Defining surgeon experience thresholds for the reduction in complications and revisions after shoulder arthroplasty: a meta-analysis of 332,542 patients

Abstract

Background

Shoulder arthroplasty utilization has increased substantially, with a growing proportion of cases performed by surgeons early in the learning curve. Surgeon-volume definitions vary widely across the literature, and although higher volume is generally associated with better outcomes, there is no consensus on the thresholds that define high volume. This study aimed to determine shoulder arthroplasty cases/yr thresholds at which complication and revision risks diverge between low- and high-volume surgeons.

Methods

PubMed, Cochrane, and Google Scholar were queried through December 2025. Inclusion criteria consisted of comparative studies evaluating the number of complications and revisions after shoulder arthroplasty across different categories of surgeon experience. Five different thresholds of cases/yr were established to perform the analysis and define high- and low-volume surgeons: 5 cases/yr, 10 cases/yr, 14 cases/yr, 20 cases/yr, and 25–28 cases/yr.

Results

Eight studies with 332,542 patients were included. At the 5 cases/yr threshold, low-volume surgeons had higher odds of complications (odds ratio [OR] = 1.41; 95% confidence interval [CI], 1.28–1.55; P< .001) and revision (OR = 1.41; 95% CI, 1.28–1.57; P< .001). At the 10 cases/yr threshold, there were no differences in complications (OR = 1.21; 95% CI, 0.94–1.56; P= .14) or revision (OR = 1.18; 95% CI, 0.90–1.54; P= .22). At the 14 cases/yr threshold, revisions were similar (OR = 1.15; 95% CI, 0.89–1.48; P= .28). At the 20 cases/yr threshold, revision did not differ (OR = 1.15; 95% CI, 0.87–1.52; P= .34). At the 25–28 cases/yr threshold, low-volume surgeons had higher odds of complications (OR = 1.39; 95% CI, 1.11–1.75; P= .004) and revision (OR = 1.22; 95% CI, 1.15–1.30; P< .001).

Conclusion

This study identified surgeon-volume thresholds for shoulder arthroplasty of 5 cases/yr and 25–28 cases/yr, which may possibly define low-to-medium and medium-to-high-volume thresholds, respectively. Patients treated by surgeons below these thresholds had higher risks of complications and revision compared with those treated by surgeons above the respective thresholds.

The incidence of shoulder arthroplasty has increased substantially, accompanied by a corresponding increase in the number of surgeons performing these procedures. Between 2011 and 2017, the number of shoulder arthroplasty increased by over 100% in the United States.³⁰ Consequently, revision procedure volumes have also increased, carrying critical financial implications.⁵ The presence of a learning curve with the use of a new prosthesis or surgical technique is well recognized,⁵ with early phases characterized by longer operative times, higher complication rates, and increased risk of reoperation.^2,33 These trends highlight the growing importance of understanding how surgeon experience influences outcomes in modern shoulder arthroplasty.

Thresholds used to define surgeon volume vary widely across the existing literature, with low-volume categories defined as fewer than 2 to fewer than 10 cases annually and high-volume definitions spanning more than 5 to 20 cases per year.^5,16,29,32 Ramkumar et al defined surgeons performing 0–4 cases annually as low volume, 5–14 as medium volume, and ≥15 as high volume.²³ Higher surgeon volume has been associated with improved outcomes and economic benefit and vice versa.^13,17,23 Surgeons performing ≥10 shoulder arthroplasties annually demonstrate a 27% reduction in the odds of revision, with a 33% reduction observed among those performing ≥29 cases compared with surgeons performing fewer than 4 cases per year.⁴ Further evidence confirms that surgeons performing fewer than 10 cases annually have significantly higher revision rates in the early postoperative period for both anatomic and reverse shoulder arthroplasty.⁵ Similarly, length of stay and hospital cost decreased significantly across progressively higher surgeon and hospital volume categories.²³ Collectively, these findings demonstrate that surgeon volume influences not only clinical outcomes but also economic efficiency, while highlighting the lack of consensus regarding the optimal volume thresholds that define proficiency.

