Search

Many important factors impact reproductive outcome in Irish seasonally calving pasture-based dairy cows. Two of the factors that were investigated in this thesis were; 1) Reproductive tract disease (RTD) and 2) Estrus detection inaccuracy (EDI). The study on RTD involved the retrospective analysis of 5,049 pre-breeding examination (PBE) records from 2,460 spring-calved cows on 8 farms over 5 years. PBEs analysed were conducted once between 25-86 days in milk (DIM) with two practical cow-side diagnostics which were;1) visual grading of vaginal discharge obtained with a Metricheck® device [purulent vaginal discharge (PVD) scoring on a scale of 0 to 3] and visual grading of ultrasound images of the uterus to assess evidence of endometritis [ultrasonographic endometritis (UE) scoring on a scale of 0 to 4]. The main objectives were; 1) to determine cow and calving related factors that were associated with increased risk of RTD, 2) to analyse the association between the test results of both diagnostics and subsequent reproductive outcome and 3) to the determine the diagnostic test parameters and select optimal thresholds for RTD at the time of the PBE. Multivariable logistic regression was used to identify factors that were associated with an increased risk of RTD when diagnosed by either PVD or UE scoring alone, or in combination. Sharded risk factors for RTD identified between both diagnostics were DIM at the at PBE, retained fetal membranes, twins, and cow genetics. Other risk factors that were specific for each diagnostic were dystocia, parity, BCS, breed, and ovarian findings. The association between PVD or UE score at the PBE and subsequent reproductive outcome was analyzed using Cox proportional hazards model. Cows with PVD score =2 (i.e. =mucopurulent discharge) and a UE score of score =1 (i.e. = trace intrauterine fluid) were less likely to conceive than cows with healthy scores. To determine the test parameters and optimal thresholds for PVD and UE scoring for diagnosis of RTD at the time of PBE, a Bayesian latent class model was fitted. The optimal test thresholds selected were score =1 for both PVD (i.e. flecks of pus) and UE. At these thresholds, median sensitivity and specificity tend to be higher for UE than PVD but this was dependent on DIM at PBE. Two observational studies were used to investigate EDI. In both studies, milk samples were collected on the same day as artificial insemination (AI) and analysed for milk progesterone (MP4) concentration using a radioimmunoassay. Cows that had a milk sample above a critical threshold of MP4 were determined to have been inseminated inaccurately, i.e., they were falsely diagnosed in estrus and AI’d. The main objectives of these studies were; 1) to estimate the cow-level true and apparent prevalence of EDI and 2) to determine potential cow-level risk factors for EDI. The first study involved the analysis of 576 milk samples from dairy cows in 125 herds whom received an AI on the same day as their routine milk recording. Using a Bayesian latent class model, the median estimate for cow-level true prevalence of EDI was 4.4% (1.7–9.0 %). In the second study, 1071 milk samples were obtained from 984 cows on 19 farms at the time of AI. Using multivariable logistic regression models that were stratified by insemination number (i.e., first versus repeat) cow-level risk factors for EDI were determined. Repeat insemination had a higher apparent prevalence of EDI than first inseminations (14.1% vs 3.3%) and cows with no signs of mounting abrasions, had no observed standing estrus event or had an abnormal preceding interval since their last insemination were at increased risk of EDI at repeat insemination. The results of this thesis will assist farmers and advisors in developing strategies that could help mitigate the negative impacts that both RTD and EDI have on reproductive outcome at cow-level but further research is required to validate these approaches. 2022-06-21 JG: Signature removed from PDF