Functional Outcomes After Anatomic and Reverse Shoulder Arthroplasty Assessed using a Novel Objective Method

1. INTRODUCTION

The goal of shoulder replacement surgery is to improve pain and shoulder function in patients with shoulder arthritis, whether or not they have a rotator cuff deficiency. Shoulder arthritis, both with and without rotator cuff deficiency, is a complex condition that seriously compromises the comfort and function of the shoulder. The primary indication for shoulder arthroplasty is a painful shoulder caused by glenohumeral arthritis that has failed nonsurgical management [1]. On the other hand, the presence of rotator cuff arthropathy often results in significant pain and loss of shoulder motion [2].

While pain relief and increased range of motion can be obtained in shoulders with arthritis and an intact rotator cuff with an anatomical shoulder arthroplasty, a Reverse Total Shoulder Arthroplasty (RSA) is more suited to shoulders with a deficient rotator cuff or severe bone loss [3-6]. In addition to providing pain relief and improving shoulder motion, each type of arthroplasty has been shown to enhance functional outcomes using traditional subjective functional outcome scores [5, 7]. Flurin et al. found that Total Shoulder Arthroplasty (TSA) performs better than RSA in functional outcome scores, especially Constant and American Shoulder and Elbow Surgeons (ASES) scores, and better in Range of Motion (ROM) pre- and post-operative over 30 months follow-up [8]. Simovitch et al. evaluated 1641 shoulders retrospectively and found that the functional outcome is 30% larger in RSA compared with TSA in the first 12 months post-operative [9]. Kiet et al. reported no significant difference between TSA and RSA at a two-year follow-up in functional outcome and complication rate [10].

Although several validated questionnaires can be used to assess shoulder function, such as the WOOS, Oxford shoulder score, Constant score, and 36-Item Short-Form Health Survey (SF-36) questionnaire, there is no widely adopted objective tool to compare the outcome of shoulder arthroplasty. Interactive robotic technologies offer an objective method for assessing arm function. In particular, the Kinarm Exoskeleton Lab uses interactive robots and an integrated virtual reality system, along with standardized behavioral tasks to quantify upper limb sensory and motor function (Fig. 1A); Kinarm, Kingston, Ontario, Canada). Kinarm has been used previously to assess the upper extremity sensory and motor function in stroke [11-14]. Therefore, this study was designed with two objectives. First, to compare the functional outcomes after anatomic and reverse shoulder arthroplasty. Second, to determine the usefulness of the Kinarm as a reliable tool to objectively measure shoulder functional improvement in subjects receiving a shoulder arthroplasty.

2. MATERIALS AND METHODS

This was a prospective study of patients with glenohumeral arthritis who had failed nonoperative treatment and underwent shoulder replacement surgery. After obtaining informed consent, patients were assigned to receive either anatomic or reverse shoulder arthroplasty based on several factors, including patient age, diagnosis, associated rotator cuff disease and pre-operative function. This study was approved by the Queen’s University and Affiliated Teaching Hospitals Health Sciences Human Research Ethics Board under study number: SURG-206-09. All surgeries were performed by the senior author, a fellowship-trained shoulder surgeon.

Patient evaluation included shoulder range of motion, incorporating forward elevation and lateral elevation. Patients also completed several Patient-Reported Outcome Measures (PROMs), including the WOOS, the Oxford Shoulder Score, the Constant Score, and the SF-36 questionnaire [15-19]. All PROMs were evaluated pre-operatively and at 6 weeks, 3 months, 6 months and 1 year post-operatively.

Kinarm assessments were performed pre-operatively and at 1-year post-operatively at Kingston Health Sciences Centre (Kingston, ON, Canada). Individuals sat in an adjustable-height chair, and their arms were supported by troughs connected to robotic linkages. The linkages were adjusted such that the participant’s shoulder and elbow joints were aligned with the joints of the robotic linkage. Shoulder and elbow angles and hand position were calculated from encoders within the torque motors attached to the mechanical linkage. The system permitted arm movement in the horizontal plane with the shoulder positioned at approximately 85 degrees of abduction. A virtual reality system using a television and a reflective mirror displayed spatial goals and feedback of limb position (when required) aligned in the horizontal workspace of the arms (Fig. 1A). Direct vision of the arms was occluded by a physical barrier, and peripheral vision of the arms was removed using a bib connected between the virtual reality stand and around the participant’s neck. (For a video of the basic setup for the Kinarm Exoskeleton Lab, see https://kinarm.com/solutions/exoskeleton-lab/).

