Cervical vertebral compressive myelopathy is a common cause of general proprioceptive ataxia and paresis in horses.1–4 The disease results from a diverse set of lesions, such as vertebral canal stenosis, vertebral body tipping, and articular process osteophytosis, which alone or in combination lead to extradural spinal cord compression.2–7 Factors involved in development of CVCM likely include vertebral column malformation, osteochondrosis of the articular processes, copper deficiency, dietary zinc excess, high carbohydrate ration, vertebral column instability, and chronic microtrauma.5,8–13
Male horses seem to be at increased risk for CVCM. In a series of reports on CVCM, 125 of 150,14 21 of 25,15 19 of 22,16 and 9 of 1117 horses were male (castrated or sexually intact). The reasons for this apparent overrepresentation are unclear, although increased risk for cervical vertebral column microtrauma caused by sex-based differences in use or activity is a possible explanation. In humans18,19 and dogs20,21 with cervical spondylotic myelopathy, a disease likely analogous to CVCM, a similar male predilection has been suspected.
A variety of breeds have been described as being predisposed to CVCM, including Thoroughbred, Quarter Horse, European Warmblood, and Tennessee Walking Horse.2,14,22,23 With the exception of work done by Mayhew et al2 and Levine et al,16 studies on CVCM have lacked control groups derived from the same hospital for purposes of assessing the significance of apparent breed overrepresentations. For example, the finding that most horses with CVCM diagnosed at an institution are a particular breed might simply reflect the preponderance of that breed in the population of horses referred to that center and not a true breed predilection.24 A study25 comparing CVCM-affected horses (n = 177) with ataxic horses without myelographic compression of the spinal cord (129) did not reveal differences in breed among groups. Thus, evidence regarding breed predilection for CVCM is conflicting. Moreover, explanations for possible breed predispositions to CVCM have been largely speculative, although different use and genetic profiles have been suggested.5,7,26
Traditionally, CVCM is thought to be most common in young horses. Although this is a widely held belief, few studies have investigated age predispositions in CVCM. Age was described as ≤ 4 years in 147 of 150 horses14 and 23 of 25 horses15 in 2 reports, although in 1 smaller study,27 only 8 of 13 horses were ≤ 4 years of age. In a study of 306 ataxic horses, Pageorges et al25 found that although 66% were < 3 years of age, horses with CVCM were no different in age, compared with those with other causes of myelopathy. Recently, CVCM was described in a subset of 22 horses > 4 years of age.16 It has been speculated that the development of CVCM in older horses may relate to chronic microtrauma to the vertebral column, whereas the disease in younger horses may directly result from malformation, nutritional imbalance, and developmental orthopedic disease.5,16
To the authors' knowledge, the association of CVCM with sex, breed, and age of horses has not been systematically evaluated in a large population of horses with CVCM. Thus, the purpose of this case-control study was to use data from horses registered in the VMDB to determine whether the distributions of sex, breed, and age of horses differed significantly between horses with CVCM (case horses) and a group of horses from the same hospitals admitted during the same time period as cases (control horses).
Materials and Methods
The VMDB (consisting of SNOMED and SNVDO)28 was searched for medical records of horses registered between July 1974 and August 2007 with diagnosis codes consistent with CVCM. The SNVDO query included the following codes: wobbler's malformation, wobbler's malarticulation, and cervicospinal arthropathy (wobbler's syndrome). The SNOMED query included the following codes: cervical malformation-malarticulation, cervical myelopathy, spinal stenosis in the cervical region, cervical spine instability, and cervical spinal ataxia. A horse with CVCM (case horse) was defined as one having been assigned one of these diagnostic codes. Only the first admission record for identified cases was used in analysis. A contemporaneous control group was assembled by selecting a horse from the VMDB without CVCM that was admitted on the same date to the same institution as each index case horse. When a control horse was not available for the same date of admission, only the case horse was included. Admission date, admitting institution, sex, breed, age at the time of registration in the database, weight, and discharge status were recorded for each case and control horse.
