External ventricular drainage is frequently used in neurocritical care, particularly in patients admitted for non-traumatic subarachnoid hemorrhage who develop hydrocephalus and/or intracranial hypertension. While external ventricular drainage is often initially lifesaving, its prolonged maintenance is associated with complications, especially infections and prolonged hospital length of stay. There is currently no consensus on the optimal weaning strategy. Two approaches are used in routine practice: direct clamping (the external ventricular drain is closed as soon as weanability criteria are met) and gradual weaning (the external ventricular drain level is progressively raised before final clamping). No randomized controlled trial has yet demonstrated the superiority of one strategy over the other in patients with non-traumatic subarachnoid hemorrhage.

The investigators hypothesize that a direct clamping strategy, combined with daily screening of standardized weanability criteria, will reduce the duration of external ventricular drain maintenance compared with the conventional gradual weaning strategy. SEVDVE-2 is a multicenter, randomized, controlled, parallel-group, single-blind superiority trial that will compare these two weaning strategies in 170 adult patients admitted to critical care for non-traumatic subarachnoid hemorrhage with a first external ventricular drain inserted within the previous 3 days. Patients will be randomized 1:1, stratified on the presence of an intraventricular hematoma. The primary outcome is the number of external ventricular drain-free days alive at Day 28.

SEVDVE-2 is a multicenter, randomized, controlled, parallel-group, single-blind superiority trial conducted in French university hospital neurocritical care units (research involving routine-care interventions).

Background and rationale: External ventricular drainage is commonly used in patients with non-traumatic subarachnoid hemorrhage to manage hydrocephalus and/or intracranial hypertension. Prolonged external ventricular drain maintenance is associated with infectious and other complications and prolongs hospital stay. There is no recommendation on the optimal weaning strategy. A retrospective multicenter cohort (SEVDVE) by the investigators showed heterogeneous practices (61% gradual weaning, 39% direct clamping) and suggested that direct clamping shortens the external ventricular drain maintenance duration but is associated with more clamping failures. The investigators have developed standardized daily weanability criteria to better define the timing of the weaning attempt.

Intervention: Eligible adult patients hospitalized in critical care for non-traumatic subarachnoid hemorrhage with a first external ventricular drain in place for less than 3 days are randomized 1:1 (stratified on the presence of intraventricular hematoma) to one of two strategies:

• Direct clamping (experimental arm): from Day 4, daily screening of weanability criteria (no episode of intracranial hypertension for the previous 24 hours, minimal sedation, external ventricular drain output <200 ml/24h with a 3-hour intracranial pressure tolerance test or <160 ml/24h without test). When criteria are met, the external ventricular drain is directly clamped for 48 hours.
• Gradual weaning (control arm): when the patient improves (neurological improvement for at least 48 hours, no intracranial hypertension), the external ventricular drain level is raised by 5 mmHg per day. When the external ventricular drain level is ≥20 mmHg and tolerated for 24 hours, the external ventricular drain is clamped for 48 hours.

In both arms, the external ventricular drain is removed after 48 hours of clamping if no neurological deterioration, intracranial hypertension, cerebrospinal fluid leak, or ventricular enlargement occurs. If clamping fails, drainage is resumed and a new attempt is made under the same strategy, within the 28 days after the start of weaning. After Day 14, internalization (ventriculoperitoneal or ventriculoatrial shunt) is to be discussed after two clamping failures.

Patients are blinded to their randomization group. Care providers, investigators, and outcome assessors cannot be blinded because the procedure involves device handling.

Study visits: V1 inclusion visit, daily follow-up between V1 and V2, V2 at hospital discharge or Day 28 ± 3 days, V3 telephone interview at Day 90 ± 7 days. All collected data are those routinely recorded during standard care.

Sample size: 170 patients (85 per arm), based on an expected difference of 3 EVD-free days at Day 28 (13 vs 10), standard deviation 6 days, alpha 5%, power 90%.

Inclusion Criteria:

• Adult patient (≥18 years)
• Admitted to critical care for non-traumatic subarachnoid hemorrhage for less than 3 days
• First external ventricular drain inserted within the last 3 days for hydrocephalus and/or intracranial hypertension
• Patient consent, or consent from a relative, or inclusion under emergency inclusion procedure
• Patient affiliated to or beneficiary of a social security scheme

Exclusion Criteria:

• Moribund patient or patient with established treatment limitation/withdrawal decisions
• Patient with a pre-existing ventriculoperitoneal or ventriculoatrial shunt
• Patient with chronic hydrocephalus
• Pregnant, lactating, or parturient woman
• Person deprived of liberty by judicial or administrative decision
• Person under involuntary psychiatric care
• Person under a legal protection measure
• Concurrent participation in another study involving external ventricular drainage management

Data will be shared upon reasonable request. Only de-identified data will be shared. Any data collected during the study may be shared. The protocol will be shared initially. Other documents may be shared at a later date upon request (e.g., the CRF to allow a collaborator to select the data they wish to access). The recipients of the data will be researchers. The data will be available for any purpose deemed relevant by the study investigator, based on a protocol provided by the requester, after verification of the obtaining of regulatory approvals, including the favorable opinion of an ethics committee.

The data will be shared after signing a negotiated data transfer agreement ( data access agreement), for the duration specified in the agreement.

The data will be made available via secure transfer (sharing platform approved by the university hospital: BlueFiles or Oodrive).

Single-blind design (participant). Because the study evaluates two strategies for managing a medical device (external ventricular drain), it is not feasible to blind the care providers, investigators, or outcome assessors. However, participants are not informed of their randomized group and have no means to identify it during the study.