Abstract
Background
Proximal humeral fractures are common and a minority develop non-union, which can result in pain and disability. We aimed to identify the risk factors and quantify the prevalence of non-union.
Methods
A thirteen-year retrospective study of 7039 patients with proximal humeral fractures was performed and a database created. 246 patients with non-union were compared to a control group to identify risk factors. Logistic regression analyses were performed to identify significant variables obtained at presentation to predict non-union.
Results
The crude rate of non-union after any proximal humeral fracture is 3.5%, rising to 5.6% after excluding patients with isolated tuberosity fractures or those who underwent primary fixation. 179 patients with non-union were compared to a control group of 295 patients whose fractures healed. There was no significant difference between the groups and the entire study population in terms of age, gender and mechanism of injury. Univariate and multivariate logistic regression analyses identified 3 significant predictors for non-union: displacement of fracture (p<0.001); having a two-part fracture (p = 0.045); and varus angulation of the humeral head at injury (p<0.001).
Conclusions
The prevalence of non-union of proximal humeral fractures (3.2/100,000population/year) is higher than previously appreciated and a modified classification (ongoing) is needed to include the identified risk factors.
