Shunt failure clusters: an analysis of multiple, frequent shunt failures

Brandon G. RocqueDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Brandon G. Rocque in
jns
Google Scholar
PubMed
Close
 MD, MS
,
Raymond P. WaldropDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Raymond P. Waldrop in
jns
Google Scholar
PubMed
Close
 BS
,
Isaac ShamblinDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Isaac Shamblin in
jns
Google Scholar
PubMed
Close
 MSBME
,
Anastasia A. ArynchynaDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Anastasia A. Arynchyna in
jns
Google Scholar
PubMed
Close
 MPH
,
Betsy HopsonDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Betsy Hopson in
jns
Google Scholar
PubMed
Close
 MHA
,
Tammie KerrDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Tammie Kerr in
jns
Google Scholar
PubMed
Close
 MSN
,
James M. JohnstonDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by James M. Johnston in
jns
Google Scholar
PubMed
Close
 MD
,
Curtis J. RozzelleDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Curtis J. Rozzelle in
jns
Google Scholar
PubMed
Close
 MD
, and
Jeffrey P. BlountDivision of Pediatric Neurosurgery, Department of Neurosurgery, University of Alabama at Birmingham, Alabama

Search for other papers by Jeffrey P. Blount in
jns
Google Scholar
PubMed
Close
 MD
Free access

OBJECTIVE

Repeated failure of ventriculoperitoneal shunts (VPSs) is a problem familiar to pediatric neurosurgeons and patients. While there have been many studies to determine what factors are associated with the first shunt failure, studies of subsequent failures are much less common. The purpose of this study was to identify the prevalence and associated risk factors of clustered shunt failures (defined as 3 or more VPS operations within 3 months).

METHODS

The authors reviewed prospectively collected records from all patients who underwent VPS surgery from 2008 to 2017 at their institution and included only those children who had received all of their hydrocephalus care at that institution. Demographics, etiology of hydrocephalus, history of endoscopic third ventriculostomy or temporizing procedure, initial valve type, age at shunt placement, and other factors were analyzed. Logistic regression was used to test for the association of each variable with a history of shunt failure cluster.

RESULTS

Of the 465 included children, 28 (6.0%) had experienced at least one cluster of shunt failures. Among time-independent variables, etiology of hydrocephalus (OR 0.27 for non–intraventricular hemorrhage [IVH], nonmyelomeningocele, nonaqueductal stenosis etiology vs IVH, 95% CI 0.11–0.65; p = 0.003), younger gestational age at birth (OR 0.91, 95% CI 0.85–0.97; p = 0.003), history of a temporizing procedure (OR 2.77, 95% CI 1.12–6.85; p = 0.028), and smaller head circumference at time of initial shunt placement (OR 0.91, 95% CI 0.84–0.99; p = 0.044) showed significant association with shunt failure cluster on univariate analysis. None of these variables maintained significance in a multivariate model. Among children with a history of a shunt failure cluster, 21 (75%) had a shunt infection either prior to or during the shunt failure cluster. A comparison of the infecting organism between these children and 62 children with a history of infection but without a shunt failure cluster showed an association of cluster with gram-negative rod species.

CONCLUSIONS

Six percent of children in this institutional sample had at least one shunt failure cluster. These children accounted for 30% of the total shunt revisions in the sample. Shunt infection is an important factor associated with shunt failure cluster. Children with a history of prematurity and IVH may have a higher risk for failure cluster.

ABBREVIATIONS

GNR = gram-negative rod; IVH = intraventricular hemorrhage; VPS = ventriculoperitoneal shunt.

OBJECTIVE

Repeated failure of ventriculoperitoneal shunts (VPSs) is a problem familiar to pediatric neurosurgeons and patients. While there have been many studies to determine what factors are associated with the first shunt failure, studies of subsequent failures are much less common. The purpose of this study was to identify the prevalence and associated risk factors of clustered shunt failures (defined as 3 or more VPS operations within 3 months).

METHODS

The authors reviewed prospectively collected records from all patients who underwent VPS surgery from 2008 to 2017 at their institution and included only those children who had received all of their hydrocephalus care at that institution. Demographics, etiology of hydrocephalus, history of endoscopic third ventriculostomy or temporizing procedure, initial valve type, age at shunt placement, and other factors were analyzed. Logistic regression was used to test for the association of each variable with a history of shunt failure cluster.

