1 Introduction

Antegrade and retrograde intramedullary nailing (IMN) are both accepted as the standard treatment for femoral shaft fractures.¹ For these injuries, retrograde nailing (RIMN) is not only equally effective as antegrade IMN (AIMN) with shorter set-up/operative time and less blood loss but is also preferred as it allows concurrent operative management of multiple injuries without the need for repositioning.^2–7 Subsequently, RIMN is routinely utilized for patients with femoral shaft fractures at our level one trauma center.

When it comes to proximal femoral shaft and subtrochanteric fractures, RIMN is controversial as there is a shorter segment captured by the IMN.^2,5,8 A potential benefit of RIMN for proximal femur fractures, however, in addition to those mentioned above, is that the procedure is performed with the hip flexed over a radiolucent triangle which can aid in the reduction of the typically flexed proximal segment potentially reducing the need for open reduction, decreasing blood loss, and shortening operative times. For these reasons, the senior author's preference is to perform RIMN for femoral fractures within two to three cortical diameters of the lesser trochanter (LT).

The purpose of this study was to compare patients with proximal femoral shaft fractures undergoing AIMN versus RIMN and determine if the practice of RIMN of these fractures is safe and effective.

2 Methods

After institutional review board approval, a retrospective review of skeletally mature patients with proximal femoral shaft fractures treated with IMN at an urban level one trauma institution (Denver Health Medical Center, Denver, CO, USA) between January 2016 and July 2021 was performed. Measurement in a number of cortical diameters (NCD) was utilized to define proximal femur fractures. These were defined as fractures within 3 cortical diameters of the LT. This measure was used instead of the metric-based measured distance, to control for radiographic magnification and differences in patient size. Hence, NCD was obtained by measuring the outer cortical diameter of the femoral shaft 5 cm below the LT to avoid the flare of the LT. The measured distance from the bottom of the LT to the most proximal extent of the fracture was measured and then divided by the measured outer cortical diameter (Fig. 1).

Close

Fig. 1 A) Inclusion criteria of fracture within 3 NCD of the lesser trochanter. B) Anteroposterior radiograph of a 56-year-old male with a right proximal femur fracture after ground-level fall, who underwent retrograde intramedullary nailing (RIMN), and demonstrated the fracture distance in NCD. Using a pre-measured cortical diameter at 5 cm of 4.0 cm (lines not shown to simplify figure), NCD = 5.8 cm/4.0 cm = 1.45. Note: NCD, number of cortical diameters.

Export to PPT

Patients were excluded if they had fractures with extension proximal to LT, concurrent intertrochanteric or femoral neck fractures, pathological fractures, or inadequate radiographs precluding the above measurements.

Electronic medical records and patient imaging were reviewed for demographic information, comorbidities, past medical/surgical history, associated injuries, mechanism of injury, open fracture, Orthopaedic Trauma Association (OTA) fracture pattern, operative details (i.e., approach, operative time, estimated blood loss) and fracture/bone measurements at the time of injury, postoperatively, and at clinical follow-ups.

Primary outcomes measures were operative time, visual estimated blood loss (EBL), postoperative translational and angular displacement, union rate, time to union, nonunion, malunion, need for secondary procedures/reoperation, and complications. For analysis of union and complications, only patients followed until radiographic union, or a significant complication/reoperation were assessed. Analysis of associated orthopaedic injuries other than femur fractures were excluded since the study did not intend to report on weight bearing progression nor patient reported outcomes from these. Only patients undergoing isolated procedures were assessed for operative time and EBL analysis.

2.2

2.2 Radiographic analysis

Plain radiographs were evaluated by an orthopaedic surgeon (FRF) and orthopaedic research fellow (NJT) using PACS Lite Viewer software (Carestream Health Inc., Rochester, NY, USA). Anteroposterior (AP) and lateral (Lat) preoperative and postoperative radiographs were analyzed for: fracture distance from LT (mm); cortical diameter 5 cm from LT; NCD; and pre- and postoperative translation and angulation. Translational and angular displacement were, respectively, measured as the maximum distance between cortical disruptions and mid-diaphyseal lines. True translational displacement (TTD) and angular displacement (TAD) was determined by combining displacement on AP and Lat radiographs using the Pythagorean theorem [√(A² + B²)], as originally described by Paley to determine true deformity in the setting of malunion (Fig. 2).^9,10

Close

Fig. 2 A and B) Preoperative translational and angular displacement on anteroposterior (28 mm medial and 36.19° varus) and lateral radiographs (48 mm posterior and 41.63° apex anterior), respectively, in the same patient demonstrated in Fig. 1. Translational displacement was measured as the maximum distance between cortical disruptions, and angular displacement was measured using mid-diaphyseal lines.⁹

Export to PPT

Fractures were classified according to AO/OTA classification system.¹¹ Union was defined by evidence of healing of ≥3 cortices of biplanar radiographs.¹² Malunion was defined as greater than 5° of angular deformity in any plane.¹³ Nonunion was defined as lack of union with no interval healing at 6 months.¹²

3 Results

A total of 55 proximal femoral shaft fractures that underwent IMN fixation were identified through a prospective trauma registry. One patient was excluded for pathologic fracture, leaving 54 femur fractures for analysis.

