Search

We trained machine learning models to identify IMI in late-lactation cows at dry-off to guide antibiotic treatment, and compared their performance to a rule-based algorithm that is currently used on dairy farms in the United States. We conducted an observational test characteristics study using a dataset of 3,645 cows approaching dry-off from 68 US dairy herds. The outcome variables of interest were cow-level IMI caused by all pathogens, major pathogens, and Streptococcus and Streptococcus-like organisms (SSLO), which were determined using aerobic culture of aseptic quarter-milk samples and identification of isolates using MALDI-TOF. Individual cow records were extracted from the farm software to create 53 feature variables at the cow and 39 at the herd level, which were derived from cow-level descriptive data, records of clinical mastitis events, results from routine testing of milk for volume, and concentrations of SCC, fat, and protein. Machine learning (ML) algorithms evaluated were logistic regression, decision tree, random forest, light gradient-boosting machine, naive Bayes, and neural networks. For comparison, cows were also classified according to a conventional rule-based algorithm that considered a cow as high risk for IMI if she had one or more high SCC (>200,000 cells/mL) tests or ≥2 cases of clinical mastitis during the lactation of enrollment. Area under the curve (AUC) and Youden's index were used to compare models, in addition to binary classification metrics, including sensitivity, specificity, and predictive values. The ML models had slightly higher AUC and Youden's index values than the rule-based algorithm for all IMI outcomes of interest. However, these improvements in prediction accuracy were substantially less than what we had considered necessary for the technology to be a worthwhile alternative to the rule-based algorithm. Therefore, evidence is lacking to support the wholesale use of ML-guided selective dry cow therapy at the moment. We recommend that producers wanting to implement algorithm-guided selective dry cow therapy use a rule-based method., (The Authors. Published by Elsevier Inc. on behalf of the American Dairy Science Association®. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).)

The objectives for this study were to (1) describe the pathogen profile in quarters from cows with clinical mastitis and in cows with subclinical mastitis in southeastern Australia; and (2) describe antimicrobial susceptibility among isolated pathogens. As a secondary objective, we aimed to compare antimicrobial resistance prevalence in pathogens isolated from clinical and subclinical mastitis samples. A convenience sample of dairy herds (n = 65) from 4 regions in southeastern Australia (Gippsland, Northern Victoria, Tasmania, Western Victoria) were invited to submit milk samples from cows with clinical and subclinical mastitis over a 14-mo period (January 2011 to March 2012). Farmers were instructed to collect aseptic quarter milk samples from the first 10 cases of clinical mastitis for each month of the study. In addition, farmers submitted composite milk samples from cows with subclinical mastitis at 1 or 2 sampling occasions during the study period. Aerobic culture and biochemical tests were used to identify isolates. Isolates were classified as susceptible, intermediate, or resistant to a panel of antimicrobial agents based on the zone of growth inhibition around antimicrobial-impregnated disks, with antimicrobial resistance (AMR) classified as nonsusceptibility by combining intermediate and resistant groups into a single category. Generalized linear mixed models were used to compare the prevalence of AMR between clinical and subclinical mastitis isolates. For clinical mastitis samples (n = 3,044), 472 samples (15.5%) were excluded for contamination. Of the remaining samples (n = 2,572), the most common results were Streptococcus uberis (39.2%), no growth (27.5%), Staphylococcus aureus (10.6%), Escherichia coli (8.4%), and Streptococcus dysgalactiae (6.4%). For subclinical mastitis samples (n = 1,072), 425 (39.6%) were excluded due to contamination. Of the remaining samples (n = 647), the most common results were no growth (29.1%), Staph. aureus (29.1%), and Strep. uberis (21.6%). The prevalence of AMR among common isolates was low for the majority of antimicrobial agents. Exploratory analysis found that the probability of Staph. aureus demonstrating resistance to penicillin was 5.16 times higher (95% confidence interval: 1.68, 15.88) in subclinical isolates relative to clinical Staph. aureus isolates. A similar association was observed for amoxicillin with subclinical Staph. aureus isolates being 4.70 times (95% confidence interval: 1.49, 14.75) more likely to be resistant than clinical Staph. aureus isolates. We concluded that the most common bacteria causing clinical mastitis in dairy herds in Australia is likely to be Strep. uberis, whereas Staph. aureus is likely to be the most common cause of subclinical mastitis. Despite decades of antimicrobial use to control these organisms, AMR appears to be uncommon., (The Authors. Published by Elsevier Inc. and Fass Inc. on behalf of the American Dairy Science Association®. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).)

