We retrospectively reviewed the records of the combat casualties managed at two institutions during the study period of March 2003 to March 2008. We identified forty-one patients with open Gustilo-Anderson Type-III fractures12 of the proximal femur, which included femoral head, femoral neck, pertrochanteric, or subtrochanteric (defined as being within 5 cm of the base of the lesser trochanter) fractures. We reviewed the casualty patient databases, electronic medical records, radiographs, and laboratory culture data.
We recorded data to include sex, age, mechanism of injury, fracture classification, modes of provisional and definitive fixation, number of debridements before and after definitive fixation, time to definitive fixation, need for skin-grafting and/or flap coverage, and time to union. We defined fracture union as bridging callus being present on both the anteroposterior and lateral radiographs, the patient being nontender at the fracture site, and the patient being able to bear full weight on the affected extremity13. We recorded complications requiring reoperation.
Although the definition of wound infection following operative treatment of fractures has considerable variation, we utilized the following Centers for Disease Control and Prevention (CDC) criteria for surgical site infection as applied to orthopaedic wounds14. A superficial wound infection was one that was located entirely above the fascia and did not require exposure of the bone or hardware. In those cases in which the fascia was absent secondary to the initial injury or subsequent debridements prior to definitive fracture fixation, the wound infection was considered superficial if there was no exposed bone or hardware and the infection did not track down to these structures. A deep wound infection was one that tracked down to the level of the bone and/or implants. In this study, all deep infections, in addition to meeting the CDC criteria, were required to have positive deep-tissue cultures and/or were treated with six weeks of parenteral antibiotic treatment in addition to irrigation and debridement of the wound.
Statistical Methods
Descriptive statistics were calculated for the results obtained. Categorical data were analyzed with use of the Student t test or Fisher exact test for binomial data.
Source of Funding
No external funding was provided in the completion of this study.
Injury Characteristics
Forty-one patients (thirty-nine men and two women) with a mean age of 25.7 years (range, twenty to forty-nine years) were identified as receiving treatment for open proximal femoral fractures. Blast was the predominant mechanism of injury (twenty-nine patients [71%]), followed by high-velocity gunshot wound (eight patients [20%]), high-energy motor vehicle crash (three patients [7%]), and helicopter crash (one [2%]). The fractures were classified as Type IIIA (thirty patients [73%]), Type IIIB (six patients [15%]), and Type IIIC (five patients [12%]) according to the Gustilo-Anderson classification. The fractures were further classified according to the AO/OTA classification: five fractures were 31-A; two, 31-B; one, 31-C; three, 32-A-1; seven, 32-B-1; and twenty-three, 32-C-1. Subtrochanteric fractures (32-X-1) and intertrochanteric fractures with subtrochanteric extension (31-A) accounted for thirty-eight patients with injuries (93%). The three other fractures were 31-B1 (basicervical femoral neck fracture with an ipsilateral transfemoral amputation), 31-B2 (transcervical femoral neck fracture), and 31-C3 (femoral head, femoral neck, and ipsilateral acetabulum fracture).
Vascular, Peripheral Nerve, and Skin Injuries
Open proximal femoral fractures often did not occur in isolation, but frequently were associated with additional trauma to the chest, abdomen, and/or head in thirty patients (73%), in addition to multiple other orthopaedic injuries. Ipsilateral fractures to the hemipelvis or distal extremity occurred in fifteen patients (37%). Concomitant peripheral nerve and vascular injuries were often encountered. Ipsilateral peripheral nerve injuries occurred in fourteen patients (34%). Of these fourteen patients, ten had nerve injuries that were in communication with the open fracture wound (eight sciatic nerve injuries, one femoral nerve injury, and one combined sciatic and femoral nerve injury), and the other four had nerve injuries that were distal to the fracture wound (three traumatic amputations and one common peroneal nerve injury associated with an open tibial plateau fracture). At the time of the latest follow-up examination, nerve function recovered in only one patient (sciatic neuropraxia injury), with proximal nerve injury associated with the open fracture. This patient recovered M4 motor strength and protective foot sensation by five years after the time of injury. Five patients sustained open Type-IIIC fractures requiring vascular repair and/or reconstruction at the time of injury to restore distal limb perfusion.