Data from SNVDO and SNOMED were recoded prior to analysis and included in a single confluent database. Horses in SNOMED described as female, intact or female, spayed; male, intact; and male, castrated were classified as female, male, and gelding, respectively. Horses in SNVDO described as female, entire; female, spayed; and female, unknown were classified as female. Horses in SNVDO described as male, entire and male, castrated were classified as male and gelding, respectively. Breeds for both databases were reclassified as Arabian, Quarter Horse (including Quarter Horse, Quarter Horse–type, and Appaloosa), Standardbred, Tennessee Walking Horse (including Tennessee Walking Horse and Walking Horse), Thoroughbred, Warmblood (including Dutch Warmblood, Hanoverian, Holsteiner, Oldenburg, Swedish Warmblood, Trakehner, and Warmblood), and other. Breed classifications were defined a priori on the basis of apparent overrepresentations noted in other reports.2,3,14,16,22,23 Age was calculated in years for horses in the SNOMED database by subtracting the date of birth from the date of admission. The age of horses in SNOMED was then coded as follows: 0 to < 6 months, 6 months to < 12 months, 12 months to < 24 months, 24 months to < 4 years, 4 years to < 7 years, 7 to < 10 years, and ≥ 10 years. Age data from SNVDO, which were stratified as 0 to < 2 weeks, 2 weeks to < 2 months, 2 months to < 6 months, 6 months to < 12 months, 1 year to < 2 years, 2 years to < 4 years, 4 years to < 7 years, 7 years to < 10 years, 10 to < 15 years, and ≥ 15 years, were recoded in an identical manner as the SNOMED data. Weight data in SNOMED, which were expressed in kilograms as a raw number, were converted to the SNVDO weight strata, which were 0 to ≤ 3 lb, 3 to 15 lb, 15 to 50 lb, 50 to 150 lb, 150 to 300 lb, 300 to 600 lb, 600 to 1,000 lb, 1,000 to 1,300 lb, and > 1,300 lb. Discharge status in SNVDO, which was expressed as alive, diednecropsy, died-no necropsy, euthanatized-necropsy, and euthanatized-no necropsy, was recoded as alive, died, or euthanatized to be confluent with SNOMED terms.
Statistical analysis—Data were analyzed by use of descriptive and inferential methods. For descriptive purposes, categoric data were summarized with contingency tables.29 For inferential analyses, the associations between CVCM and each of the variables sex, breed, age, weight, and outcome were assessed via logistic regression analysis30; ORs were determined by exponentiating coefficients from logistic regression models, and 95% CIs for the OR were derived by use of maximum-likelihood estimators. Because standard logistic regression did not account for the correlation of observations obtained from individuals from the same institution, random effects logistic regression31 was used to estimate the OR (and 95% CI) for the association of case or control status with other variables, wherein institution was modeled as a random effect and other variables (ie, sex, breed, age, weight, and outcome) were modeled as fixed effects.a Variables significantly associated with CVCM were included in multivariable modeling to characterize the association with being a case for a given variable (eg, sex), adjusted for other variables in the model (eg, age and breed). For all analyses, values of P ≤ 0.05 were considered significant.
Results
Records of 1,616 horses from 22 institutions met inclusion criteria (811 CVCM cases and 805 controls). The cluster size of cases and controls contributed by institution ranged from 2 (1 case and 1 control) to 245 (123 cases and 122 controls). The discrepancy in the number of cases and controls contributed by a given institution never exceeded 3.
Significant sex overrepresentations were present in CVCM cases, compared with controls. The odds of being male or gelding (relative to female) were higher (P < 0.001) for cases than controls (Table 1). Moreover, the odds of being a CVCM case were significantly higher for male horses than geldings. Considering males and geldings as a single group, the odds of being a CVCM case were significantly (P < 0.001) greater for males and geldings than for female horses. The magnitude of the SD of the random effect for hospital was small, but its 95% CI did not include 0, indicating that the random effect term was significant.
Results of bivariate random-effects logistic regression analysis of factors potentially associated with CVCM in a study of 811 case horses with CVCM and 805 control horses.
