Functional outcomes of reverse total shoulder arthroplasty in Boileau class 1 and class 2 proximal humeral fracture sequelae: A meta-analysis and systematic review

Abstract

Background

Reverse total shoulder arthroplasty (RTSA) is an increasingly popular salvage treatment option for proximal humeral fracture (PHF) sequelae. This meta-analysis aimed to conduct a pooled analysis of functional outcomes of RTSA in PHF sequelae, with subgroup analysis comparing between intracapsular (Class 1) and extracapsular (Class 2) PHF sequelae.

Methods

A multi-database search (PubMed, OVID, EMBASE) was performed according to PRISMA guidelines on 27th July 2020. Data from all published literature meeting inclusion criteria were extracted and analysed.

Findings

Eleven studies were included, comprising 359 shoulders (167 Class 1 and 192 Class 2). The mean age was 68.2 years, and the mean time between injury and surgery was 49 months, (1–516 months). Constant score and forward flexion improved by 31.8 (95%CI: 30.5–33.1, p < 0.001) and 60o (95%CI: 58o-62o, p < 0.001) respectively between pre-operative and post-operative values for both groups. Constant scores were better in Class 1 patients (MD = 3.60, 95%CI: 1.0–6.2, p < 0.001) pre-operatively and post-operatively (MD = 7.4, 95%CI: 5.8–9.0, p < 0.001). Forward flexion was significantly better in Class 1 patients (MD = 13o, 95%CI: 7o-17o, p < 0.001) pre-operatively, but was slightly better in Class 2 patients post-operatively (MD = 7o, 95%CI: 4o-10o, p < 0.001). Overall complication rate was 16.8%.

Conclusion

Salvage RTSA is effective for PHF sequelae, with multiple factors contributing to the high complication rate.

Introduction

Proximal humeral fractures (PHF) are common injuries in the elderly. PHF affect at least twice as many women compared to men.^1,2 The management of such fractures remain a challenge due to a variety of fracture patterns, coexisting rotator cuff pathology and risks of fracture non-union, mal-union and osteonecrosis.^3,4 Whilst the majority of PHF are treated non-operatively, there is much variation in surgical treatment methods. ⁵

Surgical treatment options include closed reduction, open reduction and internal fixation (ORIF) with either plate or intramedullary nail, hemi-arthroplasty, or reverse total shoulder arthroplasty (RTSA). In particular, RTSA is increasingly utilized as a management option for both acute fracture and fracture sequelae, especially in patients with concurrent rotator cuff insufficiency or displaced tuberosities.^6–10

PHF sequelae can occur due to failure of either conservative or surgical management. If symptomatic, this can leave patients with debilitating outcomes such as persistent pain and reduced joint motion and strength.^11–13 In 2006, Boileau et al. ¹⁴ described a classification of PHF sequelae, distinguishing between intracapsular (Class 1) and extracapsular (Class 2) fracture sequelae, as illustrated in Figure 1. Class 1 refers to intracapsular, impacted fractures with cephalic collapse or necrosis (type 1) or locked dislocation or fracture-dislocations (type 2), while Class 2 refers to extra-capsular, dis-impacted fractures comprising surgical neck non-union (type 3) or severe tuberosity malunions (type 4).

Figure 1.

Boileau proximal humerus fracture sequelae classification [adapted from: Boileau et al. Proximal humerus fracture sequelae: Impact of a new radiographic classification on arthroplasty. Clin Orthop Relat Res. 2006;442:121-30].

RTSA has been recommended for use in acute complex PHF as well as in malunion. ¹² However there remains a paucity of literature on the outcomes of RTSA used to treat various classes of PHF sequelae. The aim of this meta-analysis is to conduct a pooled analysis of functional outcomes of RTSA in PHF sequelae, with a subgroup analysis comparing between Class 1 and Class 2 PHF sequelae.

Materials and methods

Literature search

This meta-analysis was performed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) criteria. ¹⁵ A comprehensive multi-database search (PubMed, OVID Medline, EMBASE) was conducted from date of database inception to 27^th July 2020. The Medical Subject Headings and Boolean operators utilized for this search were: [(‘reverse total shoulder replacement’ OR ‘reverse total shoulder arthroplasty’) AND (‘humerus malunion’ OR ‘humerus fracture’ OR ‘Boileau’ OR ‘fracture sequelae’)]. Identified articles and their corresponding references were reviewed and considered for inclusion according to the selection criteria.

