Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Apr 1.
Published in final edited form as: Neurocrit Care. 2021 Sep 8;36(2):536–545. doi: 10.1007/s12028-021-01343-9

Association of External Ventricular Drain Wean Strategy with Shunt Placement and Length of Stay in Subarachnoid Hemorrhage: A Prospective Multicenter Study

David Y Chung 1,2,*, Bradford B Thompson 3, Monisha A Kumar 4, Ali Mahta 3, Shyam S Rao 3, James H Lai 1,2, Aleksey Tadevosyan 4, Kathryn Kessler 4, Joseph J Locascio 5, Aman B Patel 6, Wazim Mohamed 7, DaiWai M Olson 8, Sayona John 9, Guy A Rordorf 1
PMCID: PMC9128519  NIHMSID: NIHMS1791765  PMID: 34498207

Abstract

Background:

Survivors of aneurysmal subarachnoid hemorrhage (SAH) face a protracted intensive care unit (ICU) course and are at risk for developing refractory hydrocephalus with the need for a permanent ventriculoperitoneal shunt (VPS). Management of the external ventricular drain (EVD) used to provide temporary cerebrospinal fluid diversion may influence the need for a VPS, ICU length of stay (LOS), and drain complications, but the optimal EVD management approach is unknown. Therefore, we sought to determine the effect of EVD discontinuation strategy on VPS rate.

Methods:

This was a prospective multicenter observational study at six neurocritical care units in the United States. The target population included adults with suspected aneurysmal SAH who required an EVD. Patients were preassigned to rapid or gradual EVD weans based on their treating center. The primary outcome was the rate of VPS placement. Secondary outcomes were EVD duration, ICU LOS, hospital LOS, and drain complications.

Results:

A rapid EVD wean protocol was associated with a lower rate of VPS placement, including a delayed post-hospitalization shunt, in an adjusted Cox proportional analysis (hazard ratio 0.52 [p = 0.041]) and adjusted logistic regression model (odds ratio 0.43 [95% confidence interval 0.18–1.03], p = 0.057). A rapid wean was also associated with 2.1 fewer EVD days (p = 0.007) and saved an estimated 2.5 ICU days (p = 0.049), as compared with a gradual wean protocol. There were fewer nonfunctioning EVDs in the rapid group (odds ratio 0.32 [95% confidence interval 0.11–0.92]). Furthermore, we found that the time to first wean and the number of weaning attempts were important independent covariates that affected the likelihood of receiving a VPS and the duration of ICU admission.

Conclusions:

A rapid EVD wean was associated with decreased rates of VPS placement, decreased ICU LOS, and decreased drain complications in survivors of aneurysmal SAH. These findings suggest that a randomized multicentered controlled study comparing rapid vs. gradual EVD weaning protocols is justified.

Keywords: Brain aneurysm, External ventricular drain, Intrcranial pressure, Hydrocephalus, Ventriculoperitoneal shunt, Neurosurgery, Neurocritical care

Introduction

Aneurysmal subarachnoid hemorrhage (SAH) leads to disproportionately high morbidity and mortality compared with other forms of stroke [1]. Patients who survive the initial aneurysm rupture typically require intensive care unit (ICU) admission because of a high incidence of delayed neurological decompensation [2]. A subset of patients also develop acute hydrocephalus, with cited rates ranging from 15 to 87% [1], and require cerebrospinal fluid (CSF) diversion with an external ventricular drain (EVD) [3]. Placement of an EVD can be lifesaving and necessary to prevent elevated intracranial pressure (ICP) [4]. What is controversial is determining the best management of the EVD once it is in place and the best way to discontinue it when no longer needed [57].

Recent work has suggested that EVD discontinuation protocols may influence the need for a ventriculoperitoneal shunt (VPS) as well as ICU and hospital length of stay (LOS) [8]. Decreased VPS placement is an important goal by itself, as it is a neurosurgical procedure that involves a lifelong implant. The incidence of VPS complications is as high as 33%, even in the years following placement [9]. Therefore, an intervention undertaken in the ICU to obviate the need for a VPS and decrease LOS would be beneficial for patients and health care systems.

