Journal of Dairy Research (1997) 64 23–37 Printed in Great Britain 23

Patterns of clinical mastitis manifestations in Danish organic

dairy herds

B METTE VAARST  CARSTEN ENEVOLDSEN

Danish Institute of Animal Science, Foulum, DK-8830 Tjele, Denmark

(Received 27 March 1995 and accepted for publication 26 July 1996)

S. Danish organic dairy production is characterized by a low input of

antibiotics for udder treatment and a high input of other mastitis control procedures.
A study was conducted in 14 organic dairy herds with the objectives of obtaining a
comprehensive description of clinical mastitis cases and identifying characteristic
patterns in these results. Clinical signs, inflammatory reactions and microbiological
identifications were obtained from 367 cases of clinical mastitis occurring over 18
months. Cow characteristics and preincident values such as milk yield and somatic
cell count were obtained for each cow. Signs of previous udder inflammation were
present in two-thirds of the clinical mastitis cases. Severe local inflammatory
reactions were found in 21 % of the cases and some indication of generalized signs
such as fever and reduced appetite were found in 35 % of the cases. Logistic
regression analyses were performed based on the results of an initial (exploratory)
multiple correspondence analysis. Coliform mastitis (6 % of the cases) was rarely
preceded by pathogen isolation or inflammatory reactions in the same quarter.
Coliform mastitis cases usually occurred in one quarter only. Escherichia coli
infections were typically (truly) acute cases. Bacteriologically negative mastitis
(20 % of the cases) showed strong similarities with clinical coliform mastitis.
Staphylococcus aureus cases (18 % of the cases) occurred most frequently in late
lactation or around drying-off. Prior isolation of Staph. aureus and slight decreases
in milk yield were two factors that interacted but both were strongly and positively
related to clinical Staph. aureus. Staph. aureus mastitis typically had a subclinical
debut, and increasing degrees and duration of inflammation decreased shedding of
this pathogen. Streptococcus dysgalactiae (9 % of the cases) mastitis was typically
persistent, virulent and manifest in periods of lower cow resistance. More patterns of
subclinical and clinical Str. uberis mastitis (23 % of the cases) seemed to be present.

In recent years animal welfare and the quality of animal products have attracted
more and more attention. These issues have also been a major motivation for the
establishment of certified organic farming in Denmark, for which official regulations
have been issued by the Ministry of Agriculture (Anon. 1994). In organic dairy herds
the following regulations are expected to influence health. Prophylactic medication
is prohibited and withdrawal times following application of veterinary drugs are
prolonged three times compared with conventional farms. Summer grazing is
mandatory for all animals except for the first weeks of life. Daily exercise is also
mandatory, and 85 % of the daily feed ration must be of organic origin (which in
practice often results in a high roughage : concentrate ratio). The primary aim of
these rules is to promote disease prevention and the welfare of the animals.

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 24 M. V  C. E
Preliminary studies showed that clinical mastitis was an important disease
complex in a group of organic dairy herds (Vaarst et al. 1991). It is likely that the
occurrence and the control of mastitis in organic dairy herds may differ from
conventional herds as a consequence of the legislation. The highly multifactorial
nature of mastitis is well recognized. This means that the (clinical) disease
manifestations and the pathogen pattern are inter-related with a number of cow,
herd and}or environmental factors in a complex fashion. Consequently, there is a
need to study and characterize mastitis manifestations to create a foundation for
mastitis control under the special conditions of organic dairy systems. Opportunistic
mastitis pathogens might, for instance, manifest themselves less aggressively owing
to the less intensive feeding regimens. The duration of infections may be longer owing
to restricted use of antibiotics.
Results of microbiological examinations have been applied in advisory work in
mastitis control for decades. Such data provide the dairyman as well as the
veterinarian with information for diagnosis, monitoring and prevention. Updated
knowledge about the microbiological status of the animals over time identifies the
most obvious and relevant options for mastitis control in a given herd (e.g. whether
the pathogens are related mainly to the environment or to milking routines). A study
that evaluates entire patterns of clinical mastitis processes involving data at cow and
herd level is a relevant option for understanding udder health perspectives in organic
dairy herds and, ultimately, suggesting efficient mastitis control procedures.
Multivariate statistical analysis techniques offer a relevant method to meet this
purpose. A study based on sampling of mastitic milk from new acute cases of clinical
mastitis, and knowledge about the infected animal with regard to somatic cell
counts, milk yield, previous treatments of mastitis cases and clinical manifestations
in the actual case, was conducted in 14 Danish organic herds. Based on an
exploratory statistical analysis which suggested several characteristic patterns in the
information, important relationships between risk factors and mastitis pathogens
were identified in multivariable logistic regression analyses. The results are expected
to identify and quantify those relationships that are important to take into account
in diagnostic work and practical mastitis control in dairy herds.

