Study design

This study was a review of a consecutive case series of 99 patients, undertaken in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. The STROBE statement is a checklist, established by “an international, collaborative initiative of epidemiologists, methodologists, statisticians, researchers and journal editors involved in the conduct and dissemination of observational studies”. The checklist allows for a complete and consistent approach to reporting research, which encourages an adequate peer review process.

Setting

This study was undertaken in the oral and maxillofacial unit of the study hospital in Brisbane, Australia. Patients who received surgical treatment for zygomatic fractures during the period 17 January 2011 to 16 January 2012 were included in the study. The first author undertook the collection of data during the period May to July 2014.

The oral and maxillofacial unit of the study hospital provides a tertiary referral service for facial trauma for a large metropolitan and regional area that extends to at least 1000 km. Patients with zygomatic fractures can present via the hospital’s emergency department, via referral from a general medical or dental practitioner directly to an outpatient clinic, or through an inter-hospital transfer from regional centres with no maxillofacial service.

The oral and maxillofacial team at the hospital generally advocates for definitive assessment of isolated zygomatic fractures in a non-urgent manner. Patients are usually assessed between 1 and 7 days post-injury in an outpatient setting. Patients with critical complications such as neurological injury or retro-bulbar haemorrhage are seen urgently as required. Reasons for delay are multifactorial, but commonly include geographic isolation and inter-hospital transfer, and drug and alcohol intoxication. Radiographic assessment is mandatory. The modality is at the discretion of the operator, but may include plain film or computed tomography (CT) imaging. Patients requiring surgical management are usually booked for theatre or occasionally added to the hospital theatre emergency board. Patients usually receive surgical treatment between 1 and 14 days post-injury. Consultant surgeons and registrars undertake the surgeries. Different operators employ their preferred techniques, but generally an indirect elevation either via a temporal or buccal sulcus incision is employed to reduce or partially reduce depressed zygomatic fractures. Unstable fractures are plated with varying degrees of fixation. Sites utilized include the zygomaticofrontal suture, the infra-orbital rim, the zygomaticomaxillary buttress, and rarely the zygomatic arch. Plating is facilitated via open reduction and internal fixation (ORIF) peri-orbitally, orally, or via a coronal approach. Utilization of a coronal exposure is employed when adequate local exposure would result in an unacceptable risk of scarring or facial nerve injury.

Unless patient limitations or unforeseen complications arise, postoperative review of all routine maxillofacial trauma surgeries at the facility involved is undertaken at approximately 1 and 6 weeks post-surgery. Review at 1 week allows for the removal of sutures where required, and allows the identification and rectification of early complications. The 6-week review allows for the assessment of healing and restitution of function, and allows the identification of later complications. Additional review appointments are made when clinically warranted.

Participants

All patients who required surgical treatment for a fracture or fractures of the zygoma and/or zygomatic arch, and who were treated in the oral and maxillofacial unit at the study hospital between 17 January 2011 and 16 January 2012, were included in the study. Patients were excluded if they failed to present for review at 1 and 6 weeks postoperatively. There were no children in the sample.

Variables

The following information was collected for each patient: hospital identification number, date of injury, date of operation, delay (days), diagnosis (isolated zygomatic arch fracture, or zygomaticomaxillary complex fracture), operation (indirect elevation, indirect elevation and ORIF via peri-orbital or oral approach, or indirect elevation and ORIF via coronal approach), primary operator, regular alcohol use (yes, no), regular cigarette use (yes, no), trismus (absent, present), facial symmetry (aesthetically ideal, aesthetically not ideal but acceptable, or not acceptable with corrective treatment required), facial scarring (absent, present), and radiographic outcome (equivalent to premorbid, minor deviation from premorbid, major deviation from premorbid).

Data sources/measurement

Preliminary data for all variables except radiographic outcome were collected from a pre-existing database within the oral and maxillofacial unit (the State-wide Maxillofacial Audit Database). This database was maintained by the second author. Following a primary analysis of the database by the first author, it was evident that data for several of the study variables were incomplete. The medical records for each eligible patient were subsequently requested from the hospital medical records department and the variables cross-checked, corrected, and completed. Various treating members of the oral and maxillofacial team undertook the assessment of each of the variables, except radiographic outcome, as part of routine treatment and record-keeping. The fourth author assessed pre- and postoperative radiographs in August 2014.

Bias

The first author was not a member of the oral and maxillofacial unit at the study hospital and as such had no conflict of interest in reporting the unit’s surgical results. The fourth author, who conducted the radiographic outcome review, was the primary operator for only one of the 99 patients included in the study. Furthermore, this author was unaware of the results of the analysis of other variables and as such was able to review the radiographic outcomes independently and without assumption or expectation. The third author, a biostatistician from the School of Population Health at the University of Queensland, who had no professional connection with the oral and maxillofacial unit, or the first, second, or fourth author, conducted the statistical analysis independently.

