Neurosurg Focus 32 (6):E4, 2012

Is posttraumatic fistula a predictor of posttraumatic ? A US Nationwide Inpatient Sample database study

*Ashish Sonig, M.D., M.S., M.Ch., Jai Deep Thakur, M.D., Prashant Chittiboina, M.D., M.P.H., Imad Saeed Khan, M.D., and Anil Nanda, M.D., M.P.H. Department of , Louisiana State University Health Sciences Center in Shreveport, Louisiana

Object. Various factors have been reported in literature to be associated with the development of posttraumatic meningitis. There is a paucity of data regarding fractures and facial fractures leading to CSF leaks and their as- sociation with the development of meningitis. The primary objective of this study was to analyze the US Nationwide Inpatient Sample (NIS) database to elucidate the factors associated with the development of posttraumatic meningitis. A secondary goal was to analyze the overall hospitalization cost related to posttraumatic meningitis and factors as- sociated with that cost. Methods. The NIS database was analyzed to identify patients admitted to hospitals with a diagnosis of from 2005 through 2009. This data set was analyzed to assess the relationship of various clinical parameters that may affect the development of posttraumatic meningitis using binary logistic regression models. Additionally, the overall hospitalization cost for the head injury patients who did not undergo any neurosurgical intervention was further categorized into quartile groups, and a regression model was created to analyze various factors responsible for escalating the overall cost of the hospital stay. Results. A total of 382,267 inpatient admissions for head injury were analyzed for the 2005–2009 period. Men- ingitis was reported in 0.2% of these cases (708 cases). Closed skull base fractures, open skull base fractures, cranial vault fractures, and maxillofacial fractures were reported in 20,524 (5.4%), 1089 (0.3%), 5064 (1.3%), and 88,649 (23.2%) patients, respectively. Among these patients with fractures, meningitis was noted in 0.17%, 0.18%, 0.05%, and 0.10% admissions, respectively. Cerebrospinal fluid was reported in 453 head injury patients (0.1%) and CSF otorrhea in 582 (0.2%). Of the patients reported to have CSF rhinorrhea, 35 (7.7%) developed meningitis, whereas in the cohort with CSF otorrhea, 15 patients (2.6%) developed meningitis. Cerebrospinal fluid rhinorrhea (p < 0.001, OR 22.8, 95% CI 15.6–33.3), CSF otorrhea (p < 0.001, OR 9.2, 95% CI 5.2–16.09), and major neurosurgical procedures (p < 0.001, OR 5.6, 95% CI 4.8–6.5) were independent predictors of meningitis. Further, CSF rhinorrhea (p < 0.001, OR 2.0, 95% CI 1.6–2.7), CSF otorrhea (p < 0.001, OR 2.3, 95% CI 1.9–2.7), and posttraumatic menin- gitis (p < 0.001, OR 3.1, 95% CI 2.5–3.8) were independent factors responsible for escalating the cost of head injury in cases not requiring any major neurosurgical intervention. Conclusions. Cerebrospinal fluid rhinorrhea and CSF otorrhea are independent predictors of posttraumatic men- ingitis. Furthermore, meningitis and CSF fistulas may independently lead to significantly increased cost of hospital- ization in head injury patients not undergoing any major neurosurgical intervention. (http://thejns.org/doi/abs/10.3171/2012.5.FOCUS1269)

Key Words • cerebrospinal fluid fistula • meningitis • head injury

ead injury accounts for significant mortality and ated with head injury. Additionally, the development of morbidity in the US, with approximately 1.5 mil- meningitis has shown to contribute substantially to in- lion persons per year being reported to have a head creased mortality and morbidity in such patients.1,2,23,32 injury.H26 Apart from the traumatic brain insult and associ- Various factors that are associated with the develop- ated hemodynamic alterations, various factors, including ment of posttraumatic meningitis include skull fractures, posttraumatic meningitis and CSF fistula, are recognized facial fractures (mainly involving the ), as prognostic parameters in the clinical spectrum associ- otological injuries, development of CSF fistulas, and pneu- mocephalus.6,8,13,15,16,20,25,31 Typically, fractures of the skull or facial may lead to the formation of CSF fistu- Abbreviations used in this paper: CCS = Clinical Classification 25 Software; EVD = external ventricular drain; HCUP = Healthcare las, which in turn increase the likelihood of meningitis. Cost and Utilization Project; ICD-9-CM = International Classifi- There are limited data to support an independent associa- cation of Diseases, Ninth Revision, Clinical Modification; NIS = tion of meningitis with skull fractures, facial fractures, and Nationwide Inpatient Sample. subsequent CSF leaks.8,13,16,20 Further, it is unclear whether * Drs. Sonig and Thakur contributed equally to this work. particular subtypes of skull fractures (closed skull base

