Treatment of fractures of the femoral shaft can be associated with technical errors leading to numerous iatrogenic complications during closed intramedullary nailing of the femoral shaft [4, 5]. An inappropriate entry point for nail insertion into the proximal femur could result in fracture site comminution, proximal femur and even femoral neck fracture [4, 6].
Johnson et al  found that an entry point too anteriorly to the midline of the femur results in greater hoop stresses in the femoral cortex, and probably in femoral bursting. A biomechanical study carried out by Tencer et al  has shown that placement of the starting hole 6 mm or more anterior to the neutral axis of the femur is likely to result in consistent cracking of the proximal femur. Alho et al  also reported a 26% prevalence of proximal fragment comminution with an anterior insertion of the intramedullary nail. However, none of these studies used radiographic control to investigate the diagnosis of such lesions.
The placement of the nail more anteriorly to the posterior third of the neck in this study was complicated with a prevalence of 56% of anterior femoral splitting. All of these lesions were seen during nail insertion, at its final phase. A possible explanation could be that during reaming the flexible shaft of the reamers could easily follow the curved path of the medullary canal, as the femur is convex anteriorly. On the other hand, during nail insertion the relevant stiffness of the nail, the mismatch in curvature between the inserted nail and the medullary canal, in combination with an entry point anterior to the trochanteric fossa has possibly led to the femoral splitting. As described in an article by Egol et al , the average femoral anterior radius of curvature is 120 cm (SD ± 36) and there is a significant mismatch between the anatomical radius of curvature of the femur and most intramedullary nails, including the one used in this study. In the same study, the analysis of the radii of curvature of 8 current antegrade intramedullary nails demonstrated that they all have a greater radius of curvature ranging from 186 to 300 cm (straighter) than that of the average femur. However, this is only one factor affecting nail insertion. As previously stated the placement of the starting point slightly anteriorly to the neutral axis of the medullary canal forces the nail to travel anteriorly and thus, increased hoop-stresses cause splitting of the anterior femoral cortex [5, 7].
C-arm images or plain films were used to evaluate the specimens after fixation, as this type of radiographic analysis is commonly used in clinical practice. C-arm fluoroscopy is the preferable means of monitoring IMN intraoperatively. However, C-arm images often do not reveal subtle fractures that plain radiographs might, as the quality of the images using C-arm fluoroscopy is usually worse than that of plain radiographs .
In this study, the anterior splitting was not detectable either in radiographic or fluoroscopic examination apparently because of the vertical orientation of the fractures and the overlapping density of the bone cortex and the intramedullary nail. Imaging confirmation of the splitting by using either radiographs obtained in multiple different projections other than the anteroposterior and lateral ones, or a CT-scan study, was not performed as these means are not used intraoperatively.
The bone density of the cadaveric femora was not tested. By using pairs of intact femurs from one individual donor, any variability in the mechanical properties of the paired specimens is minimized and differences in the bone strength are not expected. Moreover, there were no simulated diaphyseal fractures, as it is generally difficult to simulate identical patterns of fractures (i.e., single, comminuted, etc.).
It should be noted that this study has several limitations. The main weakness is that this study involved intramedullary nailing of intact cadaveric femora; therefore our findings may not be replicated in the surgical practice of intramedullary nailing of femur fractures for two reasons. First, cadaveric bone stripped of soft tissue, frozen, and thawed is hard and has very limited ability of compliance compared to live bone; therefore the effects of the location of the entry point and of the curvature mismatch between the nail and the medullary canal were magnified. Second, intact femora were nailed, which again augmented stresses on the femur; in the presence of a fracture, entry point variations and curvature mismatch may be accommodated to a degree by displacement at the fracture site. However, this study used matched pairs of femurs, the identical implant by a single manufacturer, and a standardized antegrade nailing technique with the exception of the entry point; all cases of anterior splitting occurred in femora with more anteriorly located entry points, which emphasizes the importance of the location of the entry point in the worst case scenario of a hard and incompliant intact femur.
Finally it should be noted that all cases of longitudinal splitting were nondisplaced cracks undetectable fluoroscopically or radiographically. Hence, it could be argued that such a nondisplaced crack may have minimal effect on the stability of fixation and, consequently, minimal clinical relevance. However, this necessitates static locking of the nail to impart rotational and longitudinal stability to the construct . Although anterior femoral splitting has never been described, detected or led to clinical relevant complications as seen by non-existing literature, we feel that surgeons should be aware of this potential complication, especially if the entry point used is too anterior. Our findings add support to the current recommendation of static IMN for all types of femur fractures .
In conclusion, this study has drawn attention to the risk of undetectable iatrogenic splitting in the anterior cortex during antegrade intramedullary nailing of intact cadaveric femora. The clinical significance, if any, of this lesion is unknown.