JDS
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Klaas, I. C.
Right arrow Articles by Houe, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Klaas, I. C.
Right arrow Articles by Houe, H.
J. Dairy Sci. 87:1217-1228
© American Dairy Science Association, 2004.

Systematic Clinical Examinations for Identification of Latent Udder Health Types in Danish Dairy Herds

I. C. Klaas1, C. Enevoldsen2, M. Vaarst1 and H. Houe3

1 Danish Institute of Agricultural Sciences, Department of Animal Health and Welfare, DK-8830 Tjele, Denmark
2 The Royal Veterinary and Agricultural University, Department of Clinical Studies, Large Animal Medicine, 1870 Frederiksberg C, Denmark
3 The Royal Veterinary and Agricultural University, Department of Animal Science and Animal Health, 1870 Frederiksberg C, Denmark

Corresponding author: I. C. Klaas; E-mail: ilka.klaas{at}agrsci.dk.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
A cross-sectional study was conducted to explore the applicability of systematic clinical examinations of udders as an additional tool for the evaluation of udder health status on dairy farms. During 2000, each of the 16 dairy farms was visited 5 times; 20 cows per farm were chosen at random at each visit for clinical udder examination immediately after milking. The clinical examination included both pathological and morphological variables. One examination per cow was included in the analysis (n = 707 cows). Principal component analysis (PCA) was performed in 3 steps. First, 19 variables characterizing udder and teats were analyzed (PCA 1). Second, the variables parity and stage of lactation were included (PCA 2). Finally, somatic cell count (SCC) and milk yield (PCA 3) were included. The PCA resulted in 4 components that explained 30% of the variation of the data: 1) small udder, 2) distressed udder, 3) mastitis udder, and 4) soiled udder. Variables with high positive correlation to the "small udder" were small udder shape, short teats, and first parity. Impaired teat surface, hard udder texture, and a long udder shape were related to the "distressed udder." The "mastitis udder" was characterized by the clinical variables asymmetry between front quarters, asymmetry between hind quarters, knotty tissue, and acute clinical mastitis. Reduced milk yield and high SCC were related to the "mastitis udder," whereas low SCC was related to the "small udder." The "soiled udder" was related to early lactation. Including this information in the assessment of udder health may be of substantial value for data analysis in farms with suspected under-reporting of clinical mastitis.

Key Words: udder health • principal component analysis • clinical investigation • udder characteristics

Abbreviation key: ECM = energy-corrected milk, MLOGSCC = mean of log-transformed SCC, MRESMILK = mean residual of observed-expected milk yield, PCA = principal component analysis


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Udder health is characterized by many different terms, such as reported cases of clinical mastitis, SCC on quarter or cow level, and occurrence of pathogens isolated from infected quarters (IDF, 1997). Field studies quantifying the occurrence of treated cases of mastitis may be biased because of factors related to farm management. For example, the farmer’s threshold relevant to when treatment should be initiated or the veterinarian should be consulted influences the incidence rate of farmer-reported cases of clinical mastitis. A farmer’s decision to treat a cow for mastitis may be determined by factors such as 1) the seriousness of symptoms, 2) characteristics of the cow, 3) infection level of the herd, and 4) preference of alternative treatments (Vaarst et al., 2002). How these factors are weighted differs from farmer to farmer. One effect is the under-reporting of mastitis in many farms; strong evidence for under-reporting of clinical mastitis has been identified in national cattle databases (Bartlett et al., 2001). Farm management can also influence test day SCC if cows with clinical or chronic mastitis are excluded from milk recordings. Quarter milk samples tested for SCC and mastitis pathogens are regarded as the most precise practice in evaluating udder health (Hamann, 2003). However, because pathogens such as Staphylococcus aureus are shed intermittently with milk (Sears et al., 1990), a very costly frequent quarter sampling is needed.

Consequently, the development of more practical, precise, and cheaper methods to assess udder health on farms is needed. Systematic clinical examination of udder and teat characteristics is relatively inexpensive and may provide valuable information. Collection of such data is quite feasible, especially in herds already applying a herd health management system including other clinical recordings, e.g., body condition scoring or other paraclinical data (Markusfeld et al., 1997, Enevoldsen et al., 2000). In such herds, the marginal costs related to systematic udder examinations are low.

The overall objective of the present study was to evaluate the applicability of systematic clinical udder assessments of individual cows as a supplement to SCC and udder treatment records. The specific objectives of the study were 1) to identify types of udders of cows by using biologically meaningful combinations of systematic clinical assessments of individual cows during cross-sectional examinations and 2) to estimate the relationship between udder traits and milk yield, as a production parameter, and SCC, as a marker of inflammation.

