Tel.+49 30 838 62618 Fax.+49 30 838 62620
Objectives: The routine diagnostic of mastitis pathogens and antimicrobial resistances mostly includes the bacteriological examination. This gold standard, however, has several limitations. Sterile milk samples are necessary to achieve reliable results and transport between farm and laboratory, often conducted without cooling, might bias the results. An inter-laboratory comparison of Finnish milk laboratories has been published in 2005, however, only spiked milk samples were used in this study. Therefore, the objective of our study was to compare bacteriological results of milk samples collected in the field and shipped to 3 different laboratories.
Materials and Methods: A total of 792 milk samples (175 clinical and 250 subclinical mastitis, 346 healthy) were collected from July 2013 to June 2014 on a commercial dairy farm in Germany. Sterile milk samples were collected according to the guidelines from the International Dairy Federation. Samples were immediately cooled and transported to the laboratory of the Clinic for Animal Reproduction. Upon arrival all samples were vortexed for 15 sec, divided into 3 aliquots and filled into tubes coated with boric acid under sterile conditions. Each tube was send to 1 of 3 different accredited milk laboratories (Lab A, B, C) that offer routine milk diagnostic for commercial dairy farms and veterinarians. Samples were send by regular mail without cooling in order to provide influences similar to those samples are exposed to under normal circumstances. All samples arrived within 2d. Samples in all labs were inoculated on agar plates and incubated. Isolated bacteria were identified based on colony and cell morphology, gram staining, hemolytic pattern, catalase activity and growth on differential and selective media. All samples were examined for the presence of bacteria, yeast and prototheca species and milk somatic cell count was evaluated. Furthermore, antimicrobial susceptibilities for 20 antimicrobials used in mastitis therapy were tested in 102 out of 792 samples. Association and agreement between labs was estimated utilizing Chi-Square test or Fisher’s Exact test, Cohens Kappa test, Spearman’s (i.e., antimicrobial susceptibility) and Pearson’s (i.e., SCC) Correlation Coefficient and paired t-test.
Results: In 42.3% of all samples the same bacteriological result (i.e., same pathogen or “no growth”) was found by all three labs. The same pathogen genus (e.g., Streptococci) was diagnosed in 45.1% of the samples. Comparing only 2 out of 3 labs, Lab A and B identified the same pathogen in 57.4% of all samples (Kappa = 0.335; P < 0.001), the differentiation between gram positive and negative was identical in 95.4% of the results. Lab B and C had an agreement of 50.0% (Kappa = 0.308; P < 0.001) and 88.7% considering pathogen identification and gram stain, respectively. Best results were achieved between Lab A and C with 72.2% agreement between pathogen identification (Kappa = 0.561; P < 0.001) and 100% for gram stain. 17.8% of the results were
gram negative. Agreement considering the decision “pathogen” or “no growth” was 62.6%, 63.6% and 74.4% between Lab A and B, B and C and A and C, respectively. Identical results by all labs were 50% no growth, 35.1% Staph. aureus, 6.8% Strep. uberis, 4.1% CNS, 2.7% Coliforms, 1.4% Strep. dysgalactiae. Identical results considering antimicrobial susceptibility (i.e., susceptible, intermediate, resistant) were provided by all labs for 57.1% of the tested antimicrobials. Lab A and B provided the same results for 67.6% (Kappa = 0.322; R = 0.432; P < 0.001), Lab B and C for 61.1% (Kappa = 0.257; R = 0.341; P < 0.05) and Lab A and C for 71.0% (Kappa = 0.308; R = 0.391; P < 0.001) of the antimicrobials, respectively. While somatic cell count results differed between labs as well (P < 0.02), the correlation was good (i.e., R = 0.854 - 0.959; P < 0.001).
Conclusions: Our study showed a fair to moderate agreement between laboratories considering pathogen identification, antimicrobial susceptibility and SCC. While our results slightly differ from previous comparisons utilizing spiked milk samples, they implicate that the bacteriological examination of milk samples might have limitations, e.g. if samples are send by regular mail or without cooling. Considering current discussions on bacterial resistances and antimicrobial residues, further research is warranted to identify confounders that affect bacteriological results and to improve the reliability of results acquired by bacteriological examination.