J. Dairy Sci.

TBC
https://doi.org/10.3168/jds.2023-23976
© TBC, 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/).

Associations of inflammatory and reproductive tract disorders

postpartum with pregnancy and early pregnancy loss in dairy cows
T. C. Bruinjé,1 E. I. Morrison,1 E. S. Ribeiro,2 D. L. Renaud,1 and S. J. LeBlanc1*
1
Department of Population Medicine, University of Guelph, Canada N1G 2W1
2
Department of Animal Biosciences, University of Guelph, Canada N1G 2W1

ABSTRACT 29, P/AI was less likely in cows with Hp ≥1.54 g/L at
2 DIM (38 vs. 55%) or PVD (35 vs. 56%). Both metritis
Our objective was to describe associations of post- and ENDO were associated with decreased P/AI at d
partum health with pregnancy and pregnancy loss (P- 40 after AI. Cows diagnosed with metritis had greater
LOSS) from d 19 to 40 after first postpartum artificial risk of P-LOSS from d 19 to 29 (43 vs. 22%) or from
insemination (AI) in lactating Holstein cows. In 2 com- d 33 to 40 (37 vs. 7%) than cows without metritis.
mercial dairy herds in Ontario, Canada, 468 Holstein From d 29 to 33, the risk of P-LOSS was greater in
cows were enrolled 21 ± 3 d before expected parturi- cows with NEFA ≥0.73 mM at 2 DIM (13 vs. 5%) or
tion when body condition score (BCS) and lameness BCS ≤2.75 at 63 DIM (14 vs. 5%). The concentration
were assessed. Serum total Ca, haptoglobin (Hp), and of P4 on d 8 after AI was positively associated with P/
nonesterified fatty acids (NEFA) were measured at 2 AI at d 29, 33, and 40, and negatively associated with
and 6 ± 2 DIM. Blood β-hydroxybutyrate (BHB) mea- P-LOSS from d 19 to 29. Postpartum health disorders,
surement and metritis detection were done at 4, 8, 11, particularly reproductive tract disease, can have detri-
and 15 ± 2 DIM. Cows were examined for endometritis mental impacts on early pregnancy establishment and
(ENDO; ≥ 11.5% polymorphonuclear cells in endome- on pregnancy maintenance from d 19 to 40 after AI.
trial cytology) and purulent vaginal discharge (PVD) Key words: disease, fertility, reproduction, transition
at 35 ± 3 DIM. Lameness was assessed again at 21 and health
49 ± 3 DIM and BCS at 63 ± 3 DIM. First postpartum
AI occurred primarily (86%) based on detection of es-
INTRODUCTION
trus by activity monitors, on average (±SD) at 65 ± 9
DIM, and the remaining cows received timed AI at 86 Poor fertility in dairy cows is an important element
± 18 DIM. Serum progesterone (P4) was measured on of economic losses to dairy producers because it ex-
d 8 and 12 after AI, and pregnancy at first AI (P/AI) tends calving interval, reduces milk productivity, and
was estimated by the expression of ISG15 in peripheral increases risk of culling (Ribeiro et al., 2012). As sum-
blood leukocytes at d 19 after AI and by pregnancy-as- marized by Fricke and Wiltbank (2022), reproductive
sociated glycoprotein in serum at d 29, 33, and 40 after performance decreased in the USA from 1960 to the
AI. Each metabolite (Ca, Hp, NEFA, and BHB) was 2000s, then phenotypic trends for P/AI improved from
categorized above or below a cut-point identified with about 33% in 2005 to over 46% in 2020. Still, a sub-
receiver operating characteristic curve analysis associ- stantial proportion of cows may lose their pregnancy
ated with P/AI confirmed by ultrasound at d 33 from either before or after the first pregnancy diagnosis at
a larger data set. Data were analyzed using multivari- approximately 30 d after artificial insemination (AI)
able mixed logistic regression models, accounting for (Wiltbank et al., 2016), resulting in further economic
parity, health variables, covariates (season at calving losses (De Vries, 2006).
and at AI, milk yield at first DHIA test (categorized Following AI, fertilization and production of a viable
into terciles), AI method, and DIM), and herd. The embryo is expected in most cows, but assessment of
proportions of cows classified pregnant at d 19, 29, 33, pregnancy status in the early stages is difficult. There
and 40 after AI were 64, 54, 50, and 45%, respectively. is considerable variation among studies, with 52 to 71%
At d 19 after AI, P/AI was less likely in cows diagnosed of lactating cows reported to have a viable embryo
with ENDO (52 vs. 69%) or PVD (54 vs. 67%). At d recovered at uterine flush by 6 to 7 d after AI (Wilt-
bank et al., 2016). At approximately d 18, pregnancy
establishment depends on maternal recognition of the
Received July 15, 2023.
Accepted September 17, 2023. conceptus or “rescue” of the corpus luteum (CL), char-
*Corresponding author: sleblanc@​uoguelph​.ca
 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

acterized by secretion of interferon-tau (IFN-τ) by the MATERIALS AND METHODS

trophectoderm that will inhibit expression of oxytocin
receptors in the endometrium and prevent endometrial This study was approved by the Animal Care Com-
luteolytic pulses of prostaglandin-F2α (PGF2α) (Spencer mittee of the University of Guelph (AUP #4385) and
and Bazer, 2004). Using the expression of an interferon- the manuscript is reported using the STROBE-Vet
stimulated gene (ISG15) in peripheral blood leukocytes checklist (Strengthening the Reporting of Observa-
(PBL) as an indirect method to assess exposure to tional Studies in Epidemiology – Veterinary Extension)
IFN-τ by d 19 after AI, pregnancy losses by 27 to 32 d (O’Connor et al., 2016). The data reported here are
after AI averaged 27% (Albaaj et al., 2023). These data a subset from a larger observational study on factors
suggest that substantial pregnancy loss (P-LOSS) oc- associated with detection of estrus for first AI and
curs during or just after the period of maternal recogni- pregnancy outcomes (Bruinjé et al., 2023).
tion of pregnancy and before first pregnancy diagnosis,
commonly around 30 d after AI. Description of herds
Besides cow biology, genetics, herd management, or
environmental factors that contribute to pregnancy Two commercial dairy herds in southwestern Ontario
success (Santos et al., 2010; Wiltbank et al., 2016), were selected for enrollment due to their proximity to
postpartum health disorders are critical risk factors for the University of Guelph, consistent record keeping for
reproductive losses, and can affect nearly 50% of cows reproductive management decisions, and willingness to
(LeBlanc, 2012; Ribeiro et al., 2013). In a compilation participate in the study. Herds milked approximately
of data from 5,500 cows, P/AI at first diagnosis at ~30 450 Holstein cows each using either a parallel parlor
d was 51% in healthy cows, with 9% of them undergo- system (3 milking/d; herd A) or an automated milking
ing P-LOSS by d 60 after AI (Santos et al., 2010). system (3.2 milking/d; herd B), with average lactation
Pregnancy loss was more likely in cows that had a calv- milk yield (305 d) of 11,558 kg for primiparous and
ing problem (16%), purulent vaginal discharge (PVD; 14,256 kg for multiparous cows in herd A, and 9,931 kg
15%), mastitis (20%), or lameness (26%). Similarly, for primiparous and 12,495 kg for multiparous cows in
others reported a 9% probability of P-LOSS between herd B. Lactating cows were housed in freestall barns
first pregnancy diagnosis at 30 d and second diagnosis in both herds and feed was delivered 2 or 3 times daily
at 60 or 65 d of gestation in healthy cows but 1.4 to in herd A and once daily in herd B.
1.7 greater odds of loss in cows that experienced one Overall reproductive performance metrics during
or 2 reproductive tract diseases, respectively (Pinedo the study, retrieved from DairyComp305 (Valley Ag
et al., 2020), and 3.6 times greater odds in cows that Software) in herd A and B, respectively, were 25 and
experienced multiple postpartum clinical diseases (Ri- 23% 21-d pregnancy rate, 62 and 63% 21-d insemina-
beiro et al., 2013). Although the association between tion rate, and 44 and 39% P/AI. Management for first
postpartum health and pregnancy maintenance after postpartum AI was primarily based on detection of
pregnancy diagnosis is well established, less is known estrus by automated activity monitors with pedom-
regarding the effects of health disorders on early preg- eters (AfiAct, Afimilk, Kibutz Afikim) starting at 55
nancy maintenance and P-LOSS before first pregnancy DIM (herd A) or with collars (SmartTag Neck, Nedap
diagnosis. Livestock Management) starting at 50 DIM (herd B).
Therefore, our objective was to describe associations Cows were inseminated within 12 h (herd A) or 24 h
of postpartum health with P/AI and P-LOSS from 19 (herd B) following an estrus alert generated by the ac-
to 40 d after first AI in lactating Holstein cows. Sec- tivity system. In herd A, cows that were not detected
ondary objectives were to assess the relationships of in estrus received a PGF2a (500 μg cloprostenol sodium;
serum progesterone (P4) concentrations after AI with Bioestrovet, Vetoquinol or 25 mg dinoprost trometh-
P/AI and P-LOSS, and of serum pregnancy-associated amine; Lutalyse, Zoetis) injection at 73 DIM and, if
glycoprotein (PAG) levels in cows classified as pregnant not detected in estrus by 87 DIM, received a timed AI
based on PAG with subsequent P-LOSS. We hypoth- protocol [Ovsynch: GnRH (100 μg gonadorelin acetate;
esized that cows affected by health disorders would Fertiline, Vetoquinol) – 7 d – PGF2a – 56 h – GnRH – 12
have reduced P/AI and be more likely to undergo early h – AI]. Cows without a CL at ultrasound examination
P-LOSS than unaffected cows, and that lesser serum at the start of Ovsynch received an intravaginal proges-
P4 concentrations and PAG levels would be associated terone device (1.38 g progesterone; CIDR, Zoetis) from
with subsequent P-LOSS. d 0 to 7, and multiparous cows received a second PGF2a
at d 8. In herd B, primiparous cows not detected in es-
trus by 78 DIM were administered an Ovsynch protocol
if they had a CL, or received a GnRH (Fertiline, Veto-

