Advertisement

Within-herd prevalence thresholds for herd-level detection of mastitis pathogens using multiplex real-time PCR in bulk tank milk samples

Open ArchivePublished:August 02, 2017DOI:https://doi.org/10.3168/jds.2016-12385

      ABSTRACT

      The objective of the study was to assess the value of quantitative multiplex real-time PCR examination of bulk tank milk samples for bovine mastitis pathogens as a tool for herd level diagnosis. Using a logistic regression model, this study is aimed at calculating the threshold level of the apparent within-herd prevalence as determined by quarter milk sample cultivation of all lactating cows, thus allowing the detection of a herd positive for a specific pathogen within certain probability levels. A total of 6,335 quarter milk samples were collected and cultured from 1,615 cows on 51 farms in Germany. Bulk tank milk samples were taken from each farm and tested by bacterial culture as well as the commercial PCR assay Mastit 4A (DNA Diagnostic A/S, Risskov, Denmark) identifying Staphylococcus aureus, Streptococcus dysgalactiae, Streptococcus agalactiae, and Streptococcus uberis. In addition, PCR was performed on pooled herd milk samples containing milk aliquots from all lactating cows in each of the 51 herds. Only 1 out of the 51 herds was found PCR positive for Streptococcus agalactiae in bulk tank and pooled herd milk samples, and cultured quarter milk samples. Spearman's rank correlations between the cycle threshold value of bulk tank milk PCR and the apparent within-herd prevalence were calculated in regard to Staphylococcus aureus, Streptococcus dysgalactiae, and Streptococcus uberis. For these pathogens, significant correlations were found. If 1 bulk tank milk sample per herd was tested, the estimated within-herd prevalence thresholds for 90% probability of detection were 27.6% for Staphylococcus aureus, 9.2% for Streptococcus dysgalactiae, and 13.8% for Streptococcus uberis on the cow level. On the quarter level, the within-herd prevalence had to be at least 32.6% for Staphylococcus aureus, 1.7% for Streptococcus dysgalactiae, and 4.3% for Streptococcus uberis to detect a herd as positive using a single bulk milk sample. The results indicate that mastitis pathogens in bulk tank milk can be identified by the applied PCR assay. Bulk tank milk examination is not a reliable tool for the identification of the named pathogens by single testing, but might be a valuable monitoring tool when used frequently with repeated testing. Furthermore, this approach could be a useful monitoring tool for detecting new pathogen occurrence in the herd.

      Key words

      INTRODUCTION

      Intramammary infections are the most common diseases in dairy cattle (
      • Gundling N.
      • Ruddat I.
      • Prien K.
      • Hellerich B.
      • Hoedemaker M.
      Erkrankungshäufigkeit von Milchviehherden in Schleswig-Holstein. Einfluß der Milchleistung der ersten Milchleistungsprüfung der Laktation, der Herdenmilchleistung und der Laktationsnummer.
      ) causing large economic losses. In Germany, in 2009, the damage to the national economy caused by mastitis was estimated to be €1.4 billion (

      DVG (Deutsche Veterinärmedizinische Gesellschaft). 2012. Leitlinien Bekämpfung der Mastitis des Rindes als Bestandsproblem. Page 68. DVG. 5. Aufl, Gießen, Germany.

      ). Mastitis therapy and drying off account for the greatest amount of antibiotics used in dairy farming (
      • DANMAP (Danish Integrated Antimicrobial Resistance Monitoring and Research Programme)
      ). Dairy herd managers consider good udder health vital, not only for economical and animal welfare reasons, but also to minimize antibiotic use to prevent antimicrobial resistance. Knowledge about the mastitis pathogens prevalent in a certain herd is necessary for the control, reduction, and prevention of udder diseases (
      • Krömker V.
      • Friedrich J.
      Empfehlungen zum diagnostischen Aufwand im Rahmen der Mastitisbekämpfung auf Bestandsebene.
      ). Reliable diagnostic tools are essential for developing a farm-specific mastitis prevention plan. Bacterial culture (BC) of quarter milk samples is considered the gold standard for mastitis diagnosis, and isolation of the bacteria involved is the prerequisite to perform an antibiotic susceptibility test. Some disadvantages of cultivation are that it is labor intensive and takes a minimum of 24 to 48 h cultivation time (
      • Riffon R.
      • Sayasith K.
      • Khalil H.
      • Dubreuil P.
      • Drolet M.
      • Lagace J.
      Development of a rapid and sensitive test for identification of major pathogens in bovine mastitis by PCR.
      ;
      • Gillespie B.E.
      • Oliver S.P.
      Simultaneous detection of mastitis pathogens, Staphylococcus aureus, Streptococcus uberis, and Streptococcus agalactiae by multiplex real-time polymerase chain reaction.
      ;
      • Koskinen M.T.
      • Wellenberg G.J.
      • Sampimon O.C.
      • Holopainen J.
      • Rothkamp A.
      • Salmikivi L.
      • van Haeringen W.A.
      • Lam T.J.G.M.
      • Pyörälä S.
      Field comparison of real-time polymerase chain reaction and bacterial culture for identification of bovine mastitis bacteria.
      ). A substantial proportion of clinical and subclinical mastitis samples are negative for bacterial growth (
      • Bradley A.J.
      • Leach K.A.
      • Breen J.E.
      • Green L.E.
      • Green M.J.
      Survey of the incidence and aetiology of mastitis on dairy farms in England and Wales.
      ;
      • Taponen S.
      • Salmikivi L.
      • Simojoki H.
      • Koskinen M.T.
      • Pyörälä S.
      Real-time polymerase chain reaction-based identification of bacteria in milk samples from bovine clinical mastitis with no growth in conventional culturing.
      ). Quantitative real-time PCR (qPCR) technique is an alternative method for examining milk specimens for mastitis pathogens. It provides results within only a few hours, but is more expensive than BC. Polymerase chain reaction shows a higher sensitivity than BC for detecting bacteria in milk samples according to some authors (
      • Taponen S.
      • Salmikivi L.
      • Simojoki H.
      • Koskinen M.T.
      • Pyörälä S.
      Real-time polymerase chain reaction-based identification of bacteria in milk samples from bovine clinical mastitis with no growth in conventional culturing.
      ;
      • Bexiga R.
      • Koskinen M.T.
      • Holopainen J.
      • Carneiro C.
      • Pereira H.
      • Ellis K.A.
      • Vilela C.L.
      Diagnosis of intramammary infection in samples yielding negative results or minor pathogens in conventional bacterial culturing.
      ;
      • Spittel S.
      • Hoedemaker M.
      Mastitis diagnosis in dairy cows using PathoProofTM real-time polymerase chain reaction assay in comparison with conventional bacterial culture in a Northern German field study.
      ). On the other hand, qPCR does not differentiate between viable, growth-inhibited, or dead bacteria. Whereas the staphylococcal β-lactamase gene can be detected with qPCR, a BC is still necessary for a full antibiotic susceptibility test.
      For herd level monitoring, bulk tank milk (BTM) testing is an easy-to-use and cost-saving alternative to individual sample testing. The SCC of the BTM, which is frequently tested by dairy companies at milk collection in some countries, is a prompt, but rough indicator for the prevalence of subclinical mastitis in the herd. Should the SCC in BTM increase, it is not possible to detect the causing pathogen by a single examination of a BTM sample by BC (
      • Ruegg P.L.
      • Reinemann D.J.
      Milk quality and mastitis tests.
      ). Sensitivity for the detection of mastitis pathogens in BTM is known to be low (
      • Godkin M.A.
      • Leslie K.E.
      Culture of bulk tank milk as a mastitis screening test: A brief review.
      ;
      • Olde Riekerink R.G.
      • Barkema H.W.
      • Scholl D.T.
      • Poole D.E.
      • Kelton D.F.
      Management practices associated with the bulk-milk prevalence of Staphylococcus aureus in Canadian dairy farms.
      ;
      • da Costa L.B.
      • Rajala-Schultz P.J.
      • Schuenemann G.M.
      Management practices associated with presence of Staphylococcus aureus in bulk tank milk from Ohio dairy herds.
      ). Therefore, our hypothesis is that it is advantageous to detect major mastitis pathogens via single or repeated testing of BTM by qPCR. The objective of this study was to identify and describe the relation between the apparent within-herd-prevalence (WHPapp) and the cycle threshold (Ct) values of the qPCR aiming at quantification of the bacteria in BTM, and to calculate a threshold of the WHPapp that allows the identification of a herd as positive for the relevant pathogen. Besides BTM, we also used pooled herd milk samples as an alternative specimen.

