Individuals sustaining damage to the spinal cord due to non-traumatic causes are often treated in specialized inpatient SCI rehabilitation centers more commonly associated with those with SCI due to traumatic etiologies. Various reports have estimated that one-quarter to one-half of all cases seen in specialized SCI rehabilitation centers are associated with non-traumatic etiologies (McKinley et al. 1999; Muslumanoglu et al. 1997; van der Putten et al. 2001). Despite these significant numbers, relatively little systematic research is directed at non-traumatic SCI (McKinley et al. 2002; van der Putten et al. 2001). Common causes of non-traumatic SCI include space-occupying lesions such as tumours or prolapsed intervertebral discs, spondylosis such as that seen with degenerative spinal changes resulting in compression of the spinal cord, vascular ischemia as in arteriovenous malformations or spinal infarction, inflammation (e.g., idiopathic transverse myelitis, tropical spastic paraparesis, sarcoid) and those associated with congenital or familial etiologies (Adams & Salam-Adams 1991; McKinley et al. 2001; McKinley et al. 1999). Although estimates of the incidence of non-traumatic SCI have been provided (e.g., 8 per 100,000) (Kurtzke 1975; McKinley et al. 2001), it is difficult to ensure accuracy given the heterogeneous nature of non-traumatic SCI and the variety of facilities and programs where these patients may receive care.
Studies comparing those with damage to the spinal cord due to non-traumatic versus traumatic etiologies have demonstrated a variety of systematic differences between these 2 patient groups. In general, those with non-traumatic SCI are more likely to be older, female, have paraplegia, and have an incomplete injury than those with traumatic SCI (McKinley et al. 1996; McKinley et al. 2001; McKinley et al. 2002; New 2005).
Studies examining non-traumatic SCI typically make use of retrospective case series designs describing rehabilitation outcomes directly (Citterio et al. 2004; McKinley et al. 1996; New 2005; New 2006; van der Putten et al. 2001) or involve case control designs employing matching techniques to make comparisons with traumatic SCI while controlling for such things as age and level and completeness of injury (McKinley et al. 2008; McKinley et al. 1999; McKinley et al. 2001; McKinley et al. 2002). As noted above, those with non-traumatic SCI were more likely to be older, female, have paraplegia, and have an incomplete injury than those with traumatic SCI (McKinley et al. 1996; McKinley et al. 2001; McKinley et al. 2002; New 2005). No difference in age, marriage, education, socioeconomic factors, length of stay, and functional outcome was reported for a case control analysis originating from India ((Gupta et al. 2008), N=76).
Patients with non-traumatic SCI (Citterio et al. 2004; McKinley et al. 1996) or traumatic SCI (Dionne et al. 2020; Franceschini et al. 2020; Halvorsen et al. 2019b) were primarily discharged home after rehabilitation. Citterio et al. (2004) found that discharge to home was predicted by many factors including marital status, completeness of injury, clinical improvement, efficient bowel and bladder management, and absence of pressure ulcers. Another important predictor was shown to be a longer length of stay. This was due to the finding that there is an increased probability of functional and neurological improvement after a long hospital stay. Dionne et al. (2020) found that failure to return home was predicted by living alone, higher neurological level of injury, and comorbidities. In contrast to Citterio et al. (2004), longer acute length of stay and longer rehabilitation stay were associated with failure to return home, suggesting more serious injury with greater length of stay.
Ones et al. (2007) and Yokoyama et al. (2006) showed no significant difference in length of stay between individuals with traumatic versus non-traumatic spinal cord injury. Conversely, when direct comparisons of traumatic and non-traumatic SCI of various etiologies have been conducted using matching procedures, it is clear that shorter rehabilitation LOS was seen for those with non-traumatic SCI (McKinley et al. 2001; Osterthun et al. 2009). In addition, this shorter LOS was associated with reduced hospital charges on both an overall and a per diem basis (McKinley et al. 2001). These findings were replicated with similar studies examining subsets of those with non-traumatic SCI including those with stenosis (McKinley et al. 2002) and those with neoplastic cord compression (McKinley et al. 1999) although this was not the case for a review involving infection-based SCI (McKinley et al. 2008). Most of these findings have been established with data from the United States Model Systems, although at least two reports from other jurisdictions have reported longer rehabilitation LOS (New 2005; van der Putten et al. 2001).
None of the studies employing matching procedures noted differences in discharge destinations for those with non-traumatic SCI as compared to those with traumatic SCI (McKinley et al. 1999; McKinley et al. 2001; McKinley et al. 2002). Although New et al. (2005) did note that among those with non-traumatic SCI, individuals who were male, younger, more mobile, more independent with bowel and bladder function, and had less severe AIS grades were more likely to be discharged home. In addition, the relatively poor prognosis and low survival rate of those with neoplastic cord compression have specific implications for discharge disposition (McKinley et al. 1996) although no specific differences were noted in a matched comparison (McKinley et al. 1999).