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

quinol) injection followed by Ovsynch 7 d later if did tent was sampled using the Metricheck and cows with
not have a CL. Multiparous cows not detected in estrus mucopurulent or purulent discharge [score 2 or 3 as in
and with a CL received an Ovsynch protocol at ~65 Sheldon et al. (2006)] were classified as having PVD.
DIM, or were administered GnRH injection followed by At the same time, endometrial cytology was obtained
an Ovsynch 7 d later if did not have a CL. During the using a cytobrush at the dorsal aspect of the uterine
study, 86% of cows were bred after detection of estrus body and slides were prepared and stained using a com-
on average (±SD) at 65 ± 9 DIM, with the remainder mercial kit (Hema 3 Stat Pack; Fisher Scientific). The
bred following synchronization at 86 ± 18 DIM. proportion of polymorphonuclear cells (PMN) among
endometrial cells was evaluated over 300 cell structures
Assessment of postpartum health with an optical microscope at 400 × magnification.
Serum P4 concentrations were measured at 21, 35, 49,
Herds were visited twice weekly between May 2019 and 63 ± 3 DIM with a commercial ELISA (Ovucheck
and February 2021 by the research team. A total of Plasma, Biovet) validated for use in serum (Broes and
1,646 Holstein cows were initially enrolled, as part of LeBlanc, 2014). The threshold to determine cyclicity
a larger study, at 257 ± 3 d of gestation, from which a was 1.0 ng/mL and time to onset of cyclicity was esti-
subsample of 469 cows [150 nulliparous (79 in herd A mated by the first sample with P4 ≥ 1.0 ng/mL. The
and 71 in herd B) and 319 parous (185 in herd A and LLOQ was 0.89 ng/mL and inter- and intraassay coef-
134 in herd B) were randomly selected for the pres- ficients of variation were 6.7% and 8.9%, respectively.
ent study for data collection after first AI. Following Lameness (locomotion score ≥3) was assessed again at
first postpartum AI, the present subsample of cows (n 21 and 49 ± 3 DIM, and BCS at 63 ± 3 DIM. The
= 469) was randomly selected for this study based on median prepartum BCS was 3.50 and prepartum BCS
the day of the week that first AI occurred, such that was categorized as ≤3.50 or ≥3.75, BCS at 63 DIM was
sampling on farm visit days occurred ± 1 d from the categorized as ≤2.75 or ≥3.00, and the difference be-
target sampling day after AI, as described below. At tween pre- and postpartum BCS (median = 0.50) was
initial enrollment, prepartum cows were assessed for categorized as no loss, 0.25-point loss, or ≥0.50-point
body condition score [BCS (0 to 5 scale; Edmonson et loss. Additional information on calving ease, retained
al., 1989)] and locomotion score (0 to 5 scale; Sprecher placenta, milk fever, displaced abomasum (DA), and
et al., 1997). At 2 and 6 ± 2 DIM, blood samples were clinical mastitis (with a recorded treatment) by 30 DIM
taken from the middle coccygeal vein or artery in 10- were obtained from farm records.
mL tubes without anticoagulant (Vacutainer Precision All multiparous cows received oral Ca bolus supple-
Glide, Becton Dickinson), centrifuged at 1,500 × g for mentation immediately after calving, and cows were
15 min within 4 h, and stored at −20°C until analyzed treated for elevated BHB (≥1.2 mM) and for clinical
with an automated biochemistry analyzer (Cobas 6000 diseases according to each farm’s protocols as previ-
c 501, Roche Diagnostics) to measure serum concen- ously described (Bruinjé et al., 2023). Milk yield data
trations of total calcium (Ca), haptoglobin (Hp), and at the first DHIA test at 38 (±18) DIM were obtained
NEFA at the Animal Health Laboratory (University of from Lactanet, and categorized as lowest, middle, or
Guelph). Total Ca was measured using the Cobas Cal- highest tercile of production (kg/d) for each parity
cium Gen.2 kit (Roche Diagnostics), NEFA using the group in each herd. In the lowest, middle, and highest
Randox Manual Rx Monza kit (Randox Laboratories), tercile groups in herd A, primiparous cows averaged
and Hp according to a previously described protocol 31.4, 38.0, 44.3 kg/d and multiparous cows averaged
based on methemoglobin reduction (Makimura and 47.0, and 57.6, and 66.7 kg/d, respectively. In herd B,
Suzuki, 1982; Skinner et al., 1991). The lower limit of primiparous cows averaged 23.3, 31.2, and 38.2 kg/d
quantification (LLOQ) was 0.63 mM for total Ca, 0.05 and multiparous cows averaged 34.2, 45.4, and 54.7
mM for NEFA, and 0.02 g/L for Hp, and the intraas- kg/d respectively.
say coefficients of variation were 1.8%, 6.7%, and 7.7%,
respectively. Blood β-hydroxybutyrate (BHB) was mea- Determination of pregnancy status
sured at 4, 8, 11, and 15 ± 2 DIM using a hand-held
device (Precision Xtra Meter and B-ketone test strips, Blood samples were taken at d 8 and 12 ± 1 after
Abbott Laboratories). first AI for measurement of serum P4 in a subset of 403
Uterine health was evaluated using an intravaginal cows, when cows were expected to be in early- and mid-
device (Metricheck, Simcro) at 4, 8, 11, and 15 ± 2 diestrus with intermediate and high P4 concentrations
DIM, and a fetid, watery, and brownish or reddish (Molina-Coto et al., 2020). At d 19 ± 1, blood samples
vaginal discharge was classified as metritis, regardless were obtained from 398 cows with tubes containing
of rectal temperature. At 35 ± 3 DIM, vaginal con- 6 mL RNA stabilizing reagent (Tempus Blood RNA
Journal of Dairy Science Vol. TBC No. TBC, TBC
 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Tube, Applied Biosystems), kept on ice, and stored at pregnancy compared with cows that maintained the
−20°C within 4 d of collection until processed for isola- pregnancy (Ribeiro et al., 2014). Around these time
tion of blood leukocytes and determination of ISG15 points, cows randomly selected to be enrolled in our
expression. This gene was chosen because it is among sampling scheme after AI, as described above, would
the interferon-stimulated genes shown to have a distin- have samples taken at d 29, 33, and 40 ± 1. The dif-
guishable heightened expression in PBL on d 18 to 20 ference between the adjusted optical density (OD) of
of pregnancy compared with nonpregnant cows (Gif- the sample (OD at 635 nm minus OD at 450 nm) and
ford et al., 2007). The RNA extraction was performed the adjusted OD of the negative control was calculated,
with the Tempus Spin RNA Isolation Kit (Applied Bio- and the inter- and intra-assay coefficients of variation
systems) as per the manufacturer’s instructions, with of this arbitrary unit were 1.7% and 0.9%, respectively.
an extra DNA removal step (RapidOut DNA Removal Samples with a ratio ≥0.3 were classified as positive for
Kit, Thermo Scientific). An average of 82 (±38) ng/ pregnancy according to the manufacturer’s instructions,
μL of RNA were isolated with relative concentration and levels of PAG were interpreted as the ratio between
of nucleic acid to purified proteins (A260:​A280) of 2:1. sample and positive control values (sample-to-positive
Quantitative real-time PCR were done at the Genom- ratio; S:P). The accuracy of diagnosing pregnancy at
ics Facility (Advanced Analysis Centre, University of similar time points based on plasma PAG was >91%
Guelph) using a QuantStudio 7 Pro Real Time PCR with a different assay (Sinedino et al., 2014), and the
system (Thermo Fisher Scientific Inc.). Briefly, 20-μL positive predictive value was 100% at d 32 and 39 after
reactions comprised 1 μg RNA and 10 μL 2 × SsoAd- AI with the same assay used here (Ricci et al., 2015).
vanced Universal Inhibitor-Tolerant SYBR Green Su- Pregnancy status based on PAG at d 29 was used as
permix (Bio-Rad), 0.8 μL of PCR forward and reverse the reference to identify the optimal cut-point of ISG15
primer mix at 5 μM (final primer concentration at 200 ΔCT to predict P/AI at d 19 with the maximum com-
nM), 4.2 μL of water, and 5 μL of diluted cDNA. The bined sensitivity and specificity (Youden’s index) using
cDNA was obtained from 1 μg RNA using a High- ROC curve analysis. Cows with ISG15 ΔCT value below
Capacity cDNA Reverse Transcription Kit (Applied the cut-point, as well as those above the cut-point but
BioSystems). The reactions were run in duplicate and classified as pregnant by PAG at d 29 (false negatives
included a polymerase activation step for 3 min at 98°C, by ISG15) were categorized as pregnant at d 19.
followed by 40 cycles of a 2-step qPCR (10 s of 98°C
denaturation, 30 s of 60°C combined annealing/exten- Statistical analysis
sion). The expression of ISG15 (target gene) and of 2
reference genes [β-actin (ACTB) and ribosomal protein Data were analyzed using SAS (Studio 3.8 Edition;
L19 (RPL19)] were analyzed (Table 1). The difference version 9.04; SAS Institute Inc.) and graphs were gen-
between the threshold cycle (CT) of the ISG15 and the erated using GraphPad Prism (version 8.0.0; Graph-
geometric mean of the reference genes (Vandesompele Pad Software). In a study with a smaller sample size
et al., 2002) was used to obtain the ISG15 ΔCT values. (Molina-Coto et al., 2020), the estimated proportion of
Serum PAG was assessed at d 29, 33, and 40 (±1) cows classified pregnant at d 20 based ISG15 but not
after AI using a commercial ELISA (Bovine Pregnancy pregnant at d 32 based on US was 23 vs. 9% in cows
Test Kit, IDEXX Laboratories Inc., Westbrook, ME). with vs. without endometritis. It would require at least
Sampling days for PAG was based on a previous study 107 cows per group to detect that difference with 80%
reporting plasma pregnancy-specific protein B concen- power and 95% confidence, or 535 cows in total for a
trations increasing in pregnant cows by d 28 after AI, health disorder affecting 20% of cows. Nonetheless, for
followed by a peak between d 31 to 35, and a subtle the present study, the sample size was based on the
decrease by d 42 in cows that subsequently lost the

