1 Introduction

Shoulder arthroplasty volume in the United States has experienced a significant increase over recent decades. This burden is expected to continue to grow due to an aging population and the expanding indications of shoulder arthroplasty.¹ Projections estimate that, by 2030, the incidence of shoulder arthroplasty will increase 755% among patients aged 55 and above and also by 333% among patients younger than 55 years old.² The considerable rise in prevalence and the earlier average age of shoulder arthroplasty procedures imply an inevitable increase in the volume of revision surgeries.³

Revision shoulder arthroplasty is a surgical procedure performed to address complications or failures of a primary shoulder arthroplasty. Revision from anatomic total shoulder arthroplasty (TSA) to reverse total shoulder arthroplasty (RSA) is commonly advised for cases involving implant loosening or rotator cuff pathology, whereas revising from RSA to RSA is typically pursued to address instability or fracture.⁴ Conversely, the less common transition to hemiarthroplasty (HA) subsequent to failed prior interventions often signifies a salvage approach in the presence of significant unamenable bone loss.⁵ Recent studies have sought to elucidate various factors which predispose implant failure and require revision shoulder arthroplasty, the most common of which include rotator cuff insufficiency, joint dislocation, and implant loosening.⁶ Although existing research characterized the incidence and risk factors for revision surgery, there is a paucity of information documenting the postoperative functional outcomes and mid- to long-term survivorship of the revision surgery.^7–9 Additionally, current studies exhibit relatively small sample sizes and/or short-term follow-up, warranting the need for more extensive evaluation.

The primary aim of this study is to explore the functional outcomes and durability of implants after revision shoulder arthroplasty for aseptic indications. Additionally, we seek to characterize the common indications for primary and secondary revisions. We hypothesize that patients who undergo revision shoulder arthroplasty will exhibit improved objective and subjective outcomes after the operation, with minimal occurrences of re-revision surgery.

2 Methods

3 Results

During the study period, 137 revision shoulder arthroplasties were identified. 99 out of the 137 (72.3%) identified shoulders had undergone a revision shoulder arthroplasty and met the inclusion criteria. Thirty-eight patients were excluded for undergoing revision for prosthetic joint infection, inaccessible operative reports, had not met the minimum requirement of 2-year follow-up appointments with their orthopedic healthcare provider, or misclassified as a CPT coding error. There were 57 female (57.6%) and 42 male (42.4%) subjects in the cohort. They had a mean BMI of 31.1 ± 7.2 kg/m², with the mean age at index and revision surgeries of 60.5 ± 12.1 years and 64.8 ± 11.1 years, respectively. BMI did not significantly affect age of index surgery or interval before revision. Thirteen were current smokers, 29 were former smokers, and 55 were never smokers. Current smokers underwent index surgery at a significantly younger age (current: 49.8 ± 10.9 vs former: 63.9 ± 10.6 vs never: 62.1 ± 11.7 years; p < 0.001) but had a longer average interval before requiring a revision surgery (current: 6.4 ± 4.1 vs former: 6.4 ± 3.7 vs never: 4.5 ± 3.5 years; p = 0.0185). The average total follow-up for the cohort was 5.5 ± 3.5 years. Full demographic data can be found in Table 1. Among revisions to TSA/HA (n = 30), the most common indications included progressive arthritis (n = 11), glenoid loosening (n = 5), instability (n = 4), and repair of rotator cuff pathology (n = 4). Among revisions to RSA (n = 69), the most common indications included instability (n = 25), rotator cuff pathology (n = 14), peri-prosthetic fracture (n = 10), and component loosening (n = 9) (Table 2).

The 99 revision shoulder arthroplasty procedures were subcategorized into 4 groups: 1) TSA/HA to TSA/HA (n = 28); 2) TSA/HA to RSA (n = 51); 3) RSA to RSA (n = 18); 4) RSA to HA (n = 2). Revision TSA/HA to TSA/HA significantly improved forward elevation (pre: 98.2° ± 42.8° vs post: 120.8° ± 33.9°; p = 0.0295), but improvements in external rotation, internal rotation, and all strength testing were insignificant. Patients receiving TSA/HA to RSA benefitted the most from revision surgery, with significant improvements in forward elevation (pre: 95.5° ± 36.7° vs post: 133.3° ± 29.5°; p < 0.001), internal rotation (pre: greater trochanter vs post: S1; p = 0.0069), and all strength testing (pre: 4/5 vs post: 5/5). RSA to RSA only significantly improved forward elevation (pre: 102.9° ± 46.0° vs post: 130.8° ± 33.1°; p = 0.0124), but differences in external rotation, internal rotation, and strength testing were not significant. As only 2 patients received RSA to HA, the sample size was not large enough for a powerful analysis. Full functional measurement data, stratified by the four groups, can be found in Table 3. The average postoperative VAS pain, SANE, ASES, and SST scores were 2.2 ± 2.9, 72.6 ± 21.5, 73.3 ± 20.4, and 7.7 ± 2.8, respectively. There were no significant differences between revision procedure types (Table 4).

