Pertrochanteric, subtrochanteric and femur shaft fractures represent common injuries, with increasing incidence owing to an ageing global population. The management of such fractures usually employs the insertion of cephalomedullary (CM) nails. Despite the widespread use of this approach, the exact positioning for the entry point of the CM nail, which is critical for optimal fracture reduction and the avoidance of complications, remains a matter of debate [1].

While surgical technique guides from leading manufacturers of commonly utilized nails, such as proximal femoral nail antirotation (PFN-A) and trochanteric fixation nail advanced (TFN-A) from DePuy Synthes, AUTOBAHN™ from Globus Medical and HERACLES from 7s Medical, do offer entry point recommendations, these guidelines are often general and not definitively precise. For instance, in the anterior–posterior (AP) view, the correct entry point is usually described as on the tip or slightly lateral to the tip of the greater trochanter, factoring in the nail’s medio-lateral (ML) angle [2,3,4,5]. The lateral view, however, advises an entry point centred within the trochanter, aligning with the axis of the intramedullary canal [2,3,4,5].

Studies examining the optimal entry point have yielded variable findings. One study investigating the correct entry point for subtrochanteric fractures suggested the optimal universal entry point to be at the tip of the trochanter or slightly medial to that [6]. In contrast, a more recent publication from 2020 recommended an entry point for PFN-A to be 5 mm medial to the greater trochanter tip, challenging the conventional wisdom provided by the surgical technique guide [7].

The quality of fracture reduction and nail positioning is intrinsically linked to the CM nail entry point. The defining criteria of correct nail positioning are multifaceted and complex, including factors, such as specific distances between the cortex and the nail, the central positioning of the spiral blade, avoidance of distal tip-cortex impingement and more [8,9,10].

Given the complexity of the criteria for correct nail positioning, this study seeks to explore the variability in entry point selection by different trauma surgeons. We hypothesize that entry point selections will differ between surgeons, and that these chosen entry points may not consistently allow for optimal fracture reduction or may cause contact between the nail and the cortical bone. We also aim to examine whether surgeon experience correlates with the consistency and quality of entry point selection.

Methods

This prospective multicenter trial was conducted at a university hospital (level I trauma centre) in collaboration with another level II trauma centre. A total of 16 surgeons, all actively engaged in trauma and orthopaedic surgery, participated in the study. The participant group comprised eight residents, three specialists and five senior specialists/attending physicians. Participant experience in terms of performed implantations varied: eight participants had performed fewer than 50 implantations themselves, four had completed between 100 and 500, two had done between 500 to 1000 and two had undertaken more than 1000 implantations.

The sample size was determined through power analysis, taking into account the expected variations of entry points. On the basis of a pre-study conducted during the development of the simulation software, more experienced surgeons were anticipated to have a mean distance to the mean entry point of 3.5 mm (SD 2.5 mm), while less experienced surgeons were expected to have a mean of 4 mm (SD 2.5 mm). This analysis determined that a minimum of 13 participating surgeons would provide the study with 80% power at an alpha level of p ≤ 0.05.

The trial spanned from September to December 2022, with each participant dedicating approximately 90 min to a custom-developed simulation software. All surgeons were asked to provide the optimal entry point for implantation of a PFN-A (DePuy Synthes) for different virtual femora.

The custom developed software uses digitally reconstructed radiographs (DRRs) to provide X-ray views from computed tomography (CT)-datasets. To simulate these images, a method called ‘ray casting’ is employed. This technique involves casting rays from a virtual focal point towards an image plane, capturing data along the ray’s path via trilinear interpolation [11]. The attenuation of each pixel is computed from the CT data and then transformed into an 8-bit grayscale image, with certain adjustments to enhance realism and contrast. DRRs are all but indistinguishable from actual X-rays (voxel size: coronary 1.02 mm, sagittal 1.02 mm axial 0.55 mm).

Utilizing this method, the software displays two-dimensional images akin to those provided by a fluoroscopy machine. The surgeons could manipulate these images in all dimensions, just as they would during actual surgery. The guiding wire was visible in all images to replicate reality, and the simulated fluoroscopy machine could be adjusted by 0.5 °. Surgeons could always switch between the AP and ML view to ensure that the entry point was determined from at least two views. The guiding wire could be repositioned as needed until the participant confirmed the final position. The participant’s view is shown in Fig. 1.

Simulation software interface. This figure illustrates the user interface of the simulation software, showcasing a realistic femur image. Participants determine the optimal implant entry point by clicking directly on the image. The control panel on the right allows image manipulation, such as rotation and zooming, providing a realistic surgical experience for evaluating entry point variability

For the entirety of the analysis, the PFN-A by DePuy Synthes was utilized as it was the primary implant employed at the participating hospitals.

Additionally, the simulation software established a randomized pattern. This was achieved by autonomously selecting the order and orientation of the 19 proximal femora, each presented to the participants at least three times in a mirrored or unmirrored form. In total each surgeon provided at least 57 entry points. This methodology allowed for the examination of whether the same surgeon would consistently select the same entry point on identical femora. The distance between the chosen entry points for each femur was subsequently calculated.

The mean entry point for each femur was calculated from the 90% closest entry points determined by all surgeons, and the distance from each individual entry point to this mean point was measured at the bone surface (definition of mean entry point). This enabled a comparison of entry point variability and facilitated an investigation into potential differences owing to the experience level and training stage of the surgeons.

Importantly, certain entry points were deemed unsuitable for achieving an anatomical reduction. These are entry points where the distance between the surface of the implant and the surface of the bone was less than 1 mm in some place, which makes an intramedullary position of the implant impossible (negative values indicated an implant position outside the bone). Consequently, in the event of a fracture, an anatomical or near-anatomical reduction is impossible using these entry points. The software was used to calculate the implant’s position relative to the bone, thereby identifying these unsuitable entry points.

Confidentiality was preserved by only documenting the stage of surgical training and experience of each participant without including any identifying information. All participants granted their written consent for publication.

The demographic profile of the used CT datasets is characterized by a mean age of 59.37 ± 20.84 years, with a range extending from a minimum age of 22 to a maximum of 85 years. The sample includes a total of 13 male participants and 6 female participants.

This study was conducted in strict adherence to the Declaration of Helsinki and all of its amendments, ensuring the ethical conduct of research involving human subjects. Prior to its commencement, the study received the necessary approval from the ethics committee of Ludwig-Maximilians-University of Munich.

The statistical analysis primarily involved the calculation of inter-observer and intra-observer variability to assess the variability of entry points among the surgeons and between the entry points provided by the same surgeon. In addition, a comprehensive descriptive statistical analysis was conducted, encapsulating data on the chosen entry point, the surgeon’s experience level and training stage. The statistical significance was tested using Fisher’s exact test, t-test for independent samples and analysis of variance (ANOVA).