A binary logistic regression analysis of screw penetration after proximal humeral fracture surgery

Abstract

Objective To analyze the incidence of screw penetration after surgery for proximal humeral fractures (PHF) and to explore its prognostic factors and preventive countermeasures. Methods From June 2006 to June 2012, 187 PHF patients were treated with open reduction-internal fixation with locking plates at our department. They were 78 males and 109 females, 18 to 88 years of age (average, 55.9 years). Screw penetration was identified according to the follow-up X-ray films of the patients' shoulders. In our analysis, screw penetration was taken as the dependent variable while gender, age, Neer fracture type, medial cortex fracture, bone mineral density, time interval from trauma to surgery, bone graft, reduction quality, screw number in the humeral head, and medial column support were taken as covariates. Univariate logistic regression analysis was adopted to pick up the factors with P< 0.05 which were then analyzed by multiple stepwise logistic regression. Results All the patients were followed up for a mean time of 17.8 months (from 10 to 45 months). Screw penetration happened in 27 patients (incidence, 14.4%). The univariate logistic regression analysis showed that Neer fracture type, medial cortex fracture, bone graft, reduction quality, screw number in the humeral head, and medial column support had a significant association with screw penetration (P < 0.05). The multiple stepwise logistic regression showed that there were significant differences between Neer 3-part and 2-part fractures [P = 0.028, OR =4.406,95% CI(1.176-16.515)], 4-part and 2-part fractures[P = 0.004, OR=l.799, 95% CI(l.924~31.620)] respectively. Medial cortex fracture [P = 0.008, OR = 0.287, 95% CI (0.114-0.718)] was a key factor affecting screw penetration. Conclusions Neer fracture type and medial cortex fracture may be the chief prognostic factors affecting screw penetration after surgery for PHF. The best treatment should be made on the basis of individuals to reduce the incidence of screw penetration. Key words: Shoulder fractures; Postoperative complications; Factor analysis, statistical; Regression analysis