Despite extensive literature demonstrating associations between surgeon volume and outcomes, clinically meaningful and reproducible thresholds remain inadequately defined. Moreover, it remains uncertain whether a surgeon volume threshold exists beyond which complication rates no longer differ meaningfully between lower- and higher-volume surgeons. Accordingly, the purpose of this study was to perform a systematic review and meta-analysis to determine shoulder arthroplasty cases/yr thresholds at which complication and revision risks diverge between low- and high-volume surgeons.

Material and methods

Search strategy

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, PubMed, Cochrane, and Google Scholar (pages 1-20) were accessed and explored until December 2025 using the following keywords “Experience,” “volume,” “shoulder,” “arthroplasty,” and “replacement” to find articles assessing the impact of surgeon's volume on complications and revisions after shoulder arthroplasty. Additional articles were identified by reviewing reference lists from relevant articles. Two reviewers determined the eligibility of the included studies independently. While one author extracted the data, another author confirmed the accuracy of the extracted data. If present, differences between the investigators were resolved by tiebreaker via a third independent reviewer throughout the search and inclusion stages (Fig. 1).

Figure 1.

PRISMA flowchart for article selection process. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

We specifically looked at articles comparing the number of complications and revisions after primary shoulder arthroplasty (hemiarthroplasty, anatomic total shoulder arthroplasty, or reverse shoulder arthroplasty) across different categories of surgeon experience. Inclusion criteria consisted of comparative English and non-English-language articles. There were no restrictions on language. Nevertheless, all of the included articles were in English. Therefore, specific handling for non-English articles was not required. Exclusion criteria consisted of articles comparing the outcomes across different categories of hospital volume, studies not reporting on complications/revisions, studies assessing outcomes for the same surgeon across the different stages of the learning curve, or studies using the same cohort of patients.^1,6, 7, 8, 9^,12,14, 15, 16^,18,22, 23, 24, 25, 26, 27^,29,31

Data extraction

The eligibility of the included studies was independently determined by 2 reviewers. Extracted data included complications and revisions. For studies that only reported revisions without complications, we considered the rate of complications is similar to the rate of revisions to increase the power in the complications analysis. To accommodate the heterogeneity of the definition of high- and low-volume surgeons (HVS/LVS) in the included studies, we used 5 different thresholds of cases/yr to perform the analysis: 5 cases/yr, 10 cases/yr, 14 cases/yr, 20 cases/yr, and 25–28 cases/yr. The selected thresholds were based on the surgeon-volume cutoffs reported in the included studies. Given the heterogeneity in how surgeon volume was defined across the literature, we grouped the most commonly represented and clinically comparable thresholds to facilitate pooled analysis.

Risk of bias assessment

The Risk Of Bias In Non-randomized Studies - of Interventions (ROBINS-I) tool was used to assess the risk of bias in the included studies by 2 authors independently.²⁸ Studies were excluded if they had a critical risk of bias.

Statistical analysis

Review Manager 5.4 (The Cochrane Collaboration, 2020) was implemented for the statistical analysis. Odds ratio (OR) was utilized for dichotomous data. Heterogeneity was evaluated by Q tests and I² statistics. If considerable heterogeneity was indicated by P< .05 or I²> 50%, a random-effects model was used. Otherwise, the fixed-effect model was implemented. Statistical significance was defined as P≤ .05.

Results

Eight retrospective articles met the inclusion criteria including 332,452 patients.^{4,5,10,11,13,}19, 20, 21

Five cases/year threshold

Characteristics of the included studies

Four retrospective articles met the inclusion criteria.^4,13,19,20 These studies included 129,582 patients, with 18,995 in the LVS group and 110,587 in the HVS group. The thresholds to define LVS in the included studies ranged from 1.6 to 5 cases/yr, while the thresholds for HVS ranged from 5 to 10 cases/yr. The main characteristics of the included studies are summarized in Table I.