To identify the major udder types, principal component analysis (PCA) was used to reduce the number of observed variables to a limited number of artificial variables, the "principal components."


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Herds
Sixteen dairy farms from Southern Jutland in Denmark were selected as a convenience sample from herds enrolled in the "Kongeå Projekt" (Andersen et al., 2000). The farms were selected to represent a broad spectrum of Danish milk production units and management levels and have previously been described (Vaarst et al., 2002). The farms were visited 5 times from January to December 2000. Table 1Go gives an overview of herd characteristics.


View this table:
[in this window]
[in a new window]
  		Table 1. Herd characteristics of 16 dairy herds selected for examination of udder health.
 
Clinical Examinations
Two veterinarians and 2 experienced and specially trained research technicians performed the clinical examinations. After the first visits to each farm, the results were discussed to increase inter- and intra-observer agreement, and the recording sheet was adjusted. The research technicians performed the clinical examinations during the subsequent farm visits. The cows were examined within 2 h after milking. During every herd visit, 20 cows were chosen at random for clinical examination from the milking herd on that day. The clinical examination was conducted based on the method described in Rosenberger (1979) and further developed by Houe et al. (2002). In addition to udder and teat characteristics, the degree of soiling with manure of thigh, udder, and teats was recorded (Table 2Go). All quarters of a cow were examined. If the degree of a condition differed between quarters within a cow, the most critical condition was recorded (e.g., if poor skin quality was observed on one teat, this was the recorded result from this cow).


View this table:
[in this window]
[in a new window]
  		Table 2. Clinical and morphological variables and their definitions selected for inclusion in principal component analysis (n = 707 cows). Definitions were based on Rosenberger (1979) and Houe et al. (2002). All quarters of a cow were examined, and in cases of different findings, the quarter with the worst condition was recorded.
 
A total of 1578 records of 1018 cows were collected. The number of examinations per cow ranged from 1 to 5. Because of the adjustments that were made in the recording scheme after the first farm visit, records of the first farm visit were omitted from further analysis.

Statistical Analysis
Principal component analysis was performed using the factor procedure of SAS (Proc factor, SAS, 1999). The number of components to retain was determined by evaluating the eigenvalue-one criterion, the scree test, the proportion of variance accounted for, and the interpretability criteria (Hatcher, 1994). Because each observed variable contributes one unit of variance to the total variance in the data set, a component displaying eigenvalues >1 accounts for a greater amount of variance than one variable alone and is worthy to be retained. The scree test displayed 4 components with the largest eigenvalues before a visible ‘break.’ Further, a component was regarded as important and kept for further analysis if at least 3 variables had high loadings on that component and shared the same conceptual meaning. A factor loading greater than ± 0.35 was considered important.

A loading represents the correlation of a given variable with the underlying factor. Loadings range from –1 to +1. Principal component analysis attempts to identify a minimum number of components explaining the majority of the variation in the original data. The first component extracted in a PCA accounts for a maximal amount of total variance in the observed variables (Hatcher, 1994). Typically, the second component will be correlated with some of the observed variables that did not display strong correlations with the first component. Further, the second component is not correlated with the first component. The successive components identified explain progressively smaller portions of the total sample variance. Linear combinations of observed variables are formed. In contrast to common factor analysis, in PCA, no assumptions about the underlying common parts (communalities) and unique parts of the variables are made (Rummel, 1970).

Because ordinary PCA does not account for the effect of the individual cow, only the last observation per cow was included in the analysis (n = 707). Therefore, only clinical examinations conducted by the 2 technicians were analyzed. Principal component analysis was carried out with categorical and continuous variables (Table 2Go). Udder shape, teat shape, parity, and stage of lactation were treated as dummy variables with normal udder and teat shape, second parity, and middle part of lactation as reference level. In cases of these dichotomous variables, the loadings on a factor were interpreted as the probability of its presence (positive loading) or absence (negative loading), given that the factor exists (Rummel, 1970).

Three different PCA were performed. First (PCA 1), only clinical variables were included in the analysis to simulate the situation of the practitioner on farm with no further information about the cows. Second (PCA 2), the variables parity and stage of lactation were added as dichotomous variables. Third (PCA 3), milk yield and previous SCC were added. The non-orthogonal promax rotation with a power of 3 was used to enhance the interpretability of the factors.