  

Herds and applied management policies
The study was conducted in 14 Danish organic dairy herds from November 1991
to April 1993. The herds were visited four times each year by one of the authors (a
veterinarian) and about twice monthly by a research technician. Herd sizes were
between 24 and 180 calvings per year. Housing systems were tie stalls, deep bedded
pack barns, and traditional free stall barns with slatted floors. The breeds were
Jersey or dual purpose. The rate of mastitis treatment conducted by veterinarians
varied between 0 and 0±23 cases per cow year (Vaarst et al. 1991 ; Vaarst &
Enevoldsen, 1994). All the farmers involved claimed to treat some cases of ‘ non-
severe ’ mastitis cases themselves. Non-severe was in most cases found to be cases
without systemic symptoms such as reduced feed intake, abnormal behaviour and}or
fever. As estimated from interviews, the incidence of such cases was 0±1–1 per year.
Data recording
Detailed records of mastitis and other disease control procedures were kept
during the study period. Dates of calvings, cullings, inseminations, pregnancy

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 25
Table 1. Time aspects related to 367 cases of clinical mastitis. Definitions of variables
and percentages of cases by variable category
Percentage
Variables and definitions of cases
Parity†
First lactation 35
Second 26
Older 39
Stage of lactation
Peripartum : 21 d before calving to 8 d after calving 10
Early lactation : 9–30 d after calving 21
Mid lactation : 31–210 d after calving 43
Late lactation}early dry period : 211 d after calving to 22 d 26
before calving}culling
Number of incidents recorded for the cow
First and only 68
First of two or more 14
Second or later 18
Time of week
Weekend 19
Monday 19
Other 62
Time of day for milk sampling at incident
During milking periods 70
Other day or night hours 20
Recording missing 10
† If the incident occurred from 21 to 0 d before calving, the subsequent calving was used.

checks, disease treatments and milk yield testings were recorded. Somatic cell counts
(SCC) and milk yields were available through the Danish national milk recording
scheme. All recordings were verified continuously by the technicians and the authors.
The clinical mastitis manifestations were described on multiple choice forms by
the farmer. Abnormal changes with regard to mastitis were defined as deviation from
normal for the cow in question at the actual point in time. A mastitis case was
regarded as new when abnormal changes were observed in milk, teat and}or udder
quarter of a cow. A case was also defined as new when the affected quarter(s) had
been declared recovered and had shown no signs of mastitis for 1 month.
Milk samples from the new clinical cases and biannually from all quarters in all
cows in a given herd were mailed to the Danish Dairy Board Laboratory and
analysed bacteriologically and cytologically (California mastitis test, CMT) according
to the procedures of the official Danish mastitis control scheme which is in
accordance with internationally accepted guidelines (Anon. 1992).
Statistical analysis
The study was a proportional morbidity study (Martin et al. 1987). This means
that this study population comprised diseased animals only. The variation being
studied was that between various kinds of mastitis manifestations. Records were
available only for cows that had experienced a new case of clinical mastitis. In
consequence, no exact estimates of mastitis incidence could be provided from data
in this study.
Univariate analysis
All variables were categorized. Continuous variables were categorized by taking
biological reasonable considerations into account.
An acutely elevated SCC was defined as a measurement that increased " 2 

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 26 M. V  C. E
Table 2. Clinical manifestations characterizing 367 cases of clinical mastitis.
Definitions of variables and percentages of cases by variable category
Percentage
Variable of cases
Quarter affected
Right front 18
Left front 19
Right hind 35
Left hind 28
Number of quarters affected at the incident
One 61
Two or more 39
Quarter abnormality preceding incident
Yes 11
No 89
Teat lesion or abnormality at incident
Yes 11
No 89
Degree of inflammation before incident
No information before incident 11
No sign of inflammation 32
Measurable signs of inflammation 19
Inflammation most likely 22
Clear signs of inflammation 16
Milk abnormalities at incident
None 10
Floccules only 53
Floccules in combination with watery milk 6
Purulent (with or without floccules) 12
Watery only 5
Other symptoms or combinations 14
Inflammatory symptoms in quarter tissues
None 37
Slightly swollen 37
Severely swollen 21
Other symptoms or combinations 5
Rectal temperature at incident
Missing 84
Normal (38±0–39±0 °C) 6
Subfebrile or febrile (& 39±1 °C) 10
General condition (appetite and general condition)
Not affected 65
Slightly affected 24
Severely affected 11

between one milk yield measurement and the next. The deviation of milk yield from
an estimate of the expected milk yield (adjusted for the effects of parity and stage
of lactation) at each test day was calculated as described by Enevoldsen et al. (1995).
The variable ‘ degree of inflammation before the incident ’ was created from
records of several inter-related variables, veterinary treatments of the case–cow prior
to the incident and three variables describing SCC in milk (measured monthly on cow
level at monthly milk yield testings and adjusted for the effects of milk yield, parity
and stage of lactation) and CMT measurements at cross sectional examinations
within 6 months before the incident.
Multivariate analysis
A multiple correspondence analysis (MCA) of all the variables of interest in
relation to udder health was performed by means of Proc. CORRESP in SAS (1990).
MCA is a weighted principal component analysis that identifies dimensions in the

