A re-evaluation of the Endoscopic Third Ventriculostomy Success Score: a Hydrocephalus Clinical Research Network study

Leonard H. Verhey Division of Neurosurgery, Department of Clinical Neurosciences, Spectrum Health, Michigan State University, Grand Rapids, Michigan;

Search for other papers by Leonard H. Verhey in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Abhaya V. Kulkarni Division of Neurosurgery, Hospital for Sick Children, University of Toronto, Ontario, Canada;

Search for other papers by Abhaya V. Kulkarni in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Ron W. Reeder Department of Pediatrics, University of Utah, Salt Lake City, Utah;

Search for other papers by Ron W. Reeder in
jns
Google Scholar
PubMed
Close
 PhD
,
Jay Riva-Cambrin Division of Neurosurgery, Alberta Children’s Hospital, University of Calgary, Alberta, Canada;

Search for other papers by Jay Riva-Cambrin in
jns
Google Scholar
PubMed
Close
 MD, MSc
,
Hailey Jensen Department of Pediatrics, University of Utah, Salt Lake City, Utah;

Search for other papers by Hailey Jensen in
jns
Google Scholar
PubMed
Close
 MStat
,
Ian F. Pollack Department of Neurosurgery, UPMC Children’s Hospital of Pittsburgh, University of Pittsburgh, Pennsylvania;

Search for other papers by Ian F. Pollack in
jns
Google Scholar
PubMed
Close
 MD
,
Brandon G. Rocque Department of Neurosurgery, Children’s of Alabama, University of Alabama, Birmingham, Alabama;

Search for other papers by Brandon G. Rocque in
jns
Google Scholar
PubMed
Close
 MD, MS
,
Mandeep S. Tamber Division of Neurosurgery, UBC Department of Surgery, BC Children’s Hospital, Vancouver, British Columbia, Canada;

Search for other papers by Mandeep S. Tamber in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Patrick J. McDonald Section of Neurosurgery, Department of Surgery, University of Manitoba, Winnipeg, Manitoba, Canada;

Search for other papers by Patrick J. McDonald in
jns
Google Scholar
PubMed
Close
 MD, MHSc
,
Mark D. Krieger Department of Neurosurgery, Children’s Hospital Los Angeles, Keck School of Medicine of the University of Southern California, Los Angeles, California;

Search for other papers by Mark D. Krieger in
jns
Google Scholar
PubMed
Close
 MD
,
Jonathan A. Pindrik Division of Pediatric Neurosurgery, Nationwide Children’s Hospital, Department of Neurological Surgery, The Ohio State University College of Medicine, Columbus, Ohio;

Search for other papers by Jonathan A. Pindrik in
jns
Google Scholar
PubMed
Close
 MD
,
Jason S. Hauptman Department of Neurological Surgery, Seattle Children’s Hospital, University of Washington School of Medicine, Seattle, Washington;

Search for other papers by Jason S. Hauptman in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Samuel R. Browd Department of Neurological Surgery, Seattle Children’s Hospital, University of Washington School of Medicine, Seattle, Washington;

Search for other papers by Samuel R. Browd in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
William E. Whitehead Department of Neurosurgery, Texas Children’s Hospital, Baylor College of Medicine, Houston, Texas;

Search for other papers by William E. Whitehead in
jns
Google Scholar
PubMed
Close
 MD
,
Eric M. Jackson Department of Neurosurgery, Johns Hopkins Medicine, Baltimore, Maryland;

Search for other papers by Eric M. Jackson in
jns
Google Scholar
PubMed
Close
 MD
,
John C. Wellons III Department of Neurosurgery, Vanderbilt University Medical Center, Nashville, Tennessee;

Search for other papers by John C. Wellons III in
jns
Google Scholar
PubMed
Close
 MD, MSPH
,
Todd C. Hankinson Department of Neurosurgery, Children’s Hospital Colorado, University of Colorado, Aurora, Colorado;

Search for other papers by Todd C. Hankinson in
jns
Google Scholar
PubMed
Close
 MD
,
Jason Chu Department of Neurosurgery, Children’s Hospital Los Angeles, Keck School of Medicine of the University of Southern California, Los Angeles, California;

Search for other papers by Jason Chu in
jns
Google Scholar
PubMed
Close
 MD, MSc
,
David D. Limbrick Jr. Department of Neurosurgery, St. Louis Children’s Hospital, Washington University School of Medicine in St. Louis, Missouri; and

Search for other papers by David D. Limbrick Jr. in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Jennifer M. Strahle Department of Neurosurgery, St. Louis Children’s Hospital, Washington University School of Medicine in St. Louis, Missouri; and

Search for other papers by Jennifer M. Strahle in
jns
Google Scholar
PubMed
Close
 MD
,
John R. W. Kestle Department of Neurosurgery, University of Utah, Salt Lake City, Utah

Search for other papers by John R. W. Kestle in
jns
Google Scholar
PubMed
Close
 MD
, and
for the Hydrocephalus Clinical Research Network
Search for other papers by for the Hydrocephalus Clinical Research Network in
jns
Google Scholar
PubMed
Close
for the Hydrocephalus Clinical Research Network
Free access

OBJECTIVE

The Hydrocephalus Clinical Research Network (HCRN) conducted a prospective study 1) to determine if a new, better-performing version of the Endoscopic Third Ventriculostomy Success Score (ETVSS) could be developed, 2) to explore the performance characteristics of the original ETVSS in a modern endoscopic third ventriculostomy (ETV) cohort, and 3) to determine if the addition of radiological variables to the ETVSS improved its predictive abilities.

METHODS

From April 2008 to August 2019, children (corrected age ≤ 17.5 years) who underwent a first-time ETV for hydrocephalus were included in a prospective multicenter HCRN study. All children had at least 6 months of clinical follow-up and were followed since the index ETV in the HCRN Core Data Registry. Children who underwent choroid plexus cauterization were excluded. Outcome (ETV success) was defined as the lack of ETV failure within 6 months of the index procedure. Kaplan-Meier curves were constructed to evaluate time-dependent variables. Multivariable binary logistic models were built to evaluate predictors of ETV success. Model performance was evaluated with Hosmer-Lemeshow and Harrell’s C statistics.