Selection criteria

All articles of any study design reporting on the outcomes of RTSA patients performed for PHF sequelae initially managed non-operatively or with ORIF were considered for inclusion. Only studies that specifically stated use of Boileau classification were included in our study. Studies were not excluded based on the outcome measures they used. Non-English language studies, duplicate studies, non-peer reviewed studies, unpublished manuscripts, conference abstracts and studies not directly studying the outcomes of RTSA in PHF sequelae were excluded. Two independent authors reviewed records from the initial search twice and excluded irrelevant ones. The titles and abstracts of remaining articles were then screened against the inclusion criteria. Subsequently, full-text articles were assessed for inclusion. Included articles were critically reviewed according to a predefined data extraction form. In this study, we considered RTSA to be a salvage procedure if done greater than 1 month from initial injury and/or if mentioned specifically for PHF sequelae or malunion.

Data extraction

Extracted data parameters included details on study designs, publication year, patient numbers, basic demographics, initial treatment received, time between initial management and RTSA, classification of PHF sequelae, outcomes and complications. Various outcome measures were utilized by these studies, however only Constant scores and shoulder flexion-extension arc of motion had adequate data for analysis. Complications included dislocations, infections, peri-prosthetic fractures, hematomas, brachial plexopathies and glenoid component failure.^16–24

Methodology assessment

Methodology quality of included studies was assessed with the Methodological Index for Non-Randomized Studies (MINORS). ²⁵ MINORS uses nine criteria for non-randomized single-arm cohort studies. Each criterion was scored with a 3-point system from 0 to 2 (0: not reported, 1: inadequately reported and 2: adequately reported). The ideal score for non-comparative studies is 18 points. Two independent reviewers performed the assessment separately and any discrepancy was resolved with discussions involving a third reviewer.

Statistical analysis

Non-comparative meta-analysis was primarily performed based on weighted pooled proportions (PP) and pooled weighted mean, while comparative meta-analysis was primarily performed with odds ratio (OR) and weighted mean difference (MD) as summary statistics. In this meta-analysis, both fixed- and random-effects models were tested. Fixed-effects model assumed that treatment effects in each study were identical, while random-effects model assumed that variations were present between studies. X² tests were used to study heterogeneity between studies. I² statistic was used to estimate the percentage of total variation across studies, owing to heterogeneity rather than chance. Values greater than 50% was regarded as substantial heterogeneity. I² can be calculated as: I² = 100% × (Q - df)/Q. Q was defined as Cochrane’s heterogeneity statistics and df defined as degree of freedom. If substantial heterogeneity was present, the possible clinical and methodological reasons were explored qualitatively. This meta-analysis presented results with a random-effects model to account for clinical diversity and methodological variation between studies. Subgroup analyses comparing between Boileau Class 1 and 2 PHF sequelae were compared using Z-test statistics. All p values were two-sided. OpenMetaAnalyst (Brown University, USA, 2012) and Review Manager (version 5.3, Copenhagen, The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) were used for statistical analysis.

Results

Literature search

A selection process flowchart to include relevant studies is illustrated in Figure 2. A total of 881 studies were identified from the initial search, of which 346 duplicates and 47 non-English language articles were removed. Titles and abstracts of 488 remaining studies were screened according to the pre-defined inclusion criteria. 435 studies were excluded. Thirty-five full-text articles were assessed for eligibility and no additional articles included after citation search. In total, 11 studies^{16–24,26,27} were included for qualitative analysis, of which 8^{16,19–24,27} were included for quantitative analysis.

Figure 2.

PRISMA selection process flowchart.

Methodology assessment

Aggregate MINORS score for included comparative studies ranged from 10 to 12, with an average of 11.1. Individual scores are detailed in Appendix 1.

Demographics

A total of 359 shoulders (167 Class 1 and 192 Class 2) were included in the study. The mean age of all patients was 68.2 years, ranging from 27 to 91 years,^{16–24,26,27} while mean time from injury to salvage RTSA was 49 months, ranging from 1 to 516 months.^{16,19–21,23,24,27} Further details of individual studies are shown in Table 1.

Table 1.

Demographic details and outcome measures of included studies.