The objective of this observational study was to compare 4 cow-level algorithms to predict cow-level intramammary infection (IMI) status (culture and MALDI-TOF) in late-lactation US dairy cows using standard measures of test performance. Secondary objectives were to estimate the likely effect of each algorithm, if used to guide selective dry cow therapy (SDCT), on dry cow antibiotic use in US dairy herds, and to investigate the importance of including clinical mastitis criteria in algorithm-guided SDCT. Cows (n = 1,594) from 56 US dairy herds were recruited as part of a previously published cross-sectional study of bedding management and IMI in late-lactation cows. Each herd was visited twice for sampling. At each farm visit, aseptic quarter-milk samples were collected from 20 cows approaching dry-off (>180 d pregnant), which were cultured using standard bacteriological methods and MALDI-TOF for identification of isolates. Quarter-level culture results were used to establish cow-level IMI status, which was considered the reference test in this study. Clinical mastitis records and Dairy Herd Improvement Association test-day somatic cell count data were extracted from herd records and used to perform cow-level risk assessments (low vs. high risk) using 4 algorithms that have been proposed for SDCT in New Zealand, the Netherlands, United Kingdom, and the United States. Agreement between aerobic culture (reference test; IMI vs. no-IMI) and algorithm status (high vs. low risk) was described using Cohen's kappa, test sensitivity, specificity, negative predictive value, and positive predictive value. The proportion of cows classified as high risk among the 4 algorithms ranged from 0.31 to 0.63, indicating that these approaches to SDCT could reduce antibiotic use at dry-off by 37 to 69% in the average US herd. All algorithms had poor agreement with IMI status, with kappa values ranging from 0.05 to 0.13. Sensitivity varied by pathogen, with higher values observed when detecting IMI caused by Streptococcus uberis, Streptococcus dysgalactiae, Staphylococcus aureus, and Lactococcus lactis. Negative predictive values were high for major pathogens among all algorithms (≥0.87), which may explain why algorithm-guided SDCT programs have been successfully implemented in field trials, despite poor agreement with overall IMI status. Removal of clinical mastitis criteria for each algorithm had little effect on the algorithm classification of cows, indicating that algorithms based on SCC alone may have similar performance to those based on SCC and clinical mastitis criteria. We recommend that producers implementing algorithm-guided SDCT use algorithm criteria that matches their relative aspirations for reducing antibiotic use (high specificity, positive predictive value) or minimizing untreated IMI at dry-off (high sensitivity, negative predictive value)., (Copyright © 2021 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The objectives of this study were to (1) use partial budget analysis to estimate the cash impact for herds that switch from blanket dry cow therapy (BDCT) to culture- or algorithm-guided selective dry cow therapy (SDCT) and (2) conduct a sensitivity analysis to investigate effects in situations where SDCT increased clinical and subclinical mastitis risk during the subsequent lactation. A partial budget model was created using Monte Carlo simulation with @Risk software. Expenditures associated with dry-off procedures and health outcomes (clinical and subclinical mastitis) during the first 30 d in milk were used to model herd-level effects, expressed in units of US dollars per cow dry-off. Values for each economic component were derived from findings from a recent multisite clinical trial, peer-reviewed journal articles, USDA databases, and our experiences in facilitating the implementation of SDCT on farms. Fixed values were used for variables expected to have minimal variation within the US dairy herd population (e.g., cost of rapid culture plates) and sampling distributions were used for variables that were hypothesized to vary enough to effect the herd net cash impact of one or more DCT approach(es). For Objective 1, herd-level udder health was assumed to be unaffected by the implementation of SDCT. For culture-guided SDCT, producers could expect to save an average of +$2.14 (-$2.31 to $7.23 for 5th and 95th percentiles) per cow dry-off as compared with BDCT, with 75.5% of iterations being ≥$0.00. For algorithm-guided SDCT, the mean net cash impact was +$7.85 ($3.39-12.90) per cow dry-off, with 100% of iterations being ≥$0.00. The major contributors to variance in cash impact for both SDCT approaches were percent of quarters treated at dry-off and the cost of dry cow antibiotics. For Objective 2, we repeated the partial budget model with the 30-d clinical and subclinical mastitis