Rapid and gradual weans are the predominant paradigms to discontinue an EVD [10]. Once the decision is made to initiate a wean, a rapid wean typically consists of immediately clamping the drain to determine whether a patient is dependent on CSF diversion. A gradual wean increases the EVD ICP threshold daily in a stepwise fashion. The existing evidence for the best approach is mixed and prospective multicenter studies are lacking [3]. Therefore, we conducted a multicenter prospective study of EVD management in patients with aneurysmal SAH and hydrocephalus. Institutional and surgeon preferences for the type of EVD approach led to preassignment into a rapid or gradual wean group. The a priori aim of the study was to determine the effect of EVD wean approach on the incidence of VPS placement and ICU and hospital LOS.

Methods

The study was approved by individual institutional review boards at each of the following six participating centers: Massachusetts General Hospital (2017P000008), University of Texas Southwestern Medical Center (STU 072017102), Rush University Medical Center (17052306-IRB01), Rhode Island Hospital (1108230-9), Wayne State University (120617MP2E), and the Hospital of the University of Pennsylvania (17052306). We used the Strengthening the Reporting of Observational Studies in Epidemiology checklist for study design and reporting of results [11] (Supplemental Data).

Patients were enrolled prospectively between June 2016 and June 2019 from five centers that took a preassigned rapid or gradual EVD wean approach and from one center that took both approaches (preassigned group based on the treating neurosurgeon). Prior to data collection, each center established EVD management guidelines developed internally (Supplemental Data). The study was observational, but the patients did not self-select into treatment arms. Therefore, although patients were not randomly assigned to a treatment arm, they were assigned in a way to avoid treatment bias or other types of selection bias. A rapid wean was defined as immediate closure of the drain, regardless of starting EVD height in two of the three rapid wean centers (Supplemental Data). A rapid wean was defined as closure of the drain within 48 h of an attempted wean trial in one of the three rapid wean centers. Patients were analyzed as part of their predetermined rapid or gradual EVD wean group in an intention-to-treat like manner. We also performed separate, secondary analyses based on the as treated EVD wean. As treated wean approaches were determined based on data collected from the daily EVD collection tool. For the purposes of the as treated analysis, a rapid wean was defined as clamping of the drain within 48 h of starting a trial of wean. A wean trial failure was defined as follows: EVD reopening or halting of a wean trial due to (1) increased ICP (typically sustained > 25 mm Hg for > 5 min), clinical worsening, or acute hydrocephalus, (2) development of a CSF leak, or (3) repeat computed tomography scan that showed worsening hydrocephalus. All the participating centers allowed for multiple wean trial attempts prior to placement of a VPS. Following a wean trial failure, it was left to the treating clinical team whether to plan for VPS placement or allow for additional wean trials. Institutional guidelines suggested multiple weans (typically up to two or three failed wean trials) prior to considering placement of a VPS (Supplemental Data).

Study Sample

Inclusion criteria were adult patients ≥ 18 years old with suspected aneurysmal SAH requiring an EVD during their ICU stay. All participating centers placed an EVD only for a clinical suspicion of acute hydrocephalus. Centers did not place EVDs prophylactically and did not place EVDs solely for monitoring ICP or to optimize cerebral perfusion pressure for vasospasm management. EVDs were not cared for outside of the ICU. There were 185 patients who met inclusion criteria and were enrolled in the study (Fig. 1). Of these, 46 patients were excluded: (1) 25 patients were made comfort measures only or died in the ICU prior to initiation of a wean trial (four of these patients also never had a culprit lesion found), (2) one patient had an ICU course complicated by an EVD-related epidural hematoma on insertion that required surgical treatment, and (3) there was no culprit aneurysm found in 24 patients. This left 66 patients in the gradual wean group and 73 patients in the rapid wean group.

Fig. 1.

Fig. 1

Flowchart of enrolled rapid versus gradual wean protocol patients. Italics indicate patients who were excluded from analysis. Patients in the gradual and rapid wean groups were analyzed in an intention-to-treat like manner. VPS were placed acutely in patients who received a VPS during their initial hospitalization. A VPS was placed in a delayed manner in patients who were discharged from the ICU and required an unplanned VPS after the EVD was discontinued. *There were four patients categorized as CMO or who died while in the ICU and who were angio negative such that a symptomatic aneurysm was never found. There was one patient excluded for having an EVD-associated epidural hematoma requiring a craniotomy. angio, angiogram, CMO, comfort measures only, EVD, external ventricular drain, ICU, intensive care unit, VPS, ventriculoperitoneal shunt

Measurements

Baseline, daily EVD, and discharge variables were predefined and recorded prospectively using a REDCap database [12, 13]. Each patient was followed up at 3 months by chart review to determine whether a delayed VPS was placed. A version of the collection tool is included in the Supplemental Data.

