Edited by: Francisco Ruiz-Fons, Spanish National Research Council (CSIC), Spain
Reviewed by: Luke O'Grady, University College Dublin, Ireland; Maya Gussmann, University of Copenhagen, Denmark
This article was submitted to Veterinary Epidemiology and Economics, a section of the journal Frontiers in Veterinary Science
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Prolonged calving-to-conception interval (CCI) can increase economic loss in cattle. We investigated the effect of post-calving abnormal reproductive tract discharge (ARTD) on CCI in dairy cows and quantified the relationship of ARTD and associated risk factors with CCI. The source population was dairy cows that calved in the study period on three pasture-based, year-round calving farms in the Riverina, NSW, Australia. Farm records and records from veterinarians' visits were analyzed. ARTD was defined as the presence of reproductive tract discharge according to the following classification: per vaginum purulent discharge ≥21 days post-calving or mucopurulent discharge >26 days post-calving. The incidence of ARTD was calculated. A Kaplan-Meier survivor function was used to estimate median time to conception post-calving dependent on the presence or absence of ARTD. Mixed effects Cox-proportional hazard models were used to estimate the direct and indirect effects of ARTD, and other potential risk factors on CCI such as body condition score (BCS), ambient temperature, and milk yield. Model structures were guided by a directed acyclic graph of potential risk factors for ARTD. The incidence of ARTD in lactations was 16% (95% CI 13.8–18.5%) and did not differ significantly between the three farms (
Minimizing the duration of the calving-to-conception interval (CCI)—the period between parturition and the following conception—of dairy cows has been found to be economically beneficial due to a variety of factors. These include increased milk yield relative to labor and feed costs (
Endometritis is defined as inflammation of the endometrium (
The detrimental effect of endometritis on reproductive performance—measured as increased CCI—occurs due to direct adverse effects on the uterine environment as well as disruption of the hormonal pathways involved in the hypothalamic (GnRH)-pituitary (FSH and LH)—reproductive tract (progesterone, estrogen, inhibin, and prostaglandin) axis and subsequent effects on ovulation, conception and embryo survival (
In addition, there are also risk factors other than endometritis for prolonged CCI, such as retained fetal membranes and metabolic disease. These risk factors can also contribute to the occurrence of endometritis. For example, retained fetal membranes are a risk factor for endometritis (
Body condition score (BCS), energy deficit, and high milk yield influence CCI and the occurrence of endometritis (
The association between parity and endometritis in dairy cows is not clear. Whilst some studies found no association between parity and endometritis (
Environmental and management factors such as season, oestrus detection and artificial insemination also influence CCI. Hotter seasons resulted in reduced oestrus expression and consequently, reduced conception rates in some studies (
Herd health management programs (HHMP) have become commonplace in many livestock systems worldwide, including dairy, as the economic margin between farm income and production costs has decreased (
Overall, there are many factors that influence CCI, either directly or indirectly, and determining the relative importance of these factors is difficult. The objective of the following study was to investigate the incidence of ARTD in dairy cows on farms on which regular veterinary examinations post-calving occurred (in the Riverina, New South Wales, Australia), and to explore the relationship of ARTD and other risk factors with CCI. To attempt to untangle factors that influence CCI relative to the detection and treatment of ARTD, we use directed acyclic graphs (DAGs) to inform investigation of risk factors for CCI using mixed-effects Cox proportional hazard models.
The source population was Holstein-Friesian cows on three dairy farms in the Riverina region of New South Wales, Australia. This was a convenience sample due to selection based on farms which had regular herd health visits by authors AG and AA. Each farm implemented a pasture-based, year-round calving system. The study population was cows which calved during the study period (January 2015—August 2017 inclusive). Farms 1, 2, and 3 maintained approximately 140, 65, and 240 lactating cows annually during the study period, respectively. Artificial insemination was used for all breeding on each farm, and all farms observed a voluntary waiting period for submission for insemination of 50 days to account for uterine involution. Retrospective data were obtained from fortnightly veterinary visits to each farm. All cows were initially scheduled for examination to identify abnormal reproductive tract discharge (ARTD) at ~30–40 days in milk (DIM), then again at ~50–60 DIM. If ARTD was detected, the cow was re-examined at the next veterinary visit. The examinations were conducted by one of two cattle medicine and reproduction specialist-trained veterinarians (AG and AA) who attended all farms regularly throughout the study period.
