Keeled versus pegged glenoid components in total shoulder arthroplasty for primary osteoarthritis: a meta-analysis
Article information
Abstract
Background
Glenoid fixation strongly influences longevity and outcomes in anatomic total shoulder arthroplasty. Keeled and pegged designs are common, but comparative evidence for function, range of motion, and survivorship is inconsistent.
Methods
A systematic search of PubMed, Scopus, the Cochrane Library, and Google Scholar was conducted from database inception through December 2025. Thirteen comparative studies (n=3,219) met the inclusion criteria. Evaluated outcomes included patient-reported outcome measures (American Shoulder and Elbow Surgeons [ASES], Constant-Murley score, Simple Shoulder Test[SST], visual analog scale [VAS] pain scores), shoulder range of motion (flexion, external rotation, internal rotation), and implant-related outcomes, including failure and revision. Follow-up ranged from short to long-term (up to 26 years).
Results
Across pooled analyses, no significant differences were observed between pegged and keeled glenoid components in ASES (P=0.69), Constant-Murley (P=0.61), SST (P=0.74), or VAS pain scores (P=0.88). Postoperative flexion (P=0.61) and external rotation (P=0.35) were comparable between groups. Pegged glenoid components demonstrated a statistically significant but small improvement in internal rotation (mean difference, 1.60°; 95% CI, 0.54°–2.66°; P=0.003). Revision rates were significantly lower with pegged glenoid components compared with keeled designs (risk ratio, 0.49; 95% CI, 0.28–0.86; P=0.01), while overall failure rates did not differ significantly between groups (P=0.07).
Conclusions
Both designs yield similar pain relief, functional gains, and shoulder motion across most patient-reported outcome measures. Pegged components show lower revision risk and slightly better internal rotation, although survivorship findings are primarily supported by observational evidence. Selection should be individualized to anatomy, bone quality, and surgical factors, not expectations of superior overall function.
Level of evidence
II.
INTRODUCTION
Glenohumeral osteoarthritis is a common and debilitating condition that significantly impairs shoulder mobility, function, and quality of life [1]. It represents one of the leading indications for shoulder arthroplasty in adults over 55 years, with prevalence increasing markedly with age and showing a slight female predominance [2]. Degenerative changes of the glenohumeral joint are strongly associated with age-related cartilage wear, altered joint biomechanics, rotator cuff integrity, glenoid morphology, and prior trauma or instability [3,4]. Before surgical intervention is considered, patients typically undergo conservative management, including activity modification, physiotherapy, nonsteroidal anti-inflammatory drugs, intra-articular corticosteroid or hyaluronic acid injections, and analgesic regimens [5]. These measures aim to reduce pain and preserve function but often provide diminishing benefit as structural degeneration progresses [6]. For patients with advanced disease that is refractory to nonoperative treatment, anatomic total shoulder arthroplasty (TSA) remains the gold-standard surgical option [7].
Glenoid component fixation represents a critical determinant of implant longevity and clinical success in TSA [8]. Keeled and pegged glenoid components are the two most widely used designs. Both aim to restore joint congruency, alleviate pain, and improve function, yet they differ fundamentally in fixation geometry, bone preparation requirements, cement distribution, and theoretical load transfer to the glenoid bone [9,10]. These design differences have prompted ongoing debate over their relative performance, particularly with respect to radiographic loosening, implant survivorship, and the risk of revision [11].
Despite extensive primary comparative studies, the remaining clinical uncertainty is not whether patients improve after anatomic TSA, as short to mid-term patient-reported outcomes with keeled and pegged components are similar in most studies, but whether one design confers a clinically relevant survivorship advantage when contemporary data are pooled [12]. Prior evidence syntheses have been constrained by heterogeneous implant eras and methods. Critically, many of the comparative cohorts used for pooled estimates of revision outcomes have limited long-term follow-up, which reduces power to resolve small but consequential differences in durability [13]. Since those earlier syntheses, large contemporary observational datasets and newer long-term comparative cohorts have added revision-specific information capable of shifting pooled survivorship estimates and improving precision [14]. Accordingly, an updated meta-analysis is needed to focus specifically on primary osteoarthritis and anatomic TSA. Emphasizing revision-related endpoints in addition to patient-reported outcome measures (PROMs) and range of motion (ROM) would better inform implant selection when the clinical trade-off is “equivalent function” but potentially different long-term durability.
