Our retrospective study found 34 (6.6%) isolated GTFs which is in line with other studies that found GTFs in 4.1–6.3% following HA in patients with a femoral neck fracture [12,13,14]. The only patient-related risk factor for intraoperative GTF identified in our study was a decreased prothrombin time. Although more patients were on warfarin in the GTF group, the difference was not significant (15% vs. 9%, p = 0.358). However, prothrombin time may have also been influenced by vitamin K deficiency related to malnutrition or medical disease. Vitamin K plays an important role in bone health and is highly prevalent in patients with hip fracture [15, 16]. In addition, 25OH-vitamin D3 levels tended to be lower in the GTF group supporting the hypothesis that reduced bone health may result in increased susceptibility to iatrogenic fractures. Otherwise, a lower prothrombin time may indicate increased bleeding intraoperatively, especially during rasping the intramedullary canal, which may limit a clear view and urge the surgeon to conclude surgery promptly. This in turn increases the stress level of the surgeon, potentially altering performance with a higher rate of GTFs.
In contrast to our study, no differences of 25OH-vitamin D3 levels were found between patients suffering an intraoperative periprosthetic fracture compared to patients without fracture in a study of 271 treated with HA for femoral neck fracture [12]. However, patients in this study presented with mean 25OH-vitamin D3 levels of 20 ng/ml, whereas hip fracture patients in our region exhibit significantly lower levels with a median of 8.4 ng/ml as shown previously [17]. Hong et al. found no difference of T-Scores between groups, but identified Dorr C type femoral canals [18] as a risk factor for intraoperative fractures as did Bellova et al. [14]. Since we focused on isolated GTF, the Dorr types of the femoral canals were not determined. Bellova et al. [14] found female sex as a risk factor for intraoperative iatrogenic fractures in their series of 481 HA. We and Hong et al. [12] could not confirm sex as a risk factor.
Preoperative waiting time was significantly shorter in the GTF group. Shorter preoperative waiting times may limit optimal preparation and assembly of a well-rehearsed team. Although not significantly, HAS were less frequently involved in the GTF group compared to controls in our study. Moreover, 51.6% of all patients were operated beyond 24 h after admission and 81.1% during routine daytime hours at our institution—which more likely allows allocation of a suitable specialized team. However, surgery performed on-call had no influence on GTF as confirmed by others [14].
Although not by significance, highly specialized hip arthroplasty surgeons (HAS) were represented less frequently in the GTF group than in the control group, implicating that surgical experience may be associated with favorable outcome with regard to this specific complication. Nevertheless, no reduction of GTF was observed after establishment of a certified arthroplasty center, despite implementation of additional quality standards with reference to clinical pathways, qualification of surgeons, and quality management. 50% of the GTFs were recorded after implementation of our arthroplasty center, whereas in the non-fracture group, only 31% of HAs were performed in the period of the arthroplasty center. An explanation could be that by restructuring the arthroplasty service new surgeons had to be integrated into acute fracture arthroplasty. They had to adapt to hip implants and instruments specifically used for acute HAs in femoral neck fracture patients, but not for elective total hip arthroplasty at our institution. In addition, a significantly lower number of HAs was implanted during the considerably shorter period of the hip arthroplasty center underlined by a substantially lower case load per surgeon. Moreover, an implant or instrumentation (i.e., rasp) immanent issue might be responsible for the relatively high rate of GTF in this period. This is supported by Laflamme et al. [13] who found an abnormal number of iatrogenic intraoperative fractures following HA for femoral neck fractures after they switched to the same implant inserted at our institution. They observed 22 (6.3%) iatrogenic GTFs in 348 HAs which is almost identical to our results. Other factors potentially affecting the risk of GTF, as, for example, femoral offset influenced by implant size and design, were not analyzed in our study. However, Laflamme et al. [13] found no relationship between implant sizes and GTF. Moreover, only limited information was found in our operation reports regarding the stage of the surgery at what a GTF occurred. On the other side, it is not always possible to differentiate exactly, whether a GTF originated during hip adduction to expose the proximal femoral canal, preparation of the femoral canal, stem insertion, or reduction of the hip after implantation.
Patients with GTFs after THA were demonstrated to experience a worse outcome with limping, increased pain, and functional impairment [8,9,10]. In addition, Homma et al. [19] hypothesized that even minor GTFs with a small fragment (chip fracture) may result in increased bleeding and formation of hematoma. This is confirmed in our study. Seroma and hematoma requiring surgical revision clearly tended to be more frequently after GTFs. A significant association was shown between GTFs and early postoperative periprosthetic infections requiring revision.
This study is limited by its retrospective design including patients over a long period of time. Thus, it may be subject to changes and confounders not identified. However, annual case load and patient characteristics remained constant (Fig. 1). Another limitation may be that beyond postoperative radiographs usually no additional or routine follow-up radiographs were ordered and thus not considered. Thus, non-displaced GTFs not identified on postoperative radiographs, but dislocating secondarily may have been missed. Further limitations may be seen in the unequal distribution of the two different approaches used as well as not addressing anatomical aspects of the proximal femur (i.e., varus or valgus hip) which may also have an impact on risk of GTF. Finally, the number of patients with a GTF is relatively small compared to the non-fracture group potentially impairing statistical analysis. The strength of the study lies in a high number of patients that received the same type of implant.