Introduction 

   Hydrocephalus is a well-known sequelae of aneurysmal subarachnoid hemorrhage (SAH), and the incidence after the initial rupture of aneurysm has been reported to range from 6 to 67%^{12)14)39)56)57)} Many factors associated with the onset of hydrocephalus after SAH have been reported,^33)44)45)55) but the precise period of onset and the pathophysiologic mechanism of hydrocephalus after aneurysmal SAH are not clearly understood yet. The mechanical obstruction of blood products and the fibrotic adhesion of ventricular drainage system may attribute to impaired CSF circulation.^{3)4)10)16)19)32)57)} Late onset of hydrocephalus sometimes can be misdiagnosed as psychological problem, may disturb the neurological improvements of poorer state of patients. Untreated and sustained state of hydrocephalus may cause futher 
irreversible damage to the brain.^{24)39)49)50)61)62)} But, not all patients with SAH suffer from hydrocephalus. Some group of patients, even with SAH and IVH, do not suffer from this sequelae. 
   Our aim of study is to review whether the intial perioperative extraventricular drainage has any effect on the late onset of shunt-dependent hydrocephalus, and to offer clarification of need for initial extraventricular drainage. 

Materials and Methods 

Patient Population 
   Between January, 2004 and December, 2006, total 349 patients with aneurysmal SAH were treated at our two seperated institutions. Patient's neurological conditions were classified according to the WFNS⁹⁾ and Fisher scales.²⁶⁾ Patients who died within a week after the onset of hemorrhage were excluded from the study. Remained 311 of 349 patients were included in this study. Total 283 patients were treated with surgical aneurysmal clipping, 28 patients were treated with endovascular intervention. 
   All patients underwent computerized tomography (CT) angiography²⁷⁾ or cerebral angiography before the treatment. Each case was evaluated by a vascular neurosurgeon as well as an interventional neuroradiologist, and the decision to treat the aneurysm via surgical or endovascular approach was based on the anatomical location of the aneurysm, presence or absence of intracerebral or intraventricular hemorrhage, and the inclination of patient's family. Patient ages ranged from 26 to 82 years, with a mean age of 56±11 years. The female/male ratio was 1.6 : 1. 
   Patients selected for surgical treatment underwent pterional craniotomy with opening of the cisterns, hematoma evacuation if necessary, and clip occlusion of the aneurysm. In most of the surgery, the sylvian fissure was opened before aneurysmal clipping, according to the technique of Yasargil⁶⁰⁾ and the endovascular proedure was performed in 28 patients by using Guglielmi Detachable Coil. 

External Ventricular Drainage 
   External ventricular drainage (EVD) was performed selectively in patients with brain swelling, especially patients who were in WFNS Grade II or greater, alteration of consciousness, radiological evidence of acute hydrocephalus or thick intraventricular hemorrhage. Total 128 of 311 patients (41.2%) underwent initial perioperative EVD (Table 1). Most of EVD procedures were performed intraoperatively prior to the clipping, maintained for a mean 4.79±0.14 days postoperatively. The EVD drip chamber was positioned within above or below 10cm from the level of foramen of Monroe, and daily checked the amount of drainage simultaneously. 

Late onset of Hydrocephalus 
   The late onset of hydrocephalus was determined in the following manner. Before discharge, all patients underwent at least one CT scan. Pre- and Post-operative ventricular volume in the frontal horn was ascertained using the Evans Index (ratio of frontal horn to maximal biparietal diameter). As long as the Evans index remained over 30% in follow up CT scan, triad symptoms like gait disturbance, dementia, and urinary incontinence were presented after 2 weeks later the onset of SAH, the patients were considered to be in late onset of hydrocephalus.²⁸⁾²⁹⁾ If symptoms prolonged or devastated after the conservative treatment in some period, ventricular-peritoneal (VP) shunt was performed. Asymptomatic patients with enlarged ventricular size was followed up via out-patient doctor. Most of the shunt surgery was procedured using programmable valve (Aesculap rproGAV. programmable valve), and the suitable valve pressure was modified as the follow up CT scan. No surgical or mechanical complication related with the shunt surgery was detected in this periods. There were two cases of shunt catheter infection which needed re-operation. 