Keeled glenoid components use a single central keel to achieve fixation within the glenoid vault and have a long history of clinical use. They are often considered technically straightforward and provide robust initial fixation, particularly in patients with adequate bone stock [15,16]. Pegged glenoid components, in contrast, rely on multiple smaller pegs, typically cemented, to enhance fixation and potentially reduce stress shielding and micromotion at the bone–implant interface [17]. The theoretical advantages in bone preservation and load distribution of pegged designs have increased their use, although concerns about cement mantle integrity, radiolucent line formation, and long-term durability persist [18]. Differences in glenoid morphology, bone quality, and patient age further complicate implant selection and may influence outcomes [19].
These contrasting design philosophies translate into distinct risk profiles and postoperative considerations. While both keeled and pegged glenoid components generally provide reliable pain relief and functional improvement, glenoid loosening remains the most common cause of failure following anatomic TSA. Advances in implant design, cementing techniques, and surgical instrumentation have improved outcomes for both designs, yet conflicting evidence persists regarding radiographic performance, revision rates, and long-term survivorship. However, the most clinically relevant unresolved question is whether any survivorship differences become detectable once newer long-term cohorts and large contemporary datasets are incorporated, particularly for revision outcomes that individual studies are often underpowered to detect. Therefore, this review updates prior syntheses by extending the search through December 2025 and pooling contemporary comparative evidence, including large registry-based revision risk analyses and newer long-term matched cohorts, while maintaining a focused indication (primary osteoarthritis) and procedure type (anatomic TSA). In addition to PROMs and ROM, we prioritize implant survivorship endpoints (revision and failure) to translate the keeled-versus-pegged debate into decision-relevant risk estimates for surgeons counseling patients about durability.
METHODS
This study did not require institutional review board approval because it is a meta-analysis of previously published studies.
Search Strategy
A comprehensive literature search was conducted in PubMed, Scopus, the Cochrane Library, and Google Scholar, covering all available studies from database inception through December 2025. The search strategy combined predefined keywords and Medical Subject Headings (MeSH) using Boolean operators (“AND”, “OR”). The following terms and their variations were used: “total shoulder arthroplasty,” “anatomic total shoulder arthroplasty,” “TSA,” “glenoid component,” “keeled glenoid,” “pegged glenoid,” “glenoid fixation.” Reference lists of all the included articles and relevant systematic reviews were screened manually to identify any additional eligible studies not retrieved through the initial search.
Eligibility Criteria and Study Selection
Eligible studies included adult patients undergoing primary anatomic TSA for glenohumeral osteoarthritis based on clinical and radiographic criteria. The intervention of interest was pegged glenoid components implanted during primary anatomic TSA using any cemented design. The comparator was keeled glenoid components used in primary anatomic TSA, including standard or modified keel designs. Only peer-reviewed full-text comparative clinical studies, including randomized controlled trials (RCTs), prospective or retrospective cohort studies, and registry-based analyses, that directly compared outcomes between pegged and keeled glenoid components were included.
Exclusion criteria included: non-comparative designs (e.g., single-arm cohorts, case series, case reports); cadaveric, biomechanical, or animal studies; reviews, systematic reviews, meta-analyses, editorials, or conference abstracts; non-English publications; studies evaluating inflammatory arthropathies (e.g., rheumatoid arthritis), fracture-related pathology, tumor, or revision shoulder arthroplasty; studies involving reverse TSA or mixed arthroplasty designs; and studies where separate data for keeled and pegged glenoid components could not be extracted.
Data Extraction
Two independent reviewers extracted data from each of the included studies using a standardized spreadsheet developed for this meta-analysis. Extracted variables covered study-level characteristics such as author and publication year, study design, surgical group sample size (pegged vs keeled), age (mean±standard deviation), sex distribution (male:female), implant fixation/type (cemented vs mixed cemented/noncemented), and additional study-level descriptors when reported (e.g., ASA status, glenoid size categories, laterality, and diagnosis labels such as degenerative joint disease/osteoarthritis).