RESULTS

Seven hundred sixty-one children underwent a first-time ETV. The rate of 6-month ETV success was 76%. The Hosmer-Lemeshow and Harrell’s C statistics of the logistic model containing more granular age and etiology categorizations did not differ significantly from a model containing the ETVSS categories. In children ≥ 12 months of age with ETVSSs of 50 or 60, the original ETVSS underestimated success, but this analysis was limited by a small sample size. Fronto-occipital horn ratio (p = 0.37), maximum width of the third ventricle (p = 0.39), and downward concavity of the floor of the third ventricle (p = 0.63) did not predict ETV success. A possible association between the degree of prepontine adhesions on preoperative MRI and ETV success was detected, but this did not reach statistical significance.

CONCLUSIONS

This modern, multicenter study of ETV success shows that the original ETVSS continues to demonstrate good predictive ability, which was not substantially improved with a new success score. There might be an association between preoperative prepontine adhesions and ETV success, and this needs to be evaluated in a future large prospective study.

ABBREVIATIONS

DCC = Data Coordinating Center; ETV = endoscopic third ventriculostomy; ETVSS = Endoscopic Third Ventriculostomy Success Score; FOR = fronto-occipital horn ratio; HCRN = Hydrocephalus Clinical Research Network; IVH = intraventricular hemorrhage.

OBJECTIVE

The Hydrocephalus Clinical Research Network (HCRN) conducted a prospective study 1) to determine if a new, better-performing version of the Endoscopic Third Ventriculostomy Success Score (ETVSS) could be developed, 2) to explore the performance characteristics of the original ETVSS in a modern endoscopic third ventriculostomy (ETV) cohort, and 3) to determine if the addition of radiological variables to the ETVSS improved its predictive abilities.

METHODS

From April 2008 to August 2019, children (corrected age ≤ 17.5 years) who underwent a first-time ETV for hydrocephalus were included in a prospective multicenter HCRN study. All children had at least 6 months of clinical follow-up and were followed since the index ETV in the HCRN Core Data Registry. Children who underwent choroid plexus cauterization were excluded. Outcome (ETV success) was defined as the lack of ETV failure within 6 months of the index procedure. Kaplan-Meier curves were constructed to evaluate time-dependent variables. Multivariable binary logistic models were built to evaluate predictors of ETV success. Model performance was evaluated with Hosmer-Lemeshow and Harrell’s C statistics.

RESULTS

Seven hundred sixty-one children underwent a first-time ETV. The rate of 6-month ETV success was 76%. The Hosmer-Lemeshow and Harrell’s C statistics of the logistic model containing more granular age and etiology categorizations did not differ significantly from a model containing the ETVSS categories. In children ≥ 12 months of age with ETVSSs of 50 or 60, the original ETVSS underestimated success, but this analysis was limited by a small sample size. Fronto-occipital horn ratio (p = 0.37), maximum width of the third ventricle (p = 0.39), and downward concavity of the floor of the third ventricle (p = 0.63) did not predict ETV success. A possible association between the degree of prepontine adhesions on preoperative MRI and ETV success was detected, but this did not reach statistical significance.

CONCLUSIONS

This modern, multicenter study of ETV success shows that the original ETVSS continues to demonstrate good predictive ability, which was not substantially improved with a new success score. There might be an association between preoperative prepontine adhesions and ETV success, and this needs to be evaluated in a future large prospective study.

In Brief

The objective of this work was to critically evaluate the Endoscopic Third Ventriculostomy Success Score (ETVSS) in a large, multicenter modern cohort of patients and determine if prediction could be improved. The authors found that neither re-categorization of variables nor the addition of imaging variables substantially improved prediction compared with the original ETVSS. This study confirms that the ETVSS demonstrates good predictive ability in a modern ETV cohort.

The Endoscopic Third Ventriculostomy Success Score (ETVSS) was published in 2009 and provided a tool to predict the percentage chance of endoscopic third ventriculostomy (ETV) success using the preoperative demographic factors of age, etiology of hydrocephalus, and presence of a previous shunt. The ETVSS has been independently externally validated by others.15 The ETVSS was developed from an international cohort of patients, the vast majority of whom had undergone ETV between 1995 and 2006. Anecdotally, there have been questions about the accuracy of the ETVSS in current practice, and previous work has suggested that the ETVSS may underestimate ETV success in the midrange success levels6,7 and overestimate success in patients with an ETVSS ≤ 70.2

The goals of this study were 3-fold: 1) to determine if a new, better-performing version of the ETVSS could be developed de novo from a large, modern ETV cohort; 2) to explore the performance characteristics of the original ETVSS in a modern ETV cohort to identify areas of weakness; and 3) to determine if the addition of radiological variables to the ETVSS improves its predictive abilities. This study was conducted by the Hydrocephalus Clinical Research Network (HCRN), a 14-institution network of pediatric neurosurgery centers within North America that has prospectively collected standardized detailed data on ETV and surgical outcomes since its inception in 2008.

Methods

Study Cohort

The cohort consisted of patients who were 17.5 years of age or younger at the time of ETV. All patients were treated and followed at one of 14 HCRN centers: Children’s Hospital of Alabama, Birmingham, Alabama; Alberta Children’s Hospital, Calgary, Canada; BC Children’s Hospital, Vancouver, Canada; Children’s Hospital of Colorado, Aurora, Colorado; Johns Hopkins Hospital, Baltimore, Maryland; Children’s Hospital of Los Angeles, California; Monroe Carell Jr. Children’s Hospital, Nashville, Tennessee; Nationwide Children’s Hospital, Columbus, Ohio; UPMC Children’s Hospital of Pittsburgh, Pennsylvania; Primary Children’s Hospital, Salt Lake City, Utah; Seattle Children’s Hospital, Seattle, Washington; The Hospital for Sick Children, Toronto, Canada; St. Louis Children’s Hospital, St. Louis, Missouri; and Texas Children’s Hospital, Houston, Texas. Data for all patients who underwent an ETV procedure were prospectively collected using standardized data collection instruments by the treating surgeons and trained research coordinators at each center. For this study, only patients who underwent a first-time ETV were included. Repeat ETV procedures were not included. We excluded children who had a prior CSF shunt, as the number of these patients in the present cohort was extremely low. We made the a priori decision to focus these analyses on the majority of patients, namely, those in whom ETV was considered the first-time CSF diversion intervention. Those who underwent choroid plexus cauterization in addition to ETV were also excluded. Patients have been enrolled and followed prospectively from as early as April 23, 2008, at some centers, with a staggered later start date at other centers. All children with an ETV performed on or before August 31, 2019, were included in the study, and the data set was locked for analysis on February 29, 2020, allowing for at least 6 months of follow-up data from the time of the ETV procedure. Data were stored centrally and securely at the Data Coordinating Center (DCC) in Salt Lake City, Utah. Standardized audits are routinely performed to ensure data accuracy and timeliness. All data were anonymized. Data collection was adherent to each center’s research ethics protocols. Extraction of data for this study from the main data repository was coded by two independent DCC statisticians to ensure accuracy and reliability.

