Facial Fractures: Independent Prediction of Neurosurgical Intervention

Abstract

Context:

Over half of patients with facial fractures have associated traumatic brain injury (TBI). Based on force dynamic cadaveric studies, Lefort type 2 and 3 fractures are associated with severe injury. Correlation to neurosurgical intervention is not well characterized.

Aims:

This study characterizes fracture pattern types in patients requiring neurosurgical intervention and assesses whether this is different from those not requiring intervention.

Settings and Design:

Retrospective data were collected from the trauma registry from 2010 to 2019.

Subjects and Methods:

Patients over 18, with confirmed facial fracture, reported TBI, available neuroimaging, and hospital admission were included.

Statistical Analysis Used:

Retrospective Contingency Analysis with Fraction of Total Comparison was used with Chi-square analysis for demographic and injury characteristic data.

Results:

One thousand and one patients required no neurosurgical intervention and 171 required intervention. The intervention group had a significantly greater number of patients with Glasgow Coma Scale (GCS) <8 compared to the nonintervention group. Subset analysis revealed a twofold increase in Lefort type 2 fractures and notable increase in Lefort type 3 and panfacial fractures in the intervention group. Patients requiring craniectomy, craniotomy, or burr holes were much more likely to have Lefort type 2 or 3 fractures compared to those only requiring external ventricular drains or intracranial pressure monitoring. Subset analysis accounting for GCS supported these results.

Conclusions:

Lefort type 2 and type 3 fractures are significantly associated with requiring neurosurgical intervention. An improved algorithm for managing these patients has been proposed in the discussion. Ongoing work will focus on validating and refining the algorithm to improve patient care.

Introduction

Roughly one of seven trauma patients admitted to the emergency room had maxillofacial fractures.[1,2] Studies have suggested an association between maxillofacial fractures and TBI.[2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20] Depending on the severity, TBI may be difficult to detect using current technology, potentially delaying treatment and worsening prognosis for patients.[4]

A recent study suggested an association between Le Fort type fractures and more severe TBI.[4,21] This is likely due to diffuse axonal injury, epidural, and subdural hematomas secondary to the high-velocity facial trauma required to produce these fractures.[4,22] Despite these findings, little is known about how fracture types predict TBI severity and which patients eventually require neurosurgical intervention. Thus, the present study is designed to develop an improved algorithm for the management of TBI in the context of known facial fractures with a hypothesis that patients with mid-face fractures are at increased risk for severe TBI warranting more aggressive neurosurgical intervention. Furthermore, we grouped by Glasgow Coma Scale (GCS) to look at trends warranting improved management strategies.

Subjects and Methods

The study was submitted for Institutional Board Review at the University of Florida and abided by the highest international ethical research standards. Retrospective analysis of patients from 2010 to 2019 was obtained through our trauma registry. Inclusion criteria were adults over 18, confirmed facial fracture with available neuroimaging, reported traumatic brain injury (TBI), and admission to ICU or floor bed. Exclusion criteria were patients <18 years old, patients with no neuroimaging, and patients that were deceased prior to initiation of neurosurgical intervention.

In addition to basic demographic data, data collected included presenting GCS score, mechanism of injury, facial fracture type, TBI type, and type of neurosurgical intervention. Age was grouped into seven categories: 1 (18–24 years old), 2 (25–34 years old), 3 (35–44 years old), 4 (45–54 years old), 5 (55–64 years old), 6 (65–74 years old), and 7 (>75 years old). The race was grouped into five categories: 1 (Caucasian), 2 (Black), 3 (Asian), 4 (Hispanic), and 5 (Other). Sex was classified as male or female. GCS score was arranged: mild (14–15), moderate (9–13), or severe (8 or less). Mechanism of injury was grouped into seven categories: 1 (assault), 2 (all-terrain vehicle or dirtbike accident), 3 (gunshot wound or knife injury), 4 (bicycle or moped accident), 5 (motorcycle collision or motor vehicle collision), 6 (fall), and 7 (other). Type of TBI: 1 (contusion), 2 (diffuse axonal injury), 3 (epidural hematoma), 4 (subdural hematoma), 5 (traumatic subarachnoid hemorrhage), 6 (intracranial hemorrhage or intraventricular hemorrhage), and 7 (penetrating injury). Additional radiographic findings divided into four categories: 1 (edema), 2 (herniation), 3 (pneumocephalus), and 4 (cerebral/cerebellar laceration).

Patients were divided into those with facial fracture and TBI without neurosurgical intervention and those with facial fracture and TBI with neurosurgical intervention. Graphpad Prism 8.0 Software was used for analysis. Retrospective Contingency Analysis with Fraction of Total Comparison was used with Chi-square analysis for demographic and injury characteristic data. P < 0.05 was considered statistically significant.

