Skip Navigation LinksHome > December 2013 - Volume 73 - Issue 6 > Inferior Short-term Safety Profile of Endoscopic Third Ventr...
doi: 10.1227/NEU.0000000000000129
Research-Human-Clinical Studies: Editor's Choice

Inferior Short-term Safety Profile of Endoscopic Third Ventriculostomy Compared With Ventriculoperitoneal Shunt Placement for Idiopathic Normal-Pressure Hydrocephalus: A Population-Based Study

Chan, Andrew K. BS; McGovern, Robert A. MD; Zacharia, Brad E. MD; Mikell, Charles B. MD; Bruce, Sam S. MA; Sheehy, John Paul BA; Kelly, Kathleen M. BA; McKhann, Guy M. II MD

Free Access
Supplemental Author Material
Editor's Choice
Article Outline
Collapse Box

Author Information

Department of Neurological Surgery, Columbia University Medical Center, New York, New York

Correspondence: Andrew K. Chan, 630 West 168th Street, Room 5-454, New York, NY 10032. E-mail:

* These authors contributed to this manuscript equally.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (

Received April 16, 2013

Accepted July 31, 2013

Collapse Box


BACKGROUND: In small series, endoscopic third ventriculostomy (ETV) has been shown to potentially have efficacy similar to that of ventriculoperitoneal shunting (VPS) for idiopathic normal-pressure hydrocephalus (iNPH). Therefore, some clinicians have advocated for ETV to avoid the potential long-term complications associated with VPS. Complication rates for these procedures vary widely based on limited small series data.

OBJECTIVE: We used a nationwide database that provides a comprehensive investigation of the perioperative safety of ETV for iNPH compared with VPS.

METHODS: We identified discharges with the primary diagnosis of iNPH (International Classification of Diseases, Ninth Revision code 331.5 [ICD-9]) with ICD-9 primary procedure codes for VPS (02.34) and ETV (02.2) from 2007 to 2010. We analyzed short-term safety outcomes using univariate and hierarchical logistic regression analyses.

RESULTS: There were a total of 652 discharges for ETV for iNPH and 12 845 discharges for VPS for iNPH over the study period. ETV was associated with a significantly higher mortality (3.2% vs 0.5%) and short-term complication (17.9% vs 11.8%) rates than VPS despite similar mean modified comorbidity scores. On multivariate analysis, ETV alone predicted increased mortality and increased length of stay when adjusted for other patient and hospital factors.

CONCLUSION: This is the first study that robustly assesses the perioperative complications and safety outcomes of ETV for iNPH. Compared with VPS, ETV is associated with higher perioperative mortality and complication rates. This consideration is important to weigh against the potential benefit of ETV: avoiding long-term shunt dependence. Prospective, randomized studies are needed.

ABBREVIATIONS: CI, confidence interval

ETV, endoscopic third ventriculostomy

ICD-9, International Classification of Diseases, Ninth Revision

iNPH, idiopathic normal-pressure hydrocephalus

NIS, Nationwide Inpatient Sample

VPS, ventriculoperitoneal shunting

Idiopathic normal-pressure hydrocephalus (iNPH) is defined by the clinical triad of gait impairment, urinary incontinence, and cognitive decline in the presence of radiologic evidence of hydrocephalus—enlarged ventricles without significant atrophy—and normal cerebrospinal fluid (CSF) pressure.1,2 Once clinically diagnosed, patients frequently undergo a large-volume lumbar puncture or extended trial of CSF drainage via a lumbar drain with clinical assessment before and after the procedure to help determine whether CSF diversion would be beneficial. Surgical options typically include ventriculoperitoneal shunting (VPS), ventriculoatrial shunting, and lumboperitoneal shunting, with VPS being the most commonly used surgical procedure for iNPH.3-5 Although believed to have low perioperative morbidity and mortality rates, shunting procedures require permanent implantation of hardware with resultant risks of infection and malfunction requiring additional surgical procedures. Conversely, endoscopic third ventriculostomy (ETV), a procedure shown to be efficacious for obstructive hydrocephalus,6 obviates the need for long-term shunt dependence and has been reported as a possible treatment for iNPH as early as 1999 by Mitchell and Mathew.7

ETV has been shown to have efficacy similar to that of VPS for iNPH, with success rates generally reported around 70%.5,7-13 This is comparable to the overall reported success rates of VPS in retrospective studies (61%-75%)14-16 and prospective studies (∼80%).17,18 However, other studies report the superiority of VPS.19,20 Still, as ETV spares patients from shunting, some advocate using ETV with placement of a VPS only in the absence of subsequent clinical improvement.8 However, patients and surgeons must balance the similar efficacy of ETV and VPS with the perioperative safety outcomes, including short-term complications and mortality.19

Studies that report on series of patients from the perioperative period to longer term follow-up estimate the overall rate of complications of VPS for iNPH to range from 3% to 37.9%,8,10,15,17,18 with perioperative mortality rates estimated at 0.4%.18 For ETV, several small series have reported varying complication rates. A report early in the experience of ETV for iNPH reported a group of 15 iNPH patients in which pneumatocephalus developed in 1 patient and an ischemic thalamic lesion developed in 1 patient.10 Subsequently, 3 studies collectively including 49 iNPH patients report no mortalities, and aside from a single case of intracerebral hemorrhage, no major intra- or perioperative complications.8,13,21 A larger, more recent study reports complication rates of 6.4% (in a series of 110 patients),9 and another revealed significantly lower complication rates for ETV for iNPH patients.12 Because of the small size of these studies, the generalizability to the use of ETV in the iNPH population is limited.

