*2.3. Radiological Assessments*

Radiological assessments were carried out preoperatively (Walch and Favard classification), and postoperatively at day 1, 12 weeks and yearly [14–16]. In order to assess both the humeral and glenoid components, radiographs were taken according to a standardized protocol in multiple planes (axillary, true lateral, standard anteroposterior (AP) and true AP view of glenoid with the arm in 20◦ external and internal rotation). Radiographs were assessed independently and separately by two orthopaedic surgeons (MJKS and MC) not

involved in the patient surgeries. Disagreements were referred to a third independent experienced surgeon (HC) for final decision.

The aTSR was radiologically evaluated and assessed in a standard technique for perihumeral component osteolysis with assessment of five zones, as previously described [17]. The glenoid component was assessed using the Lazarus classification quantifying radiolucent lines (RLL) between the cement surrounding the glenoid pegs and bone interface [18].

The rTSR was radiologically assessed in a standard technique for scapular notching according to the classification by Sirveaux et al. for the glenoid component [19,20]. The humeral component was assessed for RLL around the implant in seven different zones as previously described by Levigne et al. and Bell et al. [16,21,22].

#### *2.4. Implants and Postoperative Rehabilitation*

All shoulders were replaced via a standard deltopectoral approach. For the aTSR, a stemless Affinis® short humeral ceramic head component was utilized in all cases. This was articulated with a double-pegged, cemented, all-polyethylene glenoid component made of standard, not cross-linked, polyethylene (Mathys AG, Bettlach, Switzerland) in 34 cases and more recently in 11 cases with a highly cross-linked polyethylene (HXLPE)—Affinis® Glenoid Vitamys prosthesis (Mathys AG, Bettlach, Switzerland).

For rTSR, a Grammont-style humeral prosthesis was used—Aequalis Reversed II Shoulder System (Tornier SAS (Montbonnot Saint Martin, France) part of Stryker). For the glenoid, a 25 mm baseplate was utilized in all cases, together with either a 36 mm or 42 mm glenosphere made of cobalt–chromium alloy. An appropriately sized humeral liner of standard polyethylene was utilized.

Post operation, all patients wore a sling for 5 to 6 weeks. A structured physiotherapy programme was commenced on the first postoperative day.

#### *2.5. Statistical Methods*

The data were analysed using a mixture of parametric procedures (*t*-tests and general linear models, including analysis of covariance) and nonparametric procedures (Mann– Whitney test, sign test, Fisher's exact test and logistic regression), as appropriate. Analyses were conducted using either Minitab statistical software version 18 (Minitab, Inc., State College, PA, USA) or R (R Core Team) [23].
