Você está na página 1de 10

Comparison of two methods of detecting purulent vaginal

discharge in postpartum dairy cows and effect of intrauterine


cephapirin on reproductive performance
avj_469

369..378

DJ Runciman,a* GA Andersonb and J Malmoc

Objective Part 1: compare the use of a Metricheck device (a


stainless steel probe with a semi-spherical rubber cup attached at
one end) to sample the contents of the anterior vagina with a
vaginal speculum examination for the diagnosis of pus in the
vagina of postpartum dairy cows and to investigate the association
of that pus with reproductive performance. Part 2: assess the eect
of a single intrauterine infusion of 500 mg cephapirin in cows diagnosed with vaginal purulent or mucopurulent discharge 7 to 28
days after calving on reproductive performance.
Procedure Six herds were visited fortnightly to examine cows
that had calved between 7 and 28 days (n = 423) with both the
Metricheck device and a vaginal speculum to score, by each
method, the vaginal discharge from 0 (clear or absent) to 3 (purulent) for each animal included in the study. Half of the cows that had
a positive discharge score (1 to 3 by either examination method)
were then treated with an intrauterine infusion of 500 mg of cephapirin. The relationship between Metricheck score, vaginoscopy
score, treatment and reproductive performance was assessed.
Results There was a substantial measure of agreement between
each method when scores were analysed by status. Cows that
were positive with either method had inferior reproductive
performance compared with cows with a score of zero. Treatment
of cows diagnosed with a purulent or mucopurulent discharge
with intrauterine cephapirin improved reproductive performance
in both the vaginoscopy and Metricheck groups.
Keywords dairy cattle; endometritis; Metricheck; reproduction;
vaginoscopy
Abbreviations BCS, body condition score; CIDR, controlled intravaginal drug-releasing device; CMSDI, calving to mating start date
interval (weeks); Metricheck, Metricheck; VV, visual vaginal (vaginoscopic examination)
Aust Vet J 2009;87:369378

doi: 10.1111/j.1751-0813.2009.00469.x

ndometritis in dairy cattle reduces reproductive performance.1,2 It is generally caused by bacterial infection post
partum and is dened as uterine inammation that is conned
to the endometrium.3 The denitive diagnosis of endometritis relies
on histological examination of the endometrium, or cytological
examination of uterine uid,1,2,4,5 but these techniques are of little
practical economic benet to dairy producers, because they are time
*Correspondence author.
a
University of Melbourne Rural Veterinary Unit, Mara, Victoria, Australia;
drunciman@maravet.com.au
b
University of Melbourne, Werribee, VIC, Australia
c
Mara Veterinary Centre, Mara, VIC, Australia

2009 The Authors


Journal compilation 2009 Australian Veterinary Association

PRODUCTION ANIMALS

PRODUCTION ANIMALS

consuming and costly. Clinical practitioners have used more cost


eective methods of diagnosis that are less sensitive or specic, such
as visual vaginoscopy (VV) and/or transrectal palpation of the uterus.
Of these, VV has been found to be better at predicting cows with
reduced reproductive performance and, has been associated with positive uterine cultures and histological evidence of endometrial
infection.2,69 The use of a gloved arm inserted into the vagina to
detect pus, as well as vaginal lesions, has been well described for
detecting cows with infections within the uterus post partum and does
not result in further contamination of the uterus.1,10 The diagnosis of
endometritis has been dened in terms of its eect on reproductive
performance, rather than measuring diagnostic criteria against histological and microbiological analysis.2,1214 Those studies have shown
that vaginoscopic examination of postpartum dairy cows can be useful
in predicting individuals with reduced reproductive performance that
can be treated with intrauterine antibiotics.12,13
Recently, a Metricheck device (Simcrotech, Hamilton, New Zealand)
was developed to make the diagnosis of vaginal pus easier and faster
than VV (Figure 1). It is a stainless steel rod 50 cm in length and 8 mm
diameter with a handle on one end and a semi-spherical black rubber
cup 2.5 cm diameter on the other that is designed to be introduced
into the cranial vagina to scoop any vaginal contents from the anterior
vagina for gross examination (Figure 1). It is reportedly quick and easy
to use and has only recently been validated against other diagnostic
procedures.11,15 McDougall et al used statistical techniques for two
tests for which there are no gold standards for estimating the
sensitivity and specicity of each method to detect endometritis and
found Metricheck to be of higher sensitivity but lower specicity than
vaginsocopy.15
Ultrasonography of the uterus has been found to be a more sensitive
method of detection than VV and Metricheck.11
Previous research has also demonstrated that treatment of cows with
intrauterine antibiotic infusions may help those with detected vaginal
discharge and improve their reproductive performance.7,12,14 We found
that the timing of treatment of cases diagnosed using VV was more
eective if given to cows that had calved for less than 28 days;
however, in that trial the cows were also late-calving cows and so had
less time to self-cure without the aid of antibiotics.14 Therefore, it was
proposed that another trial be devised to determine whether cows that
calved early in the calving period would also benet from treatment.
The eect of treatment of cows diagnosed with reproductive tract
disease using the Metricheck device has not been reported previously.
We report our comparison of Metricheck and VV for the diagnosis
of vaginal pus and the association between the vaginal pus status for
each method and reproductive performance. The results of the study

Australian Veterinary Journal Volume 87, No 9, September 2009

369

PRODUCTION ANIMALS

PRODUCTION ANIMALS
demonstrate the eect on reproductive performance of treating cows
with a purulent or mucopurulent discharge with intrauterine cephapirin when treatment is administered 7 to 28 days after calving.
Materials and methods
In the spring of 2004, six seasonal calving dairy herds that milked 2163
cows were enrolled in the clinical trial. From these herds, cows were
selected if they had one or more of the following conditions that may
predispose to the development of endometritis or indicate the presence of endometritis: retained fetal membranes (24 h after calving),
dystocia, twin or other multiple birth, stillbirth (or calf death within
24 h) or a vulval discharge observed 7 days or more after calving. Each
cows age, calving date and calving induction status was recorded.
The selected cows were examined at fortnightly visits to each herd,
commencing approximately 21 days after the calving start date for
each herd such that cows would be examined between 7 and 21 days
after calving. Cows that had calved up to 28 days were admitted to the
nal analysis because a previously reported study indicated that cows
examined and treated with intrauterine cephapirin within 28 days of
calving had improved reproductive performance compared with
untreated cows.14 The distribution of the calving to examination
intervals is shown in Figure 2.
On the day of examination cows were moved into a race in groups of
6 to 22 cows and body condition score (BCS) was determined using

