1 Introduction

Shoulder arthroplasty is a successful procedure that provides pain relief and improvements in range of motion. Both anatomic and reverse total shoulder arthroplasty (TSA) have been shown to be reliable procedures in the elderly population.¹ While reverse TSA (rTSA) has increased in popularity in the US in the last decade, anatomic TSA provides more natural range of motion and function; it is still indicated in higher-demand patients with a competent rotator cuff who hope to continue using their arm to its full effect.^2,3

A number of complications necessitating revision surgery have been described for these procedures including infections, such as Cutibacterium acnes (previously Propionibacterium acnes), which is a slow-growing gram-positive bacteria often in moist skin areas like the axilla.⁴ Furthermore, it has been shown that the rate of infection in multiple revision cases is significantly higher than in first revision cases.⁵ Other indications for revision surgery include instability, implant loosening, periprosthetic fractures, glenoid complications, cuff incompetence and others.⁶ Multiple studies have shown that there is an improvement in shoulder function after revision rTSA, although outcomes may not be ideal.^7–10

In this study, we look at the outcome of revision reverse total shoulder arthroplasty and compare it to the outcomes of primary reverse and anatomic total shoulder arthroplasty. We hypothesize that those with revision reverse TSA will have worse long-term outcomes both in range of motion and ability to perform activities of daily living (ADLs) than after primary reverse or anatomic total shoulder arthroplasty.

2 Material & methods

We identified patients who had one or more revision total shoulder arthroplasties at our institution between the years of 2010 and 2020. The surgeries were completed by a single surgeon and all shoulders were converted to Grammont-style reverse total shoulder arthroplasties. Data was prospectively collected and retrospectively reviewed for post-operative range of motion, strength, and ADLs in patients with revision surgery and compared to primary controls. Prospective measurements were collected preoperatively and postoperatively including range of motion and strength in the affected and unaffected shoulder. Range of motion for external rotation at 0°, 90°, abduction, and forward elevation were recorded in degrees. Internal rotation, crossbody adduction and external rotation above head were reported qualitatively and converted to numbers for statistical analysis. Reported numbers are the average levels of improvement on the appropriate scale. Internal rotation behind the back was reported as above T7; T7; T8-12; L1-L3; L4-L5; the lumbosacral junction; the buttock; the hip; and below the hip. Crossbody adduction was reported as full, AC joint, anterior acromion, deltoid insertion, antecubital fossa, and chest. External rotation above the head was reported as full elevation from the top of head; hands on top of head with elbows back; hands on top of head with elbows forward; hands behind head with elbows back; and hands behind head with elbows forward. Primary anatomic and reverse shoulder arthroplasties with complete prospective data and follow-up were then identified and analyzed for our control groups.

We collected patient reported outcome measures (PROMs) in person and via phone to identify subjective outcomes of total shoulder arthroplasty. Average follow-up was 5.27 years. Specifically, we investigated Visual Analog Scale (VAS) pain scores along with two functional tests: the Simple Shoulder Test (SST) and the American Shoulder and Elbow Surgeons Evaluation (ASES). Exclusion criteria included axillary nerve palsy, other neuromuscular conditions, latissimus dorsi, or other muscle transfer prior to primary arthroplasty. We eliminated those who lacked a significant amount of pre-op and post-op range of motion data. Patients with minor individual data points missing but otherwise intact follow up at 12 and 24 months were still included in the analysis resulting in some small discrepancies in subject number at the various time points.

Means and standard deviations were used to summarize continuous variables; frequencies and percent values were used for categorical variables. Analysis of covariance (ANCOVA) was used to investigate improvement over time in each group as well as to compare groups separately at one and two years controlling for the baseline value. We used SAS statistical software version 9.4 (SAS Institute, Cary, NC, USA) and to reduce familywise type 1 error rate we set declared statistical significance when p-values were equal or below 0.01.

3 Results

3.2

3.2 Range of motion improvements from pre-operative baseline

All three groups had significant improvement after two years in abduction (PA: 37.06°, PR: 47.77°, RR: 39.90°) and forward elevation (PA: 31.98°, PR: 51.05°, RR: 46.08°) from pre-op to post-op (Fig. 1 and Table 2). External rotation above the head (1.49 levels for PA and 1.38 levels for PR), external rotation at 90° (PA: 15.32°, PR: 16.91°), and crossbody adduction (PA: 1.24 levels, RR: 0.59 levels) had significant improvement in the PR and PA groups from pre-op to 2 years post-op (Table 2). External rotation at 0° improved by 15.55° and internal rotation improved by 1.41 lumbar levels in the PA group at two years post-op (Table 2).

Close

Fig. 1 Improvement in abduction over time.

Export to PPT

Table 2 Changes in range of motion from preoperative baseline ± standard error. Significant values bolded.