RESULTS

Of the 465 included children, 28 (6.0%) had experienced at least one cluster of shunt failures. Among time-independent variables, etiology of hydrocephalus (OR 0.27 for non–intraventricular hemorrhage [IVH], nonmyelomeningocele, nonaqueductal stenosis etiology vs IVH, 95% CI 0.11–0.65; p = 0.003), younger gestational age at birth (OR 0.91, 95% CI 0.85–0.97; p = 0.003), history of a temporizing procedure (OR 2.77, 95% CI 1.12–6.85; p = 0.028), and smaller head circumference at time of initial shunt placement (OR 0.91, 95% CI 0.84–0.99; p = 0.044) showed significant association with shunt failure cluster on univariate analysis. None of these variables maintained significance in a multivariate model. Among children with a history of a shunt failure cluster, 21 (75%) had a shunt infection either prior to or during the shunt failure cluster. A comparison of the infecting organism between these children and 62 children with a history of infection but without a shunt failure cluster showed an association of cluster with gram-negative rod species.

CONCLUSIONS

Six percent of children in this institutional sample had at least one shunt failure cluster. These children accounted for 30% of the total shunt revisions in the sample. Shunt infection is an important factor associated with shunt failure cluster. Children with a history of prematurity and IVH may have a higher risk for failure cluster.

In Brief

In this study, the authors defined a new variable, shunt failure cluster, to perform the first analysis of multiple and frequent shunt failures. They found that 6% of children with ventriculoperitoneal shunts have a cluster of shunt failures, and that these patients account for 30% of all shunt revisions. Risk factors for shunt failure cluster include prematurity and history of shunt infection.

Ventriculoperitoneal shunts (VPSs) are the mainstay of treatment for hydrocephalus, and they have been very successful at improving the natural history of childhood hydrocephalus. However, VPSs have a relatively high complication rate, including infection and shunt obstruction or failure. There have been many studies of the survival of first-time shunts that have identified risk factors for failure of the initial VPS, such as comorbid medical conditions, use of endoscopy for ventricular catheter placement, prematurity, and low birth weight.1,2 Widely quoted data from a large prospective trial put the failure rate for shunts placed in infancy at approximately 40% in the 1st year and 50% in the first 2 years.3 However, a single shunt failure is often not a major burden. It is the children who have multiple and frequent shunt failures who have the highest burden of suffering and require disproportionate attention from pediatric neurosurgeons.

Risk factors for multiple shunt failures have been identified in previous single-institution studies. They include history of traumatic brain injury, slit-like ventricles, high CSF protein, and younger age at initial placement.4–6 In addition, these studies have shown that after a shunt failure, subsequent failures become less likely as a function of increasing time from surgery. However, none of these studies specifically consider the patient who has multiple shunt failures in a short time period.

The purpose of this study was to better understand the prevalence of multiple, rapid VPS failures. To do this, we have defined a new variable: the shunt failure cluster. We examined children who have had multiple and frequent shunt failures in an attempt to both define the prevalence of this phenomenon and identify risk factors.

Methods

Data Collection

A review of prospectively collected records from all patients who underwent VPS placements from 2008 to 2017 at Children’s of Alabama was performed to identify study participants. This study was approved by the institutional review board of the University of Alabama at Birmingham. Children were eligible for inclusion if they had received all of their hydrocephalus care at our institution. Prospectively collected data included age at the time of shunt placement, sex, race, gestational age at birth, etiology of hydrocephalus, history of temporizing procedure (such as a ventriculosubgaleal shunt or ventricular reservoir), history of endoscopic third ventriculostomy prior to shunt placement, type of shunt valve, and entry site for the proximal catheter (anterior or posterior). Also included in the prospectively collected data set were the dates of every shunt operation, allowing determination of the time between initial shunt placement and each failure. Note was made of whether any shunt operation was a shunt removal for treatment of infection.

We then performed a retrospective medical record review to determine the head circumference (reported as the Z-score from WHO head circumference growth charts) and ventricle size (reported as the frontal–occipital horn ratio).7,8 We recorded head circumference and ventricle size both immediately prior to initial shunt placement and at 1 year after shunt placement, when available.