Milk Yield
Milk yield from each test day of all cows in the herds during 1999 that had at least 9 DIM was transformed to energy-corrected milk (ECM) using the following formula (Anonymous, 2002):


([1])

When analyzing ECM from test day records to model the lactation curve, the change in slope was expected to occur around 60 d after calving (Enevoldsen et al., 2000). A standard lactation milk curve of the current lactation was estimated for each cow using a piecewise linear regression model (Proc mixed; SAS, 2000) as described by Enevoldsen et al. (2000). The peak of the lactation curve was set to d 60 after calving. It was the intercept of the model and was set to 0 (DIM 60):


([2])

The second variable (DIMB60) was defined, which has the value 1 if DIM is below 60, otherwise 0. This variable models the slope from calving until 60 DIM:


([3])

The model accounted for the effect of parity and stage of lactation and was


([4])

where

Yijk=estimated ECM

µ=expected mean at 60 DIM among third parity cows

parityi=fixed effect of parity (i = 1, 2, and 3+)

DIM 60(parity)j=fixed effect of DIM within parity, and

DIMB60(parity)k=fixed effect of ‘before or after d 60’ within parity.

The intercept is an estimate of the peak milk production at d 60 after calving and was 32.4 kg for cows in the third parity group (Table 3Go). The estimate of DIM 60(parity) is the estimated difference in milk production between exactly 60 DIM and 305 DIM within the different parity groups. Cows in first lactation had the lowest difference of –4.1 kg ECM between DIM 60 and DIM 305, indicating the highest persistency. The estimated difference in milk yield between the day of calving and 60 DIM in second parity cows was not significant, indicating a falling line from calving until d 305 instead of a peak at 60 DIM, which is common in models of milk yield where milk is not adjusted to ECM.


View this table:
[in this window]
[in a new window]
  		Table 3. Parameter estimates with standard error and significance level from the piecewise linear regression model estimating energy-corrected milk (ECM) yield within parity and stage of lactation across herds.
 
The normally distributed residuals of the model represent the parity-specific deviation of the expected milk yield from the measured milk at any point during lactation according to the lactation curve of all cows in a specific parity. The mean residual milk yield of the complete lactation (observed—expected) for each cow was used as a continuous variable, MRESMILK, in the PCA 3. It characterizes the cow as yielding higher, lower, or as expected in comparison with all cows in that parity across herds. The mean residuals of the cows included in the PCA ranged from –15.0 to 16.1 kg ECM below or above the expected ECM; the mean was 0.50 ± 4.3 kg.

SCC
The previous SCC was the log-transformed (log10) SCC of all test days of the examined cows from calving until the day of their clinical examination. The mean gives information about the infection status in a cow during lactation and was included as MLOGSCC in the third step, PCA 3.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Descriptive statistics of the clinical examination are given in Table 4Go. Normal udder shape was recorded in 66.4% of cows. Nine percent of cows had small udders, 13.6% had deep udders, and 10.8% of cows had other udder shapes. Ninety-two percent of teats were of normal shape, 3.1% were short, 1.8% were fleshy, and 3.4% had other teat shapes.


View this table:
[in this window]
[in a new window]
  		Table 4. Descriptive statistics of udder and teat characteristics of cows analyzed in the principal component analysis (n = 707 cows).
 
In all 3 PCA, 4 components were extracted that explained 27% of the variation in the data. The rotated factor patterns of PCA 1-PCA 3 show the standard regression coefficients (Tables 5Go, 6Go, and 7Go). Principal component analysis 3 was regarded as the "full model" and was used to label the factors with respect to their biological meaning.


View this table:
[in this window]
[in a new window]
  		Table 5. Principal component analysis 1. Correlations (standardized regression coefficient) between udder health parameters and 4 components after oblique rotation (promax rotation, power = 3).
 

View this table:
[in this window]
[in a new window]
  		Table 6. Principal component analysis 2. Correlations (standardized regression coefficient) between udder health parameters, parity, and stage of lactation and 4 components after oblique rotation (promax rotation, power = 3).
 

View this table:
[in this window]
[in a new window]
  		Table 7. Principal component analysis 3. Correlations (standardized regression coefficient) between udder health parameters, parity, stage of lactation, expected milk yield, SCC, and 4 components after oblique rotation (promax rotation, power = 3).
 
The first component seems to characterize the "small udder" in contrast to the deep udder. Variables with the highest positive loadings on the component were first parity, small udder, and short teats. Variables with high negative loadings were parity 3+, deep udder, and MLOGSCC. The second component could best be described as a "distressed udder," as variables describing an impaired teat surface such as wounds on warts, poor skin quality, and teat wounds were loading high on the component. Further, a long or backward-bulging udder and fleshy teats were related to the component. The third component was labelled "mastitis udder" because it consisted of variables describing acute alterations (clinical mastitis) and chronic alterations (asymmetry, knotty tissue) of the udder. Cows loading high on the "mastitis udder" had a higher MLOGSCC as well as a lower MRESMILK. The fourth component was labelled "soiled udder." Cows loading high on "soiled udder" had soiled udder and teats and were more likely to be in early lactation than in other stages of lactation.