In multivariate cox regression hazard ratio analysis, adjusting for demographic and comorbidity status, no variable (age, sex, smoking status, BMI, CCI, or revision type) was identified as a significant risk factor for reoperation after revision (Fig. 1). The overall 2-year, 5-year, and 10-year post-revision implant survival rate was 85.5%, 83.1%, and 79.8%, respectively, with a mean time to re-revision of 2.8 ± 2.9 years. Patients who had an initial RSA and were revised to RSA were at higher risk subsequent re-revision (RSA to RSA: 1.5 ± 2.5 years vs. TSA/HA to RSA: 2.5 ± 2.1 years vs. TSA/HA to TSA/HA: 4.0 ± 3.5 years; p = 0.0046) (Fig. 2). Complications requiring re-revision or reoperation occurred in 15 cases, including 5 instability, 4 infection, 3 component loosening (2 glenoid and 1 humeral), 1 periprosthetic fracture, 1 progressive arthritis, and 1 rotator cuff tear (Table 5).

Close

Fig. 1 Multivariable cox regression analysis of hazard ratios associated with implant failure after revision shoulder arthroplasty. * CCI: Charlson Comorbidity Index; BMI: body mass index.

Export to PPT

Close

Fig. 2 Adjusted analysis of implant survival following revision total shoulder arthroplasty, by patient gender.

Export to PPT

4 Discussion

The prevalence of shoulder arthroplasties has experienced a rapid escalation in recent decades.² Consequently, there has been a corresponding surge in the number of revision shoulder arthroplasties performed, necessitating a heightened demand for resources and an unprecedented dedication to meet these evolving requirements.¹⁴ Medical decision making after the failure of primary arthroplasty is a complex and patient-specific process, requiring consideration of various demographic factors, rotator cuff integrity, and radiological findings affecting the type of revision arthroplasty performed.^15–18

Herein, we examined the outcomes and complications associated with aseptic revision shoulder arthroplasty procedures. Revision technique may vary based on clinical factors and advantages provided by each approach. For example, revision from TSA to RSA is often indicated for glenoid loosening or rotator cuff failure, while revision from RSA to RSA is typically done for instability or periprosthetic fracture.⁴ Conversion to HA after failed TSA or RSA is rarer, and usually indicative of a salvage operation in light of significant glenoid bone loss that is not amenable to reconstruction.⁵ Of note, the most common indications for revision shoulder arthroplasty were instability and rotator cuff pathology after TSA. Previous research has indicated the association of rotator cuff deficiency with glenoid loosening and implant failure for anatomic TSA, suggesting the importance of post-operative rehabilitation in implant longevity.¹⁹ Our results further suggest that revision shoulder arthroplasty significantly improved functional outcomes observed postoperatively. This was evident through improvements in active range of motion and related strength measures, which was most significant among patients revised from TSA/HA to RSA. Notably, while revision to from TSA/HA to RSA resulted in marked functional improvements in almost every category, revision from TSA/HA to TSA/HA or from RSA to RSA only significantly affected forward elevation. Similarly, a systematic review and meta-analysis of 15 studies including 593 revision shoulder arthroplasties conducted by Davies et al. revealed that revision to both HA and TSA did not yield consistent successful improvements in functional shoulder outcomes.⁷ Conversely, existing literature has suggested that functional outcomes after revision RSA lead to significant functional and radiographic improvements compared to preoperative measurements.^17,20,21 A study by Abdel et al. investigated outcomes after revision arthroplasty for instability, in which 31 patients treated with revision RSA showed decreased mean VAS pain score, improved active forward elevation, and substantially improved stability in 94% of cases.²² This suggests that revision from TSA to RSA may be functionally superior to alternative revision types in cases of soft-tissue related failure of previous arthroplasty types, which accounted for the majority of the indications for revision analyzed here.