Variable Definition

As noted above, to analyze children with multiple and frequent shunt failures, we defined a new variable: shunt failure cluster. We defined a cluster of shunt failures as 3 or more shunt revision operations (not counting the initial shunt placement or shunt removals) within a 3-month period. A cluster was defined as 3 operations, so that cases of a shunt revision that are complicated by infection, followed by replacement, and no other operations are not counted as clusters. We also performed a sensitivity analysis, changing the definition of a shunt failure cluster to 4 failures within 3 months. The choice of 3 months as the time threshold is derived from standard definitions of the 90-day postoperative period commonly used by payors and for definitions of surgical complications.

Patients experiencing a shunt infection are typically treated with shunt removal and external ventricular drain placement, followed by shunt replacement after antibiotic treatment. To ensure that shunt infection alone would not lead to a shunt revision cluster, we did not count shunt removal as a distinct operation. Therefore, a patient who had a shunt revision (first operation), followed by an infection and shunt removal (not counted), and then shunt replacement (second operation), would not meet the definition of a shunt failure cluster. A third operation within 3 months would be required for this episode to be considered a cluster.

Data Analysis

Logistic regression analysis was performed to determine an association of each independent variable (except for shunt infection) with the presence of a shunt failure cluster. Because all of these variables are present at the time of shunt placement, none is considered time-dependent. However, the effect of a shunt infection on the primary outcome, shunt failure cluster, may depend on the time of the infection, making modeling of this variable substantially more challenging. Small numbers of patients with shunt failure clusters also limit the power to perform complex models of infection. Therefore, we have analyzed shunt infections using descriptive statistics and cross tabulations. We identify the following 3 possible relationships between shunt infection and shunt failure cluster: 1) no infection prior to or during a shunt failure cluster; 2) infection occurring before cluster or complicating at least one of the surgeries that defines a cluster; and 3) infection occurring in a patient without a history of shunt failure cluster. As a post hoc analysis, we determined the organism responsible for each infection, and compared groups 2 and 3 above using a chi-square statistic.

As noted above, most data were prospectively collected, with the exception of head circumference and ventricle size. For these 2 variables, there were some missing data. We performed a sensitivity analysis for missing data, repeating the univariate logistic regression analysis with all missing data replaced by the median, the first quartile, and the third quartile. This was also necessary for gestational age, as some patients who underwent shunt treatment later in life did not have gestational age data available.

Results

A total of 465 patients met inclusion criteria: 254 males and 211 females. The racial and ethnic breakdown was as follows: White, non-Hispanic 53.5%; Black 35.7%; Hispanic 5.6%; and unknown or other race 5.2%. The median gestational age at birth was 36 weeks (IQR 27–39 weeks). The most common etiologies of hydrocephalus were prematurity-related intraventricular hemorrhage (IVH) and myelomeningocele. A detailed summary of demographic and clinical variables is displayed in Table 1. The mean (± SD) total clinical follow-up was 5.2 ± 3.4 years.

TABLE 1.