Figure 1. Metricheck (Simcrotech, Hamilton, New Zealand; www.


simcrotech.co.nz).

a scale of 1 to 8.16 A coin toss was used to determine whether VV


or Metricheck was used rst for the rst half of the cows in the race.
Once they were examined, the other procedure was used to examine
the second half of the cows restrained. Once each race full of cows
had been examined, they were examined again using the alternative
procedure.
During the examination process, the cows tail was held and the vulva
wiped with a disposable paper towel. For the VV examination, a
vaginal speculum was then introduced into the vagina and illumination was provided by ash light. Multiple Metricheck devices and
vaginoscopes were used at each visit so that equipment was rinsed and
immersed in 10% iodine solution for at least 2 min between examinations of the cows.
The presence and consistency of vaginal discharge was scored using
the same criteria for both VV and Metricheck:
0.
1.
2.
3.

no discharge or clear mucus only


mucus and 50% pus
mucus and >50% pus
entirely pus.

Immediately following examinations, cows that were positive (score 1,


2 or 3) were assigned to receive treatment with intrauterine cephapirin
(Metricure, Intervet Australia) or left as untreated controls. The randomisation procedure for treatment used a coin toss to decide the
treatment group until half of the cows diagnosed as having a positive
score were assigned to one or other treatment group within each row
of cows examined. Once half of the cows were allocated to one group,
the remaining cows were allocated to the other group. In the case of an
uneven number of cows diagnosed with a positive discharge in each
row, the extra cow was balanced for the next row that contained an
uneven number of positive cows.
Records of mating and other reproductive treatments were kept by
herd managers and cattle were pregnancy tested 6 to 8 weeks following the end of the articial insemination period. Cows not detectably
pregnant at this time were pregnancy tested again 7 to 9 weeks following the end of the natural service period. Pregnancy testing was
performed by transrectal palpation of the uterus and estimated dates
were correlated with recorded joining dates to ensure accuracy.
All procedures were carried out by veterinarians from the Mara
Veterinary Centre. The VV, Metricheck and body condition scoring
was performed by one veterinarian (DJR) in order to maintain
consistency.
Cows were excluded and examination data were not recorded if they
had:
uterine pathology other than endometritis or metritis (such as pyometra or perimetritis)
ambiguous identication
calved for longer than 28 days at time of examination
previous intrauterine treatment.

Figure 2. Distribution of calving to examination intervals for cows


examined (n = 404, mean SD 14.2 4.6 days).

370

Australian Veterinary Journal Volume 87, No 9, September 2009

Cows were excluded from analysis of reproductive performance if


they were culled prior to pregnancy testing (n = 13) or received
2009 The Authors
Journal compilation 2009 Australian Veterinary Association

treatment for anoestrus using intravaginal progesterone (CIDR,


Genetics Australia) and intramuscular oestradiol (Cidrol, Genetics
Australia) (n = 42).
The sample size to show a signicant (P < 0.05) dierence between a
kappa statistic of 0.7 and 0.6 (power 0.8) based on positive versus
negative test with an assumption that 50% of cows would be positive
was calculated using WINPEPI.17,18 Based on these assumptions, 503
cows were required for examination. To detect a dierence in average
mating start date to conception interval of 20 days (assuming SD of 55
days and power of 90%) between treatment groups, a sample size of
320 cows that were to be VV positive or Metricheck positive was
required (160 cows per treatment group).14 If 50% of examined cows
were positive, 640 cows were therefore required for examination. It
was anticipated that we would obtain 640 cows from seven herds
(milking 3000 cows in total) if 23% of cows had conditions that
matched the selection criteria for the trial.13 However, a herd withdrew
from the trial before the rst examination day, leaving only six herds
with 2163 cows.
The University of Melbourne Animal Ethics Committee granted
approval for the procedures involving experimental use of animals
(approval number 04093).
Statistical analysis
The Metricheck scoring was compared with VV by a two-way table
using an unweighted and quadratic weighted kappa statistic as a
measure of agreement and McNemars test to compare the proportion
of cows positive to each method.19 McNemars test was also adjusted
for clustering of observations by herd.17,18,20 The kappa statistic was
interpreted as greater than 0.80 being almost perfect agreement; 0.61
to 0.80 substantial agreement, 0.41 to 0.60 moderate agreement; 0.21
to 0.40 fair agreement and 0.20 or less indicating slight agreement
between the two methods.19 The two methods were compared for each
order of observation, as well as combined (regardless of which method
was used rst), because there was concern that one method may alter
the result for the next method.
Multiple logistic regression was used to analyse variables with
dichotomous outcomes including:
proportion of cows that conceived to their rst service to articial
insemination
proportion of cows submitted for service within 3 weeks of mating
start date
proportion of cows pregnant within 6 weeks of mating start date
proportion of cows pregnant within 21 weeks of mating start date
Metricheck status
VV status.
Herd was a xed eect in all models.21
Factors tested in reproductive performance models included herd,
BCS, age, breed, calving to mating start date interval (CMSDI), treatment group and Metricheck/VV score or status. In the Metricheck and
VV status models, calving to examination interval in weeks was used
instead of CMSDI. Initially, risk factors were tested against each
outcome variable in a univariable analysis.
2009 The Authors
Journal compilation 2009 Australian Veterinary Association