Data Collection

Baseline data collected for this specific study included corrected patient age at the time of ETV procedure, etiology of hydrocephalus (defined as postinfectious, post–intraventricular hemorrhage [IVH] secondary to prematurity, myelomeningocele, aqueductal stenosis, spontaneous hemorrhage [intracerebral, intraventricular, subarachnoid], brain tumor [infratentorial, supratentorial, midbrain], posttraumatic brain injury, encephalocele, posterior fossa cyst [e.g., Dandy-Walker syndrome], other intracranial cyst [e.g., arachnoid cyst, porencephalic cyst], communicating congenital hydrocephalus, other congenital disorder [e.g., schizencephaly, holoprosencephaly], and craniosynostosis), each center’s ETV procedural volume, and ETVSS.8 Using the brain imaging scan obtained immediately prior to the ETV procedure, the following data were obtained: imaging type (defined as ultrasound, CT, and MRI), fronto-occipital horn ratio (FOR),9,10 maximal third ventricle width, third ventricle floor concavity (downward displacement), third ventricle floor thickness, third ventricle morphology index11 (sum of the distance from mammillary body to the lamina terminalis and anterior commissure to the lowest point of the third ventricle floor divided by biparietal diameter), cerebral aqueduct morphology (defined as normal, narrowed, and obstructed), and presence of prepontine adhesions (defined as none, few, many). Figure 1 illustrates these ventricular morphology parameters. The number of ETV procedures performed at each center per year of data collection was computed and divided into tertiles, with high-volume centers defined as those in the upper tertile.

FIG. 1.
FIG. 1.

Measures of ventricular morphology. A: FOR = (A + B)/(2C). B: Maximum width of the third ventricle (arrow). C: Third ventricle floor concavity, scored as either concave/down (black line following the third ventricle floor) or normal. D: Maximum third ventricle floor thickness, measured in this example as a vertical line at the arrow. E: Third Ventricular Morphology Index = (distance from mammillary body to point of maximal displacement of lamina terminalis [A]) + (distance from anterior commissure to point of lowest displacement of the third ventricle floor [B])/(biparietal diameter). Biparietal diameter = line C in panel A.

The outcome—ETV success—was defined as the absence of ETV failure within 6 months of the index procedure. Failure of ETV was defined as: 1) recurrent symptoms of hydrocephalus following ETV, and either 2) the need for repeat surgery (i.e., repeat ETV or CSF shunt placement), as determined by the treating surgeon, within 6 months of the index ETV procedure, or 3) CSF infection, significant intraoperative complication, or mortality related to hydrocephalus management within 6 months of the index ETV. Originally the cohort was randomly divided into a derivation set (80%) and a validation set (20%). However, after we determined that we would not be developing a new scoring system, the two sets were combined to provide maximum statistical power for all analyses.

Statistical Analyses

Analysis 1: Developing a New ETV Success Prediction Model and Comparing it With the Original ETVSS

In this analysis, corrected age and hydrocephalus etiology categories were created using a combination of clinical judgment and the desire to provide as many useful discrete categories as possible, while considering the sample size within each category, resulting in 10 age and 11 etiology categories. For etiology, the "other" category included spontaneous intracerebral, intraventricular, or subarachnoid hemorrhage; posttraumatic brain injury; encephalocele; craniosynostosis; and other congenital etiology (e.g., schizencephaly). We intentionally tried to provide more focused and specific categorizations compared with the original ETVSS, which has 5 age and 7 etiology categories.

The two newly categorized age and etiology variables were first tested in a univariate logistic regression model and then combined into a multivariate model. Binary logistic regression models were built without any forward selection or backward elimination. Successful ETV at 6 months was the dependent variable. Covariates were entered into the model as categorical or continuous variables, as appropriate. We tested the adequacy of the model in two ways. First, goodness of fit was tested with the Hosmer-Lemeshow statistic, which compares the observed event rate to the event rate predicted by the model. A nonsignificant p value indicates retention of the null hypothesis that the model is well-fitted to the data.12 Second, the discriminating power of the model was assessed by computing the area under the receiver operating characteristic curve, otherwise known as Harrell’s C-statistic.13 This statistic represents the probability that the model predicts a higher chance of ETV success in an actual successful case versus a failed case. A value of 1.0 represents perfect discrimination, whereas a value of 0.5 represents no discrimination.14 These performance metrics were then compared with those of a logistic regression model that included age and etiology as categorized in the original ETVSS, to determine if the newly devised, more granular and specific categorization of age and etiology appreciably improved predictive performance.

To explore areas of weakness within the ETVSS, its performance within the different ranges of prediction was assessed graphically by plotting actual percentage success (y-axis) against ETVSS predicted success (x-axis), stratified by age (< 12 and ≥ 12 months); this enabled us to determine whether the performance characteristics differed between infants and older children.

Analysis 2: Determining if Preoperative Radiological Variables Improve Prediction of ETV Success

The HCRN data set contained a limited number of preoperative radiological variables. Of these variables, we identified 3 that were highly endorsed and had the fewest missing values: FOR, maximal third ventricle width, and third ventricle floor concavity (downward displacement). For this analysis, we built multivariate logistic regression models (as in analysis 1) in which these 3 radiological covariates were added to age and etiology. These were tested using both the ETVSS categorization of age and etiology, and the new, more granular categorization of age and etiology from analysis 1. As in analysis 1, the Hosmer-Lemeshow and C statistics were used to compare performance of the two models.

Analysis 3: Determining if the Presence of Prepontine Adhesions on Preoperative Imaging Improved the Performance of the ETVSS

We tested a fourth radiological variable, the presence of prepontine adhesions on preoperative MRI, as determined by the treating surgeon, and categorized it as none, few, or many (Fig. 2). Because of the high degree of missingness of data for this variable, we performed this analysis separately. Because we could not be certain that the missingness was purely random, we created a multivariate model that included age, etiology, and a dummy variable for prepontine adhesions (i.e., prepontine adhesions assessed, categorized as yes or no), allowing us to determine whether ETV success was systematically different in the subgroup in whom prepontine adhesions was assessed. We then built a multivariate model containing age, etiology, and prepontine adhesions (categorized as none, few, or many) to determine if the latter was an important predictor of ETV success, and if the addition of this variable improved the performance of the ETVSS compared with a model without it.

