Clinical Orthopaedics and Related Research ®

A Publication of The Association of Bone and Joint Surgeons ®

Does Postoperative Glenoid Retroversion Affect the 2-Year Clinical and Radiographic Outcomes for Total Shoulder Arthroplasty?

Benjamin C. Service MD, Jason E. Hsu MD, Jeremy S. Somerson MD, Stacy M. Russ BA, Frederick A. Matsen MD



While glenoid retroversion and posterior humeral head decentering are common preoperative features of severely arthritic glenohumeral joints, the relationship of postoperative glenoid component retroversion to the clinical results of total shoulder arthroplasty (TSA) is unclear. Studies have indicated concern for inferior outcomes when glenoid components are inserted in 15° or more retroversion.


In a population of patients undergoing TSA in whom no specific efforts were made to change the version of the glenoid, we asked whether at 2 years after surgery patients having glenoid components implanted in 15° or greater retroversion had (1) less improvement in the Simple Shoulder Test (SST) score and lower SST scores; (2) higher percentages of central peg lucency, higher Lazarus radiolucency grades, higher mean percentages of posterior decentering, and more frequent central peg perforation; or (3) a greater percentage having revision for glenoid component failure compared with patients with glenoid components implanted in less than 15° retroversion.


Between August 24, 2010 and October 22, 2013, information for 201 TSAs performed using a standard all-polyethylene pegged glenoid component were entered in a longitudinally maintained database. Of these, 171 (85%) patients had SST scores preoperatively and between 18 and 36 months after surgery. Ninety-three of these patients had preoperative radiographs in the database and immediate postoperative radiographs and postoperative radiographs taken in a range of 18 to 30 months after surgery. Twenty-two patients had radiographs that were inadequate for measurement at the preoperative, immediate postoperative, or latest followup time so that they could not be included. These excluded patients did not have substantially different mean age, sex distribution, time of followup, distribution of diagnoses, American Society of Anesthesiologists class, alcohol use, smoking history, BMI, or history of prior surgery from those included in the analysis. Preoperative retroversion measurements were available for 11 (11 shoulders) of the 22 excluded patients. For these 11 shoulders, the mean (± SD) retroversion was 15.8° ± 14.6°, five had less than 15°, and six had more than 15° retroversion. We analyzed the remaining 71 TSAs, comparing the 21 in which the glenoid component was implanted in 15° or greater retroversion (mean ± SD, 20.7° ± 5.3°) with the 50 in which it was implanted in less than 15° retroversion (mean ± SD, 5.7° ± 6.9°). At the 2-year followup (mean ± SD, 2.5 ± 0.6 years; range, 18–36 months), we determined the latest SST scores and preoperative to postoperative improvement in SST scores, the percentage of maximal possible improvement, glenoid component radiolucencies, posterior humeral head decentering, and percentages of shoulders having revision surgery. Radiographic measurements were performed by three orthopaedic surgeons who were not involved in the care of these patients. The primary study endpoint was the preoperative to postoperative improvement in the SST score.


With the numbers available, the mean (± SD) improvement in the SST (6.7 ± 3.6; from 2.6 ± 2.6 to 9.3 ± 2.9) for the retroverted group was not inferior to that for the nonretroverted group (5.8 ± 3.6; from 3.7 ± 2.5 to 9.4 ± 3.0). The mean difference in improvement between the two groups was 0.9 (95% CI, − 2.5 to 0.7; p = 0.412). The percent of maximal possible improvement (%MPI) for the retroverted glenoids (70% ± 31%) was not inferior to that for the nonretroverted glenoids (67% ± 44%). The mean difference between the two groups was 3% (95% CI, − 18% to 12%; p = 0.857). The 2-year SST scores for the retroverted (9.3 ± 2.9) and the nonretroverted glenoid groups (9.4 ± 3.0) were similar (mean difference, 0.2; 95% CI, − 1.1 to 1.4; p = 0.697). No patient in either group reported symptoms of subluxation or dislocation. With the numbers available, the radiographic results for the retroverted glenoid group were similar to those for the nonretroverted group with respect to central peg lucency (four of 21 [19%] versus six of 50 [12%]; p = 0.436; odds ratio, 1.7; 95% CI, 0.4–6.9), average Lazarus radiolucency scores (0.5 versus 0.7, Mann-Whitney U p value = 0.873; Wilcoxon rank sum test W = 512, p value = 0.836), and the mean percentage of posterior humeral head decentering (3.4% ± 5.5% versus 1.6% ± 6.0%; p = 0.223). With the numbers available, the percentage of patients with retroverted glenoids undergoing revision (0 of 21 [0%]) was not inferior to the percentage of those with nonretroverted glenoids (three of 50; [6%]; p = 0.251).


In this small series of TSAs, postoperative glenoid retroversion was not associated with inferior clinical results at 2 years after surgery. This suggests that it may be possible to effectively manage arthritic glenohumeral joints without specific attempts to modify glenoid version. Larger, longer-term studies will be necessary to further explore the results of this approach.

Level of Evidence

Level III, therapeutic study.

Back to top