Neurosurg Focus / Volume 32 / June 2012 1

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC A. Sonig et al. fractures, open skull base, cranial vault fractures) are as- Patients Without Fracture. All the patients with head sociated with increased risk of meningitis.5,24 injury who did not have a fracture were included in this We analyzed the NIS database to study the factors category. associated with the development of meningitis during the hospital stay in patients admitted for head injury. Addi- Patients With Meningitis. This variable was con- tionally, cases in which patients were admitted for head structed on the basis of CCS code 76 (meningitis [except injuries and did not undergo any major neurosurgical pro- that caused by tuberculosis or sexually transmitted dis- cedure were studied to assess the escalation of total hos- ease]). pital cost due to posttraumatic meningitis, CSF otorrhea, Patients With CSF Otorrhea. This variable was con- or CSF rhinorrhea. structed on the basis of ICD-9-CM code 388.61 (cerebro- spinal fluid otorrhea). Methods Patients With CSF Rhinorrhea. ICD-9-CM code The NIS is part of HCUP, which is sponsored by the 349.81 (cerebrospinal fluid rhinorrhea) was used to form Agency for Healthcare Research and Quality, formerly this category variable. the Agency for Health Care Policy and Research. The Neurosurgical Procedure Group. The following NIS is the largest all-payer inpatient care database that ICD-9 codes were used to identify patients who under- is publicly available in the US. It contains data from 5–8 went major neurosurgical procedures: 0101 (cisternal million hospital stays in about 1000 hospitals, which ap- proximate a 20% stratified sample of US community hos- puncture), 0123 (reexploration of craniotomy site), 0124 pitals. The NIS is drawn from those states participating (other craniotomy), 0125 (other craniectomy), 0202 (ele- in HCUP. vate fragments), 0203 (skull flap formation) Nationwide Inpatient Sample data from the 2005– 022 (ventriculostomy), and 0221 (insertion/replacement 2009 period were analyzed. Both CCS and ICD-9-CM of EVD). codes were used. The CCS coding system was developed Comorbidity Accumulation Index at the Agency for Healthcare Research and Quality as a tool for clustering patient diagnoses and procedures into Data from 382,267 cases were assessed on an indi- a manageable number of clinically meaningful catego- vidual basis, and a comorbidity accumulation index was ries. It collapses diagnosis and procedure codes from the calculated. The following comorbidities were studied: IDC-9-CM. We used single-level CCS codes 228 (skull acquired immune deficiency syndrome (AIDS), alcohol and face fractures) and 233 (intracranial injury) to extract abuse, deficiency anemias, rheumatoid arthritis/collagen data from 25,669 hospitals. vascular diseases, chronic blood loss anemia, congestive heart failure, chronic pulmonary disease, coagulopathy, Categorical Variables uncomplicated diabetes, diabetes with chronic compli- The following categorical variables were used in this cations, drug abuse, hypertension, hypothyroidism, liver study. disease, lymphoma, fluid and electrolyte disorders, meta- static cancer, other neurological disorders, obesity, paral- Patients With Only Skull Vault Fracture. This vari- ysis, peripheral vascular disorders, pulmonary circulation able was constructed on the basis of ICD-9-CM code disorders, renal failure, solid tumor without metastasis, 800 (fracture of vault of skull) and comprised the follow- peptic ulcer disease excluding , valvular disease, ing ICD-9 subcategories: 80000, 80001, 80002, 80003, and weight loss. 80004, 80005, 80006, 80009, 80050, 80051, 80052, The comorbidity accumulation indices of individual 80053, 80054, 80055, 80056, and 80059. These catego- patients ranged from 0 to a maximum of 12 (Table 1). ries included open and closed fractures of the skull vault. Quartiles were calculated, and patients with comorbidity Patients With Closed Skull Base Fractures. This indices above the 75th percentile (comorbidity index of variable was constructed on the basis of ICD-9-CM code 2 or more) were assigned to the high comorbidity index 801 (fracture of ) and comprised the follow- group. ing ICD-9 subcategories: 80100, 80101, 80102, 80103, Hospital Cost 80104, 80105, 80106, and 80109. Hospital costs were determined by the coded vari- Patients With Open Skull Base Fractures. This vari- able “total charges” in the NIS data. Total hospital cost able was also constructed on the basis of ICD-9-CM code was analyzed in the cohort of patients who did not un- 801 (fracture of base of skull) and comprised the follow- dergo a major neurosurgical procedure, and quartiles ing ICD-9 subcategories: 80150, 80151, 80152, 80153, were created to stratify the admissions by cost. Patients 80154, 80155, 80156, and 80159. whose total hospital expenditure was less than $11,624.00 Patients With Maxillofacial Injury. This variable formed the lower quartile group (expenses below the was constructed on the basis of ICD-9-CM codes 8020, 25th percentile), those whose hospital expenditure was 8021, 80220, 80221, 80222, 80223, 80224, 80225, 80226, between $11,624.00 and $21,730.00 were in the lower 80227, 80228, 80229, 80230, 80231, 80232, 80233, 80234, middle quartile (expenses between the 25th and 50th per- 80235, 80236, 80237, 80238, 80239, 8024, 8025, 8026, centiles), those whose hospital expenditure was between 8027, 8028, and 8029. $21,730.00 and $42,111.00 were in the upper middle