Summary of included variables and association with shunt failure cluster

VariableValueShunt Failure ClusterShunt Failure Cluster Univariate Analysis, OR (95% CI)p Value
Age at initial shunt, mos
 Mean (SD)20.66 (43.88)0.98 (0.95–1.01)0.12
 Median (IQR)4 (1–11.5)
Sex, n (%)
 Male254 (54.6)17 (60.7)1.30 (0.60–2.85)0.51
 Female211 (45.4)11 (39.3)Ref
Race, n (%)
 White, non-Hispanic249 (53.5)10 (35.7)Ref
 Black166 (35.7)15 (53.6)2.34 (1.04–5.42)0.04
 Hispanic26 (5.6)2 (7.1)1.99 (0.41–9.62)0.39
 Other/unknown24 (5.2)1 (3.6)1.04 (0.13–8.48)0.97
Gestational age at birth, wks*0.91 (0.85–0.97)0.003
 Mean (SD)33.5 (6)
 Median (IQR)36.0 (27–39)
Etiology of hydrocephalus, n (%)0.008
 Prematurity IVH140 (30.1)17 (60.7)Ref
 Myelomeningocele60 (12.9)2 (7.1)0.25 (0.56–1.12)0.069
 Aqueductal stenosis43 (9.2)1 (3.6)0.17 (0.02–1.33)0.092
 Other222 (47.7)8 (28.6)0.27 (0.11–0.65)0.003
History of ETV before 1st shunt, n (%)
 Yes75 (16.1)3 (10.7)Ref
 No390 (83.9)25 (89.3)1.64 (0.48–5.59)0.43
Initial shunt valve type, n (%)
 Delta (1.0, 1.5, or 2.0)369 (79.4)23 (82.1)Ref
 Orbis Sigma78 (16.8)4 (14.3)0.81 (0.27–2.42)0.71
 Other18 (3.9)1 (3.6)0.89 (0.11–6.95)0.91
History of temporizing procedure, n (%)§
 Yes54 (11.6)7 (25.0)2.77 (1.12–6.85)0.028
 No411 (88.4)21 (75.0)Ref
Entry site of initial shunt, n (%)
 Anterior245 (52.7)13 (46.4)1.31 (0.61–2.81)0.50
 Posterior220 (47.3)15 (53.6)Ref
Mean (SD) Z-score for head circumference at initial shunt placement−0.11 (4.19)0.91 (0.84–0.99)0.044
Mean (SD) FOHR at time of initial shunt**0.56 (0.21)1.64 (0.46–5.83)0.44

ETV = endoscopic third ventriculostomy; FOHR = frontal–occipital horn ratio.

Boldface type indicates statistical significance.

Data were missing for 61 patients.

Other etiologies include congenital communicating (n = 37), craniosynostosis (n = 2), encephalocele (n = 15), congenital abnormality (e.g., schizencephaly, holoprosencephaly; n = 14), unknown (n = 23), intracranial cyst (n = 13), posterior fossa cyst (n = 16), post head injury (n = 26), postinfectious (n = 19), intracranial hemorrhage (not prematurity-related) (n = 20), supratentorial tumor (n = 19), and posterior fossa tumor (n = 12).

Other valves include Strata programmable (n = 5), Codman programmable (n = 6), distal slit (n = 4), and Certas (n = 1).

Ventriculosubgaleal shunt or reservoir.

Data were missing for 110 patients.

Data were missing for 26 patients.

A total of 671 VPS revisions were recorded for the full sample of 465 children (median 0, IQR 0–2). The median number of shunt revisions per patient was 1.0 (IQR 0–2). Twenty-eight patients (6.0%) met the primary outcome criteria of having at least one shunt failure cluster (3 or more shunt surgeries within 3 months). The median age at the time of the first revision of the first shunt failure cluster was 246 days (IQR 57.5–828 days). The median number of shunt revisions prior to the first revision of a shunt failure cluster was 0 (16 patients had 0 revisions, 4 had 1 revision, 5 had 2 revisions, 2 had 3 revisions, and 1 had 9 revisions). A total of 204 shunt failures occurred in the 28 children with a history of at least one shunt failure cluster. This represents 30% of all shunt failures (204 of 671). These children had an average of 7.29 shunt revisions during the study period, compared with 1.07 revisions per patient among children without a history of shunt failure cluster (p < 0.001 by independent-samples t-test).

On univariate logistic regression analysis, etiology of hydrocephalus (OR 0.27 for other etiology vs IVH, 95% CI 0.11–0.65; p = 0.003), younger gestational age at birth (OR 0.91, 95% CI 0.85–0.97; p = 0.003), history of a temporizing procedure (OR 2.77, 95% CI 1.12–6.85; p = 0.028), and smaller head circumference at the time of initial shunt placement (OR 0.91, 95% CI 0.84–0.99; p = 0.044) were associated with a shunt failure cluster. Full results of univariate analysis can be found in Table 1. Multivariate analysis including gestational age, head circumference, and etiology of hydrocephalus showed no variables that were independently significant in their association with shunt failure cluster (Table 2). Because these variables (gestational age, head circumference, and temporizing procedure) are all potentially associated with prematurity and IVH, testing for collinearity was performed to ensure that assumptions of multivariable logistic regression were met. None of these variables showed significant collinearity.