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 27
information and where all variables are dependent variables. The correlations
between each variable category and the dimensions were interpreted like the
dimensions in factor analyses (Greenacre, 1993). The correlations do not directly
imply the direction and magnitude of the associations among the variables. In
traditional factor analysis (which requires continuous variables) a major question is
which number of factors to choose (Rummel, 1970). The same problem arises in a
factor analytic approach to MCA, but in this case there is much less experience on
which to base a decision. In the current study the number of dimensions was chosen
from a visual inspection of the rate of decline in the values of the principal inertias
for the dimensions in the data. This approach is analogous to ‘ scree plot analysis ’ in
factor analysis (Rummel, 1970).
Logistic regression analyses
A ‘ null model ’ was specified based on the results of the MCA. A null model is a
prior expectation (or hypothesis) of the relationships among a given mastitis
manifestation (say, Staphylococcus aureus isolation) and causes or risk factors (say,
prior mastitis). Null models were based on all variables included in the most relevant
dimensions of the correspondence analysis.
The logistic regression analyses examined all possible two-factor interactions.
The herd was also included in the model as a fixed effect to examine systematically a
possible herd effect. A logistic regression analysis was performed by SAS Proc.
CATMOD (SAS, 1985) for each of the following microbiological diagnoses : coliforms,
Staph. aureus, Streptococcus uberis, Str. dysgalactiae, and bacteriologically negative
clinical mastitis. Variables or combinations of variables (interaction terms) were
excluded in a stepwise backward fashion based on a 5 % (alpha) significance level. In
some cases, biological considerations were allowed to influence the decision to keep or
exclude a variable. The end results of the analyses were final models comprising those
variables (risk factors) that were significantly associated with each type of mastitis.
The final models were described in terms of odds ratios, 95 % confidence intervals,
and underlying frequency distributions from which the unadjusted odds ratios can
be calculated. For a dependent variable Y with values 1 and 0 and a single inde-
pendent variable X with values 1 and 0, the unadjusted odds ratio is calculated as
(a}(a­b))}(c}(c­d)) where a, b, c and d are the numbers of observations for the four
possible combinations Y ¯ 1 and X ¯ 1, Y ¯ 0 and X ¯ 1, Y ¯ 1 and X ¯ 0, and
Y ¯ 0 and X ¯ 0 respectively. In these analyses, odds ratios expressed the odds of
having mastitis involving a certain mastitis pathogen in the group disposed to the
risk factor in question compared with the odds of having mastitis involving all other
mastitis pathogens being disposed to the risk factor in question. The risk factor had
a significant impact on the outcome in the final model if the confidence interval did
not include the value 1.
A likelihood ratio statistic was calculated finally to indicate the goodness of fit of
the final model, that is, whether the variables and the corresponding odds ratios (the
parameters of the model) provided a satisfactory description of the information. A
high value indicated a good fit of the final model. It is generally accepted that a value
! 0±05 would be regarded as an unsatisfactory fit.


Univariate analysis
Records from 367 new clinical mastitis cases were made available. Univariate
descriptions of the edited variables are given in Tables 1–3. Table 1 contains

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 28 M. V  C. E
Table 3. Udder health variables expressed as percentages before and at acute cases of
mastitis among 367 quarter cases
Cross sectional analyses before
incident
At incident ;
Udder health variable Other quarters Same quarter same quarter
Microbiology
No sample 19 19 7
No growth 47 59 20
Contaminated — — 4
Staphylococcus aureus 13 11 18
Streptococcus dysgalactiae — 3 9
Str. uberis — 3 23
Coliforms — 1 6
Coagulase-negative staphylococci 11 4 —
Other pathogens 10 2 5
Multiple pathogens — — 8
Inflammatory reaction (California mastitis test)
None 50 7
Trace 9 4
High 17 59
Not analysed 5 24
6
7
24
No sample 8 7
Inflammatory reaction (SCC) Composite sample
Missing 16
Lower than expected 7
As expected 34
Slightly elevated 22
Clearly elevated 22
Acutely elevated SCC
Missing 16
Yes (SCC peak present during 180 d) 15
No (no peak SCC present during 180 d) 69
Milk yield
Missing 13
Higher 26
As expected 25
Slightly reduced 22
Markedly reduced 14
SCC, somatic cell count.

variables mainly related to time aspects of the case occurrence and sampling, Table
2 contains descriptions of clinical manifestations and Table 3 provides descriptions
of the bacteriological diagnoses and the inflammatory reactions (direct and indirect
(CMT) SCC), and describes udder health states before and at clinical mastitis.
Multivariate analysis
MCA revealed eight dimensions in the data. The estimates of ‘ correlations ’
between each variable category and each of the data dimensions were between one
and eight.
Using a biological and epidemiological interpretation of the variable categories in
each dimension, the eight dimensions (not shown) in the data were assigned the
following labels according to one or more dominant features in the group of variable
categories in each of the dimensions : ‘ sampling regimen ’ (dimension 1), ‘ time of
incident ’ (dimension 2), ‘ coli ’ (dimension 3), ‘ severe inflammation ’ (dimension 4),
‘ milk yield and lesion ’ (dimension 5), ‘ dysgalactiae ’ (dimension 6), ‘ aureus ’
(dimension 7), and ‘ uberis ’ (dimension 8). The same variable category was usually

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 29
Table 4. Risk factors of 23 new clinical mastitis cases involving coliforms (6 % of 367
cases) revealed in a logistic regression analysis. Associations between the variables
included in the final models and mastitis involving coliforms are described in terms of
odds ratios and 95 % confidence intervals. Results given in the Table provide a complete
bivariate description of each association stated in the final models and allow calculation
of (unadjusted) odds ratios
95 % Risk factor
Confidence plus coliforms
Odds ratio interval present}absent
Only one quarter affected 4±4 1±3–15±3 20}205
Pathogens isolated in same quarter before incident
No pathogens isolated 4±7 1±4–16±4 18}198
Coliforms† 2}0
Inflammatory reaction before incident
Most likely present 0±2 0±02–1±3 1}80
Likelihood ratio statistics 0±90
Variables included in initial model based on multiple correspondence analysis and found non-significant :
isolation of Staphylococcus aureus in same quarter before incident, no pathogens isolated from other quarters
before incident, cow related pathogens isolated in other quarters before incident, milk yield as expected, milk
yield slightly reduced, no teat lesions, no quarter abnormalities, inflammatory reaction, somatic cell count not
affected, right front quarter affected, right hind quarter affected.
† Difference (®2LL) when excluded from the model ¯ 12±0: P ! 0±001.