Range of Motion	PA 1 year post op (P value)	PA 2-year post op (P value)	PR 1 year post op (P value)	PR 2-year post op (P value)	RR 1 year post op (P value)	RR 2-year post op (P value)
Internal Rotation (Spinal levels)	1.44 ± 0.17 ( < 0.01)	1.41 ± 0.19 ( < 0.01)	−0.72 ± 0.22 ( < 0.01)	−0.32 ± 0.22 (0.14)	0.37 ± 0.36 (0.31)	0.16 ± 0.38 (0.68)
Cross Body Adduction (Levels)	0.99 ± 0.17 ( < 0.01)	1.24 ± 0.18 ( < 0.01)	0.51 ± 0.18 (0.01)	0.59 ± 0.18 ( < 0.01)	0.53 ± 0.24 (0.04)	0.50 ± 0.27 (0.08)
External Rotation Above Head (Levels)	1.42 ± 0.19 ( < 0.01)	1.49 ± 0.22 ( < 0.01)	1.46 ± 0.24 ( < 0.01)	1.38 ± 0.25 ( < 0.01)	1.56 ± 0.61 (0.03)	1.95 ± 0.66 (0.02)
External Rotation at 0°	19.08° ± 2.04° ( < 0.01)	15.55° ± 2.28° ( < 0.01)	4.29° ± 2.60° (0.10)	4.86° ± 2.58° (0.06)	3.31° ± 3.74° (0.38)	6.66° ± 4.36° (0.14)
Forward Elevation	34.01° ± 3.79° ( < 0.01)	31.98° ± 4.18° ( < 0.01)	48.62° ± 4.49° ( < 0.01)	51.05° ± 4.49° ( < 0.01)	41.47° ± 6.02° ( < 0.01)	46.08° ± 7.06° ( < 0.01)
External Rotation at 90°	17.13° ± 2.52° ( < 0.01)	15.32° ± 2.78° ( < 0.01)	18.01° ± 3.53° ( < 0.01)	16.91° ± 3.44° ( < 0.01)	20.21° ± 6.39° (0.01)	15.05° ± 7.20° (0.05)

3.4

3.4 Patient reported outcome measures

For PROMs, there was a significant increase in ASES of 42.73 for PA, 46.39 for PR, and 28.78 for RR groups from pre-op to their most recent follow up averaging 5.27 years (p < 0.01 for all, Table 4). The Simple Shoulder Test also showed improvements in all groups from pre-op to most recent follow-up: 7.74 for PA, 5.22 for PR; 3.10 for RR (p < 0.01 for all, Fig. 2). Pain scores also decreased significantly in the PA and PR groups at two years post op, whereas the pain significantly decreased in the RR group at one-year post-op (p < 0.01, Table 4).

Table 4 Improvements in patient reported outcomes ± standard error. Significant values bolded.

Patient Reported Outcome	PA 1 year post op (P value)	PA 2-year post op (P value)	PA Last FU (P value)	PR 1 year post op (P value)	PR 2-year post op (P value)	PR Last FU (P value)	RR 1 year post op (P value)	RR 2-year post op (P value)	RR Last FU (P value)
ASES	44.13 ± 2.78 ( < 0.01)	44.40 ± 2.61 ( < 0.01)	42.73 ± 2.91 ( < 0.01)	44.88 ± 2.82 ( < 0.01)	45.37 ± 2.78 ( < 0.01)	46.39 ± 2.82 ( < 0.01)	35.88 ± 8.33 ( < 0.01)	37.17 ± 7.78 ( < 0.01)	28.78 ± 8.69 ( < 0.01)
Pain	4.09 ± 0.36 ( < 0.01)	3.94 ± 0.39 ( < 0.01)	--	4.22 ± 0.43 ( < 0.01)	4.02 ± 0.43 ( < 0.01)	–	3.30 ± 0.81 ( < 0.01)	3.44 ± 1.24 (0.02)	–

Close

Fig. 2 Improvement in Simple Shoulder Test over time.

Export to PPT

As far as pairwise comparisons between the groups, ASES and pain scores had no significant differences, but for the SST there were several differences between groups (Table 5). Specifically, the PA group had significantly higher SST scores than the PR group (2.43 at one year, 2.70 at two years, and 2.22 at last follow-up, p < 0.01 for all). The PA also had better outcomes than the RR group (4.96 at one year and 2.22 at last follow-up, p < 0.01 for both). Importantly, RR was worse than PR at last follow-up (3.64, p = 0.01), but not significant at one- or two years post-op (Table 5).

Table 5 Pairwise comparisons of improvement in patient reported outcomes ± standard error. Significant values bolded. Negative values indicate that the second group in the comparison had more improvement than the first group listed.

