Introduction

Proximal humeral fractures (PHF) represent between 4 and 10% of all bone fractures and occur more frequently in elderly patients [18, 24, 39]. Their incidence is expected to increase considerably over the coming years owing to aging of the population [16, 34]. Acute three- and four-part PHF can be challenging to treat, particularly if they involve the humeral head or tuberosities [9, 11, 21, 23].

Decision-making in the treatment of PHF is influenced by different factors, including patient age and functional status, bone quality, fracture pattern, tuberosity involvement, as well as surgeon preferences [1, 34]. Traditionally, when fixation is not possible because of poor bone quality and risks of nonunion or osteonecrosis, acute PHF can be managed by shoulder hemiarthroplasty (HSA; Fig. 1; [11, 34, 39]). Although HSA has fewer complications and grants better function than internal fixation systems, it remains associated with inconsistent outcomes, particularly in elderly patients, owing to proximal humeral migration as well as osteolysis, malunion, nonunion, or tuberosity resorption [15, 33, 34, 39].

Fig. 1
figure 1

Shoulder hemiarthroplasty radiographs: a preoperative; b at 6‑week follow-up

Full size image

In recent years, reverse shoulder arthroplasty (RSA; Fig. 2) gained acceptance as an alternative treatment for complex PHF in elderly patients [11, 15, 39], because it relies on the deltoid muscle to restore shoulder function and hence circumvents poor tuberosity healing or concomitant rotator cuff tears [24, 34]. Despite encouraging results, many surgeons hesitate to treat acute PHF by RSA, because it is more invasive and could lead to potentially more complications in the long term [4, 7, 11, 15, 33, 34, 39].

Fig. 2
figure 2

Reverse shoulder arthroplasty radiographs: a preoperative; b at 6‑week follow-up

Full size image

To date there is no clear consensus on the superiority of HSA or RSA for the treatment of acute PHF. We therefore aimed to review the relevant literature for studies of high levels of evidence, reporting clinical outcomes and revision rates for HSA or RSA for the treatment of acute PHF.

Material and methods

Search strategy

An electronic literature search was conducted using the Medline (PubMed) and Embase libraries on 22 September 2018 for articles on HSA and RSA using the following keywords: (shoulder AND arthroplasty) AND (hemi* OR reverse*) AND (proximal* AND fracture* AND humer*) AND (score* OR revision* OR outcome*; Table 1). The electronic literature search returned 741 records that were screened to determine relevance in accordance with the established guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA; [26]):

Inclusion criteria

  • Level I or II studies reporting clinical and/or radiographic outcomes of primary HSA or primary RSA as treatments for acute PHF

Exclusion criteria

  • Level III, IV, or V studies

  • Studies reporting outcomes of revision HSA or revision RSA

  • Studies reporting outcomes for treatments of non-acute PHF, such as sequelae fractures or PHF with diaphyseal extension

  • Articles written in languages other than English, French, German, Italian, or Spanish

  • Articles published before 1998

Table 1 Keyword search terms
Full size table

A total of 148 articles were duplicates, 569 were excluded by reading their titles and/or abstracts, and a further 15 articles were excluded by reading their full text, because they were not level of evidence I or II studies. No additional relevant articles were identified from citations in the full-text articles. This left a total of nine articles from which data were extracted for this review (Fig. 3; [2, 3, 6, 8, 11, 22, 28, 32, 37]).

Fig. 3
figure 3

PRISMA flow diagram

Full size image

Data extraction and quality assessment

Two investigators (LN and AM) independently tabulated and verified the following data from each article: first author, study design, etiology, treatment, cohort size, age at surgery, follow-up, clinical and radiographic outcomes, reported complication, and revision rate. In cases of discrepancies between abstract and full text, data were extracted from the most comprehensive source. Disagreements were discussed until consensus was reached.

Level I studies were evaluated for bias using the Jadad Scale [20] as modified by Gummesson et al. [19], which evaluates randomization method, blinding, and description of dropouts/withdrawals with preset scores for each question from −1 to 1 point. Total scores range from −2 (lowest quality) to 5 (highest). Level II studies were evaluated for quality using the Methodological Index for Non-randomized Studies (MINORS) checklist [35], which assesses articles on 12 methodological items. Scoring is as follows: 0, not reported; 1, reported but poorly done and/or inadequate; and 2, reported in a well-done and adequate manner. Total scores range from 0 (lowest quality) to 24 (highest quality). Each study was scored by two independent investigators (LN and AM). Disagreements were discussed until consensus was reached.