Cross-checking of data from the State-wide Maxillofacial Audit Database with the historical medical records was undertaken to minimize information bias.

Study size

The study size of 99 patients was arrived at with the preliminary assessment of the State-wide Maxillofacial Audit Database. The State-wide Maxillofacial Audit Database had been completed effectively for a 12-month period spanning 17 January 2011 to 16 January 2012.

Quantitative variable handling

Delay was measured in days, as this was the smallest and most practical increment of time that could be reliably measured from injury to surgery. It was determined that analyzing delay as a continuous variable, rather than delineating arbitrary groups of delay, would elicit the most meaningful information.

The diagnosis of zygomatic type fractures was divided into two groups: zygomaticomaxillary complex fractures and isolated zygomatic arch fractures. Patients with concomitant fractures of a zygomaticomaxillary complex and the ipsilateral zygomatic arch were assessed within the zygomaticomaxillary complex group. The authors concluded that the treatment, prognosis, and potential complications of isolated zygomatic arch fractures were significantly different to zygomaticomaxillary complex fractures and therefore warranted an independent grouping. Restrained by the data available, a single grouping for zygomaticomaxillary complex fractures without the further delineation of features such as fracture subtype and degree of comminution was considered most appropriate.

Operation type was divided into three groups: indirect elevation, indirect elevation and ORIF via peri-orbital or oral approach, and indirect elevation and ORIF via coronal approach. These groups were delineated based on the decidedly dissimilar degree of technical difficulty and potential for complication, as well as for ease of statistical analysis.

The primary operator was recognized as the surgeon who primarily performed the operation, regardless of level of expertise or level of supervision.

Due to the difficulty in accurately quantifying alcohol and cigarette use, and the undefined influence of different levels of use in different populations, it was decided that alcohol and cigarette use would be analyzed in two groups: regular use and not regular use. The identification of regular use for both alcohol and cigarettes was via a routine hospital questionnaire completed by all patients upon admission.

Trismus was evaluated as either present or absent at the final outpatient review appointment, which was regularly undertaken at 6 weeks postoperatively. Trismus was noted as present when there was an obvious restriction in mouth opening, or the patient complained of subjective jaw stiffness. Facial symmetry relative to the zygomaticomaxillary complex and zygomatic arch was evaluated clinically at the final outpatient review appointment by the reviewing clinician, and the results were divided into three groups: aesthetically ideal, aesthetically not ideal but acceptable, and not acceptable with corrective treatment required. Facial symmetry was measured by direct visualization of the patient’s facial projections from in front and above. The contralateral malar projection was used for comparison and the patient’s opinion was sought. It was recognized that difficulty with clinical comparison would likely arise with premorbid facial asymmetry or bilateral injury. Fortunately bilateral zygomatic fracture did not arise within this series.

Facial scarring, specifically as a result of surgical incision, was evaluated at the final outpatient review appointment, and the results were originally divided into three groups: no scarring present, scarring present (visible to the naked eye) with no corrective treatment required, and scarring present with corrective treatment required. Corrective treatment was identified by a return to theatre primarily for the purposes of scar revision surgery. Only one patient from 99 required corrective treatment for scarring. To avoid statistical complications it was decided to adjust to two groups: scarring present and scarring absent.

The fourth author evaluated the radiographic outcome in August 2014. The hospital identification number for each of the 99 patients was recorded in isolation in a spreadsheet by the first author and given to the fourth author. The fourth author, a consultant oral and maxillofacial surgeon, was able to access pre- and post-surgical radiographs for 97 of the 99 patients included in the study. Outcomes were divided into three groups: equivalent to premorbid, minor deviation from premorbid, and major deviation from premorbid. Equivalency to the premorbid state was determined by the fourth author’s personal judgement, such that the fracture segments appeared in the most appropriate position with respect to the remainder of the bony mid-face and the fracture lines themselves. Radiographic comparison with the contralateral structures was undertaken, but not completely relied upon.

Statistical methods

Stata version 13.1 software (StataCorp LP, College Station, TX, USA) was used to perform all statistical analyses. Multivariable logistic regression was the statistical method of choice to estimate the relationships between the variables collected. Four outcomes/dependent variables were utilized to analyze the effects of treatment timing with respect to zygomatic fractures: facial symmetry, facial scarring, trismus, and radiographic outcome. Each dependent variable was analyzed initially in relation to the primary independent variable ‘delay’. In an attempt to control for potentially confounding variables, five additional independent variables were analyzed for interactions with respect to delay: operation, diagnosis, primary operator, regular alcohol use, and regular cigarette use. All tests were two-sided with P -values of 5%.

Goodness-of-fit for non-ordinal (binomial and multinomial) and ordinal logistic regression models was ensured using the χ ² test and Brant test, respectively.