A Systematic Review Of Learning Curves in Orthopaedic Sports Surgery

Abstract

Background: As surgeons enter practice or senior surgeons embark on new procedures, we must weigh the potential risks to our patients. Despite the concept of a "learning curve" gaining traction, there is limited guidance on the specific learning curves within orthopaedic sports medicine. The purpose of this article is to explore the ways in which learning curves are defined within orthopaedic sports medicine and report the number of cases required to overcome them.

Methods: Pubmed-MEDLINE, Scopus, and Embase databases were queried for "Learning Curves" pertaining to orthopaedic sports medicine procedures according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis guidelines. Abstracts were reviewed by two independent reviewers for inclusion and subdivided into categories including: Shoulder Arthroscopy and Reconstruction, Shoulder Arthroplasty, Hip Arthroscopy, Knee Arthroscopy and Reconstruction, and Knee Osteotomy.

Results: 4,558 articles were reviewed. 14 articles for shoulder arthroscopy and reconstructive procedures, 10 articles for shoulder arthroplasty, 17 articles for hip arthroscopy, 7 articles for knee arthroscopy and reconstructive procedures, and 2 articles for knee osteotomy were ultimately included. The net defined learning curve for each respective surgical sub-category was 22 for shoulder arthroscopy and reconstruction, 28 for shoulder arthroplasty, 71 for hip arthroscopy, 28 for knee arthroscopy and reconstruction, and 32 for knee osteotomy.

Conclusion: Surgeons should consider the synthesis of the described learning curves for shoulder, hip, and knee surgery when incorporating these procedures into their practice. Level of Evidence: II.

Keywords: learning curves; medical education; orthopedic residency; orthopedic surgery; residency; surgical education; surgical training.

Figure 1

Graphical depiction of a “steep” versus a “shallow” learning curve. A steep learning curve is associated with fewer cases to achieving a steady state of quality, which is preferable to a shallow curve.

Figure 2.

Flowchart of PRIMSA methodology for manuscript review and inclusion.

Figure 3A to 3D.

Graphical depiction of multiple methodologies for measuring a learning curve. (1A) group split method, (1B) Cumulative Sum (CUSUM) analysis, (1C) Slope plateau, (1D) Segmented Linear Regression. Arrow denotes case at which a learning curve is overcome.

Figure 4:

Number of cases described to overcome the learning curve with respect to various outcomes in shoulder arthroscopy and reconstructive procedures. A single point is reported when studies identified a point at which a learning curve plateaued. A range is reported when a study identified a learning curve between a series of consecutive groups. For example, if a difference was found between the first 30 and second 30 cases, the curve is surmised to fall in a range between 0 and 30 cases.

Figure 5:

Number of cases described to overcome the learning curve with respect to various outcomes in shoulder arthroplasty. A single point is reported when studies identified a point at which a learning curve plateaued. A range is reported when a study identified a learning curve between a series of consecutive groups. For example, if a difference was found between the first 30 and second 30 cases, the curve is surmised to fall in a range between 0 and 30 cases.

Figure 6:

Number of cases described to overcome the learning curve with respect to various outcomes in hip arthroscopy. A single point is reported when studies identified a point at which a learning curve plateaued. A range is reported when a study identified a learning curve between a series of consecutive groups. For example, if a difference was found between the first 30 and second 30 cases, the curve is surmised to fall in a range between 0 and 30 cases.

Figure 7:

Number of cases described to overcome the learning curve with respect to various outcomes in knee arthroscopy and reconstructive procedures. A single point is reported when studies identified a point at which a learning curve plateaued. A range is reported when a study identified a learning curve between a series of consecutive groups. For example, if a difference was found between the first 30 and second 30 cases, the curve is surmised to fall in a range between 0 and 30 cases.

Figure 8:

Number of cases described to overcome the learning curve with respect to various outcomes in osteotomies about the knee. A single point is reported when studies identified a point at which a learning curve plateaued. A range is reported when a study identified a learning curve between a series of consecutive groups. For example, if a difference was found between the first 30 and second 30 cases, the curve is surmised to fall in a range between 0 and 30 cases. *Denotes number of cases for overcoming learning curve for coronal correction (27), slope correction (47), and lateral hinge fracture (42).