If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Despite the rising incidence of anatomic total shoulder arthroplasty (ATSA) and reverse total shoulder arthroplasty (RTSA) among surgeons, little is known about the learning curve associated with these procedures. The purpose of this systematic review was to (1) identify the learning curves associated with ATSA and RTSA, (2) evaluate the effect of the learning curves on clinical outcomes, and (3) determine the number of cases needed to achieve proficiency.
Materials and Methods
Four online databases [PubMed (NLM), MEDLINE (OVID), Cochrane Library (Wiley), and Scopus (Elsevier)] were systematically searched and screened according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses [PRISMA] guidelines. The search included results from the inception of each database to May 18, 2022. Data regarding study characteristics, patient demographics, learning curve analyses, patient reported outcome measures (PROMs), range of motion (ROM), complication rates, and reoperation rates were collected. A quality assessment for each article was performed according to the Methodological Index for Non-Randomized Studies (MINORS) criteria.
Results
A total of 13 studies of fair to good quality were included for analysis (one of level II evidence, five of level III, and seven of level IV) with the majority originating from the United States [n=8, 61.5%]. Overall, there were a total of 3,381 cases (1,861 RTSA and 1,520 ATSA), with a mean patient age of 72.6 years [range: 45-92 years]. From the studies analyzed in this systematic review, for RTSA, the approximate average number of cases surgeons need to perform to move to an acceptable position on the RTSA learning curve is 25 cases. For ATSA, a wider range of 16-86 cases was derived as only two studies reported on ATSA.
Conclusion
Progression along the learning curve for reverse and anatomic total shoulder arthroplasty results in decreased operative times, improved patient reported outcomes, and fewer complications. However, a true learning curve is difficult to quantify given the heterogeneity of reported outcome measures, individual surgeon experience at time of data collection, and statistical analyses used across studies.
The incidence of anatomic total shoulder arthroplasty (ATSA) and reverse total shoulder arthroplasty (RTSA) in the United States have risen dramatically over the past decade.
Similarly, the incidence of shoulder arthroplasty among early-career surgeons within the first two years of independent practice has increased substantially over the past decade.
This variability suggests a lack of reproducibility among surgeons and presents an opportunity to characterize the learning curve for these procedures.
The surgical learning curve was first coined by Luft et al in 1979.
: (1) a rapid ascent in a measured outcome at the onset of training; (2) a zone of diminishing returns, in which further experience only confers marginal improvements in the outcome; (3) a plateau, in which further experience has no additional benefit on the measured outcome; and (4) an age-related decline in the measured outcome. This can be depicted graphically by plotting an outcome (e.g., operative time, complication rate, re-operation rate, clinical outcome, etc.) against the number of procedures performed over time. The graph can then be analyzed to find a case number or time point in which performance plateaus, which represents the end of the learning period
. Learning curves have important clinical implications as procedural experience has been correlated with cost-effectiveness, improved clinical outcomes, and patient safety.
. The purpose of this systematic review was to (1) identify the reported learning curves associated with ATSA and RTSA (2) evaluate the effect of the stated learning curves on outcomes such as complication rates, operative time, reoperation rates, patient reported outcome measures (PROMs), and range of motion (ROM), and (3) determine a point on the learning curve after which a surgeon can be considered to have achieved proficiency. We hypothesize that with increased surgeon experience, there will be fewer complications and reoperations, improved patient outcomes, and reduced operative time.
Materials and Methods
Search Strategy
Two reviewers conducted a systematic search strategy of the online databases PubMed (NLM), MEDLINE (OVID), Cochrane Library (Wiley), and Scopus (Elsevier) in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis [PRISMA] guidelines. The search focused on literature discussing the learning curve of ATSA and RTSA. The results of the search from the inception of each database to May 18, 2022 were considered. Inclusion criteria were: (1) all levels of evidence, (2) studies performed on human patients, (3) operative studies using primary or revision RTSA and/or ATSA with or without the use of navigation (4) a formal discussion or analysis of the learning curve based on the results of the study. Exclusion criteria were: (1) review articles, opinion pieces, editorials, or basic science studies and (2) multiple studies reporting on the same group of patients (only the most recent study was included).
An initial limited search of PubMed and Scopus was conducted, followed by an analysis of keywords contained in the title and abstract and indexed terms using MESH. Once applicable keywords were identified, they were used to conduct thorough searches of each database. The search was comprised of terms to identify articles evaluating the learning curves and outcomes associated with anatomic and reverse shoulder arthroplasty (Appendix 1). A second search was conducted using keywords to measure the outcomes and complications related to the surgeon's experience. The references of included studies were checked to identify potentially eligible studies that were missed with the initial search.