incidence increasing by 1, 2, and 5% (i.e., risk difference = 0.01, 0.02, and 0.05) in both SDCT groups compared with BDCT. For algorithm-guided SDCT, average net cash impacts were ≥$0.00 per cow dry-off (i.e., cost effective) when mastitis incidence increased slightly. However, as clinical mastitis incidence increased, economic returns for SDCT diminished. These findings indicate that when SDCT is implemented appropriately (i.e., no to little negative effect on health), it might be a cost-effective practice for US herds under a range of economic conditions., (Copyright © 2021 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The objective of this prospective cohort study was to explore associations between intramammary infection (IMI) in late-lactation cows and postcalving udder health and productivity. Cows (n = 2,763) from 74 US dairy herds were recruited as part of a previously published cross-sectional study of bedding management and IMI in late-lactation cows. Each herd was visited twice for sampling. At each visit, aseptic quarter milk samples were collected from 20 cows approaching dry-off (>180 d pregnant), which were cultured using standard bacteriological methods and MALDI-TOF for identification of isolates. Quarter-level culture results were used to establish cow-level IMI status at enrollment. Cows were followed from enrollment until 120 d in milk (DIM) in the subsequent lactation. Herd records were used to establish whether subjects experienced clinical mastitis or removal from the herd, and DHIA test-day data were used to record subclinical mastitis events (somatic cell count >200,000 cells/mL) and milk yield (kg/d) during the follow-up period. Cox regression and generalized estimating equations were used to evaluate the associations between IMI and the outcome of interest. The presence of late-lactation IMI caused by major pathogens was positively associated with postcalving clinical mastitis [hazard ratio = 1.5, 95% confidence interval (CI): 1.2, 2.0] and subclinical mastitis (risk ratio = 1.5, 95% CI: 1.3, 1.9). Species within the non-aureus Staphylococcus (NAS) group varied in their associations with postcalving udder health, with some species being associated with increases in clinical and subclinical mastitis in the subsequent lactation. Late-lactation IMI caused by Streptococcus and Streptococcus (Strep)-like organisms, other than Aerococcus spp. (i.e., Enterococcus, Lactococcus, and Streptococcus spp.) were associated with increases in postcalving clinical and subclinical mastitis. Test-day milk yield from 1 to 120 DIM was lower (-0.9 kg, 95% CI: -1.6, -0.3) in late-lactation cows with any IMI compared with cows without IMI. No associations were detected between IMI in late lactation and risk for postcalving removal from the herd within the first 120 DIM. Effect estimates reported in this study may be less than the underlying quarter-level effect size for IMI at dry-off and postcalving clinical and subclinical mastitis, because of the use of late-lactation IMI as a proxy for IMI at dry-off and the use of cow-level exposure and outcome measurements. Furthermore, the large number of models run in this study (n = 94) increases the chance of identifying chance associations. Therefore, confirmatory studies should be conducted. We conclude that IMI in late lactation may increase risk of clinical and subclinical mastitis in the subsequent lactation. The relationship between IMI and postcalving health and productivity is likely to vary among pathogens, with Staphylococcus aureus, Streptococcus spp., Enterococcus spp., and Lactococcus spp. being the most important pathogens identified in the current study., (Copyright © 2021 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Selective dry-cow therapy (SDCT) could be used to reduce antibiotic use on commercial dairy farms in the United States but is not yet widely adopted, possibly due to concerns about the potential for negative effects on cow health. The objective of this study was to compare culture- and algorithm-guided SDCT programs with blanket dry-cow therapy (BDCT) in a multi-site, randomized, natural exposure, non-inferiority trial for the following quarter-level outcomes: antibiotic use at dry-off, dry period intramammary infection (IMI) cure risk, dry period new IMI risk, and IMI risk at 1 to 13 d in milk (DIM). Two days before planned dry-off, cows in each of 7 herds were randomly allocated to BDCT, culture-guided SDCT (cult-SDCT), or algorithm-guided SDCT (alg-SDCT). At dry-off, BDCT cows received an intramammary antibiotic (500 mg of ceftiofur hydrochloride) in all 4 quarters. Antibiotic treatments were selectively allocated to quarters of cult-SDCT cows by treating only quarters from which aseptically collected milk samples tested positive on the Minnesota Easy 4Cast plate (University of Minnesota, St. Paul, MN) after 