Statistical Analysis

The overall approach was to equate the groups statistically on potentially relevant preexisting confounding covariates. A power calculation for the primary outcome based on unadjusted VPS rates (35% gradual and 13% rapid) from a recent retrospective study [14] estimated that approximately 60 patients per gradual or rapid EVD wean group—with a crossover rate of 36% and 8%, respectively—were needed to detect differences in VPS rate for ≥ 80% power with α = 0.05. We did not perform a power calculation on secondary outcomes. The main findings were analyzed with an intention-to-treat-like approach.

We assessed several covariates to include in our analysis by building forward stepwise logistic regression models. EVD protocol was the primary variable of interest. Age and sex were common demographic baseline factors thought to influence outcomes in SAH [1]. We included the number of EVD weans and time to first EVD wean, given that recent work suggests these factors influence VPS rates and LOS [1416]. We evaluated the GRAEB score [17], which is a scale of extra-axial blood previously described as an independent predictor of VPS dependence following SAH [18], and found that it behaved linearly and had an independent predictive value for likelihood of VPS placement. We also evaluated the related modified GRAEB score [19], but the original GRAEB had greater predictive value for our cohorts. We considered using the radiographic modified Fisher score, but this scale did not behave linearly, and the degree of intraventricular hemorrhage was already accounted for by the GRAEB score. The final adjusted logistic regression model included age, sex, GRAEB score, time to first wean trial, number of EVD wean trials, and EVD protocol (rapid vs. gradual). We used the same covariates for inclusion in Cox proportional hazards and general linear model regressions.

SAS (Version 9.4; SAS Institute), JMP (Version 15.0.0; SAS Institute), and Prism 8 (GraphPad Software Inc) were used for statistical analysis and figure preparation. Wilcoxon and Kruskal–Wallis rank sums tests were used to determine significance for nonnormally distributed variables (e.g., number of wean attempts). Student’s t-test was used for continuous variables with normal distributions. χ2 testing was performed to compare categorical variables.

Results

Baseline demographics and severity scores were similar between rapid and gradual wean protocol groups (Table 1). The exceptions were fewer posterior inferior cerebellar artery aneurysms and a greater number of attempted weans for the rapid EVD protocol group. Crossover from gradual to rapid was 21%. Crossover from rapid to gradual was 4%. We also determined baseline differences between anonymized centers and found significant differences in race, GRAEB scores, and number of posterior inferior cerebellar artery aneurysms (Supplemental Table 1). There were also significant differences in the timing of the first wean trial, number of weans, and (as expected) as treated EVD protocol.

Table 1.

Baseline demographics

Demographics Gradual wean (n = 66) Rapid wean (n = 73) p
Age, mean ± SD (yr)a 55.0 ± 14.3 56.8 ± 12.6 0.44
Sex, n (%) 0.85

   Male 18 (27) 18 (25)

   Female 48 (73) 55 (75)

Race, n (%) 0.25

   White 48 (73) 57 (78)

   Black 7 (11) 9 (12)

   Asian American 2 (3) 3 (4)

   Other 6 (9) 4 (5)

   Unknown 3 (5) 0

Ethnicity, n (%) 0.59

   Hispanic 14 (21) 12 (16)

   Non-Hispanic 50 (76) 60 (82)

   Unknown 2 (3) 1 (1)

Hunt and Hess score, n (%) 0.16

   1 6 (9) 1 (1)

   2 13 (20) 22 (30)

   3 29 (44) 27 (37)

   4 12 (18) 15 (21)

   5 6 (9) 8 (11)

WFNS score, n (%) 0.24

   1 16 (24) 23 (32)

   2 24 (36) 15 (21)

   3 2 (3) 4 (5)

   4 17 (26) 18 (25)

   5 7 (11) 13 (18)

GCS, mean ± SD 11.4 ± 4.1 11.3 ± 4.0 0.93
GRAEB, mean ± SD 3.8 ± 3.1 3.7 ± 3.1 0.72
Modified GRAEB, mean ± SD 10.8 ± 8.1 9.1 ± 8.0 0.20
Treatment, n (%) 0.42

   Clipping 17 (26) 14 (19)

   Coiling 47 (71) 54 (74)

   Other 2 (3) 5 (7)

EVD placement complications, n (%) 3 (5) 1 (1) 0.26
mF, n (%) 0.42

   1 5 (8) 3 (4)

   2 7 (11) 4 (5)

   3 28 (42) 39 (53)

   4 26 (39) 27 (37)

Aneurysm location, n (%)

   ICA 6 (9) 8 (11) 0.71

   MCA 7 (11) 14 (19) 0.15

   ACA 5 (8) 6 (8) 0.89

   Vertebral 1 (2) 0 0.22

   Basilar 10 (15) 5 (7) 0.11

   PCA 2 (3) 4 (5) 0.47

   Pcomm 14 (21) 14 (19) 0.77

   PICA 6 (9) 1 (1) 0.03

   AComm 17 (26) 20 (27) 0.83

   Other 1 (2) 2 (3) 0.62

   Angio negative 0 0

Aneurysm size, mean ± SD (mm) 6.2 ± 3.6 6.0 ± 2.7 0.67

First wean trial day, mean ± SD 9.9 ± 5.2 8 ± 5.7 0.38

Number of weans, median (IQR) 1 (1–2) 1 (1–2.5) 0.007

ACA anterior cerebral artery, AComm anterior communicating artery, EVD external ventricular drain, GCS Glasgow coma scale, ICA internal carotid artery, IQR interquartile range, MCA middle cerebral artery, mF modified Fisher scale, PCA posterior cerecral artery, Pcomm posterior communicating artery, PICA posterior inferior cerebellar artery, SD standard deviation, WFNS world federation of neurological surgeons

a

Age range was 19–87 years (min–max). There were three patients aged > 80 years

Incidence of VPS Placement

The primary outcome measure was incidence of VPS placement, analyzed in an intention-to-treat-like manner. A VPS was placed in 39% (26 of 66) of gradual and 32% (23 of 73) of rapid protocol patients (Fig. 1 and Supplemental Table 2). This included VPS placed acutely while the patient was still in the hospital (n = 33) and in a delayed manner after the patient had been discharged (n = 16). No unadjusted outcomes reached the threshold of p < 0.05 for statistical significance (Supplemental Table 2). We tabulated unadjusted outcomes by center and observed no appreciable differences in VPS placement by center (Supplemental Table 3). We observed differences in the number of EVD tract hemorrhages, nonfunctioning EVDs, posthospital disposition, number of EVD days, and ICU LOS. Because the majority of centers took a single EVD wean protocol approach, these differences could be due to the assigned EVD protocol rather than characteristics specific to an individual institution.

A logistic regression model (see Methods) demonstrated that a rapid EVD wean was associated with a decreased odds ratio (OR) of requiring a VPS up to 90 days after placement (OR 0.43 [95% confidence interval {CI} 0.18–1.03], p = 0.057) (Supplemental Table 4). There was no appreciable difference in the placement of late, posthospital discharge VPS (OR 1.46 [95% CI 0.47–4.53] p = 0.52). Early VPS placement while the patient was still in the hospital was less frequent in the rapid group (OR 0.19 [95% CI 0.061–0.62] p = 0.0056).

Data were further analyzed to assess the relationship between wean protocol and time to placement of a VPS up to 90 days by carrying out an adjusted Cox proportional hazards analysis (Fig. 2). We found a difference between the rapid and gradual EVD wean protocols in VPS-free interval with a hazard ratio of 0.52 (p = 0.041) favoring a rapid EVD wean. Our model estimated that 69% of patients in gradual and 83% in rapid wean groups remain VPS-free by day 90.

Fig. 2.

Fig. 2

Rapid EVD wean trials are associated with fewer VPS. Adjusted Cox proportional hazards model for VPS placement in patients undergoing rapid vs. gradual weans. EVD, external ventricular drain, VPS, ventriculoperitoneal shunts

An association between the number of wean trial attempts and the rate of VPS placement was noted. More wean trial attempts were associated with a greater likelihood of total VPS placement (95% CI, OR 2.9 [1.8–4.8] p < 0.0001) (Supplemental Table 4). Timing of the first wean trial was also associated with whether a VPS was placed. The longer it took until the first trial of wean, the more likely it was for patients to receive a VPS at any time (OR 1.13 [95% CI, 1.03–1.23] p = 0.0067) (Supplemental Table 4).

LOS

Data were analyzed to determine the effect of assigned wean protocol, number of wean trials, and time to the first wean trial on EVD duration and ICU and hospital LOS in a general linear regression model (Fig. 3). A rapid EVD wean was associated with 2.09 ± 0.76 (mean ± standard error of the mean) fewer EVD days (p = 0.0066) and 2.49 ± 1.25 fewer ICU days, compared with a gradual protocol (p = 0.049) (Fig. 3a). There was no appreciable effect on hospital LOS.

Fig. 3.

Fig. 3

Rapid, less frequent, and early EVD wean trials are associated with shorter EVD duration and shorter ICU LOS. Adjusted estimates for number of days saved in EVD duration and ICU or hospital LOS in relation to rapid vs. gradual wean approaches a, frequency of wean trials b, and the timing to the first trial of EVD wean c. Statistics are for the null hypothesis being no difference between wean protocol, fewer or greater number of weans, and earlier or later weans. EVD, external ventricular drain, ICU, intensive care unit, LOS, length of stay. *p < 0.05, **p < 0.01, ****p < 0.0001

There was a large effect of frequency of wean trials on all duration measures. For each attempted EVD wean trial, the model predicted 4.07 ± 0.39 additional EVD days (p < 0.0001), 3.61 ± 0.65 additional ICU days (p < 0.0001), and 4.71 ± 1.99 additional hospital days (p = 0.020) (Fig. 3b).

There was an effect of timing to the first wean trial on EVD duration and ICU LOS. For each additional day that passed prior to the first trial of wean, the model predicted 0.88 ± 0.074 additional EVD days (p < 0.0001) and 0.89 ± 0.12 additional ICU days (p = < 0.0001). The model also predicted 0.68 ± 0.38 additional hospital days for each day prior to the first wean, but this finding did not meet the predetermined threshold for statistical significance (p = 0.07) (Fig. 3c).

EVD Complications

Next, the effect of wean protocol, number of weans, and timing to first wean on select EVD-associated complications were determined in an adjusted logistic regression model (Supplemental Fig.). EVD tract hemorrhage that became apparent in the days following EVD placement did not emerge as an outcome associated with any of the examined approaches. A rapid wean protocol was associated with fewer nonfunctioning (i.e., clogged) EVDs with OR 0.32 (95% CI 0.11–0.92) (Supplemental Fig. A). An increased number of wean attempts was associated with a greater likelihood of a nonfunctioning EVD, with an OR 1.67 (95% CI 1.05–2.64) (Supplemental Fig. B). There was no detectable relationship between timing to first wean trial on EVD tract hemorrhage or incidence of a non-functioning EVD (Supplemental Fig. C).

As Treated Analysis

The above analyses were repeated using as treated EVD wean approaches to supplement the primary intention-to-treat-like analysis (Supplemental Table 5). In general, effect sizes for VPS placement and LOS were increased and the effect on complications decreased.

Longitudinal EVD Status

Finally, the daily EVD status (height at which the EVD was set to drain, as an example 10 cm H2O above the tragus, and whether the EVD was primarily open or closed) for every analyzed patient was examined (Fig. 4), organized by wean protocol and whether a VPS was placed at any time. EVDs were not always started at the same level, and their status was variable across patients. Review of individual EVD time courses confirmed many of the findings seen on unadjusted and adjusted analyses, namely, that rapid EVD weans are associated with fewer VPSs and shorter EVD durations.

Fig. 4.

Fig. 4

Individual EVD height and closure status trajectories. EVD status is organized by preassigned wean protocol and whether a VPS was placed. EVD, external ventricular drain, VPS, ventriculoperitoneal shunts

Discussion

The optimal management of the EVD in the setting of aneurysmal SAH has been controversial. We found in the studied cohort from multiple centers that a rapid EVD wean approach is associated with fewer VPS placements. A rapid EVD wean was also associated with shorter EVD duration and shorter ICU LOS, but we did not observe a statistically significant signal for hospital LOS. Furthermore, we observed that the more wean trials that are attempted or the longer it takes to perform a first wean trial, the more likely the patient will require a permanent VPS and the longer the patient will remain in the ICU and hospital.

This study is the only prospective multicenter study of EVD management approaches in patients with aneurysmal SAH [3, 58]. One single center trial randomly assigned patients to a rapid or gradual wean protocol and found a reduced ICU LOS in the rapid wean group [20]. However, the investigators did not carry out an adjusted statistical analysis and only 1 EVD wean trial failure was tolerated prior to placing a VPS. Another single center study took a retrospective before-and-after approach and found that after switching from a gradual to rapid EVD weaning protocol that a rapid wean was associated with fewer VPS placements [14]. However, a recent two-center retrospective study found that a gradual wean was associated with fewer VPS placements [21]. Surveys of individuals and institutions have demonstrated either a lack of a consensus approach or a trend toward a gradual wean approach, which has a similar degree of supportive evidence as a rapid wean approach [10, 2224]. We therefore designed the current study to address potential confounders inherent in prior studies while assessing both VPS rate and LOS.

There is a distinct and important EVD management paradigm not directly addressed in our study. An additional randomized study compared continuous versus intermittent CSF drainage and found that intermittent drainage was associated with fewer complications, driven by nonpatent EVDs [25]. Continuous drainage refers to keeping the EVD open by default. Intermittent (on demand) drainage refers to maintaining the EVD closed by default and only draining CSF when ICP reaches a set threshold or when the patient becomes symptomatic. EVD wean trials with continuous drainage can be gradual or rapid, whereas intermittent drainage is effectively a rapid EVD wean trial performed repeatedly [8]. In the current study, we found that a rapid wean protocol was associated with fewer nonpatent EVDs. This is consistent with findings in the intermittent management group in the randomized trial, but the two studies are not directly comparable. Therefore, future prospective multicenter trials involving comparisons between intermittent and continuous EVD management could be helpful in illuminating the effect of EVD management on drain complications.

For the current study, we focused on the relationship between EVD wean protocol and VPS placement. We interpret the whole of our results as demonstrating a clinically important effect of a rapid EVD wean associated with fewer VPS. Unadjusted and adjusted logistic regression analyses of EVD wean protocol did not reach the predefined threshold of p < 0.05 for statistical significance for total VPS placement, most likely because of study power [26]. However, Cox proportional hazards testing revealed an effect favoring a rapid wean that met the threshold for significance. The difference in effective statistical power may have been due to inclusion of longitudinal data in the Cox proportional hazards procedure.

This study was not designed to address how an EVD wean protocol might lead to decreased VPS placement. One proposed physiological explanation is that higher relative pressures drive flow through arachnoid granulations and other pathways of CSF efflux such that natural resorption mechanisms may be reprimed. Another theory involves the effect of ICP on the rate of CSF secretion by the choroid plexus [14]. There might also be a potential nonphysiological reason for the observed association between wean protocol and VPS rate. Clinicians who perform a rapid wean might also tolerate a degree of asymptomatic, radiographic hydrocephalus and are less likely to place a shunt. Likewise, clinicians who perform a gradual wean might tend to believe that low pressure hydrocephalus can have detrimental effects on cognition and are more likely to place a VPS. If these two schools of thought are prevalent, then the relationship between wean protocol and VPS placement may only reflect physician practice and not physiology. Questions raised by the above can only be addressed with additional, carefully designed future preclinical and clinical studies.

Our study was not a strictly observational study. We assigned patients to EVD wean group (rapid or gradual) prior to their treatment and analyzed the results in an intention-to-treat-like manner to avoid selection bias. There could have been bias introduced, given the absence of blinding, but each center voluntarily chose their wean approach in the belief that their approach was the best. Therefore, treatment bias would likely have been equally distributed among the centers. As such, we would argue that the level of evidence our study represents lies somewhere between traditional prospective randomized and prospective observational studies.

We did not disclose results from each individual center. The study was not designed to determine interinstitutional differences, and we wanted to avoid drawing inferences that may have had a high likelihood of being due to low sample size or chance alone. We assumed that each of the tertiary care neurological ICUs provided equally high-quality clinical care. Regardless, there may have been institutional bias and culturally delineated differences in care that is not measurable between centers, though we performed some preliminary tests of differences between centers within each treatment condition (gradual and rapid) on VPS placement and found nonsignificant results (Supplemental Data).

Our study has the following additional limitations: Patients were preassigned to rapid or gradual EVD weans, but the actual way this was carried out may have differed by center. A common operational definition of acute hydrocephalus was not provided for each center. We did not quantify barriers to transfer from the ICU or from the hospital, which could have led to outsize effects on the positive ICU LOS findings or have led to decreased effective power for hospital LOS. We did not have the central adjudication necessary to address the fair-to-moderate interrater reliability in diagnosing delayed cerebral ischemia [27] and so were unable to include it as an outcome measure. The study size limited the number of factors we could include in the multivariable analysis. There is a lack of long-term functional outcome data.

Conclusions

The optimal EVD management approach may involve a rapid EVD discontinuation strategy of prompt drain closure, prioritizing an early time to the first trial of wean, and expeditious placement of a VPS after the patient has proved to be a difficult wean. Future randomized multicentered studies will be critical to address persistent uncertainties in EVD management after SAH.

Supplementary Material

supplemental material

Acknowledgements

The authors thank Alexa Collins and Sara Gray (UTSW) for assistance with data collection and Matthew Cobler-Lichter, Joanna Yang, and Mabel Chung for helpful comments. Special thanks to Marek Mirski for helpful comments on the manuscript.

Source of support

Dr. Chung and Dr. Locascio have received support from the National Institutes of Health (R25NS065743, KL2TR002542, and K08NS112601 to DYC and UL1TR001102 to JJL). This work was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Advancing Translational Sciences, National Institutes of Health Award UL1TR002541) and financial contributions from Harvard University and its affiliated academic health care centers. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health. Dr. Chung has also received support from the American Heart Association and American Stroke Association (18POST34030369), the Andrew David Heitman Foundation, the Aneurysm and AVM Foundation, and the Brain Aneurysm Foundation’s Timothy P. Susco and Andrew David Heitman Foundation Chairs of Research.

Footnotes

Conflicts of interest

The authors declare no conflicts of interest.

Ethical approval/informed consent

We adhered to ethical guidelines. Ethical approvals (institutional review board) were obtained. Use of informed consent was not applicable. Prospective observational studies require a statement regarding institutional review board approval, which we have obtained, as stated in the article.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s12028-021-01343-9.

References

  • 1.Connolly ES Jr, Rabinstein AA, Carhuapoma JR, et al. Guidelines for the management of aneurysmal subarachnoid hemorrhage: a guideline for healthcare professionals from the American Heart Association/american Stroke Association. Stroke. 2012;43(6):1711–37. [DOI] [PubMed] [Google Scholar]
  • 2.Maher M, Schweizer TA, Macdonald RL. Treatment of spontaneous subarachnoid hemorrhage: guidelines and gaps. Stroke. 2020;51(4):1326–32. [DOI] [PubMed] [Google Scholar]
  • 3.Chung DY, Olson DM, John S, et al. Evidence-based management of external ventricular drains. Curr Neurol Neurosci Rep. 2019;19(12):94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Chung DY, Abdalkader M, Nguyen TN. Aneurysmal subarachnoid hemorrhage. Neurol Clin. 2021;39(2):419–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Gigante P, Hwang BY, Appelboom G, et al. External ventricular drainage following aneurysmal subarachnoid haemorrhage. Br J Neurosurg. 2010;24(6):625–32. [DOI] [PubMed] [Google Scholar]
  • 6.Fried HI, Nathan BR, Rowe AS, et al. The insertion and management of external ventricular drains: an evidence-based consensus statement : a statement for healthcare professionals from the Neurocritical Care Society. Neurocrit Care. 2016;24(1):61–81. [DOI] [PubMed] [Google Scholar]
  • 7.Capion T, Lilja-Cyron A, Juhler M, Mathiesen TI, Wetterslev J. Prompt closure versus gradual weaning of extraventricular drainage for hydrocephalus in adult patients with aneurysmal subarachnoid haemorrhage: a systematic review protocol with meta-analysis and trial sequential analysis. BMJ Open. 2019;9(10):e029719. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Chung DY, Mayer SA, Rordorf GA. External Ventricular drains after subarachnoid hemorrhage: is less more? Neurocrit Care. 2018;28(2):157–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Merkler AE, Ch’ang J, Parker WE, Murthy SB, Kamel H. The rate of complications after ventriculoperitoneal shunt surgery. World Neurosurg. 2017;98:654–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Chung DY, Leslie-Mazwi TM, Patel AB, Rordorf GA. Management of external ventricular drains after subarachnoid hemorrhage: a multi-institutional survey. Neurocrit Care. 2017;26(3):356–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Rao SS, Chung DY, Wolcott Z, et al. Intermittent CSF drainage and rapid EVD weaning approach after subarachnoid hemorrhage: association with fewer VP shunts and shorter length of stay. J Neurosurg. 2019; pp. 1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Ascanio LC, Gupta R, Adeeb N, et al. Relationship between external ventricular drain clamp trials and ventriculoperitoneal shunt insertion following nontraumatic subarachnoid hemorrhage: a single-center study. J Neurosurg. 2018;130(3):956–62. [DOI] [PubMed] [Google Scholar]
  • 16.Perry A, Graffeo CS, Kleinstern G, et al. Quantitative modeling of external ventricular drain output to predict shunt dependency in aneurysmal subarachnoid hemorrhage: cohort study. Neurocrit Care. 2020;33(1):218–29. [DOI] [PubMed] [Google Scholar]
  • 17.Graeb DA, Robertson WD, Lapointe JS, Nugent RA, Harrison PB. Computed tomographic diagnosis of intraventricular hemorrhage. Etiology and prognosis Radiology. 1982;143(1):91–6. [DOI] [PubMed] [Google Scholar]
  • 18.Czorlich P, Ricklefs F, Reitz M, et al. Impact of intraventricular hemorrhage measured by Graeb and LeRoux score on case fatality risk and chronic hydrocephalus in aneurysmal subarachnoid hemorrhage. Acta Neurochir (Wien). 2015;157(3):409–15. [DOI] [PubMed] [Google Scholar]
  • 19.Morgan TC, Dawson J, Spengler D, et al. The modified Graeb score: an enhanced tool for intraventricular hemorrhage measurement and prediction of functional outcome. Stroke. 2013;44(3):635–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Klopfenstein JD, Kim LJ, Feiz-Erfan I, et al. Comparison of rapid and gradual weaning from external ventricular drainage in patients with aneurysmal subarachnoid hemorrhage: a prospective randomized trial. J Neurosurg. 2004;100(2):225–9. [DOI] [PubMed] [Google Scholar]
  • 21.Jabbarli R, Pierscianek D, Rolz R, et al. Gradual external ventricular drainage weaning reduces the risk of shunt dependency after aneurysmal subarachnoid hemorrhage: a pooled analysis. Oper Neurosurg (Hagerstown). 2018;15(5):498–504. [DOI] [PubMed] [Google Scholar]
  • 22.Allan D Intracranial pressure monitoring: a study of nursing practice. J Adv Nurs. 1989;14(2):127–31. [DOI] [PubMed] [Google Scholar]
  • 23.Olson DM, Lewis LS, Bader MK, et al. Significant practice pattern variations associated with intracranial pressure monitoring. J Neurosci Nurs. 2013;45(4):186–93. [DOI] [PubMed] [Google Scholar]
  • 24.Olson DM, Batjer HH, Abdulkadir K, Hall CE. Measuring and monitoring ICP in neurocritical care: results from a national practice survey. Neurocrit Care. 2014;20(1):15–20. [DOI] [PubMed] [Google Scholar]
  • 25.Olson DM, Zomorodi M, Britz GW, et al. Continuous cerebral spinal fluid drainage associated with complications in patients admitted with subarachnoid hemorrhage. J Neurosurg. 2013;119(4):974–80. [DOI] [PubMed] [Google Scholar]
  • 26.Amrhein V, Greenland S, McShane B. Scientists rise up against statistical significance. Nature. 2019;567(7748):305–7. [DOI] [PubMed] [Google Scholar]
  • 27.Schupper AJ, Eagles ME, Neifert SN, Mocco J, Macdonald RL. Lessons from the CONSCIOUS-1 study. J Clin Med. 2020;9(9):2970. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplemental material

RESOURCES