Results

Based on the above inclusion/exclusion criteria, 1985 patients were pooled from the overall trauma registry. On further review, 316 were too young, 403 had no TBI, and 94 had no facial fracture. One thousand one hundred and seventy-two patients therefore met the criteria for inclusion into the study. One thousand and one had facial fracture and TBI with no neurosurgical intervention, 171 had facial fracture and TBI with neurosurgical intervention. Table 1 shows baseline demographic data. No significant difference was seen between groups for age (χ² = 8.08, P = 0.23), race (χ² = 0.6, P = 0.96), or gender (χ² = 1.33, P = 0.25).

Table 1.

Demographics

	Nonintervention (n=1001), n (%)	Intervention (n=171), n (%)	P
Age (years)
18-24	176 (17)	36 (21)	>0.05
25-34	165 (17)	39 (23)
35-44	151 (15)	21 (12)
45-54	169 (17)	33 (19)
55-64	128 (13)	28 (16)
65-74	94 (9)	10 (6)
75+	118 (12)	4 (3)
Race
Caucasian	826 (83)	135 (79)	>0.05
Black	118 (11)	24 (14)
Asian	5 (1)	1 (1)
Hispanic	34 (3)	7 (4)
Other	18 (2)	4 (2)
Gender
Male	723 (72)	136 (79)	>0.05
Female	278 (28)	35 (21)

Injury characteristics are compared in Table 2. A significant difference was seen between groups for presenting GCS (χ² = 67.71, P < 0.001). Of note, in the nonintervention group, 64% had mild GCS score (14–15) compared to 10% of the intervention group. Conversely, 74% of the intervention group had severe GCS score (3–8) compared to 22% of the nonintervention group. No significant difference was seen between groups for mechanism of injury (χ² = 7.58, P = 0.27), type of TBI (χ² = 3.09, P = 0.8), or additional radiographic findings (χ² = 1.71, P = 0.63).

Table 2.

Injury characteristics

	Nonintervention (n=1001), n (%)	Intervention (n=171), n (%)	P
GCS
1. Mild (14-15)	636 (64)	17 (10)	<0.001
2. Moderate (9-13)	140 (14)	27 (16)
3. Severe (3-8)	226 (22)	127 (74)
Mechanism of injury
1. Assault	160 (16)	16 (9)	>0.05
2. ATV/dirtbike	44 (4)	11 (6)
3. GSW/knife	46 (5)	9 (5)
4. Bicycle/moped	74 (7)	18 (11)
5. MCC/MVC	381 (38)	77 (45)
6. Fall	212 (22)	19 (12)
7. Other	84 (8)	21 (12)
Types of TBI
1. Contusion	61 (6)	21 (12)	>0.05
2. DAI	31 (3)	23 (13)
3. EDH	49 (5)	22 (13)
4. SDH	201 (20)	63 (37)
5. tSAH	295 (29)	79 (46)
6. ICH/IVH	66 (7)	27 (16)
7. Penetrating injury	11 (1)	1 (1)
Additional radiographic findings
1. Edema	25 (2)	20 (12)	>0.05
2. Herniation	32 (3)	25 (15)
3. Pneumocephalus	34 (3)	10 (6)
4. Cerebral/cerebellar laceration	5 (1)	4 (2)

GCS - Glasgow Coma Scale; TBI - Traumatic brain injury; ATV - All-terrain vehicle; GSW - Gunshot wound; MCC - Motorcycle collision; MVC - Motor vehicle collision; DAI - Diffuse axonal injury; EDH - Epidural hematoma; SDH - Subdural hematoma; tSAH - Traumatic subarachnoid hemorrhage; ICH - Intracranial hemorrhage; IVH - Intraventricular hemorrhage

Fracture type patterns were similar between the nonintervention and intervention group (χ² = 4.518, P = 0.92) as seen in Figure 1. Subset analysis did however reveal a twofold increase in Lefort type 2 and panfacial fractures in the intervention group compared to nonintervention group. In the intervention group, 136/171 required an ICP monitor or external ventricular drain (EVD) only, 12/171 required a craniotomy, craniectomy, or burr holes only, 23/171 required a craniotomy, craniectomy, or burr holes with EVD or ICP monitor [Figure 2]. A significant difference was seen in type of intervention depending on presenting facial fracture pattern (χ² = 20.02, P = 0.03). Of note, 24% of the craniotomy, craniectomy, and burr hole group had Lefort type 2 fracture compared to only 9% in the ICP monitor only group. About 15% of the craniotomy, craniectomy, and burr hole group had Lefort type 3 fracture compared to only 7% in the ICP monitor only group. Only 29% of the craniotomy/burr hole group had panfacial fractures compared to 7% of the ICP monitor only group [Figure 3].

Figure 1.

Patients in intervention vs. nonintervention group by type of facial fracture. No overall significant difference in aggregate fracture pattern between groups. However, there were more Lefort type 2, Lefort type III, and panfacial fractures in the intervention group

Figure 2.

Type of neurosurgical intervention for trauma patients with facial fractures. ICP – Intracranial pressure; EVD – External ventricular drain; Crani – decompressive craniectomy, craniotomy, or burr holes

Figure 3.

