Anatomic Total Shoulder Arthroplasty versus Hemiarthroplasty for Glenohumeral Osteoarthritis: A Systematic Review and Meta-Analysis

: Purpose: Both anatomic total shoulder arthroplasty (TSA) and shoulder hemiarthroplasty (SHA) are used for the management of end-stage glenohumeral osteoarthritis (GHOA). The present study compared TSA and SHA in terms of clinical outcome and complication rate. Methods: This meta-analysis followed the PRISMA guidelines. In October 2021, the following databases were accessed: Web of Science, Google Scholar, Pubmed, Scopus. All clinical trials comparing anatomical TSA versus SHA for GHOA were considered. Results: Data from 11,027 procedures were retrieved. The mean length of the follow-up was 81.8 (16 to 223.20) months. The mean age of the patients was 61.4 ± 8.6 years, and 56.0% (5731 of 10,228 patients) were women. At last follow-up, the age-adjusted constant score was greater following TSA ( p < 0.0001), as were active elevation ( p < 0.0001), ﬂexion ( p < 0.0001), abduction ( p < 0.0001), and American Shoulder and Elbow Surgeons Shoulder Score ( p < 0.0001). Postoperative pain ( p < 0.0001) and revision rate ( p = 0.02) were lower in the TSA group. Conclusions: Anatomic TSA performed better than SHA in patients with GHOA.


Introduction
Glenohumeral osteoarthritis (GHOA) is a degenerative disease characterized by the progressive consumption of the articular cartilage of the shoulder [1,2], leading to gradual loss of function and persistent pain [3]. The incidence of GHOA increases exponentially after the age of 50, with greater prevalence in women [1,2]. In the United States, the Nationwide Inpatient Sample database recorded an increasing incidence of shoulder arthroplasty from 2002 to 2011 of 267% [4]. In the elderly population, the demand is expected to increase by up to 755% by 2050 [4]. In the Korean population over 65 years of age, a radiological prevalence of GHOA of 16.1% was found [5]. Especially in young and active adults, nonoperative management should be attempted before considering surgery [6]. When conservative management fails, shoulder replacement may be required [7]. Both anatomic total shoulder arthroplasty (TSA) and shoulder hemiarthroplasty (SHA) are routinely used in patients with GHOA [8][9][10][11]. However, it is still controversial which type of implant provides better outcomes [12][13][14][15][16][17][18][19][20][21][22]. Previous systematic reviews and meta-analyses reported better outcomes for TSA [23][24][25][26]. However, several recent clinical trials, which were not considered in these meta-analyses, have been published and an update of current evidence is necessary [27][28][29][30][31][32]. Therefore, the present meta-analysis aimed to compare TSA and SHA through the examination of patient reported outcome measures (PROMs), range of motion (ROM), and the rate of revision between the two implants. A multivariate analysis was also conducted to investigate whether the patient characteristics at baseline influenced the surgical outcome.

Search Strategy
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA guidelines [33]. The PICOT algorithm was first considered:

Data Source and Extraction
Two authors independently (G.V. and A.P.) performed the literature search in October 2021. PubMed and Google scholar were accessed. Afterwards, Web of Science and Scopus were also accessed to identify further articles. The following keywords were used in combination: shoulder, arthroplasty, hemiarthroplasty, total, glenohumeral, osteoarthritis, prosthesis, implant, clinical, functional, outcomes, humeral, head, replacement, surgery, constant, score, index, complication, revision, pain (Table 1). The same two authors separately screened the resulting titles from the above searches. If the title and abstract matched the topic, the full text of the article was accessed. A cross-reference of the bibliographies of the full text was also performed. Disagreements were resolved by a third author (N.M.).

Eligibility Criteria
All the published studies comparing the outcomes of anatomic TSA and/or SHA for GHOA were accessed. Given the authors' language capabilities, articles in English, German, Italian, French, and Spanish were eligible. Levels I to IV of evidence, according to Oxford Centre of Evidence-Based Medicine [34], were considered. Editorials, reviews technical notes, expert opinion, comments, and letters were excluded. Only anatomic TSA implants were considered. Only studies that involved patients with advanced GHOA associated with severe pain and functional impairment were included. Only studies that reported a minimum of 12 months follow-up were eligible. Only articles reporting quantitative data under the outcomes of interest were considered for inclusion.

