|
|
||||||||

* Department of Population Medicine, Ontario Veterinary College, University of Guelph, Ontario, N1G 2W1 Canada
Department of Population Medicine and Diagnostic Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY 14853
1 Corresponding author: gerard{at}mecnrec.ca
| ABSTRACT |
|---|
|
|
|---|
Key Words: foot lesion prevalence tie stall free stall
| INTRODUCTION |
|---|
|
|
|---|
The majority of lameness cases are due to lesions on the foot (Murray et al., 1996). Therefore, regular monitoring of lesions at the hoof level (Mills et al., 1986; Guard, 2001) allows for earlier interventions by producers and their advisors (Noordhuizen, 2003; Shearer et al., 2004). Through the use of earlier interventions, the number of severe lameness cases should decrease, and in turn, animal well-being should improve. To regularly monitor disease in herds, a standardized case definition (Kelton et al., 1998) and a target level at which diagnostic or corrective action is taken (Radostits, 2001) are required. In North America, these data have been generated for use in monitoring lameness through locomotion scoring (Cook, 2003; Espejo et al., 2006). Similar monitoring data have not been reported for foot lesions. Nevertheless, in several countries in Europe, the foot lesion data for these target levels have been generated (Manske et al., 2002; Sogstad et al., 2005; Holzhauer et al., 2006). Yet given the difference in breeds, management, feeding, and environments among countries, it is unclear whether data from these studies are applicable to Ontario or other North American conditions.
Considering the lack of current information on foot lesion prevalence data, a 15-mo project was designed to describe the state of foot health in Ontario dairy cows. The objective was to describe the cow- and herd-level prevalence, as well as the within-herd variance components, of foot lesions in Ontario dairy cattle.
| MATERIALS AND METHODS |
|---|
|
|
|---|
During the training session, the HT were trained in standardized lesion identification and completed 2 lesion identification quizzes, 1 before training and 1 after training. To ensure that lesion identification remained consistent throughout the 15 mo, HT were provided with a lesion identification chart and descriptions for future reference. At the conclusion of the project, the HT completed another lesion identification quiz. Hoof trimmers were provided with carbon-copy lesion recording sheets and were asked to record all lesions on all cows that were trimmed. Selection of the cows for trimming was at the discretion of the producer. No attempt was made to select cows based on trimming history, lactation status, or any other criteria. In addition to recording lesion data, HT completed a 3-page questionnaire for each farm during the study period (Cramer, 2007).
Hoof trimmers identified the presence of foot lesions and applicable treatments by using the criteria described in Table 1
. The foot lesion nomenclature was based on a proposal by the American Association of Bovine Practitioners Lameness Committee (Shearer et al., 2004). Hoof trimmers recorded foot lesion data on a form that was based on the forms used by most of the participating HT.
|
Data Management and Statistical Analysis
Both questionnaire and lesion data were entered into a database (MySQL 4.1, MySQL AB, Uppsala, Sweden) via the Internet. Data management and descriptive analysis were done with Microsoft Excel and a commercially available statistical program (Stata 9.1, StataCorp, College Station, TX). Individual cows with duplicate, unreadable, or missing cow identification were removed from the data set.
For each type of foot lesion, prevalence was calculated as the number of affected animals divided by the number examined. Data on lesions at the foot level were collapsed first into front and hind limbs and then into cow. A cow was considered affected if one hoof had a lesion present. Only the first occurrence of a cow in the data set was used to calculate lesion prevalence at the cow and herd level. For each foot lesion, the difference in prevalence between free-stall and tie-stall housing types was compared with a chi-squared test at the cow level and with a Mann-Whitney test at the herd level.
Before the foot lesion data were collapsed into cow lesion data, the data structure had 4 different levels: HT, within-HT herds, within-herd cows, and within-cow feet. Data were recorded at the lowest level (foot) on a binary scale. For variance component analysis, intraclass correlation coefficients (ICC) were calculated at the herd level in Stata 9.1 by using a random effect generalized linear mixed model with a binomial distribution and a logit link. In Stata, this was accomplished by using the generalized linear and latent mixed model procedure. This procedure used adaptive quadrature to determine the maximum likelihood estimates. The generalized linear and latent mixed model was fitted without restrictions on the number of iterations.
All variance component models contained only cows housed in tie-stall or free-stall herds because of meager data in the other housing categories. In addition, these models contained both HT and housing type as fixed effects. Furthermore, the models were stratified according to HT for each lesion to allow for comparisons between HT.