Author	Year	Study Design	No. of shoulders	Age (year)	Male: Female Ratio	Mean time from injury (month)	Constant score			Forward flexion			External Rotation
							Pre-op	post-op	P value	Preop	Post op	P value	Preop	Post op	P Value
Boileau Classification 1
Audige (Type 1)	2019	Retrospective	40	71 ± 7	11:29	-	30.0 ± 11.0	-	-	79 ± 31	-	-	15 ± 17	-	-
Grubhofer	2017	Retrospective	27	68 ± 14	32:12	-	-	61.0 ± 9.3	-	-	-	-	-	-	-
Hattrup	2016	Case series	3	-		-	-	-	-	77 ± 19	140 ± 36	<0.001	−13	57 ± 11	0.0002
Holschen	2019	Retrospective	11	-	-	8	-	64.7 ± 12.5	-	-	122 ± 28	-	-	-	-
Raiss (Type 1)	2018	Retrospective	38	73 ± 9	-	108 ± 126	25.0 ± 12.8	57.0 ± 15.4	<0.0001	73 ± 31	117 ± 30	<0.0001	1 ± 24	11 ± 17	0.0065
Raiss (Type 2)	2017	Retrospective	22	71 ± 10	-	-	13.6 ± 10.7	47.4 ± 22.5	<0.001	38 ± 32	103 ± 5	<0.001	−1 ± 13	15 ± 15	<0.03
Schliemann	2017	Retrospective	26	75 ± 6	8:18	-	44.0 ± 12.0	73.0 ± 13.3	<0.001	-	-	-	-	-	-
Boileau Classification 2
Audige (Type 3)	2019	Retrospective registry	14	67 ± 11	5:9	-	27 ± 12	-	-	57 ± 28	-	-	8 ± 12	-	-
Audige (Type 4)	2019	Retrospective registry	23	66 ± 11	9:14	-	23 ± 11	-	-	55 ± 31	-	-	7 ± 12	-	-
Dezfuli	2016	Case Control	13	73	4:9	-	-	64	-	-	109	-	-	22	-
Grubhofer	2017	Case series	17	68 ± 14	32:12	20 ± 23	-	54 ± 13	-	-	-	-	-	-	-
Hattrup	2016	Case series	23	67 ± 15	6:20	-	-	-	-	37 ± 34 (n = 23)	132 ± 37 (n = 20)	<0.001	1 ± 20 (n = 21)	34 ± 37 (n = 20)	0.0002
Holschen (Type 3)	2019	Retrospective	3	-	-	8 ± 4	-	61 ± 13.0	-	-	125 ± 48	-	-	-	-
Holschen (Type 4)	2019	Retrospective	7	-	-	8 ± 4	-	59.1 ± 9.7	-	-	120 ± 22	-	-	-	-
Martinez	2012	Retrospective	18	79 ± 2	6:12	11 ± 1	26	55	<0.0001	35	90	<0.0001	15	30	<0.0001
Raiss (Type 3)	2014	Retrospective	32	68 ± 9	4:28	26 ± 30	14.2 ± 8.3	46.6 ± 17.3	<0.001	43 ± 40	110 ± 43	<0.001	1 ± 25	13.1 (-30-60)	<0.005
Raiss (Type 4)	2016	Retrospective	42	68 ± 14	12:30	61 ± 66	19.7 ± 10.8	54.9 ± 15.8	<0.0001	54 ± 25	121 ± 33	<0.0001	-	-	<0.0001

Functional outcome

Pooled weighted analysis revealed an overall mean improvement in Constant score of 31.8 (95%CO: 30.5–33.1, p < 0.001) between final follow-up and preoperative scores. Subgroup analysis revealed that patients with Boileau Class 1 PHF sequelae had a mean Constant score improvement of 33.0 (95% CI: 30.5–35.5, p < 0.001), while those with Class 2 PHF sequelae had improvement of 29.2 (95% CI: 27.4–31.0, p < 0.001). (Figure 3) When directly comparing between both classes of PHF sequelae, pre-operative Constant scores were significantly better in Class 1 PHF sequelae patients (MD = 3.60, 95%CI: 1.0–6.2, p < 0.001) and remained so post-operatively (MD = 7.4, 95%CI: 5.8–9.0, p < 0.001). (Table 2)

Figure 3.

Pre-operative versus post-operative constant scores. Note: Asterix (*) denotes a significant change between pre-operative and post-operative scores.

Table 2.

Pooled weighted analysis of constant score and forward flexion.

Constant Score	Pooled pre op score	n	Pooled post op score	n	Mean Difference (post op versus pre op)
Class 1	28.2 (12.4)	126	61.2 (7.2)	124	33.0, 95%CI: 30.5-35.5, P < 0.001
Class 2	24.6 (8.0)	111	53.8 (4.8)	101	29.2, 95%CI: 27.4-31.0, P < 0.001
Class 1 and 2	26.1 (8.1)	237	57.9 (5.5)	225	31.8, 95%CI: 30.5-33.1, p < 0.001
Class 1 vs 2	3.60 (0.97-6.23), p < 0.001	237	7.4 (5.8-9.0), p < 0.001	225	3.8, 95%CI: 3.3- 4.3, p < 0.001^
Forward Flexion	Pooled pre op degrees	n	Pooled post op degrees	n	Mean Difference (post op versus pre op)
Class 1	64 (22)	100	112 (13)	71	49, 95%CI: 43-54, p < 0.001
Class 2	51 (8)	134	119 (8)	107	68, 95%CI: 66-70, p < 0.001
Class 1 and 2	57 (12)	234	117 (9)	178	60, 95%CI: 58-61, p < 0.001
Class 1 vs 2	13 (7-17), p < 0.001	234	-7 (-3.5, -10.0), p < 0.001	178	-20, 95%CI: -19- -21, p < 0.001*

