External ventricular drain (EVD) placement is one of the most commonly used interventions in life-saving neurosurgical care, with over 25,000 EVDs being placed annually nationwide¹⁾. EVDs are often used to treat traumatic brain injuries, infection, primary and secondary hydrocephalus, and intracranial space occupying lesions, such as tumors, cysts, and ischemic/hemorrhagic strokes^2-4). During EVD placement, a ventricular catheter is passed through a burr hole in the skull and utilized for draining blood, infectious debris and cerebral spinal fluid (CSF), instilling medications, and continuously monitoring intracranial pressure (ICP)⁵⁾. ICP monitoring is essential in managing neuro-critical patients and EVDs are considered the gold standard.

EVDs are widely researched for their safety and efficacy. EVDs have been proven to reduce mortality in intracerebral hemorrhages with intraventricular extension⁶⁾ reduce mortality in subarachnoid hemorrhages⁷⁾ and are a suggested treatment guideline by the Brain Trauma Foundation⁸⁾ and the Neurocritical Care Society³⁾. Infection rates amongst EVD insertion can be as high as 7.9% with a hemorrhage incidence can be as high as 8.4%⁹⁾. Intracranial hemorrhages and infections can prolong hospital stay, reduce patient independent ambulation capacity and overall increase patient morbidity and mortality. While there is a plethora of research investigating the complications associated with EVDs, there is limited research regarding overall patient outcomes in patients with no neurosurgical history requiring an EVD emergently. Our study investigates patient outcomes associated with emergent EVD placement in a patient population with no prior history of neurosurgery. This will hopefully provide objective insight for clinicians and patients alike on what to expect when an EVD is used for therapeutic management.

We retrospectively reviewed the charts of all patients aged 18 and above that had an EVD placed from January 1^st, 2016 through December 31^st, 2022 at Allegheny General Hospital in Pittsburgh, PA. We excluded patients with intra-operative EVD placements and patients with any prior neurosurgical history. We collected patient age, sex, body mass index (BMI), etiology of primary brain pathology, EVD OP measurement, and Glasgow coma scale (GCS) on admission. The etiology of primary brain pathology for each patient was reported as tumor, space occupying lesion, infection, hemorrhage, or other. Brain pathology locations and volume status were assessed using patients’ initial head CT and provider notes. Locations were reported as supra-tentorial, infra-tentorial, epidural, subdural, subarachnoid, intraparenchymal, and intraventricular, with no locations being mutually exclusive.

To assess the severity of each brain pathology, CT images were analyzed for the presence of transependymal flow, sulcal/gyral effacement, and enlarged temporal horns. To evaluate patient outcomes, we used independent ambulation capability of the patient at discharge, length of hospital stay, subsequent neurosurgical intervention, and patient discharge disposition. Subsequent neurosurgical interventions were stratified into six subcategories: EVD removal only, subsequent EVD placement, ventriculoperitoneal shunt (VPS) placement, endoscopic third ventriculostomy (ETV), craniotomy, and other. Patients may have undergone multiple interventions following EVD placement, and this was accounted for in multivariate regression analysis. Patient disposition included home, inpatient rehab, skilled nursing facility, long-term acute care, hospice, and expiration.

Descriptive statistics are presented as mean and standard deviation for continuous variables or count and percentage for categorical variables. Univariate and multivariate logistic regression models were created to determine factors associated with binary outcomes, including independent ambulation, EVD removal only, subsequent EVD placement, VPS placement, ETV, craniotomy, other subsequent neurosurgical interventions, home discharge, inpatient rehab discharge, SNF discharge, long-term acute care discharge, and patient expiration. Univariate and multivariate linear regression models were created to determine factors associated with hospital length of stay. For all multivariate regression models, a subset of predictor variables was chosen by akaike information criterion (AIC) stepwise selection. A p-value of < 0.05 was defined as statistically significant for all statistical tests. All statistical analyses were performed using R version 4.2.0.

We reviewed a total of 223 patient charts; 100 (45%) patients were male, and 123 patients (55%) were female (Table 1). The average age at admission was 59.31 years with a standard deviation (SD) of 14.46, and the average GCS on admission was 11.13 (SD 4.24). The average BMI was 28.94 kg/m² (SD 7.16), and most of the population were either obese (n=85, 38%) or overweight (n=69, 31%). Complications upon insertion were negligible amongst the patient population.