30 to 40 h of incubation. For alg-SDCT cows, antibiotic treatments were selectively allocated at the cow level, with all quarters receiving antibiotic treatment if the cow had either a Dairy Herd Improvement Association test somatic cell count >200,000 cells/mL during the current lactation or 2 or more clinical mastitis cases during the current lactation. All quarters of all cows were treated with an internal teat sealant. Intramammary infection status at enrollment and at 1 to 13 DIM was determined using standard bacteriological methods. The effect of treatment group on dry period IMI cure, dry period new IMI, and IMI risk at 1 to 13 DIM was determined using generalized linear mixed models (logistic), with marginal standardization to derive risk difference (RD) estimates. Quarter-level antibiotic use at dry-off for each group was BDCT (100%), cult-SDCT (45%), and alg-SDCT (45%). The crude dry period IMI cure risk for all quarters was 87.5% (818/935), the crude dry period new IMI risk was 20.1% (764/3,794), and the prevalence of IMI at 1 to 13 DIM was 23% (961/4,173). Non-inferiority analysis indicated that culture- and algorithm-guided SDCT approaches performed at least as well as BDCT for dry period IMI cure risk. In addition, the final models indicated that the risks for each of the 3 IMI measures were similar between all 3 treatment groups (i.e., RD estimates and 95% confidence intervals all close to 0). These findings indicate that under the conditions of this trial, culture- and algorithm-guided SDCT can substantially reduce antibiotic use at dry-off without negatively affecting IMI dynamics., (Copyright © 2020 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The objective of this study was to compare culture- and algorithm-guided selective dry-cow therapy (SDCT) programs with blanket dry-cow therapy (BDCT) in a multi-site, randomized, natural exposure clinical trial for the following cow-level outcomes: clinical mastitis, removal from the herd, and Dairy Herd Improvement Association (DHIA) test-day milk yield and SCC measures during the first 120 d in milk (DIM). Two days before planned dry-off, cows in each of 7 herds were randomly allocated to BDCT, culture-guided SDCT (cult-SDCT), or algorithm-guided SDCT (alg-SDCT). At dry-off, BDCT cows received an intramammary antibiotic (500 mg of ceftiofur hydrochloride) in all 4 quarters. Antibiotic treatments were selectively allocated to quarters of cult-SDCT cows by only treating quarters from which aseptically collected milk samples tested positive on a rapid culture system after 30 to 40 h of incubation. For alg-SDCT cows, antibiotic treatments were selectively allocated at the cow level, with all quarters receiving antibiotic treatment if the cow met at least one of the following criteria: (1) any DHIA test with a somatic cell count >200,000 cells/mL during the current lactation, and (2) ≥2 clinical mastitis cases during the current lactation. All quarters of all cows were treated with an internal teat sealant. Clinical mastitis and removal from the herd events (i.e., culling or death) and DHIA test-day data from dry-off to 120 DIM were extracted from herd records. Hazard ratios (HR) for the effect of treatment group on clinical mastitis and removal from the herd during 1 to 120 DIM were determined using Cox proportional hazards regression. The effects of treatment group on test-day log e -transformed SCC and milk yield were determined using linear mixed models. Final models indicated that either SDCT program was unlikely to increase clinical mastitis risk (HR cult-SDCT/BDCT = 0.82, 95% CI: 0.58, 1.15; HR alg-SDCT/BDCT = 0.83, 95% CI: 0.63, 1.09) or test-day log e SCC (cult-SDCT minus BDCT = 0.05, 95% CI: -0.09, 0.18; alg-SDCT minus BDCT = 0.07, 95% CI: -0.07, 0.21). Risk of removal from the herd and test-day milk yield were similar between treatment groups. Findings from this study indicate that culture- or algorithm-guided SDCT can be used at dry-off without negatively affecting cow health and performance in early lactation., (Copyright © 2020 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The use of an internal teat sealant (ITS) at dry-off has been repeatedly shown to improve udder health in the subsequent lactation. However, almost all ITS research conducted in North America has evaluated one product (Orbeseal, Zoetis, Parsippany, NJ). The objective of this study was to evaluate a new ITS product (Lockout, Boehringer-Ingelheim Animal Health, Duluth, GA), by comparing it directly to Orbeseal in a multi-site, randomized, positively controlled equivalence trial for health indicators during the dry period [quarter-level new intramammary infection (IMI) risk, IMI cure risk, and IMI risk at 1 to 13 d in milk, DIM] and during the first 100 DIM [clinical mastitis and culling or death risk and test-day milk