1 Introduction

Intertrochanteric femur fractures occur at a rate of 150,000 per year in the United states, accounting for about 50 % of all proximal femoral fractures with an incidence of 63 per 100,000 population in elderly females and 34 per 100,000 in elderly males.¹ These fractures have a bimodal incidence with the majority occurring in elderly patients due to low-energy mechanisms and those occurring in younger populations as a result of high-energy trauma.^1,2 Classically, these fractures were treated with extramedullary plate and screw devices (PSD) (e.g. Dynamic Hip Screw) but their use in unstable fracture patterns leads to increased rates of fixation failure and malunion.^3,4 Treatment with cephalomedullary nails (CMN) has become increasingly popular among surgeons due to the ability to use a less invasive approach and more reliable fixation in unstable fracture patterns.^3,5

In general, cephalomedullary implants come in short (SCMN; 17–24 cm) and long (LCMN; 26–48 cm) options. Guidelines on when to choose a CMN over a PSD are well established but choosing the optimal length of CMN is largely left to surgeon preference and judgment. The American Academy of Orthopaedic Surgeons (AAOS) has several guidelines on the treatment of intertrochanteric and peritrochanteric fractures but do not specify what length of intramedullary device to use (Fig. 1).⁶

Close

Fig. 1 Selected AAOS Clinical Practice Guidelines pertaining to intertrochanteric/pertrochanteric femur fractures and recommended treatment options (3, 6).

Export to PPT

Current decision-making processes include looking at the extent of fracture propagation distal to the lesser trochanter in the shaft of the femur. Some recent literature indicates that a short CMN is adequate for up to 3 cm of subtrochanteric extension.⁷ During implant selection, the idea of “protecting the entire femur” leads some surgeons to select the longer implant for fracture fixation. This method of thinking may be due historically (at least in part) to first generation short implants being associated with a higher risk of peri-implant fracture.⁸

However, recent clinical and biomechanical studies comparing SCMN and LCMN have shown decreased operative times, intraoperative blood loss, and need for transfusion with short nails. No difference in postoperative implant failure or peri-implant fracture rates has been demonstrated when comparing the different lengths--calling into question the efficacy of "protecting the whole femur" with a long nail.^8–15 They also highlight a need to evaluate the morbidity of revising these peri-implant fractures as a long CMN often precludes the possibility of switching to a longer intramedullary implant for lack of adequate distal bone stock and will often require more invasive surgical fixation (e.g. lateral femoral locking plate). The purpose of the current study is to review patients with peritrochanteric fractures treated with SCMN, LCMN, or PSD and determine whether there were any differences between the three groups in terms of secondary fracture, need for blood transfusion, or increased dependence following the secondary surgery for fracture.

2 Methods

This is a retrospective review of a retrospective multi-center database of de-identified charts of orthopaedic patients with intertrochanteric femur fractures between January 1, 2016 to September 31, 2021. The database includes 151 centers in the United States ranging from Level IV to Level I trauma centers. Data was gathered by CPT and ICD-9/ICD-10 codes. Patients age 18–89 years old with intertrochanteric femur fracture were reviewed for peri-implant fracture and associated revision surgeries within 12 months after their index procedure. Peri-implant fracture was identified among the cohort by readmission for CPT codes corresponding to repeat cephalomedullary nail or open reduction and internal fixation of the ipsilateral limb. Outcome measures included death, deep vein thrombosis (DVT) or pulmonary embolism (PE) within 12 months of surgery, transfusion during hospitalization, length of stay, and rates of discharge to home and were measured via retrospective analysis of the electronic medical record (EMR) and medical claims data. Complications were defined as peri-implant fracture, all cause implant removal, DVT or PE within 12 months of surgery, death, and transfusion. Patients were stratified by medical comorbidities (utilizing the Elixhauser Comorbidity Index [ECI]), gender, age, race, and smoking status. Inclusion criteria were patients aged at least 18 years up to 89 years (due to database restrictions) with intertrochanteric and subtrochanteric femur fractures. Patients excluded from analysis were those who were pregnant, had diagnosis of any cancer, were revised for nonunion, age less than 18 or over 89 years, or those who had a DVT or PE or implant from prior surgery present on initial encounter.