Outcomes extracted included American Shoulder and Elbow Surgeons (ASES), Constant-Murley, Simple Shoulder Test (SST), Western Ontario Osteoarthritis of the Shoulder index (WOOS), Single Assessment Numeric Evaluation (SANE), visual analog scale (VAS) scores, ROM measures (flexion, external rotation, internal rotation, abduction), radiographic endpoints (radiographic lucency/loosening, translation), complications (e.g., dislocation), failure, and revision rate. For implant survivorship endpoints, “revision” was defined as any postoperative surgical reintervention reported by the original study that involved revision arthroplasty and/or implant removal, exchange, or addition (including component-specific revision when applicable). “Failure” was characterized as clinical failure (e.g., revision arthroplasty or component removal) and/or radiographic failure indicating clinically relevant loosening risk (e.g., component migration/shift or circumferential radiolucency meeting predefined thresholds).
This review examined postoperative outcomes across functional, clinical, ROM, and implant survival domains. Functional outcomes included ASES, Constant-Murley score, and SST. ROM outcomes encompassed postoperative shoulder flexion, external rotation, and internal rotation. Clinical outcomes included postoperative VAS pain scores. Implant survival outcomes evaluated failure events and revision surgery, with revision-free survival assessed through pooled revision risk analysis.
All extracted data were independently cross-verified for accuracy. Disagreements between reviewers were resolved through discussion or, when necessary, adjudication by a third reviewer.
Risk of Bias Assessment
Study quality was assessed using the RoB 2 tool for RCTs and Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) for non-randomized studies. RoB 2 evaluates five domains: randomization, deviations from intended interventions, missing data, outcome measurement, and selective reporting [20]. ROBINS-I addresses similar domains adapted for non-randomized designs (bias in selection of participants, classification of interventions, measurement of outcomes, selection of the reported results, and missing data) [21]. Each domain was rated as low risk, some concerns, or high risk, with the overall judgment based on the highest rating. Assessments were performed independently by two reviewers, with disagreements resolved by a third reviewer.
Statistical Analysis
All statistical analyses were performed using Review Manager 5.4 (Cochrane Collaboration, 2020). Continuous outcomes were analyzed using either mean differences (MD) or standardized MD, both reported with 95% CIs, depending on the consistency of the measurements. Dichotomous outcomes were evaluated using risk ratios (RRs) with 95% CIs. Heterogeneity was assessed with the Q test and the I2 statistic; a P-value ≤0.10 or an I2 value > 50% was considered indicative of significant heterogeneity. A random-effects model was used when heterogeneity was substantial (I2 >50%). When heterogeneity was low to moderate (I2 ≤50%), a fixed-effect model was applied. Statistical significance was defined as P<0.05. When feasible, sensitivity analyses were conducted for outcomes demonstrating statistically significant and clinically relevant between-group differences. Large registry-based cohorts were excluded when they disproportionately influenced pooled estimates.
RESULTS
Study Selection
The systematic literature search resulted in 301 relevant preliminary records across electronic databases. Following deduplication, 206 records were assessed for title and abstract screening, leaving 86 studies for full-text evaluation. Of these, 13 studies met all inclusion criteria and were included in the quantitative synthesis. A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart (Fig. 1) provides a detailed overview of record exclusion at each stage.
Characteristics of Included Studies
A total of 13 comparative studies published between 2007 and 2023 were included, encompassing 3,219 patients who underwent primary anatomic TSA with either pegged or keeled glenoid components. The included studies were six retrospective cohort studies, three prospective cohort studies, one retrospective case–control study, and three RCTs.
Sample sizes varied, ranging from small single-center cohorts with 10–26 patients per group to large registry-based cohorts including 1,448 pegged and 202 keeled glenoid implants. Overall, 2,232 patients received pegged glenoid components, whereas 987 patients received keeled glenoid components. Follow-up durations ranged from short-term assessments of 6–12 months to long-term analyses extending beyond 10 years, with several studies reporting implant survivorship and failure outcomes at mid to long-term follow-up.
Across the included studies, patients undergoing TSA with pegged glenoid components had mean ages ranging from 57.0 to 71.8 years, whereas keeled glenoid cohorts had mean ages ranging from 60.7 to 71.0 years. Sex distribution varied across studies, with pegged glenoid cohorts comprising approximately 1,060 males and approximately 1,120 females, whereas keeled glenoid cohorts included approximately 570 males and approximately 400 females, indicating a slightly higher female representation overall. Body mass index reporting varied across studies, generally falling within the overweight to obese range, with some studies excluding patients above predefined thresholds.
