External ventricular drains (EVD) are small tubes used in neuro-critical care inserted to measure pressure and treat acute build-up of fluid in the brain by draining the cerebrospinal fluid (CSF) in the ventricles, often following an event of traumatic or spontaneous bleeding. While essential to the care of these patients, EVDs run the risk of introducing bacteria into the brain of the patient, causing an EVD associated infection (EVDI). EVDIs are feared complications that are difficult to identify and predict in an intensive care setting. In order to allow for early identification of these infections, CSF is routinely sampled from the EVDs and its constitution analyzed for signs of infection. However, the constitution of the CSF in neuro-critical care patients are often difficult to assess as it is frequently mixed with blood that often clouds clinical decision making. No fast parameter has been found to yet reliably predict or identify these infections, resulting in excessive treatment with broad-spectrum antibiotics in this patient group.

EVDI diagnostics rely on mainly CSF analyses and cultures (growth of bacteria in the laboratory). Growing bacteria in the lab may take many days and can seldom guide early decision-making for these infections. Thus, EVDI diagnostics mainly rely on the analysis of the CSF constitution. Many diagnostic criteria rely on the relationship between white and red blood cells in the CSF, with red blood cells being introduced in the CSF following the brain bleed , and white blood cells being seen as a response to infection. These criteria assume that the blood is homogeneous in the CSF. However, from computed tomography (CT) imaging of these patients, it is seen that blood can settle in the brain ventricles.

In this study we aim to test the assumption that blood is homogeneously distributed in the CSF by sampling from the CSF in patients. Two samples are serially drawn allocated to a period between where patients are planned for a clinical repositioning, or not. We hypothesise that a heterogeneous distribution of blood in the CSF (as seen on CT imaging) may allow for the CSF constitution to change in serially drawn CSF samples, and that these changes may be exacerbated in repositioned patients as it may disturb the blood that has settled at the bottom of the ventricles as a result of gravity sedimentation. We further believe that these changes may affect clinical decision making and further complicate EVDI diagnostics.

Introduction:

External ventricular drains (EVD) are essential tools in neuro-critical care to treat acute hydrocephalus following traumatic or spontaneous cerebral hemorrhage. While essential, EVDs are associated with feared infections (EVDIs) that may result in long term sequelae and affect long term outcome. It is therefore important to promptly identify and treat EVDIs. EVDI diagnostics are based on cerebrospinal fluid (CSF) analyses and CSF bacterial cultures. Causal agents are commonly pathogens colonizing the skin, which makes suspected infection and true infection difficult to distinguish based on the cultured bacteria alone. Additionally, bacterial cultures may take days to finalize and seldom contribute to early decision making or guiding of antibiotic treatment. Diagnostics are thus mainly based on CSF analyses of cells, protein, lactate, and glucose. CSF cellularity, in particular, is frequently confounded in this patient group as a result of intraventricular hemorrhage (IVH), which introduces both erythrocytes and leukocytes into the CSF. To account for IVH, the ratio of erythrocytes and leukocytes (LE ratio) is used and is a central metric in the diagnostics of EVDIs. The LE ratio is based on the assumption that as bacteria is introduced into the CSF, leukocytes will increase in relationship to erythrocytes as a part of the immune response. The LE ratio further assumes that blood is homogeneously distributed in the CSF. However, it is clear from computed tomography (CT) imaging that blood is congregated to the bottom of the ventricles as a result of gravity sedimentation, indicating a heterogeneous sample space. Additionally, leukocytes and erythrocytes exhibit different properties in terms of density which causes a separation during gravity sedimentation. Leukocytes, and granulocytes in particular, may also exhibit upwards floatation due to density changes as they become activated as part of the inflammatory response. Thus, not only is an intraventricular separation of blood and CSF probable, but also a separation of individual cell types. As CSF is sampled from the EVD, a heterogeneous sample space may affect the constitution of sampled CSF depending on the location of the EVD in the ventricles and the location and distribution of intraventricular blood.

Hypothesis:

We hypothesized that blood is heterogeneously distributed in the CSF, and as a result, may introduce CSF parameter variability as the intraventricular distribution of blood and CSF may vary between samples. Furthermore, we believe that routine patient care, such as repositioning of the patient may be enough to cause such a variability.

Study protocol:

Paired CSF samples are collected from patients included in the study, instead of the routine single sample, bi-weekly or during workup for suspected infection. The resulting extra collected CSF from paired samples will amount to approximately 1-2 mL per session. Patients will be allocated to the collection of paired CSF samples around a routine patient repositioning, or a 10 minute wait time between paired CSF samples as identified as the average time between samples when a repositioning was performed during the burn a period. Patients in a lateral side position 2-3 hours prior to CSF sampling will be sampled on either side of a clinical repositioning. Open drains will be closed five minutes prior to the first sample as per clinical routine. 1,5 mL of CSF is drawn and discarded prior to each individual sample.

The following information should be documented for each sample session:

Patient ID:

Study sample nr for this patient:

Date:

Time in a lateral side position prior to sample 1:

Time of sample 1:

Time of repositioning (if repositioned):

Time of sample 2:

GCS:

GCS motor score:

Antibiotic treatment, yes/no:

If suspected/confirmed EVDI, date of diagnosis:

For sample 1 and sample 2:

CSF-Leukocytes:

CSF-Monocytes:

CSF-Granulocytes:

CSF-Erythrocytes:

CSF-Albumin:

CSF: Lactate:

CSF: Glucose:

Additional information is collected for each patient at study inclusion:

Name and social identification number:

Sex:

Main admission diagnosis:

Date of admission:

Date of study inclusion:

Based on the first CT scan after EVD insertion:

Intraventricular hemorrhage score 1-23:

Measured intraventricular blood volume:

Depth of intraventricular sediment in the dorsal horns of the lateral ventricles:

Two dimensional length from the tip of the EVD catheter to blood or sediment:

Patient inclusion:

Over18 years of age in treated in the NICU with inserted EVDs. Patient inclusion was based on written informed consent from nearest of kin.

Statistical analyses:

One-way tests such as the Wilcoxon signed rank test and the Fligner-Killen test of variance will be used to evaluate pair-wise differences and differences between sample groups. Mixed effects regression analyses will be used to evaluate independent predictors of CSF parameter change and to evaluate the correlation between change and demographic variables, such as CT derived variables. Patient ID will be included as a random effect to eliminate inter-individual differences.

Inclusion Criteria:

• patients treated with EVDs at the NICU at the Karolinska University Hospital in Stockholm Sweden during the inclusion period with with next of kin consent.

Exclusion Criteria:

• patients admitted with bacterial or viral CNS infections

Neuro Intensive Care patients in University Hospital Setting. This is the only NICU unit in the Stockholm region and covers the full population requiring ICU care with EVDs in the area