Outcomes of Interest
Data extraction was performed by two authors (G.V. and A.P.). The following data were collected: author, year, journal, type of study, number of prostheses, mean age, mean length of the follow-up, type of implant. The following data were collected at baseline and at last follow-up for each implant (TSA and SHA): Constant score [35], flexion, abduction, ASES questionnaire [36], VAS, ROM. Rates of revision were also retrieved. The primary outcome of interest was to compare the outcomes between TSA and HSA. The second outcome of interest was to perform a multivariate analysis to investigate whether the patient characteristics at baseline had an influence on the clinical outcome.

Methodology Quality Assessment
Two authors (G.V. and A.P.) independently performed the methodological quality assessment using the Coleman Methodology Score (CMS) [37]. The CMS is a validated tool to assess the quality of the methodology in systematic reviews and meta-analyses. The CMS is based on several endpoints: study size, length of the follow-up, surgical approach, type of study, description of diagnosis, surgical technique, and rehabilitation. Additionally, outcome criteria assessment, procedures for assessing outcomes, and the recruitment process were also evaluated. The CMS rates articles with values between 0 (poor) and 100 (excellent). Articles with values >60 are considered satisfactory.

Statistical Analysis
The statistical analyses were performed by the main author (F.M.). The meta-analyses were performed using Editorial Manager Software version 5.3 (The Nordic Cochrane Collaboration, Copenhagen, Denmark). Dichotomic data were analyzed through the Mantel-Haenszel method and the odds ratio (OR) effect measure. Continuous data were analyzed using the inverse variance method, with the mean difference (MD) effect measure. The confidence interval was set at 95% for all comparisons. Heterogeneity was evaluated through Higgins-I 2 and χ 2 tests. If χ 2 < 0.05, statistically significant heterogeneity was detected. Values of Higgins-I 2 were interpreted as low (<30%), moderate (30% to 60%), or high (>60%). A fixed-effects model was set as default. If moderate or high heterogeneity was detected, a random-effects model was adopted. For the multivariate analyses, STATA software (StataCorp, College Station, TX, USA) was used. Multiple pairwise correlations according to the Pearson Product-Moment Correlation Coefficient (r) were conducted. According to the Cauchy-Schwarz inequality, the final effect was ranked between +1 (positive linear correlation) and −1 (negative linear correlation). Values of 0.1 < |r| < 0.3, 0.3 < |r| < 0.5, and |r| > 0.5 were considered to have poor, moderate, and strong correlation, respectively. Overall significance was evaluated using the χ 2 test. Values of p < 0.05 were considered statistically significant.

Search Results
The initial search resulted in 1346 articles. Of them, 423 were excluded as they were duplicates. Another 874 studies were excluded as they did not match the eligibility criteria:

Search Results
The initial search resulted in 1346 articles. Of them, 423 were excluded as they were duplicates. Another 874 studies were excluded as they did not match the eligibility criteria: study design (N = 349), not matching the topic (N = 385), language limitations (N = 4), use of experimental rehabilitation (N = 5), short duration of the follow-up (N = 31), revision setting (N = 56), unclear source of data or criteria (N = 44). An additional 15 studies were excluded because of the lack of reporting of quantitative data related to the outcomes of interests. This left 34 studies for analysis. A schematic of the number of literature search results is shown in Figure 1.

Risk of Publication Bias
The funnel plot of the most commonly reported outcome (revision) was performed to assess the risk of publication bias ( Figure 2). The referral points demonstrated symmetrical distribution, mostly within the pyramidal shapes, indicating a low risk of publication bias.

Risk of Publication Bias
The funnel plot of the most commonly reported outcome (revision) was performed to assess the risk of publication bias ( Figure 2). The referral points demonstrated symmetrical distribution, mostly within the pyramidal shapes, indicating a low risk of publication bias.