Additional wound coverage with split-thickness skin-grafting was necessary in eighteen patients (44%), and seven patients (17%) had extensive thigh musculature loss with substantial exposed femur that required muscle or fasciocutaneous flap coverage. Five rotational flaps (two biceps femoris, one vastus lateralis, one sartorius, and one cross-leg fasciocutaneous) and four free flaps (three rectus abdominis and one latissimus dorsi) were used.
Fracture Treatment
The patients arrived at our institutions with a mean evacuation time of 5.9 days (range, two to twenty-five days) from the time of injury. A mean of 2.7 irrigation and debridement procedures (range, one to six procedures) were performed prior to arrival at our institutions, and an additional mean of 3.2 irrigation and debridement procedures (range, one to twelve procedures) were performed at our facilities prior to definitive fracture fixation. Perioperative antibiotic prophylaxis (Ancef [cefazolin] and gentamycin in the first seventy-two hours, followed by Ancef alone) was started in the operating room and typically was maintained until wound coverage. Intraoperative deep wound culture specimens were initially obtained in thirty-five patients (85%). Of these thirty-five patients, twenty-one (60%) had initial positive cultures, and sixteen of these twenty-one patients with initial positive wound cultures were managed with different and/or additional antibiotics (most commonly imipenem or amikacin) tailored to initial culture results and antibiotic sensitivities. Antibiotic-impregnated cement beads (1 g of vancomycin and 1.2 g of tobramycin per 40 g of cement) were utilized during debridement procedures in thirty-seven patients. Definitive fracture fixation occurred at a mean of 12.3 days (range, zero of thirty-one days) after injury. Postoperative rehabilitation was according to the preference of the surgeon on the basis of fracture pattern, fixation quality, and comorbid injuries. The mean number of total operating room encounters for treatment of the open proximal femoral fracture was 10.3 procedures (range, two of thirty-four procedures), which include a mean of six procedures prior to definitive stabilization and four procedures after definitive stabilization for ongoing soft-tissue management and/or to treat complications. We had a minimum two-year follow-up evaluation in thirty-nine (95%) of forty-one patients with a mean follow-up time of 56.5 months (range, twenty-four to eighty-four months).
Of the forty-one patients, forty underwent provisional fixation of the fractures on the day of injury (thirty-nine patients had external fixation and one had a skeletal traction pin), and one directly underwent definitive fixation (a displaced femoral neck fracture initially treated with open reduction and internal fixation utilizing four cannulated screws). Of these thirty-nine patients, seventeen (44%) underwent external fixation that spanned the hip joint (Fig. 1) and twenty-two (56%) underwent monolateral external fixation with half-pins in the femoral neck of the proximal fracture fragment and half-pins in the femoral shaft distally.
Definitive treatment of the fractures was performed predominantly in our patient group of forty-one patients with a cephalomedullary nail or reconstruction nail in thirty-four patients (83%) (Fig. 2). Other fixation methods utilized included using a proximal femoral locking plate (two patients [5%]), spanning pelvifemoral external fixation (two patients [5%]), open reduction internal fixation with cannulated screws (one patient [2%]), hip fusion (one patient [2%]), and hip disarticulation (one patient [2%]). No patients underwent primary bone-grafting at the time of definitive fixation. For patients who underwent provisional external fixation, conversion from external to internal fixation was performed in a single stage. At the time of internal fixation, the extremity underwent a sterile surgical preparation, the external fixation pins were removed, and the pin tracts were curetted. The used instruments were then removed from the surgical field. The extremity had an additional sterile preparation and drape, and conversion to internal fixation was performed. None of the external fixation pins exhibited obvious signs of infection before conversion to internal fixation.