AIMN and RIMN fixation was utilized in 31 (57.4%) and 23 (42.6%) patients, respectively. The median age was 30 years (IQR: 23.5 to 47.3), 35 were male (64.8%), median body mass index (BMI) was 26.5 (IQR: 23.4 to 26.9), 17 were smokers (31.5%), 2 had diabetes mellitus (3.7%), 23 had an American Society of Anesthesiology score (ASA) > 2 (42.6%), and 8 were open fractures (14.8%, 1 type I and 7 type IIIa, 5 ballistic). The most common mechanisms of injury were motor vehicle collision (n = 21, 38.9%), fall from height (n = 8, 14.8%), auto-pedestrian accident (n = 5, 9.3%), ballistic injury (n = 5, 9.3%), ground-level fall (n = 5, 9.3%), ski/snowboard injury (n = 4, 7.4%), and motorcycle collision (n = 3, 5.6%). From the 54 surgery cases, 46 were performed by 6 trauma fellowship-trained surgeons, and the remaining 8 (14.8%) by 4 on-call non-trauma trained faculty. There was no intergroup difference among this variable (AIMN: 81.8%, n = 18 vs. RIMN: 92.3%, n = 12, proportional difference: 10.5%, CI: 31.2%–16.2%, p = 0.63).

AIMN and RIMN groups were similar for age, sex, BMI, tobacco use, diabetes, ASA classification >2, preoperative TTD or TAD, open fractures, or ballistic fractures (Table 1). The RIMN group had a trend of more comminuted fractures (65.2% vs. 38.7% AO/OTA classification B/C, p = 0.05). The AIMN group, compared to the RIMN group, had a shorter measured distance from the LT (47.0 vs. 66.1 mm, MD: 17.0, CI: 36.0 to −1.0, p = 0.04) but the difference in NCD was not significant (1.4 vs. 2.0 NCD, MD: 0.5, CI: 1.1 to 0.1, p = 0.07).

Among patients with isolated IMN procedures, the RIMN group had shorter operative times (142.3 vs. 178.5 min, MD 36.2, CI: 8.3 to 64.1, p = 0.01) and less EBL (100 vs. 250 mL, MD: 150, CI: 50 to 250, p = 0.008). Regarding displacement, there was no intergroup difference in postoperative TTD and TAD (Table 2).

Thirty-five patients (22 AIMN and 13 RIMN) were followed to an endpoint (union, complication, reoperation) with a mean follow up of 7.4 months (CI: 5.8 to 9.0). Both groups were similar in age, sex, BMI, smoking, diabetes, ASA score, open fractures, AO fracture classification, fracture distance to LT, NCD, and preoperative TTD/TAD (Table 3). There was no observed intergroup difference in terms of union rate, time to union, complications, or reoperations (Table 4).

4 Discussion

The study found that RIMN fixation of proximal femoral shaft fractures had shorter operative times and had less EBL while leading to similar reduction quality, union rate, time to union, complications, and reoperations.

In this study, we defined proximal femoral shaft fractures as those within 3 cortical diameters (NCD) of the LT. This definition was chosen based on the authors' preference for RIMN fixation of proximal femoral shaft fractures within 2–3 NCD, with AIMN reserved for more proximal fractures. Cortical diameters, instead of measured distance, were utilized to create a standardized relative definition of proximal femoral shaft fractures that was not subject to radiographic magnification or patient size. For example, a fracture within 5 cm of the LT, the current definition of a subtrochanteric femur fracture, in a tall vs. short patient will be two different relative distances in relation to the size of the patient's femur.^15,16 In this study the median NCD of fractures 5 cm from the LT was 1.6 (range 1.2–2.0).

Observing outcomes, this study found that RIMN for proximal femoral fractures performs similarly to AIMN in union rate (76.9% vs. 90.9%, p = 0.3), time to union (12 vs. 20 weeks, p = 0.3), and complications (e.g., 3 vs. 2 nonunions; 2 vs. 1 hardware failure) despite the RIMN group having a higher proportion of AO type B or C fractures (Fig. 3). The overall incidence of open fractures and ballistic injuries was low (14.8% and 9.3%, respectively). There was no intergroup difference for these treated injuries. There were 8 open fractures for the whole cohort with a 75% union rate and 37.5% complication rate. Regarding the 5 ballistic injuries, there was an 80% union rate and 20% complication rate. Moreover, in the whole cohort, no varus malalignment was identified in either group, and time to union was similar to previously reported antegrade technique studies.^17–19 RIMN was faster and had less EBL, in concordance with existing literature.^2,4,6,7 In critically injured polytrauma patients with femur fractures, choosing RIMN to decrease EBL and operative time may be considered to minimize the procedure's morbidity.^20–22 Comparative literature was searched for these same outcomes of union rate, time to union, and complications to evaluate how this data fits within the current evidence on IMN of proximal femur fractures.