Methodological Quality Assessment
The study size and the length of the follow-up were acceptable in most studies. Surgical approach and rehabilitation were well described, and the outcome measures and timing of assessment were often defined, providing moderate reliability. The diagnosis was poorly described. The procedure for assessing outcomes and subject selection was often biased and not satisfactorily described. Overall, the CMS was 70.4 points, attesting an acceptable methodological assessment (

Methodological Quality Assessment
The study size and the length of the follow-up were acceptable in most studies. Surgical approach and rehabilitation were well described, and the outcome measures and timing of assessment were often defined, providing moderate reliability. The diagnosis was poorly described. The procedure for assessing outcomes and subject selection was often biased and not satisfactorily described. Overall, the CMS was 70.4 points, attesting an acceptable methodological assessment ( Table 2).

Patient Demographics
Data from 11,027 procedures were retrieved. The mean follow-up was 85.4 ± 49.1 months. The mean age of the patients was 61.4 ± 8.6 years of age, and 56.0% (5731 of 10,228) were female. Baseline comparability between the two groups was detected for length of follow-up (p = 0.3), active elevation (p = 0.7), and abduction (p = 0.1). Study generalities and patient baseline values are shown in Table 3.

Patient Demographics
Data from 11,027 procedures were retrieved. The mean follow-up was 85.4 ± 49.1 months. The mean age of the patients was 61.4 ± 8.6 years of age, and 56.0% (5731 of 10,228) were female. Baseline comparability between the two groups was detected for length of follow-up (p = 0.3), active elevation (p = 0.7), and abduction (p = 0.1). Study generalities and patient baseline values are shown in Table 3.

Meta-Analysis
The Constant score and the rate of revision were included in the meta-analysis. The Constant score, evaluated in four studies [21,27,29,50], was higher in the TSA group (MD 7.57; 95% CI 0.57, 14.58; p = 0.03; Figure 3).

Discussion
Based on the main findings of this study, anatomic TSA performed better than SHA for GHOA. At last follow-up, the TSA group demonstrated greater flexion, abduction, and elevation. ASES and VAS were both greater in the TSA group. TSA demonstrated a higher age-adjusted Constant score and a lower rate of revision surgery. The results of the multivariate analysis suggested that the surgical outcome was strongly influenced by the preoperative performance status.
A previous systematic review, which analyzed 1952 patients, showed that TSA was effective in reducing pain and improving ROM and patient satisfaction. The same study evidenced that TSA resulted in a significantly lower rate of revision surgery (6.5% versus 10.2%) [23]. A more recent review involving 2111 procedures (1783 TSA and 328 SHA) also found a lower revision rate in TSA (7% versus 13%) [24]. A previous meta-analysis involving 62 TSA procedures revealed a higher University of California at Los Angeles (UCLA) Shoulder Score than SHA [25]. Similarly, in a more recent meta-analysis including 153 procedures, TSA provides better clinical results than SHA [26]. Indeed, higher UCLA Shoulder Score and ASES were found in the TSA group [26]. In contrast, there was no significant difference between TSA and HA in revision, Western Ontario Osteoarthritis of