FIG. 2.
FIG. 2.

Prepontine adhesions. A and B: Sagittal FIESTA MR images from 2 patients with few prepontine adhesions. C and D: Sagittal FIESTA MR images from 2 patients with many prepontine adhesions.

Results

The CONSORT diagram (Fig. 3) illustrates the status of enrollment into the HCRN Registry at the time that the data set was locked for analysis, as well as the derivation of the final cohort of this study. Our cohort consisted of 761 patients, and their characteristics are listed in Table 1. Overall ETV success at 6 months was 76% (n = 575). Longer-term success of the cohort is shown in the Kaplan-Meier survival curve (Fig. 4).

FIG. 3.
FIG. 3.

CONSORT diagram.

TABLE 1.

Cohort characteristics

VariableValue
Age, n (%)
 <1 mo37 (4.9)
 1 to <3 mos22 (2.9)
 3 to <6 mos28 (3.7)
 6 to <12 mos52 (6.8)
 12 to <18 mos41 (5.4)
 18 to <24 mos27 (3.5)
 24 to <36 mos35 (4.6)
 3 to <5 yrs61 (8.0)
 5 to <10 yrs166 (21.8)
 ≥10 yrs292 (38.4)
Etiology, n (%)
 Postinfectious18 (2.4)
 Post-IVH (prematurity)21 (2.8)
 Myelomeningocele21 (2.8)
 Aqueductal stenosis148 (19.4)
 Posterior fossa tumor83 (10.9)
 Supratentorial tumor42 (5.5)
 Midbrain tumor/lesion232 (30.5)
 Posterior fossa cyst including Dandy-Walker21 (2.8)
 Other intracranial cyst (including arachnoid)51 (6.7)
 Communicating congenital24 (3.2)
 Other*98 (12.9)
Median FOR (IQR)0.5 (0.4–0.6)
Median 3rd ventricle maximum width (IQR), mm15 (11–20)
3rd ventricle floor shape, n (%)§
 Normal257 (43.3)
 Concave down337 (56.7)
Prepontine adhesions, n (%)
 None146 (70.2)
 Few53 (25.5)
 Many9 (4.3)

IQR = interquartile range.

This category included spontaneous intracerebral, intraventricular, or subarachnoid hemorrhage; posttraumatic brain injury; encephalocele; craniosynostosis; and other congenital etiology (e.g., schizencephaly).

Missing data for 3 patients.

Missing data for 6 patients.

Missing data for 167 patients.

Missing data for 514 patients and unable to be assessed on the scan in 39 patients.

FIG. 4.
FIG. 4.

Kaplan-Meier survival curve showing time to treatment failure for the entire ETV cohort.

Analysis 1: Developing a New ETV Success Prediction Model and Comparing it With the Original ETVSS

Using the strategy described in the Methods, we classified age into 10 categories and etiology into 11 categories (Table 1). We initially generated a derivation data set by randomly selecting 80% of the entire cohort and holding the remaining 20% of the data set as a validation set. Initial analyses in the derivation cohort demonstrated that the new categorizations of age and etiology did not substantially improve model performance compared with models containing the 5 age and 7 etiology categories of the original ETVSS. We therefore decided not to proceed with the creation of a new prediction score and combined the derivation and validation cohorts for the remainder of the analyses described below. Using the entire cohort, we built a multivariate model that included age and etiology. The Hosmer-Lemeshow p value and the C-statistic were similar between the model that used the new, more granular categorizations of age and etiology (Hosmer-Lemeshow p = 0.52, C-statistic = 0.68) and the model using the original categorizations of the ETVSS (Hosmer-Lemeshow p = 0.75, C-statistic = 0.65), suggesting that the new categorizations did not improve prediction of ETV success. Additionally, ETV proficiency, measured by case volume per year of data collection, did not significantly predict ETV success (p = 0.88), and therefore was not carried forward in the logistic models.

To graphically explore the prediction characteristics of the original ETVSS, we plotted actual ETV success against predicted success for each level of the ETVSS from 10 to 90, including only ETVSS levels that contained at least 5 patients. These graphical analyses are shown in Fig. 5. In children ≥ 12 months of age, actual ETV success was higher than that predicted by the ETVSS in the 50–60 range, but these included only 7 patients in each category of ETV success.

FIG. 5.
FIG. 5.

A: Graph of ETVSS versus actual ETV success for infants < 12 months old. This plot shows the relationship between ETVSS (horizontal axis) and actual (i.e., observed) ETV success for 138 patients < 12 months of age at the time of ETV (vertical axis). The dotted line delineates perfect prediction. The number of patients within each level of ETVSS is as follows: 10 (3 patients), 20 (1 patient only, not shown), 30 (6 patients), 40 (38 patients), 50 (41 patients), 60 (9 patients), 70 (40 patients), 80 (0 patients), and 90 (0 patients). B: ETVSS versus actual ETV success for children ≥ 12 months old. This plot shows the relationship between ETVSS (horizontal axis) and actual (i.e., observed) ETV success (vertical axis) for 623 patients ≥ 12 months of age at the time of ETV. The dotted line delineates perfect prediction. The number of patients within each level of ETVSS is as follows: 10 (0 patients), 20 (0 patients), 30 (0 patients), 40 (0 patients), 50 (7 patients), 60 (7 patients), 70 (87 patients), 80 (295 patients), and 90 (227 patients). Figure is available in color online only.

Analysis 2: Determining if Preoperative Radiological Variables Improved Prediction of ETV Success

We tested the performance of a multivariate logistic model that included age (10 categories), etiology (11 categories), and 3 preoperative radiological variables: FOR, maximum width of the third ventricle, and concavity (downward displacement) of the floor of the third ventricle. None of these radiological variables were significant predictors of ETV success in the multivariate analysis (FOR, p = 0.37; maximum width of the third ventricle, p = 0.39; and concavity of the floor of the third ventricle, p = 0.63), nor did these radiological variables improve the performance of the multivariate model (Hosmer-Lemeshow p = 0.23, C-statistic = 0.62) when compared with the model containing only age (10 categories) and etiology (11 categories) from analysis 1.

