Keywords
traumatic brain injury - Le Fort type 2 and 3 fractures - neurosurgical intervention
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 traumatic
brain injury (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 midface 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 University of Florida and
abided by the highest international ethical research standards. Retrospective analysis
of patients from 2010 to 2019 obtained through our trauma registry. Inclusion criteria:
adults over 18 years, confirmed facial fracture with available neuroimaging, reported
TBI, and admission to intensive care unit or floor bed. Exclusion criteria: patients
less than 18 years old, patients with no neuroimaging, and patients who 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). Race was grouped into 5 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 7 categories: 1 (assault), 2 (all-terrain vehicle or dirt
bike 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: 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 into 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. A p-value of < 0.05 was considered statistically significant.
Results
Based on the above inclusion/exclusion criteria 1,985 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. Note that 1,172 therefore met the criteria for inclusion
into the study. A total of 1,001 patients had facial fracture and TBI with no neurosurgical
intervention, while 171 had facial fracture and TBI with neurosurgical intervention.
[Table 1] has baseline demographic data. No significant difference was seen between groups
for age (chi-square = 8.08, p = 0.23), race (chi-square = 0.6, p = 0.96), or gender (chi-square = 1.33, p = 0.25).
Table 1
Demographics
|
Age in years
1. 18–24
2. 25–34
3. 35–44
4. 45–54
5. 55–64
6. 65–74
7. 75+
|
Nonintervention (n = 1,001)
1. (176) 17%
2. (165) 17%
3. (151) 15%
4. (169) 17%
5. (128) 13%
6. (94) 9%
7. (118) 12%
|
Intervention
(n = 171)
1. (36) 21%
2. (39) 23%
3. (21) 12%
4. (33) 19%
5. (28) 16%
6. (10) 6%
7. (4) 3%
|
p > 0.05
|
|
Race
1. Caucasian
2. Black
3. Asian
4. Hispanic
5. Other
|
1. 826 (83%)
2. 118 (11%)
3. 5 (1%)
4. 34 (3%)
5. 18 (2%)
|
1. 135 (79%)
2. 24 (14%)
3. 1 (1%)
4. 7 (4%)
5. 4 (2%)
|
p > 0.05
|
|
Gender
1. Male
2. Female
|
1. 723 (72%)
2. 278 (28%)
|
1. 136 (79%)
2. 35 (21%)
|
p > 0.05
|
Injury characteristics were compared in [Table 2]. A significant difference was seen between groups for presenting GCS (chi-square = 67.71,
p < 0.001). Of note, in the nonintervention group 64% had mild GCS score (14–15) compared
with 10% of the intervention group. Conversely, 74% of the intervention group had
severe GCS score (3–8) compared with 22% of the nonintervention group. No significant
difference was seen between groups for mechanism of injury (chi-square = 7.58, p = 0.27), type of TBI (chi-square = 3.09, p = 0.8), or additional radiographic findings (chi-square = 1.71, p = 0.63).
Table 2
Injury characteristics
|
Glasgow Coma Scale
1. Mild (14–15)
2. Moderate (9–13)
3. Severe (3–8)
|
Nonintervention (n = 1,001)
1. (636) 64%
2. (140) 14%
3. (226) 22%
|
Intervention
(n = 171)
1. (17) 10%
2. (27) 16%
3. (127) 74%
|
p < 0.001
|
|
Mechanism of Injury
1. Assault
2. ATV/Dirt bike
3. GSW/Knife
4. Bicycle/Moped
5. MCC/MVC
6. Fall
7. Other
|
1. 160 (16%)
2. 44 (4%)
3. 46 (5%)
4. 74 (7%)
5. 381 (38%)
6. 212 (22%)
7. 84 (8%)
|
1. 16 (9%)
2. 11 (6%)
3. 9 (5%)
4. 18 (11%)
5. 77 (45%)
6. 19 (12%)
7. 21 (12%)
|
p > 0.05
|
|
Types of TBI
1. Contusion
2. DAI
3. EDH
4. SDH
5. tSAH
6. ICH/IVH
7. Penetrating injury
|
1. 61 (6%)
2. 31 (3%)
3. 49 (5%)
4. 201 (20%)
5. 295 (29%)
6. 66 (7%)
7. 11 (1%)
|
1. 21 (12%)
2. 23 (13%)
3. 22 (13%)
4. 63 (37%)
5. 79 (46%)
6. 27 (16%)
7. 1 (1%)
|
p > 0.05
|
|
Additional radiographic findings
1. Edema
2. Herniation
3. Pneumocephalus
4. Cerebral/Cerebellar laceration
|
1. 25 (2%)
2. 32 (3%)
3. 34 (3%)
4. 5 (1%)
|
1. 20 (12%)
2. 25 (15%)
3. 10 (6%)
4. 4 (2%)
|
p > 0.05
|
Abbreviations: ATV, all-terrain vehicle; DAI, diffuse axonal injury; EDH, epidural
hematoma; GSW, gunshot wound; ICH, intracranial hemorrhage; IVH, intraventricular
hemorrhage; MCC, motorcycle collision; MVC, motor vehicle collision; SDH, subdural
hematoma; TBI, traumatic brain injury; tSAH, traumatic subarachnoid hemorrhage.