Across the included studies, the reported indication for TSA was primarily degenerative joint disease/osteoarthritis, with some studies also including other degenerative diagnoses when specified in the study characteristics table. Additional baseline descriptors were variably reported, including ASA status, glenoid size categories, and laterality.
Baseline clinical parameters, such as age and sex distribution, were commonly reported, whereas baseline functional scores were not consistently reported as preoperative values (Table 1) [14,22-33]. Reported outcome measures across studies included ASES, Constant-Murley, SST, SANE, WOOS, VAS pain, and postoperative ROM measures (e.g., flexion, external rotation, internal rotation, and abduction) when available. Detailed baseline characteristics and study-level parameters are provided in Table 1.
Methodological Quality Assessment
Non-randomized studies (ROBINS-I)
All non-randomized studies were evaluated using the ROBINS-I tool. As illustrated in Fig. 2A, most studies demonstrated an overall low to moderate risk of bias, with only a minority showing serious concerns, primarily reflecting the inherent limitations of retrospective observational study designs. The domain most frequently affected was bias due to confounding (D1), indicating incomplete control for baseline differences between comparison groups.
Risk of bias assessment across included studies. (A) Risk of bias for non-randomized studies assessed using the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I). (B) Risk of bias for randomized controlled trials assessed using the RoB 2.
Bias due to selection of participants (D2) was generally rated as low, although serious concerns were identified in a small number of studies with less clearly defined cohort selection. Other domains, including classification of interventions (D3), deviations from intended interventions (D4), and missing data (D5), were predominantly assessed as low risk, reflecting clear identification of glenoid component type, consistent surgical practice, and acceptable follow-up completeness. Bias in measurement of outcomes (D6) showed occasional moderate concerns, particularly in studies relying on radiographic assessment or patient-reported outcomes, and selective reporting (D7) was generally judged as low risk. Overall, the visual summary suggests that although residual confounding remains the principal methodological limitation, the quality of the non-randomized evidence is acceptable for comparative synthesis (Fig. 2A).
Randomized studies (RoB 2)
The RCTs were assessed using the RoB 2 tool, and the results indicate that the overall risk of bias was low to moderate across studies. Most trials demonstrated low risk of bias across most domains, with some studies exhibiting concerns primarily related to outcome measurement, where lack of blinding of outcome assessors or reliance on radiographic interpretation introduced potential subjectivity. Bias arising from the randomization process, deviations from intended interventions, missing outcome data, and selection of the reported result were consistently rated as low risk, reflecting adequate random sequence generation, standardized surgical techniques, complete follow-up, and reporting of predefined outcomes. Overall, the included RCTs demonstrated good methodological quality with limited concerns that were unlikely to significantly influence the comparative assessment of keeled versus pegged glenoid components (Fig. 2B).
Patient-Reported Outcome Measures
ASES score
Three non-randomized comparative studies reported postoperative ASES scores following TSA, comprising 146 patients in the pegged group and 81 in the keeled group. No RCTs reported this outcome. The mean follow-up duration across the included studies was approximately 9.2±2.7 years (range, 5.0–26 years). ASES is a validated composite outcome measure assessing shoulder pain and function, with higher scores indicating better clinical outcomes. Pooled random-effects meta-analysis demonstrated no statistically significant difference in postoperative ASES scores between pegged and keeled glenoid components (MD, 2.08 points; 95% CI, −8.12 to 12.27 points; P=0.69). Moderate heterogeneity was observed (I2=67%) (Fig. 3).
Constant-Murley score
Two studies assessed postoperative Constant-Murley scores following TSA, including one RCT and one non-randomized comparative study, comprising 32 patients in the pegged group and 31 in the keeled group. The mean follow-up duration across the included studies was approximately 3.5±2.2 years (range, 2.0–8.6 years). The Constant-Murley score is a clinician-based composite outcome evaluating shoulder pain, activities of daily living, strength, and ROM, with higher scores indicating superior functional outcomes. The RCTs demonstrated no statistically significant difference between pegged and keeled components (MD, −3.00; 95% CI, −15.46 to 9.46; P=0.64). Similarly, the non-randomized study also showed no statistically significant between-group difference (MD, 1.40; 95% CI, −2.59 to 5.39; P=0.49) and contributed most of the weight (90.7%) to the pooled estimate. When both studies were pooled, the combined fixed-effect analysis confirmed no statistically significant difference in Constant-Murley scores between pegged and keeled glenoid designs (MD, 0.99; 95% CI, −2.80 to 4.79; P=0.61), with low observed heterogeneity (I2=0%). Subgroup analysis demonstrated no statistically significant difference between RCT and non-RCT estimates (P for subgroup differences=0.51) (Fig. 4).