Table I.

Characteristics of the included studies in the 5 cases/year threshold analysis.

Author, yr	Study design	Database and period	Threshold	Surgery	Patients		Age (yr)		Follow-up
					LV	HV	LV	HV	LV	HV
Best et al 20234	Retrospective	Centers for Medicare & Medicaid Services Database 2015-2021	LV: ≤3 cases/yr HV:>10 cases/yr	Anatomic and reverse total shoulder arthroplasty	11,812	107,890	72	73	2 yr
Hammond et al 200318	Retrospective	Maryland Health Services Cost Review Commission 1994-2000	LV: ≤1.6 cases/yr HV:>7.5 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	1393	475	68		NA
Jain et al 201325	Retrospective	Nationwide Inpatient Sample 2001-2008	LV: ≤5 cases/yr HV:>5 cases/yr	Hemiarthroplasty for proximal humerus fractures	4,361	1435	NA		NA
Kim et al 202326	Retrospective	The Commercial Claims and Encounters and the Medicare Supplemental and Coordination of Benefits Marketscan Research databases 2017-2019	LV: ≤5 cases/yr HV:>5 cases/yr	Anatomic and reverse total shoulder arthroplasty	1429	787	64	68	1 yr

LV, low volume; HV, high volume.

Outcomes

Four studies reported data on complications including 129,582 patients (LVS: 18,995 patients; HVS: 110,587 patients), and 2 studies reported data on revisions including 121,918 patients (LVS: 13,241 patients; HVS: 108,677 patients). The LVS group had a higher rate of complications (3.7% vs. 2.6%; OR = 1.41; 95% confidence interval [CI]: 1.28–1.55; P< .001, Fig. 2A) and revisions (3.5% vs. 2.6%; OR = 1.41; 95% CI: 1.28–1.57; P< .001, Fig. 2B) compared to the HVS group.

Figure 2.

(A) Forest plot showing the difference in the rate of complications for a threshold of 5 cases/yr. (B) Forest plot showing the difference in the rate of revisions for a threshold of 5 cases/yr. LVS, low-volume surgeons; HVS, high-volume surgeons; CI, confidence interval.

Ten cases/year threshold

Characteristics of the included studies

Four retrospective articles met the inclusion criteria.^4,5,19,20 These studies included 176,784 patients, with 55,140 in the LVS group and 121,644 in the HVS group. The thresholds to define LVS in the included studies ranged from 5 to 10 cases/yr, while the thresholds for HVS ranged from 10 to 15 cases/yr. The main characteristics of the included studies are summarized in Table II.

Table II.

Characteristics of the included studies in the 10 cases/year threshold analysis.

Author, yr	Study design	Database and period	Threshold	Surgery	Patients		Age (yr)		Follow-up
					LV	HV	LV	HV	LV	HV
Best et al 20234	Retrospective	Centers for Medicare & Medicaid Services Database 2015-2021	LV: ≤9 cases/yr HV:>10 cases/yr	Anatomic and reverse total shoulder arthroplasty	44,002	107,890	72	73	2 yr
Brown et al 20206	Retrospective	Australian registry 2004-2017	LV: ≤10 cases/yr HV:>10 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	5,348	13,048	NA		9 yr
Jain et al 201325	Retrospective	Nationwide Inpatient Sample 2001-2008	LV: ≤5 cases/yr HV:>15 cases/yr	Hemiarthroplasty for proximal humerus fractures	4,361	419	NA		NA
Kim et al 202326	Retrospective	The Commercial Claims and Encounters and the Medicare Supplemental and Coordination of Benefits Marketscan Research databases 2017-2019	LV: ≤5 cases/yr HV:>15 cases/yr	Anatomic and reverse total shoulder arthroplasty	1429	287	64	69	1 yr

LV, low volume; HV, high volume.

