Introduction

Chronic pain limits activities, decreases quality of life, and leads to significant impairment in individuals with spinal cord injury (SCI) [1,2,3]. Individuals with pain after SCI have a $22,545 increased cost burden per year compared to their SCI peers without pain [4] and the difficulty in treating this pain has been documented for years [5]. Pain after SCI is typically classified as nociceptive (which includes visceral and musculoskeletal pain), at-level-neuropathic, and below level neuropathic [6]. Neuropathic pain is generally regarded as the most frequent type of pain after SCI [7], although this remains disputed, with studies having up to 59% of those with SCI reporting musculoskeletal nociceptive pain [8]. The prevalence of chronic pain in this population varies from 13–96% depending on the study [3, 8,9,10,11] and of severe pain from 20–58% [8, 12]. The International SCI Pain Classification System was developed in 2009, but experts were unable to estimate the prevalence of pain after SCI due to the variability between studies, suggesting the need for more and better data [6] which has since been reiterated [7, 13]. Some data suggest incomplete lesions result in more chronic pain [8, 11, 14, 15], though other studies paradoxically suggest complete lesions result in more chronic pain [16,17,18]. A metanalysis recently found no difference between groups [13]. Many studies suggest the level of injury does not affect the prevalence of chronic pain, although others have suggested lumbosacral injuries are more painful [19].

This confusion originates from the lack of consistently used and validated instruments to measure pain. In a recent metanalysis of neuropathic pain after SCI that included 17 studies, Burke et al. found only two studies used validated instruments to measure neuropathic pain [20]. To further delineate and add to the body of literature surrounding chronic pain after SCI, the authors of the current study developed a survey that included basic demographic questions, instruments to measure neuropathic pain after SCI, and specific questions assessing other types of pain that may be present in this population. Since pain is a subjective finding notoriously difficult to measure, a survey with validated pain instruments is one good way to assess the issues at hand and accurately evaluate the demographics surrounding pain after SCI.

Methods

Survey development

The authors evaluated existing reviews and metanalyses addressing the demographics of pain after SCI [6, 13, 20]. We then interviewed experts in the areas of SCI medicine, pain management, and survey statistics to solicit input on validated tools and potential questions that could be answered by this work. Based on this preliminary work, we developed a survey including general demographic questions, questions of specific interest to the authors, the standardized Spinal Cord Injury Pain Instrument (SCIPI), International Spinal Cord Injury Pain Data Set, Basic form (ISCIPDS:B), Patient Reported Outcomes Measurement Information System (PROMIS) neuropathic 5a (PROMIS-Neur), PROMIS nociceptive 5a (PROMIS-No), and PROMIS pain interference short form 8a (PROMIS-Int). The PROMIS instruments were designed to compare groups of individuals to the general population of the United States. A score of 50 represents the average population with a standard deviation of 10. The mean PROMIS-Neur scores for the surveyed population was 55.2 while the mean PROMIS-No score was 52.0. Of note, the level of completeness of injury (AIS classification) was self-reported and not confirmed by physician examination or medical records review. The final survey consisted of a possible 80 questions, although most combination of answers did not result in the participant answering all 80 questions. A pilot test was conducted by the authors of the study as well as others from within the department to evaluate question clarity and software functionality. The goals of this survey were to (a) collect demographic data as it relates to pain after SCI utilizing standardized tools; (b) assess demographic information of specific interest to the authors; and (c) compare multiple demographic parameters as they relate to measured outcomes using the validated instruments. The survey was thoroughly examined and approved by the Thomas Jefferson University Hospital institutional review board. We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during the course of this research. A formatted copy of the survey is available for review in Appendix A.

Survey distribution

The survey was distributed via Survey Monkey to SCI consumer mailing lists maintained by Thomas Jefferson University as part of the SCI model systems program. Each participant received a link via email and completed the survey online. One patient was unable to independently complete the survey online and therefore completed it over the telephone. A reminder was sent 6 weeks after the initial invitation to participate. All survey responses remained anonymous and repeat responses were discounted by canceling any duplicate IP addresses. There was no time limit to complete the survey. There were no incentives offered for completing the survey. There were 705 individuals queried on the initial email and 711 on the second email.

Data analysis

To answer study objectives, participant data were grouped into neuropathic (SCIPI > 2) or nociceptive (SCIPI < 2). For specific analyses—particularly in questions with multiple answers, similar answers were grouped for analysis (i.e., multiple original age groups were combined to form <55 years of age and >55 years of age). Data on level of injury, completeness of injury, and other similar demographics were combined when no significant between group differences were found. The data were analyzed using individual Chi-Square or t-tests for continuous data. Fisher’s Exact tests were used if expected values in categories fell below five in any cell. All data were calculated using SPSS v.25 (Armonk, NY).

Results

One hundred seventy-one responses were received, giving a response rate of 24%. 81% of respondents had chronic pain. As classified by the SCIPI, 86% of respondents with chronic pain were classified as having neuropathic pain. The mean PROMIS-Neur scores for the surveyed population was 55.2 while the mean PROMIS-No score was 52.0. Eighty-two percent of participants report having experienced pain during their initial hospitalization after their injury, 81% reported having chronic pain since that time, and 66% reported their primary chronic pain started immediately after their SCI. Most (56%) had constant and continuous pain that was unpredictably intense (45%), continued on a daily basis (90%), and has gotten worse since initial injury (54%). Seventy percent of individuals with chronic pain had at least three separate body areas with pain. The median reported daily pain on the Stanford pain scale was 5/10 or “very distressing.” The mode at initial injury was 3/10 or “tolerable” and the current mode of the surveyed sample was 4/10 or “distressing.”

Seventy percent of the respondents were >55 years of age. There was no significant difference in the development of nociceptive vs neuropathic pain as categorized by the SCIPI based on age category (χ2 = 0.942). Seventy-five percent of respondents were male. There was no difference between the type of pain experienced (neuropathic versus nociceptive) when comparing males and females (χ2 = 0.341), however, females mean neuropathic pain scores (\({\bar{x}} = 58.0\)) was significantly higher than the male mean neuropathic pain score as measured by the PROMIS-Neur [(\(\bar x =\) 54.0) (T = −2.053; p = 0.043)] (Table 1, Fig. 1).

Table 1 Categorical variation in pain quality.
Full size table
Fig. 1: PROMIS-No and PROMIS-Neur scores by gender.
figure 1

Women report higher levels of neuropathic pain but not nociceptive pain after SCI.

Full size image

The most common mechanism of injury was motor vehicle accident (35%) followed by falls (30%). Twelve percent were due to penetrating injuries. Penetrating injuries did not influence the development of neuropathic or nociceptive pain on the SCIPI when compared to non-penetrating injuries (χ2 = 0.138). However, those with penetrating injuries reported non-significantly higher PROMIS-No [(\(\bar x = 55.9\)) and PROMIS-Neur (\(\bar x = 57.5\)) scores compared to their non-penetrating peers (\(\bar x =\) 51.5 and 55.1; T = −1.692 and −0.869; p = 0.094 and 0.387) (Table 2).

Table 2 Pain severity.
Full size table

Fifty-two percent of respondents had cervical spine injuries and most (68%) were incomplete injuries as classified by the international standards for the classification of spinal cord injury (ISNCSCI) (grades B, C, D, or E.) The breakdown of ASIA classification was as follows: 30% were ASIA A, 19% were ASIA B, 20% were ASIA C, 20% were ASIA D, and 0.58 % (1 responder) was ASIA E. There was no significant difference in pain type (neuropathic vs nociceptive) based on ASIA classification (χ2 = 0.112). Those classified as ASIA C were more likely to be classified as nociceptive pain by the SCIPI than those in other ASIA classifications (20.5 % in C vs 12.2%, 9.1%, 10.7% for A, B, and D, respectively). Mean nociceptive pain scores, as measured by the PROMIS-No, remained relatively stable across all ASIA classifications (A = 52.5, B = 51.0, C = 52.8, D = 50.2; p = 0.691; Table 3). However, subjects with progressively more incomplete injuries had higher mean PROMIS-Neur scores and trended toward significance (A = 52.1, B = 55.0, C = 55.8, D = 57.7; p = 0.161; Fig. 2). Subjects classified as motor incomplete (ASIA C & D) had similar PROMIS-No scores (\(\bar x\) = 51.8) when compared to motor complete (AISA A & B) pain scores (\(\bar x\) = 52.0). Although not statistically significant, those with motor incomplete injuries reported higher PROMIS-Neur scores (\(\bar x\) = 56.5) than motor complete (\(\bar x\) = 53.2) with a trend toward significance (p = 0.061).

Table 3 Pain severity by ASIA classification.
Full size table
Fig. 2: PROMIS-No and PROMIS-Neur scores by ASIA classification.
figure 2

With more incomplete injuries there was a trend towards more pain.

Full size image

There was no difference in reported pain type when grouping higher (cervical and thoracic) and lower (lumbar and sacral) levels of injury (χ2 = 0.767). Lumbar and sacral injuries were associated with higher PROMIS-Neur (\(\bar x\) = 58.9 vs \(\bar x\) = 54.7) and PROMIS-No (\(\bar x\) = 56.7 vs \(\bar x\) = 51.5) scores when compared to cervical and thoracic injuries. Although this trend was noted, the difference was not significant for PROMIS-Neur (T = −1.476, p = 0.143) but trended toward significance for PROMIS-No (T = −1.977, P = 0.051).

Eighty-two percent of the surveyed population were >5 years from initial injury. Length of time since injury did not significantly impact the type of pain (χ2 = 0.726), nor overall pain scores experienced. There was, however, a trend that respondents with injuries >5 years old had higher average PROMIS-Neur (\(\bar x\) = 55.3) and PROMIS-No (\(\bar x = {{{\mathrm{52}}}}{{{\mathrm{.8}}}}\)) scores when compared to injuries <5 years old (\(\bar x\) = 54.4 and \(\bar x = {{{\mathrm{49}}}}{{{\mathrm{.2}}}}\)), though this difference was not significant (T = −0.376, P = 0.708; T = −1.656, P = 0.101).

Similarly, patients with higher degrees of reported pain(>5) during their initial hospital stay did not have significantly different breakdown of pain type (nociceptive vs neuropathic pain) when compared to those with lower levels of reported pain (χ2 = 0.320). Participants recalling >5 on the Stanford pain scale at initial hospital stay reported significantly higher PROMIS-Neur (\(\bar x\) = 56.35) than respondents with Stanford pain scales <4 at initial hospital stays (\(\left( {\bar x = 52.1} \right)\) (T = −2.114; p = 0.037) (Fig. 3). Similarly, individuals who recall having >5 on the Stanford pain scale during their initial hospital stay reported significantly higher PROMIS-No scores \(\left( {\bar x = 53.2} \right)\)than those with <4 during their initial hospital stay \(\left( {\bar x = 48.8} \right)\) (T = −2.413; p = 0.018) (Fig. 3).

Fig. 3: PROMIS-No and PROMIS-Neur scores during initial hospitalization.
figure 3

Individuals with more pain recall having more pain during initial hospitalization.

Full size image

Thirty-nine percent of respondents were employed to some degree, but there was no significant difference in type of pain based on employment status (χ2 = 0.957). There was no difference in PROMIS-No (\(\bar x\) = 52.4 vs \(\bar x\) = 51.4) or PROMIS-Neur (\(\bar x\) = 55.5 vs \(\bar x\) = 54.7 based on employment status (T = −0.429, P = 0.668; T = −0.632, P = 0.529).

Fifty-five percent of respondents reported shoulder pain. Those with tetraplegia were more likely than those with paraplegia (thoracic, lumbar, or sacral injuries) to have shoulder pain (p = 0.049). Respondents who reported having their worst or second worst pain affect their shoulders had significantly higher PROMIS-No scores (\(\bar x\) = 54.3 vs \(\bar x\) = 50.6) (T = 2.136; p = 0.030) but not PROMIS-Neur scores.

A summary of SCIPI groups and variables can be reviewed in Table 1. PROMIS-Neur and PROMIS-No scores for each group can be reviewed in Table 2.

Discussion

Most individuals with SCI experience chronic pain regardless of the mechanism of injury or ISNSCI scores. The demographics of this survey population are generally consistent with the population demographics of those with SCI in the United States. Results of the current study suggest that most individuals with SCI (81%) have chronic pain and most of those (86%) experience neuropathic pain, which is within the range reported for most of the SCI pain literature [7, 10, 13, 14, 21].

Neuropathic pain after SCI is likely a unique phenotype of neuropathic pain that originates from disruption of spinal modulation pathways as opposed to similar “neuropathic” conditions like a peripheral nerve injury or post-stroke neuropathic pain [22]. Neuropathic pain manifests differently at and below the level of injury. Neuropathic pain at the level of injury is likely caused by injury to the nerve roots and spinal cord at that level as compared to neuropathic pain below the level of injury, which is likely related to disruption of longer neuronal pathways from the lesion [10, 21].

Our results suggest that the completeness of the SCI correlates inversely with the degree of neuropathic pain experienced—complete injuries had a lower mean pain score compared to those with progressively more incomplete injuries (Table 2). This may be related to the way descending modulation pathways in the spinal cord are disrupted by injury, creating intermittent, incomplete, and abnormal transmission of signal across the damaged area of the cord. A similar mechanism is proposed to explain why spasticity is worse in incomplete spinal lesions [23]. As previously noted, the level of completeness was recorded by patient report and not confirmed with examination or medical record review as such measures are unlikely to significantly impact the overall accuracy of the data collected. This population is knowledgeable about their injuries and the aforementioned classification system. Given the frequency that patients report such scores, there is a high degree of confidence in the accuracy of these responses. Though, a small degree of error is introduced and may contribute to some uncertainty in our final data analysis.

In addition to pain at and distal to the level of injury, patients with SCI often develop shoulder pain, regardless of the level of injury. Those who rated the shoulder as their first or second most painful area reported higher PROMIS-No scores than the rest of our population. Individuals with paraplegia often develop nociceptive shoulder pain from overuse [24]. Years of relying on the shoulder girdle for weight shifts, transferring, and mobility (propelling a manual wheelchair) can lead to a spectrum of rotator cuff pathology. From acute tendonitis to chronic complete rotator cuff tears, these injuries can all result in chronic shoulder pain [25].

In the current study, those with tetraplegia were significantly more likely than those with paraplegia to report shoulder pain. Alternatively, individuals with higher cervical injuries (C3-5) may develop shoulder pain secondary to spasticity and shoulder subluxation. In addition to pain, a weak shoulder struggles to position the hand in space to perform activities of daily living [26]. This abnormal scapular kinesis may lead to the entire spectrum of rotator cuff pathology seen in paraplegia. Scapular dyskinesis is a well-known etiology of shoulder pain, but may be secondary to other conditions [27, 28]. Higher levels of injury may result in decreased shoulder range of motion. This has been linked to increased shoulder pain in this population [29]. Functional substitution of stronger muscle groups such as the trapezius may lead to suboptimal positioning of the scapula, further predisposing the shoulders to injury. Taken in total, the current study suggests the shoulder is a common pain generator and the shoulder pain experienced by both those with paraplegia and tetraplegia is more consistent with nociceptive pain than neuropathic pain.

With regard to level of injury, lumbar or sacral injuries trended towards having more nociceptive pain and also reported the highest PROMIS-Neur scores of any subgroup analyzed, although the mean score was not significantly different from that of the cervical/thoracic group. There is some literature that cauda equina injuries are particularly painful [14]. It is suspected that both of these differences would have been significant if the number of subjects was higher, as there were only 17 lumbar/sacral injuries in our sample.

Females reported significantly higher levels of neuropathic pain (PROMIS-neur), but not nociceptive pain (PROMIS-no). However, there was a similar distribution of females and males with neuropathic and nociceptive pain (χ2 = 0.341). As such, sex did not predispose patients to develop neuropathic or nociceptive pain. It is possible this is not a true reflection of the demographics of women with SCI as our sampled population was heavily skewed in favor of males. It has been noted in prior studies, however, that women report more below level neuropathic pain after SCI in the past [30]. This phenomenon has also been noted with other neuropathic conditions such as polyneuropathy [31]. Some suggest sex may be an important factor in the modulation of pain [32, 33]. Additionally, a review on the prevalence of chronic pain after SCI found sex to have a small impact on the experience of pain [13].

Understanding the trajectory of the pain course is of vital importance to those who treat SCI related pain. The current study was not designed to track pain over time, however, there was a correlation between the recollection of a painful acute hospital stay and current levels of neuropathic and nociceptive pain. This may suggest that those with more pain at the onset of injury will also experience more chronic pain. Alternatively, there could be recall bias where those who develop more chronic pain recall always being in more pain. This distinction is important as it may impact patient prognosis, goals, and expectations. A longitudinal study tracking pain severity over time would help elucidate this question.

The current study is not without limitations. This survey was distributed through our model system database, which covers the Delaware Valley and could introduce regional bias. There were a number of statistical categories, mainly those assessing nociceptive pain, where our sample size was small enough to introduce the possibility of Type II error. Additional studies with larger samples size spanning a broader part of the country would be warranted to eliminate the possibility of a regional bias, better understand how sex impacts pain in patients with SCI, and compare the quality and severity of pain to the level of injury.

In summary, this survey suggests neuropathic pain is the predominate pain after SCI. In our sample, 81% of individuals experience chronic pain and 86% of those with pain are classified as having neuropathic pain. Overall, individuals with SCI report higher levels of neuropathic and nociceptive pain compared to the general United States population. Those who reported higher levels of current nociceptive and neuropathic pain were more likely to report higher levels of pain during their initial hospital stay. Females were more likely to report higher levels of neuropathic pain but not nociceptive pain than males. Incomplete injuries trended toward producing a phenotype with more neuropathic pain and possibly nociceptive pain than complete injuries and lumbar/sacral injuries trended toward producing a phenotype with more nociceptive pain. Shoulder pain afflicted 55% of individuals surveyed. Those with tetraplegia were more likely to develop shoulder pain than those with paraplegia, and those who reported their first or second worst pain to be shoulder pain had significantly higher nociceptive pain scores. Understanding these pain demographics will enable physicians to better predict complications, take down barriers to improvement, and optimize care for patients with SCI.