2 Neurosurg Focus / Volume 32 / June 2012

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC Role of posttraumatic CSF leak in development of meningitis

TABLE 1: Descriptive frequency and percentages of comorbidity kal-Wallis test to determine the statistical significance (p accumulation index values in 382,267 admissions for head injury < 0.05). Statistical analysis was done using IBM SPSS Statistics 20 and JMP 9 (SAS Institute, Inc.). Comorbidity Accumulation Index Frequency (%) 0 128,419 (33.6) Results 1 97,486 (25.5) Study Population 2 72,091 (18.9) 3 43,957 (11.5) A total of 382,267 inpatient admissions for head in- 4 23,167 (6.1) jury were recorded from 2005 through 2009. Of these patients, 280,140 had head injuries and no skull or facial 5 10,509 (2.7) fractures, whereas 102,127 admissions were for head in- 6 4,160 (1.1) juries with skull and facial fractures. Meningitis was re- 7 1,659 (0.4) ported in a total of 0.2% of the patients admitted with 8 586 (0.2) head injury (708 patients). Among the patients admitted 9 186 (0) for head injury with fractures, closed skull base fractures, 10 32 (0) open skull base fractures, cranial vault fractures, and maxillofacial fractures were reported in 20,524 (5.4%), 11 12 (0) 1089 (0.3%), 5064 (1.3%), and 88,649 (23.2%) cases, re- 12 3 (0) spectively. In these patients, meningitis was reported in 0.17%, 0.18%, 0.05%, and 0.10%, respectively. Cerebrospinal fluid rhinorrhea was reported in 453 quartile (expenses between the 50th and 75th percentiles), head injury admissions (0.1%) and CSF otorrhea in 582 and those whose expenditure was more than $42,111.00 head injury admissions (0.2%). Of the patients reported formed the upper quartile group (expenses higher than to have CSF rhinorrhea, 35 (7.7%) developed meningitis, the 75th percentile). The cohort whose incurred hospital whereas in the cohort with CSF otorrhea, 15 (2.6%) devel- cost was above the 75th percentile was compared with oped meningitis. In the cohort of patients who underwent each of the other cohorts. a major neurosurgical procedure, meningitis was seen in 0.76% cases, whereas in the conservatively managed Statistical Methods group it was seen in 0.11%; this difference was statisti- Meningitis in the head injury patients was the depen- cally significant (p < 0.0001). The incidence of posttrau- dent variable studied. Preliminary analysis for associa- matic meningitis in these 2 groups stratified by presence tion was done using the Pearson chi-square test. The re- of absence of CSF leak is show in Table 2). lationships between dependent and independent variables were studied using a multivariate binary logistic-regres- Factors Predicting Posttraumatic Meningitis sion model through which p values, odds ratios, and con- Based on the review of literature and considering fidence intervals were generated. To study the intergroup the availability of clinical factors in the NIS database, differences in the mean of total hospitalization charges, variables were selected to analyze the association with we used a box plot (Fig. 1) to determine the distribution the development of posttraumatic meningitis. They in- of the data. Since the data were significant for outliers, we cluded: 1) closed skull base fractures; 2) open skull base used the nonparametric Wilcoxon rank sum test or Krus- fractures; 3) cranial vault fractures; 4) CSF rhinorrhea; 5) CSF otorrhea; 6) major neurosurgical procedures; 7) an comorbidity index of more than 2; 8) skin or infections; 9) maxillofacial fractures; and 10) head in- jury without skull or facial fracture. A test of association was done using the Pearson chi-square test followed by