TABLE 2.

Shunt failure cluster multivariate analysis

VariableOR (95% CI)p Value
Gestational age at birth0.95 (0.83–1.08)0.42
Head circumference at birth0.99 (0.87–1.12)0.86
Etiology of hydrocephalus
 Prematurity IVHRefRef
 Myelomeningocele0.29 (0.03–3.39)0.33
 Aqueductal stenosis0.47 (0.04–5.66)0.55
 Other0.86 (0.19–3.87)0.86

Collinearity testing shows no significant collinearity of these variables.

Eighty-three children had a VPS infection at any time during their history (83 of 465, 17.8%). Among the 28 children with a shunt failure cluster, 21 (75%) had a history of shunt infection before or during their shunt failure cluster: 15 children had an infection before but remote from their cluster, and 15 had an infection during their cluster (9 had an infection both before and during). Seven children (25%) had a shunt failure cluster with no history of infection. Among children with a shunt failure cluster, we tested for association between shunt infection and other clinical and demographic variables and observed no significant associations (Table 3).

TABLE 3.

Shunt infection and shunt failure cluster: comparison of patients with versus patients without associated infection

VariableShunt Infection Before or During Shunt Failure Cluster (n = 21)No Infection Associated w/ Cluster (n = 7)Statistical Analysis
Mean (SD) age at initial shunt, mos7.0 (11.5)1.9 (1.6)p = 0.25*
Sex, n (%)χ2 = 0.45, p = 0.50
 Male12 (57)5 (71)
 Female9 (43)2 (29)
Race, n (%)χ2 = 5.96, p = 0.11
 White, non-Hispanic6 (29)4 (57)
 Black13 (62)2 (29)
 Hispanic2 (9.5)0 (0)
 Other/unknown0 (0)1 (14)
Mean (SD) gestational age at birth, wks29.4 (7.0)31.3 (6.8)p = 0.55*
Etiology of hydrocephalus, n (%)χ2 = 2.16, p = 0.54
 Prematurity IVH14 (67)3 (43)
 Myelomeningocele1 (5)1 (14)
 Aqueductal stenosis1 (5)0 (0)
 Other5 (24)3 (43)
History of ETV before 1st shunt, n (%)χ2 = 1.12, p = 0.29
 Yes18 (86)7 (100)
 No3 (14)0 (0)
Initial shunt valve type, n (%)χ2 = 2.03, p = 0.36
 Delta (1.0, 1.5, or 2.0)16 (76)7 (100)
 Orbis Sigma4 (19)0
 Other1 (5)0
History of temporizing procedure, n (%)χ2 = 0.063, p = 0.80
 Yes16 (76)5 (71)
 No5 (24)2 (29)
Entry site of initial shunt, n (%)χ2 = 0.048, p = 0.83
 Anterior10 (48)3 (43)
 Posterior11 (52)4 (57)
Mean (SD) Z-score for head circumference at initial shunt placementMean −1.8 (5.2)−1.6 (3.6)p = 0.93*
Mean (SD) FOHR at time of initial shunt0.60 (0.17)0.56 (0.076)p = 0.57*

Independent-samples t-test.

Ventriculosubgaleal shunt or reservoir.

We performed post hoc analysis of the causative organism for all shunt infections. Infections caused by Staphyloccoccus aureus were most common (27 of 83 cases, 33%). Other common causative organisms were coagulase-negative staphylococci and gram-negative rod (GNR) species. Some patients did not have positive cultures associated with their shunt infection. These patients were treated as presumptively infected due to CSF leak, exposed hardware, or other criteria as specified by the Hydrocephalus Clinical Research Network infection prevention protocol.9 Children who had an infection prior to or as part of a shunt failure cluster were more likely to have an infection with GNR species and less likely to have a culture-negative infection than children with an infection but no history of cluster (p = 0.009, chi-square test). These results are shown in Table 4.

TABLE 4.

Shunt infection organisms

No. of Patients (%)
OrganismTotal Infections*Infections Before or During Shunt Failure Cluster (n = 21)Infections Not Associated w/ Cluster (n = 62)
Staphylococcus aureus278 (38)19 (31)
Coagulase-negative staphylococci142 (9.5)12 (19)
GNR species106 (29)4 (6.5)
Other143 (14)11 (18)
Culture negative but treated as infection160 (0)16 (26)

p = 0.009, chi-square test.

