1 Introduction

Subtrochanteric fractures (ST) are proximal femur fractures located from the lesser trochanter to 5 cm distal to it and commonly occurs either due to low energy trauma in elderly or high energy trauma in young patients.^1,2 Current practice of treating these fractures is surgical fixation either in the form of intramedullary nailing or extramedullary plating. However, these fractures pose significant challenge to their management because of intense concentration of tensile and torsional stresses along with decreased vascularity at this region which results in high rate of malunion, delayed and non-union and metal work failure.³ Muscle pulls by abductors and flexors to the proximal fragment and by adductors on distal fragment often causes significant displacement between the fracture fragments and reduction of such fragments is often a daunting task.

Fracture healing is described on basis of various clinical and radiological parameters. There is lack of agreement regarding definition of delayed union and non-union. For most fractures, delayed union can be defined as the cessation of the periosteal response before the fracture successfully has been bridged whereas non-union is the cessation of both the periosteal and endosteal healing responses without bridging.⁴ Fracture union time may differ for different regions as fracture in some regions tends to heal quickly than others and hence the terms delayed and non-union become relative. For long bones in most of the studies, fracture union is defined as delayed if fracture fails to consolidate in first six months.^5,6

Subtrochanteric fractures are quite common especially in elderly age group and delayed or non-union of these fractures can cause significant morbidity and mortality. The best outcome for these fractures is timely union which allows comfortable and early weight bearing. Varus malreduction of the fracture have been cited as important factor for delayed union in this area.^7,8

Through this study, we tried to evaluate the risk factors associated with delayed union in this group of fractures using radio-morphometric analysis of various parameters.

2 Methods and material

This study is a retrospective observational study conducted in two major trauma centres. Approval was obtained from the Trust Clinical Governance team. (Local approval number 10719) All patients with fracture of proximal femur were reviewed from electronic data base using the graphics package present on the hospital's picture archiving and communication system (Carestream Vue PACS, Kodak, Rochester, NY) over a period of 5 years between October 2014 and October 2019. The proximal femur fractures were classified according to AO classification and fracture with A3 or B1 according to AO classification were evaluated. Those fracture having both AP and lateral radiographs available intraoperatively (or immediate post-operatively) and 6 months' time were included in final analysis.

Inclusion criteria were age >16 years, traumatic fractures treated with cephalomedullary nailing with at least 6 months follow up. Exclusion criteria were incomplete clinical/radiological data, pathological fractures due to malignancy, stress fractures, loss to follow-up or lack of available radiographs for the opposite side.

All fracture were fixed with cephalomedullary nailing and the type of nail used was recorded. The proximal diameter of nail or screw according to manufacturing guide was used as a calibration to perform measurements on PACS. TraumaCad® digital orthopaedic templating (Brainlab AG, Germany) was used to calculate CCD (Caput-collum-diaphyseal) angle and PACS was used to calculate Radiographic Union Score for Hip (RUSH) score, displacement and three-point relationship. The following parameters were recorded by two authors (GK and NN) under the supervision of senior authors in the intraoperative or immediate post-operative radiographs: (Details in appendix)1.CCD (Caput-collum-diaphyseal) angle on operated and uninjured side. CCD (caput-collum-diaphyseal) angle difference between the operated and uninjured side was calculated from above measurements (Figs. 1 and 2

Close

Fig. 1 Showing the measurement of CCD on surgical side (see appendix for details).

Export to PPT

Close

Fig. 2 Showing the measurement of CCD on contralateral side (see appendix for details).

Export to PPT

Close

Fig. 3 Showing the measurement of residual angulation (see appendix for details).

Export to PPT

Close

Fig. 4 Showing the measurement of residual displacement (see appendix for details).

Export to PPT

The Radiographic Union Score of Hip (RUSH score) was assessed by the same authors using a 6-month radiograph and the average of two scores was taken. Fractures were deemed to be un-united if RUSH score was <18.¹²

CCD (caput-collum-diaphyseal) angle difference between the operated and uninjured side was calculated from above measurements Observational analysis was performed on all failed fixations and revision surgeries. Note was also made if the implant broke or bent, and the fracture healed later without any intervention in case if the patient was not keen to proceed for any surgical intervention.

3 Statistical analysis

Student's t-test was used to compare CCD angle difference between operated and uninjured sides between delayed union and united fractures group as our data was distributed normally. The same statistical test was also used for fracture displacement and alpha angle between the proximal and distal fragments on both AP and lateral views (data distributed normally). Chi-square test was used for 3-point relationship and association with sex. All values were considered statistically different if p < 0.05. Interrater reliability (Kappa) was calculated between the CCD, residual displacement, and residual angulation by two different authors using SPSS.