Results

The nine selected articles were published between 2011 and 2016, four of which were level I studies according to their level of evidence statements [2, 6, 28, 32, 37], three of which were level II studies according to their level of evidence statements [3, 8, 11, 22], and the remaining two were deemed to be level II studies, although they did not explicitly state the level of evidence ([2, 6, 28, 32, 37]; Table 2). Only three studies [3, 11, 32] directly compared outcomes of acute PHF treated by HSA versus RSA. The remaining six studies compared outcomes of acute PHF treated by conventional HSA versus HSA + biceps tenodesis [37], trabecular metal HSA [22], intramedullary fibular allograft with locking compression plates [8], open reduction with internal fixation (ORIF; [6]), and nonoperative treatment [2, 28]. It is worth noting that two studies [22, 32] performed biceps tenodesis as a routine procedure on all patients. On a scale of −2 (poorest) to 5 (highest) for quality assessment [19], of the four level I studies, three scored 3 points [2, 28, 32] and one scored 5 points [37]. On a scale of 0 to 24, of the four prospective level II studies, one scored 15 points [3], one 17 points [8], one 21 points [6], and two 23 points [11, 22].

Table 2 Literature review of level I or II studies comparing treatment options for acute proximal humeral fractures
Full size table

The cohort sizes ranged from 12 to 69 patients per group while the mean follow-up ranged from 1 to 4.6 years. The mean patient age was >70 years in six studies [2, 3, 6, 11, 28, 32], between 60 and 70 years in two studies [8, 22], and between 50 and 55 years in one study [37]. In all studies, the fracture etiology was classified according to the Neer classification [27] as acute three-part, four-part, or head-splitting PHF. The most frequently reported clinical outcomes were the absolute or adjusted Constant scores (Fig. 4), the Disability of the Arm, Shoulder and Hand (DASH) Score [17], the University of California–Los Angeles (UCLA) Score [14], and the American Shoulder and Elbow Surgeon (ASES) Score [25]. Forward flexion [2, 3, 8, 11, 22, 32, 37] and external rotation [2, 3, 6, 8, 11, 22, 32] were each reported in seven articles, while internal rotation was reported in four articles [2, 8, 22, 32], and abduction was reported in four articles [2, 3, 8, 32]. Five studies [2, 11, 22, 28, 32] reported tuberosity malunion or resorption. Finally, all studies reported complications and revisions (Fig. 5), while only one study [32] reported on the Kaplan–Meier survival (Table 3).

Fig. 4
figure 4

Reported postoperative absolute and/or adjusted Constant scores. BT biceps tenodesis

Full size image
Fig. 5
figure 5

Reported complication and revision rates. BT biceps tenodesis, TM-HSA trabecular metal shoulder hemiarthroplasty

Full size image
Table 3 Review of studies directly comparing shoulder hemiarthroplasty (HSA) vs. reverse shoulder arthroplasty (RSA)
Full size table

There was some heterogeneity across studies in reporting outcomes, complications, and revisions. Li et al. [22] only reported radiographic complications related to the greater tuberosity and Soliman et al. [37] merely reported pain complications. Clinical outcomes were different (e. g., Constant score vs. ASES scores) as well as measurement references (external rotation alone vs. external rotation as part of the Constant score). Two studies also presented subgroup analyses of their cohorts based on radiographic findings of the greater tuberosity [11, 22]. Boyer and coworkers’ level II study presented considerable heterogeneity across HSA and RSA cohorts in mean follow-up (2.1 vs. 1.3 years) and in mean patient age (68 vs. 78 years; [3]).

HSA vs. RSA

Only three studies [11, 32] directly compared HSA with RSA, one of which [28] compared locked-stem HSA with locked-stem RSA. All three studies reported outcomes in patients with a mean age of >70 years and found that RSA grants better clinical scores, less pain, higher forward elevation, and greater patient satisfaction. Neither Sebastiá-Forcada et al. [32] nor Cuff and Pupello [11] reported significant differences in internal rotation, while Sebastiá-Forcada et al. [32] reported a significantly better external rotation in the RSA group. The two studies without locked stems [11, 32] also agreed that tuberosity malunion or resorption negatively influenced outcomes for the HSA groups but not for the RSA groups. Furthermore, compared with RSA, both complications and revisions were more frequent following HSA without locked stems [11, 24], with only one revision among all RSA patients. Among HSA patients, a total of 11 revisions were described due to resorption of the tuberosities and resultant pseudoparesis of the shoulder [11], to an acromion-head distance of ≤7 mm, reflecting insufficiency or rupture of the rotator cuff [32], or to secondary cuff rupture ([39]; Table 3).