Screening of Studies
Titles, abstracts, and full texts of all search results were screened in duplicate by two independent reviewers. Any disagreements at the title and abstract stages resulted in automatic inclusion for the next stage of screening. Disagreements at the full-text stage were resolved through a consensus decision between the reviewers and the senior author.
Quality Assessment of Included Studies
All included studies were assessed for quality using the Methodological Index for Non-Randomized Studies (MINORS) score. The MINORS score is a validated tool consisting of 12 items, each scored 0, 1, or 2. The maximum score is 16 for non-comparative studies and 24 for comparative studies.
The same two reviewers independently abstracted data from the full texts of included studies. Extracted data were entered into an electronic database [Excel; Microsoft]. Data regarding study characteristics, patient demographics, learning curve analyses, PROMs, ROM, complication rates, and reoperation rates were collected.
Statistical Analysis
In an attempt to optimize inter-reviewer agreement throughout the screening process, the Cohen kappa statistic (k) was calculated at each stage. Substantial agreement was defined as k>0.60, moderate agreement was defined as 0.21<k<0.60, and slight agreement was defined as k<0.21, according to recent literature.
Inter-rater agreement was evaluated by calculating an intraclass correlation coefficient (ICC) for MINORS scores. It was decided a priori to categorize the quality of the evidence using the MINORS score as previously described
Of the 3,591 studies initially identified, 13 were included in the final analysis (Figure 1). There were seven case series of level IV evidence, five retrospective cohort studies of level III evidence, and one prospective cohort study or level II evidence. There was moderate to high agreement among reviewers at the title [k= 0.74, 95% CI, 0.7-0.9], abstract [k= 0.75, 95% CI, 0.7-0.97], and full-text screening stages [k=1.00, 95% CI, 1.00-1.00].
Figure 1PRISMA Literature Search and Inclusion Results
The mean MINORS score for non-comparative studies (n=7) was 12.4 +/- 1.4 [out of a possible 16], whereas that of the comparative studies (n=6) was 19.5 +/- 2.6, [out of a possible 24]. This indicates that the non-comparative studies were of fair quality while comparative studies were of good quality. Inter-rater agreement was high as evidence by an ICC of 0.95 (CI 0.5-0.99).
Patient Demographics
Overall, a total of 3,381 cases were performed (1,861 RTSA and 1,520 ATSA) across included studies (Table I). The mean age of included patients was 72.6 years [range: 45-92 years]. The included studies were conducted in the United States of America (eight studies), South Korea (two studies), the United Kingdom (one study), Australia (one study), and the Netherlands (one study).
Key: ATSA=anatomic total shoulder arthroplasty; RTSA=reverse total shoulder arthroplasty; NR=not reported; ORT=operative time; CR=complication rate; RO=reoperation rate; PROMs=patient reported outcome measures; CUSUM=cumulative sum plots; J Orthop Surg Res=Journal of Orthopaedic Surgery and Research; Ann R Coll Surg Engl=Annals of the Royal College of Surgeons of England; JSES=Journal of Shoulder and Elbow Surgery; Clin Orthop Surg=Clinics in Orthopedic Surgery; JSES Int=Journal of Shoulder and Elbow Surgery International; Clin Orthop Relat Res=Clinical Orthopaedics and Related Research; Shoulder Elbow=Journal of Shoulder and Elbow Surgery; Am J Orthop=American Journal of Orthopedics
reported data on consecutive patients and presented a true “learning curve”, while the remaining seven studies divided patients into two groups (“early” group versus “late” group) of consecutive patients
performed a retrospective cohort study that investigated eight surgeons, four of whom were considered “early-career” (i.e., <7 years post-fellowship) and four that were considered “late-career” (i.e., >10 years post-fellowship).
performed a retrospective cohort following two separate surgeons in the same institution; the authors did not comment on years of experience of each surgeon.