To estimate the variance components from the model, the following equation was used:
![]() |
where var(Z) is the total variation of the model, var(herd) is the variance of the herd random effects, and var(error) is
2/3 because an underlying latent logistic distribution is assumed (Dohoo et al., 2003). From these variance components, an ICC was calculated by using the following equation:
![]() |
Herd ICC represent within-herd correlations. A herd ICC of 0 means that there is no correlation between animals of the same herd and an ICC of 1.0 means a perfect correlation between animals (Holzhauer et al., 2006).
Cohens kappa was used to calculate the level of agreement beyond chance between HT for each foot lesion from the lesion identification quizzes (Dohoo et al., 2003) to evaluate the agreement between HT. To further investigate the potential of misclassification bias, ICC were compared between HT.
| RESULTS |
|---|
|
|
|---|
Average herd size and 305-d milk production for the 157 herds enrolled in DHI was 86 [95% confidence interval (CI): 62 to 112] and 9,254 kg (CI: 8,938 to 9,571 kg), respectively, for free-stall herds and 42 (CI: 39 to 44) and 9,179 kg (CI: 8,973 to 9,385), respectively, for tie-stall herds. Herd size for herds not enrolled in DHI was not significantly different from herds using DHI services.
Tables 2
and 3
present the cow- and herd-level prevalence for tie stalls and free stalls, respectively. Cow-level prevalence is presented for the first hoof trimming in the data set. Mean herd-level prevalence is presented for the first hoof trimming date for each herd. Infectious lesions were the most common in both types of housing systems, with digital dermatitis as the most common lesion overall. In both housing systems, back feet had a greater lesion prevalence than did front feet. Herd prevalence varied, yet for most lesions, there were herds that had a prevalence of zero.
|
|
Average between-HT kappa from the lesion identification quizzes are shown in Table 4
. Even though quiz 2 was completed after the training session, the kappa statistic for 7 of the 10 foot lesions actually decreased. Numerical ratings for quizzes 1, 2, and 3 were 65, 71, and 94%, respectively. Substantial to perfect agreement existed between HT for quiz 3. Within-herd ICC are given in Table 5
. Most herd ICC showed variation between HT, but only 1 HT (HT 4) had consistently low ICC for 7 of the 9 lesions.
|
|
| DISCUSSION |
|---|
|
|
|---|
Prevalence of Foot Lesions
Comparing herd- and cow-level prevalence estimates of foot lesions from this study with the results of other studies was difficult because of the differences in foot lesion scoring systems and overall herd management styles. This study is the first report of foot lesion prevalence estimates across multiple herds and management styles for anywhere in North America. Most North American studies have reported lameness prevalence or foot lesion prevalence for lame cows only (Cook, 2003; Melendez et al., 2003; Booth et al., 2004).
The prevalence estimates of foot lesions showed a greater prevalence than in Norway (Sogstad et al., 2005), a lesser prevalence than in Sweden (Manske et al., 2002) and a prevalence similar to the Netherlands (Holzhauer et al., 2006). The Dutch research was similar in study design to the current study, but the Dutch work included only free-stall herds. The greater prevalence reported for free-stall herds was consistent with other studies (Manske et al., 2002; Sogstad et al., 2005).
Similar to other studies (Somers et al., 2003; Holzhauer et al., 2006), infectious lesions were the most prevalent in both housing systems. This finding raises questions about the efficacy, implementation, or management of current recommended therapies. Unlike in the study by Sogstad et al. (2005), the prevalence of heel horn erosion was relatively consistent across housing types. This was somewhat surprising because moisture and manure are known risk factors (Toussaint Raven et al., 1985) for heel horn erosion, and exposure to these factors was greater in most free-stall barns. The prevalence of digital dermatitis was consistent with other reports (Somers et al., 2003; Holzhauer et al., 2006) that have illustrated the difficulty of producers in managing the disease in spite of relatively effective therapies and control strategies (Laven and Logue, 2006). The greater prevalence of digital dermatitis in free stalls (22.9 vs. 9.3%) reflects the increased exposure these cows would have to manure and moisture.
The recording of sole hemorrhages (11.0% for free-stall barns and 7.1% for tie-stall barns) in was relatively subjective. Other reports required HT to judge the size of the area affected (Holzhauer et al., 2006). Therefore, it was likely that the lesser prevalence estimate was an assessment of the prevalence of severe sole hemorrhages and did not include minor discoloration of the sole. This is supported by the prevalence estimates for ulcers (9.3% in free stalls and 4.7% in tie stalls), white line separations (5.2% in free stalls and 1.0% in tie stalls), and white line abscess (2.0% in free stalls and 0.6% in tie stalls), which were similar to the estimates of others (Manske et al., 2002; Sogstad et al., 2005; Holzhauer et al., 2006), and these foot lesions were considered gradients of a similar disease process (Leach et al., 1998; Lischer et al., 2002).
