Traumatic brain injury (TBI) is a major cause of morbidity and mortality,1–3 and post-traumatic cerebrospinal fluid (CSF) leakage is a complication of TBI. Approximately 84% of cases of CSF leakage are caused by head injuries involving skull fractures,4 and up to 30% of patients with skull base fractures develop CSF fistulae.5–7
CSF rhinorrhea and otorrhea can potentially lead to fatal complications such as persistent CSF fistulae, central nervous system infections, and meningitis. The occurrence rate of meningitis among patients with post-traumatic CSF leakage has been reported to range from 10% to 37%.4,8 These complications usually cause prolonged hospital stays and even higher mortality rates.
However, the occurrence rate, fracture sites, and mortality rates in patients of TBI with post-traumatic CSF leakage have seldom been reported. Therefore, we conducted this case-control cohort study to compare TBI patients with and without post-traumatic CSF leakage drawn from the Traumatic Brain Injury Registry database of Taiwan. We hypothesized that cases with post-traumatic CSF leakage after TBI would have increased intracranial damage and a higher mortality rate compared with those without CSF leakage.
2.1. Data collection
From January 1st, 1993, to December 31st, 2008, a survey on TBI was conducted by the Head & Spinal Cord Injury Research Group of the Taiwan Neurosurgical Society. The data of patients from 24 major hospitals (80% of the hospitals in Taiwan capable of treating TBI) were retrospectively collected in this study. The Head & Spinal Cord Injury Research Group of the Taiwan Neurosurgical Society then created a database, named the Traumatic Brain Injury Registry (TBIR). Data in the TBIR were recorded by experienced neurosurgeons from each hospital and extracted by research assistants. International Classification of Diseases, Ninth Revision (ICD-9) codes 800–804 (fracture of the skull or facial bones), 850.0 (concussion without loss of consciousness), 850.1–850.2 (concussion with loss of consciousness), 851–851.1 (cerebral/cerebellar/brain stem contusion or laceration), 852.0–853 (extracerebral and/or intracerebral hematoma, subarachnoid hemorrhage, subdural hemorrhage, epidural hemorrhage, intracranial hemorrhage), 854.0 (unspecified intracranial injury), 900.0 (injury to blood vessels of the head and neck), and 950.0–951.5 (injury to cranial nerves) were used to identify TBI cases. In total, 174,236 cases were included in the database. The principles outlined in the Declaration of Helsinki were followed strictly, and there was no patient consent sought because the retrospective data were analyzed anonymously and no personal information could be connected in this study.
2.2. Study population
We retrospectively analyzed the database of the TBIR. The inclusion criteria for the study cohort were cases who had sustained a closed TBI, and who were diagnosed with new-onset CSF leakage and tension pneumocephalus associated with the injury. Cases that sustained a penetration brain injury, died at the scene, suffered out-of-hospital cardiac arrest, or underwent cardio-pulmonary cerebral resuscitation were excluded. A total of 1773 cases matched these criteria and were included as the study cohort.
The comparison cohort (control group) was those patients who had sustained a closed TBI without CSF leakage or tension pneumocephalus during the same period. The control cohort cases were randomly selected at a ratio of 5:1 to the study cohort matched by age (<18 years, 18–30 years, 31–45 years, 46–65 years, and >65 years) and sex. In total, 8865 cases were included in the control cohort.
The demographic data, mechanisms of injury, Glasgow Coma Scale (GCS) score on admission and discharge, brain computerized tomography, cranial nerve injury, associated injury other than central nervous system, skull fractures, intracranial hemorrhage, and management were recorded. All cases were followed for 1 year, and the survival for each at the end of the year was assessed and recorded.
2.3. Statistical analysis
Pearson χ2 tests were used to explore the differences in the characteristics (age, sex, GCS, cranial nerve lesion, fracture of the skull, associated injuries, and intracranial hemorrhage) between the two cohorts. We calculated the 1-year survival rate using a Cox regression plot to examine the differences in the survival rate between the two cohorts. Cox proportional hazard regression analysis was performed to compute the 1-year survival rate, after adjusting for TBI characteristics and selecting comorbid medical disorders. Each dichotomous variable in the model was checked for proportionality while adjusting for the other covariates in the model by examining diagnostic log–log survival plots. Multinomial logistic regression models were performed to analyze the risk factors of the 1-year death or vegetative outcome in patients with traumatic brain injuries. A significance level of p < 0.05 was used to determine the significance of predictors in the models. All statistical analyses were performed using SPSS (Statistical Package for the Social Sciences) Version 17.0 (SPSS Inc., Chicago, IL, USA).
