Introduction: The three or four-part fracture of the proximal humerus remains a challenging fracture in the case of elderly patient. In this continuing prospective study we compared the outcome after implantation of a shoulder hemiprosthesis of the anatomical generation and a reversed prosthesis.

Material and Methods: 34 shoulder prosthesis were implanted in patients older than 70 between june 1996 and june 2004. All patients were evaluated by a surgeon not involved in treatment concerning activities of daily living (DASH scoring), clinical outcome (Constant-Murley Score), radiological results, and a summary of complications.

Group 1: anatomical prosthesis (Tornier): At an average follow-up of 16,5 (range 6–55) months, 13/17 patients with an average age of 78,6 years (70–95), were evaluated.

Results: concerning age, follow up and dash scoring there were no difference between two groups. In reversed group: the outcome was better (Constant-Murley: 79,5 (57,8–100) vs 57,1 (21–85) p=0.005), the average active elevation was better by 30° (p< 0.001) and the average active abduction was better by 53° (p< 0.001). 7/13 cases of glenoid notching stage 2 or 3 were noted. In anatomical group: the average active external rotation was better by 10° (p=0.01). Tubercle fixation’s failed in 6/13 cases.

Discussion: No comparative study has still been published between the two sort of implant in elderly patients. Our results of shoulder arthroplasty in acute injury to the proximal humerus with reversed prosthesis are in agreement with the preliminary results reported by Sirveau. At short follow up, reversed prosthesis allow to reach early mobilisation with best functional results. Rotation remains the key point: with a significant gain in active abduction and elevation the reversed prosthesis group do not reach a better dash scoring. Attachment fixation of the posterior rotator cuff must probably be discussed in reversed prosthesis in acute fracture.