Predicting Proximal Humerus Fracture Mechanical Complications: Are Computed Tomography Hounsfield Units the Answer?

Introduction: The purpose was to determine whether computed tomography (CT) Hounsfield units (HU) as a proxy for bone quality can predict postoperative complications following surgical treatment of proximal humerus fractures.

Methods: Sixty-six patients with 2-, 3-, or 4-part proximal humerus fractures who underwent surgical fixation at single institution and had complete radiographic data available were included. Radiographic measurements included the deltoid tuberosity index (DTI) on preoperative anterior-posterior shoulder radiographs, and the HU value from the surgical proximal humerus was determined by measuring the humeral head at the midaxial/coronal/sagittal CT image using a circle-type region of interest (≥35 mm2). Postoperative complications recorded were implant failure, development of osteonecrosis, nonunion, and acute periprosthetic fracture. Patients with and without complications were statistically compared, and binary logistic regression was performed to determine whether preoperative proximal humerus CT HU were predictive of complications.

Results: Eight patients (12.1%) developed 11 overall complications, with three patients experiencing multiple complications each. Complications included osteonecrosis (4), implant failure (5), nonunion (1), and acute periprosthetic fracture (1). No difference was observed in demographics or Neer or AO/OTA classification between those with and without complications. Patients with complications had markedly lower DTI and overall HU as well as HU in the coronal and sagittal planes. Regression analysis for average DTI demonstrated a higher DTI and had a 10 times decreased risk of complication (P = 0.040, odds ratio = -10.5, 95% confidence interval, 0.000 to 0.616). Regression analysis for average total HU also found a higher HU associated with a decreased risk of complications (P = 0.034, odds ratio = -0.020, 95% confidence interval, 0.980 to 0.962). Logistic regression analysis, including age, age-adjusted Charlson Comorbidity Index, mean DTI, and mean total HU, only found mean total HU to be notable within the model.

Discussion: CT HU may identify patients with poorer bone quality and thus help predict postoperative complications.

Level of evidence: Diagnostic Level III.