Return to Sport after Anatomic and Reverse Total Shoulder Arthroplasty in Elderly Patients: A Systematic Review and Meta-Analysis

The aim of this systematic review and meta-analysis was to evaluate the rate of return to sport in elderly patients who underwent anatomic (ATSA) and reverse (RTSA) total shoulder arthroplasty, to assess postoperative pain and functional outcomes and to give an overview of postoperative rehabilitation protocols. A systematic search in Pubmed-Medline, Cochrane Library, and Google Scholar was carried out to identify eligible randomized clinical trials, observational studies, or case series that evaluated the rate of return to sport after RTSA or ATSA. Six retrospective studies, five case series, and one prospective cohort study were included in this review. The overall rate of return to sport was 82% (95% CI 0.76–0.88, p < 0.01). Patients undergoing ATSA returned at a higher rate (90%) (95% CI 0.80–0.99, p < 0.01) compared to RTSA (77%) (95% CI 0.69–0.85, p < 0.01). Moreover, the results showed that patients returned to sport at the same or a higher level in 75% of cases. Swimming had the highest rate of return (84%), followed by fitness (77%), golf (77%), and tennis (69%). Thus, RTSA and ATSA are effective to guarantee a significative rate of return to sport in elderly patients. A slightly higher rate was found for the anatomic implant.


Introduction
Total shoulder arthroplasty (TSA) is the third most common replacement procedure after hip and knee arthroplasty and is considered the elective treatment for patients affected by advanced shoulder pathology with loss of function and severe pain [1]. Indeed, the main indications for shoulder replacement are primary and secondary glenohumeral osteoarthritis, osteonecrosis, and fractures of the proximal epiphysis of the humerus or its sequelae [1][2][3][4][5].
Three main different designs of shoulder prostheses allow surgeons to decide which is the best option for each specific case, trying also to meet patients' health needs [6]. The reverse shoulder arthroplasty (RTSA) has a specific design that determines the medialization of the rotation center, permitting the recruitment of a large part of the deltoid muscle [7]. Thus, even if the rotator cuff is damaged, the arc of movement is preserved [7]. Conversely, the biomechanics of the anatomic total shoulder arthroplasty (ATSA) is based on rotator cuff integrity. The third implant option includes scores. The tertiary endpoint of this systematic review was to give an overview of the proposed rehabilitation protocols.

Search Methods for Identification of Studies
A systematic literature search in Pubmed-Medline, the Cochrane Library and Google Scholar databases was carried out between September 2019 and April 2020. For Pubmed, the following search strategy was used: ((("Shoulder Joint"[Mesh] OR ("shoulder"[All Fields] AND "joint"[All Fields]) AND ("Arthroplasty"[Mesh] OR ("Arthroplasty"[All Fields]) OR ("Replacement"[All Fields])) AND ("Sports" [Mesh] OR ("return to sport" [All Fields] OR ("return" [All Fields] AND ("sport" [All Fields])) AND "Aged"[Mesh])). No time interval was set for publication date. Two independent reviewers (M.C. and C.D.A.) conducted the electronic search identifying the potentially relevant studies. Firstly, the retrieved articles were screened by title and, if relevant, by reading the abstract. After the exclusion of non-eligible studies, the full-text of the remaining articles was evaluated for eligibility. To minimize the risk of bias, the authors reviewed and discussed all the selected articles, the references, as well as the articles excluded from the study. If any disagreement between the reviewers was found, the senior investigator (R.P.) made the final decision. At the end of the process, further potentially missed studies were manually searched for among the reference lists of the included papers and the relevant systematic reviews.