ICD10 and CPT codes identified 29,786 patients for analysis. Patients were excluded who had multiple surgical procedures during initial hospitalization (263), had inadequate information on index procedure implant due to varied methods of input and implant description (some descriptions involved serial numbers rather than a description of implant type) (11,821), were over 89 years of age (3,412), had missing demographic data (136), were pregnant (1), had a diagnosis of cancer (643), or had a DVT present on initial encounter (313). The final cohort included 13,197 patients.

Chi-squared analysis with Bonferroni correction as needed to adjust for multiple comparisons (smoking status, transfusion, DVT/PE, and readmission rates). Levene's test was used to determine homogeneity of variance among the groups (Elixhauser, Length of stay). ANOVA was performed when non-significant results were obtained and Kruskal-Wallis test was conducted for comparisons without homogeneous variance. Post hoc analysis was required for evaluation of Elixhauser Comorbidity index scores and a Tukey-Kramer correction was used to adjust for multiple comparisons due to unequal sample size among the 3 surgery types. A negative binomial regression was conducted to predict the length of hospital stay based on surgery type and no adjustments were needed. Fischer's exact test was used in the analysis of discharge destination after surgery.

3 Results

A total of 13,197 patients were included in the study for analysis. Of these patients, 5924 were treated with a LCMN, 6464 with a SCMN, and 809 with a PSD. The majority of patients in this study were white (11,780; 89 %; p = 0.0067), aged 65–89 years old (10,976; 83 %; P < 0.0001), and female (8692; 66 %; p < 0.0001) (Table 1).

Demographic analysis showed similarity between the groups with the exceptions of smoking status and ECI. The PSD group had higher rates of current smokers (20.9 %) compared to LCMN (16.3 %; p = 0.003) and SCMN (15.9 %; p = 0.001) and slightly lower Elixhauser Comorbidity Index Scores (PSD 2.97) compared to both LCMN (3.24, p < 0.001) and SCMN (3.23, p < 0.001).

There was no statistically significant difference noted between any of the groups with respect to the primary outcome of peri-implant fracture (LCMN 50, 0.84 %; SCMN 57, 0.88 %; PSD 6, 0.74 %; p = 0.91) (Table 2). Likewise, there was no significant difference in the rates of all-cause implant removal between the three groups (LCMN 76, 1.28 %; SCMN 61, 0.94 %; PSD 9, 1.11 %; p = 0.20). Overall complication rates comparing the LCMN to SCMN (OR 1.41, 0.962.1, p = 0.08), LCMN to PSD (OR 0.99, 0.49–2.0, p = 0.98), and SCMN to PSD (OR 1.43, 0.69–2.9, p = 0.34) did not reach statistical significance.

The rates of overall complications were significantly higher in patients aged 30–49 compared to those 65–89 (OR 2.9, p = 0.01). However, the rate of overall complications, including all-cause mortality, did not differ significantly between groups (LCMN vs PSD p = 0.97; LCMN vs SCMN p = 0.078; PSD vs SCMN p = 0.34). Neither was a significant difference found in overall complications when comparing non-white to white patients (OR 0.55, 0.27–1.14, p = 0.11) or female to male patients (OR 1.33, 0.88–2.0, p = 0.18).

Following the initial surgery, rates of blood transfusion were significantly higher with use of the LCMN (0.6 %) and PSD (0.9 %) devices compared to SCMN (0.2 %; p = 0.0015) but not significantly different when comparing LCMN to PSD. There was no statistically significant association between any implant type and rates of DVT or PE in the first 12 months after surgery (LCMN 91, 1.54 %; SCMN 83, 1.28 %; PSD 13, 1.61 %; P = 0.44).