SST score
Four non-randomized comparative studies assessed postoperative SST scores following TSA, comprising 288 patients in the pegged group and 139 in the keeled group. No RCTs reported this outcome. The mean follow-up duration across included studies was approximately 8.9±2.2 years (range, 5.0–26 years). The SST is a patient-reported outcome measure evaluating shoulder function based on the ability to perform specific daily activities, with higher scores indicating better functional outcomes. Pooled random-effects meta-analysis demonstrated no statistically significant difference in SST scores between pegged and keeled glenoid components (MD, 0.15; 95% CI, −0.74 to 1.05; P=0.74). Moderate heterogeneity was observed (I2=57%) (Fig. 5).
VAS score
Three non-randomized comparative studies assessed postoperative VAS pain scores following TSA, comprising 146 patients in the pegged group and 81 in the keeled group. No RCTs reported this outcome. The mean follow-up duration across the included studies was approximately 9.2±2.7 years (range, 5.0–26 years). The VAS is a subjective patient-reported measure of pain intensity, in which lower scores indicate less pain. Pooled random-effects meta-analysis demonstrated no statistically significant difference in postoperative VAS pain scores between pegged and keeled glenoid components (MD, 0.08; 95% CI, −0.98 to 1.15; P=0.88). Moderate heterogeneity was observed (I2=66%) (Fig. 6).
Shoulder ROM
Flexion
Two non-randomized comparative studies assessed postoperative shoulder flexion following TSA, comprising 122 patients in the pegged group and 69 in the keeled group. No RCTs reported this outcome. The mean follow-up duration across included studies was approximately 9.2±3.7 years (range, 2.0–26 years). Pooled fixed-effect meta-analysis demonstrated no statistically significant difference in postoperative flexion between pegged and keeled glenoid components (MD, −1.84°; 95% CI, −8.87° to 5.19°; P=0.61). Low heterogeneity was observed (I2=30%) (Fig. 7).
External rotation
Three non-randomized comparative studies assessed postoperative external rotation following TSA, comprising 146 patients in the pegged group and 81 in the keeled group. The mean follow-up duration across the included studies was approximately 9.2±2.7 years (range, 2.0–26 years). No RCTs reported this outcome. Pooled fixed-effect meta-analysis demonstrated no statistically significant difference in postoperative external rotation between pegged and keeled glenoid components (MD, −2.22°; 95% CI, −6.86° to 2.42°; P=0.35). Low heterogeneity was observed (I2=0%) (Fig. 8).
Internal rotation
Two non-randomized comparative studies (Pegged=126; Keeled=54) reported postoperative internal rotation. No RCTs reported this outcome. The mean follow-up duration across the included studies was approximately 10.5±2.0 years (range, 2.0–26 years). Pooled analysis demonstrated a statistically significant difference favoring pegged glenoid components (MD, 1.60°; 95% CI, 0.54°–2.66°; P=0.003), with low heterogeneity (I2=0) (Fig. 9).
Implant-Related Complications
Revision
Seven studies assessed revision rates following TSA, including one RCT and six non-randomized comparative studies, comprising 1,988 patients in the pegged group and 818 in the keeled group. Across contributing studies, revision was reported as postoperative reintervention, including component-specific revision/removal, revision arthroplasty procedures (including conversion procedures), and, in some cases, isolated component removal for loosening. Indications for revision varied across studies and included aseptic loosening, structural failure (e.g., fracture), instability, infection, painful loosening/persistent pain prompting reoperation, and wear. The mean follow-up duration across the included studies was approximately 7.2±3.0 years (range, 2.0–26 years).