Table 1. Primer sequences for the target gene (ISG15) and reference genes (ACTB and RPL19) for quantification of mRNA in blood leukocytes
19 d after first insemination

Gene NCBI sequence Primer sequence

Interferon-stimulated gene 15 (ISG15) NM_174366 Forward: 5′-GGTATGAGCTGAAGCAGTT-3′
Reverse: 5′-ACCTCCCTGCTGTCAAGGT-3′
β-actin (ACTB) AY141970 Forward: 5′-CTGGACTTCGAGCAGGAGAT-3′
Reverse: 5′-GGATGTCGACGTCACACTTC-3′
Ribosomal protein L19 (RPL19) NM_001040516 Forward: 5′-ATCGATCGCCACATGTATCA-3′
Reverse: 5′-GCGTGCTTCCTTGGTCTTAG-3′

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

availability of resources and was a subset from a larger tivariable models were built but including only cows
study (Bruinjé et al., 2023). that had P4 concentration ≥1.0 ng/mL at d 8 or 12
Descriptive statistics were obtained using the MEANS among the subset that had these P4 measurements. As
and FREQ procedures in SAS. Continuous variables of described below, this excluded a small number of cows
postpartum health were categorized based on cut-points that were unlikely to be pregnant.
obtained in ROC curve analysis (LOGISTIC procedure
in SAS) associated with P/AI confirmed by ultrasound RESULTS
at d 33 in the larger study (n = 1,454; Bruinjé et al.,
2023) that included data from the same cows. For total In total, 383 cows were evaluated for P/AI at d 19 ±
Ca, ROC curve analysis was first tested for each parity 1 based on ISG15, and 469 for P/AI at d 29, 33, and
group, but because it was less discriminatory (smaller 40 ± 1 based on PAG. Cows were excluded from the
area under the curve) than when combining parity final models if they had missing measurements on the
groups, a common cut-point was used for all cows. Cut- predictor variables (n = 3 for P/AI at d 19, n = 4 for
points at 2 and 6 ± 2 DIM for total Ca were 2.09 and P/AI at d 40, and n = 1 for P-LOSS from d 29 to 33).
2.28 mM, 0.73 and 0.63 mM for NEFA, and 1.54 and Among 403 cows with serum P4 measurement after AI,
0.62 g/L for Hp, respectively. Cut-points of BHB were 6 and 3%, had P4 < 1.0 ng/mL on d 8 and 12 after AI,
0.8 mM at 4 ± 2 DIM, 0.9 mM at 8 ± 2 DIM, and 0.7 respectively, suggesting that these cows were not in-
mM at both 11 and 15 ± 2 DIM, respectively, and for seminated at the correct time or failed to ovulate after
endometritis (ENDO) was 11.5% PMN. AI. The prevalence of clinical health disorders in the
Outcomes of P/AI at d 19 and 40 and of P-LOSS larger data set was reported previously (Bruinjé et al.,
were analyzed in mixed logistic regression models using 2023). The incidence or prevalence of health variables
the GLIMMIX procedure in SAS. Predictor variables classified based on blood markers of examined clini-
evaluated were parity, metabolic events (hypocalcemia, cally among cows included in the analysis are presented
elevated NEFA, elevated Hp, hyperketonemia), pre- in Table 2. Fifty-eight percent (271/469) of cows were
partum BCS, BCS change, BCS at 63 DIM, calving clinically healthy and 29% (135/469) and 13% (63/469)
difficulty, RP, metritis, PVD, ENDO, DA, mastitis, were diagnosed with a single or with >1 clinical disor-
lameness, interval to onset of cyclicity (by 49 or 63 ± 3 der, respectively. The incidence risk of UTD and NUTD
DIM), and milk yield, in addition to the covariates of was 37% (175/469) and 20% (95/469), respectively, and
season at AI, method for first AI (detected estrus vs. 8% (40/469) of cows were diagnosed with both UTD
timed AI), and DIM at first AI. Concentrations of P4 and NUTD.
at d 8 and 12 after AI and levels of PAG at d 29 (in Based on ROC curve analysis, the cut-point of ISG15
models of P/AI and P-LOSS beyond d 29) were tested to predict P/AI at d 19, using pregnancy based on PAG
in separate models as they were considered potential at d 29 as the reference (outcome), was 7.24 ΔCT. This
intervening variables. cut-point resulted in 88% sensitivity (12% false nega-
For each outcome, predictors or covariates were tives), 70% specificity, 74% positive predictive value,
analyzed in univariable logistic regression models, and and 86% negative predictive value to predict pregnancy,
those with P ≤ 0.20 were selected for the multivari- with an area under the ROC curve of 82%. The total
able model that included herd as a random effect. Col- proportions of cows classified pregnant at d 19 (based
linearity among predictors was checked using variance on ISG15), and on d 29, 33, and 40 (based on PAG)
inflation factors, and the final model was selected by were 64% (245/382), 54% (252/469), 50% (236/469),
backward manual stepwise elimination maintaining all and 45% (209/469), respectively.
variables with P ≤ 0.10. The variance inflation fac- Among cows with P4 ≥ 1.0 ng/mL on d 8 or 12
tor was always <1.6, which indicated no issues with after AI, variables associated with reduced P/AI at d
collinearity. Removed variables were individually re- 19 in the multivariable model were parity (multiparous;
entered in the final models and retained as confounders P = 0.02), ENDO (P = 0.02), and PVD (P = 0.07)
if their inclusion changed the regression coefficient of (Table 3). Variables associated with reduced P/AI at
the significant variables >20%. In addition, separate d 29 were parity (multiparous; P < 0.01), elevated Hp
models that included disease variables combined either (≥1.54 g/L) at 2 DIM (P = 0.02), and PVD (P <
as clinical disease (any of milk fever, DA, mastitis, 0.01) (Table 4). Similarly, risk factors for reduced P/AI
lameness, RP, metritis, or PVD vs. none), or type of at d 33 were parity (multiparous; P < 0.01), elevated
disease [uterine (UTD; any of RP, metritis, ENDO, or Hp (≥1.54 g/L) at 2 DIM (P = 0.05), and PVD (P <
PVD), non-uterine clinical disease (NUTD; any of milk 0.01) (Table 5), while risk factors for reduced P/AI at
fever, DA, lameness, or mastitis), both, or none] were d 40 were parity (multiparous; P < 0.01), metritis (P
also analyzed. For pregnancy outcomes, similar mul- < 0.01), and ENDO (P = 0.05) (Table 6). From d 19
Journal of Dairy Science Vol. TBC No. TBC, TBC
 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Table 2. Incidence of transition health disorders categorized based on linear associations for P/AI on d 29 and 40, accounting
blood markers or examined clinically in Holstein cows enrolled from
two commercial dairy herds in Ontario, Canada
for covariates, are presented in Figure 3.
Risk factors for P-LOSS from d 19 to 29, 29 to 33,
Prevalence or incidence (%); and 33 to 40 are presented in Table 7. The odds of
1
Variable Definition Affected / Total (n/n)
P-LOSS from d 19 to 29 were nearly 3 times greater
Hypocalcemia2 in cows affected by metritis or by >1 clinical disease
2 ± 2 DIM ≤2.09 mM 55% (258/467) compared with unaffected cows. From d 29 to 33, the
6 ± 2 DIM ≤2.28 mM 60% (274/460)
Elevated Hp2 odds of P-LOSS were greater in cows that had NEFA
2 ± 2 DIM ≥1.54 g/L 18% (82/467) ≥0.73 mM at 2 ± 2 DIM or BCS ≤2.75 at 63 DIM.
6 ± 2 DIM ≥0.62 g/L 39% (178/460) From d 33 to 40 after AI, P-LOSS was more likely in
Elevated NEFA2
2 ± 2 DIM ≥0.73 mM 23% (105/467) cows diagnosed with metritis, DA, or with >1 clinical
6 ± 2 DIM ≥0.63 mM 32% (145/460) disease, in a multivariable model that controlled for
Hyperketonemia2 season of AI. Overall, from d 19 to 40, 36% (88/245) of
4 ± 2 DIM ≥0.8 mM 44% (205/466)
8 ± 2 DIM ≥0.9 mM 23% (108/462) cows had P-LOSS, which was 4 to 5 times more likely
11 ± 2 DIM ≥0.7 mM 54% (250/467) in cows diagnosed with metritis, DA, or >1 clinical
15 ± 2 DIM ≥0.7 mM 57% (263/462) disease (Table 8).
Retained placenta 3 8% (35/469)
Metritis 3 14% (65/469) Each 1 ng/mL increase in serum P4 concentration on
Endometritis 3 ≥11.5% PMN 23% (105/465) d 8 after AI was associated with 30% decreased odds
PVD 3 19% (90/469) (AOR: 0.70; 0.56 to 0.88; P < 0.01) of P-LOSS from d
Lameness 4 17% (79/469)
Displaced abomasum 3% (13/469) 19 to 29 and 18% decreased odds (AOR: 0.82; 0.68 to
Uterine disease 5 37% (175/469) 0.99; P = 0.04) of P-LOSS from d 19 to 40 (Figure 4).
Clinical disease 6 For each 1.0 unit increase in serum PAG S:P, the odds
One 29% (135/469)
Multiple 13% (63/469) of P-LOSS from d 29 to 33 decreased by 75% (AOR:
1
Thresholds were selected based on the maximum combined sensitivity
0.25; 0.14 to 0.44; P < 0.001) and the odds of P-LOSS
(Se) and specificity (Sp) values to predict pregnancy at d 33 after first from d 33 to 40 decreased by 58% (AOR: 0.42; 0.27 to
insemination in a larger data set (n = 1,454 cows). 0.65; P < 0.01) (Figure 5). There were no associations
2
Based on total Ca, haptoglobin (Hp), and nonesterified fat- of other health variables (hypocalcemia, hyperketone-
ty acids (NEFA) concentrations measured in serum, and blood
β-hydroxybutyrate. mia, prepartum BCS, BCS change, difficult calving,
3
Retained placenta = failure to expel fetal membranes by 24 h af- RP, mastitis, delayed cyclicity by 49 or 63 DIM) or
ter calving; Metritis = detection of fetid and colored (brownish/red- covariates (milk yield, method for first AI, DIM at first
dish) discharge at any of four examinations (4, 8, 11, and 15 ± 2 AI) with P/AI or P-LOSS in the multivariable models.
DIM); Endometritis = based on percentage of polymorphonuclear cells
(PMN) assessed in endometrial cytology at 35 ± 3 DIM; PVD = mu-
copurulent or purulent vaginal discharge at 35 ± 3 DIM. DISCUSSION
4
Lameness: locomotion score ≥3 (1 to 5 scale) at 21 or 49 ± 3 DIM.
5
Diagnosed with at least one of retained placenta, metritis, endome- Uterine disease was the major risk factor for de-
tritis, or PVD.
6 creased odds of early pregnancy and increased risk of
Diagnosed with at least one of milk fever, retained placenta, metritis,
PVD, displaced abomasum by 30 DIM, mastitis by 30 DIM, or lame- early P-LOSS. Specifically, P/AI at d 19 was reduced in
ness. cows diagnosed with ENDO or with any UTD, and P/
AI at d 33 and 40 was reduced in cows diagnosed with
metritis, ENDO, > 1 clinical disease, or with any UTD.
through 40 after AI, P/AI was lesser in cows diagnosed Cows diagnosed with metritis or >1 clinical disease
with UTD only (excluding those with a NUTD) com- were more likely to have P-LOSS from d 19 to 29 and
pared with clinically healthy cows (Figure 1), or cows from d 33 to 40, while P-LOSS from d 29 to 33 was
diagnosed with >1 clinical disease (Figure 2). There more likely in cows that had NEFA ≥0.73 mM at 2 ±
was no association between NUTD and P/AI. 2 DIM or BCS ≤2.75 at 63 DIM.
Each 1.0 ng/mL increase in serum P4 concentration The proportions of cows classified pregnant from d 19
on d 8 after AI was associated with 18% greater odds to 40 were comparable to other reports. As early as 6
(AOR: 1.18; 1.04 to 1.34; P = 0.01) of P/AI at d 29, to 7 d after AI, between 52 and 71% of lactating dairy
22% greater odds (AOR: 1.22; 1.07 to 1.38; P < 0.01) of cows had a viable embryo in uterine flushing under
P/AI at d 33, and 12% greater odds (AOR: 1.12; 0.99 thermoneutral conditions (as summarized in Wiltbank
to 1.27; P = 0.07) of P/AI at d 40. The associations of et al., 2016). Here, 64% of cows were estimated to be
P4 concentration on d 8 and the model-predicted prob- pregnant at d 19 after AI based on ISG15, and 54%,
ability of P/AI on d 29 and 33 were similar, so only the 50%, and 45% were diagnosed pregnant at d 29, 33,
and 40 based on PAG, respectively. In a summary of 4
Journal of Dairy Science Vol. TBC No. TBC, TBC
 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Table 3. Final logistic regression models of variables associated with presumed pregnancy at d 19 after first
artificial insemination (AI) based on expression of ISG15 in blood leukocytes in Holstein cows from two herds

