The present cohort was extracted from a single-center data set of aSAH patients. The assessment, treatment, and follow-up were performed according to center-specific standard procedures for patients with aSAH. Institutional Review Board (IRB) and local ethics committee (EKNZ, Basel, Switzerland) approval was obtained and accredited (Project-ID 2018–02129). The ethics committee waived the requirement for written informed consent (justification: disproportionality) for patients recruited prior to 2014, when the new Swiss Human Research Act went in force. As of 2014, written informed consent was obtained from all participants.
Study design: This is a cohort study retrospectively collected in the year 2009 and prospectively collected in the years 2010–2018. The dataset was anonymized consecutively for analysis. Because of the purely observational nature of the study, no registration in a trial registry was necessary.
Study center: All patients within the present data set were treated at the University Hospital of Basel, an accredited tertiary neurovascular center.
Study population: Data were collected from all patients admitted to the study center with aSAH from a documented ruptured intracranial aneurysm. Patients with non-aneurysmal SAH and patients who died on the day of admission were excluded.
Data collection: A set of variables of interest was predefined. These variables were from hospital charts and pooled in a secured, local registry. A recollection of EVD-related variables and infectious complications as well as CSF-laboratory investigations was conducted in 2016.
Study variables: For the present study, the following variables were extracted from the local data set: patient characteristics (age, sex); aSAH specific admission scores (Glasgow Coma Scale [GCS] [29], World Federation of Neurosurgical Societies [WFNS] score [30]); radiological characteristics (Fisher grade [31], Barrow Neurological Institute (BNI) scale [32], presence of intraventricular or intracerebral hemorrhage, presence of acute hydrocephalus [occurring within 72 hours after ictus]) [16]; aneurysm characteristics (location of the ruptured aneurysm and maximal aneurysm diameter treatment modality [open surgical clipping or endovascular treatment]); information on temporary and permanent CSF diversion procedures (time point of implantation, duration of external drainage, weaning period, time point of conversion to a permanent CSF diversion); infectious complications (EVDAI); laboratory parameters of infections such as leucocyte count, proteins, and hemoglobin measured in CSF and blood); information on DCVS and DCI (time of onset, clinical or radiological manifestation, treatment); outcome variable (modified Rankin scale [mRS] [33] and in-hospital death); and duration of hospital stay.
The indication for external ventricular drainage and the timing of EVD weaning were at the treating neurosurgeon’s discretion. Generally, the weaning process consisted of incremental raising of the extracorporeal CSF diversion system for 48 hours, followed by drain closure for 24 hours. Unsuccessful weaning was defined as either clinical manifestation of hydrocephalic symptoms or radiological demonstration of ventricular enlargement on a CT scan within this period.
Cases of EVDAI were confirmed in accordance with the modified criteria for nosocomial infections of the Centers for Disease Control and Prevention (CDC) [34]. A positive culture of CSF or a ventricular catheter tip without other abnormal CSF findings or clinical signs suggesting infection was designated as colonization [9, 34]. EVDAI was diagnosed if a positive culture of the CSF or catheter tip was present and accompanied by further pathological CSF findings or at least one new sign or symptom of CNS infection in the absence of any other known cause [9, 34]. EVDAI was also diagnosed in some cases with negative cultures in which there were pathological CSF findings and/or clinical manifestations of infection and antimicrobial therapy was given [9, 34].
DCVS was defined as local vasospasm visible on computed tomographic (CT) angiography or digital subtraction angiography accompanied by neurological worsening for at least 2 hours [35]. DCI was defined according to the multidisciplinary research group criteria [36] as brain infarction visible on CT or magnetic resonance imaging in the first 6 weeks after SAH.
Primary and secondary endpoints: The primary endpoint was the occurrence of EVDAI. The secondary endpoints were the occurrence of symptomatic or DCVS, the presence of a stroke on the last CT obtained before hospital discharge defined as DCI, and a poor mRS score (4–6) at discharge, or in-hospital death.
Follow-up: To assure the availability and consistency of data for statistical analysis, we only collected standardized variables (occurrence of EVDAI and DCVS) during hospitalization. Unambiguous time points were defined, e.g. the day before a permanent CSF diversion procedure (CSF sampling) and patient discharge (DCI, mRS, in-hospital death). A detailed patient inclusion profile including the number of participants with available data at each time point is provided in Fig. 1.
Statistical analysis: We provide descriptive statistics for a set of predefined variables of interest by reporting means and 95% confidence intervals, or median and interquartile ranges for continuous variables, as appropriate. For categorical variables, counts and percentages were reported. The patients were divided into two groups: the “early group,” in which a permanent CSF diversion procedure was performed ≤ 14 days after the ictus, and the “late group,” in which it was performed > 14 days after the ictus. To obtain the association between the outcome of interest and important variables (known to be associated with temporary and permanent CSF diversion or influence defined outcomes), the logistic regression model was used. The variables used in the univariable model(s) are dichotomized age („higher age“ [age ≥ 60] vs. „lower age“ [age < 60]), BNI scale (“high BNI scale” [BNI 4–5] vs. “low BNI scale” [BNI 1–3]), WFNS (“low WFNS” [WNFS 1–3] vs. “high WFNS [WNFS 4–5]), and mRS (“Good mRS” [mRS 1–3] vs. “Poor mRS” [mRS 4–6]). The duration of external ventricular drainage (in days) was used as a continuous variable. Risk factors that had more than 10% increase or decrease in OR (OR ≤ 0.9 or ≥ 1.10) and which were considered clinically relevant for the endpoints of interest were included in the multivariable model (see additional Tables I-III; Additional file 1). Due to the small size of this cohort and also the low number of events, a maximum of three variables was included in the multivariable model(s). The duration of external ventricular drainage and late permanent CSF diversion are not mutually exclusive variables and were analyzed separately. Measures of accuracy were calculated for the sample estimates with bootstrapping statistics. Confidence intervals were calculated with the profile likelihood method based on the Wald test statistic. Statistical significance was set at P ≤ .05. Statistical analyses were performed in SPSS (IBM SPSS Statistics 22, 2013, New York, USA).