Fracture type patterns were similar between the nonintervention and intervention group
(chi-square = 4.518, p = 0.92) as seen in [Fig. 1]. Subset analysis did, however, reveal a twofold increase in Le Fort type 2 and panfacial
fractures in the intervention group compared with the nonintervention group. In the
intervention group, 136/171 required an intracranial pressure (ICP) monitor or external
ventricular drain (EVD) only, 12/171 required a craniotomy, craniectomy, or burr holes
only, and 23/171 required a craniotomy, craniectomy, or burr holes with EVD or ICP
monitor ([Fig. 2]). A significant difference was seen in type of intervention depending on presenting
facial fracture pattern (chi-square = 20.02, p = 0.03). Of note, 24% of the craniotomy, craniectomy, and burr hole group had Le
Fort type 2 fracture compared with only 9% in the ICP monitor-only group. Fifteen
percent of the craniotomy, craniectomy, and burr hole group had Le Fort type 3 fracture
compared with only 7% in the ICP monitor-only group. Also, 29% of the craniotomy/burr
hole group had panfacial fractures compared with 7% of the ICP monitor-only group
([Fig. 3]).
Fig. 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 Le Fort type 2, Le Fort type 3, and panfacial fractures in the intervention
group.
Fig. 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.
Fig. 3 Le Fort type 2, and panfacial fractures, and 3 fractures were common in the craniotomy,
craniectomy, and burr hole group compared to the intracranial pressure (ICP) monitor-only
group. These results were statistically significant.
Further subset analysis was done to compare Le Fort type 2 and 3 fractures based on
GCS score for each group. A significant difference was seen (chi-square = 8.44, p = 0.01). Of patients with GCS 14 to 15, 7% of the nonintervention group had Le Fort
type 2 and 3 fractures compared with 6% of the intervention group. A notable difference,
however, was seen for patients presenting with GCS 9 to 13 with only 11% of patients
in the nonintervention group having Le Fort type 2 and 3 fractures compared with 40%
in the intervention group. Additionally, for patients with GCS less or equal to 8,
17% of the nonintervention group had Le Fort type 2 and 3 fractures compared with
19% of the intervention group ([Fig. 4]).
Fig. 4 A significant difference was seen between the nonintervention and intervention groups
in regards to Le Fort type 2 and 3 fractures when subanalysis was done based on Glasgow
Coma Scale (GCS). 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 to 13
were more likely to require subsequent intervention if they had Le Fort type 2 and
3 fractures lending credence to the individual predictive indication of these fracture
types.
Patients with Le Fort type 2, type 3, 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 with only EVDs or ICP monitors.
This is consistent with the initial hypothesis that the high-velocity impacts required
to produce the more severe type 2 and 3 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.
[Fig. 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 Le
Fort type 2 and 3 facial fractures. This algorithm is influenced by this study's findings
and Czerwinski et al's proposition for mandatory early computed tomography head to
rule out TBI in patients presenting with facial fractures. This argument was also
supported by Shibuya 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.
Fig. 5 Improved algorithm for managing patients with suspected traumatic brain injury (TBI)
and facial fractures. Patients with Le Fort 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 ICP 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 study will look at the prospective implementation
of the algorithm for patients with facial fractures and TBI. 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]