Lefort type 2, and panfacial fractures, and 3 fractures were common in the craniotomy, craniectomy, burr hole group compared to intracranial pressure monitor only group. These results were statistically significant

Further subset analysis was done to compare Lefort type 2 and 3 fractures based on GCS scores for each group. A significant difference was seen (χ² = 8.44, P = 0.01). Of patients with GCS 14–15, 7% of the nonintervention group had Lefort type 2 and 3 fractures compared to 6% of the intervention group. A notable difference however was seen for patients presenting with GCS 9–13 with only 11% of patients in the nonintervention group having Lefort type 2 and 3 fractures compared to 40% in the intervention group. In addition, for patients with GCS, less or equal to 8 17% of the nonintervention group had Lefort type 2 and 3 fractures compared to 19% of the intervention group [Figure 4].

Figure 4.

A significant difference was seen between the nonintervention and intervention groups in regard to Lefort type 2 and 3 fractures when subanalysis was done based on Glasgow Coma Scale. 1 = GCS 14-15, 2 = GCS 9-13, 3 = GCS < 9. **P < 0.01, *P < 0.05

Discussion

Traumatic injury is a leading cause of death and disability worldwide.[23] A significant number of trauma victims present with maxillofacial fractures, with roughly half of these patients presenting with TBI.[1,2,24] There is a growing body of literature suggesting that maxillofacial fractures serve as predictors for the presence and severity of TBI.[4,7,9,10,13,17,21] This study aimed to determine whether certain types of maxillofacial fractures can predict the need for neurosurgical interventions in an effort to produce an algorithm for the management of TBI patients presenting with known facial fractures.

A retrospective analysis of patients admitted to a major academic hospital trauma center between 2010 and 2019 with known facial fractures and concurrent TBI was performed. Most patients were victims of motor vehicle collisions, motorcycle accidents, and falls. Most suffered orbital, nasal, and maxillary fractures in a distribution similar to the study done by Menon et al. with patients sharing similar demographics.[25] Patients were not more likely to require intervention based on age, race, gender, mechanism of injury, or specific radiologic findings of edema, herniation, pneumocephalus, or laceration. As anticipated, those taken for intervention had higher GCS scores. When a subset analysis was performed, however, even patients with GCS scores 9–13 were more likely to require subsequent intervention if they had Lefort type 2 and 3 fractures lending credence to the individual predictive indication of these fracture types.

Patients with Le Fort type II, type III, and panfacial fractures were more likely to undergo neurosurgical intervention. Further, those presenting with these severe fractures were more likely to receive higher levels of neurosurgical intervention involving craniotomies, craniectomy, and burr holes compared to only EVDs or ICP monitors. This is consistent with the initial hypothesis that the high-velocity impacts required to produce the more severe type II and III Le Fort type fractures[22,26] are also more likely to lead to more severe neurologic injury. These results also support previous investigations that showed associations between midface fractures and more severe TBI.[21]

Improved algorithms to identify and triage patients with facial fractures that are more likely to require neurosurgical interventions are being designed in collaboration with the hospital trauma, plastic, and oral and maxillofacial surgeon colleagues. Figure 5 shows an improved decision tree algorithm for managing patients with severe TBI and to have greater clinical suspicion for decline in moderate TBI patients with Lefort type 2 and 3 facial fractures. This algorithm is influenced by this study's findings and Czerwinski's et al.'s proposition for mandatory early computed tomography of the head to rule out TBI in patients presenting with facial fractures. This argument was also supported by Shibuya and et al.'s study which showed that 11% of patients who underwent facial fracture repair had worsened GCS score following intervention due to underlying TBI.[27,28] Early surgical interventions improve outcomes.[29] The hope is that through initiation of the improved algorithm, early imaging can be obtained, improved interactions between specialties can enhance patient care and ultimately allow providers to quickly intervene when indicated.

Figure 5.

Improved algorithm for managing patients with suspected TBI and facial fractures. Patients with Lefort type 2 or 3 fractures are at greater likelihood for requiring neurosurgical intervention and should be grouped accordingly

This study's retrospective nature serves as a limitation. The study was also limited by the trauma registry. For example, it was not possible to separately consider the use of EVDs versus other intracranial pressure monitoring devices because providers sometimes did not use specific International Classification of Diseases (ICD) coding. ICD coding also sometimes did not specify fracture types, only indicating the presence of a facial fracture. These injuries were successfully categorized by chart review. However, it is possible that some facial fractures for patients within this group were not recorded.

We hope that the data and results from this initial study will serve as a catalyst for prospective investigations. Subsequent studies may prospectively implement the algorithm in patients with facial fractures and TBI to directly evaluate its effectiveness. This will allow direct evaluation of effectiveness. Future studies can also implement the finite element head models reviewed and developed by Tse et al. to further determine fracture patterns more likely to be associated with severe TBI.[30,31]

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.