Thus, it remains necessary to investigate the safety of ETV specifically as a surgical option for iNPH, compared with VPS—outside the confines of small series studies—to ensure a fully informed decision on the optimal surgical treatment for iNPH. To investigate the perioperative safety outcomes of ETV, we used a nationwide database that provides a comprehensive investigation of ETV for iNPH. We examined the nationwide inpatient sample (NIS) for ETV compared with VPS in patients with iNPH since the ICD-9 code for iNPH was introduced in October 2007.

Back to Top | Article Outline


The NIS was developed as part of the Healthcare Cost and Utilization Project from the Agency for Healthcare Research and Quality (Rockville, Maryland). It is the largest publicly available inpatient hospital database that catalogs discharge data from a stratified sample of approximately 20% of all nonfederal hospitals in the United States. For this study, we obtained the NIS data for the period October 2007 to 2010. This period was selected because October 2007 marked the implementation of a distinct ICD-9 code expressly for iNPH. Because the NIS is drawn from a sample that includes all patients discharged from sampled hospitals, the data can be used to estimate total annual case volumes, complications, and deaths among other variables for all nonfederal hospitals in the United States. The dataset presented in this study is thus extrapolated using the weighting variables provided by Healthcare Cost and Utilization Project.

Back to Top | Article Outline
Patient Population

ICD-9 codes were used to identify discharges with a primary diagnosis of iNPH (code 331.5). iNPH discharges with a primary procedure code for ventriculostomy (02.2) or ventricular shunt to abdominal cavity and organs (02.34) were then identified. Although the code for ventriculostomy is used for both catheter ventriculostomy (external ventricular drain) and ETV, we assume only a negligible few are not ETVs given a concomitant primary diagnosis code of iNPH. For clarity, we subsequently refer to discharges generated in this manner as discharges pertaining to ETV. All discharges since the introduction of the iNPH ICD-9 code on October 1, 2007, are included in the aggregate descriptive, univariate, and hierarchical analysis. However, for analyses comparing annual data, only the full-year data from 2008 through 2010 are used.

A small proportion of discharges with a primary diagnosis code for ETV held a secondary concomitant diagnosis code for VPS. These discharges are presumed to be either (1) previously shunted patients now receiving ETV or (2) ETV patients who were subsequently shunted during the same hospitalization. There were no discharges with a primary diagnosis code for VPS that had a secondary, concomitant diagnosis code for ETV.

Back to Top | Article Outline
Patient Characteristics

Age, sex, race, income quartile of the patient's ZIP code, and payer status were coded in the database. Medical comorbidities were defined using a modified version of the Elixhauser comorbidity score22 that excluded the 2 neurological comorbidity variables “other neurological deficit” and “paralysis,” such that the highest possible comorbidity score was 28. Payer status was defined by combining the primary and secondary payer variables and placing discharges into the exclusive categories: private insurance, Medicaid without private insurance, Medicare with neither private insurance nor Medicaid, and other.

Back to Top | Article Outline
Hospital Characteristics

The NIS provides data on hospital size (small, medium, large), region (Northeast, Midwest, South, West), location (urban or rural), teaching hospital status, as well as hospital caseload for a specific procedure by discharge number. Hospital caseload was calculated by number of discharge records containing a primary procedure code of either ventriculostomy (02.2) or ventriculoperitoneal shunt placement (02.34) with a concomitant primary diagnosis code of iNPH (331.5).

Back to Top | Article Outline
Complications/Outcomes Definitions

We identified a number of potential complications resulting from VPS or ETV. The presence of any of these complications was included in an outcome variable that we created termed any complication. The complications included in our analysis and their respective ICD-9 codes are included in Table 1.

Table 1
Table 1
Image Tools

We also examined secondary outcomes provided in the dataset including length of stay, discharge status, and mortality. We examined the status of all other outcomes (short-term hospital, skilled nursing facility, and death), referred to as nonhome discharge.

Back to Top | Article Outline
Statistical Analysis

We used univariate comparisons to illustrate trends in our data, primarily comparing the population in different years. Analyses were performed with Wilcoxon rank sum test or a χ2 test, with Yates correction for continuity for tests using 1 df, using built-in and custom scripts (Matlab; Mathworks, Natick, Massachusetts). For the univariate analyses, we adjusted discharge-level data by the associated weighting variable provided by the NIS.