Analysis 3: Determining if the Presence of Prepontine Adhesions on Preoperative Imaging Improved the Performance of the ETVSS

There were 191 patients in whom prepontine adhesions were assessed on MRI preoperatively. Whether or not prepontine adhesions were scored on preoperative imaging (i.e., the dummy variable described in the Methods) was not associated with a significant difference in 6-month ETV success (p = 0.94 in a multivariate model that included age and etiology).

When the categorial variable prepontine adhesions was added to the original ETVSS variables of age (5 categories) and etiology (7 categories) in a multivariate model, the degree of prepontine adhesions (i.e., none, few, many) was not significantly associated with 6-month ETV success (p = 0.15). However, there was a strong trend in the expected direction, with the lowest ETV success in those with many prepontine adhesions (p = 0.06; Table 2). This association was further confirmed graphically in Kaplan-Meier survival curves, showing substantially lower success for those with many prepontine adhesions, unadjusted for age or etiology (p = 0.03, log-rank test; Fig. 6).

TABLE 2.

Multivariate logistic regression of age, etiology, and prepontine adhesions as predictors of 6-month ETV success

PredictorβOR95% CIp Value
Age0.190
 <1 mo−1.270.280.05–1.500.137
 1 to <6 mos−1.160.310.09–1.090.068
 6 to <12 mos−1.090.340.09–1.300.114
 1 to <10 yrs−0.250.780.35–1.730.540
 ≥10 yrsRefRefRefRef
Etiology0.945
 Aqueductal stenosis0.171.180.37–3.720.778
 Other brain tumor−0.280.760.25–2.300.622
 Post-IVH−0.011.000.15–6.440.996
 Midbrain/tectal tumor0.051.050.43–2.580.915
 Myelomeningocele0.932.520.27–23.930.420
 Postinfectious−0.640.530.07–3.950.534
 OtherRefRefRefRef
Prepontine adhesions*0.154
 Many−1.450.240.05–1.030.055
 Few−0.070.940.42–2.110.872
 NoneRefRefRefRef

Ref = reference.

ETV success rates within each category of prepontine adhesions were as follows: many (3/7, 43%), few (39/51, 77%), and none (106/133, 80%).

FIG. 6.
FIG. 6.

Kaplan-Meier survival curve demonstrating time to treatment failure stratified by prepontine adhesions (none, few, or many) on preoperative MRI.

Discussion

The results of this study represent a modern multicenter prospective evaluation of predictors of ETV success in North American children with hydrocephalus. We showed that age and hydrocephalus etiology remain strong predictors of 6-month ETV success, and that more granular categorizations of these factors might not enhance prediction of success. Additionally, ETV success might be better than what the ETVSS predicts for children over 1 year of age in midrange success levels (i.e., ETVSS of 50 and 60), although this finding was based on a small sample. We also showed that the presence of many prepontine adhesions on preoperative MRI may be independently associated with ETV failure. The unique strength of these findings lies in the standardized, detailed, and prospective collection of data as part of the HCRN.

We reviewed all clinical, demographic, and preoperative radiographic variables collected in the HCRN Registry, and using the existing literature and consensus expertise, identified a priori potential preoperative predictors of ETV success. We did not evaluate intraoperative variables, as this area was studied previously by our group,6 and our objective was to evaluate variables that would inform on ETV success in preoperative patient counseling. Similar to our original paper,8 we considered each center’s ETV procedure volume as a predictor of ETV success. However, this variable did not significantly predict ETV success. This finding is not surprising, given the lack of heterogeneity in this variable due to significant interval experience with ETV since our original publication, and the level of mastery of the technique by surgeons participating in the HCRN Registry. Another predictive variable in the original ETVSS was a prior CSF shunt. This variable was not considered in the current analysis because we only included patients for whom ETV was their first hydrocephalus intervention. This decision was based partly on prior work of the HCRN,15 in which we observed a confounding effect of a previous CSF shunt, as it is strongly associated with increased patient age and typically unfavorable etiologies. This confounding is difficult to mitigate. Therefore, to have as clean an analysis as possible, we restricted the cohort to ETV in only those without a prior shunt. Age at the time of ETV remained the strongest predictor of success, with those < 12 months of age having the lowest predicted success. As with the ETVSS, we observed greater success with increasing age. Regarding hydrocephalus etiology, we found that aqueductal stenosis and posterior fossa pathologies were more likely to be associated with success, and postinfectious, post-IVH of prematurity, and myelomeningocele etiologies were least associated with success. Because a more granularized categorization of age and etiology did not enhance the predictive performance of the multivariate models when compared with a model containing the categorizations of the original ETVSS, there was no statistical or predictive rationale to revise the ETVSS.

Anecdotally, there have been questions about the accuracy of the ETVSS in current practice. Our data suggest that current ETV practice may outperform success predicted by the ETVSS for patients ≥ 12 months of age and in the mid-range success levels (i.e., 50–60) of the ETVSS. This finding, however, was based on only 14 patients, and therefore needs to be interpreted with caution. In addition, the generalizability of this observation might be further limited because these data come exclusively from high-volume pediatric neurosurgery centers with high levels of expertise and experience with the technique. Regardless, even if taken at face value, the degree to which the ETVSS underestimates success in this subset of patients likely would not materially affect preoperative decision-making.

The prediction of ETV success is important not only for appropriate patient selection, but also for counseling families regarding benefits (and risks) of ETV versus CSF shunting. Other groups have evaluated predictors of ETV success. Using an artificial neural network analysis, Azimi and Mohammadi16 showed that etiology, age, use of choroid plexus cauterization, communicating versus noncommunicating hydrocephalus, and history of a CSF shunt were the most important success predictors. Foroughi et al.11 showed that third ventricle morphology on preoperative imaging was predictive of successful ETV and that postoperative improvements in morphology were associated with successful ETV. Greenfield et al.17 devised a pre- and intraoperative assessment tool in which the following factors were associated with ETV success: age; history of neonatal hemorrhage, meningitis, or CSF shunt infection; abnormal third ventricle anatomy; thickened or scarred subarachnoid membranes; and absence of pulsation of the third ventricle ostomy. The HCRN previously evaluated intraoperative predictors of ETV success and found that visualization of a "naked" basilar artery was associated with success.6 Together, the variables of age, and perhaps to a lesser degree etiology, remain key determinants of ETV success. Some have suggested that a higher weight on age may improve the predictive power of the ETVSS, particularly in the moderate-range success categories.3 We did not evaluate intraoperative variables or ETV complications in this study; these were addressed previously by the HCRN,6 and the objective of this study was to re-evaluate the predictive power of preoperative factors that are known to the surgeon and family prior to the surgery. The preoperative measures of third ventricle morphology were not predictive of success in our cohort. We encountered a high degree of missingness for some of the radiological variables, partly because these variables were added later to the data-reporting forms, and partly due to variation in practice patterns for preoperative imaging. These radiological predictors will be evaluated in future iterations of analyses.

