Prevalence of CHSD in various subgroups of tested dogs
In total, 899 Australian Cattle Dogs from 35 breeding kennels were enrolled in this study. These 899 study dogs consisted of 696 dogs from litters where all pups in the litter were presented for BAER testing, 114 dogs from litters where not all pups were presented for BAER testing, and 89 single dogs tested as replacement breeding stock. Common reasons for presentation of incomplete litters included the pups being dead at birth, being killed by the mother or overlain, dying due to respiratory or gastrointestinal infections, and suspected cardiac abnormality. None of the reasons for puppy mortality appeared likely to be associated with the puppies' hearing status as detected by breeders. No breeder reported suspected deafness as a reason for non-presentation of pups for testing.
Prevalences of deafness were similar for dogs tested as part of a complete litter, an incompletely-tested litter or as single dogs (Table
). Distribution of deafness status (normal hearing, unilateral or bilateral deafness) did not differ significantly between dogs where the entire litter was tested and either where the incomplete litter was tested (P = 0.702) or where single dogs were tested (P = 0.147). However, there was some evidence that these proportions differed between dogs where incompletely-tested litters were tested compared to the testing of single dogs (P = 0.078), with a higher proportion of single dogs having normal hearing and lower proportions having unilateral deafness and bilateral deafness. After accounting for clustering of outcomes within litters and kennels using a multilevel logistic model, the odds of deafness (any deaf ears rather than normal hearing) were similar in dogs tested as incompletely-tested litters relative to those tested as complete litters (odds ratio 1.2; 95% CI 0.5 to 2.8; P = 0.693).
Table 1 Prevalences of congenital hereditary sensorineural deafness in Australian Cattle Dogs (and numbers of dogs affected) for dogs where the complete litter was BAER tested, for dogs from incompletely tested litters and for dogs tested individually (single (more ...)
Of the 52 unilaterally deaf dogs where the complete litter was BAER tested, 27 were deaf in the left ear (51.9%; 95% confidence interval 37.6% to 66.0%) and this was not significantly different from 50% as expected under the null hypothesis (goodness-of-fit P value 0.890). Of the eight unilaterally deaf dogs from partially tested litters whose affected side was recorded, two were deaf in the left ear (25.0%; 95% confidence interval 3.2% to 65.1%) and this was not significantly different from 50% as expected under the null hypothesis (goodness-of-fit P value 0.289).
Associations between potential risk factors and CHSD
Associations between exposure variables and CHSD were assessed using only the 810 dogs tested as part of completely- or incompletely-tested litters (ie we excluded single dogs tested as replacement breeding stock from these analyses). These results are shown in Tables
(univariable models), and Table
(multivariable models). Heritability of the presence of deafness (at least one deaf ear) estimated from the multivariable logistic animal model was 0.21 (standard error 0.09).
Univariable associations in Australian Cattle Dogs for congenital hereditary sensorineural deafness (CHSD) from multilevel logistic models and animal models
Table 3 Univariable relative risk ratios and odds ratios for each of having one or two ears affected by congenital hereditary sensorineural deafness (CHSD) rather than no ears affected in Australian Cattle Dogs from multinomial logistic models and cumulative (more ...)
Multivariable analysis showing association between congenital hereditary sensorineural deafness (CHSD) in Australian Cattle Dogs (n = 534) and sex, mask and pigmented body spots from a multilevel logistic model and an animal model
The prevalence of CHSD and coat markings
From both univariable and multivariable analyses, bilateral facial masks and pigmented body spots were independently associated with a reduced risk of CHSD. The prevalence of deafness in dogs with bilateral facial masks was 4.4% (3.1% unilaterally and 1.3% bilaterally deaf), compared with a prevalence of deafness of 14.6% in clear faced dogs (12.4% unilaterally and 2.2% bilaterally deaf). The odds of deafness in dogs with bilateral masks were estimated to be 0.2 times that for dogs with a clear face (P ≤ 0.001) in both univariable and multivariable analysis. Unilateral masks were not detectably associated with a reduced risk of CHSD. The prevalence of deafness in dogs with pigmented body spots was 4.8% (3.8% unilaterally deaf and 1.0% bilaterally deaf) compared with a prevalence of 12.8% deafness (9.3% unilateral and 3.3% bilaterally deaf) in dogs without body spots. The odds of deafness in animals with pigmented body spots were estimated to be 0.3 times that for other dogs in the univariable (P = 0.034) and multivariable (P = 0.038) multilevel logistic models and 0.4 times that for other dogs in the univariable (P = 0.041) and multivariable (P = 0.05) logistic animal models. White head or body patches were not associated with an increase in the prevalence of deafness in this analysis. However, these were recorded in only 27 individuals, resulting in substantial imprecision in the estimated effect.
