Shunt complications and revisions in children: A retrospective single institution study

Abstract Introduction Shunt surgery in children is associated with high revision and complication rates. We investigated revision rates and postoperative complications to specify current challenges associated with pediatric shunt surgery. Methods All patients aged < 18 years admitted to St. Olavs University Hospital, Norway, from January 2008 through December 2017, who underwent primary shunt insertions, were reviewed. Follow‐up ranged from 1 to 10 years. Ventriculoperitoneal, cystoperitoneal, and ventriculoatrial shunts were included. All subsequent shunt revisions and 30‐day postoperative complication rates were registered. Results 81 patients underwent 206 surgeries in the study period. 47 patients (58%) required minimum one revision during follow‐up. In 14 (29.8%), the first revision was due to the misplacement of hardware. Proximal occlusion was the most common cause of revision (30.4%), followed by misplacement (18.5%) and infection (9.6%). Young age and MMC were associated with revision surgery in a univariable analysis, but were not significant in multivariable analyses. Congenital hydrocephalus was associated with infection (p = .028). In approximately 30% of procedures, complications occurred within 30 days postoperatively, the most common being revision surgery. In approximately 5% of the procedures, medical complications occurred. Conclusion Children are prone to high revision and complication rates, and in this study, misplacement of hardware and proximal occlusion were the most common. Complication rates should not be limited to revision rates only, as 30‐day complication rates indicate a significant rate of other complications as well. Multi‐targeted approaches, perhaps focusing on measures to reduce misplacement, may be key to reducing revision rates.


INTRODUCTION
The global prevalence of hydrocephalus (HC) in the pediatric population has been estimated in several studies to be approximately 1 in 1000 births, with medium-to-low-income countries having a significantly higher incidence than high-income countries (Dewan et al., 2019;Isaacs et al., 2018;Kahle et al., 2016). The most common etiologies in high-income countries are HC after intraventricular hemorrhage (IVH) in the premature, aqueduct stenosis, myelomeningocele (MMC), and tumors/malignancy, while in lowincome countries, CNS infection is a dominant culprit (Kahle et al., 2016).
In the literature, complications rates are often limited to revision rates as a measure of shunt failure or success. However, in an adult population, we found that complications extend beyond revision surgeries alone, with overall 30-day complication rates of 45% in a cohort with 20.7% revision rates (Mansoor et al., 2020). We believe that reporting revision rates alone might represent a great underrepresentation of the actual challenges hydrocephalic patients face. Also, failure to assess all aspects of shunt surgery and its complications may limit the ability to improve outcomes by targeting a wider range of complications.
In the current study, we aimed to similarly examine the pediatric population, and in addition to assessing revision rates, we sought to systematically classify the presence of other complications such as postoperative extracranial infections and other medical and surgical complications.

METHODS
Ethical approval and waiver of the requirement for obtaining patient consent were granted by the Regional Committee for Medical Research, REK 2017/1796.

Data collection
In the current study, we retrospectively reviewed all children ages

Patient demographics
An overview of patient demographics and univariable comparisons of patients undergoing revision surgery and the non-revised group are presented in Table 1. A total of 81 patients were included in the study. A univariable cox analysis showed that age was inversely associated with the need for shunt revision (p = .003, HR 0.88), while the diagnosis of MMC was directly associated with shunt revision (p = .008, HR 3.0).
Gender and other etiologies were not significant in a univariable analysis. Age was still inversely associated (p = .019, HR 0.90) with the need for revision in a multivariable analysis testing for age, gender, and the most common etiologies, compromising 54.3% of all patients (congenital HC, tumor/malignancy, arachnoid cyst). Congenital HC was associated with infection in a univariable analysis (p = .028, HR 4.61), and was also significant in a multivariable analysis testing for age, sex, and the most common etiologies, congenital HC, tumor/malignancy, and arachnoid cyst (p = .033, OR 6.09).

Shunt surgery characteristics
An overview of shunt surgery characteristics and differences between the revision and non-revision group is listed in

Revision surgery
A total of 125 revision surgeries were performed during the follow- for all revisions regardless of whether it was a first or subsequent revision was proximal occlusion followed by misplacement. Infection was the cause of 9.6% of all revisions, and occurred in 5.8% of all procedures. In terms of all procedures, proximal occlusion was the most common cause for revision, followed by misplacement and infection.
Time from primary shunt insertion to first revision was median 18 weeks, range 0-425 weeks.

Shunt surgery complications
Overall complication rates in accordance with the grading system introduced by Ibanez