Discussion
Based on the main findings of this study, anatomic TSA performed better than SHA for GHOA. At last follow-up, the TSA group demonstrated greater flexion, abduction, and elevation. ASES and VAS were both greater in the TSA group. TSA demonstrated a higher age-adjusted Constant score and a lower rate of revision surgery. The results of the multivariate analysis suggested that the surgical outcome was strongly influenced by the preoperative performance status.
A previous systematic review, which analyzed 1952 patients, showed that TSA was effective in reducing pain and improving ROM and patient satisfaction. The same study evidenced that TSA resulted in a significantly lower rate of revision surgery (6.5% versus 10.2%) [23]. A more recent review involving 2111 procedures (1783 TSA and 328 SHA) also found a lower revision rate in TSA (7% versus 13%) [24]. A previous meta-analysis involving 62 TSA procedures revealed a higher University of California at Los Angeles (UCLA) Shoulder Score than SHA [25]. Similarly, in a more recent meta-analysis including 153 procedures, TSA provides better clinical results than SHA [26]. Indeed, higher UCLA Shoulder Score and ASES were found in the TSA group [26]. In contrast, there was no significant difference between TSA and HA in revision, Western Ontario Osteoarthritis of the Shoulder (WOOS) index, and instability [26]. The present systematic review and meta-analysis confirmed similar findings on a larger scale, analyzing more procedures and outcomes of interest compared with previous studies.
Pain relief and full functional recovery are the primary goals of shoulder prostheses [62]. TSA aims to restore the physiological biomechanics of the shoulder joint: spherical size, orientation, center of rotation, capsular tension, joint stability, and range of motion [63]. Neck inclination, humeral head diameter, thickness, height, retroversion, offset, and acromion-humeral distance are the main parameters to follow for a successful implant [64][65][66]. Polyethylene wear, aseptic loosening, and the erosion of the glenoid component are the most common complication of TSA [24,[67][68][69][70]. SHA is easier to perform and requires shorter surgical duration and learning curve [71]. HA is advantageous in the early stages of OA with eccentric GHOA [50]; however, given the higher risk of glenoid bone erosion, its use in advanced GHOA is controversial [72]. TSA, in contrast, is more demanding, with greater tissue damage, blood loss, and recovery time [73], but humeral component loosening is uncommon [74,75]. However, TSA provides greater joint stability, less pain, and increased ROM, especially in internal rotation [76,77], as highlighted by long-term follow-up studies [22,78]. In addition, TSA showed a lower rate of revision surgery, and is more cost effective than SHA [30,79,80]. Indeed, although TSA has higher initial costs, primarily from the longer operative time and the expense of the glenoid component [81], the lower revision rate of surgery and higher health-related quality-of-life (calculated by quality-adjusted life years, QALYs) make TSA more effective and less costly than the alternative [79]. Glenoid component failure is the most common complication leading to revision surgery following TSA [68,82,83], and may discourage the use of glenoid components in younger patients desiring higher activity levels [68,82,[84][85][86].
The present study has several limitations. The retrospective design of most of the included studies has a negative effect on the reliability of the conclusions. Given the limited quantitative data available in the current literature, surgical techniques, approaches, and implants were not analyzed separately. Between-study heterogeneity in the eligibility criteria was shown. The included studies reported that patients presenting severe pain unresponsive to nonoperative management and associated with functional limitation and radiographic signs of GHOA were eligible for replacement. Patients with secondary GHOA, such as history of trauma, instability, rheumatoid arthritis, avascular necrosis of the humeral head, and prior shoulder surgery were not eligible in most of the studies [21,27,30,38,41,48,56,57,61]. Some authors required a period of persistent pain [19,31,39,43,44,48,52,53,55], others required 3 to 6 months of nonoperative managements, including analgesics, anti-inflammatory medications, physiotherapy, and home range-of-motion exercise programs [27,31,45,49]. Two studies performed the intervention in patients younger than 50 years of age [39,54]. However, many studies did not report any inclusion or exclusion criteria [17,22,29,40,42,46,47,50,[58][59][60]. The status of the rotator cuff influences the outcome of shoulder replacement [82]. In rotator cuff-deficient patients, the unopposed action of the deltoid causes superior migration of the humeral head toward the acromion and coracoacromial arch, leading to poor functional results, glenoid component loosening, and reduced implant survivorship [87]. The average arm forward elevation of patients with impaired rotator cuff was 88 • , whereas 150 • was achieved by patients with intact rotator cuff [88]. The status of the rotator cuff in patients included in the studies considered for the present investigation was homogeneous and often biased, thus representing a further limitation. In some of the studies included in our article, the rotator cuff was intact before surgery [21,22,27,28,30,32,38,41,48,49,51,56,57,61]. Other authors repaired intraoperatively rotator cuff lesions [17,19,29,31,39,40,42,44,46,47,50,[52][53][54][55][58][59][60].
Two studies did not specify the status of the rotator cuff in their patients [43,45]. The age-adjusted Constant score has been used for analysis; however, although widely used to assess outcome for shoulder implants, the score itself is not validated [35]. Given these limitations, the data from the present study must be interpreted with caution.

Conclusions
Anatomic TSA performed better than SHA in patients with glenohumeral osteoarthritis.
Author Contributions: F.M.: writing, data items, statistical analyses; G.V.: data source and extraction, selection and data collection, data items, methodology quality assessment; A.B.: writing; A.P.: writing, data source and extraction, selection and data collection, data items, methodology quality assessment;