Keywords
Femur fractures; Fracture healing; Trauma mechanism
Abstract
Introduction: Nonunion rates of distal femur fractures range between 10 and 20%. Previous studies have tried to identify parameters that predict impaired bone healing. These factors include local changes after major trauma such as open fractures and highly comminuted fractures. In addition to these local factors, increasing evidence suggests that the systemic inflammatory response induced by major trauma also impairs bone regeneration. We retrospectively studied patients with distal femur fractures and aimed to identify parameters that predict impaired fracture healing.
Patients and methods: All patients with distal femur fractures treated at a level one trauma center and two large teaching hospitals with locked plating techniques between January 2007 and December 2014 were included. Using multivariable logistic regression, we determined which parameters were independent predictors of impaired fracture healing.
Results: A total of 58 fractures in 56 patients were analysed. 26 fractures were caused by high-energy trauma (45%) and 26 patients developed impaired healing (45%). Impaired fracture healing occurred more frequently after High Energy Trauma (p<0.001), open fractures (p<0.001), comminuted fractures (p=0.001) and in younger patients (p<0.001). High Energy Trauma remained an independent predictor of impaired fracture healing when open fractures and comminution were included in the multivariable logistic regression.
Conclusion: High energy trauma, open fractures and comminution were all identified as independent predictors of impaired fracture healing. This indicates that high energy trauma, regardless of the fracture type that results, may negatively affect fracture healing.
Introduction
Fractures of the distal femur account for 3-6% of all femur fractures and less than 1% of all fractures [1-4]. They occur in a bimodal distribution of high-energy trauma in younger patients, mostly men, and low-energy trauma in the elderly, mostly women. The gold standard in treating distal femur fractures is internal fixation, which yields good functional outcome [4-7]. Internal fixation is achieved through locked plating and to a lesser extent through intramedullary nailing, both preferably performed in a minimally invasive manner to preserve the fracture vascularization and improve outcome in fracture healing [8].
Since the introduction of lateral locked plating this technique has become increasingly popular since the first publications showing promising levels of union ranging between 90-100% [8-14]. However, later studies that include more complex fractures due to High Energy Trauma showed higher nonunion rates of up to 20 % [4,15].
Factors that increase the risk of impaired fracture healing include open fractures, comminuted fractures, diabetes and increased Body Mass Index (BMI) [4,6,16]. The trauma mechanism is another risk factor that has received increasing attention over the last years. Although the impairment of fracture healing as seen in High Energy Trauma is directly related to an increased incidence of open and comminuted fractures, there is increasing evidence that the systemic inflammatory response caused by major injury negatively affects the outcome of fracture healing [17-19].
Based on these findings, we hypothesized that High Energy Trauma is an independent predictor of impaired fracture healing, in addition to open and comminuted fractures. We therefore retrospectively studied patients with distal femur fractures and determined which clinical and fracture related parameters correlate with the outcome of bone regeneration.
Patients and Methods
A retrospective cohort analysis has been performed of all patients aged 18 and older undergoing locked plating for distal femur fractures between January 2007 and December 2014 at one level I trauma center and two level II trauma centers. Patient demographics, trauma mechanism, fracture characteristics, surgical technique and follow-up were retrieved from the medical records. Exclusion criteria were periprosthetic fractures, pathological fractures or follow-up duration less than 9 months. The primary endpoint was time to fracture healing. Fracture healing was defined as full weight-bearing without pain.
Time to healing was defined as the time in weeks between injury and the first documented time at which patients were able to bear full weight without pain. Normal healing was defined as a healing time not exceeding six months. Impaired healing was defined as no healing at six months after injury. Postoperative complications in healing were scored, including deep infections, malunions, and osteosynthesis-related complications requiring secondary surgical intervention. The trauma mechanism was either defined as High Energy Trauma (HET) or Low Energy Trauma (LET).
High-energy trauma included traffic accidents or falls from height. Fractures caused by fall from a standing position or sports injuries not involving motorized vehicles were classified as low-energy trauma. All fractures were classified according to the AO/OTA system [20]. To improve interobserver agreement fractures were not subdivided beyond A, B and C [21]. The difference between diaphyseal (shaft) and metaphyseal (distal) fractures of the femur was determined by applying the rule of squares as described by Müller [22]. The influence of all parameters on the outcome, fracture healing, was investigated as follows.
First, fracture healing was categorized as normal healing (healing within six months after injury) or impaired healing (no healing after six months, both delayed union and nonunion). By comparing the characteristics of the two groups (normal vs. impaired healing) using the tests described below, those factors that showed a significant difference were selected and subsequently included in a multivariable logistic regression model to determine which parameters were independent predictors of impaired fracture healing.
Statistical analysis was performed with IBM SPSS 22 (IBM SPSS for Windows version 22.0, Armonk, NY). Two-tailed pearsons’ Chisquare was used to compare categorical variables and an independent T-test or Mann-Whitney U test were used for continuous variables depending on whether the data had a binomial distribution. A p-value <0.05 was considered statistically significant (Table 1).
Table 1: Statistical analysis performed with IBM SPSS 22.
Results
Between January 2007 and December 2014, 96 patients with a distal femur fracture were treated with locked plating in the three participating centers. Two patients suffered bilateral distal femur fractures after high-energy trauma. Forty patients were excluded; 19 patients were excluded since their fractures were periprosthetic, 9 patients due to pathological fractures, 1 patient due to a lower leg amputation, 1 patient due to paraplegia, 1 patient moved abroad, 1 patient was lost to follow up and 8 patients died within 6 months after injury and were therefore excluded. A total of 56 patients with 58 fractures were available for final analysis.