All variables that were associated with an outcome of interest (P < 0.2)
were included in a full model and a reverse stepwise model building
process using a likelihood-ratio test statistic was used to arrive at a
nal model, with threshold P-values for inclusion being <0.05 and
>0.10 for exclusion. For analysis of reproductive indices, Metricheck
or VV score or status (depending on model analysed), treatment, and
herd were forced to remain in each model to calculate the nal odds
ratio (OR) and 95% condence interval of the eect of treatment and
method of examination on reproductive performance. Interactions
involving endometritis/treatment status and CMSD were assessed
within the full dataset and in subsets (VV positive or Metricheck
positive) of treated and control cows. No interaction terms were signicant (P > 0.5). Goodness of t was examined using the Hosmer
Lemeshow test, and the ability of each model to discriminate between
those cows that experienced the outcome of interest versus those who
did not was estimated by calculating the receiver operating characteristic value for each model. The relative risk and 95% CI were calculated
using a modied Poisson regression with the variables from the nal
model.22,23 The log-binomial model was not used because convergence
could not be achieved.23

PRODUCTION ANIMALS

PRODUCTION ANIMALS

Visual vaginal (VV) and Metricheck status (test negative score of 0


vs test positive score of 13) and VV and Metricheck scores were
initially tested in all models of reproductive performance accounting
for treatment. This was to determine which was a better model by both
testing the P-value of each variable in their respective models and
calculating the Akaikes information criteria statistics for each of the
models.24 P-values for VV and Metricheck status, and Akaikes information criteria for the model including VV or Metricheck status were
lower than their corresponding scores. Therefore, the results are presented using status.
Once scores were categorised into status, the variables test status and
treatment group were categorised into three groups for each examination method, to simultaneously measure the association between
test status and treatment on reproductive performance. These were
test-positive control cows, test-positive treated cows and test-negative
control cows.
The small subset of cows that were negative to the method being
examined, but were positive to the other method, were excluded from
reproductive performance analysis within their respective groups as
sample sizes were small (VV n = 13, Metricheck n = 8).
The hazard ratio for pregnancy was estimated using Coxs proportional hazards regression model stratied by herd. Stratication was
used to account for the possible non-proportionality of herd within
the model, as comparisons between herds were not of interest, but the
eect of herd needed to be accounted for.25,26 Factors found to be
associated (P < 0.2) with the risk of pregnancy in the univariable
analysis were included in a reverse stepwise process using a
likelihood-ratio test statistic, with threshold P-values for inclusion
being <0.05 and >0.10 for exclusion. Coxs proportional hazard
assumption was evaluated for categorical predictors by graphical representation of log-cumulative hazard by log time plots to determine if
lines were parallel. For all variables, scaled Schoenfeld residuals were
plotted against log of time to graphically assess any trend of a
smoothed line with log of time, and the scaled Schoenfeld residuals

Australian Veterinary Journal Volume 87, No 9, September 2009

371

PRODUCTION ANIMALS

PRODUCTION ANIMALS
were also tested for non-zero slope against log of time using a generalised linear regression. Model terms were also tested for interaction
with time.19,25,27 Harrells C index of concordance was calculated
and has values between 0 and 1, with 0.5 indicating no predictive
discrimination.28
The mating start date to conception interval was dened as the date a
cow conceived minus the mating start date plus one, such that a cow
conceiving on mating start date had a mating start date to conception
interval of 1. The nal day of the mating period was used for all cows
that survived until the nal pregnancy testing but were not pregnant.
Cows were categorised as 2, 3, 4 to 8 and >8 years old. BCS was
categorised as <4.5, 4.5, 5.0 and >5.0, as well as categories 4.5 and
> 4.5. CMSDI (in weeks) was tested in the statistical models as a linear
term. Breeds were categorised as FF (Friesian), FJ (Friesian and Jersey
Cross) or JJ (Jersey).
The software packages SPSS version 5.0 (SPSS Inc., Chicago, IL, USA),
and Stata version 9.2 (StataCorp, College Station, TX, USA) were used
for analyses. Statistical signicance was set as P < 0.05.
Results
We examined 423 cows in 6 herds (range 26151; mean 67.3, SD 50.3).
Of these, 19 cows were excluded (2 with severe uterine or vaginal
pathology, 3 with ambiguous identication, 6 with data irregularities,
6 had calved more than 28 days at time of examination, and 2 had
previous intrauterine treatment), leaving 404 cows eligible for inclusion in the study. Of the 404 cows, 202 were examined with VV as the
rst procedure and 202 were examined rst with Metricheck.
The mean age ( SD) for included cows was 5.0 2.5 years (range
212), BCS was 5.0 0.6 (range 3.06.5) and calving to examination
interval was 14.2 4.6 days (range 727; Figure 2).
The range of at risk conditions by age group are presented in Table 1.
Comparison of VV and Metricheck for diagnosis of vaginal pus
Tables 2 and 3 show that the method that was performed rst had little
bearing on the outcome of the comparison between the two diagnostic procedures. The measure of agreement between the two methods
was moderate (unweighted kappa) or almost perfect (quadratic
weighted kappa) with regard to measuring individual scores, and substantial when assessing the status (pus or no pus) of the discharge. Of
the 404 cows examined, only 109 (27.0%) had a negative score for both

methods (Table 3); 124 (30.7%) cows were Metricheck negative, and
143 (35.4%) were VV negative, therefore performing both methods
increased the sensitivity for detecting pus.
Factors associated with positive VV or Metricheck status
A model for the predictors of a positive status (Table 4) shows that the
longer cows had calved, the lower the risk of a positive Metricheck or
VV status. Primiparous cows were signicantly more likely to have a
positive Metricheck and VV status. Cows with a BCS 4.5 were more
likely to have positive VV status than cows with a BCS > 4.5 (OR 1.68,
P = 0.04).
Analysis of reproductive performance and eect of treatment
Of the 391 cows with reproductive data, 42 were administered CIDR
and intramuscular oestradiol (Cidrol) as part of the treatment for
non-visible oestrous. They were removed from analysis of reproductive performance because CIDR/oestradiol treatment reduces time to
conception in non-cycling cows.29 Analysis including CIDR cows was
performed to ascertain whether intrauterine cephapirin treatment had
any bearing on cows requiring CIDR treatment. In a regression model
including herd (P = 0.19) and CMSDI (P = 0.001), treatment with
cephapirin was not a signicant predictor of CIDR treatment (OR =
0.59, P = 0.18) in the 42 cows within the two herds that used it.
Excluding CIDR-treated cows left 349 cows for reproductive analysis.
For reproductive outcomes, for the analysis of VV, the sample size was
reduced such that cows that were VV-negative, but treated because
they were Metricheck positive were removed (n = 13), leaving 336
cows in three categories:
VV positive control cows (n = 113, 33.6%)
VV positive treated cows (n = 113, 33.6%)
VV-negative untreated cows (n = 110, 32.7%).
For the analysis of Metricheck, the sample size was similarly reduced
by the cows that were Metricheck-negative, but treated because they
were VV positive (n = 8), leaving 341 cows in three categories:
Metricheck positive control cows (n = 120, 35.2%)
Metricheck positive treated cows (n = 118, 34.6%)
Metricheck negative untreated cows (n = 103, 30.2%).
Tables 5 to 7 show the eect on reproductive outcomes of a negative
test status and treatment of positive cows with intrauterine cephapirin
compared with positive untreated cows. Method of detection/
treatment category had no signicant eect on proportion of cows