Close

Fig. 3 A) Anteroposterior radiographs of a 19-year-old male, involved in a motor-vehicle collision, who sustained a left femur fracture (NCD: 1.98), left posterior column posterior wall acetabular fracture, and inferior pubic ramus fracture. He underwent RIMN for his femur fracture and additionally received percutaneous posterior column screws. The posterior wall and inferior ramus fractures were stable and did not require fixation. B) Femoral implant was complicated by implant failure at fracture site 1.5 months after index surgery. C) Patient subsequently underwent hardware revision. Note: (A and B) are stitched radiographs at the distal third, due to the lack of a full femur image in the electronic medical records system.

Export to PPT

Over time, RIMN has become a feasible, convenient technique for treatment of femoral shaft fractures given its aforementioned broader indications, faster setup/positioning, decreased EBL, and shorter operating/radiation time.^2,4,6,7,23 However, for proximal femoral shaft fractures or subtrochanteric fractures, RIMN may not be ideal secondary to short segment fixation in an area of high deforming forces.^13,24–27 While this current study found no observed difference between RIMN and AIMN in terms of complications, revision surgery, and nonunion rate, these results are limited by adequate follow-up which resulted in a small number of patients and complications available for analysis. This is demonstrated by the relatively large absolute differences and wide confidence intervals between groups. The complication rate reported on both groups represent absolute numbers for those followed to endpoint criteria with some overlapping regarding types of complication, hence having patients with more than one complication (e.g., blood loss and hardware revision). Therefore, the complication rate needing reoperation in the RIMN was 38.5% and for the AIMN was 13.6%, but for the whole cohort was down to 21.7% and 9.7%, respectively. The current RIMN group in this study had a similar union rate (76.9%) and complication rate, no malunions, and a shorter time to union as compared to earlier studies (12 weeks).^5,13,28,29 Prior, comparably small studies have examined this same question and found conflicting results.⁵ DiCicco et al. reviewed 16 patients with proximal femoral shaft fractures treated with RIMN and found: a union rate of 64.7%, malreduction rate of 35% (varus average deformity of 5.06°) and a longer time to union of 22 weeks.⁵ Similarly, Ricci et al. found in 16 patients who underwent RIMN for proximal femur fractures a malreduction rate of 38%.¹³ Kuhn et al. found no intergroup differences for complication rate, union rate, and time to union among antegrade and retrograde nailing of supraisthmal proximal femur fractures.²⁹ The same researchers, in a larger cohort of patients treated with RIMN, also found a union rate of 91%, malunion rate of 3% (5–10° of varus), and a longer time to union of 26 weeks.²⁸

A major argument for AIMN of high-energy proximal femoral shaft fractures is the ability to use proximal interlocking screws in reconstruction mode to protect the femoral neck from occult or future femoral neck fractures.^30,31 It has been demonstrated that even routine high-quality computed tomography (CT) scans will miss femoral neck fractures leading one center to recommend rapid sequence magnetic resonance imaging to rule out occult femoral neck fractures which were identified in 12% of these injuries.^32,33 In the event of performing a RIMN for a proximal femoral shaft fracture, it is essential to maintain a high degree of suspicion of an associated femoral neck fracture intraoperatively and to understand that CT scans alone cannot identify all occult fractures.

This study is limited by its retrospective design and variation among surgeons with differing approaches/techniques. Consequently, there is an inherent risk of selection bias, which was most prominently seen in this study with complex fracture patterns being found marginally more often in the RIMN group (p = 0.05) which may have predisposed the group to potentially worse outcomes. Additionally, many patients were lost to follow-up as is typical of the trauma population. The observed differences between groups had large confidence intervals so true differences cannot be excluded. Future studies with larger cohorts or a prospective randomized study are needed to further assess the findings of this study.

5 Conclusion

RIMN appears to be a viable treatment option for proximal femoral shaft fractures, as compared to AIMN, resulting in less operative time, less blood loss, and no detectable differences in union, reoperations, or complications.

Funding

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

Informed consent

Not applicable.

Institutional ethical committee approval

Approved.

CRediT authorship contribution statement

Francisco Rodriguez-Fontan: Conceptualization, Methodology, Investigation, Resources, Writing – original draft, Writing – review & editing, Visualization. Nicholas J. Tucker: Conceptualization, Methodology, Investigation, Resources, Writing – original draft, Writing – review & editing, Visualization, Formal analysis, Resources. Katya E. Strage: Writing – review & editing, Visualization. Cyril Mauffrey: Supervision, Writing – review & editing. Joshua A. Parry: Conceptualization, Writing – review & editing, Supervision, Project administration.