The RCT demonstrated no statistically significant difference in revision risk between pegged and keeled glenoid components (RR, 0.34; 95% CI, 0.08–1.57; P=0.17). In contrast, the non-randomized studies demonstrated a significantly lower risk of revision in the pegged group (RR, 0.52; 95% CI, 0.29–0.94; P=0.03) and contributed most weight (86.8%) to the pooled estimate. When all studies were pooled, the combined fixed-effect meta-analysis demonstrated a significantly lower risk of revision in patients receiving pegged glenoid components compared with keeled components (RR, 0.49; 95% CI, 0.28–0.86; P=0.01), with low statistical heterogeneity (I2=0%). Subgroup analysis revealed no statistically significant difference between RCT and non-RCT estimates (P for subgroup differences=0.62) (Fig. 10).
To evaluate the influence of high-weight observational data, a sensitivity analysis excluding the large registry-based cohorts (Dillon et al. [14] and Fox et al. [24]) was performed. Compared with the primary analysis, the pooled estimate continued to favor pegged components (RR, 0.46; 95% CI, 0.21–1.00; P=0.04), but precision decreased, with a wider CI, reflecting reduced sample size and event count.
Failure
Three studies assessed failure events following TSA, including two RCTs and one non-randomized comparative study, comprising 105 patients in the pegged group and 107 in the keeled group. Across the contributing studies, failure was reported using both clinically treated failure events (e.g., revision arthroplasty or component removal) and radiographic loosening-risk endpoints (e.g., component migration/shift and/or circumferential radiolucency meeting predefined thresholds). The mean follow-up duration across the included studies was approximately 4.7±2.9 years (range, 1.0–12 years).
The RCT-only analysis demonstrated no statistically significant difference in failure rates between pegged and keeled glenoid components (RR, 0.31; 95% CI, 0.08–1.20; P=0.09) with low observed heterogeneity (I2=0%). Similarly, the non-randomized study also showed no statistically significant between-group difference (RR, 0.76; 95% CI, 0.51–1.12; P=0.17). When all studies were pooled, the combined fixed-effect meta-analysis demonstrated no statistically significant difference in failure rates between pegged and keeled glenoid components (RR, 0.71; 95% CI, 0.49–1.03; P=0.07), with low observed heterogeneity (I2=0%). Subgroup analysis revealed no statistically significant difference between RCT and non-RCT estimates (P for subgroup differences=0.21) (Fig. 11).
DISCUSSION
The principal finding of this meta-analysis is that pegged and keeled glenoid components in anatomic TSA provide comparable patient-reported outcomes and overall shoulder motion, and pegged designs are associated with a significantly lower risk of revision surgery. Across pooled analyses, no meaningful differences were observed in validated functional scores or pain measures, and ROM outcomes were largely equivalent, except for a modest advantage in internal rotation favoring pegged components. Collectively, these findings suggest that while both glenoid designs reliably achieve pain relief and functional restoration, the primary goals of TSA, implant survivorship may differ in clinically relevant ways.
This meta-analysis demonstrated no significant differences between pegged and keeled glenoid components across PROMs, which included ASES, Constant-Murley, SST, and VAS pain scores. These findings indicate that both designs are equally effective in improving pain, daily function, and overall patient satisfaction following anatomic TSA [34]. The absence of PROM differences may reflect the multifactorial nature of patient-perceived outcomes, which are strongly influenced by rotator cuff integrity, humeral component positioning, soft-tissue balancing, and postoperative rehabilitation rather than glenoid fixation geometry alone [35]. Prior comparative studies and registry analyses have also reported equivalent PROMs between glenoid designs, reinforcing the view that modern cemented keeled and pegged components both achieve reliable clinical improvement [36,37].
However, preoperative baseline functional scores and radiographic disease severity were not consistently reported across the included comparative studies. Several of the included studies did report baseline PROMs [22,25,28-31,33], whereas the remaining studies were primarily survivorship or radiographic-focused and did not provide baseline PROM reporting. Moreover, even among studies that reported baseline PROMs, preoperative equivalence was not always demonstrated. For instance, baseline ASES and WOOS differed significantly between groups in Gascoyne et al. [25], and baseline pain scores differed in Moulton et al. [29]. Given that baseline function and radiographic glenoid severity (e.g., Walch classification, erosion, subluxation) influence postoperative recovery potential, this heterogeneous reporting limits the extent to which pooled PROMs can isolate the full effect of glenoid fixation design. Furthermore, statistical heterogeneity varied across PROMs (e.g., Constant-Murley I2=0% vs. ASES/SST/VAS showing moderate I2). Even when I2 is low, it should not be interpreted as the absence of clinical heterogeneity, as the included cohorts differed in baseline patient characteristics and follow-up duration. From a surgical perspective, these findings suggest that glenoid design selection alone should not be expected to drive differences in short to mid-term patient-reported outcomes, and that meticulous surgical technique and appropriate patient selection remain the dominant determinants of perceived success [38].