Outcomes

Four studies reported data on complications including 176,784 patients (LVS: 55,140 patients; HVS: 121,644 patients), and 2 studies reported data on revisions including 121,918 patients (LVS: 13,241 patients; HVS: 108,677 patients). There was no difference in the rate of complications (OR = 1.21; 95% CI: 0.94–1.56; P= .14, Fig. 3A) and revisions (OR = 1.18; 95% CI: 0.90–1.54; P= .22, Fig. 3B) between the 2 groups.

Figure 3.

(A) Forest plot showing the difference in the rate of complications for a threshold of 10 cases/yr. (B) Forest plot showing the difference in the rate of revisions for a threshold of 10 cases/yr. LVS, low-volume surgeons; HVS, high-volume surgeons; CI, confidence interval.

Fourteen cases/year threshold

Characteristics of the included studies

Three retrospective articles met the inclusion criteria.^4,5,10 These studies included 142,599 patients, with 56,722 in the LVS group and 85,877 in the HVS group. The thresholds to define LVS in the included studies ranged from 9 to 14 cases/yr, while the thresholds for HVS ranged from 14 to 29 cases/yr. The main characteristics of the included studies are summarized in Table III.

Table III.

Characteristics of the included studies in the 14 cases/year threshold analysis.

Author, yr	Study design	Database and period	Threshold	Surgery	Patients		Age (yr)		Follow-up
					LV	HV	LV	HV	LV	HV
Best et al 20234	Retrospective	Centers for Medicare & Medicaid Services Database 2015-2021	LV: ≤9 cases/yr HV:>29 cases/yr	Anatomic and reverse total shoulder arthroplasty	44,002	57,232	72	72	2 yr
Brown et al 20206	Retrospective	Australian registry 2004-2017	LV: ≤10 cases/yr HV:>20 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	5,348	6,619	NA		9 yr
Gill et al 202514	Retrospective	Australian registry 2017-2022	LV: ≤14 cases/yr HV:>14 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	7,372	22,026	NA		5 yr

LV, low volume; HV, high volume.

Outcomes

Three studies reported data on revisions including 142,599 patients (LVS: 56,722 patients; HVS: 85,877 patients), with no studies reporting data on complications. There was no difference in the rate of revisions between the 2 groups (OR = 1.15; 95% CI: 0.89–1.48; P= .28, Fig. 4).

Figure 4.

Forest plot showing the difference in the rate of revisions for a threshold of 14 cases/yr. LVS, low-volume surgeons; HVS, high-volume surgeons; CI, confidence interval.

Twenty cases/year threshold

Characteristics of the included studies

Three retrospective articles met the inclusion criteria.^4,5,10 These studies included 141,687 patients, with 63,151 in the LVS group and 78,536 in the HVS group. The thresholds to define LVS in the included studies ranged from 9 to 20 cases/yr, while the thresholds for HVS ranged from 20 to 29 cases/yr. The main characteristics of the included studies are summarized in Table IV.

Table IV.

Characteristics of the included studies in the 20 cases/year threshold analysis.

Author, yr	Study design	Database and period	Threshold	Surgery	Patients		Age (yr)		Follow-up
					LV	HV	LV	HV	LV	HV
Best et al 20234	Retrospective	Centers for Medicare & Medicaid Services Database 2015-2021	LV: ≤9 cases/yr HV:>29 cases/yr	Anatomic and reverse total shoulder arthroplasty	44,002	57,232	72	72	2 yr
Brown et al 20206	Retrospective	Australian registry 2004-2017	LV: ≤20 cases/yr HV:>20 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	11,777	6,619	NA		9 yr
Gill et al 202514	Retrospective	Australian registry 2017-2022	LV: ≤14 cases/yr HV:>26 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	7,372	14,685	NA		5 yr

LV, low volume; HV, high volume.

Outcomes

Three studies reported data on revisions including 141,687 patients (LVS: 63,151 patients; HVS: 78,536 patients), with no studies reporting data on complications. There was no difference in the rate of revisions between the 2 groups (OR = 1.15; 95% CI: 0.87–1.52; P= .34, Fig. 5).

