The ideal timing for treatment of facial fractures has not been well established. The objective of this systematic review was to examine the effects of treatment delay on outcome in the management of facial fractures. The PubMed database was used to search for relevant English-language articles published between 1979 and 2013. Cross-referencing identified additional studies. There were no selection restrictions for study type. The first author, using pre-defined data fields, extracted information independently. Studies were assessed by study type, evidence level, sample size, data collected, outcome variables, control of confounding variables, and findings. Thirty studies were identified. Inconsistency was identified with data collected, outcome variables, and findings. Of the 30 studies identified, 28 were case series, thereby providing a low level of evidence overall. The majority of case series were retrospective and sample sizes were predominantly small. Control of confounding variables was poor. Eighteen studies found no statistically significant relationship between treatment delay and treatment outcome. Nine studies found a statistically significant relationship between treatment delay and worse treatment outcomes. There were three studies with conflicting results. With the current body of evidence, definitive conclusions cannot be drawn on the timing of treatment for facial fractures.
Facial fractures are a common presentation to hospitals worldwide. Their treatment has evolved dramatically in recent decades, particularly since the introduction of open reduction and internal fixation. However, many uncertainties remain. The relevance of delay from injury to treatment is commonly disputed amongst surgeons and in the literature. Intuitively, delaying the treatment of facial fractures could increase the risk of infection, the likelihood of technical difficulties, and the discomfort experienced by patients. For these reasons, treatment delay has historically been minimized where possible. Prominent surgeons such as Champy, Cawood, and Maloney have previously advocated delays from injury to surgery of no more than 24, 48, and 72 h, respectively. However, there are many reasons why delay beyond 72 h may be practical or unavoidable. Managing facial fractures usually involves the administration of a general anaesthetic, surgery, a hospital stay, and a rehabilitation period. Due to the multifactorial nature of the management process, outcomes may be affected by a multitude of factors. Patient factors such as age, medical co-morbidities, mental history, compliance, concomitant injuries, and financial status may be relevant. Additionally, health system factors such as inter-hospital transfer policies, funding, resource allocation, staff training, and availability have an impact.
This systematic review aims to examine broadly the effects of treatment delay on outcomes in the management of facial fractures, by identifying studies of any type that have examined the effects of timing of treatment on outcomes of any type, in the treatment of fractures of the human facial skeleton by widely accepted treatment methods.
Should it be found that treatment timing significantly alters the outcome of facial fracture management, new protocols and recommendations could be proposed. Conversely, a poor correlation may further justify the planning of facial fracture management in accordance with the conveniences of the relevant health care system.
Materials and methods
Studies of any type that examined the effects of timing of treatment on outcomes of any type, in the treatment of fractures of the human facial skeleton by widely accepted treatment methods, were reviewed ( Fig. 1 ). There were no restrictions imposed for participant selection or length of follow-up. Studies dated prior to 1979 were excluded due to the considerable differences in surgical and medical methods of management employed before this time. Studies were assessed by study type, evidence level, sample size, data collected, outcome variables, control of confounding variables, and findings.
The search for studies was developed and conducted by the first author. Studies were identified by an electronic search of the PubMed database from 1 February to 1 April 2013. In addition, cross-referencing was utilized – the reference lists of the studies identified in the preliminary search were checked for additional suitable studies.
Search terms for the preliminary search included the following: facial, fracture, treatment, management, outcome, mandible, maxilla, zygoma, orbit, frontal, nasal, delay, and timing.
An eligibility assessment was performed independently in an un-blinded manner by the first author. Studies were identified as relevant by title and abstract. After identification, full text publications were sourced. The first author reviewed the full text publication of every identified study to determine relevance with respect to the aforementioned pre-determined assessments. Every type of study that was relevant was included; study type was analyzed in the systematic review assessment. Full text publications were available for all identified studies, from both the preliminary search and the cross-referencing process.
Methods for assessing the risk of bias in individual studies were largely deemed impractical by the authors, due to the paucity of quality studies identified. Methods of control of confounding variables were identified and recorded for each study by the first author. Bias and confounding in relation to this systematic review were deliberated further in the discussion.