3.1. Study group
Of the 174,236 cases in the TBIR database, we identified 1773 patients with post-traumatic CSF leakages and tension pneumocephalus as the study cohort. The occurrence rate of post-traumatic CSF leakages was 1.0% (1773/174,236) in our series. In the study group, there were 1364 (76.9%) males and 409 (23.1%) females. We also divided the study cohort into five age subgroups, as follows: <18 years, 18–30 years, 31–45 years, 46–65 years, and >65 years. They were 202 (11.4%) cases, 609 (34.3%) cases, 387 (21.8%) cases, 359 (20.2%) cases, and 216 (12.1%) cases, respectively, in each of the five subgroups. More than half of the injury cases (56.1%) were aged 18–45 years (Table 1).
3.2. Severity of injury
In TBI, cases with CSF leakage commonly have more damage to their neurological system. In the study cohort, 50.0% had mild TBI (GCS ≥ 13), 21.8% moderate TBI (GCS 9–12), and 28.2% severe TBI (GCS ≤ 8). In the comparison group (n = 8865), 79.9%, 10.2%, and 9.9% had mild, moderate, and severe TBI, respectively. There were significantly higher rates of moderate and severe TBI in the CSF leakage group than in the control group (p < 0.001). In addition, cases in the CSF leakage group had more cranial nerve injuries (26.5% vs. 9.0%, p < 0.001) and intracranial hemorrhage (64.7% vs. 28.8%, p < 0.001) than did cases in the comparison group (Table 1).
3.3. Higher mortality rate in the CSF leakage group
There were more severe neurological injuries in the CSF leakage group, with higher morbidity and lower survival rates. Of the total 10,638 sampled cases, 406 (3.8%) died during the 1-year follow-up period. In the CSF leakage group, 159 (9.0%) cases died and 1614 (91.0%) cases were still alive, compared with 247 (2.8%) and 8618 (97.2%) cases respectively, in the comparison group. The crude hazard ratios for death between these two cohorts showed a significant difference. The hazard ratio of death in the CSF leakage group during the 1-year follow-up period was 3.35-fold [95% confidence interval (CI) = 2.75–4.09, p < 0.001] greater than for those in the control group. After adjusting for age, sex, GCS, cranial nerve lesion, associated injuries and intracranial hemorrhage, the adjusted hazard ratio of death in the CSF leakage group was still 1.44-fold (95% CI = 1.16–1.79, p < 0.001) higher compared with the control cohort. The CSF leakage group had a significantly higher 1-year mortality rate than the comparison group (p < 0.001; Table 2).
3.4. Mortality rates in the CSF otorrhea, rhinorrhea, and tension pneumocephalus subgroups
We divided the CSF leakage group into three subgroups: otorrhea (n = 568), rhinorrhea (n = 302), and tension pneumocephalus (n = 903). During the 1-year follow-up period, the mortality rates were 8.5% (48/568) in the otorrhea subgroup, 10.9% (33/302) in the rhinorrhea subgroup, and 8.6% (78/903) in the tension pneumocephalus subgroup (Table 3).
We compared the crude hazard ratios for death between these three subgroups to the control group. During the 1-year follow-up period, the hazard ratios of death were 3.17 (95% CI = 2.33–4.32, p < 0.001) in the otorrhea subgroup, 4.14 (95% CI = 2.88–5.95, p < 0.001) in the rhinorrhea subgroup, and 3.44 (95% CI = 2.72–4.36, p < 0.001) in the tension pneumocephalus subgroup, compared with the control group.
Similarly, after adjusting for age, sex, GCS, cranial nerve lesion, associated injuries, and intracranial hemorrhage, the adjusted hazard ratios of death during the 1-year follow-up period were 1.32 (95% CI = 1.02–1.95, p < 0.05) in the otorrhea subgroup, 2.29 (95% CI = 1.55–3.38, p < 0.001) in the rhinorrhea subgroup, and 1.35 (95% CI = 1.04–1.74, p < 0.05) in the pneumocephalus subgroup (Table 3). All three subgroups had significantly lower 1-year survival rates than the control group during the 1-year follow-up period, and the rhinorrhea subgroup had the lowest survival rate among three CSF leakage subgroups (Fig. 1).