Figure 5.

Forest plot showing the difference in the rate of revisions for a threshold of 20 cases/yr. LVS, low-volume surgeons; HVS, high-volume surgeons; CI, confidence interval.

Twenty-five–28 cases/year threshold

Characteristics of the included studies

Four retrospective articles met the inclusion criteria.^4,10,11,21 These studies included 304,266 patients, with 194,120 in the LVS group and 110,146 in the HVS group. The thresholds to define LVS in the included studies ranged from 25 to 28 cases/yr, while the thresholds for HVS ranged from 25 to 29 cases/yr. The main characteristics of the included studies are summarized in Table V.

Table V.

Characteristics of the included studies in the 25–28 cases/year threshold analysis.

Author, yr	Study design	Database and period	Threshold	Surgery	Patients		Age (yr)		Follow-up
					LV	HV	LV	HV	LV	HV
Best et al 20234	Retrospective	Centers for Medicare & Medicaid Services Database 2015-2021	LV: ≤28 cases/yr HV:>29 cases/yr	Anatomic and reverse total shoulder arthroplasty	94,660	57,232	72	72	2 yr
Gill et al 202514	Retrospective	Australian registry 2017-2022	LV: ≤25 cases/yr HV:>25 cases/yr	Anatomic total shoulder arthroplasty and hemiarthroplasty	14,713	14,685	NA		5 yr
Girdler et al 202315	Retrospective	Centers for Medicare & Medicaid Services Database 2016-2018	LV: ≤28 cases/yr HV:>28 cases/yr	Reverse total shoulder arthroplasty	61,971	28,347	75	74	3 mo
Liu et al 202527	Retrospective	Premier Healthcare Database 2016-2019	LV: ≤25 cases/yr HV:>25 cases/yr	Anatomic total shoulder arthroplasty	22,776	9,882	67	67	2 yr

LV, low volume; HV, high volume.

Outcomes

Three studies reported data on complications including 152,374 patients (LVS: 99,460 patients; HVS: 52,914 patients), and 3 studies reported data on revisions including 213,948 patients (LVS: 132,149 patients; HVS: 81,799 patients). The LVS group had a higher rate of complications (1.7% vs. 1.5%; OR = 1.39; 95% CI: 1.11–1.75; P= .004, Fig. 6A) and revisions (2.5% vs. 2.2%; OR = 1.22; 95% CI: 1.15–1.30; P< .001, Fig. 6B) compared to the HVS group.

Figure 6.

(A) Forest plot showing the difference in the rate of complications for a threshold of 25–28 cases/yr. (B) Forest plot showing the difference in the rate of revisions for a threshold of 25–28 cases/yr. LVS, low-volume surgeons; HVS, high-volume surgeons; CI, confidence interval.

Discussion

Several studies have established the link between surgeon volume and experience and outcomes after shoulder arthroplasty.^4,5,19,20 However, most of the studies do not use the same thresholds to define separate high-, medium-, and low-volume surgeons. Therefore, in our meta-analysis, we were able to establish 5 cases/yr and 25–28 cases/yr as 2 thresholds which may possibly define transition points from low-to-medium and medium-to-high-volume thresholds since patients operated on by surgeons performing less than these thresholds had higher risks of getting a complication or revision after shoulder arthroplasty compared to surgeons performing more cases than the defined thresholds.

By performing a stratum-specific likelihood ratio analysis, Ramkumar et al reported that shoulder arthroplasty surgeons could be stratified into low (0-4), medium (5-14), and high (15 or more) volume in terms of both length of stay and costs.²³ In our study, we were able to define different categories as well in terms of complications and reoperations using a different statistical method. Our findings show that the relationship between volume and outcomes is nonlinear with 2 distinct inflection points. Performing more than 5 cases/yr could be the learning curve that is required to achieve competency in shoulder arthroplasty, followed by a plateau phase between 5 and 25 cases/yr. However, beyond 25–28 cases/yr, the second inflection point, surgeons may become more specialized (ie, high volume) and may be able to perform shoulder arthroplasty with further reduction in complications and revisions, as they acquire experience and refine surgical judgment in this procedure.