Complete data to assess fracture union were available in thirty-eight patients (93%), as one patient underwent hip disarticulation and two patients were lost to follow-up one month after definitive fixation. The mean time to union in this group was 5.0 months (range, 2.8 to 16.0 months). The mean time to union for subtrochanteric and pertrochanteric fractures (thirty-five patients) was 5.1 months (range, 2.8 to 16.0 months) (Fig. 3).
Associated Amputations
At the time of the latest follow-up, lower-extremity amputations (proximal to the ankle joint) had been performed in twelve (29%) of forty-one patients for a total of sixteen lower-extremity amputations. Traumatic amputations occurred in twelve limbs in nine patients. The traumatic amputations were subdivided into three bilateral lower-extremity traumatic amputations and six contralateral lower-extremity traumatic amputations. Six patients had amputations that occurred ipsilateral to the open femoral fracture. These amputations were related to the open fracture injury in two patients (one patient undergoing hip disarticulation as definitive treatment in a Type-IIIC injury with acute complicated wound infection and sepsis, and one patient undergoing a delayed transtibial amputation for a dysesthetic foot secondary to a concomitant sciatic nerve injury). The remaining four patients had ipsilateral amputations for lower-extremity injuries sustained distal to the open femoral fracture wound (three traumatic transtibial amputations and one delayed amputation for failed limb salvage of open hindfoot fractures).
Complications
Twenty-nine complications requiring reoperation were observed in twenty-two (56%) of thirty-nine patients (Table I). The complications included infection in twelve patients (31%), symptomatic heterotopic ossification requiring excision in ten patients (26%), malunion requiring osteotomy and revision open reduction and internal fixation in one patient (3%), malreduction requiring revision open reduction and internal fixation in one patient (3%), nonunion requiring delayed bone-grafting in one patient (3%), flap failure requiring a repeat flap procedure in one patient (3%), hematoma in one patient (3%), wound dehiscence in one patient (3%), and a heel decubitus ulcer from an insensate foot requiring debridement in one patient (3%).
Infection occurred after definitive fixation in five (18%) of twenty-eight Type-IIIA fractures, four (67%) of six Type-IIIB fractures, and three (60%) of five Type-IIIC fractures. When comparing the Type-IIIA fractures (five of twenty-eight [18%]) with the Type-IIIB and IIIC fractures combined (seven of eleven [64%]), patients who had Type-IIIA fractures demonstrated a significantly lower rate of infection (p = 0.009). The mean time to diagnosis of postoperative infection was twenty-nine days after fixation, with a median of sixteen days (range, zero to 174 days). As may be expected, the total mean number of operative encounters for proximal femoral fracture treatment per patient was significantly higher in those patients with postoperative infection (17.2 operative encounters) compared with those patients without infection (7.4 operative encounters) (p = 0.0002).
Comparing available intraoperative culture data by the Gustilo-Anderson classification of the fractures, positive cultures were present in fourteen (56%) of twenty-five patients with Type-IIIA fractures and seven (70%) of ten patients with Type-IIIB and IIIC fractures (p = 0.36). When evaluating initial culture results compared with the development of infection, positive cultures were present in eleven of twelve patients who developed infection and in ten of twenty-three patients who did not develop infection (p = 0.006). Presented another way, a negative wound culture on presentation had a 93% negative predictive value for development of infection in our patient population, but a positive wound culture only carried a positive predictive value of 52%. The organisms isolated from the twelve intraoperative wound culture specimens in patients with postoperative infection were gram-negative in eight cultures (two Klebsiella pneumoniae, two Escherichia coli, two Pseudomonas aeruginosa, and two Acinetobacter baumannii), gram-positive in one culture (Enterococcus faecium), and polymicrobial gram-negative in three cultures (two Klebsiella and Escherichia coli, and one Pseudomonas and Candida yeast species).