3.5. Skull fracture sites
The locations of the skull fractures had a significant impact on the types of CSF leakage. In the CSF leakage group, 1305 (73.6%) cases had skull fractures. Among them, 347 (26.6%) cases were in the otorrhea subgroup, 185 (14.2%) in the rhinorrhea subgroup, and 773 cases in the pneumocephalus subgroup (59.2%). We then analyzed the relationship between the fracture sites and CSF leakage. In the otorrhea subgroup, the two most common fracture sites were the temporal bone and skull base with odds ratios (ORs) of 2.31 (95% CI = 1.70–3.12, p < 0.001) and 3.04 (95% CI = 2.15–4.29, p < 0.001), respectively. In the rhinorrhea subgroup, the two most common fractures sites were the frontal bone (OR 1.51, 95% CI = 1.01–2.25, p < 0.05) and skull bases (OR 3.58, 95% CI = 2.34–5.47, p < 0.001), although the frontal bone was the most common fracture site in the pneumocephalus subgroup (OR 1.72, 95% CI = 1.38–2.15, p < 0.001; Table 4).
3.6. Risk factors for 1-year death or vegetative outcome
After multivariate analysis, the significant risk factors for 1-year death or vegetative outcome compared with those for full recovery or moderate disability in patients with TBIs were found to include age, intracranial hemorrhage, cranial nerve lesion, GCS, and CSF leakage, including otorrhea, rhinorrhea, and pneumocephalus (Table 5). Risk factors for severe disability compared with those for full recovery or moderate disability were significant in age, intracranial hemorrhage, cranial nerve lesion, GCS, and pneumocephalus (Table 5).
4.1. The occurrence of post-traumatic CSF leakage
Head trauma is a frequent cause of CSF leakage, and 30–80% of cases of CSF leakage are caused by closed head injuries and skull base fractures.5–7,9 Bell et al10 reported that 4.6% of 735 traumatic cases developed post-traumatic CSF leakage over a 10-year study period, among whom post-traumatic otorrhea (75.8%) was more common than post-traumatic rhinorrhea (26.5%). Friedman et al4 also reported that approximately 2% of 51 cases treated for TBI had post-trauma CSF leakage. However, these previous studies were limited by the small number of cases or being localized to a single institute or particular region. Conversely, the TBIR database in Taiwan is one of the biggest TBI registry databases in the world (n = 174,236).1,3,11 Our cases were collected from multiple centers nationwide, including rural and urban areas, covering approximately 80% of the 23-million population of Taiwan.
Furthermore, this study did not only focus on a particular TBI group, such as those patients with severe (GCS ≤ 8) or moderate (GCS 9–12) TBI, but also covered the whole TBI population, including mild (GCS ≥ 13), moderate, and severe TBI. Therefore, we believe that the results of this study reflect the real occurrence of post-traumatic CSF leakage in patients with TBI. In addition, the number of cases with post-traumatic CSF leakage (n = 1773) was larger than in previous studies. The occurrence rate of post-traumatic CSF leakage was 1.0%, which is far less than in other series.10,12 However, we believe this rate more closely reflects the actual clinical situation because the cases in our study were distributed nationally, not locally, and the severity of injury was studied more wholly, not partially.
4.2. More intracranial hemorrhage and cranial nerve injuries in the CSF leakage group
Few reports have discussed intracranial damage and cranial injuries in patients with post-traumatic CSF leakage. Our results showed that the traumatic CSF leakage group had a higher rate of intracranial hemorrhage (64.7%) compared with the group without CSF leakage (28.8%; ) (p < 0.001). There were two potential explanations for this finding: (1) the presence of post-traumatic CSF leakage or tension pneumocephalus means a loss of integrity of the skull bone and skull base. An impact strong enough to break a solid skull bone involves more energy than that which does not break the skull bone. This also means more damage to the brain parenchyma and vulnerable cranial nerves; and (2) the presence of post-traumatic CSF leakage or tension pneumocephalus reflects the dural tearing and breakdown of the subarachnoid space. In this situation, both the loss of the protection of brain parenchyma from firm dura matter and the loss of the cushion effect of the subarachnoid CSF system would cause more damage to parenchyma and cranial nerves, thereby leading to the increased intracranial hemorrhage and cranial nerve injuries in our series. This is consistent with the recent report by Neville et al,13 who reported that 90% of their patients with frontal depressed skull fractures had concomitant focal lesions, 82.5% had subarachnoid hemorrhage, and 70% presented with pneumocephalus.
