Background: A radiological audit of the local use of the Dynamic Hip Screw in extracapsular proximal femur fractures. Study aim: to identify cases of mechanical failure and revision, to determine predictors of fixation failure.

Methods: A retrospective radiological review of 567 consecutive cases at Western Australian tertiary hospitals over a 3 year period (2002 – 2004) using the Picture Archive Computer System (PACS).

Results: Female: male ratio was 2.79: 1. Evan’s classification: 418 fractures stable (73.7%), 149 unstable (26.3%). Failure of fixation occurred in 14 cases (2.5%); ten due to hip screw cut out (1.8%) and four due to plate pull off (0.8%). All cases of cut out had a significantly higher mean tip apex distance (TAD) (31 vs 20mm, P < 0.001) and an unstable fracture configuration; 8 of 10 had a poor reduction. Bivariate logistical regression revealed TAD of 25mm or more to be most predictive of cut out; followed by mean TAD, superior anterior and inferior posterior screw placement, unstable fracture configuration and poor reduction. Unassociated factors included gender, age, American Society of Anesthesiologists’ score, plate angle and length, operation time and surgeon level. A three-variable model found TAD of 25mm or more and unstable fracture configuration to be predictive, but not poor reduction. Cases with a TAD of 25mm or more with unstable fracture configuration and a poor reduction had a 21.6% chance of cut out (8 of 29).

Conclusions: This is the first multifactorial multivariate analysis of a single implant sliding hip screw series. Compared with the literature, the rate of failure is low. Possible reasons include appropriate choice of implant for fracture type, improved performance with use of a single model of implant, and low exclusion rates due to the use of PACS.