Variable | No. (%) cases | No. (%) controls | OR (95% Cl) | |
---|---|---|---|---|
Sex | 807 | 802 | ||
Gelding | 252 (31) | 263 (33) | 1.6 (1.3–2.0) | < 0.001 |
Sexually intact male | 323 (40) | 150 (19) | 3.6 (2.8–4.6) | < 0.001 |
Geldings and sexually intact males | 575 (71) | 413 (52) | 2.3 (1.9–2.9) | < 0.001 |
Female | 232 (29) | 389 (48) | 1 (Referent) | NA |
Breed | 811 | 805 | ||
Arabian | 27 (3) | 68 (8) | 0.4 (0.3–0.7) | < 0.001 |
Standardbred | 67 (8) | 94 (12) | 0.8 (0.5–1.1) | 0.131 |
Thoroughbred | 262 (32) | 135 (17) | 2.1 (1.6–2.7) | < 0.0001 |
Tennessee Walking Horse | 34 (4) | 16 (2) | 2.3 (1.2–4.2) | 0.009 |
Warmbloods | 52 (6) | 37 (4) | 1.5 (< 1.0–2.4) | 0.078 |
Other breeds | 79 (10) | 144 (18) | 0.6 (0.4–0.8) | 0.001 |
Quarter Horse | 291 (36) | 311 (39) | 1 (Referent) | NA |
Age | 799 | 794 | ||
< 6 mo | 46 (6) | 84 (11) | 2.1 (1.3–3.3) | 0.002 |
6 to 11 mo | 66 (8) | 32 (4) | 7.9 (4.7–13.2) | < 0.001 |
12 to 23 mo | 233 (29) | 54 (7) | 16.5 (10.9–24.9) | < 0.001 |
2 to < 4 y | 258 (32) | 146 (18) | 6.8 (4.7–9.6) | < 0.001 |
4 to < 7 y | 115 (14) | 149 (19) | 2.9 (2.0–4.3) | < 0.001 |
7to 10 y | 34 (4) | 115 (14) | 1.1 (0.7–1.8) | 0.623 |
10 y | 47 (6) | 214 (27) | 1 (Referent) | NA |
Weight* | 348 | 321 | ||
< 300 Ib | 17 (5) | 38 (12) | 1 (Referent) | NA |
300 to < 600 Ib | 35 (10) | 27 (8) | 2.9 (1.3–6.2) | 0.007 |
600 to < 1,000 Ib | 137 (39) | 91 (28) | 3.4 (1.8–6.3) | < 0.001 |
1,000 to <1,300 Ib | 146 (42) | 135 (42) | 2.4 (1.3–4.5) | 0.005 |
≥ 1,300 Ib | 13 (4) | 30 (9) | 1.0 (0.4–2.3) | 0.943 |
Discharge status | 811 | 805 | ||
Alive | 516 (64) | 700 (87) | 1 (Referent) | NA |
Died | 19 (2) | 25 (3) | 1.0 (0.6–1.9) | 0.922 |
Euthanatized | 276 (34) | 80 (10) | 4.7 (3.6–6.2) | < 0.001 |
System international units for weight ranges are < 136.4 kg, 136.4 to 272.3 kg, 272.7 to 454.1 kg, 454.5 to 590.5 kg, and > 590.9 kg, respectively.
Referent = Referent category for comparison within a variable. NA= Not applicable.
The distribution of breeds of CVCM horses (n = 811) and controls (805) was summarized (Table 1). Relative to Quarter Horses, the odds of having CVCM were significantly greater for Thoroughbreds and Tennessee Walking Horses (P < 0.001 and P = 0.009, respectively) and were significantly lower for Arabians and horses categorized as other breeds (P < 0.001 and P = 0.001, respectively). Although not significant (P = 0.078), the odds of cases being Warmbloods were greater than for controls. The magnitude of the SD of the random effect for hospital was small, but its 95% CI did not include 0, indicating that the random effect term was significant.
Age was significantly associated with CVCM (Table 1). Relative to horses ≥ 10 years of age, horses < 6 months, 6 to < 12 months, 12 to < 24 months, 2 to < 4 years, and 4 to < 7 years of age had significantly higher odds of being in the CVCM group. The age groups with the highest odds of being cases of CVCM were horses 12 to < 24 months (OR = 16.5; P < 0.001), 6 to < 12 months (OR = 7.9; P < 0.001), and 2 to < 4 years (OR = 6.8; P < 0.001) of age. There was no significant difference between the proportions of cases and controls missing weight data. Horses weighing 300 to < 600 lb, 600 to < 1,000 lb, and 1,000 to < 1,300 lb all were significantly more likely to be in the CVCM group than the reference population (Table 1). Horses weighing 600 to < 1,000 lb had the highest odds of being in the CVCM group (P < 0.001).