Clinical and Radiological Conditions in Patients 
   Three subgroups of without the late onset of hydrocephalus, with the late onset of hydrocephalus but did not need shunt operation, and with the late onset of hydrocephalus that needed shunt operation were compared based on clinical conditions, according to the WFNS grading scale, and the amount of subarachnoid hemorrhage, according to the Fisher scale. According to the initial amount of IVH, patients were categorized into 4 groups; 1) No IVH, 2) scanty (IVH not exceed one ventricular chamber), 3) moderated (IVH not exceed half of whole ventricle), 4) large (IVH exceed half of whole ventricle). Clinical and radiological data was retrospectively reviewed with one of the authors, and a subgroup analysis was performed in patients with better clinical and radiological conditions (that is, WFNs Grades I-III, Fisher Grades 1-3) to diminish statistical bias because patients in poor condition are more likely to undergo embolization and to have more chance of EVD.²⁶⁾ Preoperative patient's consciousness level was also evaluated with Hunt-Hess classification,^35)36)37) and patient's outcomes were assessed with the modified Ranking scale.⁴⁰⁾

Statistical Analysis 
   A comparison between EVD groups was made using the chi-square test of independence or Fisher's exact test. These tests were two-sided, and a probability value of 0.05 or less was regarded as statistically significant. Categorical variables like patient's age, sex, Hunt-Hess grade, Fisher group, WFNS grade on admission, initial amount of IVH, perioperative EVD, and the site of ruptured aneurysm were cross-tabulated analyzed with dependent variables like an onset of asymptomatic hydrocephalus, or shunt-dependent hydrocephalus. To assess effects of initial perioperative EVD, a logistic regression model was fit using the response variables listed in Table 3. In these models we controlled for risk factors by including them in the models: patients sex, age, WFNS grade on admission (WFNS Grades I-III compared with Grades IV-V), Fisher grade (Grade 1,2 compared with Grade 3,4), Hunt-Hess Grades (Grades I-III compared with Grades IV-V), and additional measuring of the amount of IVH (Group 1,2 compared with Group 3,4) were included in the models because these factors may influent the risk of shunt-dependent hydrocephalus. Because there were two possible outcomes for the onset of hydrocephalus or the shunt dependency (yes or no), we applied an ordinal logistic regression models, the OR for the onset of asymptomatic hydrocephalus or shunt-dependent hydrocephalus was interpreted as the relative odds (Table 3). Consequently, an OR significantly lower than 1 would indicate that the occurrence would be less likely to occur, whereas an OR significantly greater than 1 would be the risk factor for the onset of asymptomatic or shunt-dependent hydrocephalus. 

Results 

   The late onset of hydrocephalus in initial EVD placement group was observed in 76 (24.4%), and 42 (13.5%) in patient group without initial EVD placement. Among patient group with initial EVD placement, 23 (7.4%) underwent shunt placement, and 25 (8.0%) needed shunt operation in patient group without initial EVD placement. Based on univariate analysis, the rate of shunt dependency was 49.4% (23 of 128 patients) in initial EVD group, and 13.7% (25 of 183 patients) in without EVD group. But there were no statistical significance between the perioperative EVD and the onset of shunt-dependent hydrocephalus (p=0.190). 
   Results of chi-square test or Fisher's exact test, female patients, Hunt-Hess grade, Fisher group, WFNS grade, Amount of IVH on admission, and initial EVD placement showed statistical significance related with the late onset of hydrocephalus. But, in consider with shunt-dependency, female patients, Fisher group, WFNS grade, amount of IVH, and site of ruptured aneurysm were identified as related factors statistically (Table 2). 
   Results of logistic regression analysis controlling for patient's Hunt-Hess grade, WFNS grade, Fisher grade, amount of IVH, and EVD placement are featured in Table 4. Overall, initial perioperative EVD placement showed statistical significance related with the late onset of hydrocephalus, but risk odd ratio was lower than other factors like poor Hunt-Hess grade, Fisher grade, and large amount of IVH (OR 0.408, 95% CI 0.224-0.745). In related with shunt-dependent hydrocephalus, initial EVD placement showed OR 0.931, 95% CI 0.448-1.934, but there were no statistical significance (p=0.847). 
   Among various factors used in logistic regression analysis, poor neurological condition on admission (Hunt-Hess grade IV, V) was shown as the valuable risk factor (OR 5.966, 95% CI 1.858-19.157) related with the late onset of hydrocephalus, and higher Fisher grade (Fisher grades III~IV) identified as the most higher risk factor of the late onset of shunt-dependent hydrocephalus (OR 3.546, 95% CI 1.537-8.181) (Table 4).