Upon readmission and revision surgery for peri-implant fracture, rates of discharge home versus to care facilities were similar between all three groups (p = 0.88). 5 of 50 in the LCMN group were discharged home, 5 of 57 in the SCMN group were discharged home, and 0 of 6 in the PSD group were discharged home (p = 0.14). However, PSD was associated with increased length of stay by a factor of 2.03 compared to SCMN (p < 0.009) during readmission for periimplant fracture after adjusting for age, sex, race, comorbidity score, and smoking status (Table 3).

4 Discussion

In this database review of 13,197 patients, we found no difference in the rates of peri-implant fracture between SCMN (0.88 %), LCMN (0.84 %), and PSD (0.74 %) (p = 0.91). Our findings suggest that the choice of fixation (LCMN, SCMN, PSD) for intertrochanteric femur fractures does not affect the risk of peri-implant fracture or the need for implant removal or revision. These findings contradict the notion that LCMNs protect the femur from future fracture compared to SCMNs.

It is worth noting that the incidence of peri-implant fracture was very low in our study population, comprising <1 % of each patient group, although consistent with current reported rates in the literature ranging from 0 to 4%.^9,12,16 Patients who did sustain a peri-implant fracture were not more likely to have greater overall medical comorbidities at baseline. However, they were likely to be in the oldest age group. Our results add to the growing evidence that the use of a longer implant does not protect the femur in treatment of peritrochanteric fractures.

In contrast, the findings reported by Vaughn et al. showed an increased rate of distal-peri-implant fracture after use of SCMN (2/60 patients, 3.33 %) compared to LCMN (0/196 patients, 0 %) in 2014—well past the time of the first-generation short nails.¹² However, their reported difference in rates did not reach statistical significance with a p value of 0.054. It is also worth noting that they exclusively examined third generation Gamma nails whereas our study and others reported here include several different companies and designs. Dunn et al., in a systematic review showed no difference in refracture rates between the different length intramedullary devices.¹⁰ Shannon et al. in a prospective randomized trial likewise reported no difference in peri-implant fracture rates between long and short nails with a mean follow up of 13.8 months.⁷ These findings may indicate that, while there is no overall difference in peri-implant fracture rates between the different intramedullary device lengths, a subset of devices may be associated with an increased rate of peri-implant fracture.

In the current study, the rates of transfusion after the initial surgery were greater in the LCMN and PSD groups compared to the SCMN group, but there was no statistically significant difference between the LCMN and PSD group. These findings are different than those reported by Pandarinath et al. and Cai et al., who reported increased rates of bleeding and need for transfusion in their cephalomedullary nail fixation group compared to the sliding hip screw group.^17,18 However, those two studies (Pandarinath and Cai) did not differentiate between LCMN and SCMN, instead grouping them all into one category.^17,18 Our findings are, however, consistent with several other studies where intramedullary devices were separated by length for analysis. Boone et al. reported that both estimated blood loss (EBL) and transfusion rates were significantly greater in their patients treated with LCMNs when compared to those treated with SCMNs.¹⁹ Moreover, Womble et al. reported increased rates of post-operative anemia and transfusion among the patients treated with LCMNs compared to intermediate and SCMN groups.²⁰ Dunn et al. found decreased blood loss and transfusion rates with shorter operating room (OR) times for SCMNs.¹⁰ Shannon et al. also reported shorter OR times in the short nail group.⁷ Several mechanisms may be responsible for the apparent increase in blood loss when comparing the different devices. A PSD typically requires a larger incision than intramedullary implants. When comparing the long and short cephalomedullary nails, longer nails tend to be used for more complex fractures (e.g. subtrochanteric fractures) which are more likely to require accessory incisions and open reduction. Additionally, a LCMN usually requires reaming, which may not necessarily affect the EBL for a case but could cause an increased transfusion rate. Looking forward, improved implant designs may further improve the performance of short nails while maintaining the benefits noted to OR time and blood loss.