Data Collection
All reviewers discussed the relevant items for data extraction before starting the process in order to avoid data omission. Data were independently extracted by two reviewers (M.C. and C.D.A.) and divergences were discussed with the third reviewer (R.P.) if necessary. All data related to primary, secondary and tertiary outcomes were summarized in standardized tables. Specifically, the following variables were recorded: authors, year of publication, type of study, level of evidence, number of participants, mean age, dominant or not-dominant limb, surgical approach, mean follow-up, complications, patients returned to sport and type of activity, secondary outcome measures, and rehabilitative protocols. Among the outcomes, we analyzed functional outcomes and severity of pain.

Risk of Bias Assessment
The quality of the included studies was independently evaluated by two reviewers (C.D.A. and M.C.) using the Methodological Index for Non-randomized Studies (MINORS) score [15]. The following domains were assessed: a clearly stated purpose, inclusion of consecutive subjects, prospective data collection, endpoints appropriate to the purpose of the study, unbiased assessment of the study endpoints, follow-up period appropriate for the study, loss to follow-up of less than 5%, prospective calculation of the study size, adequate control group, contemporary group, baseline group equivalence, and adequate statistical analysis. The last four items are specific to comparative studies. Each item was scored from 0 to 2 points, with a global ideal score of 16 points for non-comparative studies and 24 points for comparative studies.

Statistical Analysis
Meta-analysis was performed to determine the overall proportion of subjects returning to sport and the functional and pain level after shoulder arthroplasty across all the retrieved studies. Raw, i.e., untransformed, proportions and means were used to report the pooled proportions and means that were obtained with the inverse variance method. Heterogeneity was evaluated using Q statistic, expressed as the p value for the χ 2 test under the null hypothesis that the between-study variance (τ 2 ) equals 0, and I 2 test. All the conducted meta-analyses evidenced the presence of significant heterogeneity, defined as a I 2 > 55% and a Q statistic p value below 0.05. Accordingly, random effect models were applied. Finally, the likelihood of publication bias was estimated with a visual inspection

Study Selection
The initial database searches identified 235 potentially eligible papers. After reviewing title and abstract, 217 papers were excluded and 18 were selected for full-text evaluation. Out of these, eight papers were excluded for the following reasons: mean age of the cohort < 65 years (n = 1), German language (n = 1), return to sport not clearly stated (n = 4), and insufficient outcomes data (n = 2). One paper was added from hand search. At the end of the selection process, 11 studies were included in this systematic review and 11 papers were included in the meta-analysis [16][17][18][19][20][21][22][23][24][25]. The search process is summarized in the PRISMA flowchart ( Figure 1) [14].

Study Characteristics and Demographic Details
Of the included studies, one was a single-center prospective cohort study (PCS) of level of evidence (LOE) III [19], four were retrospective studies (RS) of LOE III [18,20,23,25], one was a retrospective study of LOE IV [26], and five were case series of LOE IV [16,17,21,22,24]. All studies were published between 2010 [21] and 2018 [26]. The 11 included studies reported on 1254 shoulder arthroplasties in 1238 patients. Within the included studies, the number of subjects varied from 35 [24] to 276 [26]. The mean age of the cohorts was 72.5 years. The mean duration of follow-up was 3.7 years, ranging from 2.4 [26] to 6.2 years [17]. Four out of the 11 studies had a mean follow-up longer than four years [16,17,19,23]. The indications for surgery were several: rotator cuff arthropathy (522 patients), primary osteoarthritis (270 patients), fracture sequelae (37 patients), and rheumatoid arthritis (15 patients). Pre-operative diagnosis was not specified for 367 patients [19,22,23,26].

Study Characteristics and Demographic Details
Of the included studies, one was a single-center prospective cohort study (PCS) of level of evidence (LOE) III [19], four were retrospective studies (RS) of LOE III [18,20,23,25], one was a retrospective study of LOE IV [26], and five were case series of LOE IV [16,17,21,22,24]. All studies were published between 2010 [21] and 2018 [26]. The 11 included studies reported on 1254 shoulder arthroplasties in 1238 patients. Within the included studies, the number of subjects varied from 35 [24] to 276 [26]. The mean age of the cohorts was 72.5 years. The mean duration of follow-up was 3.7 years, ranging from 2.4 [26] to 6.2 years [17]. Four out of the 11 studies had a mean follow-up longer than four years [16,17,19,23]. The indications for surgery were several: rotator cuff arthropathy (522 patients), primary osteoarthritis (270 patients), fracture sequelae (37 patients), and rheumatoid arthritis (15 patients). Pre-operative diagnosis was not specified for 367 patients [19,22,23,26].

