Preventing secondary screw perforation following proximal humerus fracture after locking plate fixation: a new clinical prognostic risk stratification model.

INTRODUCTION

After locking plate (LP) fixation, secondary screw perforation (SSP) is the most common complication in proximal humerus fracture (PHF). SSP is the main cause of glenoid destruction and always leads to reoperation. This study aimed to identify independent risk parameters for SSP and establish an individualized risk prognostic model to facilitate its clinical management.

METHODS

We retrospectively reviewed the medical information of patients with PHF who underwent open reduction and internal LP fixation at one medical center (n = 289) between June 2013 and June 2021. Uni- and multivariate regression analyses identified the independent risk factors. A novel nomogram was formulated based on the final independent risk factors for predicting the risk of SSP. We performed internal validation through concordance indices (C-index) and calibration curves. To implement the clinical use of the model, we performed decision curve analyses (DCA) and risk stratification according to the optimal cutoff value.

RESULTS

A total of 232 patients who met the inclusion criteria were enrolled. The incidence of SSP was 21.98% at the last follow-up. We found that fracture type (odds ratio [OR], 3.111; 95% confidence interval [CI], 1.223-7.914; P = 0.017), postoperative neck-shaft angle (OR, 4.270; 95% CI 1.622-11.239; P = 0.003), the absence of calcar screws (OR, 3.962; 95% CI 1.753-8.955; P = 0.003), and non-medial metaphyseal support (OR,7.066; 95% CI 2.747-18.174; P = 0.000) were independent predictors of SSP. Based on these variables, we developed a nomogram that showed good discrimination (C-index = 0.815). The predicted values of the new model were in good agreement with the actual values demonstrated by the calibration curve. Furthermore, the model's DCA and risk stratification (cutoff = 140 points) showed significantly higher clinical benefits.

CONCLUSIONS

We developed and validated a visual and personalized nomogram that could predict the individual risk of SSP and provide a decision basis for surgeons to create the most optional management plan. However, future prospective and externally validated design studies are warranted to verify our model's efficacy.