Prevalence of endometritis diagnosed by vaginal discharge scoring or uterine cytology in dairy cows and herds.

Endometritis, diagnosed either by assessing the proportion of nucleated cells that are neutrophils (PMN%) following cytology of the endometrium or by assessing the degree of purulent material within the vagina (purulent vaginal discharge or PVD score), is prevalent among dairy cows. However, limited data exist as to the degree of variation among herds in the prevalence of endometritis diagnosed by these 2 methods. Thus, we undertook a study involving uterine cytological and vaginal sampling at a median of 41 d in milk of 1,807 cows from 100 seasonally breeding dairy herds in New Zealand. The optimal cut-point for PMN% was determined by receiver operator characteristic (ROC) curve analysis using conception to first artificial insemination (AI) as the outcome variable. The prevalence of disease was then calculated at the cow and herd levels, and an estimate of the effect of clustering of cow within a herd was calculated. Reproductive outcomes were collated and associations between endometritis and reproductive outcomes assessed using multivariable models. The optimal PMN% cut-point was ≥2%. The correlation of results for cows within a herd (the intraclass correlation) was 0.03, which was significant; hence, subsequent modeling accounted for this clustering. The cow-level prevalence of PMN% ≥2% was 27.0% [95% confidence interval (CI): 25.0 to 29.1%], whereas the mean within-herd prevalence of PMN% ≥2% was 27.1% (95% CI: 24.7 to 29.6%; range: 5.0 to 63.6%), and the prevalence among herds varied significantly. An elevated PMN% (≥2%) was significantly associated with a reduction in the proportion of cows conceiving to first AI (45.8 vs. 54.5%), a reduced proportion of cows submitted for AI in the first 3 wk of the seasonal breeding program (83.7 vs. 89.3%), and a lower proportion pregnant in the first 3 (44.4 vs. 55.4%) and 6 wk (67.5 vs. 76.4%) of the breeding program relative to cows with a low PMN% (i.e., <2%). A total of 24.6% of cows had a PVD score ≥2. The herd mean prevalence of PVD score ≥2 was 25.1% (95% CI: 22.5 to 27.7%; range: 5.0 to 65.0%) and varied significantly among herds. The level of agreement (kappa) between the PVD score and PMN% was low (16.8%) and nonsignificant. The effects of PVD score and PMN% on reproductive outcomes were independent. The within-herd median prevalence of endometritis based on combining both diagnostic tests and using a Bayesian latent class model was 22.9% (Bayesian 95% CI: 10.4 to 40.1%). We conclude that more than one-fifth of dairy cows have endometritis diagnosed either by PMN% or PVD in seasonal breeding herds when assessed at an average of 41 DIM, which was, on average, 30 d before the start of the seasonal breeding program. There is large and unexplained variation in prevalence of endometritis among herds. The 2 diagnostic methods were both associated with reproductive outcomes but have low levels of agreement between them and their effects appear to be independent.