A total of 30 studies were determined to be relevant for inclusion in the systematic review. The preliminary search of the PubMed database yielded more than 1000 studies. After excluding non-English studies, the remainder were screened by title and abstract. Seventy-eight full text publications were obtained and assessed for inclusion by the first author, with 54 studies being discarded as non-relevant. Six additional studies were identified and included from the cross-referencing process.
Thirty studies were identified, including one systematic review, one prospective randomized controlled trial (RCT), and 28 case series.
The RCT was randomized prospectively for administration of antibiotic therapy. However, the effect of delay was evaluated retrospectively.
The systematic review included in its analysis the aforementioned RCT by Chole and Yee. However, it was not interpreted as an RCT with respect to treatment delay. In addition to the RCT, Hermund et al. identified only five studies for the systematic review that allowed for statistical analysis. Furthermore, Hermund et al. stated that none of the studies allowed for a stratified analysis to control for “confounding factors such as severity of fracture, number of fractures, alcohol or drug abuse, non-compliance or treatment delay because of an already existing infection being neglected by the patient”.
From the 28 case series identified, 24 employed a retrospective analysis. Only three case series employed a prospective analysis. One case series did not identify clearly as either retrospective or prospective.
The smallest sample size of any study reviewed, excluding the systematic review, was 21 patients. The largest sample size was 327 patients. A number of studies had larger sample sizes, but the number of patients was reduced in relation to treatment delay.
Twenty-one studies involved treatment of the mandible in isolation. One study involved treatment of the zygomatic complex. Eight studies involved treatment of multiple facial fractures.
Measure of delay
Large variation was observed with the quantification of delay. Three studies used a continuous scale for delay, measured in days. Six studies calculated mean delay for groups delineated by other variables. For example, Stone et al. calculated the mean delay for patients with and without postoperative complications following operative treatment for mandibular fractures. Eighteen studies divided delay into a maximum of four groups of varying durations, with differing definitions of ‘delayed, ‘immediate’, or ‘early’ treatment. Finally, two studies grouped the entirety of participants into a single group with respect to delay. One such study by Nakamura et al. grouped all 143 participants into a ‘delayed’ treatment group, defined as a delay from injury to treatment of more than 3 days. The group was then compared with ‘delayed’ and ‘early’ treatment groups from three other studies. Another study, by Zachariades et al. delineated delay as more than 14 days.
Other data collected
Several variables were consistently included in the design of the studies reviewed, such as age, gender, fracture aetiology, fracture location, association and disposition of teeth, use of antibiotics, treatment modality, and treatment delay. However, examination of treatment delay was not the primary variable assessed in a number of the studies reviewed. For this reason, not all of the variables were analyzed in respect to treatment delay, and vice versa. Furthermore, an obvious but significant disparity existed within many of the variables assessed over the 34-year period. Other variables that were less consistently assessed but likely to contribute to outcomes included patient compliance, operator experience, concomitant medical illness or injury, substance abuse, and severity of fracture.
Several factors were consistently assessed to determine outcome in the studies reviewed, such as infection, wound dehiscence, malocclusion, malunion, delayed union, and non-union. Other variables that were less consistently assessed included scar formation, nerve damage, need for fixation, masticatory ability, aesthetics, permanent disability, death, and hardware exposure. One study analyzed the variables of postoperative length of stay, length of intensive care unit (ICU) stay, and length of hospital stay to determine outcome. In all but one study, the authors were the evaluators. Uglesic et al. utilized a treatment score system based on both surgeon (author) and patient evaluation as a method of assessment of outcome.
Control of confounding variables
A number of studies utilized restriction when selecting participants in an attempt to minimize confounding variables. For example, Barker et al. excluded patients with condylar fractures. Condylar fractures typically result in a higher proportion of negative outcomes when compared with other mandibular fractures, such that the relationship between treatment delay and outcome may be confounded. One study utilized randomization in relation to the administration of antibiotics. Stratification was utilized by two studies, whilst two studies utilized multivariate analysis to control for confounding variables.
Eighteen studies, including the RCT by Chole and Yee, found no statistically significant relationship between treatment delay and treatment outcome ( Table 1 ).