Variable n LSM1 AOR1 95% CI1 P-value

All cows (n = 379)
Parity
Primiparous 117 71.9% Reference — —
Multiparous 262 60.3% 0.59 0.36 to 0.97 0.04
2
Endometritis
Unaffected 291 68.7% Reference — —
Affected 88 48.1% 0.42 0.26 to 0.69 <0.01
Cows with P4 ≥ 1.0 ng/mL after AI (n = 348)3
Parity
Primiparous 101 74.9% Reference — —
Multiparous 247 61.8% 0.52 0.30 to 0.89 0.02
Endometritis
Unaffected 265 68.6% Reference — —
Affected 83 52.4% 0.5 0.29 to 0.88 0.02
PVD4
Unaffected 281 67.4% Reference — —
Affected 67 54.2% 0.57 0.31 to 1.04 0.07
1
Least squares means (LSM) of predicted probability of pregnancy, adjusted odds ratio (AOR), and 95% con-
fidence intervals (CI).
2
≥ 11.5% polymorphonuclear cells in endometrial cytology performed at 35 ± 3 DIM.
3
Includes only cows that had serum progesterone (P4) concentration ≥1.0 ng/mL at d 8 or 12 after AI, among
359 cows with serum P4 measured.
4
Mucopurulent or purulent vaginal discharge (PVD) examined at 35 ± 3 DIM.

studies and over 700 cows, 70% P/AI was reported at d a study with 78 cows (Han et al., 2006), with 81% sen-
18 or 19 based on ISG15, with 20% P-LOSS from d 19 sitivity and 75% specificity, was similar to our results.
to 27 and 32% P-LOSS from d 19 to 32 (Wiltbank et Using cows not bred as negative controls, Wijma et
al., 2016). The apparent accuracy of ISG15 for P/AI in al. (2016) reported 74% sensitivity and 92% specific-

Table 4. Final logistic regression models of variables associated with pregnancy at d 29 after first artificial
insemination (AI) based on serum pregnancy-associated glycoprotein in Holstein cows from two herds

Variable n LSM1 AOR1 95% CI1 P-value

All cows (n = 469)
Parity
Primiparous 150 63.8% Reference — —
Multiparous 319 47.9% 0.52 0.35 to 0.79 <0.01
Metritis2
Unaffected 404 56.0% Reference — —
Affected 65 36.3% 0.45 0.25 to 0.81 <0.01
2
PVD
Unaffected 379 56.0% Reference — —
Affected 90 41.9% 0.57 0.34 to 0.94 0.03
3
Cows with P4 ≥ 1.0 ng/mL after AI (n = 382)
Parity
Primiparous 110 64.4% Reference — —
Multiparous 272 47.0% 0.49 0.30 to 0.80 <0.01
4
Hp at 2 DIM
<1.54 g/L 312 54.9% Reference — —
≥1.54 g/L 70 38.4% 0.51 0.29 to 0.91 0.02
PVD2
Unaffected 310 55.9% Reference — —
Affected 72 34.6% 0.42 0.24 to 0.74 <0.01
1
Least squares means (LSM) of predicted probability, adjusted odds ratio (AOR), and 95% confidence intervals
(CI).
2
Metritis = detection of fetid and colored (brownish/reddish) discharge at any of four examinations (4, 8, 11,
and 15 ± 2 DIM; PVD = mucopurulent or purulent vaginal discharge at 35 ± 3 DIM.
3
Includes only cows that had serum progesterone (P4) concentration ≥1.0 ng/mL at d 8 or 12 after AI, among
403 cows with serum P4 measured.
4
Serum haptoglobin (Hp) concentration measured at 2 ± 2 DIM.