With the introduction of non-clinical variables in PCA 2 and PCA 3, the biological meaning of the components emerges more clearly. In PCA 1, the component "small udder" starts with high positive loadings (probabilities) of small udder shape and short teats together with high negative loadings of deep udder shape and degree of teat end callosity (PCA 1). The addition of parity in PCA 2 and SCC in PCA 3 explains the "small udder" further as having a higher probability among first parity cows in contrast to cows in second (reference level) or higher parity and in contrast to cows having normal (reference level) or deep udder shape. Furthermore, cows with high loadings on that component had a lower MLOGSCC.

From PCA 1 to PCA 2, the order of the components changed. When only clinical variables were analyzed, the component "distressed udder" explained most of the variation in data. When the variables parity and stage of lactation were added, the component "distressed udder" moved to the second position, and "small udder" explained most of the variation of the data and remained the first extracted component in PCA 3. From PCA 2 to PCA 3 the components "mastitis udder" and "soiled udder" changed their order and meaning. The oblique rotation revealed a low inter-correlation between the components in general (Table 8Go). The highest inter-component correlation (–0.17) was found between component 1 "small udder" and component 4 "soiled udder". That is, in general, small udders were slightly less soiled.


View this table:
[in this window]
[in a new window]
  		Table 8. Inter-component correlations of principal component analysis 3 after oblique rotation.
 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Component 1, The Small Udder
First parity cows were related to small udders with short teats, whereas deep udders and cows in higher parities were negatively correlated with the component. Udder depth is reported to increase with parity (Seykora and McDaniel, 1985). Deep udders are likely to develop if the strength of the udder attachment decreases with increasing age (Harris et al., 1992). In general, there is a genetic correlation between improvement of milk production and an increase in udder depth (Kelm et al., 2000). The increase in milk-producing capacity requires a higher activity and storage capacity of the secretory cells of the udder (Akers, 2000) and, thus, requires space to expand. An udder with tight attachment caused by strong ligaments is higher, but milk production may be restricted mechanically (Blake and McDaniel, 1978). MRESMILK contributed with a negative loading to the component, but below the threshold of 0.3. Cows with small udders, short teats, and in first parity did not differ in milk yield from other cows in first parity.

The small udder was related to lower SCC. Deeper udders indicate a decrease in teat end to floor distance, which is correlated with higher SCC (Rogers et al., 1991; Faye et al., 1998) and is even related to a higher risk of subclinical (Ronningen and Reitan, 1990) and clinical (Slettbakk et al., 1995) mastitis. In general, cows in higher parities have higher SCC (Laevens et al., 1997; De Haas et al., 2002). However, in cows that are uninfected by mastitis pathogens, SCC is uncorrelated with parity (Laevens et al., 1997). In conclusion, cows with a high loading on the component "small udder" may be less susceptible to mastitis than cows with normal or deep udder shape.

Component 2, The Distressed Udder
In comparison with normal udder shape, long udders and backward-bulging udders were related to harder tissue and an impaired teat surface with rough teat skin and wounds on warts and teats. Rough teat skin (Burmeister et al., 1998) and wounds on skin (Larsen et al., 2000) are a reservoir for S. aureus. Even superficial lesions of the teat skin increase the risk of subclinical and clinical mastitis (Agger and Willeberg, 1986). Warts caused by viruses can result in local damage of the natural defense mechanisms of the teat and predispose them for bacterial mastitis (Wellenberg et al., 2002).

Hardness of udder tissue characterizes two different morphological features: the udder parenchyma with aggregations of alveoli (secretory cells) partitioned off into lobules by thin connective tissue septa growing into stronger connective tissue, the second morphologic feature, which builds up the udder attachment with the ligaments. The percentage of secretory tissue differs among individual dairy cows and genotypes and is influenced by parity and stage of lactation (Michel, 1994). In freshly calved first parity cows, the percentage of fat tissue is high, and the percentage of secretory tissue is low. During the first lactation, the percentage of secretory tissue increases and replaces the fat tissue, accompanied by increasing udder hardness. In subsequent parities, number and size of alveoli increase further until the fifth or sixth parity. The secretory tissue decreases again during the declining phase of lactation and during involution of the secretory tissue. In contrast to information from literature, stage of lactation and parity did not contribute to component 2. One explanation could be an interaction between connective and secretory tissue: udders appearing fleshier after milking have a heavier stroma or connective tissue than udders with a large amount of secretory tissue (Smith, 1959). Furthermore, the hardness of the udder was measured from behind with both hands palpating the udder close to the ventral abdomen where ligaments and connective tissue were strongest and percentage of secretory tissue was lowest. Cows with a strong ligament and udder capsule may also have had a stronger connective tissue in the parenchyma, leading to a higher degree in hardness than cows with a weaker ligament structure (Akers, 2000). However, previous SCC showed no relationship with high component 2 values on the "distressed udder," but cows with such "distressed udder" may be at a higher risk for subsequent clinical mastitis because of impaired teat surface.