Patient Reported Outcome	PA vs. PR 1 year (P value)	PA vs. PR 2 years (P value)	PA vs. PR Last FU (P value)	PA vs. RR 1 year (P value)	PA vs. RR 2 years (P value)	PA vs. RR Last FU (P value)	PR vs. RR 1 year (P value)	PR vs. RR 2 years (P value)	PR vs. RR Last FU (P value)
ASES	2.39 ± 3.98 (0.55)	4.43 ± 5.40 (0.42)	0.07 ± 5.72 (0.99)	0.29 ± 8.06 (0.97)	11.4 ± 8.35 (0.18)	18.7 ± 11.7 (0.12)	−2.10 ± 8.13 (0.80)	6.95 ± 8.75 (0.43)	18.6 ± 11.8 (0.12)
Pain	0.32 ± 0.49 (0.51)	0.17 ± 0.57 (0.77)	–	−0.86 ± 0.77 (0.27)	−0.42 ± 2.57 (0.87)	–	−1.18 ± 0.79 (0.14)	−0.58 ± 2.57 (0.82)	–
SST	2.43 ± 0.59 ( < 0.01)	2.70 ± 0.65 ( < 0.01)	2.22 ± 0.68 ( < 0.01)	4.96 ± 0.59 ( < 0.01)	2.38 ± 1.13 (0.04)	2.22 ± 0.68 ( < 0.01)	2.53 ± 1.52 (0.10)	−0.32 ± 1.16 (0.78)	3.64 ± 1.38 (0.01)

4 Discussion

Overall, our data suggest that there was improvement in abduction, forward elevation, ASES, and SST scores for all three groups we analyzed at a minimum of two years post-op. Revision rTSA has shown improvements in shoulder function.^11–17 Chalmers also combined several studies in a review and found improvement in ASES and forward elevation from pre-op to post-op in patients undergoing revision reverse total arthroplasty.¹⁸

As expected, there were functional improvements in the PA and PR groups given that it is established that both the anatomic and reverse approach improve shoulder function.¹⁹ Several studies have compared the outcomes of rTSA and TSA. For example, in a study with 200 patients, rTSA showed significantly better improvements in strength and active forward flexion, whereas the TSA group had larger improvements in external and internal rotation.²⁰ A more recent study showed that TSA leads to a better ROM than rTSA with the two groups having similar functional scores and revision rates.²¹ Our data generally suggests a similar conclusion with the PA group having better ROMs and PROMs compared to baseline than the PR group.

We also compared the revision reverse group to the primary reverse. In this comparison, apart from SST at most recent follow-up and abduction at one year follow-up, there were no statistically significant differences between primary and revision reverse totals shoulder ROM and patient reported outcomes. Other studies and reviews have compared these two groups and found results that suggest the revision reverse surgeries led to significantly worse outcomes than the primary reverse, although they have smaller sample size and fewer clinical data points.^22,23 It should be noted that on average, abduction and forward elevation for PR was 18.1° (p = 0.08) and 19.7° (p = 0.04) better, respectively, than the RR though these numbers did not reach statistical significance. We set our level of significance at <0.01 a priori due to the power of our study and the number of sub analyses to reduce familywise type 1 error rate, though a more liberal p-value cutoff would suggest a clinically and statistically significant difference in range of motion. Nonetheless, there were statistically and clinically significant improvements in abduction and forward elevation from their own preoperative baselines for both primary and revision reverse (Fig. 1, Table 2) demonstrating clinical benefit to revision.

We found similar outcomes in pain scores at both 1 and 2 years for primary and revision reverse; both demonstrated significant improvement in pain from preoperative baseline. Gage et al. found that there is no difference in opioid use after a revision reverse compared to a primary reverse either and suggest that a hesitancy in performing a revision may be unwarranted.²⁴

5 Limitations and strengths of the study

There are several limitations with our study. The patients undergoing revision surgeries had heterogeneity in that there were different surgeons who performed the initial surgery, some had multiple revision surgeries, and there were different indications for the revision surgeries for which we were unable to control. Additionally, some of our revisions were conversions of anatomic or hemiarthroplasty to reverse (18 patients), while others were revisions of a primary reverse (19 patients); this may ultimately lead to differences between groups and could be a point of future sub-analysis. However, Hao et al. reported no difference between primary anatomic converted to reverse compared to revision of a primary reverse.²⁵

6 Conclusion

In conclusion, our data shows a significant improvement in both primary and revision reverse total shoulder arthroplasties. The metrics we used to measure this improvement were range of motion values and patient reported outcome measures. We hypothesized that the revision total shoulder arthroplasty would have worse outcomes than the primary reverse total shoulder arthroplasty. Surprisingly, both groups had similar improvements in shoulder function. Ultimately, patients can expect similar functional outcomes after revision as their primary reverse total shoulder arthroplasty.

Institutional ethical committee approval

Institutional Review Board Approval: The study was approved by the Medical College of Wisconsin/Froedtert Hospital Institutional Review Board #5, with an approval number of PRO00037901.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Patients consent

No guardian/patient consent needed as this is original research.

CRediT authorship contribution statement

Maxwell L. Hershey: Conceptualization, Methodology, Validation, Investigation, Formal analysis, Writing – original draft, Writing – original draft, Visualization, Project administration. Anthony P. Trenga: Conceptualization, Methodology, Validation, Investigation, Supervision, Writing – original draft, Writing – original draft, Visualization, Project administration. Seth A. Roge: Investigation, Writing – original draft. Max R. Fisher: Writing – original draft. Steven I. Grindel: Conceptualization, Methodology, Writing – original draft, Visualization, Project administration.