Complications < 3 months post-operatively (early) v. complications between 3-12 months post-operatively (late) for 100 patients in ATSA and RTSA groups
Learning Curve Analysis of Reverse Total Shoulder Arthroplasty
Complication Rates
Complications were defined as dislocations, component loosening, painful hardware, fractures of the humerus, acromion, or scapula, nerve or vascular injury, or infection. Ten studies
noted a complication rate of 15.7% (6/38), where two out of the six complications occurred intra-operatively in the first 20 shoulders, and the other four occurred two months postoperatively, with learning curve plateau noted at 20 cases. Finally, Hasan et al
reported a 24.6% complication rate in the first 60 consecutive RTSA cases performed by one surgeon and suggested the learning curve spans the first 15-20 cases.
found a statistically significant increase in intraoperative complication rate amongst the first cohort compared to the second cohort (p=0.025), proposing the learning curve is seven RTSA cases. In this study
, the authors noted that after adjusting for age group and gender, individuals in the “late” cohort were only 7% as likely to have an intraoperative complication as those in the “early” cohort (odds ratio = 0.07; 9% CI, 0.01-0.92; p=0.043), but found similar (p=0.361) complication rates between the first and second cohorts postoperatively. One study
reported a higher complication rate in the first 40 patients (23.1%) compared to the last 160 patients (6.5%, p=0.005), a result largely influenced by minor complications (p=0.017). Walch et al
also noted a higher complication rate in the “early” group of 38 cases (19.1%) compared to the “late” group of 24 patients (10.8%), however this finding was not statistically significant (p=0.19). Levy et al
also failed to find statistically significant differences between the “early” group (first 20 cases) and the “late” group (last 20 cases), with complication rates of 0% and 5%, respectively (p=0.96). For the six studies that specifically identified a learning curve for RTSA based on complications, the average number of cases to achieve proficiency was 22.8.
reported that RTSA had an 10% complication rate after 100 consecutive cases, but no point on the learning curve was identified.
Patient Reported Outcome Measures and Range of Motion
Patient reported outcome measures analyzed included the University of California-Los Angeles (UCLA) shoulder scores, Constant scores, American Shoulder and Elbow Surgeons (ASES) scores, and ROM (forward flexion, abduction, external rotation, and internal rotation). Five studies
that evaluated consecutive RTSA cases, both achieved PROMs and ROM considered to be acceptable (UCLA shoulder score > 27, Constant score > 56, forward flexion/abduction > 80 degrees, internal rotation > 40 degrees, and external rotation > 30 degrees; of note ASES score has no defined “acceptable” score), except for external rotation in Blaas et al
reported a decreased mean operative time from 108.6 minutes [range, 71-147] to 87.6 minutes [range, 61-121] when comparing the first 18 cases of RTSA to the last 18 cases with a significant decrease in operative time after the 15th case. Blaas et al reported an overall mean operative time of 125.5 minutes [range, 111.3-155.3] in their analysis of 50 consecutive cases with shorter operative times (compared to the overall mean) observed after the first 10-15 cases.
plotted surgical time against case number for 62 consecutive RTSA cases and a linear regression slope was calculated. The authors demonstrated a significantly negative case slope with the first 18 cases, which subsequently leveled thereafter (cases 19-62).
plotted consecutive case number versus operative time for each early versus each late career surgeons and used a two-step regression to determine the plateau point / end of the learning period. The authors reported a significant negative linear relationship (p=0.01; m=-1.1) with moderate correlation (R2=0.49) between annual case volume and operative time for early-career surgeons, but no significant linear relationship for late-career surgeons.
specifically evaluated the learning curve of a single experienced surgeon commencing intraoperative computer navigation of the glenoid implant placement in RTSA. Mean operative time was 77.3 min and regression analysis showed the “curve of best fit” was logarithmic with a downward trend (R2=0.190, p=0.038). The significant downward trend in operative time indicated navigated RTSA did not have a learning curve, however the curve began to flatten after the 8th case.
Learning Curve Analysis of Anatomic Total Shoulder Arthroplasty
Complication Rates
Complications were defined as dislocations, component loosening, painful hardware, fractures of the humerus, acromion, or scapula, nerve or vascular injury, or infection. One study
used the complication rate to define the learning curve amongst 100 consecutive ATSA procedures. A main inflection point along the learning curve was identified at 16 cases, after which a trend of overall improved performance (i.e., significantly lower complication rate) was demonstrated at 40 cases. Based off of this study
also measured overall operative time for ATSA, noting a significant negative linear relationship (p=0.02; m=-0.8) with moderate correlation (R2=0.54) between mean annual case volume and operative time in early-career surgeons, with no such significant linear relationship noted for late-career surgeons. The authors indicated breakpoints (i.e., the number of cases at which operative time significantly decreased) along the learning curve between 16 and 86 cases.
, the average number of cases to achieve proficiency on the ATSA learning cure by operative time was between 16-86 cases.
Discussion
This review attempted to define the learning curve for RTSA and ATSA. While no single ideal measure to assess progress along the learning curve has been identified, outcomes used to establish competence have been traditionally grouped into measures of patient outcomes or surgical efficiency.
Attaining Surgical Competency and Its Implications in Surgical Clinical Trial Design: A Systematic Review of the Learning Curve in Laparoscopic and Robot-Assisted Laparoscopic Colorectal Cancer Surgery.