To evaluate the morbidity associated with surgical revision after peri-implant fracture, we looked at rates of discharge home and length of stay in the hospital following readmission for fixation of a peri-implant fracture. Patients in all groups of our study were discharged home or to another care facility at similar rates without a significant difference. In a retrospective cohort study, Darbandi et al. reported that patients with stable intertrochanteric femur fractures treated with intramedullary devices were more likely to be discharged to a place other than home compared to extramedullary devices at the time of index procedure.²¹ They ascribed the lower rate of discharge home among patients treated with intramedullary implants to increased blood loss as evidenced by greater rates of transfusion among that population.²¹ In that study, however, no distinction is made between long and short intramedullary devices and it may be that the additional reaming and extra time in the operating room for LCMNs contributed to the decreased rates of discharge to home. The findings of Darbandi et al. regarding hospital discharge focused on rates of discharge home for what would be the index procedure while we looked at discharge following revision procedures for peri-implant fracture.²¹

There are some limitations to our study. First, it is retrospective in nature and therefore carries the inherent limitations in data collection methods, establishment of causal links between variables, and selection and recall bias. In this database, patients needed to present to the same facility for their initial and secondary admissions in order to capture peri-implant fracture. We were unable to capture patients who may have presented to a second hospital for their readmission and our data, therefore, likely under-represent the true rates of peri-implant fracture in our population. Additionally, we relied on ICD-10 and CPT coding for the establishment of patient data sets as inclusion, de-indentification, and review of all pertinent imaging was not possible with the current data set due to how the data is collected and stored. Not all implants were documented in a manner that allowed for correct identification and we were unable to definitively establish the implant used in each revision surgery, which limits our overall analysis and the resultant implications of using the different implants for the primary surgery. For example, if most SCMN revisions were to a LCMN, while most LCMN revisions required distal locked plating, it may be assumed that SCMN revision carried with it decreased patient morbidity in terms of less invasive revision treatment and perhaps improved mobility following the second surgery. Unfortunately, however, these analyses were unable to be completed due to data limitations.

In conclusion, our data adds to a growing body of literature that shows no differences in the rates of peri-implant fracture for treatment of intertrochanteric femur fractures with SCMNs, LCMNs, and PSDs. There is little to no evidence that a LCMN “protects the bone” in cases of intertrochanteric femur fractures, and implants should be chosen to adequately fix the current fracture rather than in a theoretical attempt to prevent a future fracture.

CRediT authorship contribution statement

Mason Poffenbarger: Conceptualization, Methodology, Investigation, Writing – original draft, Writing – review & editing, Visualization. Neil Werthmann: Conceptualization, Methodology, Investigation, Writing – original draft, Writing – review & editing, Visualization. Kisan Parikh: Conceptualization, Writing – review & editing, Visualization, Supervision, Project administration. John Riehl: Conceptualization, Methodology, Writing – review & editing, Visualization, Supervision, Project administration.

Patient/guardian consent

No patient/guardian consent obtained. This was a retrospective review of de-identified data.

Funding/sponsorship

This research was supported (in whole or in part) by HCA Healthcare and/or an HCA Healthcare affiliated entity. The views expressed in this publication represent those of the author(s) and do not necessarily represent the official views of HCA Healthcare or any of its affiliated entities. No institutional grants or industry sourced funding was utilized in conducting the research for or in preparing this manuscript.

Institutional ethical committee approval

Documentation submitted to Journal of Orthopaedics and is available upon request.

Disclosures

The authors have no financial or other relevant disclosures. This project was supported (in whole or in part) by HCA Healthcare and/or an HCA Healthcare associated entity.

Funding statement

This research was supported (in whole or in part) by HCA Healthcare and/or an HCA Healthcare affiliated entity. The views expressed in this publication represent those of the author(s) and do not necessarily represent the official views of HCA Healthcare or any of its affiliated entities.