Utility of an in-line somatic cell count sensor for selecting cows for dry-cow therapy

Minimizing the use of antimicrobials at the end of lactation (dry-cow therapy) requires categorizing cows as being likely infected or uninfected. Although microbiology is the gold standard for such categorization, the cost of such tests means that indirect tests, such as SCC, are commonly used. An in-line SCC sensor (SenseHub In-Line Somatic Cell Count; MSD Animal Health) is commercially available, but its utility in differentiating cows eligible for dry-cow therapy has not been assessed. Our prospective diagnostic accuracy study was undertaken to define the sensitivity (Se) and specificity (Sp) of SenseHub SCC on cow-composite milk samples submitted for conventional microbiology. A secondary objective was to assess the utility of SenseHub SCC compared with the maximum or last DHI SCC determined from cow-composite milk samples collected as part of routine herd production recording at monthly intervals throughout lactation. Cows (n = 1,544) from 4 spring-calving, predominantly pasture-fed dairy herds from 3 regions of New Zealand had cow-composite milk samples collected following aseptic teat end preparation immediately before or after the final milking of lactation. These samples were submitted for routine microbiology. The microbiology data from approximately half the cows (n = 770; training dataset) were randomly selected after blocking for IMI status within herd, and these data were used to determine the optimal predictor for indicating IMI from the in-line SCC data. This determination was made by maximizing the area under the curve (AUC) for the receiver operator curve. The average of the in-line SCC over the final 12 wk of lactation (in-line 12wSCC) was found to be the best predictor and was used for further analyses. The Se and Sp of the in-line SCC for any IMI or for a major pathogen IMI (defined as the presence of Staphylococcus aureus, Streptococcus dysgalactiae, or Streptococcus uberis) was calculated using the test dataset (n = 774). The AUC for the maximum and last DHI SCC were compared with that of the in-line 12wSCC. The cow-level prevalence of any IMI or a major IMI across the entire population was 50.6% and 14.2%, respectively. At a cut-point of 150,000 cells/mL, Se and Sp of the in-line 12wSCC for any IMI was 0.68 (95% CI 0.64-0.72) and 0.71 (95% CI 0.65-0.76), respectively, and the Se and Sp for a major pathogen IMI was 0.89 (95% CI 0.82-0.95) and 0.51 (95% CI 0.47-0.55), respectively. The AUC for a major pathogen IMI was 0.82 (95% CI 0.79-0.86), 0.82 (95% CI 0.78-0.86), and 0.84 (95% CI 0.90-0.97) for in-line 12wSCC, maximum DHI SCC, and last DHI SCC, respectively. These AUC did not differ, and the AUC for the in-line 12wSCC was noninferior to that of the last and maximum DHI SCC. Given that the in-line 12wSCC had an AUC, Se, and Sp not different from DHI SCC data, we concluded that this test has utility in selecting cows for different dry-cow therapy treatments.