Results presented as mean difference (Standard deviation). Forward flexion presented in degrees (°), rounded to nearest number. ^Favours Class 1, *Favours Class 2.

Pooled weighted analysis revealed an overall mean forward flexion arc of motion improvement of 60° (95%CI: 58°-62°, p < 0.001) between final follow-up and preoperative measurements. Subgroup analysis further revealed that patients receiving RTSA for Class 1 PHF sequelae had a mean final forward flexion improvement of 49° (95% CI: 43°–54°, p < 0.001), while those with Class 2 PHF sequelae was 68° (95% CI: 66°-70°, p < 0.001). (Figure 4).When directly comparing between both classes of PHF sequelae, pre-operative forward flexion arc of motion were better in Class 1 PHF sequelae patients (MD = 12°, 95%CI: 7°-17°, p < 0.001), but was better in Class 2 PHF sequelae patients post-operatively (MD = 7°, 95%CI: 4°-10°, p < 0.001). (Table 2)

Figure 4.

Pre-operative versus post-operative forward flexion. Note: Asterix (*) denotes a significant change between pre-operative and post-operative scores.

When comparing against pre-operative values for Class 1 PHF sequelae, two studies^21,22 reported statistically significant improvement in internal and external rotation at final follow-up. In Class 2 PHF sequelae, four studies^18–21 reported improvement in external rotation from pre-operative to final follow-up values (Table 1).

Complications

Overall, there was a complication rate of 16.8%.^16–24,27 Complication rates varied amongst studies. Subgroup analysis of complication to compare between Class 1 and 2 PHF sequelae was not possible due to lack of raw data. Notably, Holschen et al. ²⁷ reported no complications during a mean of 45 months follow-up amongst 21 patients, whilst Raiss et al. ²³ reported complication rate of 41% over a mean of 48 months follow-up among 32 patients. The most common complication in salvage RTSA was prosthetic dislocations and infections. Prevalence of complications across studies is displayed in Table 3.

Table 3.

Complications of included studies.

Author	Year	Number of shoulders	Rate of complications (%)	Rate of Revisions (%)	Complications
Boileau Classification 1
Audige (Type 1)	2019	40	n/a	n/a	n/a
Grubhofer	2017	27	n/a	9	Intraoperative periprosthetic fracture, periprosthetic infection, axilla folliculitis, haematoma
Hattrup	2016	3	19	4	infection, brachial plexopathies, dislocation
Holschen (Type 1)	2019	11	0	0	n/a
Raiss	2018	38	8	5	Fracture, haematoma, infection
Raiss	2017	22	32	27	Glenoid failure due to bony defects
Schliemann	2017	26	19	19	Humeral stem loosening, periprosthetic fracture, periarticular ossifications
Boileau Classification 2
Audige (Type 3)	2019	14	n/a	n/a
Audige (Type 4)	2019	23	n/a	n/a
Dezfuli	2016	13			Infection
Grubhofer	2017	17	n/a	9	Intraoperative periprosthetic fracture, periprosthetic infection, axilla folliculitis, haematoma
Hattrup	2016	23	19	4	Infection, brachial plexopathies, dislocation
Holschen (Type 3)	2019	3	0	0	n/a
Holschen (Type 4)	2019	7	0	0	n/a
Martinez	2012	18	27	11	Dislocation, infection, transient axillary nerve palsy
Raiss	2014	32	41	28	Dislocation, infection
Raiss	2016	42	10	2	Dislocation, infection, periprosthetic fracture, loosening of components

Discussion

This meta-analysis reports on the largest pooled analysis of the current available literature to illustrate the efficacy of RTSA as a salvage procedure for PHF sequelae. The results revealed clinically and statistically significant improvements in functional outcome and range of motion. Subgroup analysis comparing outcomes of RTSA between Class 1 and 2 Boileau classifications of PHF sequelae was also performed. To our knowledge, there is no previous meta-analysis on this topic. Our results showed significant improvements in Constant scores and forward flexion arc of motion for both groups, with overall mean improvement of 31.8 and 60° respectively, with a likely greater benefit being demonstrated in forward flexion arc of motion in Boileau Class 2 PHF sequelae patients. However, higher quality studies are needed to verify this.

