Comparisons with other studies
Pelvic fractures represent approximately 8% of blunt trauma injuries and are associated with high mortality and disability [
26]. Deformity of the pelvic ring often leads to pelvic asymmetry, pelvic instability, and subsequent complications, such as residual pain and movement disorders, which are associated with poor patient outcomes [
27,
28].
Traditional pelvic classification systems for preoperative planning and reduction pathway guidance for unstable pelvic fractures are no longer sufficient for clinical and research purposes [
29], and a thorough understanding of pelvic deformity is important for further evaluation, closed reduction, and safe screw and external fixator pin placement. Currently, specific types of pelvic fractures are assessed with the aid of roentgenography and CT scans. Viegas et al. reported that 3D-CT reconstruction techniques had a volumetric validity of 94% and a linear validity of 97% [
30]. Nystrom et al. demonstrated that CT and its resulting computer-reconstructed radiographs (CRRs) showed good validity and reliability, especially for assessing rotational displacement [
31]. Although CT is regarded as the gold standard for true displacement, it is associated with high cost and excessive radiation exposure [
31,
32]. Hence, some orthopedic surgeons attempt to analyze pelvic displacement using radiographs, which are conventional and commonly used in clinical settings.
Lefaivre et al. collected pre- and postoperative radiographs from 25 patients with OTA B or C pelvic ring disruption, assessed the reliability using three representative methods, and then reviewed 31 manuscripts from 28 authors published between 1990 and 2009 involving the radiographic measurement of pelvic fractures [
33,
34]. Although there was an advantage with using three preoperative films, the absolute displacement method of Lefaivre et al. [
35] showed poor reliability in contrast with the moderate agreement of the impression of the observers. The cross measurement method of Keshishyan et al. [
36] based only on the AP view showed excellent agreement, and the inlet/outlet ratio measurement method described by Sagi et al. [
37] showed almost perfect agreement on preoperative radiographs. It was shown that a high number of views, choices, and steps available to an observer were associated with poor agreement [
34]. However, as the pelvis is a complex 3D structure, it is difficult to describe pelvic displacement completely from one or two views. Comprehensiveness and reliability appear to form a paradox; therefore, although measurement methods for pelvic deformity have been described by many authors, there is still no widely accepted standardized method for describing the complete information of pelvic fractures with high reliability and validity. The pelvic deformity measurement software program appears to be a solution for the abovementioned paradox.
The inner measurement method was shown to be reliable and valid in our previous study [
25]. However, it appears to be slightly difficult to master, and there may be errors from hand measurement. Therefore, we created a measurement software program that can semi-automatically measure the specific pelvic displacement patterns from radiographic planes (inlet, outlet, and AP view), and evaluated its reliability, validity, and preliminary clinical application status. Although the reliability of the software was slightly lower than that of CT coupled with 3D reconstruction, the software showed near perfect agreement in all views and all movement patterns and was more reliable than hand measurement. The validity of the software was assessed according to its consistency with the gold standard, which was almost perfect (0.939, 95% confidence interval 0.905–0.973). The likely reasons for the high reliability and validity of the software were well-defined landmarks, a standardized semi-automated measurement procedure, and simple operation steps, which substantially reduced human-induced measurement error. The software well resolved the paradox between comprehensiveness and reliability. The simple operation and short learning period make the software more suitable for young surgeons with limited experience of pelvic fracture. The four observers in our study took a mean of 55 ± 9 min to grasp the software and less than 10 min to analyze pelvic deformity using the three X-ray planes after practicing for a mean of 7 ± 1.4 times.
Previously, reduction of pelvic displacement depended mostly on a surgeon’s experience. In our preliminary clinical application, the surgeons took full advantage of the software and found an evident decline in the reduction time. Compared with hand measurement, the software can provide the surgeon with a more comprehensive and reliable impression of the pelvic displacement, which is very helpful for guiding closed reduction and planning a reduction pathway. Compared with CT and 3D reconstruction, the software depends on real-time fluoroscopy, which is less time-consuming.
The pelvic deformity measurement software program has several advantages. First, it comprehensively describes the spatial pelvic displacement with translational and rotational patterns in three mutually orthogonal axes based on X-ray images, and thereby covers all possible types of pelvic deformity in three dimensions. Second, it shows good reliability and validity when compared with the gold standard and has obvious advantages over hand measurement. Third, it turns a sophisticated measurement procedure into a simple measurement approach with a few clicks in the radiograph planes. It requires less operating room equipment, avoids the high cost and large amount of radiation of CT, and can be used to evaluate pelvic displacement in real time during the closed reduction process. Vertical and transverse distances are both automatically calculated once the landmarks are selected. Fourth, considering the more ubiquitous presence of fluoroscopes in the operating room and the reduced radiation exposure in comparison with CT, our software is a good choice for the measurement of pelvic displacement.
The present study has some limitations. First, our software can only assess unilateral pelvic fractures with AO/OTA classifications of type B or C. Second, the study sample was small; however, we will expand the number of clinical cases in the future. Third, our software fails to calculate the exact extent of the deformity. Nonetheless, the surgeon can take advantage of the software program for comprehensive preoperative planning and overall reduction pathway guidance. Surgeons should use real-time fluoroscopy to observe specific details and the extent of closed reduction and percutaneous fixation. The next stage in the development plan for the software is to add the capability to measure the translational distance and rotational angle.