gms | German Medical Science

Objectives: The best treatment of comminuted fractures of the proximal humerus (three-part, four-part, and split head fractures) in the elderly is an unresolved clinical problem. Case series and a small sample randomized study suggest that hemiarthroplasty may be superior to nonoperative treatment, but the scientific level of evidence is inadequate. The aim of this study was to compare the outcome of hemiarthroplasty and nonoperative treatment in elderly patients with a comminuted proximal humeral fracture.

Methods: This international randomized controlled trial enrolled patients aged 65 years or older with a comminuted proximal humeral fracture. Patients were randomized to hemiarthroplasty (HA) or nonoperative treatment and were followed for two years. Outcome measures included the Constant-Murley score (primary outcome), Disabilities of the Arm, Shoulder, and Hand (DASH), pain (Visual Analog Score, VAS), quality of life (Short Form-36 (SF-36) and EuroQoL-5D-3L (EQ-5D)), and revision operation.

Results: Between October 6, 2009 and April 26, 2017, 30 patients were assigned to hemiarthroplasty and 28 to nonoperative treatment. Patients had a median age of 77 years, and 89% was female. According to the Hertel classification, most fractures were type 12 (41%) or 9 (27%). The median Constant-Murley score increased from 22 (P25-P75 14-32) at six weeks to 48 (32-63) at 24 months in the HA group, and from 26 (19-39) to 67 (53-72) in the nonoperative group. Only at 24 months, this was significantly different in favor of the nonoperative group (p=0.020). None of the follow-up moments showed a significant difference in the DASH score between the two groups. At 24 months, DASH scores were 20.3 (5.2-40.0) and 10.0 (4.2-32.5) in the HA and nonoperative group, respectively. Although pain levels did not differ significantly between the groups, a 1-point lower VAS score is reported by nonoperatively treated patients from six weeks onwards. SF-36 and EQ-5D were similar in both groups throughout follow-up. Currently, four revision surgeries have been performed; luxations required conversion to a reversed shoulder prosthesis in two patients in the HA and one in the nonoperative group. A third patient in the HA group required an ulnar nerve release.

Conclusion: The data of this RCT do not confirm superiority of either one of the treatments compared. The only difference found was asignificantly higher Constant-Murley score in the nonoperative group at 24 months. Three out of four revision surgery were done in the HA group. Based on these results, hemiarthroplasty is not indicated for all comminuted proximal humeral fractures in the elderly.