Table 5. Risk factors in 75 new clinical mastitis cases where no pathogens could be
isolated (bacteriologically negative mastitis in 20 % of cases) revealed in logistic regression
analysis. Associations between the variables included in the final models and
bacteriologically negative mastitis are described in terms of odds ratios and 95 %
confidence intervals. Results given in the Table provide a complete bivariate description
of each association stated in the final models and allow calculation of (unadjusted) odds
ratios
95 % Risk factor plus
Confidence no pathogens
Odds ratio interval positive}negative
No pathogens isolated before incident 3±3 1±8–6±0 59}157
Cow related pathogens isolated — — 0}18
in other quarters before incident†
Right front quarter involved 1±6 0±9–3±0 17}50
Milk yield markedly reduced before incident 0±3 0±1–0±9 5}48
Likelihood ratio statistics 0±84
Variables evaluated in initial model based on multiple correspondence analysis and found non-significant :
no inflammatory reaction before incident, inflammatory reaction probably present before incident,
Staphylococcus aureus isolated from same quarter before incident, coliforms isolated from quarter before incident,
no pathogens isolated from other quarters before incident, milk yield at expected level, one quarter involved
only, no teat lesions before incident, no quarter abnormalities before incident, right hind quarter affected.
† Difference (®2LL) when excluded from the model ¯ 6±17 : P ! 0±025.

related to more than one dimension. The results from MCA can aid in developing
simple measurement scales to describe mastitis cases.
Finally, values of squared cosines for each dimension for each herd were given.
These values indicated which dimensions were most relevant for each herd. Most
herds were not clearly related to any dimension, indicating that a broad spectrum of
mastitis types had occurred compared with the number of incidents in total. Some
herds, however, tended to be related to one of the dimensions, e.g. ‘ aureus ’ or
‘ uberis ’.

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 30 M. V  C. E
Table 6. Risk factors for 66 new clinical mastitis cases involving Staphylococcus aureus
(18 % of 367 cases) revealed in logistic regression analysis. Associations between the
variables included in the final models and mastitis involving Staph. aureus are described
in terms of odds ratios and 95 % confidence intervals. Results given in the Table provide
a complete bivariate description of each association stated in the final models and allow
calculation of (unadjusted) odds ratios
95 % Risk factor plus
Confidence Staph. aureus
Odds ratio interval present}absent
Pathogens isolated from same quarter before incident
Streptococcus uberis† — — 0}11
Str. dysgalactiae‡ — — 0}11
Milk yield slightly decreased ; Staph. aureus isolated
from same quarter before incident
Staph. aureus isolated and milk yield slightly 15±2 4±1–56±2 6}5
decreased
Staph. aureus not isolated and milk yield slightly 2±0 1±0–4±1 14}57
decreased
Staph. aureus isolated and milk yield not slightly 27±6 9±1–83±9 17}12
decreased
Staph. aureus not isolated and milk yield not slightly 1±0 — 29}227
decreased
Inflammatory reaction before incident
Inflammation obviously present 0±3 0±1–0±8 11}47
Right hind quarter affected 1±7 1±0–3±2 27}101
Late lactation or dry period until 21 d 3±3 1±8–6±1 28}67
before calving
Likelihood ratio statistics 0±98
Variables evaluated in initial model based on multiple correspondence analysis and found non-significant :
no inflammatory reaction in quarter before incident, inflammatory reaction suspected before incident,
inflammatory reaction probably present before incident, environmentally related pathogens isolated from same
quarter before incident, no pathogens isolated from same quarter before incident, no pathogens isolated from
other quarters before incident, environmentally related pathogens isolated from other quarters before incident,
coagulase-negative staphylococci isolated from same quarter before incident, coliforms isolated from same
quarter before incident, cow-related pathogens isolated from other quarters before incident, no quarter
abnormalities before incident, teat lesions before incident, Staph. aureus isolated from other quarters before
incident, milk yield at expected level, left front quarter affected.
† Difference (®2LL) when excluded from the model ¯ 3±03 : P ! 0±1.
‡ Difference (®2LL) when excluded from the model ¯ 4±09 : P ! 0±05.

Table 7. Risk factors for 33 new cases of clinical mastitis involving Streptococcus
dysgalactiae (9 % of 367 cases) revealed in logistic regression analysis. Associations
between the variables included in the final models and mastitis involving Strep.
dysgalactiae are described in terms of odds ratios and 95 % confidence intervals. Results
given in the Table provide a complete bivariate description of each association stated in
the final models and allow calculation of (unadjusted) odds ratios
95 % Risk factor plus
Confidence Strep. dysgalactiae
Odds ratio interval present}absent
Pathogens isolated in same quarter before incident
Str. dysgalactiae 75±3 14±9–379±7 9}2
Periparturient period 3±3 1±1–8±1 7}31
Quarter abnormalities before incident 2±4 0±9–6±9 6}34
Likelihood ratio statistics 0±99
Variables evaluated in initial model based on results from multiple correspondence analysis and found non-
significant : no pathogens or cow related pathogens isolated from other quarter before incident, milk yield at
expected level, milk yield slightly reduced, milk yield markedly reduced, one quarter mastitic, two or more
quarters mastitic, case on a Monday, case during a weekend, no teat lesion before incident, right front quarter
affected, right hind quarter affected, inflammatory reaction obviously present.