We detected a dose-response effect in the expected direction between prepontine adhesions and ETV success—more prepontine adhesions were associated with lower ETV success across the three categories, but this was based on a smaller sample size. Grading prepontine adhesions is subjective, scored as none, few, or many, but it does accurately reflect real-world practice. There is no current quantitative measure of prepontine adhesion burden. Therefore, when surgeons use this parameter to decide whether to pursue ETV, they make a subjective decision out of necessity, and this is what we have emulated in our analyses. Interrater agreement of prepontine adhesions evaluated on preoperative imaging has not been assessed; however, interrater agreement between surgeons on intraoperative endoscopic assessment of prepontine cistern scarring has previously been studied.18 When we considered whether intraoperative visualization of a naked basilar artery and the presence of residual prepontine membranes could be surrogates of prepontine adhesions on preoperative MRI, chi-square analyses showed there was no significant association between these factors (residual membranes, p = 0.18; naked basilar artery, p = 0.56). Other groups have evaluated prepontine adhesions in the setting of ETV; for instance, this is likely what Greenfield et al. referred to as thickened or scarred subarachnoid membranes in their assessment tool.17 In their national retrospective analysis of repeat ETV, the Dutch Pediatric Neurosurgery Study Group also showed that prepontine arachnoid membranes were negatively associated with treatment success.4 The statistical trend we show is promising, but our ability to detect clear statistical significance was impeded by the small subset in which this variable was available. Therefore, this result warrants further study in the next iteration of our analysis as more data become available, where we will also attempt to better validate the preoperative assessment of prepontine adhesions using intraoperative visualization; this will likely be limited by the fact that those with the most severe burden of prepontine adhesions on MRI will probably not be offered ETV.

Limitations of the Study

We recognize important limitations to our study. Perhaps most important is that ETV failure and the decision to pursue subsequent CSF diversion procedures was made solely by the treating surgeon; there was no independent adjudication panel. This fact is particularly relevant because defining ETV failure can be more difficult and less objective than shunt failure, for example, given that ventricle size is often not a reliable indicator of success or failure. Another limitation is our sample size and missing radiographic data. Although the overall cohort size was substantial and had adequate statistical power for our analyses, the assessment of the radiographic variables was limited due to a smaller sample size. Additional prospective radiographic data collection is ongoing by the HCRN and will position us to further evaluate the predictive value of preoperative radiographic features for ETV success. Finally, the subjectivity of the grading of prepontine adhesions, while capitulating real-world practice, is a limitation. Future work within the HCRN will include an assessment of the reliability of the grading of prepontine adhesions on MRI.

Conclusions

This updated and modern account of ETV success in North America showed that the original ETVSS continues to demonstrate strong predictive ability. A more granularized categorization of age and etiology did not enhance the prediction of ETV success beyond that of the original ETVSS. Based on a limited subset analysis, it is possible that the ETVSS underestimates success for those older than 1 year and in the midrange success categories. The importance of prepontine adhesions in terms of ETV success will be evaluated in a future larger prospective study of the HCRN.

Acknowledgments

We thank our colleagues for their past and ongoing support of the HCRN: D. Brockmeyer, M. Walker, R. Bollo, S. Cheshier, J. Blount, J. Johnston, L. Acakpo-Satchivi, W. J. Oakes, P. Dirks, J. Rutka, M. Taylor, D. Curry, G. Aldave, R. Dauser, A. Jea, S. Lam, H. Weiner, T. Luerssen, R. Ellenbogen, J. Ojemann, A. Lee, A. Avellino, S. Greene, E. Tyler-Kabara, T. Abel, T. S. Park, S. McEvoy, M. Smyth, N. Tulipan, A. Singhal, P. Steinbok, D. Cochrane, W. Hader, C. Gallagher, M. Benour, E. Kiehna, J. G. McComb, P. Chiarelli, A. Robison, A. Alexander, M. Handler, B. O’Neill, C. Wilkinson, L. Governale, A. Drapeau, J. Leonard, E. Sribnick, A. Shaikhouni, E. Ahn, A. Cohen, M. Groves, S. Robinson, C. M. Bonfield, and C. Shannon.

In addition, our work would not be possible without the outstanding support of the dedicated personnel at each clinical site and the DCC: L. Holman, J. Clawson, P. Martello, N. Tattersall, and T. Bach (Salt Lake City); T. Caudill, P. Komarova, A. Arynchyna, and A. Bey (Birmingham); H. Ashrafpour, M. Lamberti-Pasculli, and L. O’Connor (Toronto); E. Santisbon, E. Sanchez, S. Martinez, and S. Ryan (Houston); K. Hall, C. Gangan, J. Klein, A. Anderson, and G. Bowen (Seattle); S. Thambireddy, K. Diamond, and A. Luther (Pittsburgh); A. Morgan, H. Botteron, D. Morales, M. Gabir, D. Berger, and D. Mercer (St. Louis); M. Stone, A. Wiseman, J. Stoll, D. Dawson, and S. Gannon (Nashville); A. Cheong and R. Hengel (Vancouver, British Columbia); R. Rashid and S. Ahmed (Calgary, Alberta); J. Yea and A. Loudermilk (Baltimore); N. Chapman, N. Rea, and C. Cook (Los Angeles); S. Staulcup (Aurora); S. Saraswat and A. Sheline (Columbus); and N. Nunn, M. Langley, V. Wall, D. Austin, B. Conley, V. Freimann, L. Herrera, and B. Miller (Salt Lake City DCC).

The HCRN acknowledges the following sources of funding: National Institute of Neurological Disorders and Stroke (NINDS) grant no. 1RC1NS068943-01 Challenge; NINDS grant no. 1U01NS107486-01A1 ESTHI; Patient Centered Outcomes Research Institute (PCORI) grant no. CER-1403-13857; The Gerber Foundation reference no. 1692-3638; private philanthropy; and the Hydrocephalus Association.