TABLE 2: Occurrence of posttraumatic meningitis and CSF leak in patients who were treated conservatively or underwent a major neurosurgical procedure

Meningitis Status Absent Present no neurosurgical procedure no CSF leak 341,988 379 (0.11%) Fig. 1. Box plots representing the mean hospitalization charges in CSF leak 694 31 (4.27%) patients with and without meningitis who were admitted with head injury and did not undergo any major neurosurgical procedure. The arrows neurosurgical procedure point toward the group means. The box plot shows significant outliers in no CSF leak 38,600 280 (0.72%) the study population for which a robust nonparametric Wilcoxon rank- CSF leak 277 18 (6.1%) sum test was used to determine statistical significance.

Neurosurg Focus / Volume 32 / June 2012 3

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC A. Sonig et al. logistic-regression model analysis. The latter showed that TABLE 4: Table highlighting the admissions in which CSF CSF rhinorrhea (p < 0.001, OR 22.8, 95% CI 15.6–33.3), rhinorrhea was present but no skull or facial fracture was CSF otorrhea (p < 0.001, OR 9.2, 95% CI 5.2–16.09), and diagnosed* major neurosurgical procedures (p < 0.001, OR 5.6, 95% CI 4.8–6.5) were independent predictors of the develop- Variable & Cohort No Fracture Reported Fracture Total ment of meningitis (Table 3). The regression model was all pts w/ head injury 280,140 102,127 382,267 adjusted with the above-described covariates so that it CSF rhinorrhea could predict independent associations. As expected, un- dergoing a major neurosurgical procedure independently absent 279,912 101,902 381,814 increased the odds of posttraumatic meningitis (by 5.6 present 228 (50.3%) 225 (49.7%) 453 times) as compared with no major neurosurgical interven- CSF otorrhea tion. The presence of CSF rhinorrhea or otorrhea inde- absent 279,767 101,918 381,685 pendently increased the odds of posttraumatic meningitis present 373 (64.1%) 209 (35.9%) 582 by 22.86 and 9.2 times, respectively (after adjusting for neurosurgical procedures and other covariates). * pts = patients. Cases With CSF Leaks and Undiagnosed Fractures was $35,109,541 (399 admissions, 11 missing values) We analyzed the database to identify the cases in and $13,912,533,097 (338,316 admissions, 4366 missing which head injury patients developed CSF leaks without a values), respectively. Using the nonparametric Wilcoxon diagnosis of skull or facial fracture. In 228 (50.3%) of the rank-sum test to adjust for outliers, the comparison of the admissions in which patients had CSF rhinorrhea and 373 means showed statistically higher cost of hospitalization (64.1%) of those in which patients had CSF otorrhea, no charges in the patients who had head injury with meningi- concomitant skull or facial fracture was reported (Table 4). tis than in those without meningitis ($42,808 vs $32,319, p < 0.0001, Z = 11.2) Association of CSF Leak and Type of Fracture In addition, the NIS head injury database was was We further analyzed the database to determine an analyzed using a quartile system (see Methods). Head association between the various types of skull fractures injury admissions with an overall cost of hospital stay or maxillofacial fractures and the development of CSF greater than the 75th percentile ($42,111.00) were catego- rhinorrhea or CSF otorrhea. We found that the presence rized as having increased hospitalization cost. Since the of closed skull base fractures was an independent predic- presence of comorbidities, wound infections, respiratory tor of the development of a CSF leak (Table 5) (p < 0.001, failure, coronary artery disease, and different types of OR 2.4, 95% CI 1.9–3.1). Cranial vault fractures and open fractures may influence the cost of hospitalization, these skull base fractures showed no significant association variables were studied together in a logistic-regression with CSF leak. model. Cerebrospinal fluid rhinorrhea (p < 0.001, OR 2.0, 95% CI 1.6–2.7), CSF otorrhea (p < 0.001, OR 2.3, 95% Impact of Meningitis and CSF Fistula on Cost of CI 1.9–2.7), and posttraumatic meningitis (p < 0.001, OR Hospitalization 3.1, 95% CI 2.5–3.8) were independent factors associated The effect of posttraumatic meningitis and CSF fis- with escalating costs of head injury in patients not requir- tula on the overall hospitalization cost was studied. Pa- ing any surgical intervention (Table 6). tients undergoing any major neurosurgical intervention Discussion (see Methods) were excluded from the cost analysis as the procedure itself could have escalated the cost. The In the literature, the incidence of posttraumatic men- total cost of hospitalization in the cohorts of head in- ingitis ranges from 0.38% to 10%.23,29 Posttraumatic men- jury patients with and without posttraumatic meningitis ingitis was noted in 0.2% of the admissions in the NIS da-