We used a mixed, or hierarchical, logistic regression model to analyze outcomes given the nested data architecture of NIS. A hierarchical multivariable logistic regression model provides a multilevel analysis that can account for the fact that patients are clustered within hospitals rather than assuming independence of observations as a traditional linear logistic regression model would. We performed 3 separate regressions using SAS (SAS Institute, Cary, North Carolina) procedures GLIMMIX and MIXED to evaluate predictors of length of stay, nonhome discharge, the presence of complications, and death. Age, sex, race, mean income of patient's ZIP code, payer status, and comorbidity score were coded as patient-related variables with hospital caseload, hospital size, geographic region, teaching status, and year performed as hospital- or nonpatient-related variables with consensus-based inclusion within the model. Length of stay was log-transformed before inclusion in the regression models. Continuous variables (age, mean modified comorbidity score, hospital experience, and length of stay) were grand mean centered. Year of discharge was mean centered by a unique hospital identification code. We imputed missing data using single imputation with deterministic regression modeling. The unique hospital identification code was coded as the nesting variable. P < .05 was considered statistically significant. Bonferroni correction was used for the univariate comparisons (α = .002). Results are reported as ± standard error.

Back to Top | Article Outline


Patient Characteristics

From October 1, 2007, through 2010, there were 652 discharges for ETV for iNPH from nonfederal hospitals in the United States. During this same time period, there were 12 845 discharges for VPS for iNPH. Figures 1A and 1B show the discharges by year from 2008 through 2010 for both ETV and VPS for iNPH. The average number of iNPH discharges for ETV annually increased significantly from 2008 to 2010, whereas the number of iNPH discharges for VPS annually did not change significantly from 2008 to 2010.

Figure 1
Figure 1
Image Tools
Back to Top | Article Outline
Endoscopic Third Ventriculostomy

Characteristics of ETV discharges have remained stable over time. The average patient age of ETV discharges was 72.4 ± 1.2 years. The mean modified comorbidity score was 2.0 ± 0.2. Of the total number of ETV discharges, 10.0% had a concomitant ICD-9 diagnosis code for VPS—putatively representing VPS before or during hospitalization.

With regard to hospital characteristics, hospitals discharging patients with ETV had an average annual caseload of 1.8 ± 0.1 surgeries per year. This has remained stable over time (mean, 1.8 cases; range, 1.7-2.1 surgeries). However, the percentage of cases performed at nonteaching hospitals has increased (18.9% vs 39.5%, P < .01) (Figure 1C). Table 2A shows the patient and hospital characteristics of iNPH discharges for ETV in 2007 through 2010.

TABLE 2-a Characteri...
TABLE 2-a Characteri...
Image Tools

The average length of stay was 6.1 ± 0.75 days. There were complications during hospitalization in 17.9% of discharged patients; 10.3% of discharged patients had intracranial complications, whereas 11.6% of discharged patients had extracranial complications. The complication percentage for all discharged patients comprised ICD-9 complication codes for (1) urinary tract infections (11.6%), (2) hemorrhage or infarction (3.9%), (3) mechanical complications (2.5%), (4) infections related directly to a mechanical device (1.7%), (5) seizures (1.5%), and (6) hematoma, hemorrhage, or seroma (0.8%). Only 45.0% of discharges were coded as “routine.” Of those patients discharged nonroutinely, 40.0% were transferred to a skilled-nursing facility or other long-term care facility, 11.0% were discharged with home health care, and 0.8% were transferred to another hospital; 3.2% of admitted patients died during hospitalization.

TABLE 2-b Characteri...
TABLE 2-b Characteri...
Image Tools
Back to Top | Article Outline
Ventriculoperitoneal Shunting

The average age of discharges for VPS for iNPH was 74.8 ± 0.2 years. The mean modified comorbidity score was 1.8 ± 0.03. None of the discharged patients received an ETV during hospitalization.

Overall, hospitals that discharged patients with VPS had an average annual caseload of 9.5 ± 0.2 surgeries, which has remained constant over time (range, 7.3-11.7). Discharges from teaching hospitals have remained constant (66.6% vs 62.0%, P = .34). Table 2B shows the patient and hospital characteristics of iNPH discharges for VPS in 2007 through 2010.

The average length of stay during hospitalization was 4.0 ± 0.1 days; 11.8% of discharged patients had a complication during hospitalization, 6.0% of discharged patients had an intracranial complication, whereas 7.0% of discharged patients had an extracranial complication. For all discharged patients, complications were coded as (1) urinary tract infections (5.7%), (2) seizures (3.1%), (3) mechanical complications (1.2%), (4) hemorrhage or infarction (1.0%), (5) other complications (0.7%), (6) abdominal injury (0.4%), (7) hematoma, hemorrhage, or seroma (0.4%), (8) infections related to a mechanical device (0.3%), (9) deep venous thrombosis/pulmonary embolism (0.1%), and (10) meningitis, ventriculitis, or intracranial abscess (0.1%). As for discharge method, 45.6% of discharges were “routine.” Of the nonroutine discharges, 38.8% of patients were discharged to a skilled-nursing facility or other long-term care facility, 14.3% were discharged to home health care, and 0.6% were transferred to another hospital; 0.5% of admissions resulted in death and 0.2% of patients were alive but discharge destination was reported as unknown.