Fourteen dogs had unilateral masks and were unilaterally deaf on an identified side. Of these, seven dogs had masks on the left-hand side, and four of these (57.1%) were deaf on the left-hand side. The remaining seven dogs had masks on the right-hand side and three of these (42.9%) were deaf on the left-hand side. There was no significant association between the side of deafness and the side of the unilateral facial pigmented mask (P = 1.0).
Neither base coat colour (red, blue, blue/black, blue black and tan, blue and tan, and red and tan) nor blue or red speckled nor blue or red mottled coat patterns were significantly associated with CHSD. However, the number of mottled dogs was too small for meaningful interpretation.
The prevalence of CHSD and sex
The prevalence of deafness in female dogs was 13.2% (9.1% unilaterally deaf and 4.1% bilaterally deaf), compared to 8.9% deafness in male dogs (5.8% unilaterally deaf and 3.1% bilaterally deaf). Female Australian Cattle Dogs had increased odds of CHSD. From the multilevel logistic models, the odds of deafness in female dogs were estimated to be 1.7 times higher than in males on univariable analysis (P = 0.044; Table
), and twice as high on multivariable analysis (P = 0.035; Table
). From the logistic animal models, the odds of deafness in female dogs were estimated to be 1.6 times higher than in males on univariable analysis (P = 0.057; Table
), and 1.9 times as high on multivariable analysis (P = 0.034; Table
). These analyses indicate that the association between sex and CHSD was probably not due to confounding by mask type or pigmented body spots.
The prevalence of CHSD and hearing status of the sire and dam
The hearing status of the sires of study dogs was known for 63% or 512 of the 810 study dogs from completely- or partly-tested litters. The sire’s hearing status was unknown for the other dogs as the study data were collected over 12 years, and early data were incomplete regarding the deafness status of sires and dams due to the previous unavailability of BAER testing facilities in the region. The prevalence of deafness (either one or both ears affected) in offspring if the sire’s hearing was normal was 8.7%, if the sire was unilaterally deaf, was 37.5%, and if the sire’s hearing status was unknown was 16.2%. Of dogs whose sire’s hearing status was known, only 10 dogs had deaf sires so effect estimates were highly imprecise but the point estimates of the odds ratios were consistent with a strong association between hearing status of the sire and CHSD on univariable analyses (multilevel logistic model: odds ratio 5.9, P = 0.085; logistic animal model: odds ratio 3.3, P = 0.200).
The hearing status of the dams of study dogs was known for only 53% or 433 of the 810 study dogs, and only 3 dogs were known to have dams with CHSD, precluding meaningful interpretation of the association between dam’s hearing status and CHSD. However, point estimates for the odds ratios were consistent with a strong association between hearing status of the dam and CHSD.
Analyses of number of ears affected
Odds ratios for having one ear affected (rather than none) were generally similar to those for having 2 ears affected (rather than none) (Table
) except when considering the mask variable. For this variable, point estimates differed substantially but were quite imprecise. Odds ratios from the univariable cumulative proportion animal models (Table
) were similar to those from the univariable logistic animal models (Table
Kennel and litter effects
Variances at kennel- and litter-levels from the null multilevel logistic model were 0.00 and 1.40 (95% CI 0.64 to 3.05), respectively. The likelihood-ratio test comparing the model to ordinary logistic regression without random effects was highly significant (P < 0.001). The proportions of total variance associated with CHSD status (at least one deaf ear or no deaf ears) at each of the kennel- and litter-levels from the null model were 0.00 and 0.30, respectively. These results indicate that there was no or little clustering of CHSD amongst dogs from different litters within the same kennel, but there was some clustering of CHSD amongst dogs from the same litter.
Unilateral and bilateral deafness
There was no consistency in the side of unilateral deafness within a litter. In total, 60 dogs in 46 litters from 20 kennels were unilaterally deaf in an identified ear. Twenty-nine dogs were unilaterally deaf in the left ear and 31 were unilaterally deaf in the right ear. In five of the 46 litters, more than one dog was unilaterally deaf. There was no consistency in the side of unilateral deafness within a litter. Three litters each contained two unilaterally deaf dogs. In each of these litters, one pup was deaf in the left ear and one in the right. In a litter with three unilaterally deaf pups, two were deaf in the left ear and one in the right. Finally, in one litter with four unilaterally deaf dogs, three were deaf in the left ear and one in the right. The proportion of total variance associated with side of deafness at litter level from the null model was 0.14. This indicates that there was little clustering of the deaf side amongst unilaterally affected dogs from the same litter.