Each patient performed three Kinarm Standard Tests (KST) in 15 minutes, including training and instruction [13].

2.3. Arm Position Matching (APM) Task (Fig. 1D) [12]

This task assesses arm proprioceptive function. During this task, the Kinarm robot attached the operative arm to one of four spatial locations, and the participant was instructed to move their other arm to the mirror-image position. The test was repeated six times for each target.

Several spatial and temporal features were used to characterize performance on each of the Kinarm tasks [11-14]. KST also calculates a Task Score that quantifies an individual’s performance relative to a large cohort of healthy controls, such that scores near 0 reflect the best performance and 1.96 represents the 95th percentile for healthy individuals [13].

Sample size was calculated using a previous Kinarm visually guided reaching task study [11]. With a mean difference of 7 degrees between affected and unaffected arms and standard deviations of 6 and 13 degrees, 30 patients are required in each of the two arthroplasty groups. To allow for a 20% loss to follow-up, we targeted 36 patients in each group. Comparisons were made within and across implant groups. Statistical analyses employed paired and independent samples t-tests, as well as one-way Analysis of Variance (ANOVAs). As there is no gold standard tool to compare the Kinarm measure with, Pearson’s correlation coefficient was used to correlate the pre- and post-operative Kinarm scores with each of the WOOS, Oxford, Constant, and SF-36 measures to assess reliability.

3. RESULTS

This study included 81 shoulders in 74 patients, with 13 patients lost to follow-up, 31 of whom received a TSA and 37 received an RSA (Fig. 2). The RSA group matched the age of the TSA group but contained significantly more females Table 1. Evaluation of shoulder range of motion included forward elevation and lateral elevation. Comparing each group with baseline findings (pre-op), the TSA group showed a significantly better range of motion at 3, 6 and 12 months post-operatively, whereas the RSA group showed significant improvement at 6 and 12 months post-operatively Tables 2 and 3. The TSA group had a better forward elevation at 6 weeks and 3 months and better lateral elevation at 3 months post-operatively compared with RSA. However, no significant difference was observed between the two groups after these points Tables 2 and 3.

Evaluation of functional outcome scores revealed that the WOOS and Oxford scores showed significant improvements from baseline to 6 weeks, 3, 6 and 12 months in both groups (p < 0.05). The Constant score showed better outcome in both groups at 3, 6 and 12 months post-operatively Tables 3 and 4. The SF-36 showed improvements in physical scores in RSA, but not in mental scores in TSA Table 4 and 5. When comparing the functional outcome measures of both groups, the TSA group showed better WOOS scores at 6 weeks and 3 months compared to the RSA group, but no differences were observed at 6 months and 1 year Table 6. The Constant score was better at 3 months in the TSA group, but there were no differences at other time points. The TSA group had higher Constant and Oxford scores at baseline and 6 weeks post-operatively Table 6, but there were no differences at other time points.

The evaluation of objective functional outcomes using the Kinarm revealed that pre-operative performance in the visually guided reaching and object-hitting tasks was quite high (poor score) (Fig. 3A). A majority of individuals were identified as impaired in reaching and object-hitting tasks Table 7. A few individuals were identified as impaired pre-operatively in the arm matching task, which assessed proprioceptive function, and even fewer were identified at the one-year follow-up Table 7. In general, both surgical groups showed a significant improvement in object hit and visually guided reaching tasks with no significant difference between groups. In contrast, in arm matching tasks, the RSA group performed better than the TSA group (p-value = 0.05) (Fig. 3B, Table 7.