Back to Top | Article Outline
Examination of Safety Outcome Differences Between ETV and VPS for iNPH Discharges Using a Univariate Analysis

Admissions for ETV had a greater number of perioperative deaths (3.2% vs 0.5%, P < .001), despite similar mean modified comorbidity scores (Figure 1D), mean age, and teaching hospital status of the hospital at which the procedure was performed. ETV was also associated with higher overall complication rates (17.9% vs 11.8%, P < .001) including when assessing for only intracranial and extracranial complications (10.3% vs 6.0%, P < .001; and 11.6% vs 7.0%, P < .001, respectively).

There is a significant difference in discharge disposition (P < .001), which is driven by the significant difference in perioperative deaths. There was an increased length of stay (6.1 vs 4.0 days, P < .001) for discharged ETV patients, which were from hospitals with less experience performing their respective primary procedure for iNPH (1.8 vs 9.5 surgeries, P < .001). Finally, ETVs were performed at the largest centers (P < .001), although teaching status and urban location did not differ.

Back to Top | Article Outline
Predictors of Short-term Outcomes for ETV Using a Hierarchical Logistic Regression Model

Length of stay increased during the study period from 2008 through 2010, although the result did not reach statistical significance (4.5 vs 8.0 days, P = .09) (Figure 2B). Examining factors important in length of stay, our multivariate model showed that increased comorbidity score and the presence of any complication increased length of stay (percentage of change per unit of change = 22.5%; 95% confidence interval [CI]: 10.7-35.6%; P < .001; and percentage of change per unit change = 126.8%; 95% CI: 51.3-239.8%; P < .001, respectively), whereas nonteaching hospital status predicted decreased length of stay (percentage of change per unit of change = −35.0%; 95% CI: −53.3 to −9.8%; P = .01, respectively). Of note, we found that each subsequent year that ETVs were performed significantly predicted increased length of stay (percentage of change per unit of change = 19.1%; 95% CI: 2.0-39.1%; P = .03). Increased annual caseload predicted increased discharge length of stay (percentage of change per of unit change = 12.2%; 95% CI: 0.6-25.1%; P = .04). None of the other patient- and hospital-related variables significantly contributed to length of stay. Table 3 shows the percentage of change per unit of change for patient- and hospital-related variables included in the hierarchical model for length of stay.

Figure 2
Figure 2
Image Tools
Table 3
Table 3
Image Tools

Nonhome discharges (those discharges excluding routine discharges and those discharges including home health care) increased significantly during the study period (32.5% vs 54.8%, P < .01) (Figure 2C). Our hierarchical model showed that mean modified comorbidity score significantly contributed to nonhome discharges (odds ratio [OR]: 1.69; 95% CI: 1.22-2.36; P = .003). Increasing age also significantly predicted non-home discharge (OR: 1.07; 95% CI: 1.01-1.13; P = .02). Notably, the presence of complications did not significantly increase the percentage of those being nonhome discharged (P = .09). None of the remaining patient- and hospital-related variables significantly contributed to nonhome discharge. Table 3 shows the OR for the unit of change for patient- and hospital-related variables included in the hierarchical model for nonhome discharge.

The number of perioperative deaths did not change from 2008 to 2010. Examining factors important in perioperative mortality, the hierarchical model showed that the mean modified comorbidity score contributed to an increase in mortality (OR: 2.50; 95% CI: 1.17-5.35; P = .02). Of note, the year in which ETV was performed, hospital teaching status, and hospital caseload had no significant effect on mortality. Table 3 shows the OR for unit of change for variables included in the hierarchical model for mortality.

The overall percentage of complications has stayed steady over time (mean, 17.9%; range, 12.2%-22.0%) (Figure 2A). Our model did not find a significant predictor of complications. Table 3 shows the OR for unit of change for patient- and hospital-related variables included in the hierarchical model for complications.

The hierarchical logistic regression models for length of stay, nonhome discharge, mortality, and complications were repeated for VPS, and the results are included in Supplemental Digital Content 1 (see Table S1,, which illustrates the outcome predictors for VPS for iNPH).

Back to Top | Article Outline
ETV Predicts Increased Mortality and Length of Stay Using a Hierarchical Logistic Regression Model

We repeated the 4 outcome analyses with the addition of a binomial procedural variable (ETV or VPS). When adjusting for other patient- and hospital-related variables, ETV predicted increased mortality (OR: 4.24; 95% CI: 1.14-15.63; P = .03) and increased length of stay (percentage of change per unit of change = 123.5%; 95% CI: 109.1-138.9%; P < .001). Of note, ETV approached, but did not reach, significance in predicting the presence of complications (OR: 1.64; 95% CI: 0.997-2.70; P = .052). Nonhome discharge was not predicted by procedure type alone. The results are summarized in Supplemental Digital Content 2 (see Table S2,, which illustrates the outcome predictors of iNPH with inclusion of procedure type as a predictor variable).

