Study design and sample

To address the research objectives, a retrospective cohort study was designed and implemented. The sample was derived from the patients treated by one surgeon (JCP) in a private practice setting, with surgery performed at a single hospital (MedStar Georgetown University Hospital, Washington, DC, USA) between 2004 and 2013. Subjects had undergone a minimum of 2 years follow-up at the close of the study (range 2–11 years). The study sample included subjects with a spectrum of bimaxillary developmental DFDs also involving the chin and with symptomatic chronic obstructive nasal breathing. The subjects then underwent a minimum of Le Fort I osteotomy, bilateral sagittal ramus osteotomies (SROs) of the mandible, osseous genioplasty, septoplasty (submucosal resection), and reduction of the inferior turbinates. Patients were excluded if their jaw deformity had previously been operated upon, or was syndromic, cleft-related, post-traumatic, or tumor-related. Patients not residing in North America were also excluded, as long-term follow-up was geographically inconsistent. Other standard exclusions for semi-elective orthognathic surgery per routine included the use of nicotine products for at least 3 weeks prior to surgery, current use of bisphosphonate medication, immunosuppressed status, and insulin-requiring diabetes. All subjects were confirmed to be cardiovascularly stable without pulmonary disease, renal disease, or a known coagulation disorder. The Georgetown University Institutional Review Board approved the study protocol.

Predictor variables

The predictor variables were grouped into the following categories: demographic, anatomical, operative, and complications.

The demographic variables collected included age of the patient at operation and sex.

The anatomical variables studied included the pattern of the developmental DFD, location of jaw osteotomy, and presence of a mandibular third molar undergoing simultaneous extraction. With regard to the pattern of the developmental DFD, each subject was classified at presentation into one of six common jaw growth patterns: primary mandibular deficiency, maxillary deficiency with relative mandibular excess, asymmetric mandibular excess, short face, long face, and bimaxillary dental protrusion. In addition, there was an atypical category that included DFDs influenced by parafunctional oral habits such as thumb sucking. The second anatomical variable was the site of the osteotomy requiring fixation; these included the maxilla at the Le Fort I level, the ramus of the mandible in association with an SRO, and the symphysis of the mandible (osseous genioplasty). The third anatomical variable was the presence of a mandibular third molar requiring simultaneous removal. This was further categorized as fully impacted (no exposure through the oral mucosa) or erupted/partially erupted (exposure through the oral mucosa).

Operative variables studied included the method and type of osteotomy fixation and the method of mandibular third molar management. The Le Fort I osteotomy fixation was with four titanium L-plates (Stryker Corporation, Kalamazoo, MI, USA) . If the degree of maxillary advancement and lengthening created a defect that required an interpositional graft, then an additional microplate was placed across each inset graft . The SRO was fixated with three bicortical titanium screws of 2.3 mm in diameter (Stryker Corporation) . The length of the screws varied between 14 mm and 18 mm according to the width and separation of the bones. The three screws were placed at locations where bicortical bone overlap was best, but also with the intention of avoiding dental roots, maximizing the distance between screws, and avoiding placement of the screws too superiorly (i.e., close to the alveolar crest) . Occasionally, if fixation with three screws was deemed inadequate, either additional screws or a plate with monocortical screws was placed to achieve fixation. The ramus osteotomy fixation was placed through a transbuccal trocar. Chin osteotomy fixation was with two surgeon-contoured straight titanium plates (Stryker Corporation) . Each plate crossed the osteotomy and was secured on each side with one screw (1.7 mm diameter and 4 mm length). If the extent of chin lengthening created a defect requiring an interpositional graft, an additional plate was secured across the graft . If for reasons of long-term dental health, extraction of a mandibular third molar was required, it was the senior surgeon’s approach (JCP) to remove it at the time of orthognathic surgery . If the mandibular third molar was fully impacted, it was removed through the osteotomy site after splitting the ramus. If the mandibular third molar was partially or fully erupted, it was removed after the ramus osteotomy incision and soft tissue dissection and just prior to splitting of the mandible . This was to limit unintended fracture of the lingual plate.

The surgical complications studied as potential risk factors for the need to remove fixation hardware included partial wound dehiscence, the presence of a SSI, lingual nerve injury, and dental injury.

Outcome variables

The primary outcome variable studied was the need to remove fixation hardware after surgery. This variable was confirmed through a review of the medical records (hospital and outpatient) and radiographs (prior to and after surgery). All study subjects had undergone a minimum of 2 years of clinical follow-up (direct face-to-face visit or telephone communication) after orthognathic surgery at the close of the study (range 2–11 years).

Collection, management, and analysis of data

The data were abstracted and recorded on a standardized data collection form by two researchers/observers (EC, AC) from the subjects’ hospital and outpatient medical records, including a review of standard interval facial radiographs. Lateral cephalometric and panoramic radiographs were obtained for each subject approximately 2 months before surgery and 5 weeks after surgery. The postoperative radiographs were reviewed to document the location and type of fixation, removal of third molars, injury to dental roots, and presence of foreign bodies. The data were entered into a database created using Microsoft Access (Microsoft Inc., Redmond, WA, USA). The database was transferred to a software package for statistical analysis (Stata 13.1; StataCorp LP, College Station, TX, USA).

Descriptive statistics were computed for all study variables. Bivariate analyses were computed to measure any associations between predictor variables and the need to remove fixation hardware. The level of significance α was set at 0.05 for a 95% confidence interval. With α set at 0.05, the probability of incorrectly rejecting a true null hypothesis (type I error) was limited to 5%. Power analysis and the probability of incorrectly accepting a false null hypothesis (type II error) could not be calculated because the population mean, from which the present sample was derived, is unknown. When meaningful, logistic regression analysis was conducted to evaluate possible confounders and modifiers. No subjects were lost to follow-up and no records or data points were missing for any of the subjects for all parameters in the study.