Intramedullary Nail Fixation
Thirty-two patients were managed with an intramedullary nail fixation for definitive fixation, with >2-year follow-up evaluation available (twenty-four Type-IIIA, six Type-IIIB, and two Type-IIIC fractures). The time to union in the intramedullary nail fixation subgroup was 5.1 months (range, 2.8 to 16.0 months). Postoperative hip-motion data were available in fifteen of the thirty-two patients who had undergone intramedullary nail fixation with a mean arc of extension-flexion of 95°. Complications were encountered in nineteen (59%) of thirty-two patients throughout the course of treatment. Infection was the most common complication, occurring in ten (31%) of thirty-two of all patients who had undergone intramedullary nail fixation (Table II). The total mean number of irrigation and debridement procedures performed prior to definitive intramedullary nail fixation was 6.0 procedures for patients who developed infection compared with 5.6 procedures for those who did not develop infection (p = 0.75). The mean time from injury to the time to definitive intramedullary nail fixation was 14.8 days (range, two to thirty-one days) for patients who developed infection, compared with 13.0 days (range, zero to thirty-three days) for those who did not develop infection (p = 0.57). After intramedullary nail fixation, the mean time to union was 6.5 months for patients who had postoperative infection, compared with 4.6 months for those who did not have infection (p = 0.040). Of the ten patients who had undergone intramedullary nail fixation and had developed infections, two had superficial infections, requiring irrigation and debridement and a short course of ten to fourteen days of antibiotics to clear the infection, and eight had deep infections and/or osteomyelitis, requiring serial irrigation and debridements and organism-specific parenteral antibiotics for at least six weeks. Patients with Type-IIIB and IIIC fractures combined had an infection rate of 63%, and patients with Type-IIIA fractures alone had an infection rate of 19% (p = 0.042). Five patients with deep infection required intramedullary nail removal to clear the infection.
Open proximal femoral fractures are severe injuries often associated with other major trauma. Our patient group, like the one described by Miric et al.7, had a high rate of concomitant injuries, including ipsilateral fractures, lower-extremity amputations, and associated neurovascular injuries, demonstrating the severity of the global injury sustained with these open fractures. Likewise, the peripheral nerve injury rate of 34% in our study was similar to the sciatic nerve injury rate of 30% seen in their 2002 study7, with recovery of nerve function occurring in only one patient in each study. We also demonstrated that obtaining fracture union can be accomplished without the need for initial bone-grafting in these extremely comminuted fractures, regardless of the fixation method utilized. However, in their study, Miric et al. evaluated the use of pelvifemoral external fixation to definitively treat open wartime proximal femoral fractures. Their reasoning for choosing definitive external fixation was that they noted that intramedullary nail fixation carried too high a risk for nonunion, collapse, and infection, although no supporting data for that opinion were presented. Miric et al. reported obtaining union via hip-spanning external fixation in all of their patients, but they also noted universal postoperative stiffness, most likely due to the long period of immobilization and time to union, at a mean of 11.5 months. Those results contrast with the results of our intramedullary nail fixation patient group, which demonstrated a time to union at a mean of 5.1 months and began hip and knee motion immediately after wound closure or as soon as other comorbid injuries allowed. Our reported time to union was very similar to previously reported civilian data describing the treatment of open femoral shaft fractures by means of intramedullary nail fixation9.
Another difference between the two studies, besides the method of fixation of the fractures, was that all seventeen fractures in the Miric et al. study7 were Gustilo-Anderson Type-IIIA injuries; in contrast, we reported on the treatment of six Type-IIIB fractures and five Type-IIIC fractures, in addition to thirty Type-IIIA fractures. Gustilo-Anderson Type-IIIB and IIIC fractures often pose an even greater treatment challenge because of the extensive soft-tissue injuries, need for flap coverage and/or vascular repair, reconstruction, and increased risk of infection9,12,15. These devastating wounds require extreme diligence and serial meticulous, aggressive debridements to remove all devitalized tissue as the wound continues to evolve after injury. In our study, patients with Type-IIIB and IIIC fractures treated with intramedullary nail fixation had a substantial infection rate of 63%, but it is unknown what the infection rate would have been in these same patients had the injuries been treated with a different mode of fixation.