Despite improvements in shoulder pain and functionality, previous studies have concluded that revision shoulder arthroplasty is a complex and challenging procedure associated with risk for prosthetic instability and significantly high complication rates, as high as nearly 40–70% in some studies, as well as poor implant longevity.^17,23–25 For instance, a study performed by Boileau et al., which analyzed 45 cases of RSA at a mean follow-up of 40 months, observed that patients undergoing revision RSA had significantly lower ASES and Constant scores than patients undergoing primary RSA; revision RSA patients accounted for 9 of 14 total complications, even though revision RSA accounted for less than half (21 of 45) of the total cases investigated.²⁶ In our study, we found an overall reoperation rate of 15% with higher success among TSA/HA to RSA, which is consistent with existing studies.^15,20 These findings suggest that conversion from TSA to RSA may provide satisfactory long-term implant survival compared to other options, as also evidenced by a recent systematic review of 5225 patients having undergone revision shoulder arthroplasty with an average follow-up of 4 years by Ravi et al.¹⁸ Further studies are warranted to optimize outcomes of revision arthroplasty procedures and to determine revision indications that are better approached by specific treatments.

The overall success rate of revision shoulder arthroplasty was moderate, as implant survival rates at 2-years, 5-years, and 10-years post-revision were 85.48%, 83.06%, and 79.84%, respectively. These rates indicate that most revised implants remained functional over a significant period. As previously mentioned, complications requiring re-revision or reoperation occurred in approximately 15% of cases, as infection was alarmingly common. This rate of infection is consistent with that of other studies, ranging from 10 to 16% in revision shoulder arthroplasties, but only occurring in 1 to 3% of primary total shoulder arthroplasties.^24,26–28 Of further note, patients revised from RSA to RSA experienced a significantly higher risk of implant failure than TSA/HA to TSA/HA. This finding raises important considerations regarding the influence of revision approach on implant survival and outcomes following revision shoulder arthroplasty. Further research is warranted to explore the underlying mechanisms contributing to these differences and to identify strategies for mitigating the increased risk in RSA patients. Overall, these results highlight the ongoing challenges and risks associated with revision arthroplasty, emphasizing the need for careful patient selection, surgical technique optimization, and most importantly, infection prevention strategies.

Several limitations should be taken into consideration when interpreting the results of this study. This patient population was sampled from a single institution, potentially limiting the external generalizability of our results. Furthermore, the relatively small sample size restricted statistical analysis. For example, TSA and HA were grouped for the sake of sample size and based on their similar mode of failure, though they are different procedures that should be considered separately in studies with larger sample size. Moreover, the sample size did not allow for a more sophisticated stratification and analysis of PROs (i.e., by revision type) due to insufficient response rate and overall cohort size. Notably, the retrospective study design itself is associated with systematic errors, notably the inability to determine causal relationships, as well as potential selection and misclassification biases due to potential inconsistencies between medical record reviews during the data collection process. Similarly, as a retrospective study, PROs were not collected prior to the operation. A prospective, multi-center randomized controlled trial incorporating a large sample size and longer, standardized follow-up time would likely lead to improvements in generalizability and allow for more detailed analysis of the underlying mechanisms governing the extent of functional recovery and implant success after revision shoulder arthroplasty.

5 Conclusion

Revision shoulder arthroplasty significantly improved patient outcomes post-primary arthroplasty failure. Infection is the most common indication for revision failure, occurring in 13% of cases. Revisions were more likely to be successful when revising from anatomic total shoulder to reverse total shoulder arthroplasties.

Disclaimer

None.

Ethical statement

This study was approved by the Biomedical Institutional Review Board of The Ohio State University.

Funding statement

The authors received no funding for this study.

Guardian/patient's consent

Consent was not required by our institution for this retrospective study.

CRediT authorship contribution statement

Erryk S. Katayama: Project administration, Data curation, Visualization, Writing – original draft. Louis W. Barry: Data curation, Writing – original draft. John S. Barnett: Data curation, Writing – original draft. Amogh I. Iyer: Formal analysis, Visualization, Writing – original draft. Akshar V. Patel: Conceptualization, Project administration, Writing – review & editing. Julie Y. Bishop: Conceptualization, Writing – review & editing. Gregory L. Cvetanovich: Conceptualization, Writing – review & editing. Ryan C. Rauck: Conceptualization, Supervision, Writing – review & editing.