Methodological Evaluation
The MINORS score ranged from 7 [17,25] to 11 [23] for non-comparative studies and from 14 [16,18] to 16 [20,26] for the comparative ones ( Table 1). The mean value was 8.5 for non-comparative studies and 14.5 for comparative studies. All papers resulted at high risk of bias.

Methodological Evaluation
The MINORS score ranged from 7 [17,25] to 11 [23] for non-comparative studies and from 14 [16,18] to 16 [20,26] for the comparative ones ( Table 1). The mean value was 8.5 for non-comparative studies and 14.5 for comparative studies. All papers resulted at high risk of bias.

Rehabilitation Protocols
Only 5 out of 11 included papers reported the postoperative rehabilitation protocol [17,18,22,23,25]. Those authors advised a shoulder sling immobilization for the first four weeks, leaving free elbow and wrist movements. [17,18,22,23,25] In general, only passive ROM was allowed for the first 4 weeks, waiting for the sixth postoperative week to start active exercises. [17,18,22,23,25] Strengthening exercises were generally allowed from the twelfth postoperative week [18,22], even if Barnes et al. started them from the eighth [23]. On the contrary, Kolling et al. [25] permitted active mobilization and water therapy for shoulder strength and coordination from the second week after surgery. The surgical approach was evaluated in order to correlate subscapular repair to restrictions in the rehabilitative protocol. Among the five surgeons who performed the subscapularis tendon repair [16,17,21,23,25], only Kolling et al. [25] chose to limit external rotation movements to protect the reinserted tendon until the end of the second postoperative week. The postoperative rehabilitation protocols and the surgical approach are reported in Table 4.

Rehabilitation Protocols
Only 5 out of 11 included papers reported the postoperative rehabilitation protocol [17,18,22,23,25]. Those authors advised a shoulder sling immobilization for the first four weeks, leaving free elbow and wrist movements. [17,18,22,23,25] In general, only passive ROM was allowed for the first 4 weeks, waiting for the sixth postoperative week to start active exercises. [17,18,22,23,25] Strengthening exercises were generally allowed from the twelfth postoperative week [18,22], even if Barnes et al. started them from the eighth [23]. On the contrary, Kolling et al. [25] permitted active mobilization and water therapy for shoulder strength and coordination from the second week after surgery. The surgical approach was evaluated in order to correlate subscapular repair to restrictions in the rehabilitative protocol. Among the five surgeons who performed the subscapularis tendon repair [16,17,21,23,25], only Kolling et al. [25] chose to limit external rotation movements to protect the reinserted tendon until the end of the second postoperative week. The postoperative rehabilitation protocols and the surgical approach are reported in Table 4. Mean VAS Figure 10. Forest plot chart of VAS score.  (2) Patients with glenohumeral osteoarthritis and rotator cuff disease being active prior to RSA surgery are able to successfully return to their level of sports participation afterwards.

ASES SCORE (overall mean change) +39
None Despite traditional sport restrictions placed on RTSA, patients undergoing RTSA can return to sports at rates higher than those undergoing HHA, with fewer postoperative complaints.

