Total shoulder arthroplasty is the reconstructive option to deal with severe shoulder joint damage. Reverse total shoulder arthroplasty (RSA) is often used in more complex shoulder derangement with coexistent rotator cuff deficiencies. RSA has been used successfully in patients with rotator cuff arthropathy, complex proximal humerus fractures and non-unions, and for revisions of failed conventional shoulder arthroplasties [1]. The main advantage of RSA over conventional arthroplasty for patient with rotator cuff deficiencies is that it uses a semiconstrained design that provides a mechanical advantage to the deltoid muscle allowing it to participate in forward elevation [1-3]. RSA provides functional improvements to a subgroup of patients for whom no other good solution previously existed [1, 2, 4]. However, given these indications, it can be problematic to decide when outcomes are optimized.

RSA is performed in patients with substantial impairment and disability prior to surgery; therefore, full function recovery is unlikely. Furthermore, since function is multifactorial, it can be difficult to ascertain functional outcome based on the evaluation of isolated impairments like range of motion and strength. For this reason, some consider patient satisfaction an essential criterion since it forms a global rating. Satisfaction, however, can be related to process as well as outcomes achieved, and does not reflect change over time [5].

Clinicians and researchers share a common goal of determining whether interventions provide sufficient benefit to warrant their implementation. These thresholds can be problematic to define in all reconstructive surgeries, but are especially difficult in RSA where the pathology is complex and leads to extensive impairment and disability. Thus, RSA provides a good model for variations in outcome definition. The purpose of this study was to examine concordance between different methods of determining outcomes following RSA.

Significant positive correlations were observed between the two shoulder function scales (Table 4). Furthermore, the CMS and SST were correlated with: 1) pain level; 2) range of motion; and, 3) isometric strength. Significant positive correlations were also observed between the CMS pain subscale and 1) age, and, 2) range of motion in internal and external rotation at 90° of abduction.

This study illustrates that outcomes of RSA varied according to patient’s perspective and the assessment method used, with varying rates of concordance between different outcome measures. Patient satisfaction with their care was the outlier in that it was not correlated with other outcomes and exceeded 90% despite substantial residual pain, range of motion and strength deficits and functional limitations. From this perspective, patient satisfaction with their care was relatively non-informative. The Global Clinical Outcome process is an extension of an adjudication process sometimes used in clinical trials and allows experts to weight multiple factors and complex considerations. The Global Clinical Outcome, while potentially the best reflection of overall outcome, requires extensive chart reviews with independent expert raters and thus is not practical for routine assessment.

From the remaining standardized outcomes, the SST and CMS seemed to be the best indicator of the overall success of a RSA surgery since they were equally related to impairment, motion and strength scores. The CMS agreed to a greater extent with the Global Clinical Outcome, which may reflect that the subscales of the CMS address the criteria that the clinicians consider when deciding whether the surgery was successful. Conversely, the SST agreed with categorical subgroups of external rotation range of motion and strength, whereas the CMS did not agree with these subgroups. This suggests that the SST is better at differencing functional subgroups, since external rotation range of motion and strength contributes to functional ability following a RSA [1, 3, 13-15].

The agreements between different outcome measures were fairly low, with only 11 of 51 comparisons exceeding chance. This suggests that conclusions on the success of the treatment would depend heavily on the particular outcome modality chosen. This is an agreement with work done on the elbow rating scales that showed similar discordance between different outcome measures [16]. Two potential reasons could explain the poor agreement between different outcome measuring tools. First, it is possible that the different outcome methods reflect different constructs or perspectives on outcome. In this case, individual results for a particular method of assessing outcome might be true for that perspective, but provide little information on other perspectives. Secondly, it is also possible that, by defining categories of success using specific cut-offs, important information was lost. However, correlations between raw scores confirmed that the relationships between different outcomes are moderate at best. The implications of this are that different outcome measures should be used in clinical practice and research in order to provide a comprehensive view of outcome.