4 Results

A total of 278 patients were identified. 22 were pathological fractures and were excluded. 130 patients did not have any follow up. This figure was alarming and has led to change in practice to improve follow up. 148 had follow-up but 63 of them had radiographs at different time scale and would not allow calculation of Radiographic Union Score for Hip (RUSH) score at 6 months mark to assess union. This left a total of 64 eligible patients for final analysis.

Delayed union group (26 patients) had a mean RUSH score of 14.12 ± 2.63 (range 10–17.5) and united group (37 patients) had a mean RUSH score of 26.34 ± 3.24 (range 18–30). The mean CCD on the operated side in the delayed union group was 124.07 ± 6.34 (Range 108.5–133.5°) and in the united group was 127.58 ± 6.63 (Range 114–148) and this was statistically different (p = 0.039). However, the average CCD on the uninjured side was 129.85 ± 5.20 (range 112.5–136) and 130.04 ± 4.01 (range 120–138.5) in delayed union group and united group respectively. This was not statistically different (p = 0.882). The mean CCD angle difference between operated and uninjured sides was −5.77° ±6.60 (range +4 to −19.5) for delayed union and −2.33° ±5.40 (+13 to −14.5) for united group. This was again statistically different between the two groups (p = 0.034) suggesting that delayed union group has more varus neck angle as compared to uninjured side.

Residual displacement at fracture site showed statistically significant difference between delayed union and united fractures on anteroposterior and lateral views. On the AP view, in the delayed group, the residual angulation and the residual displacement between the proximal and distal fragment was 5.52 ± 4.32° (range 1–13.6) and 8.11 ± 5.56 mm (range 1.2–20.05) as compared to 3.24 ± 3.52° (range 0–15.45) and 3.68 ± 4.21 mm (range 0–15.15) in the union group. This had p value of 0.031 and 0.001 respectively and was statistically significant. Similarly, on the lateral view, the residual angulation and residual displacement between the proximal and distal fragment in delayed union group was 5.59 ± 5.04° (range 0–15.13) and 8.47 ± 6.98 mm (Range 0.25–27.95) as compared to 2.94 ± 3.60° (range 0–17) and 3.20 ± 3.63 mm (Range 0–14.5) in the union group. These were again statistically significant (p value of 0.026 and 0.0001 respectively)

There was no statistical difference in 3-point relationship of the implant fixation proximally in between the two groups (p = 0.775).

The inter-rater reliability for the calculation of RUSH score, CCD angle and displacement between the two authors (GS and NN) has been shown in Table 1.

There were significant events in both groups (see Table 2). 10 revision procedures were performed for implant failures in the delayed union group. 5 of these failed prior to 6 months and had no RUSH score available but were included in delayed union group. Distal screw bent in 1 case and broke in 2 cases prior to 6 months and resulted in healing of the fractures in union group. This is in contrast to the study from Krappinger et al.⁷ where auto dynamization of nail with first 3 months was associated with implant failure and revision surgery.

5 Discussion

In our study, we found two surgeon related distinct predictors for delayed union. These were varus malalignment of the neck and poor contact between fracture fragments post reduction.

Normal CCD or NSA (neck shaft angle) of different individual may vary. In their systematic review of 26 publications to find the reported values of NSA in plain radiographs in healthy and osteoarthritic subjects, Boese et al. found that the mean NSA (neck shaft angle) was highly variable and found that the mean neck shaft angle was 128.5° (127–130.5°) for rotation-corrected and 129.5° (119.6–151°) for the non-corrected measurements.¹³ They concluded that there is high variance of reported neck shaft angles and there seems to be inconsistency of the published methods of measurements. Hence calculation of this angle in single post-operative films may not represent the change of neck shaft angulation for that particular individual from his native neck shaft angle. Therefore, we used the uninjured CCD as a reference for a particular individual. In cases where opposite sided images were not available, we used the previous hip images available on the system that were taken in the past for some other reason, hence giving us difference of change in neck shaft angle for that particular subject. In his study of 61 patients with subtrochanteric fractures, Freigang et al. found that postoperative mean operated CCD angle was significantly less as compared to mean of opposite side.⁸ They identified that varus alignment of the hip contributes to delayed healing, however, they calculated the mean average of the operated side as compared to mean average of uninjured side. We identified in our study that CCD difference between the operated and uninjured side was significant in delayed union group as compared to union group.

Contact between the fracture fragments is important determinant for union of any fracture. We also found that residual displacement of the fracture was significantly different in two groups on both AP and lateral radiographs. Residual displacement angulation on the AP view was significantly greater in delayed group. Tarantino et al. stated that the efficacy of fracture healing is determined by three ideal conditions: adequate blood supply, good contact between bone fragments and good stability.¹⁴ All these factors can be addressed by surgeon by respecting the soft tissues, adequately reducing the fracture, and using the appropriate metal work to fix the fracture. There had been argument that opening the fracture and using the cables around the fracture can strip off vascular supply to the bone and can prevent healing of these fractures. However, we found that reduction of the fracture with or without open reduction and/or cerclage cables, is an important key step in the management of these fractures.^15,16 In his study by Fauconnier et al., the authors found that average displacement of the lateral wall was larger in the no-cerclage group (p = 0.003) and concluded that wiring does not lead to higher re-operation or infection rates¹⁷.