Discussion 

   Hydrocephalus related to SAH was first described as early as 1928 and by the 1950s was well recognized.^2)12)15)17) After this recognition, many neurosurgical efforts was made to improve CSF circulation, and shunt surgery emerged as the standard treatment modality of this condition.^34)38)39)55) In the International Cooperative Study on the Timing of Aneurysm Surgery,³⁰⁾⁴⁵⁾ 22.3% of the population suffered from asymptomatic or shunt-dependent hydrocephalus after initial rupture of the aneurysm. Eight percent of the population underwent definitive shunt placement compared with 15.4% in our study. 
   Many reports have been published addressing the relationship between the age and initial neurological and radiological condition in patients with aneurysms who develop hydrocephalus, and older patients in poor neurological condition have been known to be more likely to develop shunt-dependent hydrocephalus.^2)17)43) In our series, there were no significant different age distribution in patients group with or without late onset of hydrocephalus. Hunt- Hess grades, Fisher groups, WFNS grades, amount of IVH, and the site of ruptured aneurysm were also shown as statistical related factors with both of the onset of asymptomatic hydrocephalus and shunt-dependent hydrocephalus (Table 2). 
   The effect of IVH on the development of hydrocephalus has been well established.^{1)34)38)54)56)} In our series, Fisher groups and the amount of IVH significantly related with the late onset of hydrocephalus, and also contributes as the high risk factor of shunt-dependent hydrocephalus. Some authors assert that the presence of blood clots and high CSF viscosity can lead to an obstructive form of hydrocephalus and early CSF distrubances.^21)32)59) In one study, investigators noted no relationship between the development of hydrocephalus and the number of erythrocytes in the CSF, thus indicating that flow distrubances also play an important role in late onset of hydrocephalus.⁶⁾ Authors focused on the possibility of the flow disturbance of CSF circulation and the development of hydrocephalus, analyzed clinical correlation with the initial EVD placement as a significant factor of interrupting natural CSF drainage. In the present study, initial EVD placement showed statistical significance related with the late onset of hydrocephalus, but it's risk odd ratio was too low that below 1, and futhermore had no relationship with the onset of shunt-dependent hydrocephalus. Authors can conclude that initial EVD placement in aneurysmal SAH patients plays role in developing the late onset of hydrocephalus, but had no significant relationship with shunt-dependent hydrocephalus. 
   Fuhrmeister, et al.,¹³⁾ reported CSF outflow resistance to be elevated threefold in patients with SAH. Although CSF hydrodynamics become normalized in the majority of patients, physiological CSF outflow resistance may not occur until 40 to 50 days post-SAH. In cases of arachnoid scarring and blockage of arachnoid granualations, CSF outflow resistance can remain elevated and a state of chronic hydrocephalus may develop. In the present series, definitive shunt insertion was performed a mean of 36 days after SAH, that is at a time when CSF hydrodynamics should have been normalized. 
   Many investigators have examined the association between aneurysm location and hydrocephalus.^8)17)34) The conclusions are rather confounding, with some researchers asserting that posterior circulation aneurysms are more frequently associated with shunt dependency,⁸⁾¹⁹⁾ and others showing a clear correlation between the anterior circulating aneurysms and the development of shunt dependent hydrocephalus.⁶⁰⁾ In the study by Gruber, et al.,¹⁸⁾ there was no relationship between aneurysm location and the development of shunt-dependent hydrocephalus. In our series, site of ruptured aneurysm revealed statistical significance related with the onset of shunt-dependent hydrocephalus (p=0.011), but total amount of anterior circulating aneurysms are about three times larger than posterior circulating aneurysms. There could be bias related with it. 

Limitations of Our Study 
   The patients in this study were treated using craniotomy and clip application or an endovascular approach. Although patients in both treatment groups were comparable on admission data, only random allocation of patients to the treatment arms could have ensured control for unrecognized risk factors for shunt dependency. 
   The treatment groups in the present study were different in terms of the number of patients. This discrepancy is due to the fact that the majority of ruptured aneurysms are still being treated using surgery and clip application at our institution. 
   Also, it is clearly known that patient group with poor radiological and neurological condition may have more chance to undergo initial EVD. Cohort study not only with similar patient's status but also with strictly controlled related factors may allow us more clear, sophisticated results. 

Conclusion 

   Initial EVD placement in SAH patient had no significant risk of shunt-dependent hydrocephalus, even though low risk of late onset of hydrocephalus. Patient's poor radiological and neurological condition had more higher risk value rather than EVD placement. Neurosurgeons better not to hesitate to place EVD for the patient of poor clinical condition. 

REFERENCES

1. Askenasy HM, Herzberger EE, Wijsenbeek HS. Hydrocephalus with vascular malformations of the brain; a preliminary report. Neurology 3:213-20, 1953 

2. Bailey OT, woodard JS. Small vascular malformations of the brain: their relationship to unexpected death, hydrocephalus and mental deficiency. J Neuropathol Exp Neurol 18:98-113, 1959 

3. Blasberg R, Johnson D, Fenstermacher J. Absorption resistance of cerebrospinal fluid after subarachnoid hemorrhage in the monkey, effects of heparin. Neurosurgery 9:686-91, 1981