Table 1. Prevalence of disease conditions in 404 enrolled cows by age group (some cows had more than one condition)

Age group

Dystocia

Milk fever

Induction

Vulval discharge

Dead calf

Twins

RFM

2 years
3 years
48 years
>8 years
Total

77
60
228
39
404

59.7%
43.3%
37.7%
43.6%
43.3%

0.0%
0.0%
12.7%
15.4%
8.7%

3.9%
45.0%
42.1%
38.5%
34.9%

20.8%
10.0%
4.8%
0.0%
8.2%

33.8%
40.0%
34.6%
28.2%
34.7%

2.6%
1.7%
5.7%
12.8%
5.2%

9.1%
26.7%
23.7%
17.9%
20.8%

RFM, retained fetal membranes.

372

Australian Veterinary Journal Volume 87, No 9, September 2009

2009 The Authors


Journal compilation 2009 Australian Veterinary Association

PRODUCTION ANIMALS

VV score
Metricheck score

First procedure

VV
Metricheck
Total
VV
Metricheck
Total
VV
Metricheck
Total
VV
Metricheck
Total
VV
Metricheck
Total

Total

0
55
54
109
14
12
26
4
4
8

73 (36.1%)
70 (34.7%)
143 (35.4%)

4
8
12
27
21
48
9
7
16
1
1
2
41 (20.3%)
37 (18.3%)
78 (19.3%)

2
1
3
7
12
19
23
22
45
3
4
7
35 (17.3%)
39 (19.3%)
74 (18.3%)

Total

4
1
5
14
11
25
35
44
79
53 (26.2%)
56 (27.7%)
109 (27.0%)

61 (30.2%)
63 (31.2%)
124 (30.7%)
52 (25.7%)
46 (27.8%)
98 (24.2%)
50 (24.7%)
44 (24.3%)
94 (23.3%)
39 (19.3%)
49 (24.3%)
88 (21.8%)
202
202
404

PRODUCTION ANIMALS

Table 2. Distribution of Metricheck and visual vaginoscopy (VV) scores and the procedure with which cows were rst examined

VV rst: kappa = 0.59 (95% CI 0.500.67), quadratic weighted kappa = 0.82 (95% CI 0.770.88).
Metricheck rst: kappa = 0.59 (95% CI 0.510.68), quadratic weighted kappa = 0.86 (95% CI 0.810.90).
Overall: kappa = 0.59 (95% CI 0.530.65), quadratic weighted kappa = 0.84 (95% CI 0.810.88).

Table 3. Distribution of Metricheck and visual vaginoscopy (VV) status for cows and the procedure with which cows were rst examined

VV statusa
Metricheck status

First procedure

Test + ve

VV
Metricheck
Total
VV
Metricheck
Total
VV
Metricheck
Total

Test - ve

Total

Test + ve

Test - ve

Total

123
123
246
6
9
15
129 (63.9%)
132 (65.3%)
261 (64.6%)

18
16
34
55
54
109
73 (36.1%)
70 (34.7%)
143 (35.4%)

141 (69.8%)
139 (68.8%)
280 (69.3%)
61 (30.2%)
63 (31.2%)
124 (30.7%)
202
202
404

VV rst: kappa = 0.73 (95% CI 0.630.83), McNemar P = 0.014, adjusted McNemar P = 0.09.
Metricheck rst: kappa = 0.72 (95% CI 0.620.82), McNemar P = 0.16, adjusted McNemar P = 0.37.
Overall: kappa = 0.73 (95% CI 0.660.80), McNemar P = 0.007, adjusted McNemar P = 0.17.
a
Coded as test + ve if score > 0 and test - ve if score = 0.

submitted for service within 3 weeks of mating start date (P > 0.3)
(data not shown). An increased proportion of cows were pregnant
within 6 weeks of mating start in the two groups of cows with vaginal
pus that were treated and cows without vaginal pus that were not
treated, compared with cows that had vaginal pus but were not treated.
The hazard for pregnancy with time from mating start date is shown
in Table 8.
2009 The Authors
Journal compilation 2009 Australian Veterinary Association

Both VV and the Metricheck device were equally eective in predicting cows not pregnant within 42 days of mating start date (sensitivity
60.6% and specicity of 62.5% in the case of VV; 64.5% and 60.4% for
Metricheck).
Unadjusted proportion of cows pregnant within 6 weeks of mating
start date according to VV/Metricheck status and treatment group by
dierent CMSDI quartiles is shown in Figure 3. Although the unad-

Australian Veterinary Journal Volume 87, No 9, September 2009

373

PRODUCTION ANIMALS

PRODUCTION ANIMALS

Table 4. Descriptive statistics and model estimates for a positive Metricheck or VV status from multiple logistic regression (n = 404)

Metrichecka
Age (years)
2
3
48
>8
CEI (weeks)
Herd (6 herds)
Constant
VVb
Age (years)
2
3
48
>8
CEI (weeks)
BCS
4.5
>4.5
Herd (6 herds)
Constant