Kolling et al. [25]
RTSA NR NR -Most patients carried out their main sports activity after surgery with a moderate level of intensity (83%) and between one to three times per week (69%). -42% indicated that returning to sports was among their key demands after RSA.  preoperative. Similar results were showed by Simovitch et al. [22], with a mean difference between preoperative and postoperative VAS of 6.1 points. Additionally, they showed that postoperative pain reduction was associated with an improvement of ROM and ASES scores [22]. Three studies reported the difference between preoperative and postoperative ASES and, in all of them, an improvement in postoperative values can be observed [18,20,22]. Barnes et al. [23] reported only the mean postoperative ASES which was 77.5, but even in this case, the improvement of ASES scores and VAS was associated with return to sport at the same or better preoperative level. The present meta-analysis has shown that patients returned to sport activities at the same or a higher level in 75% of cases. This confirms that most patients undergoing shoulder arthroplasty (regardless of type) can safely return to at least one sport, with many returning to the same level of play, although a 100% guarantee should not be provided. Bulhoff et al. [17] assessed that, in their cohort, the postoperative activity levels and frequencies in sports practice were higher than before surgery. Moreover, patients were satisfied with their performances. Kolling et al. [25] selected 69 patients who clearly expressed their desire to resume sports activities after surgery and 60% of these patients were satisfied with their postoperative performance level and, within a year from surgery, 86% returned to practice sport at the same preoperative level or higher.
The postoperative rehabilitation protocol was reported in five studies [17,18,22,23,25]. Available protocols provided general information about time of immobilization and gradual recovery of shoulder motion and strength. Generally, the majority of surgeons followed similar indications: sling immobilization for at least four weeks, passive ROM for the first four weeks, active exercise from about the sixth postoperative week and strength training from the 12th postoperative week. On the contrary, Kolling et al. [25] permitted active mobilization and water therapy for shoulder strength and coordination in the second week after surgery. Unfortunately, the current literature lacks a detailed description of the rehabilitative steps and specific information about training for the athletic population. Moreover, to the best of our knowledge, no high-level evidence trials have been performed to test the efficacy of different post-operative rehabilitation protocols for patients who underwent TSA. However, some authors demonstrated that patients who received a physician-directed rehabilitation program had a significantly better range of motion as compared to patients only supervised by physiotherapists [12].
This systematic review has a few limitations including the number of studies and their heterogeneous methodological approach. Moreover, designs and implantation techniques may have varied significantly across the analyzed studies, thus reflecting the sparse available evidence on the subject and the absence of randomized controlled trials. Importantly, none of these studies mentions the abilities and experience of the surgeon. Since ATSA involves greater operative time and attention, surgical experience could be a determining factor in the decision to perform a reverse or anatomic total shoulder arthroplasty. In order to create a more homogenous cohort, future studies should account for these individual surgeon factors in the methodology. Moreover, all the included studies were affected by a high risk of bias and, in some of them, the follow-up period was quite short to detect important postoperative complications after return to sport, such as loosening or periprosthetic fractures. Patients and sports were heterogeneous as well as the postoperative rehabilitation protocol assessed. Great variability was observed in the postoperative treatment protocols following shoulder arthroplasty. Therefore, it is very difficult to identify common patterns, making it impossible to do a metanalysis of postoperative rehabilitation protocols. Moreover, we performed the metanalysis only on postoperative ASES and VAS scores since their preoperative data were not reported in the included studies, hindering the assessment of significant improvements of these postoperative outcomes. Finally, important postoperative clinical outcomes, such as postoperative ROM, were often not reported.

Conclusions
After ATSA and RTSA, elderly patients can satisfactorily resume their sports activities. The rate of return to sports following ATSA is slightly higher than RTSA, probably due to differences in the patient population, surgical indication, and biomechanical issues. Most patients are able to return to practice sport at the same or a higher preoperative level. The most practiced sports after surgery are low contact activities such as fitness, swimming, golf, and tennis. Unfortunately, there is a lack of research data on the advantages and disadvantages of existing rehabilitation protocols and no standard of practice could be deduced. Therefore, more prospective randomized studies are needed to establish which kind of postoperative protocol is best following ATSA and RTSA.