. When results for the RCC plates were compared with results for standard bacteriologic testing, 100% of the false-positive results were attributable to coliforms. The predictive values of the RCC plates reflect how well this test will work in field conditions. The PPV indicates the likelihood that a milk sample with a positive result for the RCC plates is attributable to a coliform. However, there is always a question when there is a positive coliform test result whether there is truly an IMI or the coliform is a contaminant in the milk sample. The present study was performed to evaluate the effectiveness of coliform identification and not the threshold considered for a true IMI.

Use of the color change of the RCC plates at 12 hours after inoculation did not prove to be sensitive for detection of coliforms. For the frozen samples, the PPV was low and there was poor agreement between the color change for the RCC plates and results of standard bacteriologic testing. This may be explained by the nature of the RCC plates. The color change of the RCC plates was based on pH. Mastitic milk typically has a higher pH than does normal milk.¹⁵ This may negatively affect the color change of the RCC plates. The benefit of the color change at 12 hours is the potential to have results available before the next milking so appropriate treatment decisions can be made for cows with CM.

The AC plates were not effective for streptococci differentiation in mastitic milk samples in the study reported here. In the present study, SRSC and RCC plates with positive results were removed to eliminate S aureus and coliforms. This OFCS was adapted from the procedures used in another study⁸ in which mastitic milk samples were inoculated on the SRSC, coliform culture, and AC plates. Samples that had negative results for S aureus and coliforms but that had growth on the AC plates were assumed to be streptococci. The problem with this assumption is that it does not account for other pathogens (eg, CNS). In the present study, the proportion of plates on which streptococci and CNS were isolated was 57 of 287 (19.9%) and 53 of 287 (18.5%), respectively. More than 80% of the AC plates had positive results for bacterial growth. This resulted in extremely poor predictive values for the AC plates for streptococci determination. In another study,⁷ investigators used the AC plates in combination with coliform culture plates for evaluation of CM samples. The objective of that study was to differentiate gram-positive from gram-negative bacteria. Clinical mastitis samples that had negative results on the coliform culture plates but positive results on the AC plates were considered to contain gram-positive organisms, and appropriate treatment was implemented. The sensitivity and specificity of the microbiological media OFCS for differentiating gram-positive and gram-negative organisms were 93.8% and 70.1%, respectively. Herds that require streptococci differentiation for treatment and management decisions should use another available OFCS.^j

In another study,⁶ investigators determined that the test characteristics of the SRSC plates were highly dependent on an evaluator's ability to interpret the pink zone. In that study,⁶ the zone reaction was quantified as weak or distinct. The specificity and PPV for S aureus detection increased from 87.2% and 31.0% to 96.0% and 57.1%, respectively, when distinct pink zones were seen. In the present study, investigators made no attempt to classify the pink zones. In addition, evaluation of the microbiological media plates was performed by trained laboratory technicians and veterinarians to reduce variability among evaluators.

To evaluate agreement between the tests, results for the microbiological media tests were compared with results for the standard bacteriologic testing and not with results for the criterion-referenced standard. The agreement was very good to good for the SRSC and RCC plates, but there was no agreement of the AC plates for the detection of streptococci. Therefore, agreement was acceptable only between the SRSC and RCC plates and standard bacteriologic testing.

Inoculum volumes are thought to play an important role in the sensitivity and specificity of a diagnostic test.^16–18 A study¹⁶ conducted to evaluate different inoculum volumes revealed that the sensitivity and specificity for S aureus were higher when a 0.1-mL inoculum was used. Use of a single composite milk sample and an inoculum of 0.1 mL yielded sensitivity and specificity of 92% and 86%, respectively.¹⁸ Inoculum volumes have been evaluated for the diagnosis of mastitis in samples obtained from clinically affected glands.¹⁶ In that study,¹⁶ investigators found that there was no difference in culture outcomes between inoculum volumes of 0.1 and 0.01 mL. In the present study, test characteristics of the microbiological media plates varied depending on the dilution. Also, the volume of milk inoculated was 100 times (1 mL) or, if diluted, 10 times (0.1 mL) as great as the typical volume of 0.01 mL used for standard bacteriologic testing. Consequently, we could expect that the number of colonies observed on the microbiological media plates would be 10 to 100 times as high as the number of colonies isolated via standard bacteriologic testing. When colonies are too numerous to count on the microbiological media plates and the growth media is readily used, ideal growth of organisms is limited and interpretation is more difficult. Dilution of milk samples makes enumeration and interpretation easier. This is one of the main reasons that the sensitivity of the microbiological media plates is expected to be even higher than that of standard bacteriologic testing. We suggest the use of diluted samples for the microbiological media system, especially for RCC plates.

Numerous studies^19–23 have been conducted to evaluate the effect of freezing of milk samples on results of bacteriologic culture. In 1 study,¹⁹ investigators determined that freezing decreased the number of samples with positive results for E coli samples and increased the number of CNS. Also, freezing had no effect on detection of S aureus or streptococci in that study.¹⁹ The implications of these findings stress the importance of the use of fresh milk samples if coliform mastitis is suspected. Investigators had contradictory findings in another study.²⁰ Those investigators found that freezing had no effect on viability of any pathogens (S aureus, Staphylococcus hyicus, Streptococcus dysgalactiae, Streptococcus uberis, Corynebacterium bovis, and E coli).²⁰ When the effects of freezing on the isolation of S aureus were evaluated, it was found that use of incubation in broth combined with freezing and incubation resulted in the highest S aureus isolation percentage.²¹ In addition, maximal sensitivities for the detection of S aureus were obtained when milk samples were fresh or frozen premilking samples and frozen postmilking samples.²² Results of the present study supported these findings because freezing fresh mastitic milk samples increased the sensitivity for S aureus detection by 8.7% and also decreased the sensitivity for coliform detection by 3.9%.

The costs of the plates were comparable to and lower than those for standard methods, depending on the regional laboratory fees. Laboratory fees for standard bacteriologic testing in the region of the study reported here varied from $7.50 to $12.00/sample.^k An SRSC plate and SRSC disk cost approximately $2.34 and $1.80, respectively. Thus, the total cost for a positive result on the S aureus test would have been $4.14 plus materials and labor. However, the SRSC plate is limited to the detection of S aureus. The AC plates include all bacteria cultured, but they do not differentiate among the types of bacteria. An AC plate cost approximately $0.80. The RCC plates will detect only coliforms. The largest drawback to the RCC plates is the cost; RCC plates were approximately twice the cost ($2.60 vs $1.00) of coliform culture plates that only yield results at 24 hours. A combination of the SRSC, AC, and RCC plates will enable more specific bacteria identification. The total cost of the media will depend on the specific herd objectives of the OFCS. With costs slightly lower than those for standard methods, an additional main advantage of the microbiological media system is the rapid turnaround time for results. Rapid diagnosis is necessary for the successful implementation of on-farm treatment protocols.

The microbiological media systems are not intended to replace a high-quality laboratory or identify all organisms that may be present. Certain organisms, such as Mycoplasma spp, will not grow on these media. However, the microbiological media plates do allow for a more rapid diagnosis than can be obtained via standard bacteriologic testing. As such, it has a definite advantage for use with an OFCS in which rapid decisions are required for implementation of treatment of cows with CM and segregation programs for S aureus. There is no streptococci determination with the microbiological media system used in the present study.

The dairy industry has made an important shift away from the use of intramammary antimicrobial treatment at the first sign of clots in milk. A rapid OFCS can be useful in the implementation of treatment protocols and overall herd milk quality-control programs. For ideal treatment, timely and accurate tests are needed to differentiate among infections caused by gram-positive organisms, gram-negative organisms, and those that yield no growth. For mastitis control programs, the ability to rapidly identify S aureus and coliforms would be a valuable asset. The SRSC and RCC plates can be used to rapidly identify pathogens in milk. The highest test characteristics were obtained when fresh milk samples were diluted (1:10). However, freezing milk samples increased the detection of S aureus. Analysis of the results revealed that the SRSC and RCC plates were comparable with standard bacteriologic culture for the isolation of S aureus and coliforms.