The studies included in this review considered reduced operative time as a proxy for surgical efficiency and improvements in complication rate, reoperation rate, and PROMs to represent measures of patient clinical outcomes. We found that surgeons may achieve significantly decreased operative times, improved PROMs, and fewer complications/reoperations as his/her experience performing RTSA and ATSA procedures grows. Although we attempt to estimate the number of cases at which a surgeon achieves proficiency along the RTSA and ATSA learning curves, a true learning curve remains difficult to quantify given the heterogeneity of reported outcome measures and statistical analyses used across included studies.
Previously, studies for total knee arthroplasty (TKA) have demonstrated significant differences in PROMs between surgeons with differing case volumes,
Association of hospital and surgeon procedure volume with patient centered outcomes of total knee replacement in a population-based cohort of patients age 65 and older.
Association of hospital and surgeon procedure volume with patient centered outcomes of total knee replacement in a population-based cohort of patients age 65 and older.
which may help to explain the similar findings we found with ATSA and RTSA as surgeons gain more experience with these complex procedures. This is especially consequential as indications for both ATSA and RTSA continue to expand to more complex conditions which may have lower preoperative scores.
Given the technical challenges that come along with these procedures, we predicted increasing proficiency with increasing case volume, despite a slightly less steep learning curve.
Complication rates were among the most commonly reported outcomes used in the evaluation of surgical skill in this review. The overall complication rate for RTSA ranged from 0-25% (mean 17.02%) for early cases and from 5-25% (mean 9.45%) for late cases
did not specify the exact number of patients with complications in their “early” versus “late” cohort, however the group had an overall complication rate of 75%. For the studies reporting overall complication rates, these ranged from 7-75% (mean 23.3%)
The reason these studies did not identify a learning curve possibly include the surgeon already having passed the “inflection point” on the learning curve when the study was conducted as most surgeons were considered high volume
Based on this one study, the learning curve for ATSA was demonstrated at 40 cases, after which a trend of overall improved performance (i.e., significantly lower complication rate) was observed.
In addition, it is important to consider the lack of consistency in the classification (i.e., type and timing) of complications reported by included studies as this may affect the validity of the results. Three studies noted that the reduction of intraoperative or minor complications may disproportionately contribute to the decrease in complication rates observed for late cases.
Use of the complication rate to assess proficiency also neglects to account for varying surgical indications or case complexity of individual cases that may represent varied degrees of difficulty. In a review of complications after RTSA, Barco et al linked the expanding indications of RTSA to more complex conditions and the implementation of changing implant designs to a higher risk of complication.
Implant designs in this study were diverse, with studies reporting use of either the Grammont-style or glenoid-based lateralized implants, which each have unique complications at differing rates.
Lastly, there has been evolution of ATSA and RTSA knowledge base over time, allowing for fine tuning of the procedure in order to decrease complication rates of more recently performed surgeries as opposed to remote surgeries; all studies in this review were performed in varying years.
Furthermore, longer operative times for ATSA and RTSA have been associated with worse postoperative outcomes, including increased risk of infection and peripheral nerve injury.
Four studies reported a decrease in operative time with progression along the learning curve for RTSA and reported the learning curve between 10-36 RTSA cases, with an average of 13.9 cases.
also evaluated operative time for ATSA in addition to RTSA, reporting the number of cases at which operative time significantly decreased along the learning curve between 16 and 86 cases.
However, it should be noted that utilizing reduced operative time as an indicator for surgeon proficiency may be confound by external factors. In particular, as surgeons become comfortable with performing a procedure, they may require less assistance from more experienced surgeons or undertake more complex cases, ultimately affecting overall operative time.
In addition, further confounding may be the inclusion of time for anesthesia induction or patient positioning resulting in artificial inflation of operative time.
specifically evaluated the learning curve of a single experienced surgeon commencing intraoperative computer navigation of the glenoid implant placement in RTSA using operative time. Regression analysis demonstrated a significant downward trend in operative time of the “curve of best fit”, indicating navigated RTSA did not have a learning curve.
demonstrates technical change in operative technique of experienced surgeons later in careers may not require a learning period or significantly affect surgical proficiency. As not all papers explicitly defined after how many years a surgeon was considered “experienced,” operating surgeons may have had varying degrees of training prior to conducting their analysis of the learning curve. This heterogeneity may have confounded the derived learning curves, or lack thereof, as more “experienced” surgeons may have already plateaued on the learning curve at the time the study was conducted.
00119-5&id=gr1.jpg){kind=link}