Limited reliability of grading scapular notching according to Nerot–Sirveaux on anteroposterior radiographs

Introduction Incidences of scapular notching (SN) in reverse shoulder arthroplasty (RSA) range between 0% and 96%, and controversy remain as to its relevance for patient outcome. We assessed the reliability of scapular notching (SN) grading based on the Sirveaux classification system using anteroposterior radiographs. Materials and methods 206 RSA procedures with 5-year postoperative anteroposterior shoulder radiographs were classified independently by seven assessors according to Sirveaux (session 1). After a review meeting, three assessors re-classified the radiographs along with quality criteria (session 2). SN grading by the majority of assessors was taken as the reference. Classification interobserver reliability was analyzed using Kappa statistics. Results The incidence of SN was estimated at 53% and 37% at the first and second sessions, respectively. Interobserver reliability Kappa coefficients resulting from the first and second sessions were 0.27 and 0.43, respectively. Case selection based on radiographic quality criteria did not improve SN grading reliability in the second session. Conclusion Agreement between individual surgeons was low when grading SN in RSA according to Sirveaux using anteroposterior radiographs. Consensus among several assessors may increase reliability in research settings.