Learning curve for the open Latarjet procedure: a single-surgeon study

Article information

Clin Shoulder Elb. 2024;27(4):400-406

Publication date (electronic) : 2024 November 15

doi : https://doi.org/10.5397/cise.2024.00199

Aidan G. Papalia, Paul V. Romeo, Matthew G. Alben, Andrew Cecora, Dashaun Ragland, Mandeep S. Virk

Division of Shoulder and Elbow Surgery, Department of Orthopedic Surgery, NYU Grossman School of Medicine, NYU Langone Orthopedic Hospital, NYU Langone Health, New York, NY, USA

Corresponding Author: Mandeep S. Virk Division of Shoulder and Elbow Surgery, Department of Orthopedic Surgery, NYU Grossman School of Medicine, NYU Langone Orthopedic Hospital, NYU Langone Health, 246 East 20th St, New York, NY 10003, USA Tel: +1-646-356-9408 Email: mandeep.virk@nyulangone.org

Received 2024 March 1; Revised 2024 May 1; Accepted 2024 September 25.

Abstract

INTRODUCTION

Anterior shoulder instability is a common type of shoulder instability, with an incidence of 0.08 per 1,000 person-years in the United States and greater prevalence in young active males, collision sports athletes, and active military personnel [1-3]. The Latarjet procedure (LP), a partial coracoid transplantation to the anterior glenoid, first described in 1954 by Dr. Michael Latarjet, has become a widely accepted operative modality for management of anterior shoulder instability [1,4,5]. Compared to primary Bankart repair, the LP is technically challenging and involves a higher complication rate [4,6-8]. As such, LP is often reserved for patients with significant glenoid bone loss, failed primary treatment, or participation in high-risk contact sports [6-9]. Although arthroscopic LP has gained traction over recent years, open LP (oLP) has remained the predominant treatment modality [5,10].

When learning a new procedure, performance typically improves with increasing experience [11-16]. Plotting improvements in performance against increasing experience yields a learning curve [14]. The application of learning curves in surgical procedures has demonstrated a predictable relationship among number of procedures performed, operative time, and complication rate [17]. Learning curves classically demonstrate four stages: (1) a rapid improvement in performance; (2) a period of diminishing returns; (3) a plateau phase; and (4) an age-related decline in performance [5,14].

Knowledge of a procedure learning curve offers valuable contributions to clinical practice. This was first demonstrated by the General Medical Council in the United Kingdom, which found that the early phase of the learning curve for surgical procedures is associated with higher rates of complications and adverse patient outcomes [5,18]. Learning curves provide useful insight for surgeons during the integration of new procedures [14]. While learning curves have been established for arthroscopic LP, there remains a paucity of literature investigating learning curves for oLP [5,19-24]. Although it is difficult to determine a “proficiency value,” past studies have used the lowest number of procedures that yields a significant difference in operative time [20].

The purpose of this study is to determine the learning curve for the oLP and establish the threshold for proficiency of this procedure. Based on prior orthopedic and surgical studies investigating learning curves, such as Ekhtiari et al. [5], we hypothesize that the operative time will plateau after 30 or fewer procedures by a single surgeon, which is reflective of proficiency in oLP.

METHODS

RESULTS

Patient Demographics

Seventy-five oLPs conducted by the senior author met the inclusion criteria for this study. The patients were predominantly male (68 males [90.7%] vs. 7 females [9.3%]) with an average age of 28.7±8.5 years. A total of 37 procedures was performed on the left shoulder and 38 on the right. The majority of the procedures was primary oLP (61, 81.3%) for instability, while 14 (18.7%) were salvage oLPs for failed Bankart repair. A complete list of patient demographics is presented in Table 1.

Table 1.

Patient Demographic

Operative Time

The average operative time across all 75 procedures was 96.1±17.6 minutes. Fig. 1 shows the trend in operative time by case number. After 75 procedures, there was a total decrease in average operative time of 31.5 minutes. The average operative times of the chronological groups of 15 procedures decreased from 115.9±15.1 minutes in the first 15 cases to 98.5±14.8 minutes for cases 16–30, 90.3±12.7 minutes for cases 31–45, 91.5±13.8 minutes for cases 46–60, and finally to 84.4±14.2 minutes for cases 61–75 (Table 2). One-way analysis of variance demonstrated a significant difference in operative times between any two groups (P<0.001). Subsequent post hoc analysis demonstrated a decrease in operative time between cases 1–15 and 16–30 (115.9 vs. 98.5 minutes, P=0.0053). There were no significant differences in operative times between any other pair of groups. However, compared to the initial average operative time, there was a significant decrease in operative time for each group (P<0.05). Fig. 2 demonstrates the average operative times for procedure groups. The decline in operative times begins to level off after 30 procedures.

Fig. 1.

Operative time by case number.

Table 2.

Operative Time by Case Number

Fig. 2.

Average operative time by case number.