Component 3, The Mastitis Udder
Because the incidence of truly acute clinical mastitis is of short duration (Smith and Hogan, 1995), we expected to find a low frequency of acute cases in our study. In addition to signs of acute clinical mastitis, knotty tissue and asymmetry of quarters significantly contributed to the "mastitis udder," component 3.

This component appeared to be a mixture of acute and chronic symptoms of clinical mastitis. Hardness of mastitis quarter and diagnosed clinical mastitis are acute symptoms, whereas asymmetry and knotty tissue are chronic alterations of the udder. Chronic mastitis can lead to atrophy of the affected quarter, resulting in asymmetry as a post-inflammatory state. Mastitis can also develop progressively with increases in fibrous tissue so that the size of the affected quarter increases (Fuchs, 1994). Further, the farmer may have decided to dry off a mastitis quarter, resulting in involution of the affected quarter. Because these chronic alterations were longer-lasting conditions, the observed frequency in the presented investigation is much higher than the frequency of acute mastitis. The clinical variables with high loadings point toward the same result—a clinical (visible) mastitis—and can be used as additional classification of clinical mastitis. Especially in situations where under-reporting is suspected, this component can improve the quality of data because it decreases the number of false-negative non-mastitis cows. A problem when using this component to classify udder health can be that it only tells that there was previous mastitis but tells nothing about the duration or the onset of disease. Conversely, udders once affected by clinical mastitis are more susceptible to recurrent cases of mastitis with acute symptoms (Hogan et al., 1989). When there is no further information about first cases of clinical mastitis available, a distinction between acute clinical mastitis and chronic mastitis with acute episodes is difficult, especially in older cows.

As expected, the variables describing clinical signs of mastitis were related to a higher SCC and a lower milk yield in comparison with other cows in the same parity. Chronic alterations may have reduced the milk production in the affected quarter. Furthermore, the reduction in milk yield after clinical mastitis seemed to persist during the whole lactation as reported in the literature (Morris, 1973; Bartlett et al., 1991). A systematic screening for cows with high loadings on the component may be beneficial in the assessment of udder health status and in the development of a treatment strategy in the herd.

Component 4, The Soiled Udder
The soiled udder was related to early lactation and is consistent with Ward et al. (2002), who found higher counts of Eschericia coli and Staphylococcus uberis in beds of early lactation cows and a higher degree of soiling in early lactation cows. No other clinical or production variable contributed to the "soiled udder" when stage of lactation was added to the analysis. This result is in contrast to the results of a factor analysis performed by Bruun (2003), where days after calving showed no relation with degree of soiling. Contamination via manure reflects hygiene in the barn (Cook, 2002) and because degree of soiling was measured immediately after milking, hygiene practices before the milking process had an influence on the degree of soiling. These management practices affected the cows independently from their udder and teat morphology, milk production, or previous SCC.

General Discussion
Differences in the clinical examinations performed by 2 different persons might have induced a bias. Houe et al. (2002) have shown that the agreement between clinicians on pathological conditions is high, but there was a higher variation when classifying non-pathological conditions such as udder shape or hardness of udder parenchyma. Because our data recording sheets supplied the 2 clinicians with detailed descriptions and figures on specific conditions, we expect the variation between clinicians to be low. Further, the last observation of each cow was chosen to minimize bias caused by a "learning effect" of the examiners during the project.

When developing and interpreting PCA, the researcher should find biologically meaningful descriptions for the extracted components. The order of the components changed with the introduction of non-clinical variables in PCA 2 and PCA 3. Components "soiled udder" and "mastitis udder " reversed in PCA 2 and PCA 3. But, the amount of variation explained by each component was similar; a change in the order of the components was not an interpretational problem. The clinical variables with the highest loadings on a component kept their high loading and their position within a component throughout the model-building process. The stability of these variables is evidence for a meaningful pattern that was not produced by chance. When non-clinical variables were included in the analysis, the focus changed within the component "small udder," and age became the most important variable related to the size of the udder. In contrast, the component "distressed udder" consisted only of clinical variables that were not related to age, stage of lactation, milk production, or previous SCC. Therefore the "distressed udder" remains invariant from PCA 1 to PCA 3. The factor "mastitis udder" is a combination of clinical variables, previous SCC, and milk yield. However, each part of the combination adds up to the whole component, indicating that each part is of the same importance in its contribution to the "mastitis udder." The fourth component, "soiled udder," is determined only by those clinical variables describing soiling of teats and udder and stage of lactation. The presence of a low inter-component correlation with the "small udder" indicates that there is only a weak relationship between udder size and soiling.