4.3. The association between fracture sites and pattern of CSF leakage
The association between fracture sites and pattern of CSF leakage in TBI has not been thoroughly studied to date. Choi et al14 reported 115 cases who presented with post-traumatic CSF leakage, and concluded that the patients with CSF rhinorrhea had a significantly greater incidence of periorbital hematoma. In addition, frontal/ethmoid bone fractures were particularly associated with dural tearing and CSF leakage. Friedman et al4 reported that 84% of their cases with post-traumatic CSF leakage presented with skull fractures, and the frontal sinus was most commonly involved. Anterior skull base leaks are more common than middle or posterior leaks, due to the firm adherence of the dura to the anterior basilar skull. The most commonly reported sites in CSF rhinorrhea after accidental trauma are the sphenoid sinus (30%), frontal sinus (30%), and ethmoid/cribriform (23%).15,16
In our study, 73.6% of post-traumatic CSF leakages presented with skull fractures, mildly lower than the Friedman's series,4 however, we had a similar trend of more vulnerable injuries in frontal bone. In our post-traumatic CSF rhinorrhea group, the frontal bone and skull base were the two most commonly involved fracture sites (36.8% and 36.2%, respectively). The presence of frontal bone and skull base fractures had a greater risk of dural tearing and CSF rhinorrhea (OR, 1.51 and 3.58, respectively) compared with those patients without such fractures. There were similar findings in the post-traumatic CSF otorrhea group, in which temporal bone was the most commonly involved fracture site (40.3%), and the OR was 2.31, significantly higher than for those cases without temporal bone fracture (p < 0.001).
Approximately 75–80% of cases of pneumocephalus are caused by trauma.17 In our series, the frontal bone was the most common fracture site causing tension pneumocephalus (272 cases, 35.2%), compared with the other skull bone/base fractures. There were also more frontal bone fractures (OR 1.72, p < 0.001) in the tension pneumocephalus group than other skull fractures. There are two possible mechanisms to explain these findings: (1) the relationship of anatomy between the frontal/ethmoid sinus and frontal bone causes codamage of the frontal/ethmoid sinus in cases of frontal bone fractures18; and (2) motorcycle accidents were the most common cause of TBI in our database.11,19 Even with helmet protection, the anterior frontal-facial parts of these patients are the most vulnerable.
4.4. The risks of death in the CSF leakage group
For all sampled cases of TBI (10,638), the 1-year mortality rate was 3.8%. The mortality rate of those with CSF leakage was significantly higher than for those without CSF leakage (9.0% vs. 2.8%, respectively), and the crude hazard ratio of death between these two groups was 3.35 (95% CI = 2.75–4.09, p < 0.001). After adjusting for age, sex, GCS, cranial nerve lesions, associated injuries, and intracranial hemorrhage, the hazard ratio was still significant between these two groups (1.49). This means that patients with post-traumatic CSF leakage had a higher association with death than those without post-traumatic CSF leakage.
Among the three subgroups of post-traumatic CSF leakage, CSF rhinorrhea had a greater association with death than the other two subgroups after 1-year of follow-up. There are two possible explanations for this finding: (1) craniofacial trauma was a common injury in our TBIR database, in which motorcycle accidents are predominant. This trauma type causes complicated frontal base fractures and para-sinus injuries, and it is more difficult to repair these complicated fractures involving multiple sites. This may also be associated with a higher risk of CNS infections, such as meningitis and brain abscess, and also the recurrence of CSF rhinorrhea even after surgical repair16; and (2) the occurrence of CSF rhinorrhea may reflect the release of intracranial pressure via CSF leakage, regardless of post-traumatic CSF leakage or spontaneous CSF leakages,20,21 and intracranial hypertension leads to higher mortality and morbidity in TBI cases. This may also explain the outcomes in the post-traumatic CSF leakage group being poorer than in those cases without CSF leakage in the current study.
To the best of our knowledge, this study is the first population-based cohort study to assess the risk for death in patients with CSF leakage. In addition, the large sample size of our dataset allowed us to determine the risk of death for patients with different categories of CSF leakage, while controlling for major potential confounding factors.