TABLE 3: Results of the logistic regression model with dependent variable as posttraumatic meningitis*

Covariates Used in Analysis B p Value OR 95% CI skull base closed fracture 0.152 0.593 1.164 0.667–2.032 skull base −0.292 0.693 0.746 0.175–3.191 vault fracture −1.183 0.061 0.306 0.089–1.054 rhinorrhea 3.130 <0.001 22.864 15.688–33.323 otorrhea 2.220 <0.001 9.203 5.263–16.093 neurosurgical procedure 1.732 <0.001 5.651 4.850–6.586 maxillofacial injury −0.577 0.082 0.562 0.293–1.075 head injury w/o fracture −0.084 0.810 0.919 0.461–1.832 constant 1.532 0.190 4.628

* Boldface type indicates a statistically significant association with the development of posttraumatic meningitis.

4 Neurosurg Focus / Volume 32 / June 2012

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC Role of posttraumatic CSF leak in development of meningitis

TABLE 5: Logistic-regression model showing the association between CSF leak and type of fracture

Covariates Used in Analysis B p Value OR 95% CI skull base open fracture 0.528 0.155 1.695 0.819–3.510 vault fracture −0.302 0.184 0.739 0.474–1.154 maxillofacial injury −0.890 <0.001 0.411 0.313–0.539 skull base closed fracture 0.901 <0.001 2.462 1.906–3.179 head injury w/o fracture −1.072 <0.001 0.342 0.251–0.467 constant 5.919 <0.001 371.885 tabase. The median time of onset of meningitis after the Factors Predicting Posttraumatic Meningitis injury generally ranges from 5 to 13 days, although it can 2,11,12,14,23 In our study we found that no specific subtype of manifest even several years after a head injury. skull or maxillofacial fractures predicted the develop- Our study focused on the development of posttraumatic ment of meningitis. On the other , CSF rhinorrhea, meningitis during the same hospital stay (single admis- CSF otorrhea, and neurosurgical intervention did inde- sion) as the one occasioned by a head injury, which is a pendently predict development of posttraumatic menin- relatively rare event.18,21,27 Posttraumatic meningitis is as- gitis.15,16 sociated with considerable morbidity and mortality.1,2,23,32 Typically, a fracture involving the skull or neigh- What is the Role of Prophylactic Antibiotic Therapy? boring facial structures may produce a breach in the meningeal layer and can result in CSF rhinorrhea, CSF Although most studies agree that CSF leak could otorrhea, or both.24 Cerebrospinal fluid leaks following be regarded as an important factor in the development skull base fractures develop in approximately 10%–30% of posttraumatic meningitis, results of studies of prophy- of cases.8,24,32 Sometimes, due to the limitations of the lactic antibiotic therapy have been equivocal.3–5,7,9,19,24,25,30 standard imaging modalities, the skull base fractures are Meta-analyses on this topic were published in 19973 and missed. High-resolution CT, cisternography, and MRI 1998.30 Brodie3 showed that the cohort of patients with CISS 3D protocols are then used to locate the site of CSF posttraumatic CSF leakage treated with prophylactic an- leak.17,28 In the current NIS database study, for 50.3% of tibiotic therapy had a significantly reduced incidence of admissions in which the patients were reported to have meningitis. In contrast, Villalobos et al.30 found that the CSF rhinorrhea and 64.1% of those in which they were antibiotic prophylaxis in patients with basilar skull frac- reported to have CSF otorrhea, no skull or facial fracture tures (independent of whether they had a CSF leak) did was reported. Posttraumatic CSF rhinorrhea and otorrhea not significantly decrease the incidence of posttraumatic incidence increases after trauma to the frontal, ethmoid, meningitis. A major limitation of these meta-analyses and sphenoid bones.6,22,24 In our study, closed skull base was that the findings were primarily based on retrospec- fractures independently predicted the development of tive and observational studies. Recently, Ratilal et al.25 CSF rhinorrhea (Table 5). conducted another meta-analysis in which they reviewed