In our study, we noted patients in delayed healing group were significantly older than the patient in the union group and were predominantly females. This could be related to medical co-morbidities with advancing age, nutritional status, and osteoporosis. We know from the previous studies that osteoporosis in more common in elderly people especially post-menopausal females and this can also contribute to delayed union of these fractures.^18,19

The Radiographic Union Score for Hip (RUSH) is a previously validated tool that improves fracture healing agreement between radiologists and orthopaedic surgeons by using a checklist-based scoring approach.¹² In his study by Frank et al., a threshold score of <18 was associated with a 100 % specificity (95 % CI, 97%–100 %) and a positive predictive value of 100 % (95 % CI, 73%–100 %) for radio-graphic non-union. They found that the patients with a RUSH score of <18 were 10 times more likely to undergo a non-union reoperation than individuals with higher scores (relative risk [RR], 9.9; 95 % CI, 4.4–22.7). Similarly, the 18-point threshold was also predictive for reoperation for all indication (RR, 2.7; 95 % CI, 1.7–4.4). RUSH score calculation can be subjective and hence the average of two scores from different observers was taken to calculate the final score in our study. We found inter-rater reliability between two observers to be 0.574 but there was only one disagreement to categorize these patients into union or delayed union group when the threshold value of 18 was taken. This patient had an average RUSH score of 17.5 and was considered in delayed union group.

This is a large series of 63 patients analysing the factors responsible for delayed union in subtrochanteric fractures. Although the fracture union depends upon multiple factors which could be either patient related, fracture related or surgeon related, this study highlights few critical factors for delayed union which include age, female sex, varus malalignment and residual displacement. Out of this, varus malalignment and residual displacement are intra operative factors which improve union rates when adequately addressed.

Our study has some limitations as well. Being an observation study, we agree that our study might not be adequately powered to find the real difference between the two groups. However, it would be impractical to conduct a randomized study to see the adequacy of reduction and healing of such fractures.

Other limitation of the study is calculation of CCD angle on non-standardized radiographs. It is known fact that the neck shaft angle could be different if the limb is in rotation.^20–22 To counteract this, we allowed ourselves to choose either the intra-operative images or the immediate post-operative radiographs in case if intra-operative images were not good. General practice that foot is kept in neutral rotation on traction table for fixation of these fractures allowed us to get standard radiographs in most of the cases. To choose the other side, we analysed all radiographs available on the system for that patient and chose the best rotation view to calculate the CCD.

Also, the calculation of CCD, alpha angle and residual displacement is quite subjective.²³ These were done on either Traumacad (CCD) or PACS (angles and residual displacement), however, the exact angle calculation and displacement depends on where the marker is placed and lines are drawn. There could be significant intraobserver and interobserver variability in calculation of these angles and distances in spite of standardization. To minimize this error, all the measurements were calculated by two independent observers and the average of the two was taken. Calculation of dislocation on AP and lateral view radiographs could be tricky especially in case of comminution and we took the distance between maximally displaced fragments in such cases in our study.

Our study shows that old age, female sex, varus alignment and residual displacement are important factors to determine the delayed union in subtrochanteric fractures. However, fracture union is complex and other factors like timing of intervention, metabolic parameters, other co-morbidities, fracture personality, open or closed fracture reduction could also determine fracture union and were beyond the scope of this study. We were also not able to find the critical values of varus, residual angulation and displacement which could predict delayed union of the fracture from this study.

6 Conclusion

Timely union of subtrochanteric fractures can prevent long term morbidity and mortality. Our study gives valuable information and highlight the fact that fracture contact and avoidance of varus malalignment improve the union rates in subtrochanteric fractures.

Ethical committee approval

Approval for study was obtained from the Trust Clinical Governance team. (Local approval number 10719).

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Author contribution

Gurvinder Singh Kainth: Conceptualisation; Data Curation; Formal Analysis; Investigation; Methodology; Project Administration; Validation; Writing – Original Draft Preparation; Writing – Review & Editing.

Nimesh Nebhani: Data Curation; Formal Analysis; Investigation; Methodology, Writing – Review & Editing.

Basil Shah: Data Curation; Formal Analysis; Investigation; Methodology.

Gunasekaran Kumar: Conceptualisation; Data Curation; Formal Analysis; Investigation; Methodology; Supervision; Writing – Review & Editing.

Birender Kapoor: Conceptualisation; Data Curation; Formal Analysis; Investigation; Methodology; Supervision; Writing – Review & Editing.

Conflicts of interest statement

The authors declare no conflicts of interest.

Patient consent

Patient data was anonymous and only electronic records were reviewed, hence patient consent was not obtained.