*2.1. Patients*

A multicenter retrospective study among 9 centers was conducted. Inclusion criteria were patients suffering from PANHH operated with total shoulder arthroplasty (TSA) or hemiarthroplasty (HA), aged 60 years old or under at the time of the surgery. Exclusion criteria were post-traumatic avascular necrosis, and less than 2 years between surgery and last follow-up (FU) for clinical and radiological analysis.

One hundred and twenty-seven patients were operated between 1991 and 2015 with a mean age of 46 years old (SD 10, range 19–60): 108 HA and 19 TSA. The etiologies of PANHH were Churg and Strauss disease (*n* = 2), corticotherapy for Hodgkin lymphoma (*n* = 2), drepanocytosis (*n* = 1) and post-radiotherapy (*n* = 1). For the other 121 patients (95%), no specific etiologies were found, and osteonecrosis was therefore classified as idiopathic. Five patients had undergone conservative treatment prior to arthroplasty: micro-fractures (*n* = 3), arthroscopic suprasupinatus repair (*n* = 1) and acromioplasty (*n* = 1).

We evaluated clinical outcomes with passive and active range of motion using the Constant score [14] and Subjective shoulder value [15] (SSV). Radiographic evaluation consisted of true anteroposterior radiographs of the gleno-humeral joint using a standardized protocol during the preoperative evaluation and the last follow-up. The osteonecrosis severity was assessed with Ficat's [16] classification modified by Cruess [17] (Table 1).

**Table 1.** Pre-operative radiographs assessed by Ficat's classification modified by Cruess.

Radiographs were evaluated by a senior and a resident orthopedic surgeon. Preoperative radiographs were missing for 6 patients. Radiolucent lines (RLL) around the humeral stem and the glenoid component (of TSA) were assessed with Mole score [18]. All patients provided informed consent for their participation in this study, which had been approved by the institutional review board.