Specific organism data are missing for 2 patients.

Finally, we performed sensitivity analysis. First, we altered the definition of shunt failure cluster to 4 operations in 3 months. Only 8 children (1.7%) met these stricter criteria for shunt failure cluster. Because this is such a small proportion of the overall sample, no further analysis was performed for these patients. There were missing data for gestational age at birth (61 missing), head circumference at initial shunt placement (110 missing), and ventricle size at initial shunt placement (26 missing). We repeated the univariate logistic regression for each of these variables, replacing all missing values with the median, first quartile value, and third quartile value. There were no large changes to the odds ratios with any of these analyses (Table 5).

TABLE 5.

Sensitivity analysis for missing data

OR (95% CI)
VariableOriginal Univariate Logistic Regression ResultMissing Data Replaced by MedianMissing Data Replaced by 1st QuartileMissing Data Replaced by 3rd Quartile
Gestational age at birth*0.91 (0.85–0.97), p = 0.0030.90 (0.85–0.96)0.92 (0.86–0.96)0.91 (0.85–0.96)
Z-score for head circumference at initial shunt placement0.91 (0.84–0.99), p = 0.0440.90 (0.82–0.99)0.92 (0.84–1.02)0.89 (0.82–0.97)
FOHR at time of initial shunt1.64 (0.46–5.83), p = 0.441.68 (0.47–6.0)1.79 (0.53–6.1)1.57 (0.41–5.9)

Boldface type indicates statistical significance.

Data are missing for 61 patients.

Data are missing for 110 patients.

Data are missing for 26 patients.

Discussion

In this study, we have defined a new variable, shunt failure cluster, to quantify the experience of multiple frequent shunt failures. The sample is relatively large (465 patients) and has adequate follow-up (mean 5.4 years). We found that about 6% of children who have received care for their hydrocephalus at our institution have experienced at least one shunt failure cluster. Most commonly, the shunt failure cluster was early in the child’s history (the median number of failures prior to the cluster was 0) and occurred at a young age (median age at first cluster 246 days). The revisions in these patients (regardless of whether they are associated with a failure cluster) account for 30% of all revisions in our series. These data reinforce the observation that a handful of patients with difficult shunts bear a disproportionate share of shunt-related problems and make up a disproportionate share of the workload for pediatric neurosurgeons.

A history of shunt infection shows strong correlation with shunt failure cluster. Seventy-five percent of children with a cluster had at least one shunt infection either before or during the failure cluster. It is possible that children with shunt infection are more likely to have a shunt failure cluster. It is also possible that children with a cluster have a history of more operations and therefore a greater likelihood of an infection at some point in their lives. A comparison between children with a shunt failure cluster and no previous infections and those with infections before or during their cluster showed no significant relationship with any clinical or demographic factor. In other words, among children who had a shunt failure cluster, those with an infection history are similar to those without an infection history. As previous research has suggested, shunt infection is a complication to be avoided at all costs.

Analysis of the organisms causing shunt infection showed significantly higher proportions of GNR infections associated with clusters and more culture-negative infections in children with no history of cluster. This may indicate that more virulent organisms, such as GNR species, predispose children to more future shunt failures. One possible reason for this could be a higher level of immune system activity in the brains of children with GNR infections. This could in turn lead to greater inflammatory reaction to the ventricular catheter of a shunt system and more catheter occlusion. Previous laboratory work has suggested that inflammatory cells are a key component to shunt failures.10,11 Therefore, future study of methods to reduce cerebral inflammatory activity may lead to some reduction in clustered shunt failure.

Unfortunately, our data do not identify any easily modifiable risk factors for shunt failure cluster. The only baseline variables that showed association with shunt failure cluster were lower gestational age, smaller head circumference at time of shunt, and etiology of hydrocephalus. None of these maintained independent significance when tested in a multivariate model. Clearly, gestational age and hydrocephalus etiology are not modifiable. However, the relationship with head circumference warrants closer observation. If larger head circumference at initial shunt placement is protective against future shunt failure clusters, then perhaps delaying initial shunt placement as long as possible could be beneficial. However, these data are insufficient to support that approach. More importantly, these data speak only to the frequency of shunt revision and not to the outcome of the developing brain. Allowing a larger head circumference may result in impairment of brain development. Further study of the head size and ventricle size is warranted, including study of the changes in these parameters that result from shunting.