In our study, the infection rate was 19% for patients with Type-IIIA fractures treated with intramedullary nail fixation. Additionally, the majority of our patients sustained blast injuries rather than gunshot wounds, which were primarily treated in the pelvifemoral external fixation study7. Although high-energy gunshot wounds may obviously cause severe injuries, a wartime blast trauma has the potential of causing even more soft-tissue injury and contamination than an isolated gunshot wound15. As all of these details may help to explain the higher infection rate seen in our study compared with that in the pelvifemoral external fixation study7, two possible risk factors that did not demonstrate any correlation with the development of infection in our intramedullary nail fixation patients were the total number of debridements received prior to definitive fixation and the time from injury to definitive fixation.
The majority (60%) of the patients in this study presented with initial positive deep wound cultures. Although a positive culture result in our study demonstrated a sensitivity of 92%, it only showed a positive predictive value of 52% for the development of infection. However, our data may suggest that a negative initial culture result may be more clinically useful as it represented a negative predictive value of 93% for the development of infection in our patient population. Also of note, most reports have demonstrated that gram-positive organisms are the most common cause of postoperative wound infection16,17. However, gram-negative organisms grew on culture of specimens obtained from eleven of twelve patients at the time of debridement. We believe that this discrepancy may be due to several factors, including the high-energy mechanisms with gross wound contamination at the time of injury, the proximity of the fracture wounds to the groin and anus, and associated polytrauma.
Not surprisingly, the mean number of operative encounters required was much higher in patients who developed infection (17.2 encounters) compared with those who did not (7.4 encounters). However, despite requiring more operative procedures (including intramedullary nail removal in five patients) and a mean of 1.9 additional months to produce fracture union, all patients who had development of infection had successful treatment of the infection, with no new recurrences of infection at the time of the latest follow-up evaluation. Similar to previously reported data on open femoral shaft fractures treated with intramedullary nail fixation from civilian trauma, our data suggest that proper surgical debridement and eventual intramedullary nail removal in selected cases are an effective method to eradicate deep infection when it occurs9,18. Although intramedullary reaming and intramedullary nail fixation can be a potential mechanism to spread infection along the intramedullary space, the deep infection and/or osteomyelitis encountered in our patients was seen uniformly at the open fracture site and was not localized distally in the intramedullary canal.
There were several limitations to our study. As a retrospective study of a rare wartime trauma injury in U.S. service members, no standardized treatment regimen was instituted and no control group was generated for comparison. Only two patients were managed with pelvifemoral external fixation to union, and, because of this small sample size, statistical analysis was not appropriate. Throughout the course of this study, there were seven attending surgeons who definitively managed patients with these injuries. The surgical and rehabilitative protocol was at the discretion of the attending surgeon, not allowing for the evaluation of a single treatment method, but possibly making the results of this study more generalizable for various surgeons. Finally, limited functional outcome data were obtained in this patient group regarding lower-extremity or global functional status.
Cephalomedullary nail fixation of open Type-IIIA wartime subtrochanteric and pertrochanteric femoral fractures can be reliably used to effect fracture union in a timely manner, with a mean time of 5.1 months. Open Type-III wartime proximal femoral fractures occur from high-energy mechanisms and are often associated with trauma to multiple organ systems as well as considerable morbidity and complications. Because of the increased risk of deep infection in treating open Type-IIIB and IIIC injuries, especially for those patients with positive initial wound cultures, thorough patient counseling should be undertaken and caution should be exercised in managing these patients with intramedullary nail fixation. Further research initiatives are required to evaluate the functional outcomes obtained after treating wartime open proximal femoral fractures, as well as the means of potentially reducing postoperative infection in these devastating injuries.
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.