Abstract
Purpose
Femoral shaft fractures are routinely treated using antegrade intramedullary nailing under fluoroscopic guidance. Malreduction is common and can be due to multiple factors. Correct entry point identification can help minimize malreduction and the risk of iatrogenic fracture. This study aims to compare landmark identification used to guide nail entry, the piriformis fossa (PF) and the trochanteric tip (T), via computer navigation and conventional fluoroscopy.
Method
The location of the PF and T were digitized under direct visualization with a three-dimensional scribe on ten, fresh-frozen cadaveric right femora (two male, eight female) by three fellowship trained orthopaedic surgeons. To estimate inter- and intraobserver reliability of the direct measurements, an intraclass correlation coefficient was calculated with a minimum of two weeks between measurements. Under navigation, each specimen was draped and antero-posterior (AP) and lateral radiographs of the proximal femur were taken with a c-arm and image intensifier. The c-arm was positioned in a neutral position (0 for AP, 90 for lateral) and rotated in 5 increments, yielding a range of acceptable images. Images, in increments of 5, within the AP range (with a neutral lateral) were loaded into a navigation system (Stryker, MI). A single surgeon digitized the T and PF directly based on conventional fluoroscopy, and again directed by navigation, yielding two measurements per entry point per specimen. This was repeated for the lateral range. Hierarchical linear modelling and a Wilcox rank test were used to determine differences in accuracy and precision, respectively, in the identification of PF and T using computer navigation vs. conventional fluoroscopy.
Results
The average range of suitable images for both the AP and lateral images was 29 (range of 25 to 30). The location of the PF and T was found to be reliable for a single observer (0.98 and 0.99) and between observers (0.96 and 0.93). Similar accuracy was found in identifying PF under navigation and fluoroscopy (0.05 to 1.4 cm and 0.1 to 1.5 cm respectively, p = 0.26), whereas improved accuracy was found for T using fluoroscopy (0.07 to 2.5 cm) as compared to navigation (0.2 to 2.2 cm, p < 0.001). For both the PF and T, the navigation-based points had greater precision than those selected by fluoroscopy alone (p = 0.001 and p = 0.024).
Conclusion
The ideal entry point, under direct visualization, was highly repeatable, indicating that the surgeons could identify their targeted point of entry for both the PF and T. However, there is an arc of approximately 30, through which acceptable AP and lateral images can be obtained. Throughout this range, the location of the PF and T can vary up to 1.5 and 2 cm, respectively. Navigation was less accurate than fluoroscopy in the T selection, yet had greater precision for both points. Thus, while navigation may decrease accuracy in selection of the T, it is more repeatable overall and equally accurate in selecting the PF.