DISCUSSION
In this retrospective single center study, 81 children underwent primary shunt insertion in a 10-year period. A total of 125 revisions were performed in this cohort during follow-up. In total, 42% of patients required revision surgery within one year and nearly 60% required shunt revision during follow-up. Hardware misplacement and proximal occlusion were the most common causes for revision, followed by infection. The revision group was significantly younger, and patients with congenital HC and MMC were more likely to experience revision surgery. Approximately one-third experienced surgical or medical complications within 30-days as classified by Ibanez et al (2011). This highlights that shunt complications are more than shunt failure or shunt infections. Measures to improve outcomes in the pediatric shunt population should be multi-focused, and should target surgical precision in drain placement, perhaps especially in patients with higher risks of revisions. This could perhaps also improve long-term patency. Lower revision rates will naturally also reduce other complication rates.  Crowley et al., 2014;Janson et al., 2014). A review indicated level III evidence for the use of ultrasound and electromagnetic guidance for ventricular catheter placement (Flannery et al., 2014); at least one study was, however, most likely underpowered due to low number of participants. Studies that are clearly underpowered should not mistakenly be interpreted as lack of effect of said measure. In our study, neuronavigation or ultrasound guidance was only used in five patients during primary shunt insertion, but based on our findings, the routine use of neuronavigation or other tools to improve accuracy is probably the way to advance.
Postoperative scans were generally only performed in cases of clinical suspicion of shunt dysfunction. The number of revision rates due to misplacement may be an underestimation of actual misplacements.
Although one prospective study failed to show improved shunt survival by using ultrasound or stereotactic guidance (Riva-Cambrin et al., 2016), one cannot dismiss that improved placement, and avoiding the choroid plexus can reduce not only revision rates due to misplacement, but also perhaps due to occlusion. As the most frequently encountered revision causes; hence focus should be directed toward this.
In terms of infection rates already extensively examined in the literature, studies report greatly diverse infection rates, but generally ranging from 4% to 10% (Erps et al., 2018;Reddy et al., 2011;Riva-Cambrin et al., 2016;Simon et al., 2009Simon et al., , 2014, but also higher (Bir et al., 2016;Simon et al., 2012). Our results are on the lower end.
One study compared two cohorts almost 50 years apart, reporting few differences in shunt survival, and reporting 0.0% infection rate in the pediatric population, but the pediatric population was small (Mansson et al., 2017). Young age and perhaps more importantly, repeated revisions, have regularly been associated with increased risk of shunt infection (Arslan et al., 2018;Berry et al., 2008;Erps et al., 2018;Reddy et al., 2011;Simon et al., 2012Simon et al., , 2014. Some etiologies have also been associated with increased infection rates (Arslan et al., 2018;Tuli et al., 2000). Young age and etiology are undoubtedly intertwined; MMC has been suggested to be associated with infection (Arslan et al., 2018), but children with MMC usually undergo shunt insertion at a very young age, the latter probably being the most important factor.

TA B L E 4 Description of complications
The use of antibiotic-impregnated catheters (AIS) reduces infection rates (Mallucci et al., 2019;Sciubba et al., 2005). The regular use of an AIS should therefore be the gold standard. Only seven patients received AIS in the current study, but the number is too low to make any meaningful analysis on this. Although shunt infection is a serious complication, representing increased morbidity and mortality in shunt patients, it is not necessarily the most frequently encountered revision cause.
Revision surgery occurred most commonly within one year, findings similarly reported by others (Mansoor et al., 2020;Merkler et al., 2017;Stein & Guo, 2008). However, revision surgery also occurred after 8 years in the current study. One study followed patients for a minimum of 15 years and reported revision rate of 85%; 12.5% did not require their first revision before >10 years after initial shunt insertion, with the latest revision occurring as late as 17 years after shunt insertion (Stone et al., 2013). Other studies have similarly reported the need of revision surgery several decades after primary surgery (Bir et al., 2016;Paulsen et al., 2015;Vinchon et al., 2012). This indicates likely significant underreporting in a great portion of the current literature as few studies follow patients for such a long period of time.
In terms of etiologies presented in this study, the most noteworthy is perhaps the high proportion of patients with arachnoid cysts receiving shunts. Ten out of these 11 patients did undergo craniotomy with cyst fenestration prior to shunt insertion, but with inadequate results.
There were relatively few patients with IVH compared to other studies; in four out of the five patients with ICH, imaging revealed blood in the ventricle system, but the primary diagnosis was not IVH, but rather parenchymal hemorrhage from arteriovenous malformations and other pathologies, which is why the two entities were not put together.
This study did not contemplate long-term complications in terms of functional disabilities. Studies indicate that pediatric HC and shunt treatment are associated with cognitive challenges and reduced social functioning as well as mental health issues Kulkarni & Shams, 2007;Paulsen et al., 2015;Vinchon et al., 2012). Subsequent shunt complications, prolonged and frequent treatments for infections, and possibly also length of hospital stay (initial HC treatment) are factors that have been found to be associated with increased disability and reduced quality of life (Kulkarni & Shams, 2007); this also highlights the point in reducing the overall shunt complication rates, and not only one single factor.
Our study has several limitations, its retrospective design being one.
Longer follow-up would probably have indicated even higher frequency of shunt failure than currently reported. We suspect some overlap when it comes to etiologies, perhaps presenting difficulties when comparing the current results to other studies; authors have categorized HC differently in different studies, which may limit the generalizability of the results. Challenges of defining HC have previously been discussed, and due to the great variance in how authors have defined HC, it may affect the results in terms of specific etiologies being established risk factors for shunt failure.

CONCLUSIONS
We consider it worthwhile to reflect on all aspects that can reduce overall revision rates; whilst age and etiology are non-modifiable factors frequently associated with the occurrence of revisions and complications, there should be a greater focus on any modifiable factors that can contribute to increased shunt survival and reduced morbidity. In the current study, misplacement and proximal occlusion accounted for the most frequent revisions. Increased attentiveness toward the early postoperative phase, with focus on prompt mobilization and prevention of extracranial infections, and so forth, are vital for improving care for these patients. All measures, however small, should be attempted to reduce overall complication rates.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
Data available on request due to privacy/ethical restrictions.