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 31
Table 8. Risk factors for 84 new clinical mastitis cases involving Streptococcus uberis
(23 % of cases). Associations between the variables included in the final models and
mastitis involving Str. uberis are described in terms of odds ratios and 95 % confidence
intervals. Results given in the Table provide a complete bivariate description of each
association stated in the final models and allow calculation of (unadjusted) odds ratios
95 % Risk factor plus
Confidence Str. uberis
Odds ratio interval present}absent
Pathogens isolated in same quarter before incident
Coagulase-negative staphylococci 10±8 3±3–34±9 8}7
Str. uberis 25±1 5±8–107±5 8}3
No pathogens 3±3 1±7–6±4 56}160
Likelihood ratio statistics 0±49
Variables evaluated in initial model based on results from multiple correspondence analysis and found non-
significant : environmentally related pathogens in other quarters before incident, milk yield slightly decreased,
teat lesion.

Logistic regression analyses

The final models of the occurrence of Staph. aureus, Str. dysgalactiae, Str. uberis,
coliforms and cases with no pathogen isolation in acute cases of mastitis are
presented in Tables 4–8. Odds ratios and 95 % confidence intervals are given,
together with the number of mastitis cases in which the risk factor was present. These
figures allow the calculation of unadjusted odds ratios. Likelihood ratio statistic
values are also presented in the Tables. They indicated a good fit of the final models.


This study vividly illustrates the complexity of mastitis manifestations and the
difficulties of studying the dynamics of the disease in a changing population like a
dairy herd. For example, in some cases no information was available prior to the
clinical mastitis case because the acute case occurred early in the first lactation
before any cross sectional sample was available. Despite such difficulties, a
comprehensive description of a relatively large number of acute cases of clinical
mastitis was obtained in this study. Udder health values prior to the acute case in
particular are given in relatively full detail.
In nine herds virtually all cases were sampled, while less than half appeared to be
sampled in five herds. This was probably due to inconsistent recording routines
among some farmers.
Definitions of specific abnormalities in gland, teat and milk appeared to be
relatively homogeneous between herds.
The microbiological quality of the milk sampling regimen appeared to be very
satisfactory as indicated by the low frequency of contaminated samples. Indirect
SCC was not measured in 24 % of the samples because highly abnormal milk could
not meaningfully be subjected to CMT.
Repeat cases in the same udder constituted 18 % of the cases. These could be true
new infections, but chronic infection could also be suspected in such cases.
The main clinical symptoms were abnormalities of the milk, in most cases
floccules. Severe symptoms were relatively rare. The majority of samples were taken
in connection with milking. This seems logical, because the physical contact with the
udder during milking will obviously increase the likelihood of revealing new cases of

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 32 M. V  C. E
clinical mastitis. The recognition of non-severe cases of clinical mastitis between
milkings is difficult in practice. Time of week was included as a possible risk factor.
In most herds, milking during the weekends was carried out by an individual other
than the one milking from Mondays to Fridays. Furthermore, only veterinarians are
allowed to treat with antibiotics in Denmark, and during weekends the cost of a
veterinary call is twice the normal price. These two factors could influence milking
routines, observation intensity and treatment pattern during the weekends.
The predominant pathogens in the cross sectional examinations in the organic
herds as well as the cows included in this study were clearly cow-related pathogens
such as Staph. aureus, coagulase-negative staphylococci (CNS) and Str. dysgalactiae
(Table 3). In a cross sectional study conducted in a given herd, all animals are
sampled at one time. Infections of long duration will most likely dominate the
bacteriological pattern found in such biannual studies compared with infections of
short duration.
CNS, Staph. aureus and Str. dysgalactiae are all pathogens considered capable of
colonizing skin, wounds and the streak canal. The question whether all bacterial
cultures isolated in cross sectional sampling are truly intramammary infections must
remain open.
For decades, the protective effect of CNS with regard to colonization by
organisms causing mastitis has been discussed. Schukken and co-workers (1989)
found Escherichia coli to be the predominant microorganism isolated in herds
with low bulk milk SCC. Regarding bulk milk SCC as an indicator of infection status
in the herd (Reneau, 1986), the findings of Schukken and co-workers indicate that
cow-related mastitis pathogens ‘ protect against ’ coliform infections. The results of
Schukken and co-workers were, however, based on herd level analyses, which cannot
be interpreted unambiguously at the single cow level. Another recent study
(Matthews et al. 1991) similarly suggests that quarters harbouring a CNS infection
suppress colonization of the mammary gland by other mastitis-causing pathogens.
Timms & Schultz (1987) found CNS to be the predominant pathogen involved in
subclinical infections of long duration, with elevated SCC and significant reduction
in milk yield.
CNS were not isolated from any clinical cases. Aungier & Austin (1987) isolated
CNS in pure culture in 11 % of clinical mastitis cases, and they were also found in
association with other pathogens. In this study, the proportion of cases involving
cow-related pathogens (Staph. aureus, Str. dysgalactiae) and the proportion of cases
involving environmentally related microorganisms (Str. uberis, coliforms) were of
almost equal size. The distribution of pathogens from the clinical cases in the herds
consequently seemed to be different from the distribution of pathogens found in a
cross sectional study in the same herds.
A combined variable, ‘ inflammation ’, was based on more inter-related measures
in order to characterize signs of inflammation prior to the clinical case. One reason
for doing so was to concentrate inter-related information describing the same aspects
with regard to mastitis. Furthermore, not all information was available for all
animals. Cross sectional samples had not been sampled from cows in first parity,
early lactation. Other sources of information could consequently be used – although
with a lower degree of confidence – and partly cover the same aspect. More critical
points could however be raised, such as the fact that SCC and treatments before the
incident were measured at cow level, while CMT results from previous cross sectional
studies in the herd were measured on quarter level.
The level of the inflammatory reaction and the milk yield levels before and after

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 33
the clinical incidents showed a tendency for an increased frequency of inflammatory
reactions from preincident to post-incident examinations.
This analysis would not reveal an association between a risk factor and a given
mastitis type if the risk factor were related to other mastitis types as well. The odds
are calculated as the odds of the specific mastitis type out of other types of mastitis.
This usually means that only risk factors specifically related to one mastitis type
were identified. Polytomous logistic regression (Hosmer & Lemeshow, 1989) might
have given further insight into the differentiation between all the mastitis types
instead of comparing one type with all the others. Causal probabilistic nets
(Højsgaard & Thiesson, 1995) offer another option for constructing these very
complex models.
Several other risk factors would be relevant to consider in these analyses. A large
number of herd-related risk factors (e.g. calving routines, calves suckling after
calving and bedding material) must be expected to influence the development of each
mastitis case. Several studies have discussed the influence of mastitis incidence and
pathogen pattern caused by managerial factors. Bartlett et al. (1992) found, for
example, that an increased prevalence of coliform infection among others was
associated with an increased amount of milk remaining in the udder after milking.
Such information is highly relevant but would demand more detailed recordings than
those made in this study. Management factors were partly taken into account by
including the risk factor ‘ herd ’ in the regression analyses. The management routines
within each herd were, however, not characterized in detail. The number of
observations and risk factors already included did not encourage us to bring more
complex relationships into the analyses.

Coliform mastitis
Coliform mastitis cases (6 %) were rarely preceded by pathogen isolation or
inflammatory reactions in the same quarter. Coliforms were isolated from very few
quarters (2 of 367) in cross sectional analyses before clinical cases of mastitis. In both
these quarters coliforms were involved in a clinical mastitis incident within 6 months.
It is not possible from the actual records to draw any conclusion as to whether this
is due to chronic coliform infections or reinfection. Hill & Shears (1979) discussed the
phenomenon of recurrent coliform mastitis in the dairy cow, having concluded that
5 of 28 experimentally induced infections resulted in a persistent and recurring
condition. In most studies, coliforms are regarded as non-persistent microorganisms.
Hill et al. (1984) described experimental and (two) natural infections involving Esch.
coli with a duration ranging from 18 h to 13 d.
Coliform mastitis cases usually occurred in one quarter only. This suggests
coliforms to be mastitis pathogens causing a localized clinical outbreak without
further dispersal to other quarters.
Coliform mastitis is described both as a mild and as a very acute progressive type
of mastitis. Kremer et al. (1990) regarded coliform bacteria as opportunistic
microorganisms, causing serious cases of mastitis only in cases with incompetent
immune function. Burvenich et al. (1994) discussed the role of the neutrophil
leucocytes in reactions during experimentally induced Esch. coli mastitis and
concluded that the physiological qualities of neutrophils in the blood largely
determine the severity of the case in combination with resorption of toxins from the
inflamed quarters. However, in experimentally induced infections the conditions
might be different because the microorganism is forced to cross barriers which it may
not have been able to cross in the same cow if naturally exposed.

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 34 M. V  C. E
Bacteriologically negative clinical mastitis
In a relatively large part of the samples no pathogens were isolated. Because
sampling was probably conducted at an early phase of the development of the case,
a high frequency of pathogen-free samples could be regarded as quite surprising. The
high MCA correlations between the bacteriologically negative category and both the
‘ coli ’ and the ‘ severe inflammation ’ dimensions also indicated some similarity
between coliform mastitis and bacteriologically negative mastitis. An explanation
might be that the coliforms had been eliminated at the time of sampling. In a study
conducted in 201 herds from The Netherlands, where sampling was also carried out
by the herdsman, 28 % of the samples were bacteriologically negative (Miltenburg et
al. 1994). In a study in Ireland, no pathogens were isolated from 39 % of the samples
(Aungier & Austin, 1987). Between 18 and 38 % of the milk samples were culture-
negative in studies cited by Zorah et al. (1993). In their own study, bacterial antigens
were detected by enzyme-linked immunosorbent assay (ELISA) tests in 67±7 % of
culture-negative cases. Hicks et al. (1994) concluded from a study that results from
bacterial cultures in consecutive samples were more consistent than results of ELISA
performed on consecutive samples, suggesting among other possibilities cross
reactions with other antibodies in the milk. In this study, only the diagnosis of Staph.
aureus was discussed.
Staph. aureus is reported to be a mastitis pathogen often involved in chronic
cases, persistent in udder tissue, with an intermittent shedding pattern (i.e. low
sensitivity to diagnostic tests). Because of the pattern of behaviour, some ‘ pathogen-
free ’ mastitis might be Staph. aureus cases that were not detected by the
microbiological test applied. However, in this study the diagnosed Staph. aureus
mastitis cases were usually preceded by either pathogen detection or signs of
inflammation. This suggests that the bacteriologically negative mastitis may not
have involved Staph. aureus.
No biologically plausible explanation can be given for the more frequent
occurrence of bacteriologically negative cases in the right front quarter only.

Staph. aureus mastitis

Staph. aureus cases (18 %) occurred most frequently in late lactation or around
drying-off (42 % of the cases). This could be due to reduced milk yield, environmental
changes related to drying-off and the shift from two daily milkings to no milkings.
Clinical Staph. aureus mastitis never occurred when Str. uberis or Str. dysgalactiae
had been isolated prior to an incident and occurred less frequently among quarters
showing strong inflammatory reactions. Prior isolation of Staph. aureus and slight
decreases in milk yield were two factors that interacted but both were strongly and
positively related to clinical Staph. aureus. The findings showed that Staph. aureus
mastitis typically had a subclinical debut. Increasing degrees and duration of
inflammation apparently decreased occurrence or shedding of this pathogen, as was
also found in a study of subclinical Staph. aureus mastitis (Enevoldsen et al. 1995).
No biologically plausible explanation can be given for the more frequent occurrence
of Staph. aureus cases in the right hind quarter only. Aungier & Austin (1987) found
that significantly more mastitis cases were seen in the hind quarters, especially the
right hind quarter. Further investigations are required in order to confirm such
findings and provide rational explanations.

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 35
Str. dysgalactiae mastitis
Str. dysgalactiae cases represented 9 % of the clinical mastitis. Other studies of
clinical mastitis in conventional Danish herds have found a considerably higher
proportion of Str. dysgalactiae cases. Clinical mastitis involving Str. dysgalactiae
occurred in particular among cows with preceding isolation of Str. dysgalactiae. Dodd
(1983) and Harmon & Crist (1994) regard Str. dysgalactiae as a pathogen of
environmental origin and, consequently, as being impossible to eradicate from a
herd. In a review, King (1981) discussed the onset of different streptococcal
infections and concluded that Str. dysgalactiae seemed to produce clinical signs in
relatively many cases, whereas Str. uberis, for example, appeared to have a more
subclinical debut. However, if this pathogen usually has a subclinical debut, then it
should be possible to prevent the occurrence of clinical cases, because Str. dysgalactiae
is highly sensitive to antibiotics (Craven, 1987). In organic herds, the prophylactic
use of antibiotics is prohibited, but a bacteriological diagnosis will justify treatment.
The high risk related to the periparturient period might be explained largely by
the environmental and physiological changes around calving and in milking routines,
which might reduce cow resistance. In the organic systems a gentle initiation of
lactation is emphasized, and a clearly lower milk yield in early lactation has been
demonstrated (Kristensen et al. 1994). This should support the resistance of the cow
in early lactation and, consequently, lower the risk. However, the proportion of
mastitis cases involving Str. dysgalactiae in this study actually seemed to be much
lower than other reports indicate. This might be explained by the feeding routines in
early lactation in the organic system, but studies focusing specifically on this
relationship would be needed to draw any such conclusion. As with mastitis
involving Staph. aureus there was no association with teat trauma, which in the light
of results from other studies seems surprising.
Quarter abnormalities before the incident were positively related to occurrence of
mastitis cases involving Str. dysgalactiae. This suggests that Str. dysgalactiae causes
severe damage to the udder tissue. Str. dysgalactiae mastitis was indicated to be
persistent and virulent, and manifest in periods of lower cow resistance.
Almeida & Oliver (1995) found that Str. dysgalactiae seemed to be able to invade
epithelial cells, induce cell damage and persist inside bovine mammary epithelial
cells.

Str. uberis mastitis

Str. uberis occurrence (23 %) was strongly and positively related to prior isolation
of Str. uberis and CNS, but also positively related to quarters showing no pathogen
growth prior to the incident. This indicated more than one ‘ typical manifestation ’
of clinical Str. uberis mastitis : one with sudden onset (truly acute) and one with
subclinical debut. The farmers had been asked to sample and describe new clinical
cases of mastitis, so these must consequently be regarded as cases of which the farmer
had been unaware before the clinical condition.
Todhunter et al. (1994) regarded Str. uberis as an environmentally related
microorganism. Leigh (1994) described Str. uberis as a successful, specialized,
opportunistic pathogen of the mammary gland. King (1981) concluded from a review
of several research papers that Str. uberis seems to be a major cause of clinical
mastitis in the early dry period and prior to calving, indicating an increased
susceptibility, but the identification of the primary reservoir and the mode of
transmission of Str. uberis remain unsolved.

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 36 M. V  C. E
CNS were isolated before C 10 % of clinical cases caused by Str. uberis. CNS are
common skin inhabitants, and intramammary or teat duct colonization could be
indicators of risk factors related to milking procedures or hygiene. In a study
conducted by Todhunter et al. (1994), teat injuries were observed in 17 % of
infections (clinical or subclinical) with Str. uberis. All these cases were clinical.
Robinson et al. (1985) found that a large number of Str. uberis could be isolated from
the teat end (13 % of cases). Machine milking has been found to cause teat end
damage, inducing colonization by environmental pathogens (Zecconi et al. 1992).
Such teat end damage could explain the relation between CNS and subsequent Str.
uberis infection. As mentioned before, CNS has been claimed in several studies to
have a preventive effect for colonization by other mastitis pathogens, but the relation
between these two pathogens may be artificial and both should perhaps be regarded
simply as pathogens partly related to teat end damage.
The pattern of mastitis cases involving Str. uberis does allow definite conclusions
to be drawn, but leaves a number of open questions, some of which only further
studies will clarify.

We thank Anna Huda of The Danish Dairy Board Laboratory for valuable
comments and discussions on microbiological diagnoses and their interpretation
during the project period.

A, R. A. & O, S. P. 1995 Invasion of bovine mammary epithelial cells by Streptococcus
dysgalactiae. Journal of Dairy Science 78 1310–1317
A. 1992 [Ministry of Agriculture guide lines of 15 February 1992 concerning mastitis control and relevant
laboratory analyses.] Copenhagen : Danish Ministry of Agriculture
A. 1994 [Ministry of Agriculture Executive Order no. 720 of 19 August 1992.] Copenhagen : Danish
Ministry of Agriculture
A, S. P. M. & A, F. H. 1987 An investigation into the nature of clinical mastitis in some large
dairy herds in Ireland. Irish Veterinary Journal 41 304–310
B, P. C., M, G. Y., L, S. E. & H, L. E. 1992 Managerial determinants of
intramammary coliform and environmental streptococci infections in Ohio dairy herds. Journal of Dairy
Science 75 1241–1252
B, C., P, M. J., H, A. W., G, A. J., M, R. H., H, R., K, W. D. J.
& B, A. 1994 Role of the neutrophil leucocyte in the local and systemic reactions during
experimentally induced E. coli mastitis in cows immediately after calving. Veterinary Quarterly 16 45–50
C, N. 1987 Efficacy and financial value of antibiotic treatment of bovine clinical mastitis during
lactation – a review. British Veterinary Journal 143 410–422
D, F. H. 1983 Symposium. Advances in understanding mastitis. Mastitis – progress in control. Journal of
Dairy Science 66 1773–1780
E, C., G$ , Y. T. & T, I. 1995 Dairy cow characteristics related to Staphylococcus aureus
isolation from quarter samples. Journal of Dairy Research 62 69–81
G, M. J. 1993 Correspondence Analysis in Practice. London : Academic Press
H, R. J. & C, W. L. 1994 Environmental mastitis in lactating and dry cows and prepartum
heifers. Proceedings, Annual Meeting, National Mastitis Council 33 241–249
H, C. R., E, R. J. & S, W. M. 1994 Comparison of microbiologic culture, an enzyme-linked
immunosorbent assay, and determination of somatic cell count for diagnosing Staphylococcus aureus mastitis
in dairy cows. Journal of the American Veterinary Medical Association 204 255–260
H, A. W., F, A. J. & B, B. E. 1984 Progressive pathology of severe Escherichia coli mastitis in
dairy cows. Research in Veterinary Science 37 179–187
H, A. W. & S, A. L. 1979 Recurrent coliform mastitis in the dairy cow. Veterinary Record 105 299–301
H, S. & T, B. 1995 BIFROST – Block recursive models induced from relevant knowledge,
observations, and statistical techniques. Computational Statistics and Data Analysis 19 155–175
H, D. W. & L, S. 1989 Applied Logistic Regression. New York : Wiley
K, J. S. 1981 Streptococcus uberis : a review of its role as a causative organism of bovine mastitis. I.
Characteristics of the organisms. British Veterinary Journal 137 36–52
K, W. D. J., N-S, E. N. & L, J. A. C. M. 1990 Host defence and bovine
coliform mastitis. Veterinary Quarterly 12 103–113
K, T., V, M. & K, E. S. 1994 Analysis of the production in Danish organic and
conventional dairy herds. Paper presented at 10th IFOAM Conference, New Zealand

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press

 Mastitis in Danish organic dairy herds 37
L, J. A. 1994 Progress in the understanding of the pathogenesis of Streptococcus uberis for the lactating
bovine mammary gland. Proceedings, Annual Meeting, National Mastitis Council 33 96–103
M, S. W., M, A. H. & W, P. 1987 Veterinary Epidemiology. Principles and Methods. Ames,
IA : Iowa State University Press
M, K. R., H, R. J. & L, B. E. 1991 Effect of naturally occuring coagulase-negative
staphylococci infections on new infections by mastitis pathogens in the bovine. Journal of Dairy Science 74
1855–1859
M, J. D.,  L, D., C, A. P. P., B, J. H. & E, A. R. W. 1994 Estimating
the incidence of clinical mastitis in dairy herds in the southern Netherlands. Proceedings, Annual Meeting,
National Mastitis Council 33 357–359
R, J. K. 1986 Effective use of dairy herd improvement somatic cell counts in mastitis control. Journal
of Dairy Science 69 1708–1720
R, T. C., J, E. R. & M, A. 1985 Factors involved in the epidemiology and control of
Streptococcus uberis and coliform mastitis. British Veterinary Journal 141 635–642
R, R. J. 1970 Applied Factor Analysis. Evanston, WY : Northwestern University Press
SAS 1985 SAS User’s Guide : Statistics, 5th edn. Cary, NC : SAS Institute
SAS 1990 SAS User’s Guide : Statistics, 4th edn. Cary, NC : SAS Institute
S, Y. H., G, F. J.,   G, D. & B, A. 1989 Incidence of clinical mastitis on
farms with low somatic cell counts in bulk milk. Veterinary Record 125 60–63
T, L. L. & S, L. H. 1987 Dynamics and significance of coagulase-negative staphylococcal
intramammary infections. Journal of Dairy Science 70 2648–2657
T, D. A., S, K. I. & H, J. S. 1994 Environmental streptococcal mastitis. Proceedings,
Annual Meeting, National Mastitis Council 33 92–95
V, M., B, K., E, C. & K, E. S. 1991 [Health, disease control and reproduction
in organic dairy farms.] National Institute of Animal Science Report no. 699 130–187
V, M. & E, C. 1994 Disease control and health in Danish organic dairy herds. In Biological
Basis of Sustainable Animal Production, pp. 211–217 (Eds E. A. Huisman, J. W. M. Osse, D. van der Heide,
S. Tamminga, B. J. Tolkamp, W. G. Schouten, C. E. Hollingworth and G. L. van der Winkel). (EAAP
Publication no. 67)
Z, A., H, J., B, V. & R, G. 1992 Machine-induced teat tissue reactions and infection
risk in a dairy herd free from contagious mastitis pathogens. Journal of Dairy Research 59 265–271
Z, K. T., D, R. C. W. & F, A. J. 1993 Detection of bacterial antigens in milk samples from
clinical cases of bovine mastitis in which culture is negative. Veterinary Record 132 208–210

https://doi.org/10.1017/S002202999600194X Published online by Cambridge University Press