Outcome data were obtained for all cases and controls (Table 1). Necropsy was performed in 164 of 276 (59%) horses with CVCM that were euthanatized and 17 of 19 (89%) that died. Necropsy was performed in 32 of 80 (40%) control horses that were euthanatized and 20 of 25 (80%) that died. Random effects logistic regression indicated that cases had 4.7 times the odds (95% CI, 3.6 to 6.2) of being euthanatized, compared with controls (P < 0.001).
A multivariate logistic regression model was built to account for confounding among signalment variables (Table 2). Males and geldings were significantly (P < 0.001) more likely to be in the CVCM group. Thoroughbreds and Tennessee Walking Horses were significantly (P < 0.001 and P = 0.019, respectively) more likely than Quarter Horses to be in the CVCM group, with ORs of 1.7 (95% CI, 1.3 to 2.3) and 2.3 (95% CI, 1.1 to 4.7), respectively. Warmbloods had 1.9 times the odds (95% CI, 1.1 to 3.1) of being in the CVCM group, compared with Quarter Horses (P = 0.020). Arabians and Standardbreds were less likely (P = 0.035 and P < 0.001, respectively) to be in the CVCM group than Quarter Horses. Significant age relationships as observed in bivariate analysis were also evident in the multivariate model (Table 2). A separate multivariate logistic regression analysis was also performed that only included data from CVCM horses and control horses with necropsy (n = 233). The results were similar regarding the significance and magnitude of associations detected in the larger data subset, except that the significance of the Warmblood overrepresentation in the CVCM group was lost (P = 0.093).
Multivariable model of association of signalment factors with odds of having CVCM in a study of 1,618 horses from 22 veterinary teaching hospitals in North America.
OR (95% Cl) | Variable | |
---|---|---|
Sex | ||
Gelding | 2.0 (1.5–2.6) | < 0.001 |
Sexually intact male | 2.4 (1.8–3.2) | < 0.001 |
Female | 1 (Referent) | NA |
Breed | ||
Arabian | 0.6 (0.3–0.9) | 0.0345 |
Standard bred | 0.5 (0.3–0.7) | < 0.001 |
Thoroughbred | 1.7 (1.3–2.3) | < 0.001 |
Tennessee Walking Horse | 2.3 (1.1–4.7) | 0.019 |
Warmblood | 1.9 (1.1–3.1) | 0.020 |
Other breeds | 0.6 (0.4–0.8) | 0.006 |
Quarter Horse | 1 (Referent) | NA |
Age | ||
< 6 mo | 2.4 (1.4–3.9) | < 0.001 |
6 to 11 mo | 6.6 (3.8–11.5) | < 0.001 |
12 to 23 mo | 16.4 (10.5–25.8) | < 0.001 |
2 to < 4 y | 7.2 (4.9–10.5) | < 0.001 |
4 to < 7 y | 3.1 (2.1–4.6) | < 0.001 |
7 to 10 y | 1.1 (0.7–1.8) | 0.6548 |
≥ 10y | 1 (Referent) | NA |
See Table 1 for key.
Discussion
Cervical vertebral compressive myelopathy is a well recognized cause of cervical spinal cord signs in horses. Although the histopathologic features associated with this disorder and certain aspects of the etiopathogenesis have been characterized, there has been limited investigation into sex, age, breed, and weight predispositions.