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Table 5. Final logistic regression models of variables associated with pregnancy (P/AI) at d 33 after first
artificial insemination (AI) based on serum pregnancy-associated glycoprotein in Holstein cows from two herds

Variable n LSM1 AOR1 95% CI1 P-value

All cows (n = 465)
Parity
Primiparous 148 59.3% Reference — —
Multiparous 317 44.7% 0.55 0.37 to 0.84 <0.01
2
Metritis
Unaffected 401 52.6% Reference — —
Affected 64 31.0% 0.40 0.22 to 0.74 <0.01
Endometritis3
Unaffected 360 52.5% Reference — —
Affected 105 39.3% 0.59 0.36 to 0.95 0.03
Cows with P4 ≥ 1.0 ng/mL after AI (n = 382)4
Parity
Primiparous 110 60.4% Reference — —
Multiparous 272 43.4% 0.5 0.31 to 0.81 <0.01
Hp at 2 DIM5
<1.54 g/L 312 50.7% Reference — —
≥1.54 g/L 70 36.4% 0.55 0.31 to 0.99 0.05
3
PVD
Unaffected 310 52.1% Reference — —
Affected 72 31.5% 0.42 0.24 to 0.75 <0.01
1
Least squares means (LSM) of predicted probability, adjusted odds ratio (AOR), and 95% confidence intervals
(CI).
2
Metritis = detection of fetid and colored (brownish/reddish) discharge at any of four examinations (4, 8, 11,
and 15 ± 2 DIM).
3
Endometritis was defined as ≥11.5% polymorphonuclear cells in endometrial cytology performed at 35 ± 3
DIM; PVD = mucopurulent or purulent vaginal discharge at 35 ± 3 DIM.
4
Includes only cows that had serum progesterone (P4) concentration ≥1.0 ng/mL at d 8 or 12 after AI, among
396 cows with serum P4 measured.
5
Serum haptoglobin (Hp) concentration measured at 2 ± 2 DIM.

Table 6. Final logistic regression models of variables associated with pregnancy (P/AI) at d 40 after first
artificial insemination (AI) based on serum pregnancy-associated glycoprotein in Holstein cows from two herds

Variable n LSM1 AOR1 95% CI1 P-value

All cows (n = 465)
Parity
Primiparous 148 54.0% Reference — —
Multiparous 317 37.3% 0.51 0.33 to 0.77 <0.01
2
Metritis
Unaffected 401 47.2% Reference — —
Affected 64 19.1% 0.26 0.13 to 0.53 <0.01
3
Endometritis
Unaffected 360 46.0% Reference — —
Affected 105 31.3% 0.53 0.32 to 0.88 0.02
Cows with P4 ≥ 1.0 ng/mL after AI (n = 380)4
Parity
Primiparous 108 53.1% Reference — —
Multiparous 272 36.3% 0.5 0.31 to 0.81 <0.01
Metritis2
Unaffected 324 45.2% Reference — —
Affected 56 18.3% 0.27 0.13 to 0.58 <0.01
Endometritis3
Unaffected 288 43.9% Reference — —
Affected 92 31.2% 0.58 0.33 to 1.01 0.05
1
Least squares means (LSM) of predicted probability, adjusted odds ratio (AOR), and 95% confidence intervals
(CI).
2
Metritis = detection of fetid and colored (brownish/reddish) discharge at any of four examinations (4, 8, 11,
and 15 ± 2 DIM).
3
Endometritis was defined as ≥11.5% polymorphonuclear cells in endometrial cytology performed at 35 ± 3
DIM.
4
Includes only cows that had serum progesterone (P4) concentration ≥1.0 ng/mL at d 8 or 12 after AI, among
394 cows with serum P4 measured.

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

P-LOSS from d 19 to 33. Our findings support that: 1)

most cows likely have a viable embryo at d 19 after AI;
2) at least a quarter of cows undergo P-LOSS before
first pregnancy diagnosis at around 30 d of gestation;
3) P-LOSS before first pregnancy diagnosis is substan-
tially worsened by uterine disease approximately 2 mo
earlier.
There is limited literature reporting risk factors for
early P/AI and P-LOSS during the period of maternal
recognition of pregnancy, which occurs between d 16
and 25 after breeding (Wiltbank et al., 2016). On d 5 or
6 after AI, a lower proportion of live embryos were re-
covered in cows that had UTD (~50 vs. 70%) or NUTD
(~60 vs. 70%) compared with healthy cows (Ribeiro et
al., 2016). Ribeiro et al. (2016) also reported smaller
Figure 1. Predicted probability of pregnancy at first artificial in- conceptuses at d 15 and lesser concentration of IFN-τ
semination (P/AI) in lactating Holstein cows diagnosed with post- in the uterine flushing of cows that previously had UTD
partum uterine disease (any of retained placenta, metritis, purulent
vaginal discharge, or endometritis) compared with healthy cows not or NUTD than in the uterine flushing of cows that had
diagnosed with any uterine disease or non-uterine disease (*P ≤ 0.05) not experienced clinical diseases, although the propor-
evaluated in a multivariable model controlling for parity and body tions of cows with IFN-τ detected in the uterine flush
condition score as covariates, and herd as a random effect. Data ex-
clude cows diagnosed with a non-uterine clinical disease (n = 85 at d at d 15 were similar. With our findings of decreased
19; n = 99 at d 29, 33, and 40). Pregnancy status was estimated at d P/AI at d 19 in cows diagnosed with ENDO or with
19 after AI based on expression of ISG15 in blood leukocytes and at >1 clinical disease, we speculate that embryos of cows
d 29, 33, and 40 based on serum pregnancy-associated glycoprotein.
affected by uterine disease might have compromised de-
velopment and less IFN-τ secretion than embryos from
ity using ISG15 to detect the presence of an embryo. healthy cows. We also observed lesser P/AI at d 40 and
Han et al. (2006) observed 50% P/AI at d 19 based on greater risk of P-LOSS from d 19 to 29 and from d 33
ISG15 and 33% P-LOSS from d 19 to 33. Similarly, to 40 in cows diagnosed with metritis. In a previous
we observed 24% P-LOSS from d 19 to 29 and 28% study, cows diagnosed with metritis had indications of
compromised endometrial glandular development one
month postpartum and during the breeding period
compared with clinically healthy cows (Sellmer Ramos
et al., 2023). Impairment of endometrial function or of
early conceptus signaling during maternal recognition
of pregnancy could explain the decreased P/AI and
increased risk of early P-LOSS observed here in cows
that were diagnosed with metritis.
Uterine disease or the occurrence of multiple clinical
diseases were risk factors for decreased P/AI from d 29
to 40 after AI. Such associations have been similarly
observed in larger studies evaluating pregnancy at first
pregnancy diagnosis (Santos et al., 2010; Dubuc et al.,
2011; Ribeiro et al., 2016), including the larger study
from which the present data are drawn (Bruinjé et al.,
2023). Interestingly, in cows with serum P4 ≥ 1.0 ng/
mL after AI, elevated Hp at 2 DIM was associated
Figure 2. Predicted probability of pregnancy at first artificial in- with reduced P/AI at d 29 and 33. This suggests that
semination (P/AI) in lactating Holstein cows diagnosed with none excessive systemic inflammation shortly after parturi-
(clinically healthy), one, or multiple (>1) postpartum clinical diseases tion might have long-term consequences for pregnancy
(any of retained placenta, milk fever, displaced abomasum, lameness,
mastitis, metritis, or purulent vaginal discharge) (*P ≤ 0.05) evalu- establishment, even when accounting for other clinical
ated in a multivariable model controlling for parity and subclinical diseases.
endometritis as covariates, and herd as a random effect. Pregnancy Lesser serum P4 concentration during early diestrus
status was estimated at d 19 after AI based on expression of ISG15 in
blood leukocytes and at d 29, 33, and 40 based on serum pregnancy- after AI was associated with lower P/AI from d 29 to
associated glycoprotein. 40, and greater P-LOSS from d 19 to 29 and from d
Journal of Dairy Science Vol. TBC No. TBC, TBC
 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