It is likely that younger gestational age and IVH etiology are factors associated with prematurity and that this is an important risk factor for future shunt failure cluster. While not modifiable, this is still important information to be used when counseling families about the long-term risks of shunt placement. Other potentially modifiable factors, such as endoscopic third ventriculostomy prior to initial shunt placement, entry site (anterior vs posterior), and type of valve, showed no association with shunt failure cluster.

While this is the first study to specifically analyze shunt failure clusters, previous studies have evaluated multiple shunt failures. Lazareff et al. studied 244 children over 6 years, comparing those who had no shunt revisions with those with 1, 2 or 3, or > 4 revisions.4 All children with a history of shunt infection were in one of the multiple-revision groups. They also found a higher CSF monocyte count among children with multiple revisions. In a study by McGirt et al., the risk of shunt failure was shown to decline with increasing time from the last revision. This study considered shunt infection as an adverse outcome, but not as a risk factor for subsequent failure.5 Finally, in an analysis of 830 patients, Tuli et al.6 considered only the first 3 shunt failures and showed a decreased risk of second and third failures with increasing time from the last operation, similar to the findings of McGirt et al. This study included no analysis of shunt failure cluster or of shunt infection.6

Limitations

The primary limitation of this study is its retrospective nature. However, most of the included data have been collected in a prospective fashion. Logistic regression analysis requires a sufficient number of events per variable tested. Therefore, it is possible that the multivariate analysis of shunt failure cluster is underpowered. However, our analysis adheres to the 10-event-per-variable rule.12 There are 3 variables with missing data: gestational age at birth (61 missing), head circumference at time of shunt placement (110 missing), and ventricle size at time of shunt placement (26 missing). These data are collected retrospectively and were not available for all patients. All data were collected from a single institution, including only patients who have received all hydrocephalus care at that one institution, limiting the generalizability of the findings. In particular, our institution has a high proportion of children with prematurity IVH as the etiology of hydrocephalus. Given the importance of prematurity seen in our results, this may be an important limitation to the external validity of these findings. Finally, the definition of shunt failure cluster as 3 operations within 3 months is arbitrary. However, a stricter threshold of 4 operations in 3 months yields too few cases for meaningful analysis.

Conclusions

We have defined a new variable, the shunt failure cluster, and found that 6% of children in our institutional sample have experienced at least one shunt failure cluster. These children account for 30% of all shunt revisions in this cohort. Shunt infection is an important risk factor for shunt failure cluster.

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author Contributions

Conception and design: Rocque, Hopson, Kerr, Johnston, Rozzelle, Blount. Acquisition of data: Waldrop, Shamblin, Arynchyna, Hopson. Analysis and interpretation of data: Rocque, Shamblin, Arynchyna, Blount. Drafting the article: Rocque, Waldrop, Shamblin. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Rocque. Statistical analysis: Rocque. Administrative/technical/material support: Rocque, Shamblin, Arynchyna, Hopson. Study supervision: Rocque, Blount.

Supplemental Information

Previous Presentations

A portion of this work was presented in poster form at the 48th Annual Meeting of the AANS/CNS Joint Section on Pediatric Neurological Surgery, Scottsdale, Arizona, December 5–8, 2019. The poster title was “Pediatric neurosurgeon’s bane: clustered-shunt failure.”

References

  • 1

    Riva-Cambrin J, Kestle JRW, Holubkov R, et al. Risk factors for shunt malfunction in pediatric hydrocephalus: a multicenter prospective cohort study. J Neurosurg Pediatr. 2016;17(4):382390.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Bir SC, Konar S, Maiti TK, et al. Outcome of ventriculoperitoneal shunt and predictors of shunt revision in infants with posthemorrhagic hydrocephalus. Childs Nerv Syst. 2016;32(8):14051414.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Kestle J, Drake J, Milner R, et al. Long-term follow-up data from the Shunt Design Trial. Pediatr Neurosurg. 2000;33(5):230236.

