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

Abstract 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.

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. however, these might also be reduced in dogs with a T3-L3 myelopathy and concurrent spinal shock. [3][4][5] Spinal shock is characterized by reduced segmental spinal reflexes and muscle tone caudal to an injury to the spinal cord, despite local reflex arcs remaining physically intact. 1,[3][4][5][6] For dogs with a T3-L3 myelopathy, this presents as transiently reduced segmental spinal reflexes in the pelvic limbs, giving a false appearance of a lower motor neuron lesion. The condition is thought to be caused by acute injury-associated interruption in descending, primary faciliatory, motor tracts and transient localized changes in the reflex centers of the spinal cord. 3, Currently, prioritization of spinal shock over other causes of reduced pelvic limb reflexes and other lower motor neuron signs is solely based on the clinician's intuition and is highly influenced by training and experience. Specifically, the clinical presentation of spinal shock can lead to clinical confusion in dogs with T3-L3 myelopathies, and might cause the clinician to improperly localize the lesion, exclude the affected area from diagnostic imaging or to prioritize differentials with more guarded prognoses than intervertebral disc disease, such as multifocal inflammatory myelopathies or progressive myelomalacia.
Spinal shock occurs in about 50% of people with acute spinal cord injury, making neuroanatomic localization and assessment of the severity of spinal cord injury a common problem for physicians. 12,13,26,29,32,[35][36][37]52,53 While spinal shock is commonly described in people, and often clinically appreciated in veterinary medicine, only 2 studies have described the condition in dogs with clinical spinal cord injury. 3,4 Developing a better understanding of what clinical criteria are related to T3-L3 myelopathy with concurrent spinal shock vs L4-S3 myelopathy could help guide clinical decision making.
As spinal shock appears to be an underreported phenomenon in veterinary medicine, this study aimed to describe clinical and demographic factors in a large, prospectively identified cohort of dogs with T3-L3 myelopathy with spinal shock and to compare those to a contemporaneous group of dogs with L4-S3 myelopathy. We hypothesized that statistical models could be used to identify clinical variables associated with spinal shock in dogs with spinal cord injuries. This information could aid in diagnostic planning for dogs with thoracic and lumbar myelopathies.

| Animals and study design
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. This database serves as a veterinary disease registry and aims to collect clinical data on dogs with spinal cord injury that are presented to specialty referral institutions. Details related to the database and its sponsor, CANSORT-SCI, have been previously reported. 54 Data for all dogs with a confirmed T3-L3 or L4-S3 myelopathies were originally extracted from the database. Dogs were assigned to 1 of 2 groups: those with clinical signs and imaging findings consistent with a T3-L3 myelopathy with spinal shock, and those with findings suggestive of an L4-S3 myelopathy. Dogs were included in the "T3-L3 with spinal shock" group if they had lesions at or cranial to the L2-L3 disc space and no other structural lesions were identified that could explain 1 or more: reduced pelvic limb reflexes, reduced pelvic limb tone, reduced perineal reflex, or decreased perianal tone, were included in the spinal shock group. Dogs with lesions at or caudal to the L3-L4 disc space were included in the L4-S3 group. Demographic and clinical data elements prospectively collected by the CANSORT-SCI database are summarized in Table 1. 54

| Animals
A total of 518 dogs with spinal cord injury were available in the database for evaluation. Once exclusion criteria were applied, data from 72 dogs remained for inclusion in the present study; 59 dogs with T3-L3 myelopathy and spinal shock and 13 dogs with L4-S3 myelopathy ( Figure 2). Demographic and clinical features of both groups are summarized in Table 2.

| Multivariable analysis
The best fitting model with the smallest AIC included Log e weight, Log e duration, severity score and pelvic limb tone (Table 4). There was F I G U R E 2 Flowchart documenting the numbers of dogs available from the CANSORT database and exclusion criteria applied to identify the T3-L3 myelopathy with spinal shock and L4-S3 myelopathy groups included in the present study T A B L E 2 Demographic information for dogs with T3-L3 myelopathies and spinal shock group and the L4-S3 myelopathies available for inclusion from the CANSORT-SCI database When other independent variables were kept constant, the odds of spinal shock increased with decreasing weight and decreasing duration of clinical signs (Figures 3 and 4). The odds of spinal shock in Severity score ¼ 1 is paraplegic otherwise ¼ 0: Pelvic limb tone ¼ 1 if decreased=absent otherwise ¼ 0: To illustrate common clinical presentations 2 scenarios were put into the model and are listed in Table 5. For example, Scenario 1 evaluates the probability of spinal shock in a 5 kg dog that is presented for an acute onset of paraplegia of less than 12 hours duration, and decreased pelvic limb tone. In that dog, the probability of spinal shock is 97%. In contrast, a 20 kg dog that is presented for ambulatory para- Similarly, these tracts are postulated to provide input to the lower T A B L E 5 Probability of spinal shock for different Log e weight and Log e duration for dogs that are presented with paraparesis vs paraplegia and decreased pelvic limb reflexes or tone In the present study, the most common diagnosis in dogs with spinal shock was an acute intervertebral disc herniation (90%), followed by an acute noncompressive nucleus pulposus extrusion (ANNPE; 8%), and fibrocartilaginous embolic myelopathy (FCEM; 2%).
Acute intervertebral disc herniation (56%) is the most prevalent etiology in dogs with spinal shock, followed by FCE (33%), and ANNPE (11%). 3 All dogs in the L4-S3 myelopathy group were diagnosed with an acute intervertebral disc herniation. This is in contrast to a case series of dogs with spinal shock that found FCE to be the most prevalent (41%), followed by acute intervertebral disc herniation (35%) and ANNPE (24%). 4 The high prevalence of acute intervertebral disc herniations in the present study might have been influenced by the origins of the CANSORT-SCI database, which originally focused on collecting clinical data specifically from dogs with intervertebral disc herniation and was later expanded to capture information from dogs with other causes of spinal cord injury. Alternatively, it could simply reflect that acute intervertebral disc herniation is the most common cause of spinal cord injury in dogs. 56 Limitations of the present study include data collected across multiple referral institutions across North America and Europe. While definitions are provided for data elements within the CANSORT-SCI database, the potential for clinician-dependent variability in interpretation of certain aspects of the neurologic exam such as pelvic limb tone, probably exist. Additionally, as previously mentioned, diagnostic assignments collected in the database do not distinguish between disc extrusion and disc protrusion, which could impact the clinical presentation of acute spinal cord injury observed within a given group. Furthermore, the database only collects information related to initial presentation, diagnosis, and acute treatment. Therefore, we are unable to report information related to longitudinal monitoring of spinal reflexes or other clinical findings throughout recovery. While outside the original aims of the study, serial monitoring of spinal reflexes over time after presentation could assist the clinician further in identifying spinal shock vs L4-S3 myelopathy in dogs with compatible clinical signs. Lastly, further studies are required to assess the external validity of the developed model.
In conclusion, we identified several key clinical variables associated with spinal shock in dogs with spinal cord injuries. Specifically, in dogs presented with an acute onset of signs (<24 hours), a lower body weight (<10 kg), and paraplegia, reduced pelvic limb reflexes and tone are significantly more likely to be related to a T3-L3 myelopathy with spinal shock rather than to the presence of an L4-S3 myelopathy.
These findings should aid clinicians diagnostic planning for dogs presented with signs of neurologic disease caused by spinal cord injury and reduced pelvic limb reflexes and withdrawals.