Regarding shoulder ROM, pooled analyses revealed no significant differences in postoperative flexion or external rotation between groups [39]. Although statistical heterogeneity was low to moderate for these ROM outcomes, relevant clinical variability may persist across studies due to differences in rehabilitation protocols, surgical technique, and baseline stiffness. Additionally, baseline ROM was not uniformly reported across all comparative cohorts. Internal rotation demonstrated a statistically significant advantage favoring pegged glenoid components; however, this finding must be interpreted cautiously [40]. The number of contributing studies was limited, and the observed MD was small and probably below established minimal clinically important difference thresholds for shoulder motion, suggesting minimal functional relevance [41]. In addition, the low I2 estimate for internal rotation should be interpreted cautiously, given the small number of studies, which limits the ability of heterogeneity statistics to detect between-study variability. Biomechanically, improved internal rotation with pegged designs may relate to more uniform cement distribution, reduced stress concentration, or improved humeral head centering, but these effects appear modest in clinical practice [42]. Existing literature generally reports comparable postoperative ROM between glenoid designs, with variations more strongly influenced by preoperative stiffness, capsular management, subscapularis handling, and rehabilitation protocols [43,44]. Consequently, although the observed difference in internal rotation is statistically detectable, it does not support a meaningful overall ROM advantage for either glenoid design.
In contrast, implant-related outcomes revealed a clinically important difference in revision risk, with pegged glenoid components being associated with a significantly lower likelihood of revision surgery compared with keeled components. Although statistical heterogeneity was low (I2=0%), this should not be interpreted as true homogeneity, as revision outcomes remain highly dependent on clinical differences across cohorts (implant generation, cementing technique, and follow-up duration) and are disproportionately weighted by large registry-based datasets. Furthermore, stratified analysis by study design suggested that the observed revision advantage was primarily driven by observational evidence. Additionally, revision is a time-dependent endpoint, and heterogeneity in follow-up duration across studies limits the interpretation of pooled revision risk as a uniform long-term survivorship comparison. Residual confounding may also reflect implant generation and surgical era differences across cohorts, which could influence revision risk independent of keel versus peg geometry. Revision thresholds and indications differed across studies and follow-up strategies. Accordingly, pooled revision estimates reflect a range of clinically treated reinterventions. Furthermore, sensitivity analysis excluding the large registry-based cohorts demonstrated a consistent direction of effect, although with reduced precision. This supports robustness of direction, but does not eliminate residual confounding related to indication, implant era, or follow-up duration. Notably, this difference was not mirrored by a significant difference in reported failure rates, suggesting that revisions may occur due to progressive loosening, pain, or surgeon-driven decision-making rather than overt implant failure alone [45]. Interpretation of pooled failure outcomes should also account for variability in failure definitions across studies, which may not be captured by low statistical heterogeneity metrics. Pegged glenoid components may confer improved long-term fixation through more favorable load distribution, reduced micromotion, and diminished stress concentration within the glenoid vault [46]. Prior biomechanical, radiographic, and registry-based studies have similarly identified lower loosening and revision rates with pegged designs, particularly at mid to long-term follow-up [27,47]. Clinically, reduced revision risk is a highly relevant endpoint, as revision TSA carries increased morbidity, technical complexity, and inferior outcomes compared with primary procedures [48]. Therefore, although our findings suggest an association between pegged glenoid design and reduced revision risk, this advantage should be interpreted as associative rather than definitively causal, and should be integrated with patient-specific anatomy and surgeon technique when selecting implants.