Back to Top | Article Outline


Primary Outcomes: Mortality and Morbidity

Our study is the first to find that in a multicenter, nonselected hospital setting, ETV is associated with a significantly increased perioperative mortality and complication rates than VPS for iNPH. Perioperative mortality in the current study is sevenfold higher in ETV discharges compared with discharges for VPS and is more than 10-fold higher than previously reported.23 Indeed, when adjusted for patient and hospital variables, ETV alone significantly predicted increased mortality relative to VPS. Although not specific to ETV for iNPH, Bouras and Sgouros23 conducted a systematic literature review of ETV in published studies from 1998 through 2010 (in which only 1.3% were iNPH patients) and estimated that the postoperative mortality rate was 0.22% (vs 3.2% as in our study).

Importantly, the overall ETV complication rate here is also higher than the rates of previously reported in smaller series investigating ETV for iNPH (4.7%, when combining the early case series of several studies; range, 0%-13.3%), larger series (6.4%), and that estimated in the systematic review by Bouras and Sgouros including intraoperative, postoperative, and delayed adverse events (not specific to iNPH) (8.8%; range, 2.9%-16.1%).8,9,12,13,21,23 However, the study complication rates approach our rates when considering only intracranial complications (10.3% in our study). Still, we chose to report overall complications, including urinary tract infections, pulmonary embolism, and pneumonia because these remain important contributors to perioperative morbidity, especially when considering the increased length of stay of patients receiving ETV.

One crucial aspect of the consideration of ETV for iNPH is a comparison of long-term complications. However, as demonstrated in our study during the perioperative time period, the type of long-term complications may differ between the 2 treatments (eg, seizures vs hemorrhage for VPS vs ETV, respectively). In addition, the number and type of reoperations may differ as well. Indeed, Hailong et al21 report that 4 of 17 patients (23.5%) undergoing ETV for iNPH underwent subsequent VPS. The possibility of a second ETV (estimated at 3.6%9) and/or subsequent shunt placement in the absence of clinical improvement from initial ETV for iNPH remains a vital portion of the discussion of ETV for iNPH.

Recommendations to use ETV primarily in the treatment of iNPH rely on the downside of lifelong shunt dependence as a key justification.8,15 However, the long-term efficacy of ETV has not been shown to be superior. Indeed, it has been reported to be similar5,7-18 or inferior19,20 to VPS. Given the current finding of the inferiority of perioperative complications and mortality of ETV, a long-term, prospective study is vital to explore the appropriateness of ETV for iNPH, as is currently being undertaken by Pinto et al.20

Back to Top | Article Outline
Secondary Outcomes: Length of Stay and Discharge Disposition

Discharges for ETV in our study were associated with increased length of hospital stay. As ETV for iNPH was associated with a greater number of complications in our study, it is not surprising that there was an increased length of stay in the ETV population. To this point, the presence of complications was a significant predictor in our hierarchical model.

Our results indicate that an increasing number of ETV procedures are being performed annually and that there is a dispersion of cases from teaching hospitals to nonteaching hospitals. The low hospital annual caseload throughout all 3 study years (1.8 ETVs annually at hospital of discharge) despite the increasing number of total cases of ETV suggests that new centers are expanding their breadth of surgical treatments for iNPH. Given that these nonteaching, inexperienced hospitals constitute an increasingly larger proportion of ETV for iNPH centers, it is not surprising to see proxies of short-term outcome, such as length of hospital stay after ETV, to be significantly predicted by year of procedure. Somewhat paradoxically, however, our multivariate analysis reveals a systematic difference in ETV performed at nonteaching hospitals, as this setting predicted a decreased length of stay.

To assess this finding, we ran a supplemental analysis (see Table S3, Supplemental Digital Content 3,, which demonstrates the characteristics of discharges for ETV for iNPH) and did not find any obvious predictive variables. There was a greater proportion of discharges at nonteaching hospitals with Medicare than with private insurance. Other important patient and hospital variables did not differ, including mean modified comorbidity score. In fact, discharges at nonteaching hospitals had a nonsignificantly increased modified comorbidity score. Discharges at nonteaching hospitals were also more likely to be routine. Our study suggests that despite an increasing number of ETV surgeries at nonteaching hospitals in the context of an increasing, albeit nonsignificant, patient comorbidity score trend, mortality and complication rates of ETV have not changed. It will remain important to address the impact on the efficacy and safety outcomes of ETV for iNPH as a greater number of centers gain experience with the procedure.