Several reports have suggested that CVCM may be more common in males and geldings, compared with females. Data from the present study indicated that horses with CVCM were more than twice as likely to be either geldings or males than were controls. In most species, males seem to be predisposed to spinal cord injury. This has been detected, for example, in dogs with thoracolumbar disk herniation,32 humans with traumatic myelopathy,33,34 and humans with spondylotic myelopathy.18,19 The reasons for male overrepresentation in horses with CVCM and other species with spinal cord injury remain unclear. Several authors believe that males may have different behavioral patterns (and in horses, different uses), which predispose the vertebral column to chronic microtrauma or acute traumatic events.16,33,34 Data from experimental models also suggest that sex hormones may play a direct role in susceptibility to spinal cord injury. In rodent spinal cord injury models, progesterone increases the expression of brainderived neurotrophic factor and myelin basic protein, both of which may be neuroprotective.35–37 In rats with experimentally induced spinal cord injury, 17E-estradiol improves locomotor scores, increases white matter sparing, and reduces neuronal apoptosis.38 Sex may also directly affect vertebral shape and vertebral canal diameter. In clinically normal suckling and weanling foals, vertebral canal diameter as assessed by use of corrected minimum sagittal diameter is smaller in males than females. Although a study39 in humans does not reveal sex-based differences in vertebral shape, others suggest that males have more vertebral wedging and concavity than females.40,41
Multivariate logistic regression analysis used in this study indicated that Thoroughbreds, Warmbloods, and Tennessee Walking Horses were significantly more likely to have CVCM than Quarter Horses; Standardbreds and Arabians were significantly less likely to have CVCM. Familial or breed overrepresentations are not unique to equine CVCM and are also reported in cervical spondylotic myelopathy in humans and dogs.20,42 The etiopathogenesis of breed predilections in CVCM is unknown but is speculated to involve genetic factors and differences in morphometry and use.2,5,22 Despite the hypothesis that breed overrepresentations may have a genetic basis, results of 2 small studies26,43 do not indicate that foals from horses with CVCM are more likely to develop the disease. Although Thoroughbreds are commonly suggested to be predisposed to developing CVCM, until recently, Warmbloods and Tennessee Walking Horses were not identified as such.2,5,16 Similarly, the lack of overrepresentation of Quarter Horses in the CVCM group in our report is interesting because many authors have suggested this breed to be particularly susceptible.22 Although use of Quarter Horses as the reference group also impacted results; had Arabians been selected as the reference category, the odds of cases being Quarter Horses would have been significantly greater than controls. With the exception of 2 reports2,16 with small case populations, previous data on breed overrepresentation were either obtained without a contemporaneous control group or were based on clinical impressions. Considering the large number of horses, sourcing of data from multiple institutions, and multivariate logistic regression analysis that should reduce confounding, the authors believe that breed distributions in the present report are likely to be accurate. The association of certain breeds with CVCM indicates the need for further evaluation of a genetic contribution to this disorder.
Generally, CVCM has been considered to be most common among horses ≤ 4 years of age. Moore et al44 reported mean ages of 1.4 years (range, 0.25 to 3.5 years) and 1.5 years (range, 0.5 to 6 years) in 2 groups comprising 61 horses with CVCM that were treated surgically. In a study by Mayhew et al,2 11 horses with CVCM had a mean age of 1.2 years (range, 0.5 to 2.5 years). Age data in the present study could be reported only as ranges because of the manner in which data were recorded in the VMDB. All horses in age categories < 7 years had significantly greater odds of being in the CVCM group, compared with horses ≥ 10 years of age. Horses that were 12 to < 24 months of age had the highest odds of being in the CVCM group. Although horses < 6 months and 4 to < 7 years of age were still significantly more likely to be in the CVCM group, the magnitudes of the respective ORs for these categories were not as large as those for horses in age categories between 6 months and < 4 years. Young horses are likely predisposed to CVCM because the disease results in part from congenital malformation-malarticulation, osteochondrosis, and nutritional imbalance during bone growth.2,5,8–11,44 Our data clearly indicated, however, that CVCM can occur in horses > 4 years of age, and even horses in the 4 to < 7 years of age range were overrepresented, compared with horses ≥ 10 years of age. It has been postulated that the development of CVCM in horses > 4 years of age occurs because of progression of subclinical malformation-malarticulation, chronic vertebral column microtrauma, or a combination of these influences.16
Horses weighing 600 to < 1,000 lb had the highest odds of being in the CVCM group, compared with those ≥ 1,300 lb. Horses in the 300 to < 600 lb and 1,000 to < 1,300 lb ranges were also more likely to be in the CVCM group than the reference population. Mayhew et al2 have suggested that horses < 320 kg (704 lb) have smaller vertebral canal diameters than those > 320 kg; in humans with cervical spondylotic myelopathy and horses with CVCM, small vertebral canal diameter is likely related to development of compressive lesions.23,45,46 Conclusions regarding weight and risk for developing CVCM have to be made carefully, however, for 2 reasons. First, weight data were missing for almost 60% of the study population. Second, the observed association of weight with CVCM may have been confounded by age, breed, or sex. The odds of CVCM were higher in younger horses, and younger horses generally were in the lighter weight categories; thus, the apparent association of weight with CVCM might have been explained by effects of age. Similarly, certain breeds and male sex may be associated with greater weight. Adjusting for effects of age, breed, and sex in multivariate random effects logistic regression analysis to evaluate the association of weight with CVCM was deemed of limited value because of the large proportion of horses for which weight data were missing. Future studies to examine the importance of weight, adjusted for effects of age, breed, and sex, seem warranted because it is plausible that weight influences the risk of CVCM independent of potential confounders.
