Accuracy of the composite somatic cell count to detect intra-mammary infection in dairy cows using latent class analysis

Abstract

The somatic cell count (SCC) is considered an important indicator of intra-mammary infection (IMI). The purpose of this study was to determine the accuracy of both SCC and culture to detect IMI and their conditional dependence by means of latent class methods. This study involved 175 dairy cows from 2 herds with different udder infection prevalences. Quarter and composite milk samples were collected for SCC and bacteriological culture. Latent-class models using Bayesian methods were used to estimate test sensitivity (Se) and specificity (Sp) and population prevalence. The models ran involved only major mastitis pathogens and composite SCC (CSCC). Five thresholds between 100,000 and 300,000cells/mL were evaluated and the receiver operating characteristics (ROC) curve analysis was performed. Fifty-five percent of the cows had CSCC ≥200,000cells/mL and 95.4% of the cows had at least one infected quarter either with minor or major pathogens. Considering a threshold of 150,000cells/mL, the estimated Se and Sp for the CSCC were, 0.80 (95% CrI 0.71–0.88) and 0.57 (95% CrI 0.44–0.71), respectively. The estimated culture Se and Sp were 0.83 (95% CrI 0.73–0.93) and 0.89 (95% CrI 0.74–0.98), respectively. There was no evidence of dependence between CSCC and culture. The area under curve for CSCC was 0.72. To the best of our knowledge, this is the first report of the CSCC accuracy to detect IMI for major pathogens considering the effect of culture misclassification. The estimates provided here could help to examine the performance of sampling schemes based on CSCC to manage udder health.