DISCUSSION

The primary finding of this study was the learning curve for oLP. There was a considerable reduction in operative time after 15 procedures (P=0.005) and a continuing trend in decreasing operative time (P<0.05). Specifically, though operative times continued to decrease with increased experience, the learning curve plateaued after 30 procedures. The findings of this study largely corroborate those of other reports, demonstrating a steep learning curve similar to that observed in other studies on surgery of the glenohumeral joint [5,20,21,26-29].

The learning curve can be a measure of surgical process (operative time) or surgical outcome (complication, radiographic outcomes like arthritis, or recurrence rate). We established a learning curve for operative time, which is an objective measurement that has been used in prior studies to determine a learning curve (including that of arthroscopic LP) [20,21,26,29]. Additionally, we reviewed the complication and recurrence rates [27,28].

Learning curves for complex or new surgical procedures have important clinical implications. They allow surgeons who are adapting or learning a new or established complicated procedure to anticipate the number of procedures before achieving proficiency and the risks their patients may be exposed to during the learning phase. During the early phase (cases 1–15) of the learning curve for oLP, there was rapid improvement in efficiency. Improvements in this period correlate with the rapid acquisition of procedure-specific skills and implementation of workflows that improve overall efficiency. After approximately 30 cases, the learning curve began to plateau, which was consistent with our hypothesis based on prior published studies of learning curves of open orthopedic procedures [5]. During this plateau period, minor adjustments in technique driven by surgeon experience continue to contribute to small improvements in overall efficiency. The plateau phase of the learning curve demonstrates optimal surgeon efficiency. Therefore, the duration of the preceding early phase of the learning curve can be a useful guide to determine surgeon progress toward competency in this procedure. However, physician experience and skill alone are not the only variables contributing to improvements in learning curves. Learning curves likely represent a composite measure of improvements in multidisciplinary team performance [14]. Therefore, comparisons of physician performance against established learning curves should be used in conjunction with other performance metrics.

Bishai et al. [20] conducted a retrospective study using data from a single surgeon to analyze operative times for arthroscopic LP. They reported similar findings of a significant decrease in operative time after 25 procedures, with an overall complication rate of 6%. Finally, they determined that proficiency is influenced by both the number of procedures and the time between these procedures [20]. Bonnevialle et al. [26] conducted a prospective nonrandomized study that analyzed operative complications and determined an operative time learning curve. They found intraoperative and immediate postoperative complication rates for arthroscopic LP with double button fixation of approximately 10%. Additionally, their analysis of the learning curve data showed operative time and bone block position to significantly improve with surgical experience [26].

Cunningham et al. [21] conducted a prospective study to compare learning curves for arthroscopic and open LPs. They found that surgical time was significantly longer in the arthroscopic group compared to the open group. The authors emphasized the importance of this data when selecting patients to undergo arthroscopic procedures because providers must be aware of the risk of cerebral ischemia with prolonged operative times and prolonged hypotension. Cunningham et al. [21] determined that it took approximately 10 arthroscopic procedures to overcome the need for conversion and 20 arthroscopic procedures to have comparable operative times to the open technique. Regarding complications, the study concluded that arthroscopic LPs involved more numerous complications compared to open procedures. Common complications included graft non-union, material migration, and recurrences. Additionally, in the arthroscopic group, there was the benefit of no perforation of the glenoid articular surface. This demonstrates the clear benefit of the direct visualization of the articular surface with this technique. Conversely, screw angulation was more accurate in the open group [30].

Ekhtiari et al. [5] conducted a systematic review to determine the learning curves associated with LPs. They found that surgeons became proficient after 22 open procedures, whereas it took 20 to 40 arthroscopic procedures for surgeons to become proficient. They determined proficiency as a decrease in operative time on the learning curve. However, based on the small amount of evidence available, they were unable to establish a true learning curve for the procedure. Complications were few and did not vary between arthroscopic and open LPs. Finally, those authors mentioned that future studies should define operative time, as was stated in the present study [5].

There are limitations inherent to our study. As with all retrospective studies, ours is dependent on accurate record-keeping by the surgical team. Although future prospective studies should control for and analyze factors such as a consistent surgical team, our study provides the foundation for understanding the learning curve of oLP. As our study was based on the surgical times of a single surgeon with a relatively homogeneous patient population, there is limited external validity to our findings. Additionally, future studies could perform a subgroup analysis to determine if there is a difference in learning curve between a Latarjet for primary bone loss and a Latarjet for failed arthroscopic Bankart repair. Most patients in this study underwent an LP for primary bone loss, although approximately 19% were indicated for failed Bankart repair. Our study did not have adequate numbers to study the differences in learning curve for these two aforementioned indications of latarjet. Despite this, our patient population mirrored that of the typical patient who requires surgical intervention for recurrent shoulder issues [5,20,21,26,31-33].

CONCLUSIONS

Learning curves are important tools for monitoring the performance of a surgeon integrating new procedures into their practice. Concerning the oLP, there is a steep learning curve, with rapid improvement in operative time after 15 cases that plateaued after 30 cases, which indicates proficiency in this procedure.

Article information Continued