Limited information could be obtained from the VMDB with regard to outcome. When horses were classified as alive, died, or euthanatized at the time of discharge, CVCM cases had 4.7 times higher odds of being euthanatized than controls. Whether the high proportion of CVCM horses euthanatized actually reflected a poor prognosis associated with the disease is unclear. Euthanasia of affected horses may occur for other reasons, including client or veterinarian beliefs and perceptions regarding the prognosis associated with CVCM; the risks posed by affected horses to themselves, other horses, and those handling them; economic considerations; and restrictions in use of horses affected by CVCM.
The major limitations of this study included the retrospective nature of data acquisition, misclassification of cases and controls because the diagnosis of CVCM was presumptive for many of the horses coded as affected in the VMDB, and the limitations of the control population. Although a prospective, multi-institutional study would provide more definitive information, a retrospective study47 that used a large number of cases and controls from an existing database derived from multiple clinical centers was deemed to offer important information, especially when the focus of the study did not pertain to outcome after treatment. Prospective investigations are challenging to perform for diseases such as CVCM because cumulative incidence at a given institution can be low, necessitating the contribution of data from multiple centers during a long period.
Some degree of misclassification of cases and controls will occur for any disease for which a perfectly accurate diagnostic test does not exist, and this was likely to have occurred in this study as a result of using a case definition of CVCM based on diagnostic codes for stored data. We believe that misclassification had limited influence on study findings for the following reasons. First, the magnitudes of the ORs were large (> 3.5) for many of the bivariate associations and > 2 for multivariate associations. Thus, the misclassification bias would have to be fairly large to explain the observed associations. Second, the observed associations were consistent with previous controlled studies.2,14,16,44 Third, nondifferential misclassification usually leads to a loss of significant associations and shifting of ORs toward 1.0 (ie, toward the null hypothesis of no association).48 Fourth, results of multivariate analysis restricted to horses that underwent necropsy examinations and thus presumably were accurately classified with respect to disease or were alive at discharge were essentially identical in magnitude and significance of the observed OR for all variables.
Identification of an ideal control population for case-control studies is challenging, and limitations exist for most populations of controls.47,49 It is conceivable that management conditions, such as feeding practices, nutritional factors, and exercise regimens, contribute to CVCM. Use of hospital controls may introduce bias because horses evaluated for disease at referral hospitals might differ with respect to management conditions predisposing to CVCM, whereas controls from the same farm might be more similar with respect to these factors. Hospital-based controls also may be associated with other biases. For example, it is possible that the observed association of Arabians with lower odds of CVCM simply occurred because these horses were more likely to be referred for other problems (such as colic) than because they were less likely to have CVCM. Consideration of other control populations, such as horses referred for other neurologic diseases, or inclusion of multiple control populations are options that could have been used to help assess the effect of any bias introduced by including controls selected from horses admitted to hospitals. The small discrepancy between the number of cases (811) and controls (805) arose because of the means of control selection and was unlikely to have introduced bias.
ABBREVIATIONS
CI | Confidence interval |
CVCM | Cervical vertebral compressive myelopathy |
OR | Odds ratio |
SNOMED | Systemized Nomenclature of Medicine |
SNVDO | Standard Nomenclature of Veterinary Diseases and Operations |
VMDB | Veterinary Medical Database |
GLME Procedure, S-PLUS, version 8.0, Insightful Inc, Seattle, Wash.
References
- 1.
Nout YS, Reed SM. Cervical vertebral stenotic myelopathy. Equine Vet Educ 2003;15:212–223.
- 2.↑
Mayhew IG, deLahunta A, Whitlock RH, et al . Spinal cord disease in the horse. Cornell Vet 1978;68(suppl 6):1–207.
- 3.
Mayhew IG. Large animal neurology: a handbook for veterinary clinicians. Philadelphia: Lea & Febiger, 1989.
- 4.
Rush BR. Cervical stenotic myelopathy. In: Smith BP, ed. Large animal internal medicine. 3rd ed. St Louis: Mosby, 2002;971–977.
