 |
|
|
Investigation of farm-level distribution and risk
factors |
|
A cross-sectional study was performed on eight
representative dairy farms in north western England and north Wales, which were
stratified into high prevalence, medium prevalence, low prevalence and BDD-free
farms.
The morbidity, prevalence and risk factors for BDD have
been relatively well investigated and published in various cross-sectional
studies and surveys. These studies concurrently investigated different
infectious conditions causing lameness (Frankena et al., 1991; Somers et al.,
2005)), and / or were primarily interested in management, husbandry and
production effects across all herds (the same authors, and Rodriguez-Lainz et
al., (1996); Rodriguez-Lainz et al. (1998, 1999); Wells et al. (1997);
Holzhauer et al. (2006)).
The objective of our study
was to investigate the distribution of, and risk factors for, BDD within herds
(i.e. within and between management groups) and between farms. Hence, we
elected to perform a detailed study on a small number of farms, rather than a
less detailed study on a larger number of farms. Blood samples were taken for
serology of all animals present on the farms on the date of sampling (n=2215);
the feet of a randomly selected subset of animals (n=609) were inspected for
BDD lesions. We applied statistical methods to characterize the farm-level
prevalence, distribution and risk factors for BDD. Environmental and animal
hygiene scores were recorded; management and husbandry practices were recorded
by interview, and data on animal records were collected.
|
|
Fig. W5. Clinical foot inspection using the borescope |
|
We stratified the study population into high, medium and
low prevalence farms on the basis of prior clinical information. As expected,
high prevalence farms had higher clinical prevalence in the lactating and dry
cow groups (35-60%); they also had a higher proportion of acute lesions, i.e.
recent infections. The mean prevalence was approximately 40%, which is
comparable to published figures (Laven, 2003).
Appropriate statistical
analysis was performed to describe within-farm disease distribution and
investigate risk factors for clinical BDD and serology. Almost all animals with
clinical BDD belonged to the cow groups. There was substantial inter-farm
variability of clinical prevalence in these cow groups, ranging from 0% to 67%,
with an overall prevalence of 41%. Higher prevalence farms also showed a higher
morbidity of acute lesions. The highest prevalence was observed in animals
between 3 and 5 years of age; prevalence decreased with increasing age. This
may indicate development of partial immunity.
These findings are indicative
of an epidemic disease pattern on high prevalence farms, corresponding to a
higher level of exposure. The disease pattern was more endemic on lower
prevalence farms. Anecdotal information from farmers and veterinarians suggests
that fluctuations of these patterns occur within farms over time; these may be
partly due to seasonal effects, but are also due to unexplained effects.
Serology was useful for further exploring the farm-level dispersion
of the infection. While the ELISA was developed using a ‘cocktail’ of
BDD-associated Treponema spp. antigens, we could not unequivocally
assert that the IgG2 titres measured were elicited by BDD-related
challenge with these bacteria. However, the serological frequency distributions
were consistent with an infectious disease, showing evidence of a bimodal
response; this effect was stronger for high prevalence farms. Comparison of the
distributions of the clinical positives to the clinical negatives showed more
clearly that the ELISA distinguishes between positive and negative
sub-populations. However, there was substantial overlap, although the
difference between the mean titres of these subpopulations was significant.
Animals with acute or chronic lesions had significantly higher titres than
those with regressing lesions. This indicated that the antibody halflife was
short, which was consistent with previously published research (Walker et al.,
1997; Trott et al., 2003).
Antibody titres were low for
the young stock, rising gradually up to the age of first calving. Subsequently,
they rose sharply; thereafter, they remained high, presumably due to repeated
exposure in the housing environment. No reduction was found for animals over 5
years of age, as was the case with clinical disease; this indicated that
serology was a good marker for repeated exposure to infection. Insofar as this
led to development of partial immunity, it would appear that this was not
through humoral mechanisms – a conclusion shared by other authors (Walker
et al., 1997; Demirkan et al., 1999).
The statistical models
quantitatively confirmed the patterns identified by the EDA. As two outcomes
were possible (clinical inspection and serology), different models were
formulated. Comparison of different random effects structures indicated that
fitting management group nested within farm was most appropriate. For the
outcome of clinical BDD, a generalized linear mixed model (GLMM) was applied;
for the outcome of serology, a linear mixed effects model (LME) was used.
Unfortunately, no management group-level explanatory variables (housing
hygiene, housing comfort, BDD treatment and footbath protocols) were
significant in the univariable analysis, for either of the statistical models;
therefore only individual-level variables remained in the final multivariable
models.
The highest odds of clinical BDD were in the 2-3 year category;
animals >6 years had substantially lower odds. The odds ratio for lactating
cows was more than double that of dry cows. For the serological outcome, the
estimates rose gradually, showed a sharp rise for the 3-4 year category, and
continued to rise slightly with increasing age.
Hygiene in the housing is the
risk factor that has been most commonly associated with BDD. The
individual-level foot hygiene scores (FHS) and body hygiene scores (BHS) could
be considered to be proxies for environmental hygiene. The statistical models
did not associate high FHS with clinical BDD or high serology; however, BHS was
significantly identified as a risk factor by both the GLMM and the LME. This
was surprising, as FHS might be expected to exert a stronger effect. An
empirical observation is that within management groups, these scores tended to
be quite uniform; it is possible that effects on the foot level could be masked
by a lack of observable variability in FHS. |
|
 |
|
|
©
Willem Daniel Vink 2006 |