Considering that lameness is a painful, long-lasting condition (Whay et al., 1998), it is surprising that only a few cows (2.2% in free stalls and 0.3% in tie stalls) were treated with the use of a block. Reduced pressure on the affected hoof through the application of a block to the sound hoof is considered standard therapy for cows with sole ulcers or white line abscesses (Toussaint Raven et al., 1985). Alternative treatment strategies exist if enough heel is present on the sound hoof (Manabe et al., 2004). Nevertheless, it is likely that this method was not widely adopted because of the novelty of the technique. From the data, it is unclear what percentage of cows would have been considered lame and in need of a block. An attempt was made to capture these data, but it was only sporadically recorded. It is likely that with a prevalence of 11.3% combined for ulcers and white line abscesses in free-stall herds, more than 20% of these cows would have received substantial benefit from a block applied to the sound hoof. It is unclear whether the decision to place a block on a cow rests with the producer or with the HT at the time of trimming. Either way, it is likely that the cost of the block and its application decreased their use.
Effect of HT
The use of professional HT to record lesions was a potential source of misclassification bias (Sogstad et al., 2005; Holzhauer et al., 2006). To address this source of bias, HT were trained and evaluated by using digital photos at the beginning and end of the project. Although numerical grades for individual quizzes increased after the training session, the kappa statistic for some lesions actually decreased. This decrease showed that for individual HT, lesion recognition improved but that there was still a lack of agreement between HT. It was not possible to detect trends in HT lesion recognition improvement because of constraints in the number of pictures that were part of the lesion identification quizzes.
Kappa measures the extent of agreement beyond what would have been expected by chance. Some guidelines for interpreting kappa have been proposed (Dohoo et al., 2003). When these guidelines were used, the kappa values ranged from slight to substantial agreement for quizzes 1 and 2. At the conclusion (quiz 3), agreement among HT was excellent for all lesions. This high level of agreement was unexpected, especially because 11 veterinarians completed the same quiz and had slight to moderate agreement (Cramer, 2007).
The range in kappa values reported was similar to the values recorded in the Netherlands (Holzhauer et al., 2006), yet these values were lower than those in a Swedish study (Manske et al., 2002) that used a more intense degree of training. In the future, given that digital photos are only 2-dimensional and do not truly represent working conditions, a more appropriate assessment tool may be the use of live animals (Holzhauer et al., 2006).
The conflicting results obtained for lesion scoring after the training sessions were further explored by evaluating the effect of HT on herd ICC. From the ICC presented, one HT had lower ICC for lesions. Because of the low number of herds this HT contributed to the study, the effect on the whole study was likely small. Variation in the herd ICC between HT could be due to HT and other uncontrolled herd-level variables. Cow-level variation was fixed at
2/3 and would therefore not be a significant source of variation.
Based on the kappa values and herd ICC, the HT appears to be a source of misclassification bias. This finding was consistent with a Dutch study that reported high ICC for HT for sole hemorrhage, interdigital heel horn erosion, and chronic laminitis (Holzhauer et al., 2006). The present study only used a small number of HT, and it was not possible to determine ICC for HT to further evaluate this misclassification bias.
| CONCLUSIONS |
|---|
|
|
|---|
| ACKNOWLEDGEMENTS |
|---|
|
|
|---|
Received for publication February 28, 2008. Accepted for publication June 5, 2008.
| REFERENCES |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
A. H. Sanders, J. K. Shearer, and A. De Vries Seasonal incidence of lameness and risk factors associated with thin soles, white line disease, ulcers, and sole punctures in dairy cattle J Dairy Sci, July 1, 2009; 92(7): 3165 - 3174. [Abstract] [Full Text] [PDF] |
||||
![]() |
G. Cramer, K. D. Lissemore, C. L. Guard, K. E. Leslie, and D. F. Kelton The association between foot lesions and culling risk in Ontario Holstein cows J Dairy Sci, June 1, 2009; 92(6): 2572 - 2579. [Abstract] [Full Text] [PDF] |
||||
![]() |
G. Cramer, K. D. Lissemore, C. L. Guard, K. E. Leslie, and D. F. Kelton Herd-level risk factors for seven different foot lesions in Ontario Holstein cattle housed in tie stalls or free stalls J Dairy Sci, April 1, 2009; 92(4): 1404 - 1411. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |