Study Design
After institutional review board approval had been obtained, a computerized database search for paraplegic patients was performed in the Department of Orthopaedic Surgery and Rehabilitation Medicine. The department has a dedicated division for spinal cord injury and thus a large patient base. Once the patients were identified, medical records were reviewed to obtain (1) the date of injury; (2) the type of injury; (3) demographic information; (4) concomitant medical conditions; and (5) the total duration of wheelchair dependence. These patients were contacted by mail and were asked to participate in a clinical case-control trial to examine their shoulder function and to assess their shoulder joint with use of magnetic resonance imaging.
An advertisement was placed in our regional newspaper in order to include a representative sample from the normal population as controls for the study. Because of cost constraints, we only included one control per case.
All patients and able-bodied volunteers (control sample) were offered a small incentive (travel expenses and allowance) to participate in the study. Before enrollment, both patients and controls gave written informed consent to participate in the study.
A patient was included in the study if he or she (1) had been wheelchair-dependent for a minimum of thirty years; (2) was not morbidly obese (body mass index <40); (3) did not have an active shoulder infection; (4) had not previously undergone surgery on the shoulder; (5) was physically and mentally willing and able to comply with the scheduled clinical examination and magnetic resonance imaging examination; (6) did not have known cervical disc herniation, cervical or thoracic syringomyelia, tetraplegia, or advanced degenerative joint disease of the spine; (7) did not have a history of soft-tissue injury of the upper extremities (i.e., brachial plexus injury, tendon rupture, or dislocation); (8) had not had previous surgery of any kind on the upper extremities (i.e., fracture repair, tendon repair, or acromioplasty); and (9) did not present with any contraindications for undergoing magnetic resonance imaging studies. The initial search identified 320 patients who met these inclusion criteria. From the 320 available paraplegic patients, we randomly selected 100 by drawing lots.
With the exception of wheelchair dependence, the inclusion and exclusion criteria were the same for the spinal cord injury group and the control cohort. After their written agreement had been received, these individuals were included in the study.
Between October 2005 and May 2007, 100 paraplegic patients (200 shoulders) who had been paraplegic and wheelchair-dependent for a mean of thirty-three years entered the study and were matched with a cohort of 100 able-bodied volunteers (200 shoulders). The matching was done on the basis of age within five years and sex. The control cohort of able-bodied volunteers was selected in the order in which they responded to our invitation and advertisement. By virtue of the inclusion and exclusion criteria, the matching criteria seemed to be adequate for establishing comparable groups.
Both cohorts were scheduled for a clinical examination and a magnetic resonance imaging study. In addition, the able-bodied volunteers underwent a brief examination to rule out any neurological deficits affecting the upper extremities. The examinations for both groups followed standardized protocols, and the control group was used as an internal reference for the paraplegic patients.
Demographic Data
The mean age (and standard deviation) of the paraplegic patients (including seventy-two men and twenty-eight women) was 52.0 ± 9.0 years (range, twenty-nine to seventy years), and 89% (eighty-four) of ninety-four patients were right-handed. The level of spinal cord injury ranged from T2 to L3, and the patients had been fully wheelchair-dependent for an average of 33.7 ± 5.1 years (range, twenty-nine to forty-eight years). In this group, the mean body mass index was 24.8 ± 5.0 kg/m2. In the matched control group of able-bodied volunteers, the mean age was 52.0 ± 8.2 years (range, thirty-six to sixty-five years) and 95% (ninety-two) of ninety-seven patients were right-handed. In the control group, the mean body mass index was 24.9 ± 3.6 kg/m2. The demographic data are summarized in Table I.
Clinical Evaluations
Clinical evaluations included standardized testing with numerous clinical tests that were developed to examine the shoulder (the Speed test, the Yergason test, the O'Brien test, the Jobe test, the Neer impingement test, the lift-off test, and the apprehension test). A visual analog scale was used to measure current pain, with 0 indicating a painless shoulder and 100 indicating a severely painful shoulder. In addition, the Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire12 was administered. This is a thirty-item, self-reported questionnaire that is designed to measure disability related to the upper extremity, regardless of the cause.
The Constant-Murley method was used for the functional assessment of the shoulder13,14. This 100-point scoring system is based on the assessment of a number of individual subjective and objective parameters. The right and left shoulders were assessed separately.
As the upper extremity of paraplegic patients is not neurologically impaired and therefore is theoretically comparable with that of able-bodied volunteers, the visual analog scale, Constant score, and DASH questionnaire can be applied to the upper extremity in paraplegic patients.
Radiographic Analysis
Magnetic resonance imaging of the shoulder was conducted for all patients in the oblique coronal, oblique sagittal, and axial planes on a 1.0-T Gyroscan ACS-NT-T10 Intera scanner (Philips Medical Systems, Hamburg, Germany). All of the magnetic resonance imaging examinations were analyzed by the same two musculoskeletal radiologists (including one of the authors [T.G.]), who were blinded with regard to whether the individual undergoing the magnetic resonance imaging study was from the study group or the control group. The magnetic resonance images were reviewed according to an established standardized protocol for shoulder injuries.
Pathological findings in the shoulder girdle are briefly defined and summarized in a table in the Appendix.
Statistical Methods
Data were entered into a spreadsheet, and then the accuracy of the data was checked and evaluated with use of statistical software. The distribution of continuous data and scores was described on the basis of means and standard deviations as well as the median, minimum, and maximum values. For categorical data, absolute and relative frequencies were calculated. Possible differences between patients and controls were analyzed with use of the Fisher exact test for categorical data and with use of the Student t test for continuous data and scores.
The primary objective of the present study was to compare the prevalence of rotator cuff tears in paraplegic patients with that in a group of controls with use of magnetic resonance imaging as a diagnostic criterion.
The prevalence of the primary objective criterion, rotator cuff rupture in patients and controls, was used to estimate the relative risk of this rupture between cases and controls. We determined the odds ratio as an estimate of the relative risk. The level of significance (a) was set at 5%. The primary end point of the study was the prevalence of rotator cuff rupture (left or right). The statistical analysis of the other characteristics is descriptive in nature. Therefore, no adjustment of the significance level was performed.
The sample size could not be calculated as the preexisting evidence was insufficient. However, assuming the incidence of rotator cuff tear in the controls to be 15% and the estimate of the relative risk (odds ratio) for the patients to be increased threefold, a sample size of seventy-five individuals per group would be sufficient to detect these risk increases.
Source of Funding
There was no external funding source.
Constant Score, DASH Score, and Visual Analog Score for Pain
The Constant scores were significantly better for the able-bodied volunteers than for the paraplegic patients (p < 0.001). The mean Constant score for the right shoulder was 70.0 ± 15.1 (range, 28.0 to 96.0) points for the study group, compared with 89.0 ± 7.4 (range, 50 to 100) points for the control group. A similar difference in Constant scores was seen for the left shoulder, with a mean score of 74.7 ± 14.8 (range, 34 to 100) points for the study group and 88.6 ± 8.2 (range, 67 to 100) points for the control group. Statistical comparison of Constant scores revealed significantly better function for the able-bodied controls than for the paraplegic patients (p < 0.001 for both comparisons).
The results on the DASH questionnaire were significantly better for the able-bodied volunteers than for the paraplegic patients (p < 0.001). The mean DASH score for the right shoulder was 53.7 ± 26.8 points (range, 30 to 131 points) in the study group, compared with a mean of 30.7 ± 3.3 points (range, 5 to 39 points) in the control group. A similar difference was seen for the left shoulder, with a mean DASH score of 53.9 ± 28.8 points (range, 30 to 150 points) for the paraplegic patients and 31.7 ± 3.4 points (range, 23 to 53 points) for the able-bodied controls. Comparison of the DASH scores revealed significantly better function in the able-bodied controls than in the paraplegic patients (p < 0.0001).
Similarly, the visual analog scores for pain were fourfold to sevenfold lower for the able-bodied controls than for the paraplegic patients (p < 0.0001 for both comparisons). The mean visual analog score for pain in the right shoulder was 37.4 ± 30.8 (range, 0 to 85) for the paraplegic patients and 9.2 ± 20.4 (range, 0 to 76) for the volunteers. The mean visual analog score for pain in the left shoulder followed a similar pattern, with a mean score of 34.5 ± 30.7 (range, 0 to 100) for the paraplegic patients and 5.2 ± 17.8 (range, 0 to 95) for the controls. Sixty-seven percent of the patients in the study group had shoulder pain, compared with 16% of the subjects in the control group (p < 0.0001).
Prevalence
Magnetic resonance imaging revealed that the prevalence of rotator cuff tears involving the right or left shoulder was four times higher in the paraplegic group than in the control group (63% compared with 15%; p < 0.0001). The sixty-three tears in the paraplegic group included forty-nine full-thickness tears (78%) and fourteen partial-thickness tears (22%). The fifteen tears in the volunteer group included eleven full-thickness tears (73%) and four partial-thickness tears (27%). A rotator cuff tear involving the left shoulder was present in 50% of the patients in the study group, compared with 11% of the subjects in the control group (p < 0.0001). A rotator cuff tear involving the right shoulder was present in 52% of the patients in the study group, compared with 14% of the subjects in the control group (p < 0.0001). A rotator cuff tear involving the supraspinatus tendon was present in 61% of the patients in the study group, compared with 14% of the subjects in the control group (p < 0.0001). A tear involving the infraspinatus tendon was present in 19% of the patients in the study group, compared with 3% of the subjects in the control group (p = 0.0003). A tear of the subscapularis tendon was present in 12% of the patients in the study group, compared with 2% of the subjects in the control group (p = 0.0056). The results are summarized in detail in Table II.
The prevalence of glenohumeral osteoarthritis was 19% in the study group, compared with 1% in the control group (p < 0.0001). The prevalence of acromioclavicular osteoarthritis was 42% in the study group, compared with 26% in the control group (p < 0.0169) (Table II).
Allocation of Rotator Cuff Tears
Of the 100 paraplegic patients in the study group, thirty-eight had a tear of the supraspinatus tendon only; twelve had a tear of the supraspinatus and infraspinatus; five had a tear of the supraspinatus and subscapularis; five had a tear of the supraspinatus, infraspinatus, and subscapularis; one had a tear of the supraspinatus, infraspinatus, and teres minor; one had a tear of the infraspinatus only; and one had a tear of the subscapularis only.
Of the 100 able-bodied volunteers in the control group, ten had a tear of the supraspinatus only, two had a tear of the supraspinatus and infraspinatus, one had a tear of the supraspinatus and subscapularis, one had a tear of the supraspinatus and teres minor, and one had a tear of the infraspinatus and subscapularis.
Risk Analysis
The paraplegic patients had a tenfold higher risk of experiencing a rotator cuff rupture (a full-thickness or partial-thickness rotator cuff tear) (odds ratio, 9.6; 95% confidence interval, 4.9 to 19.1). Table II shows that at a value of about 5, the lower limit of the 95% confidence interval associated with the odds ratio is considerably higher than 1, thus substantiating the highly significant increase in risk (p < 0.0001).
Nearly without exception, the results of risk assessment for the dichotomous secondary objective criteria were significant. Furthermore, we distinguished between the individual muscle tears. The odds ratio for a supraspinatus tendon tear was 9.6 (95% confidence interval, 4.8 to 19.2), the odds ratio for an infraspinatus tendon tear was 7.6 (95% confidence interval, 2.2 to 26.5), and the odds ratio for a subscapularis tendon tear was 6.7 (95% confidence interval, 1.5 to 30.7).
The relative increases in the risk for bursitis, acromioclavicular arthritis, and glenohumeral arthritis are shown in Table II.
With the numbers available, the prevalence of rotator cuff tear could not be correlated with the level of neurological injury (the level of trunk innervation), and thus a specific level of injury could not be identified as a risk factor (p = 0.21).
Previous studies have indicated that 30% to 64% of patients with a spinal cord injury have reported chronic shoulder pain2-6,10,15,16. However, the reported rates of shoulder pain have varied widely among the studies. This variation in rates may be explained by differences in study population characteristics, by differences in the diagnostic criteria used, and by the lack of inter-examiner and intra-examiner consistency regarding many items on the physical examination.
In previous studies, the investigators simply attempted to determine the prevalence of shoulder pain in patients with spinal cord injury with use of postal surveys3,10,16,17 and interviews18,19. A limitation of most of those studies was that they did not distinguish between specific clinical disorders of the shoulder joint. On the basis of those studies, shoulder pain was reported to affect as many as 50% of patients with a spinal cord injury. However, another common limitation of all of those studies was the lack of a standardized shoulder examination and the lack of modern imaging modalities to detect pathological shoulder changes. Furthermore, shoulder discomfort and pain were not investigated with respect to pathological shoulder changes.
In the present study, 67% of the paraplegic patients reported shoulder pain, compared with 16% of the controls. These results are consistent with those published in the literature3,10,16,17. The total DASH score was significantly better for the control group of able-bodied volunteers than it was for the study group of paraplegic patients. The total DASH score for the control subjects implies that there was a moderate amount of disability in the control group. This is probably because of the age of the control subjects. Jester et al.20 provided data that were comparable with those for our study population and that gave insight into disability levels in a nonclinical population (normative data). Those authors reported that, when data were stratified according to age, significant differences were found in terms of the total DASH score, with the poorest function being associated with increasing age. They reported a mean total DASH score (and standard deviation) of 14.0 ± 15.4 for the individuals who were thirty to forty-nine years of age and of 19.0 ± 18.0 for those who were fifty to sixty-five years of age. The total DASH score in our control group, in which the mean age was 52.0 years, fits well with the results reported by Jester et al.20 for the group of individuals who were fifty to sixty-five years of age. The total DASH score for the controls lies within the standard deviation of the published normative data.
Sie et al.21 interviewed 103 patients with a spinal cord injury who had paraplegia and found that 64% reported upper-extremity pain and 36% specifically complained of shoulder pain. An analysis of the etiology of the shoulder pain demonstrated that it was related to a musculoskeletal disorder in 73% of the patients. In the present study, 63% of the paraplegic patients had a rotator cuff tear, which was documented on the basis of both clinical examination and magnetic resonance imaging of the shoulder joint.
Bayley et al.8 examined ninety-four patients with complete paraplegia and found that one-third had shoulder pain. In the symptomatic group, 75% of the patients had symptoms and signs that were consistent with impingement and subacromial bursitis. Evaluation with use of arthrography showed a rotator cuff tear in 65% of the patients in the symptomatic group. Escobedo et al.7 evaluated individuals who had paraplegia with use of magnetic resonance imaging. Evidence of a rotator cuff tear was noted in 73% of the shoulders in symptomatic individuals with paraplegia, compared with 9% of the shoulders in asymptomatic individuals with paraplegia. In the study by Boninger et al.22, twenty-eight paraplegic individuals who had sustained a spinal cord injury at a mean of 11.5 years previously were evaluated with use of magnetic resonance imaging. Nine (32%) of the twenty-eight subjects experienced shoulder pain in the month prior to testing. Magnetic resonance imaging demonstrated only one rotator cuff tear. This prevalence is significantly different from our findings and may be attributed to the relatively short mean duration of wheelchair dependence (11.5 compared with thirty-three years).
The sitting position in wheelchair-dependent patients requires frequent and often prolonged overhead reaching for day-to-day and vocational tasks. The association of overhead reaching with impingement has been well documented in able-bodied laborers and athletes23-25.
Age-related, primary degeneration appears to be the most common mechanism for rotator cuff tears. Failure of the rotator cuff tendon is the most common clinical problem related to the shoulder in the elderly population, accounting for more than 4.5 million physician visits per year in the United States26,27. The natural history of degenerative cuff-tendon failure is one of age-related progression28. Several studies have shown that the prevalence of rotator cuff tears increases significantly after the age of sixty years28-33.
The average age of our study population of paraplegic patients with a rotator cuff tear was only fifty-two years. The 63% rate of rotator cuff tear in this group not only was considerably higher than that in our control group (15%) but was also considerably higher than the rates reported in the aforementioned studies28-33, suggesting that an age-related etiology is much less likely a reason as compared with mechanical causes such as wear and tear. It is highly likely that wheelchair activity plays a major role in the high prevalence that we noted. This is also supported by the finding that there was no difference in the prevalence of rotator cuff tears between the right and left shoulders in our population of paraplegic patients. This finding needs to be verified in future studies.
Several common physical stressors including repetition, force, posture, and duration have been found to lead to activity-related musculoskeletal problems of the upper extremity23,34-37. Kulig et al.9 developed a three-dimensional model to determine the magnitude and direction of the forces and moments acting on the shoulder during wheelchair propulsion. The results of that study9 clearly showed that propelling a wheelchair imposed an upward force at the shoulder.
In patients with paraplegia, the upper extremity is fully innervated but innervation is variable at the level of the trunk2. In the absence of trunk innervation, there are increased biomechanical stresses on the upper extremities during functional activities2. Without synergistic trunk stability, upper-extremity tasks must be performed in isolation2, and compensatory strategies, such as using the other upper limb for stabilization or sitting in a "C" spinal posture, must be used to avoid falling over2. These functional anatomical differences may contribute to the propensity for pathological shoulder conditions to develop in patients with higher-level paraplegia2,38. In our study, we did not see a higher prevalence of rotator cuff tear in patients with higher levels of paraplegia (p < 0.21), although the study may have been insufficiently powered for this determination. However, this finding does not support those of other reports38 that have shown a relationship between the level of neurological impairment and the prevalence of rotator cuff tears.
Collinger et al.39 showed that body weight is a variable that affects the shoulder forces required to propel a manual wheelchair. We could not identify an association between body weight or body mass index and shoulder pathology in the group of paraplegic patients (p = 0.7973).
The limitations of the present study were the relatively small numbers of patients and controls and the likelihood that patients and able-bodied volunteers with shoulder pain were more likely to participate in the study than patients and volunteers without pain.
To our knowledge, this is the first study to demonstrate that the structural and functional changes of the shoulder joint are more severe and the risk of shoulder-girdle pathology is significantly higher in individuals with long-term paraplegia than in an age-matched control group. We believe that the pathology of the shoulder joint in paraplegic patients is triggered by repetitive over-the-head reaching to access the environment from a wheelchair position and the duration of wheelchair dependence rather than being primarily age-related deformation.
For these patients, it may be important that rehabilitation schemes take this into account and that aids and the environment be optimally adapted to their needs in order to reduce the incidence of secondary damage to the shoulder in individuals who are confined to a wheelchair for a long period of time.