Our methodology required setting the cut-offs for successful versus unsuccessful outcomes. We did this because the decision of success versus failure is a common judgment. We used literature on functional requirements and statistical properties of the measures to establish these cut-offs, but recognize they are somewhat arbitrary. The minimal detectable change (MDC) is a statistical measure defining the smallest difference or change that would be statistically significant when comparing samples. The MDC of SST and CMS are unknown. However, the MDC of the visual analog scale, such as the ones used in the CSM, is 20% [17] and the MDC of other self-report shoulder scales is also close to 20% [17]. Since two standard deviations is a common definition when defining normality, the approach we used for defining the threshold for a successful functional outcome was a variation from normal score that exceeded twice the MDC. Since the actual shoulder strength needed to perform different functional activities has not been determined, the same approach was used to compare strength deficits to normative data. Most of the patients undergoing a RSA will have rotator cuff deficiencies, and therefore significant preoperative weakness [18]. Post-surgical restoration of normal strength is not expected. Since we observed relatively little agreement between functional strength and either functional range of motion or self-report outcomes, it appears that even patients with low levels of strength may be functional following RSA.

Agreements beyond chance were observed between the SST and functional external rotation range of motion and strength; confirming that external rotation is an important aspect of shoulder function. Rotator cuff tear arthropathy can be associated with atrophy or fatty infiltration of infraspinatus and teres minor in patient with a combined loss of active elevation and external rotation [3, 19]. Although RSA can restore active elevation in patient with rotator cuff deficiencies, it cannot restore active external rotation when both the infraspinatus and teres minor muscles are absent or atrophied [2, 3, 20]. Simovitch and colleagues have shown that stage-3 or 4 fatty infiltration of the teres minor compromises the outcome of RSA [20]. In these patients, preliminary results suggest that the addition of a latissimus dorsi and teres major transfer to a concomitant RSA could be an option in order to improve active external rotation and, therefore, function [3].

Our data suggests that patient’s satisfaction with their care rating is a poor indicator of shoulder function. Ninety-three percent of the subjects were satisfied even though some of the satisfied patients had less than 50° of shoulder elevation, no active external rotation, moderate to severe pain and postoperative complications. Wall and colleagues observed the same percentage (93.0%) of satisfied patients following RSA [4]. Since patients undergoing RSA are typically older and percent with disability, they have substantially modified their activity expectations. Small improvements in pain relief and physical impairment may improve their ability to complete valued low demand activities. Schmitt et al. [5] demonstrated that patient satisfaction ratings do not accurately reflect true change over time since they are heavily influenced by current status. Others have indicated that patient satisfaction is highly influenced by process such as wait times or friendliness of staff. It is also possible that patients appreciate the efforts made to deal with their complex shoulder disability, particularly since other health care providers would have been unable to offer any solution. Only qualitative studies would shed light on the actual thinking of patients in determining this level of satisfaction. However, these findings suggest that healthcare providers should avoid assuming that operation was worthwhile performing because the patient is satisfied with its care.

Complication rates following RSA have been shown to be close to 25% [21]. In the present study, there were a total of eleven complications in nine patients (20.5%). At follow-up, only one patient with postoperative complications was not satisfied and there were no significant differences between patients with and without complications for function, pain, range of motion and strength. Six of the nine patients with complications, however, were considered as having a poor Global Clinical Outcome. This suggests that complications affected clinicians’ perspective on surgical success to a greater extent than it did for the patients. Some patients experience complications that require additional surgical procedures and additional medical care but ultimately were more functional than they had been prior to surgery. Complications do not necessarily lead to unsatisfied patients or to poor functional outcomes if they can be adequately resolved. This suggests that patients should be informed not only on the absolute rate of surgical risk, but also what might be required in terms of subsequent management to resolve surgical complications.

This study does have limitations. While we used clinical measurement approaches to dichotomized outcomes, these cut points could not be validated as the best inflection points for function or success. Furthermore, the wording of the satisfaction question may have influenced patients to reflect on process given that it specifically mentioned satisfaction with their care rather than asking to perform a global rating on their shoulder function or improvement of their condition. Therefore, our findings do not generalize to global rating of change scores sometimes used when evaluating patients. Strengths of this study are the use of a single surgical procedure that exemplifies the challenge in defining surgical success, and that independent evaluators were used to adjudicate outcomes.

In conclusion, since different outcome measure do not agree with each other in terms of whether a functional outcome has been achieved, it would be important to sample different constructs-i.e. both measured impairments and self-reported disability. Despite concerns about self-report measures, these were equally or more related to independent expert global assessment than were measures of physical impairments.