      MATERIALS AND METHODS

      Dairy Herds

      For this study, 51 small and medium-sized farms located in the German Federal States Thuringia (26) and Hesse (25) were selected. Herds of this size were chosen because of a common lack of routine mastitis control programs in such herds, and therefore a screening of BTM for mastitis pathogens could be beneficial. Mean herd size of the study farms was 72 (9–144) cattle. On these farms, quarter milk samples were taken from each lactating cow not excluded from milk delivery. Therefore, cows with signs of clinical mastitis, treated cows, and those within the first 5 d postcalving were not sampled. The average number of cows tested on each farm was 32 (3–61). These cows were kept in multiple housing conditions (free-stall barns with various beddings, tie-stall with or without pasture). Average herd milk yield varied between 4,700 to 11,500 kg/cow per year.

      Milk Samples

      The samples were collected from October to December 2014 by Animal Health Service of Thuringia veterinarians and the first author. The milk samples were collected during the daily milking routine after udder preparation according to the “Guidelines for aseptic collection of milk samples and for isolation and identification of mastitis pathogens” of the German Veterinary Society (

      DVG (Deutsche Veterinärmedizinische Gesellschaft). 2009. Leitlinien zur Entnahme von Milchproben unter antiseptischen Bedingungen und Isolierung und Identifizierung von Mastitiserregern. DVG. 2. Aufl, Gießen, Germany.

      ). After the milking routine was finished, the milk in the bulk tank was stirred for at least 5 min, and 2 BTM samples were taken using sterile disposable syringes. The sample tubes for all types of milk samples contained boric acid as a preservative agent. After collection, all samples were immediately cooled and transported within 48 h to the laboratory of the Animal Health Service, Thuringian Animal Diseases Fund of the State of Thuringia in Jena.

      Pooling

      At the laboratory, quarter milk samples of each cow were pooled by transferring 0.25 mL per sample to another sterile tube, yielding cow-level milk samples. Furthermore, 2 pooled herd samples were assembled from 0.2 mL of each cow-level sample for each herd. Afterward, the tubes containing pooled herd samples were frozen and maintained at −20°C (±5°C) until analysis.

      Bacterial Culture

      The quarter milk and BTM samples were cultured according to the above mentioned guidelines. Using a glass bar, 0.01 mL of each milk sample was spread on an esculin sheep blood agar plate (Oxoid, Wesel, Germany) prepared with a streak of a β-hemolysin producing Staphylococcus aureus. After the process, the glass rod with the adhering residual milk was dipped into a glucose broth (Oxoid) for enrichment. The agar plates and the broths were incubated aerobically at 37°C (±2°C). The cultures were examined after 18 to 24 and 42 to 48 h, respectively. The broths were plated out after 18 to 24 h and incubated for 1 d before examination to increase the sensitivity of BC in regard to Streptococcus agalactiae and Staph. aureus. The colonies grown from primary cultures and the broths were evaluated for morphology, hemolysis, pigmentation, and Gram stain Streptococcus species were differentiated from Staphylococcus species via catalase reaction. For identification of Staph. aureus, a coagulase test with rabbit plasma (Remel, Lenexa, KS) was carried out. Streptococcus was tested for esculin hydrolysis, Christie, Atkins, and Munch-Peterson (CAMP) reaction, and Lancefield groups. Because of the milk inoculum of about 0.01 mL, the detection limit was about 100 cfu per mL in the sample. For Staph. aureus, Strep. agalactiae, and Strep. dysgalactiae, the culture was considered positive when one or more colonies in pure or mixed culture were found either in the primary culture or in the culture from the enrichment broth. A culture was classified as positive for Strep. uberis when at least 10 colonies had grown in the primary culture. As confirmation assays for Strep. uberis, cultivation on kanamycin esculin azide agar (Oxoid) as well as an agar diffusion assay with penicillin and rifampicin were performed. Inconclusive results were checked for Lancefield group antigen D via latex agglutination. After taking the required sample material for the BC, the BTM samples were frozen and stored under the same conditions as the pool samples until qPCR analysis was performed. The WHPapp of the different mastitis pathogens at the cow level and at the quarter level were calculated from the results of the bacteriological culture of the quarter milk samples.

      qPCR

      The qPCR testing of the BTM and pooled herd samples was performed according to the manufacturer's protocol () in the laboratory of the PCR assay manufacturer, DNA Diagnostic A/S (Risskov, Denmark), by a Thuringian Animal Health Service technician. The samples were tested directly and without any enrichment. The qPCR assay Mastit 4A is a multiplex qPCR capable of detecting DNA of 4 pathogens (Staph. aureus, Strep. agalactiae, Strep. dysgalactiae, and Strep. uberis). With qPCR, a quantitative estimation of DNA in the sample is possible: The fewer cycles necessary to get a fluorescence signal, the greater the amount of the target DNA present in the sample. The obtained result was expressed as a Ct value. A sample was considered positive at Ct ≤37.

      Statistical Methods

      Bacterial culture and qPCR results and descriptive statistics were processed in Microsoft Excel 2013 (Microsoft Corporation, Redmond, WA). For the calculation of sensitivity and specificity, a herd was considered positive for the respective detection method and the tested pathogen if at least one of the 2 BTM samples or pooled herd samples showed a positive result or if the infectious agent was isolated from one of the quarter milk samples. It was classified negative if both BTM samples or pooled herd samples or all quarter milk samples were tested negative. Further statistical analyses were done using the statistical software package BMDP/Dynamic (Release 8.1;
      ) concerning Staph. aureus, Strep. dysgalactiae, and Strep. uberis. As only one herd was positive for Strep. agalactiae, herd level sensitivity, specificity, and logistic regression models were not calculated for this pathogen. Statistical significance was designated as α = 0.05 for all tests.
      Using 2 BTM samples per herd, herd level sensitivity and specificity were calculated for each pathogen (Staph. aureus, Strep. dysgalactiae, and Strep. uberis) and each applied method (BC/PCR). Herd level sensitivity was the proportion of correctly identified positive herds among all positive herds as determined by BC of quarter milk samples. Herd level specificity was the proportion of correctly identified negative herds among all negative herds as determined by BC of quarter milk samples.
      Spearman's rank correlation coefficients (rs) were calculated (program BMDP3D) for analyzing the relationship between the WHPapp and the Ct of BTM and pooled herd samples. The relationship between WHPapp and herd status (positive/negative) as determined by testing either BTM or pooled herd samples was analyzed by means of separate logistic regression models. These models were calculated using WHPapp on either the cow or quarter level as an independent variable. The outcome (dependent variable) was a herd positive or negative for the specific pathogen in BTM samples or pooled herd samples in either single or double sampling. Using the single sampling approach, a randomized selection of 1 of the 2 samples was performed. Considering double sampling, a herd was counted as positive if the infectious agent was detected in either one or both BTM samples or one or both pooled herd samples. This resulted in 8 univariate and univariable logistic models for each pathogen. For the estimation of the threshold of WHPapp to detect a herd as positive with a given probability of detection (Pd), these logistic regression models were inverted and asymptotic standard errors of the estimates were calculated. Calculations were performed for Pd = 0.9 and Pd = 0.7.

      RESULTS

      Quarter milk samples were taken from 1,615 cows. As 125 quarters were not lactating and therefore not BC tested, a total of 6,335 quarter samples were available for analysis. The BC results and the mean and range of prevalences on the cow and quarter level are given in Table 1. Staphylococcus aureus was isolated from 5.0% of the quarters with quarter level WHPapp ranging from 0.0 to 38.5%. Streptococcus uberis was detected in 2.1% of the quarters with a maximum WHPapp of 19.3%. Streptococcus dysgalactiae (0.9%) exhibited a low quarter level prevalence with a maximum WHPapp of 4.5%. Twelve samples were positive for 2 pathogens. Five herds tested negative for all the selected pathogens. Only one out of 51 herds tested positive for Strep. agalactiae by BC in quarter milk samples as well as BTM. In this herd, both BTM samples and both pooled herd samples tested positive for Strep. agalactiae by qPCR. The prevalence for Strep. agalactiae was 32.1% on the cow level and 15.6% on the quarter level. As the pathogen was found in one herd only, it was excluded from further statistical calculations in this study.
      Table 1Pathogen findings by bacterial culture regarding the herds, cows, and udder quarters
      ItemnStaphylococcus aureus positiveStreptococcus uberis positiveStreptococcus dysgalactiae positiveStreptococcus agalactiaepositive
      Herds (no.)513633241
       %70.664.747.12.0
      Cows (no.)1,6152001034718
       %12.46.42.91.1
      Udder quarters (no.)6,3353141356435
       %5.02.11.00.6
      Cow-level prevalence (%)12.86.52.70.6
       Range of WHPapp
      Apparent within-herd prevalence.
      0.0–77.30.0–35.70.0–17.90.0–32.1
      Quarter-level prevalence (%)5.02.10.90.3
       Range of WHPapp0.0–38.50.0–19.30.0–4.50.0–15.6
      1 Apparent within-herd prevalence.
      The percentage of positive herds for BC of quarter milk and BTM samples as well as qPCR of BTM samples and pooled herd samples are shown in Table 2. Considering all 4 pathogens included in the qPCR, 93 positive PCR results were achieved in the BTM samples. Of the herds classified as positive by PCR of BTM samples, 62.4% showed a positive qPCR result for both tested samples, and 37.6% for only one of the 2 tested samples. Considering pooled herd samples, the Mastit 4 qPCR kit also yielded in 93 positive results, in 75.3% of the herds simultaneously in both samples. When BC was applied on BTM samples, the percentage of positive herds was much lower compared with the other testing method.
      Table 2Total number and percentage of herds (n = 51) that tested positive by different methods of examination (bacterial culture of quarter milk samples, quantitative real-time PCR (qPCR) of pooled herd samples, qPCR, and bacterial culture of bulk tank milk samples)
      PathogenBacterial culture of quarter milk samplesPCR of pooled herd samples
      A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      PCR of bulk tank milk samples
      A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      Bacterial culture of bulk tank milk samples
      A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      Number%Number%Number%Number%
      Staphylococcus aureus3670.62956.93568.61223.5
      Streptococcus uberis3364.73466.73058.800.0
      Streptococcus dysgalactiae2447.12956.92752.923.9
      Streptococcus agalactiae12.012.012.012.0
      1 A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      Herd level results of BTM samples or pooled herd samples in relation to the herd status as determined by BC of quarter milk as well as herd level sensitivity and specificity are summarized in Table 3. Some herds classified as positive by BC were determined negative according to the results of qPCR and vice versa. For BC examination of BTM samples, a very low sensitivity (0.0–33.3%) was calculated. Further analysis of BTM BC results were waived because sensitivity was too low for calculating a meaningful correlation as well as estimating a logistic regression model in relation to WHPapp.
      Table 3Comparison of the classification of herds by bacterial culture (BC) of quarter milk samples (reference) with the classification by quantitative real-time PCR analysis or culture of 2 bulk tank milk or pooled herd milk samples
      ItemStaphylococcus aureusStreptococcus uberisStreptococcus dysgalactiae
      PositiveNegativePositiveNegativePositiveNegative
      Bulk tank milk BC positive
      A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      (no.)
      1200020
      Bulk tank milk BC negative (no.)241533182227
      Sensitivity (%)33.30.08.3
      Specificity (%)100.0100.0100.0
      Bulk tank milk PCR positive
      A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      (no.)
      305264216
      Bulk tank milk PCR negative (no.)610714321
      Sensitivity (%)83.378.887.5
      Specificity (%)66.777.877.8
      1 A herd was considered as positive if at least 1 of 2 samples showed a positive result.
      Using Spearman's rank correlation coefficients, we examined the correlation between the WHPapp on the cow level as well as on the quarter level, and the Ct value of the qPCR analysis in either 1 or 2 BTM samples, or 1 or 2 pooled herd samples (Table 4). We found a correlation (P < 0.001) for all examined associations with only small deviations between the different pathogens. Spearman's rank correlation coefficients regarding the correlation between the Ct value of a single BTM examination and the WHPapp were −0.51 to −0.56. For pooled herd samples on the cow level, the correlation coefficients of the WHPapp were closer to −1 than for the BTM samples. When both pooled herd samples or both BTM samples were tested, the correlations were stronger than when testing only one sample. When analyzing WHPapp on the quarter level instead of the cow level, the correlation coefficients were all slightly closer to −1 for all associations. The strongest correlation with rs of −0.75 was found for Strep. uberis WHPapp at the quarter level when both pooled herd samples were considered.
      Table 4Spearman rank correlation coefficients for the relationship between the apparent within-herd prevalence (WHPapp) and cycle threshold values of the quantitative real-time PCR of bulk tank milk and pooled herd samples for the different pathogens
      P-value for all calculations <0.001; number of herds with valid test results = 51.
      ItemStaphylococcus aureusStreptococcus uberisStreptococcus dysgalactiae
      WHPapp (cow level)
       Bulk tank milk (1 sample)
      One randomly chosen bulk tank milk sample of the 2 samples of a herd.
      −0.55−0.51−0.56
       Bulk tank milk (both samples)−0.60−0.55−0.68
       Pooled herd (1 sample)
      One randomly chosen pooled herd sample of the 2 samples of a herd.
      −0.59−0.72−0.57
       Pooled herd (both samples)−0.71−0.74−0.65
      WHPapp (udder quarter level)
       Bulk tank milk (1 sample)−0.59−0.55−0.61
       Bulk tank milk (both samples)−0.66−0.57−0.71
       Pooled herd (1 sample)−0.62−0.74−0.58
       Pooled herd (both samples)−0.73−0.75−0.68
      1 P-value for all calculations <0.001; number of herds with valid test results = 51.
      2 One randomly chosen bulk tank milk sample of the 2 samples of a herd.
      3 One randomly chosen pooled herd sample of the 2 samples of a herd.
      Estimated threshold values (Table 5) of the WHPapp required to detect a herd as positive for each of the 3 pathogens were calculated using a logistic regression model. The threshold values varied depending on the pathogen, the kind of specimen (BTM or pooled herd sample), and the number of examined samples (single or double sampling). The highest threshold value of the WHPapp was calculated for Staph. aureus for BTM samples as well as for pooled herd samples for single sampling. The threshold value at a given Pd = 0.9 was 27.6% for Staph. aureus if 1 BTM sample was tested. In case of double sampling, the estimated threshold value dropped to 17.7%. Lower threshold values were found when pooled herd samples were used. The threshold value for Staph. aureus was 18.0% when a single pooled herd sample was tested and 16.1% in case of both pooled herd samples of a herd. Applying Pd = 0.7, the threshold value for Staph. aureus decreased to 15.5% for a single BTM sample examination. Threshold values for Staph. aureus were lower when the calculation was based on the udder quarter level rather than on the cow level, except for the examination of a single BTM sample, which resulted in a detection limit value of 32.6%. Threshold values for the tested streptococci proved to be lower than for Staph. aureus. When only a single BTM sample was examined, the threshold value for Strep. uberis at Pd = 0.9 was 13.8% at the cow level and 5.3% at the quarter level. When 2 samples were tested, the threshold values were even lower for the BTM samples as well as for the pooled herd samples. The lowest threshold values were found for Strep. dysgalactiae. In the case of at least approximately 9.2% of the cows of a herd being infected, the herd could be tested positive at Pd = 0.9 testing only 1 BTM specimen.
      Table 5Estimates of apparent within-herd prevalence (WHPapp) threshold value ± asymptotic SE and approximate 95% CI for classifying a positive herd as positive by quantitative real-time PCR for the different pathogens using 1 or 2 bulk tank milk samples or 1 or 2 pooled herd samples, respectively, at different probabilities of detection
      The calculation of the WHPapp on the cow level and on the quarter level is based on the bacterial culture of quarter milk samples.
      Pd
      Probability of classifying a positive herd as positive.
      (%)
      Staphylococcus aureusStreptococcus uberisStreptococcus dysgalactiae
      Bulk tank milk samplePooled herd sampleBulk tank milk samplePooled herd sampleBulk tank milk samplePooled herd sample
      WHPapp ± SE95% CIWHPapp ± SE95% CIWHPapp ± SE95% CIWHPapp ± SE95% CIWHPapp ± SE95% CIWHPapp ± SE95% CI
      WHPapp on cow level
       1 Sample tested
        7015.5 ± 4.17.4–23.612.3 ± 2.27.9–16.77.7 ± 1.93.9–11.64.9 ± 1.22.5–7.35.9 ± 1.43.2–8.74.8 ± 1.32.2–7.3
        9027.6 ± 7.812.0–43.118.0 ± 3.511.0–25.013.8 ± 3.56.8–20.98.4 ± 2.04.4–12.49.2 ± 2.24.7–13.68.2 ± 2.23.7–12.7
       2 Samples tested
        708.2 ± 2.72.9–13.610.4 ± 2.16.2–14.56.4 ± 1.53.4–9.44.2 ± 1.21.9–6.63.1 ± 0.81.5–4.83.0 ± 0.91.2–4.8
        9017.7 ± 5.46.9–28.416.1 ± 3.49.3–22.911.1 ± 2.65.8–16.37.9 ± 2.13.8–12.05.4 ± 1.42.6–8.35.7 ± 1.62.4–9.0
      WHPapp on quarter level
       1 Sample tested
        7016.1 ± 9.0−2.0–34.13.9 ± 0.82.2–5.52.6 ± 0.61.4–3.81.5 ± 0.40.7–2.31.1 ± 0.30.6–1.61.6 ± 0.40.7–2.4
        9032.6 ± 19.8−7.0–72.26.2 ± 1.43.4–9.14.3 ± 1.02.3–6.42.8 ± 0.71.4–4.21.7 ± 0.40.9–2.52.6 ± 0.71.1–4.1
       2 Samples tested
        702.4 ± 0.80.9–4.03.1 ± 0.61.9–4.31.7 ± 0.40.9–2.51.2 ± 0.30.5–1.80.9 ± 0.20.4–1.40.8 ± 0.30.3–1.4
        905.0 ± 1.52.0–8.04.7 ± 1.02.8–6.62.9 ± 0.71.6–4.32.1 ± 0.61.0–3.21.5 ± 0.40.7–2.41.6 ± 0.50.6–2.6
      1 The calculation of the WHPapp on the cow level and on the quarter level is based on the bacterial culture of quarter milk samples.
      2 Probability of classifying a positive herd as positive.

      DISCUSSION

      Mastitis control in dairy herds requires the identification of major mastitis pathogens to implement adequate control measures. Testing quarter milk samples or cow level samples is labor intensive and may be expensive depending on the laboratory infrastructure available to the farmer. Therefore, attempts were made to use BTM as specimen to monitor mastitis pathogen occurrence within a herd. The BTM samples were examined via BC in many studies, but only a few have simultaneously tested pathogen prevalence in the cows. Several studies in which quarter milk samples and BTM were examined, showed a low sensitivity (9.2 to 58% for Staph. aureus) for BTM BC (
      • Godkin M.A.
      • Leslie K.E.
      The relationships between bulk tank milk culture, management factors used in mastitis control and the herd prevalence of mastitis.
      ;
      • Bartlett P.C.
      • Miller G.Y.
      • Lance S.E.
      • Heider L.E.
      Use of bulk tank and milk filter cultures in screening for Streptococcus agalactiae and coagulase positive staphylococci.
      ;
      • Wilson D.J.
      • Gonzalez R.N.
      Evaluation of milk culture, SCC and CMT for screening herd additions.
      ). This low sensitivity was also confirmed by our study. As a further result, it was ascertained that estimating the number of infected cows by the quantity of the mastitis pathogens present in the BTM was not possible (
      • Ruegg P.L.
      • Reinemann D.J.
      Milk quality and mastitis tests.
      ). The PCR testing of BTM samples was described as an easy and cost-effective herd level screening method for infectious agents such as Coxiella burnetii (
      • Rodolakis A.
      • Berri M.
      • He'chard C.
      • Caudron C.
      • Souriau A.
      • Bodier C.C.
      • Blanchard B.
      • Camuset P.
      • Devillechaise P.
      • Natorp J.C.
      • Vadet J.P.
      • Arricau-Bouvery N.
      Comparison of Coxiella burnetii shedding in milk of dairy bovine, caprine, and ovine herds.
      ).
      • Katholm J.
      • Bennedsgaard T.W.
      • Koskinen M.T.
      • Rattenborg E.
      Quality of bulk tank milk samples from Danish dairy herds based on real-time polymerase chain reaction identification of mastitis pathogens.
      examined the BTM of 4,258 Danish herds using a commercial real-time PCR and concluded that real-time PCR can be applied in the same way as BC for monitoring BTM samples, and that PCR is particularly useful for bacteria with low prevalence. Other authors stated that PCR results of BTM testing should be confirmed by BC of BTM samples and, if possible, of composite milk samples (
      • Zanardi G.
      • Caminiti A.
      • Delle Donne G.
      • Moroni P.
      • Santi A.
      • Galletti G.
      • Tamba M.
      • Bolzoni G.
      • Bertocchi L.
      Short communication: Comparing real-time PCR and bacteriological cultures for Streptococcus agalactiae and Staphylococcus aureus in bulk-tank milk samples.
      ). Consequently, it is controversial whether this approach is appropriate for herds with a low prevalence of the relevant pathogens.
      Therefore, the calculation of a WHPapp threshold that allows the identification of a herd as positive is of interest for professional use of this approach. Our study is the first one to estimate these thresholds of the WHPapp at the cow level and the quarter level as determined by BC of quarter milk samples. We examined every lactating cow of the herd that was included in milk delivery. In doing this, we ensured that each cow that contributed to the bulk tank was involved in the calculation of WHPapp. The WHPapp thresholds were calculated for different settings such as single and double sampling of either a BTM sample or a pooled milk sample. We analyzed BTM samples by qPCR and BC. Depending on the pathogen, a sensitivity of 78.8 to 87.5% for qPCR analysis of BTM was calculated using BC of quarter samples as the reference method (Table 3). In BC testing of BTM samples, sensitivity values were much lower (0.0–33.3%). The few herds detected as positive by BC of BTM samples also showed positive in the reference method, the BC of quarter milk samples, resulting in a high specificity of 100%. Because of the low sensitivity, the result regarding specificity must be interpreted with caution. Taking into account the results of other studies (
      • Godkin M.A.
      • Leslie K.E.
      The relationships between bulk tank milk culture, management factors used in mastitis control and the herd prevalence of mastitis.
      ;
      • Bartlett P.C.
      • Miller G.Y.
      • Lance S.E.
      • Heider L.E.
      Use of bulk tank and milk filter cultures in screening for Streptococcus agalactiae and coagulase positive staphylococci.
      ;
      • Wilson D.J.
      • Gonzalez R.N.
      Evaluation of milk culture, SCC and CMT for screening herd additions.
      ), qPCR seems to be a more sensitive method for BTM testing.
      In 15 cases we found pathogens in BTM using qPCR that could not be detected in the udder quarters of the herd via BC (4 × Strep. uberis, 5 × Staph. aureus, and 6 × Strep. dysgalactiae). This reduces the specificity of qPCR. This might be due to contamination of BTM samples, the different specimen used in the reference method or a limited sensitivity of the reference method. First, quantity and composition of the bacteria in BTM are influenced not only by the milk itself but also by several aspects such as the cleanliness of the udder skin, milker's hands, milking equipment, bulk tank, and cooling process (
      • Reinemann D.J.
      • Mein G.A.
      • Bray D.R.
      • Reid D.
      • Britt J.S.
      Troubleshooting high bacteria counts in farm milk.
      ). All of the examined pathogens can be found in the cow's environment as well as on the cow's udder surface (
      • Roberson J.R.
      • Fox L.K.
      • Hancock D.D.
      • Gay J.M.
      • Besser T.E.
      Ecology of Staphylococcus aureus isolated from various sites on dairy farms.
      ;
      • Wolter W.
      • Kloppert B.
      • Castaneda H.
      • Zschöck M.
      ;
      • Jørgensen H.J.
      • Nordstoga A.B.
      • Sviland S.
      • Zadoks R.N.
      • Sølverød L.
      • Kvitle B.
      • Mørk T.
      Streptococcus agalactiae in the environment of bovine dairy herds—Rewriting the textbooks?.
      ). Second, the BTM does not contain the foremilk of those cows whose milk the farmer considers fit for selling, and for the BC only foremilk was used, as recommended for BC (

      DVG (Deutsche Veterinärmedizinische Gesellschaft). 2009. Leitlinien zur Entnahme von Milchproben unter antiseptischen Bedingungen und Isolierung und Identifizierung von Mastitiserregern. DVG. 2. Aufl, Gießen, Germany.

      ). Third, tubes with boric acid as a preservative agent were used for all samples. This could influence the viability of the bacteria in the milk samples and thus also the sensitivity of BC to detect bacterial growth.
      We accounted for the selection bias by limiting sampling to cows without clinical mastitis because the selling of milk from cows with clinical mastitis is prohibited in the European Union. Regarding herd level sensitivity in large dairy herds, the dilution effect in large bulk tanks is another major problem of BTM sampling. In low prevalence situations particularly, the number of bacteria in the sample may fall below the detection limit and cause a negative BTM test result.
      Using Spearman's rank correlation coefficients, we demonstrated a correlation (P < 0.001) between the Ct value of the qPCR and WHPapp. Depending on the pathogen, specimen, and number of samplings in our study, correlations ranged between −0.51 and −0.74 on the cow level and between −0.55 and −0.75 on the quarter level. A significant correlation for the prevalence of pathogens on the quarter level for Strep. agalactiae and Strep. dysgalactiae of 0.65 and 0.36, respectively, was also found in another study (
      • Spohr M.
      • Breitenwieser F.
      Vergleichende Untersuchungen zur Mastitisdiagnostik mittels Real-Time-PCR und klassischer Bakteriologie.
      ). Contrary to our results, no significant correlation could be proven for Staph. aureus in their study. Our results lead to the hypothesis that it may be possible to detect changes in the WHPapp of a specific pathogen by alterations of the corresponding Ct values in BTM. This hypothesis is subject to further research.
      Testing a single BTM sample, we calculated cow level WHPapp thresholds of approximately 9.2% for Strep. dysgalactiae, 13.8% for Strep. uberis, and 27.6% for Staph. aureus at Pd = 0.9. On the quarter level, we found WHPapp thresholds of 1.7% for Strep. dysgalactiae, 4.3% for Strep. uberis, and 32.6% for Staph. aureus. These results clearly indicate that a relevant percentage of the cows must be BC positive to get a positive test result in BTM testing. These thresholds were calculated for specific mastitis pathogens in our study and were given for 2 different probabilities of detection. We found remarkable differences in the WHPapp threshold between Staph. aureus and the streptococci. The higher thresholds for Staph. aureus might be caused by either a lower sensitivity of the Mastit 4 qPCR kit for the target Staph. aureus compared with streptococci or a lower number of bacteria per BC positive cow or quarter secreted with the milk from 1 infected quarter. Relating to 1 infected cow, the latter might cause a lower amount of DNA copies in the BTM specimen. Further investigation is necessary to determine why the threshold was higher for Staph. aureus than for Strep. uberis and Strep. dysgalactiae.
      For all pathogens, the WHPapp thresholds, both on cow and quarter levels, were lower if 2 instead of 1 sample was examined. This increase in sensitivity of BTM testing by taking more than 1 sample is consistent with the results of other studies (
      • Godkin M.A.
      • Leslie K.E.
      Culture of bulk tank milk as a mastitis screening test: A brief review.
      ;
      • Olde Riekerink R.G.
      • Barkema H.W.
      • Scholl D.T.
      • Poole D.E.
      • Kelton D.F.
      Management practices associated with the bulk-milk prevalence of Staphylococcus aureus in Canadian dairy farms.
      ;
      • da Costa L.B.
      • Rajala-Schultz P.J.
      • Schuenemann G.M.
      Management practices associated with presence of Staphylococcus aureus in bulk tank milk from Ohio dairy herds.
      ). Examining 2 instead of just 1 sample lowers the WHPapp threshold for the pooled herd samples, too.
      In our study, both BTM samples were taken from the content of 1 bulk tank. If the pathogens were homogeneously distributed, the qPCR results of the samples should be similar and the detection limit should not be influenced by double sampling. But in only 62.4% of the herds classified as positive using bulk tank samples, the same pathogen was detected simultaneously in both bulk tank samples. This may be caused by the imperfect distribution of the pathogens in the BTM and the resulting inconsistency caused by the specimen. Therefore, the accuracy of the BTM analysis appears to be limited, independent of the applied method of examination. The “Subclinical Mastitis” expert committee of the German Veterinary Society recommended that a screening of BTM is only useful professionally for the obligatory udder-associated Strep. agalactiae, as all of the other pathogens causing IMI could also be found in the environment of the animals (
      • DVG (Deutsche Veterinärmedizinische Gesellschaft)
      Stellungnahme des Sachverständigenausschusses zur Verwendung der Polymerase-Kettenreaktion (PCR) in der Mastitisdiagnostik.
      ). However, Strep. agalactiae is not considered an obligatory intramammary pathogen anymore, as this pathogen was detected in the environment and in the gastrointestinal tract of dairy cattle (
      • Jørgensen H.J.
      • Nordstoga A.B.
      • Sviland S.
      • Zadoks R.N.
      • Sølverød L.
      • Kvitle B.
      • Mørk T.
      Streptococcus agalactiae in the environment of bovine dairy herds—Rewriting the textbooks?.
      ). After a positive BTM result, the infected cows and udder quarters should be identified by individual BC testing of quarter foremilk samples for further decisions regarding mastitis management on the herd level (
      • DVG (Deutsche Veterinärmedizinische Gesellschaft)
      Stellungnahme des Sachverständigenausschusses zur Verwendung der Polymerase-Kettenreaktion (PCR) in der Mastitisdiagnostik.
      ). Therefore, the classical tools for diagnosing subclinical and clinical mastitis remain valuable (e.g., for monitoring the presence of certain mastitis pathogens in a herd).
      An alternative method for taking bulk tank samples was used in the Danish screening program for Strep. agalactiae. The samples were taken during weighing-in the milk from the bulk tank to the truck tank. However, this approach may be prone to cross-contamination between herds if the sampling systems are not disinfected after use (
      • Andersen H.J.
      • Pedersen L.H.
      • Aarestrup F.M.
      • Chriel M.
      Evaluation of the surveillance program of Streptococcus agalactiae in Danish dairy herds.
      ).
      Most studies that evaluated test characteristics of mastitis PCR systems used BC of quarter milk samples as the reference method, although some authors stated that its sensitivity and specificity is limited (
      • Krömker V.
      • Friedrich J.
      • Klocke D.
      Ausscheidung und Nachweis von Staphylococcus aureus über Milch aus infizierten Milchdrüsenvierteln.
      ;
      • Paradis M.E.
      • Haine D.
      • Gillespie B.
      • Oliver S.P.
      • Messier S.
      • Comeau J.
      • Schall D.T.
      Bayesian estimation of the diagnostic accuracy of a multiplex real-time PCR assay and bacteriological culture for 4 common bovine intramammary pathogens.
      ). Studies in which quarter milk samples or composite samples were analyzed in parallel with both test methods detected more pathogens by qPCR than by BC (
      • Taponen S.
      • Salmikivi L.
      • Simojoki H.
      • Koskinen M.T.
      • Pyörälä S.
      Real-time polymerase chain reaction-based identification of bacteria in milk samples from bovine clinical mastitis with no growth in conventional culturing.
      ;
      • Koskinen M.T.
      • Wellenberg G.J.
      • Sampimon O.C.
      • Holopainen J.
      • Rothkamp A.
      • Salmikivi L.
      • van Haeringen W.A.
      • Lam T.J.G.M.
      • Pyörälä S.
      Field comparison of real-time polymerase chain reaction and bacterial culture for identification of bovine mastitis bacteria.
      ;
      • Spittel S.
      • Hoedemaker M.
      Mastitis diagnosis in dairy cows using PathoProofTM real-time polymerase chain reaction assay in comparison with conventional bacterial culture in a Northern German field study.
      ).
      • Spittel S.
      • Hoedemaker M.
      Mastitis diagnosis in dairy cows using PathoProofTM real-time polymerase chain reaction assay in comparison with conventional bacterial culture in a Northern German field study.
      compared the results of qPCR examination of quarter milk samples with those of pooled cow level samples. They found reduced sensitivities of 81.3% for Staph. aureus and 77.1% for Strep. uberis for pooled cow level samples, indicating that the sensitivity of the qPCR is reduced if the pathogens in the specimen were diluted by examining cow level milk samples instead of quarter milk samples.
      Beside other factors, the sensitivity of qPCR depends on the choice of the cut-off (
      • Cederlöf S.E.
      • Toft N.
      • Aalbaek B.
      • Klaas I.C.
      Latent class analysis of the diagnostic characteristics of PCR and conventional bacteriological culture in diagnosing intramammary infections caused by Staphylococcus aureus in dairy cows at dry off.
      ). In our study the manufacturer's instructions () were used, which recommended a cut-off value of 37 cycles, whereas other studies chose higher cut-off values for the examination of BTM samples (
      • Katholm J.
      • Bennedsgaard T.W.
      • Koskinen M.T.
      • Rattenborg E.
      Quality of bulk tank milk samples from Danish dairy herds based on real-time polymerase chain reaction identification of mastitis pathogens.
      ;
      • Mahmmod Y.S.
      • Toft N.
      • Katholm J.
      • Grønbæk C.
      • Klaas I.C.
      Estimation of test characteristics of real-time PCR and bacterial culture for diagnosis of subclinical intramammary infections with Streptococcus agalactiae in Danish dairy cattle in 2012 using latent class analysis.
      ). Using a similar qPCR assay,

      Bennedsgaard, T. W., L. Svennesen, and I. C. Klaas. 2016. Test characteristics of the qPCR test Mastit 4 to identify major pathogens in spiked and originally infected milk samples. Pages 180–181 in Natl. Mastitis Counc. 55th Annual Mtg. Proc., Glendale, AZ.

      stated that the Ct value not only depended on the number of colony-forming units determined by BC but also on the pathogen itself.
      • Rattenborg E.
      • Paulrud C.O.
      • Jensen S.K.S.
      • Katholm J.
      Bulk tank milk surveillance for contagious mastitis pathogens: Comparison of two commercial real-time PCR test kits.
      showed the effect of the applied qPCR assay itself by examining BTM samples with 2 different qPCR assays (PathoProof Mastitis Complete 16 Kit and Mastit 4B). Depending on the target pathogen, they found only low to moderate agreement between the qualitative test results (presence/absence) of the 2 qPCR assays. An alternative approach to consider for qPCR testing is the use of milk samples collected in the framework of a herd improvement test during the milking process representing a proportion of the whole milk output of the cow.
      Further research is necessary to evaluate qPCR testing of different milk specimens for mastitis pathogens regarding pool size, sensitivity, and specificity, as well as contamination. We were able to fill a gap of knowledge regarding the benefit of using qPCR instead of BC in testing bulk milk samples for major mastitis pathogens.

      CONCLUSIONS

      Applying qPCR on BTM samples allows the detection of herds positive for the examined pathogens when WHPapp exceeds a threshold that was calculated in this study. Regarding herd level sensitivity, qPCR outperforms BC of 0.01 mL of BTM. Double testing further increases sensitivity; however, sensitivity is still limited. Testing of BTM using qPCR could be a reliable method for detecting an occurrence of a new pathogen in the herd, but not to confirm that a distinct herd is free from a specific pathogen. Usefulness of qPCR testing applied on several specimens for monitoring changes in prevalence of mastitis pathogens needs to be investigated. Pooled herd samples are an alternative to BTM; for specific pathogens, the use of pooled milk samples may be advantageous compared with BTM samples.

      ACKNOWLEDGMENTS

      The present study is a project of the Thuringian Animal Diseases Fund. It was partly funded by the Thuringian Dairy Industry Association (Landesvereinigung Thüringer Milch e. V., Germany). The authors thank both institutions for their support. Furthermore, the authors thank the laboratory staff and the veterinarians of the Animal Health Service of Thuringia and the involved dairy farmers. The support of DNA Diagnostics was greatly appreciated.

      REFERENCES

        • Andersen H.J.
        • Pedersen L.H.
        • Aarestrup F.M.
        • Chriel M.
        Evaluation of the surveillance program of Streptococcus agalactiae in Danish dairy herds.
        J. Dairy Sci. 2003; 86 (12741548): 1233-1239
        • Bartlett P.C.
        • Miller G.Y.
        • Lance S.E.
        • Heider L.E.
        Use of bulk tank and milk filter cultures in screening for Streptococcus agalactiae and coagulase positive staphylococci.
        J. Food Prot. 1991; 54: 848-854
      1. Bennedsgaard, T. W., L. Svennesen, and I. C. Klaas. 2016. Test characteristics of the qPCR test Mastit 4 to identify major pathogens in spiked and originally infected milk samples. Pages 180–181 in Natl. Mastitis Counc. 55th Annual Mtg. Proc., Glendale, AZ.

        • Bexiga R.
        • Koskinen M.T.
        • Holopainen J.
        • Carneiro C.
        • Pereira H.
        • Ellis K.A.
        • Vilela C.L.
        Diagnosis of intramammary infection in samples yielding negative results or minor pathogens in conventional bacterial culturing.
        J. Dairy Res. 2011; 78 (21134309): 49-55
        • Bradley A.J.
        • Leach K.A.
        • Breen J.E.
        • Green L.E.
        • Green M.J.
        Survey of the incidence and aetiology of mastitis on dairy farms in England and Wales.
        Vet. Rec. 2007; 160 (17322356): 253-257
        • Cederlöf S.E.
        • Toft N.
        • Aalbaek B.
        • Klaas I.C.
        Latent class analysis of the diagnostic characteristics of PCR and conventional bacteriological culture in diagnosing intramammary infections caused by Staphylococcus aureus in dairy cows at dry off.
        Acta Vet. Scand. 2012; 54 (23164432): 65
        • da Costa L.B.
        • Rajala-Schultz P.J.
        • Schuenemann G.M.
        Management practices associated with presence of Staphylococcus aureus in bulk tank milk from Ohio dairy herds.
        J. Dairy Sci. 2016; 99 (26686713): 1364-1373
        • DANMAP (Danish Integrated Antimicrobial Resistance Monitoring and Research Programme)
        Use of antimicrobial agents and occurrence of antimicrobial resistence in bacteria from food animals, food and humans in Denmark. Statens Serum Institut, National Veterinary Institute, and National Food Institute, Technical University of Denmark, 2014: 34
      2. Dixon W.J. BMDP Statistical Software Manual. Volume 1 and 2. University of California Press, Berkeley1993
        • DNA Diagnostic
        Instruction-protocol-m4a.pdf.
        (Accessed Jul. 22, 2017.)
      3. DVG (Deutsche Veterinärmedizinische Gesellschaft). 2009. Leitlinien zur Entnahme von Milchproben unter antiseptischen Bedingungen und Isolierung und Identifizierung von Mastitiserregern. DVG. 2. Aufl, Gießen, Germany.

        • DVG (Deutsche Veterinärmedizinische Gesellschaft)
        Stellungnahme des Sachverständigenausschusses zur Verwendung der Polymerase-Kettenreaktion (PCR) in der Mastitisdiagnostik.
        Deutsches Tierärzteblatt. 2010; 2010: 914-917
      4. DVG (Deutsche Veterinärmedizinische Gesellschaft). 2012. Leitlinien Bekämpfung der Mastitis des Rindes als Bestandsproblem. Page 68. DVG. 5. Aufl, Gießen, Germany.

        • Gillespie B.E.
        • Oliver S.P.
        Simultaneous detection of mastitis pathogens, Staphylococcus aureus, Streptococcus uberis, and Streptococcus agalactiae by multiplex real-time polymerase chain reaction.
        J. Dairy Sci. 2005; 88 (16162525): 3510-3518
        • Godkin M.A.
        • Leslie K.E.
        The relationships between bulk tank milk culture, management factors used in mastitis control and the herd prevalence of mastitis.
        in: Proc. Int. Symp. Mastitis, Indianapolis, IN. Natl. Mast. Counc., Madison, WI1990: 363-374
        • Godkin M.A.
        • Leslie K.E.
        Culture of bulk tank milk as a mastitis screening test: A brief review.
        Can. Vet. J. 1993; 34 (17424304): 601-605
        • Gundling N.
        • Ruddat I.
        • Prien K.
        • Hellerich B.
        • Hoedemaker M.
        Erkrankungshäufigkeit von Milchviehherden in Schleswig-Holstein. Einfluß der Milchleistung der ersten Milchleistungsprüfung der Laktation, der Herdenmilchleistung und der Laktationsnummer.
        Berl. Münch. Tieräztl. Wochenschr. 2015; 128: 225-232
        • Jørgensen H.J.
        • Nordstoga A.B.
        • Sviland S.
        • Zadoks R.N.
        • Sølverød L.
        • Kvitle B.
        • Mørk T.
        Streptococcus agalactiae in the environment of bovine dairy herds—Rewriting the textbooks?.
        Vet. Microbiol. 2016; 184 (26854346): 64-72
        • Katholm J.
        • Bennedsgaard T.W.
        • Koskinen M.T.
        • Rattenborg E.
        Quality of bulk tank milk samples from Danish dairy herds based on real-time polymerase chain reaction identification of mastitis pathogens.
        J. Dairy Sci. 2012; 95 (22921631): 5702-5708
        • Koskinen M.T.
        • Wellenberg G.J.
        • Sampimon O.C.
        • Holopainen J.
        • Rothkamp A.
        • Salmikivi L.
        • van Haeringen W.A.
        • Lam T.J.G.M.
        • Pyörälä S.
        Field comparison of real-time polymerase chain reaction and bacterial culture for identification of bovine mastitis bacteria.
        J. Dairy Sci. 2010; 93 (21094742): 5707-5715
        • Krömker V.
        • Friedrich J.
        Empfehlungen zum diagnostischen Aufwand im Rahmen der Mastitisbekämpfung auf Bestandsebene.
        Prakt. Tierarzt. 2011; 92: 516-524
        • Krömker V.
        • Friedrich J.
        • Klocke D.
        Ausscheidung und Nachweis von Staphylococcus aureus über Milch aus infizierten Milchdrüsenvierteln.
        Tierärztl. Praxis (G). 2008; 6: 123-126
        • Mahmmod Y.S.
        • Toft N.
        • Katholm J.
        • Grønbæk C.
        • Klaas I.C.
        Estimation of test characteristics of real-time PCR and bacterial culture for diagnosis of subclinical intramammary infections with Streptococcus agalactiae in Danish dairy cattle in 2012 using latent class analysis.
        Prev. Vet. Med. 2013; 109 (23194895): 264-270
        • Olde Riekerink R.G.
        • Barkema H.W.
        • Scholl D.T.
        • Poole D.E.
        • Kelton D.F.
        Management practices associated with the bulk-milk prevalence of Staphylococcus aureus in Canadian dairy farms.
        Prev. Vet. Med. 2010; 97 (20696486): 20-28
        • Paradis M.E.
        • Haine D.
        • Gillespie B.
        • Oliver S.P.
        • Messier S.
        • Comeau J.
        • Schall D.T.
        Bayesian estimation of the diagnostic accuracy of a multiplex real-time PCR assay and bacteriological culture for 4 common bovine intramammary pathogens.
        J. Dairy Sci. 2012; 95 (22981579): 6436-6448
        • Rattenborg E.
        • Paulrud C.O.
        • Jensen S.K.S.
        • Katholm J.
        Bulk tank milk surveillance for contagious mastitis pathogens: Comparison of two commercial real-time PCR test kits.
        in: Natl. Mastitis Counc. 54th Annual Mtg. Proc., Syracuse, NY. Natl. Mast. Counc., Madison, WI2015: 215-216
        • Reinemann D.J.
        • Mein G.A.
        • Bray D.R.
        • Reid D.
        • Britt J.S.
        Troubleshooting high bacteria counts in farm milk.
        in: Proc. Natl. Mastitis Counc. 36th Annual Mtg. Natl. Mast. Counc., Madison, WI1997: 65-79
        • Riffon R.
        • Sayasith K.
        • Khalil H.
        • Dubreuil P.
        • Drolet M.
        • Lagace J.
        Development of a rapid and sensitive test for identification of major pathogens in bovine mastitis by PCR.
        J. Clin. Microbiol. 2001; 39 (11427573): 2584-2589
        • Roberson J.R.
        • Fox L.K.
        • Hancock D.D.
        • Gay J.M.
        • Besser T.E.
        Ecology of Staphylococcus aureus isolated from various sites on dairy farms.
        J. Dairy Sci. 1994; 77 (7814712): 3354-3364
        • Rodolakis A.
        • Berri M.
        • He'chard C.
        • Caudron C.
        • Souriau A.
        • Bodier C.C.
        • Blanchard B.
        • Camuset P.
        • Devillechaise P.
        • Natorp J.C.
        • Vadet J.P.
        • Arricau-Bouvery N.
        Comparison of Coxiella burnetii shedding in milk of dairy bovine, caprine, and ovine herds.
        J. Dairy Sci. 2007; 90 (18024725): 5352-5360
        • Ruegg P.L.
        • Reinemann D.J.
        Milk quality and mastitis tests.
        Bovine Pract. 2002; 36: 41-54
        • Spittel S.
        • Hoedemaker M.
        Mastitis diagnosis in dairy cows using PathoProofTM real-time polymerase chain reaction assay in comparison with conventional bacterial culture in a Northern German field study.
        Berl. Munch. Tierarztl. Wochenschr. 2012; 125 (23227767): 494-502
        • Spohr M.
        • Breitenwieser F.
        Vergleichende Untersuchungen zur Mastitisdiagnostik mittels Real-Time-PCR und klassischer Bakteriologie.
        Prakt. Tierarzt. 2012; 93: 342-350
        • Taponen S.
        • Salmikivi L.
        • Simojoki H.
        • Koskinen M.T.
        • Pyörälä S.
        Real-time polymerase chain reaction-based identification of bacteria in milk samples from bovine clinical mastitis with no growth in conventional culturing.
        J. Dairy Sci. 2009; 92 (19447993): 2610-2617
        • Wilson D.J.
        • Gonzalez R.N.
        Evaluation of milk culture, SCC and CMT for screening herd additions.
        in: Natl. Mastitis Counc., 36th Annual Mtg. Proc., Albuquerque, NM. Natl. Mast. Counc., Madison, WI1997: 127-131
        • Wolter W.
        • Kloppert B.
        • Castaneda H.
        • Zschöck M.
        Die Mastitis des Rindes. Ein Kursbuch. Staatliches Untersuchungsamt, Hessen, Germany2002
        • Zanardi G.
        • Caminiti A.
        • Delle Donne G.
        • Moroni P.
        • Santi A.
        • Galletti G.
        • Tamba M.
        • Bolzoni G.
        • Bertocchi L.
        Short communication: Comparing real-time PCR and bacteriological cultures for Streptococcus agalactiae and Staphylococcus aureus in bulk-tank milk samples.
        J. Dairy Sci. 2014; 97 (24997661): 5592-5598