TABLE 6: Results of binary logistic regression analysis for predicting factors that increase cost in head injury patients not undergoing any major neurosurgical intervention*

Covariates Used in Analysis B p Value OR 95% CI skull base closed fracture 0.101 <0.001 1.107 1.059–1.157 skull base open fracture 0.788 <0.001 2.199 1.889–2.559 vault fracture −0.780 <0.001 0.459 0.413–0.509 rhinorrhea 0.741 <0.001 2.099 1.618–2.722 otorrhea 0.838 <0.001 2.313 1.931–2.771 meningitis 1.142 <0.001 3.134 2.575–3.813 comorbidity index 0.401 <0.001 1.494 1.467–1.521 skin infection 0.460 <0.001 1.584 1.485–1.688 maxillofacial injury 0.458 <0.001 1.581 1.485–1.683 head injury w/o fracture 0.332 <0.001 1.394 1.306–1.488 CAD 0.106 <0.001 1.112 1.074–1.151 respiratory failure −1.578 <0.001 0.206 0.200–0.213 constant −2.980 <0.001 0.051

* Boldface type indicates significant association of the pertinent covariates with increased hospital cost. Abbreviation: CAD = coronary artery disease.

Neurosurg Focus / Volume 32 / June 2012 5

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC A. Sonig et al.