19 to 40. Suboptimal circulating P4 during the early outcomes including only cows with serum P4 ≥ 1.0 ng/
diestrus phase following fertilization altered expression mL on d 8 or 12 after AI. Our findings suggest that
of genes in the endometrium and worsened embryonic possibly because of impaired uterine environment and
development (Forde et al., 2009, 2011). In a larger ob- embryonic development, cows presumably pregnant at
servational study, lower milk P4 concentration during d 19 were more likely to undergo P-LOSS by d 29.
early diestrus after AI was associated with reduced P/ Few studies looked at risk factors for P-LOSS before
AI assessed at d 50 after AI (Bruinjé et al., 2019). It is d 40 of gestation. Molina-Coto et al. (2020) reported
also possible that some cows were bred at false positive greater P-LOSS from d 22 (based on ISG15) to d 45,
estrus alerts or were asynchronous at the time of AI or from d 32 to d 45 in cows diagnosed with a uterine
and ovulated later, resulting in lower P4 at d 8 after AI disease compared with healthy cows. A study with ap-
and reduced probability of pregnancy. This could have proximately 1,400 lactating cows observed greater P-
occurred if the estrus detection by AAM and AI oc- LOSS from d 31 to 45 in cows diagnosed with mastitis,
curred slightly later than the optimal timing of AI, or if but other clinical diseases were not evaluated (Chebel
cows had incomplete luteolysis or delayed luteinization et al., 2004). In the present study, cows with elevated
during the timed AI protocol. We report models evalu- NEFA at 2 DIM or with low BCS at 63 DIM were more
ating associations between serum P4 and pregnancy likely to undergo P-LOSS from d 33 to 40, although the

Figure 3. Associations between serum progesterone (P4) concentration measured at d 8 after first artificial insemination (AI) and the model-
predicted probability of pregnancy on d 29 and 40 after AI in lactating Holstein cows with serum P4 ≥ 1.0 ng/mL on d 8 or 12 after AI from 2
commercial herds. In panel A (n = 382), the multivariable model accounted for parity [primiparous (PRIM) or multiparous (MULT)], haptoglo-
bin (Hp) at 2 ± 2 DIM, and purulent vaginal discharge (PVD). In panel B (n = 380), the model accounted for parity, metritis, and endometritis
(ENDO), and all models included herd as a random effect. Data are stratified by selected levels of covariates for clarity, and different patterns
in datapoints within a category were due to levels not shown (e.g., only metritis, only ENDO, or herd). Each 1.0 ng/mL increase in serum P4
was associated with greater odds of P/AI on d 29 (AOR: 1.18; 1.04 to 1.34; P = 0.01) and 40 after AI (AOR: 1.12; 0.99 to 1.27; P = 0.07).

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Table 7. Final logistic regression models of health variables associated with presumed pregnancy loss between
d 19 and 29, 29 to 33, and 33 to 40 after first insemination in Holstein cows from two herds

Variable n LSM1 AOR1 95% CI1 P-value

2
From d 19 to 29
Metritis3
Unaffected 217 21.7% Reference — —
Affected 28 42.9% 2.7 1.20 to 6.17 0.02
Clinical disease4
None 150 22.2% Reference — —
Single 68 20.7% 0.92 0.39 to 2.13 0.97
Multiple 27 45.0% 3.13 0.99 to 10.0 0.05
From d 29 to 332
NEFA at 2 DIM 5
≤0.73 mM 199 4.5% Reference — —
≥0.73 mM 52 12.8% 3.08 1.07 to 8.85 0.04
BCS at 63 DIM6
≥3.00 219 4.9% Reference — —
≤2.75 32 14.2% 3.23 1.01 to 10 0.05
From d 33 to 402
Metritis3
Unaffected 217 7.2% Reference — —
Affected 19 38.9% 8.26 2.65 to 25.64 <0.01
DA7
Unaffected 231 8.9% Reference — —
Affected 5 76.1% 33.33 4.33 to 250.0 <0.01
Clinical disease4
None 150 8.4% Reference — —
One 68 11.0% 1.35 0.44 to 4.13 0.81
Multiple 18 28.9% 4.44 1.03 to 19.23 0.05
1
Least squares means (LSM) of predicted probability of pregnancy loss, adjusted odds ratio (AOR), and 95%
confidence intervals (CI).
2
Pregnancy at d 19 after first insemination was determined based on the expression of ISG15 in blood leuko-
cytes, and pregnancy at d 29 was based on serum pregnancy-associated glycoprotein.
3
Metritis = detection of fetid and colored (brownish/reddish) discharge at any of four examinations (4, 8, 11,
and 15 ± 2 DIM).
4
Clinical disease variable included any of retained placenta, milk fever, displaced abomasum, lameness, masti-
tis, metritis, or purulent vaginal discharge and was evaluated in a separate multivariable model.
5
Nonesterified fatty acids (NEFA) measured in serum at 2 ± 2 DIM.
6
Body condition score (BCS; 1 to 5 scale) assessed at 63 ± 3 DIM.
7
DA = displaced abomasum diagnosed at surgical correction by 30 DIM.

Table 8. Final logistic regression models of health variables associated with presumed pregnancy loss1 between
d 19 and 40 after first insemination in 245 Holstein cows from two herds

Variable n LSM2 AOR2 95% CI2 P-value

3
Metritis
Unaffected 217 33.3% Reference — —
Affected 28 69.6% 4.59 1.94 to 10.99 <0.01
DA3
Unaffected 239 37.4% Reference — —
Affected 6 75.5% 5.15 0.89 to 29.41 0.07
Clinical disease4
None 150 34.3% Reference — —
One 68 33.7% 0.97 0.46 to 2.06 1.00
Multiple 27 66.8% 3.85 1.34 to 10.99 <0.01
1
Pregnancy at d 19 after first insemination was determined based on the expression of ISG15 in blood leuko-
cytes, and pregnancy at d 29 was based on serum pregnancy-associated glycoprotein.
2
Least squares means (LSM) of predicted probability of pregnancy loss, adjusted odds ratio (AOR), and 95%
confidence intervals (CI).
3
Metritis = detection of fetid and colored (brownish/reddish) discharge at any of four examinations (4, 8, 11,
and 15 ± 2 DIM); DA = displaced abomasum diagnosed at surgical correction by 30 DIM.
4
Clinical disease variable included any of retained placenta, milk fever, displaced abomasum, lameness, masti-
tis, metritis, or purulent vaginal discharge and was evaluated in a separate multivariable model.

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Figure 4. Associations between serum progesterone (P4) concentration measured at d 8 after first artificial insemination (AI) and the model-
predicted probability of pregnancy loss from d 19 to 29 and from d 19 and 40 after AI on 230 lactating Holstein cows from 2 commercial herds.
Pregnancy status at d 19 was estimated based on expression of ISG15 in blood leukocytes and at d 29 and 40 was based on serum pregnancy-
associated glycoprotein. In panel B, the multivariable model accounted for season at AI, metritis, and displaced abomasum (DA), and all models
accounted for herd as a random effect. Data are stratified by selected levels of covariates for clarity, and different patterns in datapoints within a
category were due to levels not shown (e.g., only metritis, only DA, or herd). Each 1 ng/mL increase in serum P4 was associated with decreased
odds of pregnancy loss from d 19 to 29 (AOR: 0.70; 0.56 to 0.88; P < 0.01) and from d 19 to 40 after AI (AOR: 0.82; 0.68 to 0.99; P = 0.04).

number of cows with P-LOSS in each category is rela- the negative association of levels of PAG at d 29 with
tively small and these associations should be interpreted P-LOSS from d 29 to 33 or from d 33 to 40 observed
cautiously. From d 33 to 40, cows diagnosed with metri- here. This is consistent with previous studies in which
tis, DA, or with >1 clinical disease were more likely to lower circulating pregnancy-specific protein B at d 24
undergo P-LOSS. We speculate that such associations or 25 of gestation was associated with P-LOSS by d 61
are attributable to impaired placental function that or 67 (Ribeiro et al., 2014; Stangaferro et al., 2021).
would result in P-LOSS before 40 to 45 d of gestation. However, associations of PAG levels and P-LOSS could
After d 27, embryo adhesion is complete, placentomes be confounded by the delay of 7 to 10 d between P-
start to develop, and embryo nutrition switches from LOSS and decrease in PAG (or PSPB) levels (Wijma
uterine histotrophic to hemotrophic nutrition (Assis et al., 2016). Using PAG as a marker, it is possible that
Neto et al., 2010). Therefore, inadequate placentation placental function is altered in cows that had health
could affect nutrient supply and possibly increase the disorders, which could be one of the pathways to early
chances of P-LOSS at this stage. The concept of sub- pregnancy losses.
optimal placental function could also be supported by

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

Figure 5. Associations between serum pregnancy-associated glycoprotein [PAG; measured as sample-to-positive ratio (S:P)] at d 29 after
first artificial insemination (AI) and the model-predicted probability of pregnancy loss from d 29 to 33 and from d 33 to 40 after AI (panel A;
n = 248) and from d 33 to 40 (panel B; n = 233) in lactating Holstein cows. Multivariable models accounted for serum NEFA concentrations
at 2 ± 2 DIM and body condition score (BCS) at 63 DIM (panel A; n = 248), or season at AI, metritis, and DA (panel B; n = 233), and all
models considered herd as a random effect. Data are stratified by selected levels of covariates for clarity. Each 1.0 unit increase in serum PAG
S:P was associated with decreased odds of P-LOSS from d 29 to 33 (AOR: 0.25; 0.14 to 0.44; P < 0.001) and from d 33 to 40 after AI (AOR:
0.42; 0.27 to 0.65; P < 0.01).