Beyond glenoid design alone, multiple additional factors critically influence outcomes following anatomic TSA and warrant consideration in clinical decision-making [49]. Glenoid morphology, bone stock quality, version abnormalities, and Walch classification strongly affect fixation success and long-term stability, often outweighing design-related differences [50,51]. Surgeon experience, cementing technique, component positioning, and soft-tissue balancing play decisive roles in preventing loosening and optimizing function [52,53]. Furthermore, outcomes not specifically analyzed in this meta-analysis, including radiolucent line progression, scapular notching, polyethylene wear, patient satisfaction, cost-effectiveness, and ease of revision, have been inconsistently reported in the literature, with mixed findings between keeled and pegged designs [54,55]. Some studies suggest increased radiographic lucency with keeled components, while others report equivalent radiographic performance when modern cementing techniques are used [56,57]. These data reinforce the conclusion that glenoid design functions within a broader biomechanical and surgical ecosystem rather than acting as an isolated determinant of success.
In summary, the available evidence supports a measured, patient-centered approach to glenoid selection in anatomic TSA. Both pegged and keeled glenoid components provide comparable pain relief, functional improvement, and overall shoulder motion. Pegged designs may offer an advantage in long-term implant survivorship through reduced revision risk. However, given the equivalence of most clinical outcomes, glenoid selection should be individualized based on glenoid anatomy, bone quality, patient age and activity level, and surgeon expertise rather than an expectation of superior global function. Optimizing outcomes in anatomic TSA ultimately depends on comprehensive preoperative planning, precise surgical execution, and alignment of implant choice with patient-specific anatomical and functional goals.
This meta-analysis has several limitations that warrant consideration. First, many of the included studies were retrospective, with various follow-up durations, increasing susceptibility to selection bias and residual confounding, including confounding by indication, despite matching or statistical adjustment. Thus, pooled estimates may reflect mixed short, mid, and long-term follow-up rather than a uniform survivorship interval. In fact, the pooled survivorship outcomes were heavily influenced by non-randomized observational and registry-based studies. Second, baseline preoperative functional status and radiographic disease severity were not consistently reported across the included studies, and some studies that reported PROMs demonstrated missing baseline values or baseline group differences, limiting assessment of baseline comparability between keeled and pegged cohorts. This was particularly notable in registry/survivorship and radiographic-focused studies. Third, variation in implant design generation, cementing techniques, surgical approaches, and rehabilitation protocols may independently affect outcomes and limit direct comparability, even when statistical heterogeneity estimates are low. Implant generation and surgical era were not consistently reported, and improvements in polyethylene quality, cementing technique, and instrumentation over time may have contributed to survivorship differences independent of keel or peg geometry. Fourth, not all studies reported complete quantitative data for all outcomes, restricting pooling for certain measures and necessitating narrative synthesis in some domains. Fifth, radiographic assessment methods and definitions of loosening varied across studies, potentially contributing to measurement inconsistency. Relatedly, definitions of implant failure and revision varied across studies, which may affect the comparability of pooled survivorship endpoints, particularly revision, because pooled estimates may combine operative events arising from different indications and reintervention thresholds across cohorts. Finally, long-term outcomes remain underrepresented in high-quality randomized trials, underscoring the need for further prospective studies with standardized reporting to refine evidence on glenoid component performance.
CONCLUSIONS
In summary, this meta-analysis demonstrates that both pegged and keeled glenoid components in anatomic TSA provide effective pain relief and meaningful functional improvement for patients with degenerative glenohumeral disease. Most functional and PROMs were comparable between designs. Although a small statistical difference in internal rotation favored pegged components, its clinical relevance appears limited. In contrast, pegged glenoid components were associated with a significantly lower risk of revision, particularly in observational evidence, which may indicate a potential advantage in implant durability and long-term survivorship. However, keeled glenoid components achieved similar clinical outcomes and remain a dependable option when appropriately selected.
These findings suggest that pegged glenoid components may be preferential for patients prioritizing long-term implant durability and revision avoidance, whereas keeled glenoid components remain suitable in selected anatomical scenarios and surgical settings. Ultimately, optimal outcomes in anatomic TSA are driven less by a universal implant preference and more by individualized surgical planning, meticulous technique, and the careful alignment of implant choice with patient-specific anatomy, biomechanics, and functional expectations.
Notes
Author contributions
Conceptualization: MB, GA. Data curation: GA, JS, ZS, RAK. Formal analysis: MB. Supervision: GA. Validation: RBR, BE. Visualization: JS, ZS, RAK. Writing – original draft: M Boutros, GA. Writing – review & editing: RBR, BE. All authors read and agreed to the published version of the manuscript.
Conflict of interest
None.
Funding
None.
Data availability
Contact the corresponding author for data availability.
Acknowledgments
None.
