In the current study, we reviewed the medical records of patients at our institution between July 2004 and December 2015. The inclusion criterion was patients who underwent revision surgery for aseptic atrophic nonunion after IM nailing for management of femoral shaft fracture. Patients who previously underwent other osteosynthesis surgeries, such as plate osteosynthesis surgery, as initial management for femoral shaft fracture were excluded from the study. Moreover, patients who were suspected of having subclinical septic nonunion, underwent limb lengthening procedures during revision surgery, or had additional pathologic fracture were also excluded. In addition, in order to narrow and specify the results of the current study, patients who had nonunion classified as hypertrophic were also excluded. The medical record reviewing process was approved by our Institutional Review Board (No. 201600790B0) and processed by a single investigator (L. P.-J.).
To standardize the collection of data, we adapted several well-accepted criteria for nonunion, including septic or aseptic nonunion, type of nonunion, and anatomical location of nonunion. First, nonunion was defined as (1) a patient with persistent pain at the fracture site at least 6 months after the primary osteosynthesis surgery; (2) a fracture without complete healing at 6 months on radiographic examination; or (3) a lack of progressive healing for 3 consecutive months on radiographic follow-up [17].
Second, the current retrospective study aimed to review the outcomes of treatment for femoral shaft nonunion without infection; therefore, patients who had femoral shaft septic nonunion were excluded from the study. Diagnosis of septic nonunion was based on intraoperative tissue biopsy from a specimen obtained from non-united ends of the femoral shaft. There were 3 sets of tissue biopsies for each nonunion during surgery, and patients were enrolled in the study if all results were negative for bacterial growth.
Third, the criteria for nonunion were consistent with previous studies [18]. Hypertrophic nonunion referred to a fracture line persisting beyond the expected time for union, with callus in variable amounts about the fracture site on radiographic examination. On the contrary, atrophic nonunion referred to a fracture line persisting beyond the expected time for union, with no demonstrable callus on radiographic examination.
Last, the anatomical definition of the femoral shaft was defined according to the Arbeitsgemeinschaftfür Osteosynthesefragen (AO) classification, which refers to the area from the lower edge of the lesser trochanter to the upper border of the trans-epicondylar width of the knee. The anatomical location of femoral shaft nonunion was further divided into isthmic and non-isthmic [19].
The surgical techniques of ERN and AAP followed the descriptions in previous literature [20,21,22,23]. In the ERN group, the former IM nail was removed through a previous surgical wound. Then, the femoral medulla was prepared using a reaming technique (Fig. 1). We chose a new nail that was 1 to 2 mm larger than the previous nail, according to the size of the femur. In addition, all locking screws were placed in a static position. During the study period, there were 3 different antegrade femoral nails available at our institution (2004–2010: Russell-Taylor femoral interlocking nail; Smith & Nephew, Memphis, TN; 2010–2014: M/DN nail; Zimmer-Biomet, Warsaw, IN; since 2014: King Bo femur interlocking nail; Syntec Scientific Co, Changhwa, Taiwan). The choices of applied femoral interlocking nail during each period were based on the introduction policy in our hospital. No additional bone grafts were applied over the nonunion site.
In the AAP group, the former IM nail was left in place, whether or not the nail was broken. A new incision, usually 10 to 12 cm in length, was made over the nonunion site. After debridement, the interposed tissue between the nonunited ends was decorticated until bleeding, and a broad dynamic compression plate (DCP; DePuy Synthes, Johnson & Johnson, Raynham, MA) was applied (Fig. 2). A plate with appropriate length was chosen and fixation was achieved with compression cortical screws through near-all-cortex purchases just next to and passing by the IM nail. Autologous cancellous bone graft was harvested from the iliac crest and implanted over the nonunion site after decortication in all cases.
The primary outcome was bony union after surgery, with the endpoints of evaluation being either nonunion becoming united or any new surgical intervention being performed. Clinically, union was defined as the patient’s full-weight ambulation without pain or discomfort. Radiographically, union was defined as a continuous cortex in 3 of 4 cortices on anteroposterior and lateral radiographs.
Data were analyzed using SPSS version 18.0 statistical software (SPSS Inc., Chicago, IL). Nonparametric variables were compared using the Pearson chi-squared or Fisher exact test, while continuous variables were analyzed using the student t test. The level of significance was set at P ≤ 0.05.