As for discharge method, discharge disposition for ETV and VPS differed significantly, with the effect driven by the aforementioned difference in mortality. Interestingly, removing those who died perioperatively, virtually identical numbers of discharges were to home (routinely or to home health care) or transferred to another health care setting, suggesting that both procedures provide a comparable amount of short-term clinical benefit (ie, dependence vs independence in daily living). Our data reveal that 56.0% and 60.0% of discharges for both ETV and VPS, respectively, were discharged to home, including home health services. This number is lower than that provided by a previous study that stated discharge disposition was to home in 84.9% of nonselected patients receiving a ventricular shunt (to peritoneal, atrial, pleural, and urinary systems and for initial placement, revision, and removal) in the United States,28 although this may be a result of either patient selection or the elderly iNPH population seen in our study. Comparing more meaningful measures of long-term outcome such as the Karnofsky Performance Scale in the context of perioperative and long-term morbidity and mortality will be key in the assessment of both procedures for iNPH and must be further studied in future prospective studies.

Back to Top | Article Outline
Patient Selection: Predictor of Short-term Safety Outcomes

When adjusting for patient and hospital factors, patient comorbidity score remained the lone consistent predictor of mortality, length of stay, and nonhome discharge. The fact that healthier patients have better neurosurgical outcomes,29-31 which remains true for ETV, underlies the importance of proper patient selection for ETV for iNPH. Aside from medical comorbidities, multiple studies have attempted to better define inclusion criteria to maximize ETV benefit in iNPH. Gangemi et al,8 in their initial series on ETV for iNPH, promote initial ETV in those iNPH patients who have “short preoperative evolution and significant prevalence of gait disturbances with little or no dementia.” Other studies have considered ETV for iNPH only in those with symptoms of a short duration,5,8 in patients younger than 80 years of age (with presumably more preserved brain parenchyma compliance),5 and in those with preoperative magnetic resonance imaging scans revealing aqueductal CSF stroke volume >42 μL5 or aqueductal stenosis.11 Further prospective study is necessary to find which patient inclusion criterion will maximize the clinical benefit of ETV.

Back to Top | Article Outline

We acknowledge certain limitations of this study. The NIS is a retrospective dataset and has the associated limitations. First, the NIS complication data are limited by ICD-9 coding. As a code intended for billing purposes, this scheme may fall short of providing a full clinical picture of a perioperative patient, such as would be provided with a prospective study. Similarly, although an exhaustive search was carried out of ICD-9 coding to capture the full breadth of perioperative complications in our dataset, shortcomings and inconsistencies in physician coding will limit ability to capture all potential complications from a neurosurgical procedure. For example, we do not cover long-term complications or outcomes (eg, procedure revisions, quality of life) of ETV (which is beyond the scope of this study) or discharge-level complications that are not readily coded such as hormonal complications, delayed sudden death from ETV,23 gaze palsy, memory disorders, altered consciousness, and other nonspecific intraoperative neural tissue injury.32 Likewise, we cannot capture the many long-term complications of shunting (eg, shunt migration, infection, malfunction, under- or overdrainage and the subsequent subdural hematoma, and scalp necrosis), shunt outcomes, or perioperative complications not amenable to ICD-9 coding such as those mentioned. Indeed, we believe that our complication rates represent a robust minimum periprocedural complication rate for ETV for iNPH, a minimum that is higher than that of many previous studies reporting on ETV for iNPH.10,13,21

In addition, there is no specific ICD-9 code for ETV. We used ICD-9 code 02.2, the identifier for “ventriculostomy.” Based on demographic similarities in age, sex, comorbidities, location of procedure, ZIP code income, and form of payment to those receiving the unambiguously coded VPS, we believe that these discharges truly represent iNPH admissions for ETV rather than the alternative, catheter ventriculostomy (external ventricular drain). A key methodological safeguard, using only discharges with the primary diagnosis code of iNPH, ensures that we capture ETV, and not extraventricular drainage, for iNPH. In our experience with iNPH patients from 1998 to the present, extraventricular drainage was associated with admission primary diagnosis codes of subarachnoid hemorrhage, subdural hemorrhage, shock, or infection.

Records in the NIS account for unique patient discharges without definitively capturing unique iNPH patients or procedures. Although most patients with iNPH undergo VPS as surgical treatment, many undergo ETV or both, especially in cases in which clinical improvement is not seen with the initial treatment choice. If these “clinical failure” patients carry worse comorbidities, for example, this could skew the complication rates for ETV patients because 10.0% of our ETV dataset had concomitant coding for VPS during identical hospital admission.

Last, it remains unknown what factors led surgeons to decide on ETV rather than VPS for iNPH, limiting our ability to compare outcomes as would be possible with a prospective study. For example, at our center, patients with a diagnosis of iNPH who have evidence of aqueductal stenosis or poor flow on CSF cine magnetic resonance imaging studies are considered candidates for ETV.

ETV remains a less invasive operative option in well-selected patients. Yet, the significant increase in perioperative mortality and complications found currently must be studied further to ensure that ETV's potential long-term superiority is not offset by its short-term inferiority. The aforementioned shortcomings should be taken in the light that this remains the largest database reporting complications of ETV and VPS for iNPH patients in the United States in a multicenter, nonselected hospital group fashion, beyond the confines of meta-analyses of previously published studies.