Establishing a threshold of cases/yr where adverse events are reduced is important for surgeons and patients alike. Community surgeons can use this threshold to decide whether or not to do the surgery themselves or refer it to a specialist in order to provide the best possible care and outcomes for the patients. Furthermore, these thresholds could be used by residency and fellowship programs as benchmarks for trainee education. In addition, patients could use these criteria when selecting a surgeon to perform their shoulder arthroplasty. However, volume is only one of several other considerations that should be taken into account when choosing a surgeon or center. While a cost analysis was not performed, our results may have important economic implications. Shoulder arthroplasty accounts for more than $1.8 billion in costs to US hospitals/yr, with the rate of both primary and revision arthroplasty increasing.³ Thus, the reduction of revision rate beyond our defined thresholds may support programs such as employer healthcare plans that partner with high-volume centers of excellence.⁴ Moreover, variations in bundled payments according to surgeon volume and outcomes could be an incentive for higher-volume surgeons or a way to reduce costs.⁴

To our knowledge, this is the first meta-analysis to identify surgeon volume thresholds associated with lower adverse event rates after shoulder arthroplasty. This study offers a sizable patient cohort from several settings, facilitating the generalization of the results and directing preoperative expectations and conversations for both the surgeon and the patient. Nevertheless, this study has several limitations. As a systematic review, the results of this meta-analysis are pooled. Therefore, granular data such as implant type and intraoperative variables that might affect outcomes were not accounted for. Furthermore, due to the low number of included studies, we were not able to perform a subanalysis by indication or by type of implant, which can affect the incidence of complications. In fact, most included studies did not differentiate outcomes between anatomic and reverse total shoulder arthroplasty, precluding the identification of procedure-specific volume thresholds. Additionally, complications were frequently reported as aggregate outcomes without detailed classification, limiting our ability to determine which specific complications are most influenced by surgeon volume. Another limitation relates to the inclusion of studies conducted over a wide time span, including older studies in which surgical techniques, implant design, and perioperative management differ from contemporary practice. While these temporal differences may affect the absolute rates of complications and revisions, the relative association between surgeon volume and outcomes is likely to remain consistent. Nevertheless, the findings should be interpreted in the context of evolving surgical practice. Despite these drawbacks, this study offers a sizable patient cohort from several settings, facilitating the generalization of the results and directing preoperative expectations and conversations for both the surgeon and the patient.

Conclusion

This study identified surgeon-volume thresholds for shoulder arthroplasty of 5 cases/yr and 25–28 cases/yr, which may possibly define low-to-medium and medium-to-high-volume thresholds, respectively. Patients treated by surgeons below these thresholds had higher risks of complications and revision compared with those treated by surgeons above the respective thresholds. Future research should validate our thresholds in terms of functional outcomes and operative metrics in order to optimize patient outcomes.

Disclaimers:

Funding: No funding was disclosed by the authors.

Conflicts of interest: Scott P. Steinmann would like to disclose the following: Board of Directors or committee member, self: AAOS Fellowship Committee Chair; other professional activities: Stryker Corporation, Zimmer Biomet Holdings, Inc., Elsevier, and Sonex Health, Inc.

Joseph A. Abboud would like to disclose royalties from Osteocentric Technologies, Enovis, Zimmer Biomet, Stryker, and Globus Medical; has stock or stock options held in Shoulder Jam, Aevumed, Oberd, OTS Medical, Orthobullets, Atreon, and Restore 3D; has received research or institutional support from Enovis and Arthrex; serves as a board member, owner, officer, or committee member of American Shoulder and Elbow Society, MID Atlantic Shoulder and Elbow Society, Shoulder 360, and Pacira.

Hafiz F. Kassam would like to disclose consulting fees from Zimmer Biomet.

Any additional authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.