Factor analysis was used in 2 other studies on clinical udder examinations in Denmark. Houe et al. (2002) extracted 3 factors with common factor analysis including clinical variables only: the mastitis udder, the long udder, and the high udder. Variables loading high on the mastitis udder were asymmetric front or hind quarters and dry quarters. Similar to our results, but by using the mastitis factor as a variable in a general linear model, Houe et al. (2002) showed that an increase in factor score was correlated with a decrease in lactation milk yield (305-d energy-corrected milk). As in our study, Bruun (2003) extracted 4 factors, but used common factor analysis with maximized variance. The author labelled the first factor "chronic changes," which was similar to our component "mastitis udder." In contrast to our results, no acute signs of mastitis contributed to the factor and no relationship with milk yield were found. The researcher used the milk yield recorded immediately before clinical examination, whereas the present study worked with a milk model to classify cows as high, average, or low yielders within their parity.

Comparable with Bruun et al. (2003), the total amount of variance explained was around 27%, which indicates that several variables have high degrees of uniqueness. That is, they provide important information on their own. Further, the data structure may not have been optimal; Bruun (2003) rescaled ordinal variables with the method of optimal scaling to improve linearity. When rescaled variables are analyzed, the interpretation is more difficult with regard to the original categories. In the present study, dummy variables were used, e.g., for parity and stage of lactation, when non-linearity was suspected. Regardless of this somewhat different approach, extracted factors or components were similar in both studies when the correlation to the factor (component) was strong. In addition to a different method within factor analysis and treatment of variables, differences may also be caused by the choice of variables that were included in the analysis. Introduction of new variables with important contribution to a factor (component) may shift the focus or enhance the interpretability of a factor (component) as seen in the component "small udder." Variables related to factor "udder shape" (Bruun, 2003) and to the component "small udder" of the present study were identical, which is further evidence that the PCA produced a biologically meaningful pattern and contributes to the understanding of udder health.

Welfare and health assessment protocols, including animal-based measurements, become an issue of increasing importance in health management and quality programs in several countries (Noordhuizen and Metz, 2003; Whay et al., 2003). Increasing herd sizes increase the demand for systematic, effective, and economic tools in udder health management. Inclusion of animal-based measurements could be beneficial in advisory services to understand the udder health situation on the farm and to specify udder health risks. The extracted udder types can be implemented in decision support systems on udder health in dairy farms. The presence of cows with "distressed udder" indicates impaired teat and udder condition and could be used to evaluate the effectiveness of hygienic measures or the necessity of measures in the milking parlor. The "soiled udder" could be used as an indicator of hygiene in the barn. Classifying cows according to the "small udder" can support decision making when selecting cows for mastitis prevention on the herd level. The prevalence of cows with a "mastitis udder" provides information about the number of clinically affected cows and the severity of clinical signs, which is of prognostic value. When results from clinical examination are compared with farm records, management practices, such as blinding of mastitis quarters, will be revealed, and the quality of farm records can be evaluated. Further investigations are necessary to test the applicability and the effect of animal-based measurements in udder health monitoring.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The stepwise approach in 3 models by using PCA shows that the clinical variables form a consistent pattern that was not changed by adding more traditional variables such as milk yield and SCC. This stability in udder type patterns provides solid evidence that clinical udder examinations can be a reliable tool in veterinary research and practice. The extracted components provide information on udder health status in addition to SCC and milk yield. Both the practitioner and the scientist conducting a field study can benefit from performing systematic clinical udder examinations, especially when under-reporting of mastitis cases is expected.

Received for publication August 8, 2003. Accepted for publication December 19, 2003.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Agger, J. F., and P. Willeberg. 1986. Epidemiology of teat lesions in a dairy herd. II. Associations with subclinical mastitis. Nord. Vet. Med. 38:220–232.[Medline]

Akers, R. M. 2000. Selection for milk production from a lactation bology viewpoint. J. Dairy Sci. 83:1151–1158.[Abstract]

Andersen, H. J., K. Aagaard, F. Skjøth, E. Rattenborg, and C. Enevoldsen. 2000. Integration of research, development, health promotion, and milk quality assurance in the Danish dairy industry. Abstr. 635 in Proc. 9th Symp. Int. Soc. Vet. Epidemiol. Econ., Breckenridge, CO.

Anonymous. 2002. Handbook in Cattle Husbandry. (in Danish). Håndbog I Kvæghold. Landbrugets Rådgivningscenter. Lanbrugsforlaget, Grenaa, Denmark.

Bartlett, P. C., J. F. Agger, H. Houe, and L. G. Lawson. 2001. Incidence of clinical mastitis in Danish dairy cattle and screening for non-reporting in a passively collected nation surveillance system. Prev. Vet. Med. 48:73–83.[Medline]

Bartlett, P. C., J. van Wijk, D. J. Wilson, C. D. Green, G. Y. Miller, G. A. Majewski, and L. E. Heider. 1991. Temporal patterns of lost milk production following clinical mastitis in a larger Michigan Holstein herd. J. Dairy Sci. 74:1561–1572.[Abstract]

Blake, R. W., and B. T. McDaniel. 1978. Relationships among rates of milk flow, machine time, udder conformation, and managemental aspects of milking efficiency: A review. J. Dairy Sci. 61:363–378.[Abstract/Free Full Text]

Bruun, J. 2003. Risk factors for different health measures in Danish dairy cows—with metritis, antibacterial drug use and udder health as examples. Ph.D. dissertation, The Royal Veterinary and Agricultural University, Frederiksberg C, Denmark.

Burmeister, J. E., L. K. Fox, J. K. Hillers, and D. D. Hancock. 1998. A comparison of two methods of evaluation of teat skin pathology. J. Dairy Sci. 81:1904–1909.[Abstract]

Cook, N. 2002. The influence of barn design on dairy cows hygiene, lameness and udder health. Pages 97–103 in Proc. 35th Annu. Conf. Am. Assoc. Bovine Pract., Madison, WI.

De Haas, Y., H. W. Barkema, and R. F. Veerkamp. 2002. The effect of pathogen-specific clinical mastitis on the lactation curve for somatic cell count. J. Dairy Sci. 85:1314–1323.[Abstract]

Enevoldsen, C. E., K. B. Elgaard, N. P. Jensen, K. H. Nielsen, H. K. Nørregaard, and J. S. Philipsen. 2000. Identification of interactions and nonlinear effects, assessment of herd effects, and model validation in a dairy herd health management context. Abstr. 516 in Proc. 9th Symp. Int. Soc. Vet. Epidemiol. Econ., Breckenridge, CO.

Faye, B., L. Perochon, N. Dorr, and P. Gasqui. 1998. Relationship between individual-cow udder health status in early lactation and dairy cow characteristics in Brittany, France. Vet. Res. 29:31–46.[Medline]

Fuchs, H.-W. 1994. Pathomorphologie der Rindermilchdrüse. Pages 244–294 in Euter- und Gesäugekrankheiten. K. Wendt, H. Bostedt, H. Mielke, and H.-W. Fuchs, ed. Gustav Fischer Verlag Jena and Stuttgart, Germany.

Hamann, J. 2003. Definition of the physiological cell count threshold based on changes in milk composition. Bull. Int. Dairy Fed. 381, Mastitis Letter 25:9–12.

Harris, B. L., A. E. Freeman, and E. Metzger. 1992. Genetic and phenotypic parameters for type and production in Guernsey dairy cows. J. Dairy Sci. 75:1147–1153.[Abstract]

Hatcher, L. 1994. A Step-by-Step Approach to Using the SAS® System for Factor Analysis and Structural Equation Modeling. SAS Inst., Inc., Cary, NC.

Hogan, J. S., K. L. Smith, K. H. Hoblet, P. S. Schoenberger, D. A. Todhunter, W. D. Hueston, D. E. Pritchard, G. L. Bowman, L. E. Heider, B. L. Brockett, et al. 1989. Field survey of clinical mastitis in low somatic cell count herds. J. Dairy Sci. 72:1547–1556.

Houe, H., M. Vaarst, and C. Enevoldsen. 2002. Clinical parameters for assessment of udder health in Danish dairy herds. Acta Vet. Scand. 43:173–184.[Medline]

IDF Subgroup of Group A2, Bovine Mastitis. 1997. Recommendations for presentation of mastitis related data. Bull. Int. Dairy Fed. 321:8–33.