Though minimum clinically important difference (MCID) for PHF is not yet established, the substantial improvements in Constant scores were statistically significant in both groups and were much larger than the MCID of other commonly debilitating conditions such as humeral shaft fracture and rotator cuff tears (10.4 and 6.1 respectively).^28,29 Similarly, the gains in forward flexion arc of motion were substantial in both groups, representing both statistical and clinical significance. Notably, gains in Class 2 PHF sequelae patients were greater than those of Class 1 patients. One helpful measure of effectiveness is that both groups achieved final forward flexion arc of motion deemed adequate for activities of daily living, suggested to be approximately 110°−120°.^30,31

Overall, the results of RTSA performed for PHF sequelae were similar to RTSA performed for other indications. Kim et al. ³² reported overall post-operative Constant score and forward flexion to be 57.8 and 121.8° for multiple indications. Comparatively, Constant scores for Class 1 (61.2) were slightly better and Class 2 (53.8) shoulders were slightly worse, whilst Forward flexion in Class 1 (112°) and Class 2 (117°) were largely similar.

The Constant score, comprised of subjective and objective components, is a widely used and validated functional outcome assessment score for patients with proximal humerus fracture. ²⁸ Subjective parameters include pain severity, ability to perform activities of daily living, while objective parameters include shoulder range of motion and strength. Despite significant improvements of both Constant scores and forward flexion arc of motion, no sub-score analysis of Constant score was performed. Hence, it is unclear if significant improvement in forward flexion arc of motion was the main driver of Constant score improvement. Further sub-score analysis in future could delineate the drivers of functional improvements in patients receiving salvage RTSA for PHF sequelae.

Initial non-operative management of acute PHF is less costly than operative management.^33,34 However, there is a paucity of literature on cost-analysis for PHF sequelae and non-union requiring further surgical management. Further cost analysis should be undertaken to weigh the benefits and costs of adopting surgical management from the outset or a trial of non-operative management and then performing salvage RTSA for those who continue to suffer from debilitating symptoms and limitations.

Variable outcomes have been reported for RTSA performed as acute or salvage procedures. Torchia et al. ³⁵ found similar clinical outcomes and complications in both settings whilst Nelson et al. ³⁶ suggests better outcomes and lower complications for RTSA performed acutely. The latter study also found that salvage RTSAs in patients previously managed non-operatively may outperform those who had failed operative treatment. This information acknowledges the existing challenges on management of PHF, as it may be reasonable to opt for acute RTSA to avoid a worse clinical outcome after a failed internal fixation, yet it is also justifiable to opt for a trial of initial conservative management despite relatively high non-union rates of up to 33% with non-operative management.^4,37

The overall complication rate of RTSA in PHF sequelae is higher than the reported average of 8.9% in an analysis of 4158 patients for multiple indications. ³⁸ This is expected since delayed surgery for PHF nonunion or malunion is inherently more technically challenging than acute surgery. These challenges include the surgical approach, removal of scar tissue, fracture callus, and defining normal planes. Tuberosity osteotomies often have to be performed and these may not be anatomical in the delayed setting too. Furthermore, patients initially trialled on non-operative management were more likely to have been deemed surgically unsuitable due to severe medical comorbidities or having risk factors for poor recovery from surgical management. Hence, non-operatively managed patients who subsequently develop suffer from PHF sequelae were likely to have a higher complication risk.

There are some limitations to this study. Firstly, there was no high quality randomized controlled trials amongst included studies and the inclusion of retrospective studies meant that recall and selection bias cannot be excluded. Due to a lack of raw data, important parameters such as revisions, readmissions, various range of shoulder motions and other validated functional outcomes such as the American Shoulder and Elbow score could not be quantitatively analysed. Furthermore, potential confounders such as age, medical comorbidities, time from injury to salvage procedure and severity of injuries could not be adjusted for. Additionally, we did not have data to suggest if any of the shoulders were previously treated successfully or when they deteriorated. Inter- and intra-rater reliability in measuring shoulder range of motion or the subjectivity involved in recording functional outcomes may also be an issue and could cause discrepancies. Lastly, heterogeneity of study protocols, patient demographics and follow-up periods could have also limited the accuracy and reliability of evidence presented.

Conclusion

Salvage RTSA is an effective treatment for both classes of PHF sequelae. There are multiple factors contributing to the high complication rate. Further high-quality comparative studies and cost analyses are required to further characterize and delineate the full benefits of salvage RTSA.