Thirty-two fractures were caused by low-energy trauma (55%) and 26 fractures (45%) followed high-energy trauma. Thirtythree fractures were comminuted (57%) and 15 fractures were open fractures (26%). Prior to definitive treatment; an external fixator was used for temporary stabilization in 11 cases. In all patients the external fixator was replaced by definitive fixation within 14 days after injury.
When classified by the AO-OTA system there were 28 type A, 3 type B and 27 type C fractures. Thirty-two fractures healed adequately within 6 months after injury (55%). The 26 fractures that did not heal within 6 months (45%) included 9 delayed unions and 17 nonunions. The mean healing time was 8 months (±7.6) for the entire cohort. Patients following highenergy trauma healed significantly longer than those after low-energy trauma (11 vs. 6 months, p=0.030).
There were no postoperative complications in 37 of the patients (64%). In the remaining patients, 17 patients required secondary surgery due to non- or malunion, 2 patients suffered from leg-length discrepancies, one developed a loose screw without malalignment and one temporary loss of peroneal nerve function. Open fractures occurred more frequently in the impaired healing group compared to fractures that healed adequately within 6 months after injury (50% versus 6%, p<0.001).
Comminuted fractures were also more prevalent in the impaired healing group (81% versus 38%, p=0.001). Within the impaired healing group 19 fractures were caused by high energy trauma, compared to 7 following Low Energy Trauma (73% versus 27%, p<0.001). Patients with impaired fracture healing were significantly younger than patients with normal fracture healing (47 years versus 69 years, p<0.001). Binary logistic regression analysis showed that when including comminution, open fractures and trauma mechanism in the model, high-energy trauma remained an independent predictor of impaired fracture healing with an odds ratio of 4.3 (Table 2).
Table 2: Binary logistic regression analysis.
Discussion
We found that high-energy trauma remained an independent predictor of impaired fracture healing of the distal femur when open fractures and comminution were added to the multivariable logistic regression with an odds ratio of 4.3. Of the fractures that were analyzed, 45% showed impaired healing. Factors that were associated with impaired fracture healing were age (p30), open fractures, infection and stainless steel implants. In a retrospective review by Ricci et al. [27] a nonunion rate of 19% was found in 335 distal femur fractures.
They identified diabetes and open fractures as independent predictors of nonunion. Although some of these studies have identified open fractures and comminution as risk factors for impaired healing, trauma mechanism was identified as an independent risk factor in our study following binary logistic regression.
This increases the likelihood of a causal relationship between High Energy Trauma and impaired healing. In the interpretation of the results of the current study several limitations must be kept in mind. Firstly, the small study population of 58 fractures increases the risk of a type II error. Due to our small study population we were unable to include all 4 parameters into the logistic regression that were significantly different between normal and impaired fracture healing. We therefore excluded age from the regression since we believed that this parameter represented or closely correlated with trauma mechanism (e.g. young men are seen more often in de HET group and older women mostly within the LET group).
Poor fracture healing in younger patients also seems counterintuitive from a biological perspective. Ricci et al. [27] also found a higher incidence of nonunion in closed fractures in patients younger than 65 years and they also suggested this was due to the severity of the trauma, since younger patients represent a large part of the highenergy trauma group. In addition, the limited number of cases from 3 different centers is also a form of selection bias.
Secondly, the retrospective design of the study increases the risk of bias. Most of the patients were seen in the outpatient clinic at standard follow-up times of 6 weeks, 3, 6 and 12 or 15 months after surgery. This may have influenced the reported healing rates since the majority of the patients healed between 6 and 12 months. In addition, we defined healing as the first documented moment of pain-free full weight bearing. This may explain the difference in healing time mentioned in other studies, where other types of healing were defined.
Defining healing from a patient’s perspective was deliberately chosen as this reflects patient’s activities rather than, for example, radiological healing. Lastly, a remarkable finding in this study is the high number of patients with impaired fracture healing. The nonunion rates previously reported range from 10-20% [13,15,25,27] whereas in our study population 45% of the fractures resulted in impaired fracture healing.
To our knowledge this might be due to the following: (1) our wide definition of fracture healing combined with the retrospective design of the study (information bias), (2) the wide definition of impaired fracture healing in which not only non-union but also delayed union is included and (3) the rigid fixation generated by locking plates might as well cause a higher number of delayed- or nonunion [30]. In conclusion, our study shows that factors associated with impaired fracture healing of the distal femur are age, open fractures, comminuted fractures and high-energy trauma.
Logistic regression analysis showed that high-energy trauma remains an independent predictor of impaired fracture healing when open and comminuted fractures are added to the same model. Major trauma may therefore impair bone healing through an additional mechanism other than increased incidence of open and comminuted fractures, for instance by inducing a detrimental systemic inflammatory response.
References
1. Court-Brown CM, Caesar B. Epidemiology of adult fractures: a review. Injury. 2006; 37: 691-697.
Citation
Karhof S, Bastian OW, van Olden GDJ, Leenen LPH, Kolkman KA and Blokhuis TJ. Impaired Fracture Healing of the Distal Femur after High Energy Trauma. SM J Arthritis Res. 2017; 1(1): 1003.