Developing a predictive model for spinal shock in dogs with spinal cord injury

Background: Reduced pelvic limb reflexes in dogs with spinal cord injury typically suggests a lesion of the L4-S3 spinal cord segments. However, pelvic limb reflexes might also be reduced in dogs with a T3-L3 myelopathy and concurrent spinal shock.

Hypothesis/objectives: We hypothesized that statistical models could be used to identify clinical variables associated with spinal shock in dogs with spinal cord injuries.

Animals: Cohort of 59 dogs with T3-L3 myelopathies and spinal shock and 13 dogs with L4-S3 myelopathies.

Methods: Data used for this study were prospectively entered by partner institutions into the International Canine Spinal Cord Injury observational registry between October 2016 and July 2019. Univariable logistic regression analyses were performed to assess the association between independent variables and the presence of spinal shock. Independent variables were selected for inclusion in a multivariable logistic regression model if they had a significant effect (P ≤ .1) on the odds of spinal shock in univariable logistic regression.

Results: The final multivariable model included the natural log of weight (kg), the natural log of duration of clinical signs (hours), severity (paresis vs paraplegia), and pelvic limb tone (normal vs decreased/absent). The odds of spinal shock decreased with increasing weight (odds ratio [OR] = 0.28, P = .09; confidence interval [CI] 0.07-1.2), increasing duration (OR = 0.44, P = .02; CI 0.21-0.9), decreased pelvic limb tone (OR = 0.04, P = .003; CI 0.01-0.36), and increased in the presence of paraplegia (OR = 7.87, P = .04; CI 1.1-56.62).

Conclusions and clinical importance: A formula, as developed by the present study and after external validation, could be useful for assisting clinicians in determining the likelihood of spinal shock in various clinical scenarios and aid in diagnostic planning.