Disclosures

Dr. Reeder reported grants from the Hydrocephalus Association during the conduct of the study. Dr. Hauptman reported personal fees from BK Medical and Medtronic outside the submitted work.

Author Contributions

Conception and design: Kulkarni, Verhey, Reeder, Riva-Cambrin, Krieger, Hauptman, Wellons, Hankinson. Acquisition of data: Kulkarni, Riva-Cambrin, Pollack, Rocque, Tamber, McDonald, Krieger, Pindrik, Hauptman, Whitehead, Jackson, Hankinson, Limbrick, Strahle. Analysis and interpretation of data: Kulkarni, Verhey, Reeder, Riva-Cambrin, Jensen, Krieger, Pindrik, Whitehead, Jackson, Wellons, Hankinson, Limbrick, Strahle, Kestle. Drafting the article: Kulkarni, Verhey, Krieger. Critically revising the article: Kulkarni, Verhey, Reeder, Riva-Cambrin, Jensen, Pollack, Rocque, Tamber, McDonald, Krieger, Pindrik, Hauptman, Whitehead, Jackson, Wellons, Hankinson, Chu, Limbrick, Strahle, Kestle. Reviewed submitted version of manuscript: Kulkarni, Verhey, Reeder, Riva-Cambrin, Jensen, Pollack, Rocque, Tamber, McDonald, Krieger, Pindrik, Hauptman, Browd, Whitehead, Jackson, Wellons, Strahle, Kestle. Approved the final version of the manuscript on behalf of all authors: Kulkarni. Statistical analysis: Kulkarni, Verhey, Jensen. Administrative/technical/material support: McDonald. Study supervision: Kulkarni.

Supplemental Information

Previous Presentations

Presented in part as an oral paper at the AANS/CNS Pediatric Section Meeting in Washington, DC, December 1–4, 2022.

References

  • 1

    Durnford AJ, Kirkham FJ, Mathad N, Sparrow OC. Endoscopic third ventriculostomy in the treatment of childhood hydrocephalus: validation of a success score that predicts long-term outcome. J Neurosurg Pediatr. 2011;8(5):489493.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Naftel RP, Reed GT, Kulkarni AV, Wellons JC. Evaluating the Children’s Hospital of Alabama endoscopic third ventriculostomy experience using the Endoscopic Third Ventriculostomy Success Score: an external validation study. J Neurosurg Pediatr. 2011;8(5):494501.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Breimer GE, Sival DA, Brusse-Keizer MG, Hoving EW. An external validation of the ETVSS for both short-term and long-term predictive adequacy in 104 pediatric patients. Childs Nerv Syst. 2013;29(8):13051311.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Breimer GE, Dammers R, Woerdeman PA, et al. Endoscopic third ventriculostomy and repeat endoscopic third ventriculostomy in pediatric patients: the Dutch experience. J Neurosurg Pediatr. 2017;20(4):314323.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Furtado LMF, da Costa Val Filho JA, Dos Santos Júnior EC. External validation of the ETV success score in 313 pediatric patients: a Brazilian single-center study. Neurosurg Rev. 2021;44(5):27272734.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Kulkarni AV, Riva-Cambrin J, Holubkov R, et al. Endoscopic third ventriculostomy in children: prospective, multicenter results from the Hydrocephalus Clinical Research Network. J Neurosurg Pediatr. 2016;18(4):423429.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Kulkarni AV, Riva-Cambrin J, Browd SR. Use of the ETV Success Score to explain the variation in reported endoscopic third ventriculostomy success rates among published case series of childhood hydrocephalus. J Neurosurg Pediatr. 2011;7(2):143146.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Kulkarni AV, Drake JM, Mallucci CL, Sgouros S, Roth J, Constantini S. Endoscopic third ventriculostomy in the treatment of childhood hydrocephalus. J Pediatr. 2009;155(2):254259.e1.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Kulkarni AV, Drake JM, Armstrong DC, Dirks PB. Measurement of ventricular size: reliability of the frontal and occipital horn ratio compared to subjective assessment. Pediatr Neurosurg. 1999;31(2):6570.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    O’Hayon BB, Drake JM, Ossip MG, Tuli S, Clarke M. Frontal and occipital horn ratio: a linear estimate of ventricular size for multiple imaging modalities in pediatric hydrocephalus. Pediatr Neurosurg. 1998;29(5):245249.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Foroughi M, Wong A, Steinbok P, Singhal A, Sargent MA, Cochrane DD. Third ventricular shape: a predictor of endoscopic third ventriculostomy success in pediatric patients. J Neurosurg Pediatr. 2011;7(4):389396.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Harrison DA, Brady AR, Parry GJ, Carpenter JR, Rowan K. Recalibration of risk prediction models in a large multicenter cohort of admissions to adult, general critical care units in the United Kingdom. Crit Care Med. 2006;34(5):13781388.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Steyerberg EW, Harrell FE Jr, Borsboom GJ, Eijkemans MJ, Vergouwe Y, Habbema JD. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. J Clin Epidemiol. 2001;54(8):774781.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Miller ME, Hui SL, Tierney WM. Validation techniques for logistic regression models. Stat Med. 1991;10(8):12131226.

  • 15

    Rocque BG, Jensen H, Reeder RW, et al. Endoscopic third ventriculostomy in previously shunt-treated patients. J Neurosurg Pediatr. 2022;30(4):428436.

  • 16

    Azimi P, Mohammadi HR. Predicting endoscopic third ventriculostomy success in childhood hydrocephalus: an artificial neural network analysis. J Neurosurg Pediatr. 2014;13(4):426432.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Greenfield JP, Hoffman C, Kuo E, Christos PJ, Souweidane MM. Intraoperative assessment of endoscopic third ventriculostomy success. J Neurosurg Pediatr. 2008;2(5):298303.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    He L, Gannon S, Shannon CN, Rocque BG, Riva-Cambrin J, Naftel RP. Surgeon interrater reliability in the endoscopic assessment of cistern scarring and aqueduct patency. J Neurosurg Pediatr. 2016;18(3):320324.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand
  • FIG. 1.

