• OBJECTIVE
    • The objectives of the study were: a) to identify osteoporotic proximal humerus fractures in a large consecutive series of patients; b) to identify radiographic fracture patterns among osteoporotic and non-osteoporotic proximal humerus fractures; and c) to calculate intra- and inter-observer reliability of assessment of osteoporosis and of radiographic fracture patterns.
  • METHODS
    • This was a prospective observational study of patients admitted to the emergency department affected by a proximal humerus fracture between June 2014 and June 2016. Three researchers evaluated demographic data and comorbidities, x-rays and CT-scans. A new evaluation method for assessment of osteoporosis was proposed; 7 radiographic fracture patterns were studied. Reliabilities between intra- and inter-tester evaluations, and correlations between the presence of osteoporosis and the 7 radiologic fracture patterns were calculated.
  • RESULTS
    • Two hundred twenty-five patients with a humeral fracture were recruited. Their mean (26-95, 32) age was 58. Of those, 163 (72.4%) were identified as osteoporotic. Among the three raters, the intra- and inter-observer agreement using the proposed methods were high or excellent. Significant correlations with diagnosis of osteoporosis were found with Codman-Lego type 12(p = 0.041), metaphyseal comminution(p < 0.001), impaction of fragments(p = 0.023), comminution of tuberosities(p = 0.037), inferior subluxation(p = 0.029). Intra- and inter-tester reliability of evaluation of these osteoporotic fracture patterns were high.
  • CONCLUSIONS
    • Osteoporosis of the proximal humerus was identified in 72% of patients during a two year period; most of these patients were elderly females sustaining low energy trauma. These fractures showed to have specific radiographic patterns, as comminution of metaphysis and tuberosities, impaction of fragments, and inferior subluxation of the humeral head. These patterns can be assessed with the simple observation of a 2-plan view of a radiograph, without the use of specific software.