Other findings

Among the selected articles, a single study [22] compared outcomes of different humeral prostheses, one conventional and the other using trabecular metal, in patients treated with HSA (with adjuvant biceps tenodesis). Compared with the conventional HSA group, at a mean follow-up of 4.6 years, the trabecular metal HSA group had fewer radiographic complications related to the greater tuberosity (25.7% vs. 6.1%; p = 0.028), higher ASES (72 vs. 81; p = 0.012) and UCLA (25 vs. 28; p = 0.007) scores, greater active forward elevation (131° vs. 116°; p = 0.044) and external rotation (38° vs. 30°; p = 0.015), but no difference in internal rotation (p = 0.671) or pain on the visual analog scale (p = 0.779).

Only one study compared outcomes of HSA with and without adjuvant treatment [37]. Reporting clinical outcomes at 2 years for patients aged 50–55, Soliman et al. [37] found that, compared with HSA alone, HSA with biceps tenodesis granted significantly better age- and gender-adjusted Constant scores (74.4 vs. 69.8; p = 0.04), reduced the occurrence of pain (15.8% vs. 33.3%; p = 0.03), and tended to increase the range of forward elevation (152° ± 18° vs. 1° ± 16°; p >0.05).

Discussion

The most important finding of the present literature review is that, from the nine relevant level I and II studies that reported clinical outcomes and revision rates of primary HSA or RSA in treating acute PHF, only three studies [3, 11, 32] directly compared HSA with RSA. All three stated that RSA had better clinical scores and fewer revisions than HSA in elderly patients. Given that only one study described outcomes for patients aged <60 [37], the best surgical treatment in young and more demanding patients remains under debate [5, 10, 30].

Our analysis confirms the current consensus on the superiority of RSA over HSA for the surgical treatment of acute PHF in elderly patients [9, 15, 16, 24, 32, 34, 39]. In their recent systematic review and meta-analysis including level I–IV studies, Gallinet et al. [16] , also concluded on the superiority of RSA, although they emphasized the lack of long-term studies. In a surgeon survey, Savin et al. [31] report that most practitioners agree that RSA is at least as reliable as, and often more adapted than, HSA for managing four-part PHF in elderly patients. One clear advantage of RSA over HSA is that, by granting full function to the deltoid muscle, RSA reduces the effect of poor or slow tuberosity healing on outcomes [16]. Our analysis confirmed that tuberosity resorption is one of the most frequent causes for revision in HSA, but also suggested that in the presence of tuberosity resorption or malunion, the use of trabecular metal stems could significantly improve HSA outcomes [22]. Biceps tenodesis is another factor that may contribute to better HSA outcomes. Among the selected studies, one [37] compared outcomes of HSA alone with HSA with adjuvant biceps tenodesis. Adjuvant tenodesis improved functional outcomes and reduced pain; strikingly two cases of enduring pain after HSA alone were improved by postsurgical tenodesis.

The selected studies comparing HSA with surgical treatments other than RSA in elderly patients reported that intramedullary fibular allograft with locking compression plate versus HSA granted better clinical outcomes, greater range of motion, lower complication rates, and higher patient satisfaction [8], while ORIF and HSA did not differ significantly in functional outcomes. Therefore, ORIF could be considered an acceptable solution for less-demanding patients without humeral avascular necrosis and/or for less experienced surgeons [6]. Given the distinct risks (loss of reduction, screw cutout, intra-articular screw penetration, and avascular necrosis) and benefits (bone stock preservation, anatomic healing) of ORIF [29, 36, 38], this technique is usually preferred for younger patients, which renders comparison difficult [13]. However, a recent study of 1791 surgically treated patients with PHF, from the American Surgeons National Surgical Quality Improvement Program database [12], reported that, compared with HSA and with RSA, ORIF is an independent protective factor against complications, although patient comorbidities remain stronger predictors of short-term complications than surgical treatment.

While there is general consensus on the surgical treatment of displaced four-part fractures in younger patients, some controversy pertains to elderly patients, who are more likely to be osteoporotic, have low functional demands, and slow recovery. For them, less invasive procedures may provide adequate reduction and stable primary fixation [23]. The two level I studies that compared nonoperative treatment with HSA in elderly patients [2, 28] observed no functional difference between patients treated conservatively or surgically at 1 or 2 years; however, conservative treatment was associated with more pain at 3 months [2] and lower quality of life [28].

Limitations

The present study has three main limitations: first, despite clear objectives and quantitative comparisons, the scarcity of level I and II studies limits the scope of our findings. Second, the absence of long-term level I and II studies does not permit firm conclusions. Third, we did not replicate our search in the Cochrane Library, nor did we consider gray literature. Nevertheless, to our knowledge, the present review is the first to summarize level I and II studies comparing outcomes of primary HSA or RSA for the treatment of acute PHF.

Practical conclusion

  • Reverse shoulder arthroplasty is a reliable treatment for acute PHF in elderly patients.

  • It grants better outcomes and fewer complications than HSA at short-term follow-up.

  • Future research is needed to establish the best surgical option for acute PHF in young patients.