    Measures of ventricular morphology. A: FOR = (A + B)/(2C). B: Maximum width of the third ventricle (arrow). C: Third ventricle floor concavity, scored as either concave/down (black line following the third ventricle floor) or normal. D: Maximum third ventricle floor thickness, measured in this example as a vertical line at the arrow. E: Third Ventricular Morphology Index = (distance from mammillary body to point of maximal displacement of lamina terminalis [A]) + (distance from anterior commissure to point of lowest displacement of the third ventricle floor [B])/(biparietal diameter). Biparietal diameter = line C in panel A.

  • FIG. 2.

    Prepontine adhesions. A and B: Sagittal FIESTA MR images from 2 patients with few prepontine adhesions. C and D: Sagittal FIESTA MR images from 2 patients with many prepontine adhesions.

  • FIG. 3.

    CONSORT diagram.

  • FIG. 4.

    Kaplan-Meier survival curve showing time to treatment failure for the entire ETV cohort.

  • FIG. 5.

    A: Graph of ETVSS versus actual ETV success for infants < 12 months old. This plot shows the relationship between ETVSS (horizontal axis) and actual (i.e., observed) ETV success for 138 patients < 12 months of age at the time of ETV (vertical axis). The dotted line delineates perfect prediction. The number of patients within each level of ETVSS is as follows: 10 (3 patients), 20 (1 patient only, not shown), 30 (6 patients), 40 (38 patients), 50 (41 patients), 60 (9 patients), 70 (40 patients), 80 (0 patients), and 90 (0 patients). B: ETVSS versus actual ETV success for children ≥ 12 months old. This plot shows the relationship between ETVSS (horizontal axis) and actual (i.e., observed) ETV success (vertical axis) for 623 patients ≥ 12 months of age at the time of ETV. The dotted line delineates perfect prediction. The number of patients within each level of ETVSS is as follows: 10 (0 patients), 20 (0 patients), 30 (0 patients), 40 (0 patients), 50 (7 patients), 60 (7 patients), 70 (87 patients), 80 (295 patients), and 90 (227 patients). Figure is available in color online only.

  • FIG. 6.

    Kaplan-Meier survival curve demonstrating time to treatment failure stratified by prepontine adhesions (none, few, or many) on preoperative MRI.

  • 1

    Durnford AJ, Kirkham FJ, Mathad N, Sparrow OC. Endoscopic third ventriculostomy in the treatment of childhood hydrocephalus: validation of a success score that predicts long-term outcome. J Neurosurg Pediatr. 2011;8(5):489493.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Naftel RP, Reed GT, Kulkarni AV, Wellons JC. Evaluating the Children’s Hospital of Alabama endoscopic third ventriculostomy experience using the Endoscopic Third Ventriculostomy Success Score: an external validation study. J Neurosurg Pediatr. 2011;8(5):494501.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Breimer GE, Sival DA, Brusse-Keizer MG, Hoving EW. An external validation of the ETVSS for both short-term and long-term predictive adequacy in 104 pediatric patients. Childs Nerv Syst. 2013;29(8):13051311.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Breimer GE, Dammers R, Woerdeman PA, et al. Endoscopic third ventriculostomy and repeat endoscopic third ventriculostomy in pediatric patients: the Dutch experience. J Neurosurg Pediatr. 2017;20(4):314323.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Furtado LMF, da Costa Val Filho JA, Dos Santos Júnior EC. External validation of the ETV success score in 313 pediatric patients: a Brazilian single-center study. Neurosurg Rev. 2021;44(5):27272734.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Kulkarni AV, Riva-Cambrin J, Holubkov R, et al. Endoscopic third ventriculostomy in children: prospective, multicenter results from the Hydrocephalus Clinical Research Network. J Neurosurg Pediatr. 2016;18(4):423429.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Kulkarni AV, Riva-Cambrin J, Browd SR. Use of the ETV Success Score to explain the variation in reported endoscopic third ventriculostomy success rates among published case series of childhood hydrocephalus. J Neurosurg Pediatr. 2011;7(2):143146.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Kulkarni AV, Drake JM, Mallucci CL, Sgouros S, Roth J, Constantini S. Endoscopic third ventriculostomy in the treatment of childhood hydrocephalus. J Pediatr. 2009;155(2):254259.e1.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Kulkarni AV, Drake JM, Armstrong DC, Dirks PB. Measurement of ventricular size: reliability of the frontal and occipital horn ratio compared to subjective assessment. Pediatr Neurosurg. 1999;31(2):6570.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    O’Hayon BB, Drake JM, Ossip MG, Tuli S, Clarke M. Frontal and occipital horn ratio: a linear estimate of ventricular size for multiple imaging modalities in pediatric hydrocephalus. Pediatr Neurosurg. 1998;29(5):245249.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Foroughi M, Wong A, Steinbok P, Singhal A, Sargent MA, Cochrane DD. Third ventricular shape: a predictor of endoscopic third ventriculostomy success in pediatric patients. J Neurosurg Pediatr. 2011;7(4):389396.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Harrison DA, Brady AR, Parry GJ, Carpenter JR, Rowan K. Recalibration of risk prediction models in a large multicenter cohort of admissions to adult, general critical care units in the United Kingdom. Crit Care Med. 2006;34(5):13781388.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Steyerberg EW, Harrell FE Jr, Borsboom GJ, Eijkemans MJ, Vergouwe Y, Habbema JD. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. J Clin Epidemiol. 2001;54(8):774781.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Miller ME, Hui SL, Tierney WM. Validation techniques for logistic regression models. Stat Med. 1991;10(8):12131226.

  • 15

    Rocque BG, Jensen H, Reeder RW, et al. Endoscopic third ventriculostomy in previously shunt-treated patients. J Neurosurg Pediatr. 2022;30(4):428436.

  • 16

    Azimi P, Mohammadi HR. Predicting endoscopic third ventriculostomy success in childhood hydrocephalus: an artificial neural network analysis. J Neurosurg Pediatr. 2014;13(4):426432.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Greenfield JP, Hoffman C, Kuo E, Christos PJ, Souweidane MM. Intraoperative assessment of endoscopic third ventriculostomy success. J Neurosurg Pediatr. 2008;2(5):298303.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    He L, Gannon S, Shannon CN, Rocque BG, Riva-Cambrin J, Naftel RP. Surgeon interrater reliability in the endoscopic assessment of cistern scarring and aqueduct patency. J Neurosurg Pediatr. 2016;18(3):320324.

    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 1053 1053 1053
PDF Downloads 1182 1182 1182
EPUB Downloads 0 0 0