Comparing the rehabilitation of individuals with traumatic SCI with or without concomitant TBI, Bradbury et al. (2008) reported no significant differences in LOS and FIM change scores. However, the presence of dual diagnoses was deemed to result in clinically, but not statistically, significantly greater costs associated with the FIM change score.
All studies reviewed employed the FIM to assess the functional status of individuals and generally demonstrated improved function with rehabilitation. Typically, motor FIM scores were employed, or in the event total FIM scores were used it was acknowledged that changes were due primarily to the motor FIM subscale given a ceiling effect associated with the cognitive FIM subscale (McKinley et al. 1999; New 2005). There was conflicting evidence in admission and discharge FIM scores between traumatic and non-traumatic SCI groups. A study by Ones et al. (2007) found patients with traumatic SCI had significantly lower admission FIM scores than those with non-traumatic SCI. However, other studies found no such trend (McKinley et al. 1999; McKinley et al. 2001). FIM discharge scores were shown to be lower in the non-traumatic SCI population than traumatic (McKinley et al. 1999; McKinley et al. 2001) while Ones et al. (2007) showed no such difference. When examining only those with stenosis versus those with traumatic SCI, those with non-traumatic SCI had higher FIM scores on admission, and similar scores on discharge, resulting in reduced change scores and lower efficiency (McKinley et al. 2002). On the other hand, those with neoplastic cord compression demonstrated similar FIM scores on admission, and reduced scores on discharge, resulting in reduced change scores but no difference in efficiency (McKinley et al. 1999).
McKinely et al. (1999) and McKinely et al. (2001) found no significant difference between traumatic versus non-traumatic SCI populations in FIM efficiency. However, Ones et al. (2007) showed a significantly higher FIM efficiency for persons with a traumatic as compared to a non-traumatic etiology. Given this and other conflicting findings in these studies, it seems that it is especially important to appreciate the heterogeneity inherent in rehabilitation outcomes of persons with non-traumatic etiologies. In particular, much variation might be expected, especially between center-based reports with relatively small sample sizes and which include various non-traumatic etiologies within a single non-traumatic grouping. Future research should focus on large-scale, case control methodologies employing subject matching strategies that control for potential confounding variables or that examine the effect of potential mediating variables. It is also important to consider logical subgroups based on specific etiologies of non-traumatic SCI.
Van der Putten (2001) assessed a variety of factors using multiple linear regression techniques in order to predict those most associated with increases in FIM motor scores during rehabilitation. They included 100 consecutively admitted patients with non-traumatic SCI with rehabilitation periods of more than 1 week. The primary factors associated with improved motor FIM scores accounting for 54% of the variance were having a lower score on admission and reduced time between symptom onset to admission. Age, specific diagnostic subgroup (i.e., space-occupying, vascular, spondylosis, inflammation, or hereditary), or lesion level did not improve the prediction significantly.
In line with the topic of this discussion, guidelines established by Rapidi and colleagues (2018) for Physical and Rehabilitation Medicine suggest that therapeutic exercise programs in SCI are prescribed and adapted to SCI persons’ needs, according to the neurological level of injury, age, and comorbidity.
There is level 3 evidence (from five case control studies: McKinley et al. 2008; McKinley et al. 1999; McKinley et al. 2001; McKinley et al. 2002; Yokoyama et al. 2006) that those with non-traumatic SCI have generally reduced rehabilitation LOS and reduced hospital charges.
There is level 3 evidence (from one case control study: Dionne et al. 2020); level 4 evidence (from two case series: Citterio et al. 2004; McKinley et al. 1996); and level 5 evidence (from two observational studies: Franceschini et al. 2020; Halvorsen et al. 2019a) that those with non-traumatic SCI have similar discharge destinations as compared to those with traumatic SCI.
There is conflicting level 3 evidence (from seven case control studies: Gupta et al. 2008; McKinley et al. 2008; McKinley et al. 2001; McKinley et al. 2002; Ones et al. 2007; Yokoyama et al. 2006) that individuals with non-traumatic SCI may experience less functional gains than those with traumatic SCI, although many studies are comparing persons with different etiologies of non-traumatic SCI.
There is level 3 evidence (from one case control study: Bradbury et al. 2008) that individuals with traumatic SCI with or without concomitant traumatic brain injury have similar LOS and achieve similar FIM motor scores, but associated costs were higher in those with dual diagnosis.
There is level 4 evidence (from one case series study: van der Putten et al. 2001) that those with non-traumatic SCI are more likely to be older, female, have paraplegia and have an incomplete injury as compared to those with traumatic SCI.
There is conflicting level 4 evidence (from four case series: Citterio et al. 2004; Gupta et al. 2009; McKinley et al. 1996; New 2005) that patients with non-traumatic SCI recover significant neurological and functional improvements following rehabilitation.