Studies frequently evaluated risk factors for P-LOSS (20%), or lameness (26%). In a study with grazing cows
at later stages, based on pregnancy diagnosis by ultra- (Ribeiro et al., 2013), risk factors for reduced P/AI at
sound at the first and second pregnancy examinations d 30 were elevated NEFA (≥0.7 mM) at 7 or 14 DIM,
(30 and 60 d after AI). In a compilation of data on over digestive problems (diarrhea, bloat, or DA), metritis, or
5,000 cows, Santos et al. (2010) reported 51% P/AI at any clinical disease, while risk factors for P-LOSS from
first diagnosis in Holstein cows not diagnosed with a d 30 to 65 were calving problems (assisted calving, twin
clinical disease, with 9% of them undergoing P-LOSS birth, stillbirth, or RP), PVD, or >1 clinical disease.
by d 60. However, P-LOSS was more likely in cows that Altogether, it seems clear that postpartum health dis-
had a calving problem (16%), PVD (15%), mastitis

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

orders are important risk factors for pregnancy losses ACKNOWLEDGMENTS

from d 19 to 65 of gestation.
A limitation of our study is that the approach used to This study was supported by the Ontario Ministry
determine early pregnancy at d 19 after AI and P-LOSS of Agriculture, Food and Rural Affairs and Natural
from d 19 cannot account for potential false positives, Sciences and Engineering Research Council of Canada
where cows presumed pregnant by ISG15 were truly (NSERC), and by doctoral scholarships provided by
non-pregnant. False positives would overestimate P/AI Dairy Farmers of Ontario and the NSERC. We thank
at d 19 and P-LOSS and result in a misclassification IDEXX Laboratories Inc. for donating the Bovine
bias, likely increasing type II error and reducing our Pregnancy Test kits and Drs. Julio Giordano (Cornell
ability to find true associations. However, we observed University) and Trevor DeVries (University of Guelph)
biologically plausible associations of health disorders for providing valuable comments in the preparation of
with P/AI at d 19 and P-LOSS from d 19 to 29 that this manuscript. We also thank the dairy producers for
would likely be stronger if false positives were mini- their cooperation and willingness to participate in the
mized. A similar proportion of false negatives (12 to study, and the students and research technicians who
19%) was observed previously for estimating pregnancy assisted with data collection.
at d 19 based on ISG15 (Han et al., 2006; Wiltbank
et al., 2016). Furthermore, our sampling protocol for REFERENCES
measurement of ISG15 encompassing only d 19 ± 1
Albaaj, A., J. Durocher, S. J. LeBlanc, and S. Dufour. 2023. Meta-
after AI (i.e., d 18, 19, or 20) could have misclassified analysis of the incidence of pregnancy losses in dairy cows at differ-
cows with a later increase in circulating IFN-τ (and ent stages to 90 days of gestation. JDS Commun. 4:144–148. https:​
ISG15 expression). This could have happened due to an /​/​doi​.org/​10​.3168/​jdsc​.2022​-0278​.
Assis Neto, A. C., F. T. V. Pereira, T. C. Santos, C. E. Ambrosio,
expected variation in the timing of conceptus attach- R. Leiser, and M. A. Miglino. 2010. Morpho-physical recording of
ment between cows, as previously described with serial bovine conceptus (Bos indicus) and placenta from Days 20 to 70 of
measurements of PSPB (Middleton and Pursley, 2019). pregnancy. Reprod. Domest. Anim. 45:760–772. https:​/​/​doi​.org/​
10​.1111/​j​.1439​-0531​.2009​.01345​.x​.
The second limitation was the financial and logistical Broes, A. J., and S. LeBlanc. 2014. Comparison of commercial pro-
constraints on our sample size. Nevertheless, to our gesterone assays for evaluation of luteal status in dairy cows. Can.
knowledge, our study is one of the first describing the Vet. J. 55:582–584.
Bruinjé, T. C., M. G. Colazo, E. S. Ribeiro, M. Gobikrushanth, and
associations of postpartum health with P-LOSS from D. J. Ambrose. 2019. Using in-line milk progesterone data to char-
19 to 40 d after AI and will contribute to better under- acterize parameters of luteal activity and their association with
standing the consequences of health disorders for early fertility in Holstein cows. J. Dairy Sci. 102:780–798. https:​/​/​doi​
.org/​10​.3168/​jds​.2018​-14654.
pregnancy establishment and maintenance. Bruinjé, T.C., E.I. Morrison, E.S. Ribeiro, D.L. Renaud, R.C. Serr-
enho, and S.J. LeBlanc. 2023. Postpartum health is associated
with detection of estrus by activity monitors and reproductive
CONCLUSION performance in dairy cows. J. Dairy Sci. (Accepted Publ. Jul 2023).
Chebel, R. C., J. E. P. Santos, J. P. Reynolds, R. L. A. Cerri, S. O.
In summary, cows were less likely to be pregnant at Juchem, and M. Overton. 2004. Factors affecting conception rate
d 19 after first AI if they had postpartum uterine dis- after artificial insemination and pregnancy loss in lactating dairy
cows. Anim. Reprod. Sci. 84:239–255. https:​/​/​doi​.org/​10​.1016/​j​
ease, and less likely to be pregnant at d 29 through 40 .anireprosci​.2003​.12​.012.
after AI if they had elevated serum haptoglobin within De Vries, A. 2006. Economic value of pregnancy in dairy cattle. J.
4 d postpartum, uterine disease, or multiple clinical Dairy Sci. 89:3876–3885. https:​/​/​doi​.org/​10​.3168/​jds​.S0022​
-0302(06)72430​-4.
diseases. Serum P4 concentrations during early dies- Dubuc, J., T. F. Duffield, K. E. Leslie, J. S. Walton, and S. J. LeBlanc.
trus were positively associated with P/AI from d 19 to 2011. Effects of postpartum uterine diseases on milk production
40. Similarly, the occurrence of a uterine disease or of and culling in dairy cows. J. Dairy Sci. 94:1339–1346. https:​/​/​doi​
.org/​10​.3168/​jds​.2010​-3758.
multiple clinical diseases increased the odds of P-LOSS Edmonson, A. J., I. J. Lean, L. D. Weaver, A. T. Farver, and G.
from d 19 to 29 and from d 33 to 40, and indicators of Webster. 1989. A body condition scoring chart for Holstein Dairy
metabolic imbalance were associated with greater odds Cows. J. Dairy Sci. 72:68–78. https:​/​/​doi​.org/​10​.3168/​jds​.S0022​
-0302(89)79081​-0.
of P-LOSS from d 29 to 33. Furthermore, PAG levels in Forde, N., M. E. Beltman, G. B. Duffy, P. Duffy, J. P. Mehta, P.
serum were positively associated with pregnancy main- O’Gaora, J. F. Roche, P. Lonergan, and M. A. Crowe. 2011.
tenance. Our findings suggest that postpartum health Changes in the Endometrial Transcriptome During the Bovine
Estrous Cycle: Effect of Low Circulating Progesterone and Con-
disorders can have meaningful detrimental effects on sequences for Conceptus Elongation. Biol. Reprod. 84:266–278.
early pregnancy establishment and maintenance. https:​/​/​doi​.org/​10​.1095/​biolreprod​.110​.085910.
Forde, N., F. Carter, T. Fair, M. A. Crowe, A. C. O. Evans, T. E.
Spencer, F. W. Bazer, R. McBride, M. P. Boland, P. O’Gaora, P.
Lonergan, and J. F. Roche. 2009. Progesterone-Regulated Changes
in Endometrial Gene Expression Contribute to Advanced Concep-