somatic cell count (SCC) and milk yield]. At dry-off, cows were randomly allocated to be treated with Orbeseal or Lockout after blanket administration of a cloxacillin dry cow therapy product. Cows were then followed from dry-off until 100 DIM. Intramammary infection status at enrollment and at 1 to 13 DIM was determined using standard bacteriological methods, allowing for the measurement of IMI dynamics during the dry period (i.e., IMI cures and new IMI). The effect of ITS group on dry period IMI cure, dry period new IMI, and IMI risk at 1 to 13 DIM was determined using generalized linear mixed models (logistic). Marginal standardization was used to derive risk difference estimates. An equivalence hypothesis test was conducted to compare ITS groups for dry period new IMI risk (margin of equivalence was ±5% units). The effect of ITS group on clinical mastitis and culling or death was determined using Cox proportional hazards regression. The effect of ITS group on test-day SCC and milk yield was determined using linear mixed models. Final models indicated that measures of quarter-level IMI dynamics were similar between ITS groups (i.e., risk difference estimates and 95% confidence intervals all close to zero). Furthermore, Lockout was found to be equivalent to Orbeseal for dry period new IMI risk using an equivalence hypothesis test. Hazard ratio estimates for clinical mastitis and culling or death were close to 1 and differences in SCC and milk yield between ITS groups were close to 0, indicating negligible effects of ITS group on test-day SCC and milk yield. In most cases, these effect estimates were relatively precise (i.e., narrow 95% confidence intervals). We conclude that producers using blanket dry cow therapy could consider including Orbeseal or Lockout treatment in their programs., (Copyright © 2020 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Objectives of this study were to (1) describe the intramammary infection (IMI) prevalence and pathogen profiles in quarters of cows approaching dry-off in US dairy herds, (2) compare IMI prevalence in quarters of cows exposed to different bedding material types, and (3) identify associations between bedding bacteria count and IMI in cows approaching dry-off. Eighty herds using 1 of 4 common bedding materials (manure solids, organic non-manure, new sand, and recycled sand) were recruited in a multi-site cross-sectional study. Each herd was visited twice for sampling. At each visit, aseptic quarter-milk samples were collected from 20 cows approaching dry-off (>180 d pregnant). Samples of unused and used bedding were also collected. Aerobic culture was used to determine the IMI status of 10,448 quarters and to enumerate counts (log 10 cfu/mL) of all bacteria, Staphylococcus spp., Streptococcus spp. and Streptococcus-like organisms (SSLO), coliforms, Klebsiella spp., noncoliform gram-negatives, Bacillus spp., and Prototheca spp. in unused (n = 148) and used (n = 150) bedding. The association between bedding bacteria count and IMI was determined using multivariable logistic regression with mixed effects. Quarter-level prevalence of IMI was 21.1%, which was primarily caused by non-aureus Staphylococcus spp. (11.4%) and SSLO (5.6%). Only modest differences in IMI prevalence were observed between the 4 common bedding material types. Counts of all bacteria in unused bedding was positively associated with odds of IMI caused by any pathogen [ALL-IMI; odds ratio (OR) = 1.08]. A positive association was also observed for counts of SSLO in unused bedding and SSLO-IMI (OR = 1.09). These patterns of association were generally consistent across the 4 common bedding materials. In contrast, the association between counts of all bacteria in used bedding and ALL-IMI varied by bedding type, with positive associations observed in quarters exposed to manure solids (OR = 2.29) and organic non-manure (OR = 1.51) and a negative association in quarters exposed to new sand (OR = 0.47). Findings from this study suggest that quarter-level IMI prevalence in late-lactation cows is low in US dairy herds. Furthermore, bedding material type may not be an important risk factor for IMI in late lactation. Higher levels of bacteria in bedding may increase IMI prevalence at dry-off in general, but this relationship is likely to vary according to bedding material type., (Copyright © 2019 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Because cloth udder towels (CUT) may function as a fomite for mastitis-causing pathogens, most udder health laboratories offer towel culture services as a tool to monitor towel hygiene. However, no studies have investigated if an association exists between bacteria levels in CUT and udder health outcomes. The objectives of this cross-sectional study were to (1) describe associations between herd-level measures of towel bacteria count (ToBC) and