- 5.
van Biervliet J, Mayhew J, de Lahunta A. Cervical vertebral compressive myelopathy: diagnosis. Clin Tech Equine Pract 2006;5:54–59.
- 6.
Rantanen RW, Gavin PR, Barbee DD, et al . Ataxia and paresis in horses. Part II. Radiographic and myelographic examination of the cervical vertebral column. Compend Contin Educ Pract Vet 1981;3:S161–S172.
- 7.
Wagner PC, Grant BD, Reed SM. Cervical vertebral malformations. Vet Clin North Am Equine Pract 1987;3:385–396.
- 8.
Stewart RH, Reed SM, Weisbrode SE. Frequency and severity of osteochondrosis in horses with cervical stenotic myelopathy. Am J Vet Res 1991;52:873–879.
- 9.
Donawick WJ, Mayhew IG, Galligan DT, et al . Results of a lowprotein, low-energy diet and confinement in young horses with wobbles. Proc Am Assoc Equine Pract 1993;39:125–127.
- 10.
Gabel AA, Knight DA, Reed SM, et al . Comparison of incidence and severity of developmental orthopedic disease on 17 farms before and after adjustment of ration. Proc Am Assoc Equine Pract 1987;33:163–170.
- 11.
Reed SM, Knight DA, Weisbrode SE, et al . The relationship of cervical vertebral malformation to developmental orthopedic disease. Proc Am Assoc Equine Pract 1987;33:139–142.
- 12.
Knight DA, Gabel AA, Reed SM. Correlation of dietary mineral to incidence and severity of metabolic bone disease in Ohio and Kentucky. Proc Am Assoc Equine Pract 1985;31:445–461.
- 13.
Kronfeld DS, Meacham TN, Donoghue S. Dietary aspects of developmental orthopedic disease. Vet Clin North Am Equine Pract 1990;6:451–465.
- 14.↑
Moore BR, Reed SM, Biller DS, et al . Assessment of vertebral canal diameter and bony malformations of the cervical part of the spine in horses with cervical stenotic myelopathy. Am J Vet Res 1994;55:5–13.
- 15.↑
Powers BE, Stashak TS, Nixon AJ, et al . Pathology of the vertebral column of horses with cervical static stenosis. Vet Pathol 1986;23:392–399.
- 16.↑
Levine JM, Adam E, MacKay RJ, et al . Confirmed and presumptive cervical vertebral compressive myelopathy in older horses: a retrospective study (1992–2004). J Vet Intern Med 2007;21:812–819.
- 17.
Wagner PC, Grant BD, Gallina A, et al . Ataxia and paresis in horses part III. Surgical treatment of cervical spinal cord compression. Compend Contin Educ Pract Vet 1981;3:S192–S202.
- 18.
Wang B, Liu H, Wang H, et al . Segmental instability in cervical spondylotic myelopathy with severe disc degeneration. Spine 2006;31:1327–1331.
- 19.
Chibbaro S, Benventui L, Carnescchi S, et al . Anterior cervical corpectomy for cervical spondylotic myelopathy: experience and surgical results in a series of 70 consecutive patients. J Clin Neurosci 2006;13:233–238.
- 20.
VanGundy T. Canine wobbler syndrome. Part I. Pathophysiology and diagnosis. Compend Contin Educ Pract Vet 1989;11:144–157.
- 21.
Read RA, Robins GM, Carlisle CH. Caudal cervical spondylomyelopathy (wobbler syndrome) in the dog: a review of 30 cases. J Small Anim Pract 1983;24:605–621.
- 22.↑
Nixon AJ. Surgical management of equine cervical vertebral malformation. Prog Vet Neurol 1991;2:183–195.
- 23.
Mayhew IG, Donawick WJ, Green SL, et al . Diagnosis and prediction of cervical vertebral malformation in thoroughbred foals based on semiquantitative radiographic indicators. Equine Vet J 1993;25:435–440.
- 24.↑
Gordis L. Measuring the occurrence of disease: I. Morbidity. In: Epidemiology. 3rd ed. Philadelphia: Elsevier Saunders, 2004;32–47.
- 25.↑
Pageorges M, Gavin PR, Sande RD, et al . Radiographic and myelographic examination of the cervical vertebral column in 306 ataxic horses. Vet Radiol 1987;28:53–59.
- 26.
Wagner PC, Grant BD, Watrous BJ, et al . A study of the heritability of cervical vertebral malformation in horses. Proc Am Assoc Equine Pract 1985;31:43–50.
- 27.↑
Yovich JV, LeCouteur RA, Gould DH. Chronic cervical compressive myelopathy in horses: clinical correlations with spinal cord alterations. Vet J 1991;68:326–334.
- 28.↑
Veterinary Medical Database [database online]. Urbana, Ill: Veterinary Medical Database, 2006. Available at: www.vmdb.org. Accessed Sep 10, 2007.
- 29.↑
Rosner BA. Hypothesis testing: categorical data. Nonparametric methods. In: Fundamentals of biostatistics. Boston: Duxbury Press, 1986;302–357.
- 31.↑
Breslow N, Clayton DG. Approximate inference in generalized linear mixed models. J Am Stat Assoc 1993;88:9–25.
- 32.↑
Priester WA. Canine intervertebral disk disease—occurrence by age, breed, and sex among 8,117 cases. Theriogenology 1976;6:293–303.
- 33.
Jackson AB, Dijkers M, Devivo MJ, et al . A demographic profile of new traumatic spinal cord injuries: change and stability over 30 years. Arch Phys Med Rehabil 2004;85:1740–1748.
- 34.
Pickett GE, Campos-Benitez M, Keller JL, et al. Epidemiology of traumatic spinal cord injury in Canada. Spine 2006;31:799–805.
- 35.
De Nicola AF, Gonzalez SL, Labombarda F, et al. Progesterone treatment of spinal cord injury. J Mol Neurosci 2006;28:3–15.
- 36.
Gonzalez SL, Labombarda F, Deniselle MC, et al . Progesterone neuroprotection in spinal cord trauma involves up-regulation of brain-derived neurotrophic factor in motoneurons. J Steroid Biochem Mol Biol 2005;94:143–149.
- 37.
Labombarda F, Gonzalez S, Deniselle MC, et al . Progesterone increases the expression of myelin basic protein and the number of cells showing NG2 immunostaining in the lesioned spinal cord. J Neurotrauma 2006;23:181–192.
- 38.↑
Chaovipoch P, Jelks KA, Gerhold LM, et al . 17beta-estradiol is protective in spinal cord injury in post- and pre-menopausal rats. J Neurotrauma 2006;23:830–852.
- 39.↑
Boyle JJ, Singer KP, Milne N. Morphological survey of the cervicothoracic junctional region. Spine 1996;21:544–548.
- 40.
Goh S, Price RI, Song S, et al . Magnetic resonance-based vertebral morphometry of the thoracic spine: age, gender and level-specific influences. Clin Biomech (Bristol, Avon) 2000;15:417–425.
- 41.
Grados F, Fardellone P, Benammar M, et al . Influence of age and sex on vertebral shape indices assessed by radiographic morphometry. Osteoporos Int 1999;10:450–455.
- 42.
Yoo K, Origitano TC. Familial cervical spondylosis. Case report. J Neurosurg 1998;89:139–141.
- 43.
Falco MJ, Whitwell K, Palmer AC. An investigation into the genetics of ‘wobbler disease’ in thoroughbred horses in Britain. Equine Vet J 1976;8:165–169.
- 44.↑
Moore BR, Reed SM, Robertson JT. Surgical treatment of cervical stenotic myelopathy in horses: 73 cases (1983–1992). J Am Vet Med Assoc 1993;203:108–112.
- 45.
Mayhew IG. Radiographic diagnosis of equine cervical vertebral malformation, in Proceedings. 20th Am Coll Vet Intern Med 2002;382–383.
- 46.
Ogino H, Tada K, Okada K, et al . Canal diameter, anteroposterior compression ratio, and spondylotic myelopathy of the cervical spine. Spine 1983;8:1–15.
- 47.↑
Fisher CG, Wood KB. Introduction to techniques of evidencebased medicine. Spine 2007;32:S66–S72.
- 48.↑
Gordis L. More on causal inferences: bias, confounding, and interaction. Epidemiology. 3rd ed. Philadelphia: Elsevier Saunders, 2004.
- 49.
Gordis L. Case-control and cross-sectional studies. Epidemiology. 3rd ed. Philadelphia: Elsevier, 2004;159–176.