Kelm, S. C., A. E. Freeman, and NC-2 Technical Committee. 2000. Direct and correlated responses to selection for milk yield: Results and conclusions of regional project NC-2, "improvement of dairy cattle through breeding, with emphasis on selection." J. Dairy Sci. 83:2721–2732.[Abstract]

Laevens, H., H. Deluyker, Y. H. Schukken, L. de Meulemeester, R. Vandermeersch, E. de Muêlenaere, and A. de Kruif. 1997. Influence of parity and stage of lactation on the somatic cell count in bacteriologically negative dairy cows. J. Dairy Sci. 80:3219–3226.[Abstract]

Larsen, H. D., K. H. Sloth, C. Elsberg, C. Enevoldsen, L. H. Pedersen, N. H. R. Eriksen, F. M. Aarestrup, and N. E. Jensen. 2000. The dynamics of Staphylococcus aureus intramammary infection in nine Danish dairy herds. Vet. Microbiol. 71:89–101.[Medline]

Markusfeld, O., N. Galon, and E. Ezra. 1997. Body condition score, health, yield and fertility in dairy cows. Vet. Rec. 141:67–72.[Abstract/Free Full Text]

Michel, G. 1994. Anatomie der Milchdrüse. Pages 44–50 in Euter- und Gesäugekrankheiten. K. Wendt, H. Bostedt, H. Mielke, and H.-W Fuchs, ed. Gustav Fischer Verlag Jena and Stuttgart, Germany.

Morris, R. S. 1973. The depression of quarter milk yield caused by bovine mastitis, and the response of yield to successful therapy. Aust. Vet. J. 49:153–156.[Medline]

Noordhuizen, J. P. T. M., and J. H. M. Metz. 2003. Quality control on dairy farms, with emphasis on food safety, animal health and animal welfare. Page 167 in Abstracts of 54th Annu. Mtg. European Association for Animal Production. Rome, Italy, EAAP, Wageningen, Academic Publishers, Wageningen, The Netherlands.

Rogers, G. W., G. L. Hargrove, T. J. Lawler, Jr., and J. L. Ebersole. 1991. Correlations among linear type traits and somatic cell count. J. Dairy Sci. 74:1087–1091.[Abstract]

Ronningen, O., and A. D Reitan. 1990. Influence of static and dynamic teat characteristics and milking time on udder health in Norwegian Red cattle. J. Dairy Res. 57:171–177.[Medline]

Rosenberger, G. 1979. Clinical Examination of Cattle. Paul Parey, Berlin and Hamburg, Germany.

Rummel, R. J. 1970. Applied Factor Analysis. Northwestern University Press, Evanston, IL.

SAS. 1999. Version 8.2. SAS Inst., Inc., Cary, NC.

Sears, P. M., B. S. Smith, P. B. English, P. S. Herer, and R. N. Gonzalez. 1990. Shedding pattern of Staphylococcus aureus from bovine intramammary infections. J. Dairy Sci. 73:2785–2789.[Abstract]

Seykora, A. J., and B. T. McDaniel. 1985. Heritabilities of teat traits and their relationships with milk yield, somatic cell count, and percent two-minute milk. J. Dairy Sci. 68:2670–2683.[Abstract/Free Full Text]

Slettbakk, T., A. Jørstad, T. B. Farver, and J. C. Holmes. 1995. Impact of milking characteristics and morphology of udder and teats on clinical mastitis in first- and second-lactation Norwegian cattle. Prev. Vet. Med. 24:235–244.

Smith, K. L., and J. S. Hogan. 1995. Epidemiology of mastitis. Pages 3–13 in Proc. 3rd IDF Int. Mastitis Sem., Tel-Aviv, Israel.

Smith, V. R. 1959. Physiology of Lactation. The Iowa State University Press, Ames.

Vaarst, M., B. Paarup-Laursen, H. Houe, C. Fossing, and H. J. Andersen. 2002. Farmers’ choice of medical treatment of mastitis in Danish dairy herds based on qualitative research interviews. J. Dairy Sci. 85:992–1001.[Abstract]

Ward, W. R., J. W. Hughes, W. B. Faull, P. J. Cripps, J. P. Sutherland, and J. E. Sutherst. 2002. Observational study of temperature, moisture, pH and bacteria in straw bedding, and faecal consistency, cleanliness and mastitis in cows in four dairy herds. Vet. Rec. 151:199–206.[Abstract/Free Full Text]

Whay, H. R., D. C. J. Main, L. E. Green, and A. J. F. Webster. 2003. Assessment of the welfare of dairy cattle using animal-based measurements: Direct observations and investigations of farm records. Vet. Rec. 153:197–202.[Abstract/Free Full Text]

Wellenberg, G. J., W. H. M. Van der Poel, and J. T. Van Oirschot. 2002. Viral infections and bovine mastitis: A review. Vet. Microbiol. 88:27–45.[Medline]


This article has been cited by other articles:


Home page
 J DAIRY SCIHome page
W. Steeneveld, H. Hogeveen, H. W. Barkema, J. van den Broek, and R. B. M. Huirne
The Influence of Cow Factors on the Incidence of Clinical Mastitis in Dairy Cows
J Dairy Sci, April 1, 2008; 91(4): 1391 - 1402.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Klaas, I. C.
Right arrow Articles by Houe, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Klaas, I. C.
Right arrow Articles by Houe, H.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS