Introduction

Hydrocephalus in children remains one of the more common etiologies for admission in the pediatric neurosurgical units. [1] Ventriculoperitoneal shunt (VPS) is considered the standard of care, despite the advancements and introduction of new cerebrospinal fluid (CSF) diversion approaches in hydrocephalus’ neurosurgical management, including the endoscopic third ventriculostomy (ETV). [2, 3] Nevertheless, VPS is associated with significant complications including shunt malfunction, infections, and inconsistent long-term motor and cognitive outcomes. [4] Several studies have evaluated the current shunt success and failure rates as compared to those of the past decades; however, results are inconclusive in the literature. [5, 6]

ETV has emerged as an alternative CSF diversion procedure especially in non-communicating hydrocephalus cases [7, 8]; however, it has been suggested that ETV might also be beneficial for some pediatric patients with communicating hydrocephalus. [9, 10] Recently, the addition of choroid plexus cauterization (CPC) during ETV has been reported to improve the efficacy of the endoscopic approach. [11, 12] This increasing interest is reflected by the emerging number of studies evaluating the comparative effectiveness of VPS and ETV. [13, 14]

The aims of this study are to systematically review the literature and evaluate the comparative safety and efficacy of shunting and ETV in pediatric hydrocephalus cases. Knowing the limitations of mixed hydrocephalus etiology, geographical location, and changes in the performance of ETV, this study tries to investigate their effect by conducting subgroup analyses.

Methods

This systematic review and meta-analysis was performed according to the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines. [15]

Search strategy and selection criteria

Systematic searches were conducted in PubMed (Medline) and Cochrane Databases. The keywords used for PubMed were “endoscopic third ventriculostomy”, “shunt”, “children”, “infants”, “neonates”, and “pediatric”. The search was conducted by two independent investigators (PT, MT). Any disagreements or discrepancies were resolved by consensus. The references of the included studies were also manually reviewed in order to identify further eligible articles.

A study was included in this meta-analysis if it fulfilled four predefined criteria: (i) randomized controlled trials (RCT) or prospective and retrospective observational analyses comparing the shunting vs ETV (with or without CPC); (ii) studies that reported quantitative data on clinical outcomes of interest; and (iii) studies published up to October 2018. When duplicates were identified, the most recent analysis was included unless the earliest version reported more relevant outcomes.

Data extraction and outcomes

Two reviewers, blind to each other (PT, MT), independently extracted the relevant data from the eligible studies. Variables abstracted were first author, year of publication, years of enrollment, country, design of study, study arms, hydrocephalus etiology, age, gender, failure rates, CSF leak, infection, mortality, re-operation, shunt malfunction, and intraventricular hemorrhage. All disagreements were resolved following discussion and final decision was reached by consensus. The primary outcome was incidence of infection. A standardized definition for infection could not be provided due to lack of reporting by the included studies. Secondary outcomes were all-cause mortality, CSF leak, and the need for re-operation. Re-operation in both groups was defined as the need for any second operation either shunt or ETV during the follow-up period.

Risk of bias assessment

Risk of bias was assessed by two investigators (PT, MT) with the Robins-I tool for non-randomized studies. [16] The following domains for the non-randomized studies were evaluated: confounding, selection of participants, departure from intended interventions, missing data, measurement of outcomes, and selective reporting. RCTs were assessed with the Cochrane tool. [17] Discrepancies in quality assessment were resolved via consensus.

Statistical synthesis and analysis

Odds ratios (ORs) with the corresponding 95% confidence intervals (CIs) were used for the outcomes. A random effects model was used to account for heterogeneity among studies. Heterogeneity was assessed with the Higgins I-square statistic. [18] I2 greater than 50% indicated significant heterogeneity. [18] Forest plots were used to graphically display the effect size in each study and the pooled estimates. Meta-regression analysis was performed to adjust for the age and various follow-up intervals as a study level covariates. A p value of < 0.05 was considered significant. STATA 14.1 (StataCorp, College Station, Texas) statistical software was used for all analyses.

Results

Search results

Literature search yielded 683 potentially relevant records after duplicates were removed. After screening titles and abstracts, 18 articles were retrieved for full-text evaluation. Fourteen studies met the predetermined eligibility criteria and were included in the meta-analysis as shown in the PRISMA flow diagram (Fig. 1).

Fig. 1
figure 1

PRISMA search flow diagram

Full size image

Characteristics of the included studies

This meta-analysis of 14 studies comprised 8419 patients overall (ETV: 1994; shunt: 6425 patients). [1, 3, 13, 14, 19,20,21,22,23,24,25,26,27,28] Thirteen studies were observational and one was an RCT. None of the included observational studies or the sole RCT were assessed as having high risk of bias. A detailed assessment of risk of bias is available in the Supplemental Tables 1 and 2. Three of the included studies compared the combined ETV/CPC vs shunt approach while the rest compared ETV vs shunt. Five studies were conducted in Europe, four in Africa, four in USA/Canada, and one was international. Pertinent patient and study characteristics are presented in Table 1.

Table 1 Baseline patient and study characteristics
Full size table

Post-operative outcomes for ETV ± CPC vs shunt

Pediatric patients who had an ETV (with or without CPC) were at a statistically significant lower risk of procedure-related infection compared to shunt (ETV: 1/300 (0.33%); shunt: 35/389 (8.9%)) (OR: 0.19; 95% CI: 0.07–0.53; I2: 0%) (Fig. 2). There were no differences between patients in the ETV (43.4%, N = 864/1989) vs shunt groups (39.6% N = 2496/6295) (OR: 0.72; 95% CI: 0.39–1.32; I2: 93.5%) in terms of re-operation (Fig. 3). The funnel plot for re-operation demonstrated asymmetry and the Egger’s regression test (p = 0.033) validated that there is a high risk of publication bias for this outcome. (Supplemental Fig. 1). During the follow-up, all-cause mortality was reported in 4.5% (N = 12/262) and 8.5% (N = 38/447) of patients in the ETV and shunt groups, respectively; however statistical significance was not reached (OR: 0.77; 95% CI: 0.35–1.68; I2: 0%) (Fig. 4). The incidence of post-procedural CSF leak was similar between the ETV (5.8%; N = 10/172) and shunt groups (3.2%; N = 3/94) (OR: 1.53; 95% CI: 0.37–6.31; I2: 0%) (Fig. 5). Shunt malfunction of any type was reported in 34.7% of the patients (N = 331/954).

Fig. 2
figure 2

Forest plot comparing post-operative infection in ETV vs shunt

Full size image
Fig. 3
figure 3

Forest plot comparing re-operation rates in ETV vs shunt

Full size image
Fig. 4
figure 4

Forest plot comparing all-cause mortality in ETV vs shunt

Full size image
Fig. 5
figure 5

Forest plot comparing cerebrospinal fluid (CSF) leak in ETV vs shunt

Full size image

Subgroup analyses for ETV/CPC and ETV only

Subgroup analyses for infection showed that ETV/CPC studies were not associated with a statistically significant lower risk for infection compared to shunt (OR: 0.32; 95% CI: 0.04–2.38; I2: 23.9%) whereas ETV only studies showed a statistically significant lower risk of procedure-related infection (OR: 0.15; 95% CI: 0.04–0.52; I2: 0%) (Supplemental Fig. 2). Subgroup analyses for the other outcomes were consistent with the pooled effect estimate and were not presented.

Subgroup analyses based on the geographic region

A subgroup analysis of re-operation based on the geographic region of the patients identified that ETV was associated with statistically significant lower odds to undergo any repeat procedure compared to shunt in studies conducted in Europe (OR: 0.39; 95% CI: 0.30–0.52; I2: 1.4%) and Africa (OR: 0.13; 95% CI: 0.03–0.48; I2: 0%) (Fig. 6). Studies conducted in USA/Canada did not show differences between the study groups and were accompanied by a significant amount of heterogeneity (OR: 1.49; 95% CI: 0.85–2.63; I2: 86.2%) (Fig. 6). One study was international and was not included in any of the subgroups. The sole RCT was not included in the African subgroup as it would be inappropriate to be pooled with the other observational studies (see “Discussion”).

Fig. 6
figure 6

Subgroup analysis of re-operation for ETV and shunt divided by the geographic region

Full size image

Subgroup analyses for obstructive hydrocephalus cases only

The subgroup analysis of studies including only obstructive hydrocephalus cases showed that re-operation rates were 27.6% and 40.2% in the ETV and shunt groups, respectively; however, statistical significance was not reached (OR: 0.55; 95% CI: 0.24–1.25; I2: 66.4%) (Fig. 7). Patients with obstructive hydrocephalus in the ETV group had statistically significant lower odds for infection as compared to the shunt group (OR: 0.14; 95% CI: 0.04–0.58; I2: 0%) (Supplemental Fig. 3). No differences were identified in terms of all-cause mortality between the two groups (OR: 0.34; 95% CI: 0.09–1.34; I2: 0%) (Supplemental Fig. 4).

Fig. 7
figure 7

Subgroup analysis of re-operation for ETV and shunt including only obstructive hydrocephalus cases

Full size image

Meta-regression analysis

Meta-regression analysis suggested that there is no statistically significant effect of the length of follow-up (coefficient: − 0.003; 95%CI: − 0.015–0.009, p = 0.57) (Supplemental Fig. 5) and average age (coefficient: − 0.027; 95%CI: − 0.011–0.006, p = 0.50) (Supplemental Fig. 6) of the patient population on the pooled estimate of re-operation. This suggests that these variables were not significant sources of heterogeneity.

Discussion

This meta-analysis included 8419 patients and evaluated the comparative efficacy of ETV vs shunt. Our results showed that patients who had ETV were at a statistically significant lower risk of any procedural-related infection, which was consistent in the subgroup analysis of obstructive hydrocephalus cases only. No differences in terms of repeat operations, mortality, and CSF leak were identified between the two groups in the original pooled analyses. Interestingly, ETV patients in Africa and Europe had significantly lower odds to undergo a re-operation than shunt patients whereas no difference was identified for patients in the USA/Canada.

Even though VPS has been used for several decades, it still remains associated with multiple complications including procedure-related infections, CSF leaks, a high failure, and re-operation rate and even mortality. [1] When ETV was introduced as an attractive alternative of shunting, its use was initially restricted to hydrocephalus caused by aqueductal stenosis; however, its indications are rapidly expanding. This was also reflected in the ETV indications reported in the included studies of this meta-analysis, including obstructive and non-obstructive hydrocephalus cases. [3, 29] ETV and shunting can be used for overlapping indications but a number of individual variables can predict procedural success and influence patient selection. Because of this, the choice of ETV and shunt is not always a simple binary argument of one versus the other. We therefore, caution the reader from interpreting the results of the present study to conclude that one procedure is simply superior to the other.

One of the most important complications in CSF diversion procedures is infections. [20] Despite significant efforts from organizations such as Hydrocephalus Clinical Research Network (HCRN), shunt infection remains a major source of morbidity and mortality. [30] The current meta-analysis demonstrated that ETV is associated with a statistically significant lower risk of procedure-related infection compared to shunting, during the follow-up. Interestingly, none of the individual studies reporting on this comparison showed statistically significant differences between ETV and shunt. It is likely that these studies were underpowered to detect this difference; however, meta-analytic methods increased the statistical power substantially, and therefore, statistical significance was reached. The pooled analysis including ETV plus or minus CPC and the subgroup analysis of ETV alone vs. shunt showed statistically significant lower odds of infection when ETV was used. However, the subgroup analysis of ETV + CPC vs shunt did not reach statistical significance. This could be explained by a lack of statistical power in this subgroup analysis, as the absolute rates of infection were still lower in the ETV + CPC group as compared to shunt (Supplemental Fig. 2). Importantly, even though several studies have suggested that the up-front costs of shunting and ETV are similar, the average cost of admission due to shunt infection was $83,649. [27, 31] Future studies should investigate the potential long-term cost effectiveness of ETV compared to shunting by accounting for the cost of infections and re-operations needed in each group.

In cases where the primary CSF diversion procedure fails, a second surgery (either repeat of the first or use of an alternative modality) is commonly needed. [3, 32] In the studies included in our analysis, the specific type of second surgery was inconsistently reported. Our pooled analysis did not show any significant differences between ETV vs. shunt in terms of re-operation, but notably, this analysis was accompanied by a significant amount of heterogeneity (I2: 94.2%). The heterogeneity prompted a subgroup analysis of re-operation based on the Continent in which the procedures were performed. Re-operation following ETV was 49.9% (N = 693/1387) in USA/Canada, 29.9% (N = 116/387) in Europe, and 5.2% in Africa (N = 3/57). However, re-operation rates following shunt were 37.8% (N = 1984/5237) in USA/Canada, 52.4% (N = 476/907) in Europe, and 27.4% (N = 14/51) in Africa. Furthermore, the subgroups of patients in Europe and Africa not only showed statistically significant lower re-operation rates in the ETV group, but also demonstrated a major reduction in heterogeneity (I2 = 1.4% and I2 = 0%, respectively) as compared to the pooled analysis (I2 = 93.4%). In contrast, USA/Canada studies demonstrated the trend that shunt is associated with fewer re-operations; however, statistical significance was not reached. This is in line with a large prospective study by the HCRN which showed that ETV/CPC appears to be associated with a higher failure rate as compared to shunting. [33]

One possible explanation could be that studies conducted in the USA/Canada had shorter follow-up intervals compared to European or African, thus shunt cases did not have adequate time to develop complications and need re-operation. The two studies on patients in Africa had a follow-up of approximately 1 and 2 years, whereas the European studies reported a mean follow-up that ranged between 1 and 8 years. In contrast, the USA/Canada studies had a follow-up that ranged between 3 months and 3 years. Nevertheless, results from the meta-regression analysis showed that the duration of follow-up did not affect the size estimate of re-operation in the included studies. Another explanation may infer heterogeneities in shunt instrumentation, including valve types, catheters, anti-siphon devices, tip placement, and other technical considerations of the shunt or ETV approaches themselves. These details were inconsistently reported among the included studies. Lastly, perhaps most importantly, socioeconomic and cultural differences in availability of acute and emergent care capacities and attitudes toward repeat operations would have influenced clinicians in deciding performing ETV versus shunt.

It should be noted that three studies overall were conducted in Africa. Two of them were retrospective and when pooled, showed that ETV was associated with significantly lower odds of re-operation. The third study was a randomized trial and was not pooled with the two others, owing to the fundamental differences in their designs (randomized vs. observational). [34] In addition, even though this RCT was conducted in Africa, it was performed by North American investigators whereas the other African studies were performed by African investigators. Moreover, the RCT included only non-obstructive hydrocephalus cases for which the efficacy of ETV has been questioned. [9, 10, 14] In contrast, the two observational African studies included only obstructive hydrocephalus cases for which ETV efficacy has been well-documented. Due to these fundamental differences, the authors did not elect to pool the three African studies together.

The current study suggests that all-cause mortality and CSF leak rates are similar between the ETV and shunt groups. Our results are in agreement with the sole RCT by Kulkarni et al. who did not show any statistically significant differences between the study groups in terms of all-cause mortality. CSF leak was not reported by this RCT but was consistent with all included studies that synthesized this outcome. Future RCTs are warranted to validate the results of this study. In particular, future studies would benefit from standardized outcomes definitions for success and failure of ETV or shunt, uniform age and hydrocephalus etiology subgroups, and a predetermined long-term follow-up.

Limitations

Results of this study should be interpreted in the context of several limitations. First, the majority of the studies were observational which have an inherent risk for confounding and selection bias. Second, the follow-up intervals varied among the studies; however, our meta-regression analysis showed that re-operation rates were not affected by the duration of the follow-up. Third, patient populations were heterogeneous across the studies. The age groups were variable, but meta-regression results showed that age did not affect re-operation rates. This suggests that age was not a significant source of heterogeneity. The etiology of hydrocephalus varied across the studies; however, we were able to conduct subgroup analyses for obstructive hydrocephalus cases only, which showed similar results to those of the pooled analyses. In the pooled analyses, ETV and ETV + CPC were combined. However, to eliminate bias associated with this combination, subgroup analyses were conducted. Lastly, the sample size in the African subgroup was small, which limits the generalizability of our results.

Conclusions

This study shows that ETV was associated with a statistically significant lower risk of procedure-related infection compared to shunt. All-cause mortality and CSF leak rates were similar between the study groups. Re-operation rates were similar between the study groups in the pooled analysis. Subgroup analysis, based on the geographic region that the studies were conducted, suggested that ETV is associated with statistically significant lower odds for re-operation in Europe and Africa, but not in USA/Canada. Future RCTs are needed to validate the results of this study.

Funding statement

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.