Back to Top | Article Outline


This is the first study that robustly assesses the perioperative complications and safety outcomes of ETV specifically for iNPH. Compared with VPS, ETV is associated with increased perioperative mortality and complication rates, which may counterweigh the promise of ETV as a potential method to avoid long-term shunt dependence. This study suggests that medical comorbidity is a crucial predictor of worse short-term safety outcomes and again reinforces the importance of patient selection in ETV and VPS for iNPH. Future, multicenter, prospective, randomized studies are needed to provide definitive comparisons in efficacy, short-term complications, and, especially long-term complication rates.

Back to Top | Article Outline

Drs Chan and Sheehy received funding from the Doris Duke Charitable Foundation. The authors have no personal financial or institutional interest in any of the drugs, materials, or devices described in this article.

Back to Top | Article Outline


1. Adams RD, Fisher CM, Hakim S, Ojemann RG, Sweet WH. Symptomatic occult hydrocephalus with “Normal” cerebrospinal-fluid pressure. A treatable syndrome. N Engl J Med. 1965;273:117–126.

2. Hakim S, Adams RD. The special clinical problem of symptomatic hydrocephalus with normal cerebrospinal fluid pressure. Observations on cerebrospinal fluid hydrodynamics. J Neurol Sci. 1965;2(4):307–327.

3. Bergsneider M, Black PM, Klinge P, Marmarou A, Relkin N. Surgical management of idiopathic normal-pressure hydrocephalus. Neurosurgery. 2005;57(3 suppl):S29–S39; discussion ii-v.

4. Brean A, Fredo HL, Sollid S, Muller T, Sundstrom T, Eide PK. Five-year incidence of surgery for idiopathic normal pressure hydrocephalus in Norway. Acta Neurol Scand. 2009;120(5):314–316.

5. Fountas KN, Kapsalaki EZ, Paterakis KN, Lee GP, Hadjigeorgiou GM. Role of endoscopic third ventriculostomy in treatment of selected patients with normal pressure hydrocephalus. Acta Neurochir Suppl. 2012;113:129–133.

6. Gangemi M, Donati P, Maiuri F, Longatti P, Godano U, Mascari C. Endoscopic third ventriculostomy for hydrocephalus. Minim Invasive Neurosurg. 1999;42(3):128–132.

7. Mitchell P, Mathew B. Third ventriculostomy in normal pressure hydrocephalus. Br J Neurosurg. 1999;13(4):382–385.

8. Gangemi M, Maiuri F, Buonamassa S, Colella G, de Divitiis E. Endoscopic third ventriculostomy in idiopathic normal pressure hydrocephalus. Neurosurgery. 2004;55(1):129–134; discussion 134.

9. Gangemi M, Maiuri F, Naddeo M, et al.. Endoscopic third ventriculostomy in idiopathic normal pressure hydrocephalus: an Italian multicenter study. Neurosurgery. 2008;63(1):62–67; discussion 67-69.

10. Meier U. Shunt operation versus endoscopic ventriculostomy in normal pressure hydrocephalus: diagnostics and outcome [in German]. Zentralbl Neurochir. 2003;64(1):19–23.

11. Meier U, Zeilinger FS, Schonherr B. Endoscopic ventriculostomy versus shunt operation in normal pressure hydrocephalus: diagnostics and indication. Acta Neurochir Suppl. 2000;76:563–566.

12. Paidakakos N, Borgarello S, Naddeo M. Indications for endoscopic third ventriculostomy in normal pressure hydrocephalus. Acta Neurochir Suppl. 2012;113:123–127.

13. Rangel-Castilla L, Barber S, Zhang YJ. The role of endoscopic third ventriculostomy in the treatment of communicating hydrocephalus. World Neurosurg. 2012;77(3-4):555–560.

14. Klinge P, Marmarou A, Bergsneider M, Relkin N, Black PM. Outcome of shunting in idiopathic normal-pressure hydrocephalus and the value of outcome assessment in shunted patients. Neurosurgery. 2005;57(3 suppl):S40–S52; discussion ii-v.

15. McGirt MJ, Woodworth G, Coon AL, Thomas G, Williams MA, Rigamonti D. Diagnosis, treatment, and analysis of long-term outcomes in idiopathic normal-pressure hydrocephalus. Neurosurgery. 2008;62(suppl 2):670–677.

16. Mori K. Management of idiopathic normal-pressure hydrocephalus: a multiinstitutional study conducted in Japan. J Neurosurg. 2001;95(6):970–973.

17. Eide PK, Sorteberg W. Diagnostic intracranial pressure monitoring and surgical management in idiopathic normal pressure hydrocephalus: a 6-year review of 214 patients. Neurosurgery. 2010;66(1):80–91.

18. Poca MA, Solana E, Martinez-Ricarte FR, Romero M, Gandara D, Sahuquillo J. Idiopathic normal pressure hydrocephalus: results of a prospective cohort of 236 shunted patients. Acta Neurochir Suppl. 2012;114:247–253.