5 randomized controlled studies (involving 208 patients). brain.” As there is no separate code for “lumbar drain,” The authors concluded that available evidence does not it could not be included in the neurosurgical procedure support the prophylactic use of antibiotics in patients group; the code for cisternal puncture was used. with basilar skull fractures (with or without CSF leak). Delayed or recurrent meningitis is a known entity in Although, they found a trend of decreasing incidence of head injury patients. Since the NIS database codes a sin- meningitis in the treatment group on subgroup analysis gle inpatient admission as 1 identity, it is not possible to of the cases with CSF leakage, it was not statistically sig- follow up on a patient using this database. Therefore, our nificant. The authors mentioned that a study population study does not comment on delayed meningitis due to an of 798 patients was needed to achieve optimal statistical occult CSF fistula after the patient was discharged from power. the hospital. The usage of prophylactic antibiotic therapy Some reports suggest that the use of antibiotic thera- or the lack of it could not be studied as there is no ICD-9 py in patients with CSF leaks associated with skull base code for it. fractures may increase the chances of meningitis rather than decreasing it.4,6,7,10,30 The putative mechanism is via Conclusions the eradication of commensal organisms and subsequent colonization of pathogenic flora.6,32 This NIS database study found that CSF rhinorrhea The results of the current NIS study add to the on- and CSF otorrhea are independent predictors of posttrau- going debate regarding prophylactic antibiotic therapy in matic meningitis. Additionally, posttraumatic meningitis head injury patients with CSF leaks. We found that CSF and CSF fistula may lead to a significantly increased cost leaks independently predict the development of meningi- of hospitalization. We recommend focused prospective tis in head injury patients. We hope that this study will lay trials to study the efficacy of prophylactic antibiotics in a foundation for a focused prospective trial to study the reducing the incidence of posttraumatic meningitis, espe- efficacy of prophylactic antibiotic therapy in reducing the cially in patients presenting with posttraumatic CSF leak. incidence of meningitis in patients presenting with post- traumatic CSF leaks. Disclosure The authors report no conflict of interest concerning the mate- Impact of Posttraumatic Meningitis and CSF Fistula on the rials or methods used in this study or the findings specified in this Cost of Hospitalization paper. Author contributions to the study and manuscript preparation There is a consensus in the literature that meningitis include the following. Conception and design: Sonig. Acquisition considerably increases the odds of mortality, morbidity, of data: Sonig. Analysis and interpretation of data: Sonig, Thakur. and length of stay in the head injury patients. We evalu- Drafting the article: Sonig, Thakur. Critically revising the article: ated whether CSF fistulas and posttraumatic meningitis Sonig, Thakur. Reviewed submitted version of manuscript: Chitti- lead to increased hospitalization cost in the NIS database. boina, Khan. Approved the final version of the manuscript on behalf As described in Methods, we categorized the NIS data- of all authors: Nanda. Statistical analysis: Sonig, Thakur. Adminis- base “total charges” variable into quartiles. Admissions trative/technical/material support: Nanda. Study supervision: Nanda. with hospital cost above the 75th percentile were com- pared with the rest. To remove the bias of neurosurgical Acknowledgment intervention, we included in the analysis only those head The authors wish to thank Ms. Nidhi Setya, Texas A&M Uni- injury admissions in which patients did not undergo any versity, College Station, Texas, for her assistance with the statistical neurosurgical intervention. The mean cost of total hospi- analysis using JMP software. talization charges was significantly higher in the cohort of head injury patients with concomitant meningitis than References in those in whom meningitis was not reported. Further- more, we found that the presence of posttraumatic men- 1. Appelbaum E: Meningitis following trauma to head and face. JAMA 173:1818–1822, 1960 ingitis, CSF rhinorrhea, or CSF otorrhea significantly es- 2. Baltas I, Tsoulfa S, Sakellariou P, Vogas V, Fylaktakis M, calated the cost of hospital stay independent of subtype Kondodimou A: Posttraumatic meningitis: bacteriology, hy- of skull/facial fractures, comorbidities, and skin infection drocephalus, and outcome. Neurosurgery 35:422–427, 1994 (Table 6). 3. Brodie HA: Prophylactic antibiotics for posttraumatic cere- brospinal fluid fistulae. A meta-analysis. Arch Otolaryngol Limitations of the Study Head Neck Surg 123:749–752, 1997 4. Brown E, de Louvois J, Bayston R, Hedges A, Johnston R, When studying a population-based database like Lees P: Antimicrobial prophylaxis in neurosurgery and after NIS, coding errors are possible, although every effort was head injury. Lancet 344:1547–1551, 1994 made to extract the data on the basis of the ICD-9-CM 5. Bullock MR, Chesnut R, Ghajar J, Gordon D, Hartl R, Newell codes and the CCS codes from the database. DW, et al: Surgical management of depressed cranial frac- Previous studies have also suggested a role for pneu- tures. Neurosurgery 58 (3 Suppl):S56–S60, 2006 mocephalus in the development of posttraumatic men- 6. Choi D, Spann R: Traumatic cerebrospinal fluid leakage: risk factors and the use of prophylactic antibiotics. Br J Neuro- ingitis. This factor could not be included in our study surg 10:571–575, 1996 because there is no ICD-9 code for . It 7. Cooper MA, Kidner NL: Antimicrobial prophylaxis in neuro- is usually coded under ICD-9 code 348.89, which is a after head injury. Lancet 345:515, 1995 (Letter) heterogeneous diagnostic code for “other conditions of 8. Dagi TF, Meyer FB, Poletti CA: The incidence and prevention

6 Neurosurg Focus / Volume 32 / June 2012

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC Role of posttraumatic CSF leak in development of meningitis