ARTD was determined by the color and consistency of uterine discharge withdrawn on a gloved hand during
Data that were recorded during veterinary visits by veterinarians included calving date, DIM, pregnancy status, days in-calf (pregnant, diagnosed between 30 and 60 days gestation), the presence and characteristics of uterine discharge, reproductive cyclicity status (the presence of a CL), body condition score (BCS), and concurrent disease (periparturient disease; PPD). BCS was measured to the nearest half point (0.5) for each cow, using a previously described 5-point system (
Data were cleaned to remove duplications and correct inconsistencies (for example, if insemination date was not consistent with days in calf, farm managers were consulted to identify correct dates). All analyses were performed using the software R (
Summary statistics described the number of lactations and the incidence risk of ARTD in individual cows and in all lactations. Summary statistics also described the distributions of parity, BCS, peak milk yield and the types and risk of PPD. Due to the potential increased submissions for insemination in cows with ARTD in the same lactation (“repeat breeders”), we also summarized the number of inseminations/cow and assessed whether there was a statistically significant difference between cows with and without ARTD.
Due to the presence of right censored data (pregnancy not observed during a lactation) and the likely changing population rate of pregnancy throughout the lactation, survival analysis methods were used to investigate the influence of ARTD and other potential risk factors on CCI. Right censoring was defined when a cow was not recorded as pregnant but no longer appeared in the dataset. A Kaplan-Meier survivor function was used to visually assess and estimate median time to pregnancy post-calving, dependent on the presence or absence of ARTD in the lactation. Boxplots of the duration to censoring were compared to determine if there was a difference in censoring between groups (which could then have been related to the variable of particular interest; presence or absence of ARTD).
Mixed effects Cox-proportional hazard models were used to estimate the effects of potential risk factors on CCI. Potential causal pathways and links between these variables and CCI were described and illustrated using a directed acyclic graph (DAG;
Directed acyclic graph used to inform survival analyses to identify the influence of endometritis and other potential risk factors in the peri- and post-partum period on calving to conception interval in a study of dairy cows in the Riverina, New South Wales, Australia.
The exposure variable of primary interest in this study was the presence of ARTD in the post-calving period. The direct and total effects of this variable were investigated. The total effect quantifies the influence of the presence or absence of ARTD in the post-calving period on CCI, including its effects mediated via intervening variables. The total effect is therefore, the value of greatest interest. The direct effect of ARTD in the post-calving period on CCI was also investigated to determine the relative influence of the effect of ARTD via intervening variables. In addition, the total effects of other variables of interest (BCS, parity, peak milk yield, and temperature) were quantified to assess their influence on CCI relative to the total effect of ARTD. We considered that the effects of BCS, parity, and milk yield were unlikely to be linear and therefore, investigated their effects in plausible biological categories (
Exposure variables of interest and minimal sufficient adjustment sets of covariates used in mixed-effects Cox-proportional hazards models to quantify direct and indirect effects on calving to conception interval in dairy cows in the Riverina, Australia.
ARTD | Other periparturient diseases, parity, temperature | Total |
ARTD | BCS, milk yield, other periparturient diseases, ovarian cyclicity, parity, temperature | Direct |
BCS (≤2, 2 ≤ 4, >4) | Milk yield, other periparturient diseases | Total |
Peak Milk Yield (low ≤32 L/day, high >32 L/day) | Other periparturient diseases, parity | Total |
Parity (low ≤2, high >2) | — | Total |
Temperature | — | Total |
Peri-parturient disease | Parity, reproductive tract pathogens | Total |
Natural clustering within the study data (farm and repeated measurements of individual cows) was modeled as nested random effects to account for the effects of unmeasured farm-level variables such as diet and cow-level variables such as individual susceptibility. The assumption of proportional hazards and potential existence of time dependency of variables in the models of direct and total effects ARTD on CCI were investigated by assessing Schoenfeld residuals, plots of predicted and observed data, and cumulative log plots of observed data.
Following data preparation, this study included 232, 93, and 334 cows (total 659) during the study period from Farms 1, 2, and 3, respectively. The first and last observation dates recorded during the study were 06/01/15 and 15/08/17, respectively. There were a total of 977 lactations included in the study, with 361, 141, and 475 lactations from Farms 1, 2, and 3, respectively. The number of lactations per cow during the study period ranged from 1 to 3 (median 2 lactations/cow). The mean duration of monitored periods was 125 days (95 % range 40–343 days).
On farms for which the parity of each cow was available (Farms 1 and 2), parity ranged from 1 to 8 (median = 2;
Distributions of body condition score (BCS), parity and peak mild yield of cows from farms 1 and 2 in a study to identify the influence of abnormal reproductive tract discharge and other potential risk factors on the calving to conception interval in dairy cows in the Riverina, NSW, Australia.
There was median 1 and 2 (95 % range 0–9 and 0–8) repeat inseminations in lactations with and without ARTD, respectively, which was not a statistically significant difference (Kruskal-Wallace
The incidence risk of ARTD/lactation recorded from all three farms was 0.16/lactation (
Incidence of abnormal reproductive tract discharge (ARTD) in cows and post-calving monitored periods in a study to identify the influence of ARTD and other potential risk factors on calving to conception interval in a study of dairy cows on three farms in the Riverina, NSW, Australia.
lactations | Risk | 0.18 (0.14–0.23) | 0.11 (0.07–0.18) | 0.16 (0.13–0.20) | 0.16 (0.14–0.19) |
Cow | Risk | 0.27 (0.21–0.33) | 0.16 (0.10–0.26) | 0.20 (0.16–0.25) | 0.22 (0.19–0.25) |
The incidence risk of cows with ARTD in at least one lactation from all three farms was 0.22 (
The median number of examinations for cows with ARTD was 3 and was significantly greater than the median number of examinations for cows without ARTD difference (Kruskal-Wallace
Of 821 lactations in which ARTD was not observed, pregnancy was reported in 614 lactations (right-censored = 207 lactations). Of the 156 lactations in which ARTD was observed, pregnancy was reported in 98 lactations (right-censored = 58 lactations). Twenty-four pregnancies occurred prior to the 50 days VWP (3.4% of all pregnancies, at median 45 days post-calving), and all were in cows in which ARTD was not recorded. Censoring appeared to be independent of ARTD in that the distributions of censoring in cows with and without ARTD were similar [median 105 and 83 days (95% range 33–339 and 33–363) in cows with and without ARTD, respectively, Kruskal-Wallis
A plot of Kaplan-Meier survival functions of the probability of remaining not in-calf during a lactation with and without ARTD, is shown in
Kaplan-Meier survival curve of the probability of remaining not in-calf, dependent on the presence of abnormal reproductive tract discharge (ARTD) during lactation in a study to identify the influence of ARTD and other potential risk factors on the calving to conception interval in dairy cows on three farms in the Riverina, NSW, Australia. “Number of cows at risk” shows the number of cows that remained not in-calf at time points during lactation. “Number of censored observations” shows the number of cows that did not conceive or were not detected as having conceived during the lactation. Orange = cows with ARTD.
Estimated total and direct (ARTD only) effects of variables of interest are shown in
The estimated total and direct [abnormal reproductive tract discharge (ARTD) only] effects of variables of interest from Farms 1 and 2 in a study to identify the influence of ARTD and other potential risk factors on the calving to conception interval in a study of dairy cows on three farms in The Riverina, New South Wales, Australia.
ARTD, ref: no |
Yes ( |
Total | 0.62 | 0.18 | 0.01 |
Yes ( |
Direct | 0.65 | 0.18 | 0.02 | |
BCS, ref: >4 |
≤ 2 ( |
Total | 0.78 | 0.44 | 0.56 |
2 ≤ 4 (264) | 0.82 | 0.42 | 0.64 | ||
Parity, ref: low |
High > 2 ( |
Total | 0.75 | 0.13 | 0.02 |
Peak milk yield, ref: low (≤ 32 L/day, |
High > 32 L/day |
Total | 0.63 | 0.13 | <0.001 |
Mean maximum temperature | Continuous | Total | 1.02 | 0.01 | 0.72 |
Peri-parturient disease, ref: no |
Yes ( |
Total | 0.54 | 0.38 | 0.1 |
The rate of pregnancy was significantly decreased by approximately 0.6 times in cows in which the peak milk yield was >32 L/day (median peak yield) and 0.75 times in higher than median parity (>2) cows (Kaplan-Meier survivor function plots are shown in
Assessment of the cumulative log hazard plots of observed data indicated that the curves of the groups with and without ARTD were reasonably parallel (
The incidence of ARTD on the farms in the current study is broadly consistent with other studies in dairy herds in which an incidence of 5–43% has been estimated. This broad variation could be due to many factors such as location, management differences, diagnostic method, and definition of clinical endometritis or ARTD (
Although the association between ARTD, pregnancy rate, and prolonged CCI appears to be consistent with other studies, the magnitude of the effect of ARTD on pregnancy rate and CCI in the current study was greater than expected given the findings of other studies. When accounting for herd, parity, and ovarian status, cows diagnosed with clinical endometritis in a study in Canada had a reduced relative pregnancy rate of 27% (for example, a decrease in 21-day pregnancy rate from 20 to 14.6%), and a 32 days increase in median time to pregnancy (
As described in the introduction, there are many factors involved in the occurrence of endometritis and prolonged CCI. Many relate to management of cows in the dry and peri-partum periods and it is possible that environmental conditions in Australia (phases of drought in many locations throughout the study period, including the Riverina, NSW) are making it more difficult to maintain metabolically healthy pasture-based dairy cows in these periods. This could be resulting in increased difficulty to resolve problems such as ARTD post-calving. Further research is needed to determine if alternate protocols to resolve ARTD would be of greater benefit on the farms in the current study, and determine the influence of dry cow and calving cow management. It would also be of interest to extend this study to further dairy farms in Australia. An economic analysis of the benefit of treatment of cows with ARTD would be warranted if our findings are reproduced. The economic benefit of treatments for endometritis have been found to be herd-specific (
We found that CCI was also influenced by parity and peak milk yield. Consistent with these findings, Hillers et al. (
There are some limitations associated with the current study. Detailed information about predictors that vary over time, such as BCS and milk yield throughout lactation, were not available. In future studies we suggest the inclusion of such information in the Cox proportional hazard models as time varying covariates (including ambient temperature) to determine more accurately their effect on CCI. In particular, lactation curves could be inferred from milk records (monthly on the farms in this study) so that peak yield could be more accurately defined. Early conception is not possible to measure and if embryonic death (ED) or early fetal loss (EFL) occurred prior to pregnancy diagnosis the impact of endometritis on the CCI might have been overestimated. However, we also made the assumption that ED or EFL had causes other than endometritis; if endometritis was the cause, the impact of endometritis on CCI might have been underestimated (
Overall, the high proportion of ARTD in dairy herds found in this study and its significant influence on the CCI reinforces the importance of reproductive management of dairy cows in the pre- and periparturient period. As well as research into the treatment and prevention of endometritis, further research on farms in other areas of Australia to determine the impact of ARTD is worthwhile because the economic impact on the dairy industry might be large. The magnitude of difference in CCI between cows with and without ARTD despite regular veterinary examination of post-calving cows brings into question the value of veterinary interventions to improve herd fertility in a herd health management program.
The datasets generated for this study are available on request to the corresponding author.
The study was exempt from ethics approval by the Animal Care and Ethics Committee of Charles Sturt University because pre-existing clinical records were used.
MH collected the data. MH and VB performed the analysis. All authors conceived, designed the study, and contributed to the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors acknowledge the farmers and the CSU librarians, especially Lee-Anne McInerney, for their assistance in this study. This study is part of the Bachelor of Veterinary Biology/Bachelor of Veterinary Science (Honours) dissertation of MH.
The Supplementary Material for this article can be found online at:
Literature review undertaken to inform DAG in a study to identify the influence of endometritis and other potential risk factors on the calving to conception interval in a study of dairy cows on three farms in The Riverina, New South Wales, Australia.
log-cumulative hazard plots and observed vs. predicted hazard plots of cows with and without abnormal reproductive tract discharge (ARTD), and plots of Schoenfeld residuals of variables in Cox proportional hazards models to determine the association between ARTD and the calving to conception interval of cows in a study to identify the influence of ARTD and other potential risk factors on the calving to conception interval in dairy cows on three farms in the Riverina, NSW, Australia.
Boxplots of time to right censoring in cows with and without abnormal reproductive tract discharge (ARTD) in a study to identify the influence of ARTD and other potential risk factors on the calving to conception interval in a study of dairy cows on three farms in the Riverina, New South Wales, Australia.
Kaplan-Meier survival curve of the probability of remaining not in-calf dependent on peak milk yield. The table of “Number of cows at risk” shows the number of cows that remained non-pregnant at time points following calving. The graph of the “Number of censored cows” shows the number of cows that did not conceive or were not detected as having conceived in the allocated study period (orange = cows with low peak milk yield).
Kaplan-Meier survival curve of the probability of remaining not in-calf dependent on parity. The table of “Number of cows at risk” shows the number of cows that remained non-pregnant at time points following calving. The graph of the “Number of censored cows” shows the number of cows that did not conceive or were not detected as having conceived in the allocated study period (orange = cows with low peak milk yield).
calving-to-conception interval
retained fetal membranes
days in milk
abnormal reproductive tract discharge
body condition score
confidence interval.