Journal of Dairy Science Vol. TBC No. TBC, TBC

 Bruinjé et al.: POSTPARTUM HEALTH AND EARLY PREGNANCY LOSS

tus Development in Cattle1. Biol. Reprod. 81:784–794. https:​/​/​doi​ associated glycoprotein (PAG) levels in plasma and milk of Hol-
.org/​10​.1095/​biolreprod​.108​.074336. stein cows during early pregnancy and their effect on the accuracy
Fricke, P. M., and M. C. Wiltbank. 2022. Symposium review: The of pregnancy diagnosis. J. Dairy Sci. 98:2502–14. doi:​https:​/​/​doi​
implications of spontaneous versus synchronized ovulations on the .org/​10​.3168/​jds​.2014​-8974​.
reproductive performance of lactating dairy cows. J. Dairy Sci. Santos, J. E., R. S. Bisinotto, E. S. Ribeiro, F. S. Lima, L. F. Greco,
doi:​https:​/​/​doi​.org/​10​.3168/​jds​.2021​-21431. C. R. Staples, and W. W. Thatcher. 2010. Applying nutrition and
Gifford, C. A., K. Racicot, D. S. Clark, K. J. Austin, T. R. Hansen, M. physiology to improve reproduction in dairy cattle. Soc. Reprod.
C. Lucy, C. J. Davies, and T. L. Ott. 2007. Regulation of Interfer- Fertil. Suppl. 67:387–403. https:​/​/​doi​.org/​10​.5661/​RDR​-VII​-387​.
on-Stimulated Genes in Peripheral Blood Leukocytes in Pregnant Sellmer Ramos, I., J. G. N. Moraes, M. O. Caldeira, S. E. Poock, T.
and Bred, Nonpregnant Dairy Cows. J. Dairy Sci. 90:274–280. E. Spencer, and M. C. Lucy. 2023. Impact of postpartum metritis
https:​/​/​doi​.org/​10​.3168/​jds​.S0022​-0302(07)72628​-0. on the regeneration of endometrial glands in dairy cows Impact
Han, H., K. J. Austin, L. A. Rempel, and T. R. Hansen. 2006. Low of postpartum metritis on the regeneration of endometrial glands
blood ISG15 mRNA and progesterone levels are predictive of non- in dairy cows. JDSC. doi:​https:​/​/​doi​.org/​10​.3168/​jdsc​.2022​-0338.
pregnant dairy cows. J. Endocrinol. 191:505–512. https:​/​/​doi​.org/​ Sheldon, I.M., G.S. Lewis, S. LeBlanc, and R.O. Gilbert. 2006. Defin-
10​.1677/​joe​.1​.07015. ing postpartum uterine disease in cattle. 65:1516–1530. doi:​https:​/​
LeBlanc, S. J. 2012. Interactions of Metabolism, Inflammation, and /​doi​.org/​10​.1016/​j​.theriogenology​.2005​.08​.021.
Reproductive Tract Health in the Postpartum Period in Dairy Sinedino, L. D. P., F. S. Lima, R. S. Bisinotto, R. L. A. Cerri, and J.
Cattle. Reprod. Domest. Anim. 47(s5):18–30. https:​/​/​doi​.org/​10​ E. P. Santos. 2014. Effect of early or late resynchronization based
.1111/​j​.1439​-0531​.2012​.02109​.x. on different methods of pregnancy diagnosis on reproductive per-
Makimura, S., and N. Suzuki. 1982. Quantitative determination of formance of dairy cows. J. Dairy Sci. 97:4932–4941. https:​/​/​doi​
bovine serum haptoglobin and its elevation in some inflamma- .org/​10​.3168/​jds​.2013​-7887.
tory diseases. Nippon Juigaku Zasshi 44:15–21. https:​/​/​doi​.org/​10​ Skinner, G., R. A. L. Brown, and L. Roberts. 1991. Bovine hapto-
.1292/​jvms1939​.44​.15​. globin response in clinically defined field conditions. Vet. Rec.
Middleton, E. L., and J. R. Pursley. 2019. Short communication: 128:147–149. https:​/​/​doi​.org/​10​.1136/​vr​.128​.7​.147​.
Blood samples before and after embryonic attachment accurately Spencer, T. E., and F. W. Bazer. 2004. Conceptus signals for estab-
determine non-pregnant lactating dairy cows at 24 d post-artificial lishing and maintenance of pregnancy. Reprod. Biol. Endocrinol.
insemination using a commercially available assay for pregnancy- 2:1–15. https:​/​/​doi​.org/​10​.1186/​1477​-7827​-2​-1​.
specific protein B. J. Dairy Sci. 102:7570–7575. https:​/​/​doi​.org/​10​ Sprecher, D. J., D. E. Hostetler, and J. B. Kaneene. 1997. A lameness
.3168/​jds​.2018​-15961. scoring system that uses posture and gait to predict dairy cattle
Molina-Coto, R., S. G. Moore, L. M. Mayo, W. R. Lamberson, S. reproductive performance. Theriogenology 47:1179–1187. https:​/​/​
E. Poock, and M. C. Lucy. 2020. Ovarian function and the es- doi​.org/​10​.1016/​S0093​-691X(97)00098​-8​.
tablishment and maintenance of pregnancy in dairy cows with Stangaferro, M. L., M. Z. Toledo, R. S. Gennari, M. M. Perez, R. V.
and without evidence of postpartum uterine disease. J. Dairy Sci. Barletta, R. Wijma, E. M. Sitko, G. Granados, M. Masello, M.
103:10715–10727. https:​/​/​doi​.org/​10​.3168/​jds​.2020​-18694. E. Van Amburgh, D. Luchini, J. O. Giordano, R. D. Shaver, and
O’Connor, A. M., J. M. Sargeant, I. R. Dohoo, H. N. Erb, M. Cevallos, M. C. Wiltbank. 2021. Effects of feeding rumen-protected methio-
M. Egger, A. K. Ersbøll, S. W. Martin, L. R. Nielsen, D. L. Pearl, nine pre- and postpartum in multiparous Holstein cows: Lactation
D. U. Pfeiffer, J. Sanchez, M. E. Torrence, H. Vigre, C. Waldner, performance and plasma amino acid concentrations. J. Dairy Sci.
and M. P. Ward. 2016. Explanation and Elaboration Document 104:7583–7603. https:​/​/​doi​.org/​10​.3168/​jds​.2021​-20190​.
for the STROBE-Vet Statement: Strengthening the Reporting of Vandesompele, J., K. De Preter, F. Pattyn, B. Poppe, N. Van Roy,
Observational Studies in Epidemiology—Veterinary Extension. A. De Paepe, and F. Speleman. 2002. Accurate normalization of
J. Vet. Intern. Med. 30:1896–1928. https:​/​/​doi​.org/​10​.1111/​jvim​ real-time quantitative RT-PCR data by geometric averaging of
.14592. multiple internal control genes. Genome Biol. 3:1–12. https:​/​/​doi​
Pinedo, P., J. E. P. Santos, R. C. Chebel, K. N. Galvão, G. M. Schuen- .org/​10​.1186/​gb​-2002​-3​-7​-research0034​.
emann, R. C. Bicalho, R. O. Gilbert, S. Rodriguez Zas, C. M. Sea- Wijma, R., M. L. Stangaferro, M. M. Kamat, S. Vasudevan, T. L. Ott,
bury, G. Rosa, and W. W. Thatcher. 2020. Early-lactation diseases and J. O. Giordano. 2016. Embryo Mortality Around the Period
and fertility in 2 seasons of calving across US dairy herds. J. Dairy of Maintenance of the Corpus Luteum Causes Alterations to the
Sci. 103:10560–10576. https:​/​/​doi​.org/​10​.3168/​jds​.2019​-17951. Ovarian Function of Lactating Dairy Cows. Biol. Reprod. 95:112.
Ribeiro, E. S., R. G. S. Bruno, A. M. Farias, J. A. Hernández-Rivera, https:​/​/​doi​.org/​10​.1095/​biolreprod​.116​.142075.
G. C. Gomes, R. Surjus, L. F. V. Becker, A. Birt, T. L. Ott, J. Wiltbank, M. C., G. M. Baez, A. Garcia-Guerra, M. Z. Toledo, P. L. J.
R. Branen, R. G. Sasser, D. H. Keisler, W. W. Thatcher, T. R. Monteiro, L. F. Melo, J. C. Ochoa, J. E. P. Santos, and R. Sartori.
Bilby, and J. E. P. Santos. 2014. Low doses of bovine somatotro- 2016. Pivotal periods for pregnancy loss during the first trimester
pin enhance conceptus development and fertility in lactating dairy of gestation in lactating dairy cows. Theriogenology 86:239–253.
cows. Biol. Reprod. 90:1–12. https:​/​/​doi​.org/​10​.1095/​biolreprod​ https:​/​/​doi​.org/​10​.1016/​j​.theriogenology​.2016​.04​.037.
.113​.114694​.
Ribeiro, E. S., K. N. Galvão, W. W. Thatcher, and J. E. P. Santos.
2012. Economic aspects of applying reproductive technologies to ORCIDS
dairy herds. Anim. Reprod. 9:370–387.
Ribeiro, E. S., G. Gomes, L. F. Greco, R. L. A. Cerri, A. Vieira-Neto, T. C. Bruinjé https:​/​/​orcid​.org/​0000​-0003​-4046​-8081
P. L. J. Monteiro Jr., F. S. Lima, R. S. Bisinotto, W. W. Thatcher, E. I. Morrison https:​/​/​orcid​.org/​0000​-0003​-2602​-0336
and J. E. P. Santos. 2016. Carryover effect of postpartum inflam- E. S. Ribeiro https:​/​/​orcid​.org/​0000​-0002​-4201​-7383
matory diseases on developmental biology and fertility in lactating D. L. Renaud https:​/​/​orcid​.org/​0000​-0002​-3439​-3987
dairy cows. J. Dairy Sci. 99:2201–2220. https:​/​/​doi​.org/​10​.3168/​
jds​.2015​-10337. S. J. LeBlanc https:​/​/​orcid​.org/​0000​-0003​-2027​-7704
Ribeiro, E. S., F. S. Lima, L. F. Greco, R. S. Bisinotto, A. P. A. Mon-
teiro, M. Favoreto, H. Ayres, R. S. Marsola, N. Martinez, W. W.
Thatcher, and J. E. P. Santos. 2013. Prevalence of periparturient
diseases and effects on fertility of seasonally calving grazing dairy
cows supplemented with concentrates. J. Dairy Sci. 96:5682–5697.
https:​/​/​doi​.org/​10​.3168/​jds​.2012​-6335.
Ricci, a, P.D. Carvalho, M.C. Amundson, R.H. Fourdraine, L. Vin-
centi, and P.M. Fricke. 2015. Factors associated with pregnancy-

Journal of Dairy Science Vol. TBC No. TBC, TBC