19. Cage TA, Auguste KI, Wrensch M, Wu YW, Gupta N. Self-reported functional outcome after surgical intervention in patients with idiopathic normal pressure hydrocephalus. J Clin Neurosci. 2011;18(5):649–654.

20. Pinto FC, Saad F, Oliveira MF, et al.. Role of endoscopic third ventriculostomy and ventriculoperitoneal shunt in idiopathic normal pressure hydrocephalus: preliminary results of a randomized clinical trial. Neurosurgery. 2013;72(5):845–853; discussion 853-844.

21. Hailong F, Guangfu H, Haibin T, et al.. Endoscopic third ventriculostomy in the management of communicating hydrocephalus: a preliminary study. J Neurosurg. 2008;109(5):923–930.

22. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27.

23. Bouras T, Sgouros S. Complications of endoscopic third ventriculostomy: a systematic review. Acta Neurochir Suppl. 2012;113:149–153.

24. Schirmer CM, Heilman CB, Bhardwaj A. Pneumocephalus: case illustrations and review. Neurocrit Care. 2010;13(1):152–158.

25. Chan YP, Yau CY, Lewis RR, Kinirons MT. Acute confusion secondary to pneumocephalus in an elderly patient. Age Ageing. 2000;29(4):365–367.

26. Markham JW. The clinical features of pneumocephalus based upon a survey of 284 cases with report of 11 additional cases. Acta Neurochir (Wien). 1967;16(1):1–78.
27. Counselman FL, Derkum S. An unusual cause of altered mental status. J Emerg Med. 1995;13(6):757–763.

28. Patwardhan RV, Nanda A. Implanted ventricular shunts in the United States: the billion-dollar-a-year cost of hydrocephalus treatment. Neurosurgery. 2005;56(1):139–144; discussion 144-135.

29. Boakye M, Arrigo RT, Kalanithi PS, Chen YR. Impact of age, injury severity score, and medical comorbidities on early complications after fusion and halo-vest immobilization for C2 fractures in older adults: a propensity score matched retrospective cohort study. Spine (Phila Pa 1976). 2012;37(10):854–859.

30. Chitale R, Campbell PG, Yadla S, Whitmore RG, Maltenfort MG, Ratliff JK. International classification of disease clinical modification 9 modeling of a patient comorbidity score predicts incidence of perioperative complications in a nationwide inpatient sample assessment of complications in spine surgery [published online ahead of print]. J Spinal Disord Tech. 2012.

31. Grossman R, Mukherjee D, Chang DC, et al.. Preoperative charlson comorbidity score predicts postoperative outcomes among older intracranial meningioma patients. World Neurosurg. 2011;75(2):279–285.

32. Bouras T, Sgouros S. Complications of endoscopic third ventriculostomy. J Neurosurg Pediatr. 2011;7(6):643–649.

Back to Top | Article Outline

This paper seeks to answer an important question of whether there is a higher short-term morbidity to endoscopic third ventriculostomy (ETV) as opposed to ventriculoperitoneal shunting (VPS) for the management of normal-pressure hydrocephalus. The use of the National Inpatient Sample for this study has an important potential limitation, as the authors point out, which is that the codes for ETV and external ventricular drain (EVD) are the same. So it is not possible to be completely certain whether patients received an ETV or an EVD. A patient who had an EVD for any reason would be expected to have a higher potential morbidity. The authors have, however, addressed this concern and minimized the potential impact.

Perry A. Ball

Lebanon, New Hampshire

Back to Top | Article Outline

1. With respect to the treatment of idiopathic normal pressure hydrocephalus (iNPH), how do outcomes compare between ventriculoperitoneal shunt (VPS) and endoscopic third ventriculostomy (ETV)?

A. VPS is associated with higher mortality rates

B. VPS is associated with increased length of stay

C. VPS is associated with a lower incidence of short-term complications

D. VPS is associated with worse long-term functional outcomes

2. What is the risk of peri-operative mortality associated with endoscopic third ventriculostomy (ETV) for idiopathic normal pressure hydrocephalus (iNPH)?

A. <1%

B. 1-3%

C. 3-5%

D. 5-8%

3. What MRI sequence best evaluates anatomical eligibility for ETV in NPH in patients with triventriculomegaly?

A. Noncontrast T1-weighted axial

B. T2-weighted coronal

C. Sagittal constructive interference in steady state (CISS)

D. Noncontrast T1-weighted midsagittal

E. Flair coronal


Endoscopic third ventriculostomy; Idiopathic normal-pressure hydrocephalus; Nationwide inpatient sample; Perioperative outcomes; Safety; Ventriculoperitoneal shunt

Supplemental Digital Content

Back to Top | Article Outline

Copyright © by the Congress of Neurological Surgeons


Article Tools



Search for Similar Articles
You may search for similar articles that contain these same keywords or you may modify the keyword list to augment your search.