Proportion positive

77
60
228
39

0.84
0.78
0.64
0.54

Coe

SE

P value

OR

95% CI

Risk ratio

95% CI

0.006
0.303
1.074
1.187
0.660

0.466
0.363
0.483
0.183

2.762

0.671

0.5
0.003
0.014
<0.001
0.005

1.00
0.74
0.34
0.31
0.52

1.84
0.70
0.79
0.74

1.00
0.94
0.78
0.73
0.82

0.80
0.67
0.54
0.74

1.11
0.89
0.98
0.91

0.34
0.25
0.20
0.34

1.78
0.89
1.19
0.68

1.00
0.93
0.80
0.80
0.78

0.76
0.67
0.59
0.69

1.14
0.95
1.08
0.87

1.01

2.78

1.18

1.02

1.36

0.30
0.17
0.12
0.36

0.084
77
60
228
39

0.77
0.72
0.61
0.54

129
275

0.71
0.62

0.249
0.751
0.723
0.731

0.420
0.325
0.458
0.179

0.5
0.021
0.11
<0.001

0.517

0.257

0.044

1.00
0.78
0.47
0.49
0.48
1.68
1.00

0.006
2.027

0.629

ROC = 0.70; HosmerLemeshow P = 0.95; bROC = 0.69; HosmerLemeshow P = 0.81.


BCS, body condition score; CEI, calving to examination interval; CI, condence interval; Coe, coecient; OR, odds ratio; ROC, receiver operating
characteristic.
Table 5. Descriptive statistics and model estimates for proportion of cows that conceived to their rst service to articial insemination by visual
vaginoscopy (VV), Metricheck and treatment status groups from multiple logistic regression models that also included herd

VV/treatment groupa
Positive/control
Positive/treatment
Negative/control
Metricheck/treatment groupb
Positive/control
Positive/treatment
Negative/control

Proportion
conceived

Coe

SE

P value

103
108
104

0.24
0.39
0.43

0.674
0.816

0.320
0.330

0.036
0.013
0.10

110
111
97

0.32
0.47
0.56

0.450
0.623

0.310
0.323

0.15
0.054

OR

95% CI

Risk ratio

95% CI

0.033
1.00
1.96
2.26

1.05
1.18

3.68
4.32

1.00
1.52
1.64

1.02
1.09

2.27
2.47

1.00
1.57
1.87

0.86
0.99

2.88
3.51

1.00
1.32
1.45

0.90
0.98

1.94
2.15

ROC = 0.72; HosmerLemeshow P = 0.53; bROC = 0.71; HosmerLemeshow P = 0.50.


CI, condence interval; Coe, coecient; OR, odds ratio.

justed proportion of cows pregnant within 6 weeks of mating start


date does not account for the eect of herd or CMSDI, as are
accounted for in logistic regression, there is a trend that indicates
diagnosis of reproductive tract inammation between days 7 to 28
after calving is a predictor of reproductive performance regardless of
the time of calving relative to the mating start date in seasonal herds.
Multiple logistic regression was used to investigate the apparent

374

Australian Veterinary Journal Volume 87, No 9, September 2009

reduced eect of treatment in the later calving cows, but interaction


terms were not statistically signicant (P > 0.5).
Discussion
In the present study, VV and the Metricheck device were used to
detect intravaginal pus within a subset of high-risk cows examined 7
2009 The Authors
Journal compilation 2009 Australian Veterinary Association

PRODUCTION ANIMALS

VV/treatment groupa
Positive/control
Positive/treatment
Negative/control
Metricheck/treatment groupb
Positive/control
Positive/treatment
Negative/control

Proportion pregnant

Coe

SE

P value

113
113
110

0.32
0.48
0.55

0.696
0.835

0.286
0.294

0.015
0.005
0.003

120
118
103

0.32
0.47
0.56

0.708
0.956

0.278
0.295

0.011
0.001

OR

95% CI

Risk ratio

95% CI

0.010
1.00
2.01
2.31

1.15
1.30

3.51
4.10

1.00
1.49
1.58

1.08
1.14

2.05
2.19

1.00
2.03
2.60

1.18
1.46

3.50
4.64

1.00
1.50
1.67

1.10
1.22

2.06
2.31

PRODUCTION ANIMALS

Table 6. Descriptive statistics and model estimates for proportion of cows pregnant within 6 weeks of mating start date by VV, Metricheck and
treatment status groups from multiple logistic regression models that also included herd and the calving to mating start date interval

ROC = 0.68; HosmerLemeshow P = 0.74; bROC = 0.68; HosmerLemeshow P = 0.77.


CI, condence interval; Coe, coecient; OR, odds ratio.

Table 7. Descriptive statistics and model estimates for proportion of cows pregnant within 21 weeks of mating start date, by VV, Metricheck and
treatment status groups from multiple logistic regression models that also included herd

VV/treatment groupa
Positive/control
Positive/treatment
Negative/control
Metricheck/treatment groupb
Positive/control
Positive/treatment
Negative/control

Proportion pregnant

Coe

SE

P value

110
112
110

0.63
0.77
0.76

0.696
0.535

0.312
0.321

0.026
0.096
0.2

118
117
102

0.65
0.75
0.75

0.522
0.405

0.300
0.320

0.082
0.020

OR

95% CI

Risk ratio

95% CI

0.06
1.00
2.01
1.71

1.09
0.91

3.70
3.20

1.00
1.21
1.17

1.02
0.98

1.44
1.39

1.00
1.69
1.50

0.94
0.80

3.04
2.81

1.00
1.15
1.12

0.98
0.94

1.36
1.33

ROC = 0.69; HosmerLemeshow P = 0.99; bROC = 0.67; HosmerLemeshow P = 0.99.


CI, condence interval; Coe, coecient; OR, odds ratio.

Table 8. Hazard ratios for pregnancy following mating start date for
visual vaginoscopic (VV) or Metricheck status and treatment groups as
derived from Cox proportional hazard models stratied by herd and
accounting for calving to mating start date interval and body condition
score

VV/treatment groupa
Positive/control
Positive/treatment
Negative/control
Metricheck/treatment groupb
Positive/control
Positive/treatment
Negative/control

Hazard
ratio

113
113
110

1.00
1.47
1.54

120
118
103

1.00
1.37
1.43

P value

95% CI

0.022
0.019
0.012
0.066

1.06
1.10

2.03
2.17

0.049
0.035

1.00
1.02

1.86
2.00

Harrells C = 0.59; bHarrells C = 0.58.


CI, condence interval.

2009 The Authors


Journal compilation 2009 Australian Veterinary Association

to 28 days after calving. Of those, 64.6% or 69.3% of cows, respectively,


were diagnosed with an intravaginal purulent or mucopurulent discharge. The use of Metricheck to diagnose the discharge had a moderate level of agreement with VV using a 0 to 3 scoring system, and a
substantial agreement based on a positive versus negative scoring
system. It is the opinion of the authors that using the Metricheck is a
much quicker and easier method of diagnosing vaginal pus in dairy
cows than VV and in this trial had similar sensitivity for the detection
of purulent or mucopurulent vaginal discharge as VV.
There is no gold standard clinical test for endometritis, so Metricheck
was compared with VV, which is currently used by Australian veterinarians and has previously been shown to predict cows with reduced
reproductive performance and to be useful in assessing whether a cow
should be treated with intrauterine cephapirin.13,14 McDougall et al
assessed the sensitivity and specicity of each method for the diagnosis of endometritis and found the Metricheck had higher sensitivity,
but lower specicity than vaginoscopy.15 Others found that 45% of
cows that had no pus detected by VV were diagnosed as having
endometritis as dened using ultrasonography and uterine cytology.30
Therefore, it is improbable that the true sensitivity of VV or

Australian Veterinary Journal Volume 87, No 9, September 2009

375

PRODUCTION ANIMALS

PRODUCTION ANIMALS

guidelines also proposed that cases with mucopurulent vaginal


discharge should not be reported as endometritis until after 21 days
post partum because it may include cases of spontaneously resolving
bacterial contamination of the uterus.
The vaginal discharge detected by either VV or Metricheck device
predicted reduced reproductive performance when investigated 7 to
28 days post calving in seasonally calving dairy cows, regardless of the
time of calving. Although no signicant dierence in proportion of
cows submitted for service within 3 weeks of mating start date was
found, there were signicant and substantial dierences in the proportion of cows that conceived to their rst service to articial insemination, the proportion of cows pregnant within 6 weeks of mating
start date, and the hazard for pregnancy between VV/Metricheck
negative and untreated VV or Metricheck positive cows. With the
exception of no dierence in proportion of cows submitted for service
within 3 weeks of mating start date, these ndings are consistent with
earlier work demonstrating large and signicant detrimental association between a positive VV score and reproductive performance.13 It is
therefore likely that both Metricheck and VV were diagnosing a
disease process within the reproductive tract that delayed conception
after mating start date, even in early calving cows.

Figure 3. Observed 6WICP SE by vaginal examination method/


treatment status within each of the calving to mating start date quartiles.
Diagnosis and treatment groups: (Top) vaginoscopy (n = 336); (Bottom)
Metricheck (n = 341). (Red) Test positive control; (Blue) test positive treatment; (Green) test negative control. 6WICP, proportion of cows pregnant
within 6 weeks of mating start date.

Metricheck for the diagnosis of endometritis is as high as reported by


McDougall et al,15 based on the methods they used, rather it is the
diagnosis of purulent or mucopurulent discharge within the vagina of
postpartum cattle, not histological evidence of endometrial inammation, that is required to make a denitive diagnosis of endometritis.
The authors propose that a term such as bovine reproductive tract
inammatory disease (BRTID) be used for cases of reproductive tract
disease diagnosed using methods that only sample vaginal contents,
such as vaginoscopy, Metricheck device or a gloved hand. The term
endometritis may therefore be retained as a true pathological description of a disease process that has been investigated using more specic
diagnostic techniques such as histopathology and microbiological
analysis. A similar term, pelvic inammatory disease is commonly
used in human medicine to describe clinical cases of reproductive
tract inammation that have not been properly diagnosed.31
Previously reported guidelines classify cases with a mucopurulent or
purulent vaginal discharge as clinical endometritis only if the cow has
calved for more than 21 days, and cases that can only be diagnosed
from uterine cytology are termed subclinical endometritis.1 These

376

Australian Veterinary Journal Volume 87, No 9, September 2009

Figure 3 shows that even early calving VV and Metricheck positive


cows have reduced probability to self-cure and conceive by 42 days
after mating start date. So although Sheldon et al argued that
endometritis should not be diagnosed prior to 21 days post partum,1
the present trial demonstrates that within seasonal calving herds, these
cows are less likely to self-cure within the constraints of an annual
calving pattern and most cows had calved less than 21 days at the time
of examination (Figure 2). With 65% to 69% of high-risk cows diagnosed as having a positive discharge and 19.5% (423/2163) of cows
within herds being in this at risk group, there is potentially a large
economic impact of reduced fertility demonstrated by this and previous studies. Treatment of reproductive tract disease diagnosed using
VV or Metricheck may therefore be of economic benet, given that
signicant reproductive performance improvement was also demonstrated. Figure 3 shows the apparent reduction in treatment ecacy in
the later calving cows compared with the earlier calving cows. No
signicant interaction between CMSDI and treatment was found. The
univariable analysis shows that early calving cows with a positive
endometritis status did not self-cure, but does not account for potential confounders such as age, breed and BCS. Within the group of later
calving cows, older cows with a higher BCS were over represented
compared with earlier calving groups. Interactions between age, BCS,
breed and treatment were investigated and not found to be statistically
signicant (P > 0.5), but could account for the apparent dierence in
the eect of treatment within this subset.
We have already shown that a single intrauterine infusion of 500 mg
cephapirin given to cows with a positive VV was eective in improving
reproductive performance, with a greater eect in cows treated within
28 days of calving than in those that had calved for a longer period of
time.14 Unfortunately, in our previous study, all cows were examined
on a single visit to each herd and therefore over a wide interval from
calving to treatment, so it could not be determined whether the
reduced dierence in treatment compared with controls within
early calving cows (with long interval from calving to treatment) was
2009 The Authors
Journal compilation 2009 Australian Veterinary Association

PRODUCTION ANIMALS

In the case of VV, there was clearly a large and signicant improvement in the key reproductive performance parameters of proportion
of cows pregnant within 6 or 21 weeks of mating start date for treated
positive cows compared with untreated positive control cows. There
was also a signicant improvement in proportion of cows that conceived to their rst service to articial insemination and hazard for
pregnancy post mating start date. There was no signicant dierence
between VV-negative cows and VV positive treated cows; however, the
study lacked power in assessing the small dierences observed.
Within the subset of at risk cows examined in each herd, those with a
positive Metricheck or VV status were more likely to be rst calving
primiparous cows, and VV positive cows had lighter BCS, so the
incidence of reproductive tract infection may be reduced by addressing these risk factors. BCS has been recognised as a major factor
inuencing reproductive performance and was linked to uterine
infections in our previous trial.13 Body condition at the time of calving
has been previously shown to be associated with neutrophil function
and susceptibility to uterine infections.32,33
The association between a higher prevalence of vaginal discharge
detected using the Metricheck device or VV in primiparous cows
agrees with our previous research,13 but diers from others who found
no age disposition for endometritis,34 or an increased incidence of
endometritis with age, which was thought to be related to an impairment of neutrophil function with age.35 There was a higher proportion
of primiparous cows admitted to the trial because of dystocia and
visible vulval discharge compared with older cows. Therefore, it may
be one or both of these factors that is associated with a greater likelihood of having pus detectable within the anterior vagina by either
method used if the association is not specically related to age.
A path analysis for the associations of periparturient disease, age, BCS,
reproductive examination ndings and reproductive performance
cannot be produced from this data, as the subset of cows deemed to be
at high risk of uterine inammation was selected from herds for
examination. A further study examining all cows within herds will be
required to fully understand any association between at risk conditions, age, body condition and other variables of interest such as breed
and milk production.
The ndings of this trial conrm previous work that purulent or
mucopurulent vaginal discharge is associated with reduced reproductive performance in dairy cows examined using VV.We also found that
a diagnosis of vaginal discharge can be made using the Metricheck
device, which gives similar results to VV in terms of actual scores, as
well as its ability to predict cows with reduced reproductive performance. Also, cows with purulent or mucopurulent vaginal discharge 7
to 28 days after calving, regardless of when they calved within the
normal calving season in seasonal herds, was associated with reduced
reproductive performance.
2009 The Authors
Journal compilation 2009 Australian Veterinary Association

Acknowledgments
We thank the herd owners and managers who participated in this
study and the anonymous reviewers for their constructive comments.
We also thank Intervet Australia Ltd for providing Metricheck devices
and product for the trial.
References

PRODUCTION ANIMALS

because the disease was more dicult to treat or because the cows
were self-curing. The current study shows that even within the subset
of early calving cows there is a signicant improvement in treated
test-positive cows compared with untreated test-positive cows for
many of the indicators of reproductive performance.

1. Sheldon IM, Lewis GS, LeBlanc SJ, Gilbert RO. Dening postpartum uterine
disease in cattle. Theriogenology 2006;65:15161530.
2. LeBlanc SJ, Dueld TF, Leslie KE et al. Dening and diagnosing postpartum
clinical endometritis and its impact on reproductive performance in dairy cows.
J Dairy Sci 2002;85:22232236.
3. Smith BI, Risco CA. Clinical manifestations of postpartum metritis in dairy
cattle. Comp Cont Ved Ed 2002;24:5663.
4. Kasimanickam R, Dueld TF, Foster RA et al. Endometrial cytology and ultrasonography for detection of subclinical endometritis in postpartum dairy cows.
Theriogenology 2004;62:923.
5. Gilbert RO, Shin ST, Guard CL, Erb HN, Frajblat M. Prevalence of endometritis
and its eects on reproductive performance of dairy cows. Theriogenology
2005;65:18791888.
6. Thurmond MC, Jameson CM, Picanso JP. Eect of intrauterine antimicrobial
treatment in reducing calving-to-conception interval in cows with endometritis.
J Am Vet Med Assoc 1993;203:15761578.
7. McDougall S. Eect of intrauterine antibiotic treatment on reproductive performance of dairy cows following periparturient disease. NZ Vet J 2001;49:150158.
8. Dohmen MJW, Lohuis JACM, Huszenicza G, Nagy P, Gacs M. The relationship
between bacteriological and clinical ndings in cows with subacute/chronic
endometritis. Theriogenology 1995;43:13791388.
9. Noakes DE, Till DE, Smith GR. Bovine uterine ora postpartum: a comparison of
swabbing and biopsy. Vet Rec 1989;124:563564.
10. Sheldon IM, Noakes DE, Rycroft AN, Dobson H. Eect of postpartum manual
examination of the vagina on uterine bacterial contamination in cows. Vet Rec
2002;151:531534.
11. Mee J, Dalemans I. Diagnosis of endometritis in dairy cows using a novel
intravaginal scoop (Metricheck). In: Proceedings of the XXIV World Buiatrics Congress. World Buiatrics Congress Committee, Nice, 2006.
12. LeBlanc SJ, Dueld TF, Leslie KE et al. The eect of treatment of clinical
endometritis on reproductive performance in dairy cows. J Dairy Sci 2002;85:
22372249.
13. Runciman DJ, Anderson GA, Malmo J, Davis GM. Use of postpartum vaginoscopic (visual vaginal) examination of dairy cows for the diagnosis of endometritis and the association with reduced reproductive performance. Aust Vet J 2008;
86:205213.
14. Runciman DJ, Anderson GA, Malmo J, Davis GM. Eect of intrauterine treatment with cephapirin on the reproductive performance of seasonally calving
dairy cows at risk of endometritis following periparturient disease. Aust Vet J
2008;86:250258.
15. McDougall S, Macaulay R, Compton C. Association between endometritis
diagnosis using a novel intravaginal device and reproductive performance in
dairy cattle. Anim Reprod Sci 2007;99:923.
16. Robins C, Stockdale R, Crosby J, Morton JM, The condition magician. 2nd edn.
Department of Primary Industries, Melbourne, 2003;38.
17. Abramson JH. WINPEPI (PEPI-for-Windows): computer programs for epidemiologists. Epidemiol Perspect Innovat 2004;1:6.
18. Donner A, Eliasziw M. A goodness-of-t approach to inference procedures for
the kappa statistic: condence interval construction, signicance testing and
sample size determination. Stat Med 1992;11:15111519.
19. Dohoo I, Martin W, Stryhn H. Veterinary epidemiologic research. AVC Inc.,
Charlottetown, 2003;9193.
20. Eliasziw M, Donner A. Application of the McNemar test to non-independent
matched pair data. Stat Med 1991;10:19811991.
21. Hosmer DW, Lemeshow S. Applied logistic regression. 2nd edn. John Wiley &
Sons, New York, 2000.
22. Zou G. A modied Poisson regression approach to prospective studies with
binary data. Am J Epidemiol 2004;159:702706.
23. McNutt L, Wu C, Xue X, Hafner J. Estimating the relative risk in cohort studies
and clinical trials of common outcomes. Am J Epidemiol 2003;157:940943.

Australian Veterinary Journal Volume 87, No 9, September 2009

377

PRODUCTION ANIMALS

PRODUCTION ANIMALS
24. Akaike H. A new look at the statistical model identication. IEEE Trans Automat
Control 1974;19:716723.
25. Hosmer DW, Lemeshow S. Applied survival analysis: Regression modeling of
time to event data. John Wiley & Sons, New York, 1999.
26. Singer JD, Willett JB. Applied longitudinal data analysis: Modelling change and
event occurrence. Oxford University Press, New York, 2003.
27. Cleves MA, Gould WW, Gutierrez RG. An introduction to survival analysis using
Stata. Stata Corporation, Texas, 2004.
28. Harrell FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361387.
29. McDougall S, Compton C. Reproductive performance of anoestrus dairy
cows treated with progesterone and estradiol benzoate. J Dairy Sci 2005;88:
23882400.
30. Kasimanickam R, Dueld TF, Foster RA et al. The eect of a single administration of cephapirin or cloprostenol on the reproductive performance of dairy
cows with subclinical endometritis. Theriogenology 2005;63:818830.

BOOK REVIEW

avj_476

31. Pearce JM. Pelvic inammatory disease. BMJ 1990;300:10901091.


32. Kim IH, Na KJ, Yang MP. Immune responses during the peripartum period in
dairy cows with postpartum endometritis. J Reprod Dev 2005;51:757764.
33. Hammon DS, Evjen IM, Dhiman TR, Go JP, Walters JL. Neutrophil function
and energy status in Holstein cows with uterine health disorders. Vet Immunol
Immunopathol 2006;113:2129.
34. Hendricks KEM, Bartolome JA, Melendez P, Risco C, Archbald LF. Eect of
repeated administration of PGF2[alpha] in the early post partum period on the
prevalence of clinical endometritis and probability of pregnancy at rst insemination in lactating dairy cows. Theriogenology 2006;65:14541464.
35. Lee JY, Kim IH. Advancing parity is associated with high milk production at the
cost of body condition and increased periparturient disorders in dairy herds. J Vet
Sci 2006;7:161166.
(Accepted for publication 20 December 2008)

378..

BSAVA manual of rodents and ferrets. E Keeble and A Meredith, editors. 350 pp. A$190.00. ISBN 978-1-905319-08-4

he BSAVA manual of rodents and ferrets is the rst text by this


title, although a previously published manual, the BSAVA
manual of exotic pets, has covered many of these species in
less depth. The production of this manual is a reection of the
demand by veterinarians for more information on small mammalian pets other than dogs and cats, what we in Australia refer to as
exotics. The book is divided into two sections, the rst part devoted
to rodents and the second to ferrets. Topics covered are consistent
in both sections and include introductory chapters on biology and
husbandry and the physical examination. Chapters on diagnostic
imaging, clinical pathology, therapeutics, anaesthesia and surgery
are then followed by sections on disease using a systems-based
approach, except for the last chapter in the ferret section, Ferrets:
systemic viral diseases.
The editors, Emma Keeble and Anna Meredith, both highly
respected and qualied exotics practitioners, have gathered
together a variety of chapter authors from Europe and the
Americas; some young and emergent, others more experienced.
Unfortunately, there is only one Australian author represented,
an oversight that hopefully will be addressed in future editions.
Experienced exotic mammal practitioners, such as B Carmel,
T Donnelly, JD Lewington and D Vella, come to mind as chapter
authors for future editions. Photographs and illustrations are
generally of a high standard, particularly the chapter Ferrets:
common surgical procedures by Vittorio Capello, which contains
good-quality close-ups of surgical procedures.
Information is well presented, although the chapter on Rodents:
biology, husbandry and clinical techniques in more unusual
species could have been placed earlier in the volume. An appendix
covers dierential diagnosis from a problem-based view. Ideally,
the busy practitioner could consult the appendix rst and then refer
to the relevant chapter for more in-depth information.

378

Australian Veterinary Journal Volume 87, No 9, September 2009

Information is up to date and comprehensive, although many of


the rodent species covered are not kept as pets in Australia; for
example, chinchillas, gerbils, hamsters and degus (a small caviomorph rodent that is native to Chile). From a local perspective
the information on those species is of academic interest only. A
plethora of tables throughout the book ensures that information
is easily accessed and readable. How many of us, at short notice,
have been asked to worm rats, treat ectoparasites in mice or wondered how to treat respiratory disease in a rodent? Rodents:
therapeutics includes an extensive and useful formulary designed
for the busy clinician in these situations. The chapter on rodent
dentistry by Vladimir Jekl shows how endoscopy can be used
to diagnose oral pathology in small rodents. Ferrets: endocrine
and neoplastic diseases by Nico Schoemaker contains the latest
opinions on the diagnosis and treatment of adrenal disease. After
reading this section practitioners should have a much more
organised and positive approach to handling this challenging
condition.
This latest BSAVA publication provides up to date information for
small animal practitioners on the treatment of exotic mammals
other than rabbits. Just as rabbit medicine is now a stand alone
discipline, ferret medicine is as well. It would, therefore, be more
useful for a future manual to cover ferrets only and devote a separate volume to rodents, albeit a smaller one.
R Johnson
Robert Johnson has a rst and second opinion exotic pet practice in
Sydney and is a committee member of the Unusual and Exotic Pets
Special Interest Group of the AVA
doi: 10.1111/j.1751-0813.2009.00476.x

2009 The Author


Journal compilation 2009 Australian Veterinary Association

Você também pode gostar