  • 4

    Lazareff JA, Peacock W, Holly L, et al. Multiple shunt failures: an analysis of relevant factors. Childs Nerv Syst. 1998;14(6):271275.

  • 5

    McGirt MJ, Leveque J-C, Wellons JC III, et al. Cerebrospinal fluid shunt survival and etiology of failures: a seven-year institutional experience. Pediatr Neurosurg. 2002;36(5):248255.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Tuli S, Drake J, Lawless J, et al. Risk factors for repeated cerebrospinal shunt failures in pediatric patients with hydrocephalus. J Neurosurg. 2000;92(1):3138.

  • 7

    Das MK, Bhattacharyya N, Bhattacharyya AK. WHO child growth standards. Eur J Pediatr. 2010;169(2):253–255, 257258.

  • 8

    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.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Kestle JR, Riva-Cambrin J, Wellons JC III, et al. A standardized protocol to reduce cerebrospinal fluid shunt infection: the Hydrocephalus Clinical Research Network Quality Improvement Initiative. J Neurosurg Pediatr. 2011;8(1):2229.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Hanak BW, Bonow RH, Harris CA, Browd SR. Cerebrospinal fluid shunting complications in children. Pediatr Neurosurg. 2017;52(6):381400.

  • 11

    Harris CA, McAllister JP II. What we should know about the cellular and tissue response causing catheter obstruction in the treatment of hydrocephalus. Neurosurgery. 2012;70(6):15891602.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    van Domburg R, Hoeks S, Kardys I, et al. Tools and techniques—statistics: how many variables are allowed in the logistic and Cox regression models? EuroIntervention. 2014;9(12):14721473.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand
Figure from Coblentz et al. (pp 346–356).
  • 1

    Riva-Cambrin J, Kestle JRW, Holubkov R, et al. Risk factors for shunt malfunction in pediatric hydrocephalus: a multicenter prospective cohort study. J Neurosurg Pediatr. 2016;17(4):382390.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Bir SC, Konar S, Maiti TK, et al. Outcome of ventriculoperitoneal shunt and predictors of shunt revision in infants with posthemorrhagic hydrocephalus. Childs Nerv Syst. 2016;32(8):14051414.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Kestle J, Drake J, Milner R, et al. Long-term follow-up data from the Shunt Design Trial. Pediatr Neurosurg. 2000;33(5):230236.

  • 4

    Lazareff JA, Peacock W, Holly L, et al. Multiple shunt failures: an analysis of relevant factors. Childs Nerv Syst. 1998;14(6):271275.

  • 5

    McGirt MJ, Leveque J-C, Wellons JC III, et al. Cerebrospinal fluid shunt survival and etiology of failures: a seven-year institutional experience. Pediatr Neurosurg. 2002;36(5):248255.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Tuli S, Drake J, Lawless J, et al. Risk factors for repeated cerebrospinal shunt failures in pediatric patients with hydrocephalus. J Neurosurg. 2000;92(1):3138.

  • 7

    Das MK, Bhattacharyya N, Bhattacharyya AK. WHO child growth standards. Eur J Pediatr. 2010;169(2):253–255, 257258.

  • 8

    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.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Kestle JR, Riva-Cambrin J, Wellons JC III, et al. A standardized protocol to reduce cerebrospinal fluid shunt infection: the Hydrocephalus Clinical Research Network Quality Improvement Initiative. J Neurosurg Pediatr. 2011;8(1):2229.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Hanak BW, Bonow RH, Harris CA, Browd SR. Cerebrospinal fluid shunting complications in children. Pediatr Neurosurg. 2017;52(6):381400.

  • 11

    Harris CA, McAllister JP II. What we should know about the cellular and tissue response causing catheter obstruction in the treatment of hydrocephalus. Neurosurgery. 2012;70(6):15891602.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    van Domburg R, Hoeks S, Kardys I, et al. Tools and techniques—statistics: how many variables are allowed in the logistic and Cox regression models? EuroIntervention. 2014;9(12):14721473.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 513 0 0
Full Text Views 1119 987 120
PDF Downloads 396 272 27
EPUB Downloads 0 0 0