quarter-level intramammary infection (IMI) status in late-lactation cows, (2) establish pathogen-specific target levels of bacteria in CUT to aid the interpretation of towel culture reports, and (3) identify laundering-related risk factors for high ToBC. The study was conducted in 67 herds from 10 dairy states in the United States that used CUT. These 67 herds were originally recruited as part of a larger (80 herd) cross-sectional study of bedding management. Each herd was visited once during December 2017 to April 2018 and quarter-milk samples (n = 4,656) were collected from late-gestation (>180 d pregnant) cows (n = 1,313). Two recently laundered CUT were collected and a questionnaire was used to collect information about pre-milking teat preparation and CUT management practices. Quarter-level IMI status was determined using standard bacteriologic methods. In addition, colony-forming units of all bacteria (total bacteria), Staphylococcus spp., Streptococcus spp. or Streptococcus-like organisms (SSLO), coliforms, noncoliform gram-negatives, and Bacillus spp. were determined for each pair of CUT (log 10 cfu/cm 2 ). The association between ToBC and IMI was determined using multivariable logistic regression with mixed effects. After dichotomizing ToBC into high and low categories, associations between towel management practices and ToBC category were determined using unconditional logistic regression. The quarter-level prevalence of IMI was 19.6%, which was predominantly caused by non-aureus Staphylococcus spp. (NAS; 10.2%) and SSLO (5.1%). The predominant bacteria in CUT were Bacillus spp. (median = 3.13 log 10 cfu/cm 2 ). Total bacteria count was not associated with odds of IMI (odds ratio = 1.06), likely due to the predominance of Bacillus spp. in CUT and low number of IMI caused by Bacillus spp. In contrast, counts of Staphylococcus spp. and SSLO were positively associated with odds of IMI caused by NAS (odds ratio = 1.33) and SSLO (odds ratio = 1.45), respectively. Of 12 CUT management practices evaluated, only the failure to use a dryer was identified as a clear predictor of risk for a high ToBC (risk ratio of high coliform count = 8.17). Our study findings suggest that CUT may act as a fomite for NAS and SSLO. We recommend that herds aim to keep counts of Staphylococcus spp. and SSLO in CUT below 32 cfu/cm 2 (or 5 cfu/in 2 ), and that laundered towels be completely dried in a hot air dryer., (Copyright © 2019 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Plastic sanitary sheaths over artificial insemination (AI) guns have been used at the time of AI to improve hygiene at AI and fertility in cattle, but fertility responses have been variable in studies when AI was performed by professional inseminators. The aims of this study were to investigate whether the use of a sanitary sheath at the time of AI carried out by nonprofessional (do-it-yourself, or DIY) inseminators improves pregnancy rates to AI in pasture-based dairy cows and whether effects of sheaths are greater in cows with contaminated vulvas and in cows at increased risk of extended calving to conception intervals. Lactating dairy cows located in 10 pasture-based herds in a subtropical region of northern Australia were inseminated by herd-based DIY inseminators and assigned to be inseminated with (n = 3,655) or without (n = 3,969) a sanitary sheath, with potential effects assessed using multivariable logistic regression. Overall, use of a sheath at the time of AI did not significantly affect pregnancy rates to AI (36.3% for those inseminated without a sheath vs. 36.8% for those inseminated with a sheath; odds ratio: 1.02; 95% confidence interval: 0.92-1.11). Effects of using a sheath on pregnancy rates to AI varied by herd, with lower pregnancy rates with the use of sheaths in 1 herd and some evidence of increases in 3 herds. Unexpectedly, there was evidence that the effect of sheath on pregnancy rates was less positive (or more negative) when the vulva was classified as dirty before any cleaning of the vulva before insemination compared with when the vulva was classified as clean (interaction odds ratio: 0.75; 95% CI: 0.56-1.00). Interactions between sheath and other explanatory variables that could affect fertility were not significant; thus, there was no compelling evidence that the effect of using a sheath was modified by any of these variables. We conclude that the use of sheaths during AI of pasture-based dairy cows by DIY inseminators does not, on average, markedly improve pregnancy rates to AI. However, responses may vary between herds, and the response to sheaths may be inferior (i.e., less positive or more negative) when a cow's vulva is contaminated with feces or discharge at the time of AI compared with when the vulva is clean., (Copyright © 2019 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)