Pearson’s correlation coefficient showed that the pre- and post-operative Kinarm Task Scores were moderately to highly correlated with Constant, Oxford scores, and WOOS scores Table 8. The pre- and post-operative SF-36 scores showed a weak to moderate correlation with Kinarm scores Table 8.

4. DISCUSSION

This study found that the TSA group scored significantly better on several pre-operative measures compared to the RSA group. This is expected as the selection criteria for patients undergoing RSA included associated rotator cuff disease and poor pre-operative function. Another important finding was the significant improvement in all functional scores in both groups post-operatively compared with the pre-operative functional scores. The TSA group had better function in the early post-operative period, but there was no significant difference in functional scores between the two groups at 6 months and 1 year post-op. These results align with those observed in previous studies, which suggested a significant improvement in functional outcomes both before and after surgery, with no difference between the two groups over time [20, 21]. This is expected as the TSA group had better pre-operative function and would therefore be expected to have better function early post-operatively, whereas the RSA group took longer to regain function post-operatively.

This study found a statistically significant moderate to strong correlation between the Kinarm score and the Constant score pre- and post-operatively. A moderate correlation was found between Oxford scores pre- and post-operatively, as well as with WOOS pre- and post-operatively. However, Kinarm scores showed a low to moderate correlation with the SF-36, which is similar to a previous study that found a poor correlation between the SF-36 and shoulder-specific measures [20]. Therefore, Kinarm can be considered a good, reliable, and objective tool to assess pre- and post-operative improvement in shoulder function after shoulder arthroplasty. Thus, Kinarm may provide an objective measurement tool to assess shoulder function and may be an important addition to current functional outcome measures.

In addition, we found a significant post-operative improvement in the Kinarm limb matching task, which measures proprioception function in the RSA group. However, no significant change was found in the TSA group. This finding suggests that shoulder arthroplasty does not impair proprioceptive function and may provide some improvement, particularly as observed with the RSA group. Therefore, RSA may provide an improvement in proprioceptive function that is not observed after TSA. As mentioned earlier, the Kinarm findings were reinforced by the PROMs, which showed significant post-operative improvements compared with pre-operative scores. However, these measurements are subjective and depend mainly on the patient’s cooperation and memory, which can be prone to response fatigue. Thus, the Kinarm measurements are objective tools and can overcome this issue. Therefore, Kinarm may provide an objective measurement tool for shoulder function after shoulder arthroplasty.

One strength of this study was the standardization of procedures, as all shoulder replacements were performed by the same fellowship-trained shoulder surgeon. The Kinarm strengths include its ability to eliminate human subjectivity in quantifying the nature and severity of dysfunction. It also allows precision by quantifying subtle but measurable limb dysfunction. Although the status of the patient’s pre-operative shoulder condition prevented us from randomizing them to the implant intervention, the study was prospective, which limited the biases inherent in retrospective studies. One limitation of using the Kinarm is its inability to assess the full shoulder range of motion, as the patient requires both arms and forearms to be attached to the Kinarm exoskeleton linkages. However, this limitation was minimized in this research by the ability of the participant to move their shoulder through flexion/extension in the horizontal plane, as well as abduction/adduction around 85 degrees. The ROM findings were also in accord with the Kinarm and PROM results, showing improvements over time and no differences between the implant groups at one year.

5. STUDY LIMITATIONS

This study involved a moderate sample size and non-randomized patient allocation, which may limit the generalizability of the findings. Kinarm assessments, although objective, do not capture the full three-dimensional motion of the shoulder. Additionally, outcomes were limited to one year post-operatively.

CONCLUSION

There was an overall improvement in functional outcome scores after both anatomic and reverse shoulder arthroplasty. Patients undergoing TSA have a better range of motion and function pre-operatively and in the early post-operative period compared to patients undergoing RSA. However, both TSA and RSA have a similar range of motion and function at 6 months and 1 year post-operatively. Kinarm is a novel research tool that may be useful to objectively assess functional outcomes after shoulder arthroplasty. Therefore, this tool may have applications to pre- and post-operative objective evaluation of shoulder function. We anticipate that this will primarily be used in objective measurement in research studies, rather than routine post-operative evaluation.

References