of meningitis after . Am J Emerg Med JS, Malangoni MA: Post-traumatic meningitis: risk factors, 1:295–298, 1983 clinical features, bacteriology, and outcome. Internet J Neu- 9. Eftekhar B, Ghodsi M, Nejat F, Ketabchi E, Esmaeeli B: Pro- rosurg 2:2005 (http://www.ispub.com/journal/the-internet- phylactic administration of ceftriaxone for the prevention of journal-of-neurosurgery/volume-2-number-1/post-traumatic- meningitis after traumatic pneumocephalus: results of a clini- meningitis-risk-factors-clinical-features-bacteriology-and- cal trial. J Neurosurg 101:757–761, 2004 outcome.html) [Accessed May 3, 2012] 10. Eljamel MS: Antibiotic prophylaxis in unrepaired CSF fistu- 24. Prosser JD, Vender JR, Solares CA: Traumatic cerebrospinal lae. Br J Neurosurg 7:501–505, 1993 fluid leaks.Otolaryngol Clin North Am 44:857–873, vii, 2011 11. Friedman JA, Ebersold MJ, Quast LM: Persistent posttraumatic 25. Ratilal BO, Costa J, Sampaio C, Pappamikail L: Antibiotic cerebrospinal fluid leakage. Neurosurg Focus 9(1):e1, 2000 prophylaxis for preventing meningitis in patients with basilar 12. Hand WL, Sanford JP: Posttraumatic bacterial meningitis. skull fractures. Cochrane Database Syst Rev (8):CD004884, Ann Intern Med 72:869–874, 1970 2011 13. Helling TS, Evans LL, Fowler DL, Hays LV, Kennedy FR: In- 26. Rutland-Brown W, Langlois JA, Thomas KE, Xi YL: Inci- fectious complications in patients with severe head injury. J dence of in the United States, 2003. J Trauma 28:1575–1577, 1988 Head Trauma Rehabil 21:544–548, 2006 14. Jones SR, Luby JP, Sanford JP: Bacterial meningitis compli- 27. Sponsel C, Park JW: Recurrent pneumococcal meningitis. cating cranial-spinal trauma. J Trauma 13:895–900, 1973 Search for occult skull fracture. Postgrad Med 95:109–110, 15. Kaufman BA, Tunkel AR, Pryor JC, Dacey RG Jr: Meningi- 197, 1994 tis in the neurosurgical patient. Infect Dis Clin North Am 28. Stone JA, Castillo M, Neelon B, Mukherji SK: Evaluation 4:677–701, 1990 of CSF leaks: high-resolution CT compared with contrast- 16. Korinek A-M, Baugnon T, Golmard JL, van Effenterre R, Co- enhanced CT and radionuclide cisternography. AJNR Am J riat P, Puybasset L: Risk factors for adult nosocomial menin- Neuroradiol 20:706–712, 1999 gitis after craniotomy: role of antibiotic prophylaxis. Neuro- 29. van de Beek D, Drake JM, Tunkel AR: Nosocomial bacterial surgery 59:126–133, 2006 meningitis. N Engl J Med 362:146–154, 2010 17. La Fata V, McLean N, Wise SK, DelGaudio JM, Hudgins PA: 30. Villalobos T, Arango C, Kubilis P, Rathore M: Antibiotic pro- CSF leaks: correlation of high-resolution CT and multiplanar phylaxis after basilar skull fractures: a meta-analysis. Clin reformations with intraoperative endoscopic findings. AJNR Infect Dis 27:364–369, 1998 Am J Neuroradiol 29:536–541, 2008 31. Weisfelt M, van de Beek D, Spanjaard L, de Gans J: Nosoco- 18. Lui TN, Lee ST: Late posttraumatic meningitis with con- mial bacterial meningitis in adults: a prospective series of 50 cealed CSF otorrhea. Pediatr Neurosci 15:85–87, 1989 cases. J Hosp Infect 66:71–78, 2007 19. MacGee EE, Cauthen JC, Brackett CE: Meningitis follow- 32. Yilmazlar S, Arslan E, Kocaeli H, Dogan S, Aksoy K, Korfali ing acute traumatic cerebrospinal fluid fistula. J Neurosurg E, et al: Cerebrospinal fluid leakage complicating skull base 33:312–316, 1970 fractures: analysis of 81 cases. Neurosurg Rev 29:64–71, 2006 20. Palabiyikoglu I, Tekeli E, Cokca F, Akan O, Unal N, Erberktas I, et al: Nosocomial meningitis in a university hospital be- tween 1993 and 2002. J Hosp Infect 62:94–97, 2006 Manuscript submitted February 15, 2012. 21. Pappas DG Jr, Hammerschlag PE, Hammerschlag M: Cere- Accepted May 1, 2012. brospinal fluid rhinorrhea and recurrent meningitis. Clin In- Please include this information when citing this paper: DOI: fect Dis 17:364–368, 1993 10.3171/2012.5.FOCUS1269. 22. Park JI, Strelzow VV, Friedman WH: Current management of Address correspondence to: Anil Nanda, M.D., Department of cerebrospinal fluid rhinorrhea.Laryngoscope 93:1294–1300, Neurosurgery, Louisiana State University Health Sciences Center 1983 in Shreveport, 1501 Kings Highway, P.O. Box 33932, Shreveport, 23. Plaisier BR, Yowler CJ, Fallon WF, Likavec MJ, Anderson Louisiana 71130-3932. email: [email protected].

Neurosurg Focus / Volume 32 / June 2012 7

Unauthenticated | Downloaded 09/27/21 02:41 AM UTC