In conclusion, TBI patients with post-traumatic CSF leakage had significantly higher mortality rates than those cases without CSF leakages, and the cases with CSF rhinorrhea had worse outcomes compared with pneumocephalus or otorrhea among CSF leakages. Therefore, neurosurgeons should pay more careful attention to TBI cases with CSF leakage, especially those with the presence of CSF rhinorrhea.
We acknowledge the support of Grant 103-wf-eva-30 from Taipei Municipal Wan Fang Hospital, Taipei Medical University, Taipei, Taiwan.
1. Chiu WT, Hung CC, Shih CJ. Epidemiology of head-injury in rural Taiwan—a 4 year survey. J Clin Neurosci
2. Conroy C, Kraus JF. Survival after brain injury. Cause of death, length of survival, and prognostic variables in a cohort of brain-injured people. Neuroepidemiology
3. Tsai WC, Chiu WT, Chiou HY, Choy CS, Hung CC, Tsai SH. Traumatic brain injury
(TBI) is a major cause of morbidity and mortality in Taiwan. J Clin Neurosci
4. Friedman JA, Ebersold MJ, Quast LM. Post-traumatic cerebrospinal fluid leakage. World J Surg
5. Jennett B, Teasdale G, Fry J, Braakman R, Minderhoud J, Heiden J, et al. Treatment for severe head injury. J Neurol Neurosurg Psychiatry
6. Loew F, Pertuiset B, Chaumier EE, Jaksche H. Traumatic, spontaneous and postoperative CSF rhinorrhea. Adv Tech Stand Neurosurg
7. Marentette LJ, Valentino J. Traumatic anterior fossa cerebrospinal fluid fistulae and craniofacial considerations. Otolaryngol Clin North Am
8. Daudia A, Biswas D, Jones NS. Risk of meningitis with cerebrospinal fluid rhinorrhea. Ann Otol Rhinol Laryngol
9. Beckhardt RN, Setzen M, Carras R. Primary spontaneous cerebrospinal fluid rhinorrhea. Otolaryngol Head Neck Surg
10. Bell RB, Dierks EJ, Hormer L, Potter BE. Management of cerebrospinal fluid leak associated with craniomaxillofacial trauma. J Oral Maxillofac Surg
11. Chiu WT, Chu SF, Chang CK, Lui TN, Chiang YH. Implementation of a motorcycle helmet law in Taiwan and traffic deaths over 18 years. JAMA
12. Hegazy HM, Carrau RL, Snyderman CH, Kassam A, Zweig J. Transnasal endoscopic repair of cerebrospinal fluid rhinorrhea: a meta-analysis. Laryngoscope
13. Neville IS, Amorim RL, Paiva WS, Sanders FH, Teixeira MJ, de Andrade AF. Early surgery does not seem to be a pivotal criterion to improve prognosis in patients with frontal depressed skull fractures. Biomed Res Int. 2014, p. 6, Article ID 879286. http://dx.doi.org/10.1155/2014/879286
14. Choi D, Spann R. Traumatic cerebrospinal fluid leakage: risk factors and the use of prophylactic antibiotics. Br J Neurosurg
15. Banks CA, Palmer JN, Chiu AG, O’Malley BWJ, Woodworth BA, Kennedy DW. Endoscopic closure of CSF rhinorrhea: 193 cases over 21 years. Otolaryngol Head Neck Surg
16. Prosser JD, Vender JR, Solares CA. Traumatic cerebrospinal fluid leaks. Otolaryngol Clin North Am
17. Rouit RL, Murali R. Injuries of the cranial nerves. Cooper PR, editor. Head injury. 2nd ed. Baltimore: Williams and Wilkins; 1987: pp. 141-158.
18. Metzinger SE, Metzinger RC. Complications of frontal sinus fractures. Craniomaxillofac Trauma Reconstr
19. Lin JW, Tsai SH, Tsai WC, Chiu WT, Chu SF, Lin CM, et al. Survey of traumatic intracranial hemorrhage in Taiwan. Surg Neurol. 2006;66(Suppl.):S20-S25.
20. Badia L, Loughran S, Lund V. Primary spontaneous cerebrospinal fluid rhinorrhea and obesity. Am J Rhinol
21. Schlosser RJ, Wilensky EM, Grady MS, Bolger WE. Elevated intracranial pressure in spontaneous cerebrospinal fluid leaks. Am J Rhinol