The most common brain injury found amongst our patients were hemorrhages, with 186 (83%) patients experiencing them (Fig. 1). All brain pathologies were generally distributed evenly between the supratentorial (n=216, 97%) and infratentorial (n=198, 89%) regions of the skull with the majority being found in the intraventricular (n=201, 90%) and subarachnoid (n=153, 69%) compartments (Table 2). Further radiographic volumetric analysis also showed that most of the patients (n=181, 81%) had enlarged temporal horns and effacement of the sulcal/gyral pattern (n=130, 58%).

Patient outcomes were evaluated based on length of hospital stay, independent ambulation at discharge, subsequent neurosurgical interventions, and final disposition (Table 3). The study population’s average hospital stay was 20.47 days (SD 15.13), and only 23 (10%) patients were capable of independent ambulation upon discharge. After EVD insertion our patient population were most likely to have their EVD removed with no other intervention occurring (n=143, 64%). Nevertheless, a notable number of patients had a craniotomy (n=77, 35%) or had a VPS placed (n=54, 24%). Of the 223 patients in our cohort, only 38 (17%) were healthy enough to be discharged home, with a number of patients being discharged to an inpatient rehab facility (n=59, 26%) or passing away (n=45, 20%).

Our study aimed to examine the outcomes following emergent EVD placement in a neuro-surgically naïve patient population. Using specific outcome measures, our data helps accurately delineate a realistic clinical course of a patient requiring EVD placement. Current literature on patient outcomes following EVD placement oftentimes is only described in the context of a specific etiology leading to the EVD placement (SAH vs infection vs lesion, etc.) This limits the generalizability of the study as outcomes are likely only going to be representative of that specific brain insult pathology. Our study is specific to outcomes associated with EVD placement following any brain insult etiology described in our patient population. This allows for a comprehensive inclusion of different pathologies that lead to EVD placement, which fosters a more realistic clinical picture. Most similar to our study is from Gu et al. were they provided an observational study on intracranial hemorrhage patients requiring EVD placement¹⁰⁾. While this study accurately describes potential complications, subsequent shunt placement and GCS score on follow-up, patients with previous neurosurgical intervention were not excluded. We feel this is an import patient characteristic when delineating a clinical course, as previous neurosurgical intervention could skew outcomes and potential characterize an inaccurate representation of patient outcomes and foster inappropriate expectations in clinicians and families alike. Furthermore, only patients with intracranial hemorrhage were included in the study. While hemorrhage is one of the more common indications for EVD placement, there are other pathologies including tumors, space occupying lesions, infection and more that may require EVD placement. Including patients with all brain insult pathologies leading to EVD insertion provides a more accurate and comprehensive approach to detailing outcomes.

Despite our patients having underwent successful uncomplicated EVD placement and stabilization, favorable patient outcomes were limited. In-hospital mortality was 20% and an additional 9% were discharged to hospice care, bringing the combined mortality and end-of-life care rate to nearly 30%. Only 18% of patients were discharged home, while the remainder required continued care at rehabilitation centers, skilled nursing facilities, or long-term care institutions. Furthermore, only 10% of patients were independently ambulatory at discharge, almost one fourth (24%) required subsequent permanent management of VPS shunt placement and our average hospital stay was 20.47 days. This data reflects the high burden of morbidity following emergent EVD placement in critically ill patients. It is important to elucidate that EVDs are a safe and efficacious therapeutic management and are by no means implicated as the cause of these morbid patient outcomes. In our study EVD placement is used as a threshold for severity of a serious neurological injury occurring, leading to hydrocephalus. Once this threshold is met, we can surmise that the neurologic injury is critical and patient outcomes are affected.

This study provides novel insight into the clinical course of emergent EVD placement in patients exclusively without prior neurosurgical history. Understanding outcomes in this context is critical for neurosurgeons, intensivists, and families alike, as it frames expectations for recovery, disposition and the likelihood of subsequent procedures. These findings can guide clinical decision-making and help bridge a gap in health literacy between clinicians and families. This would allow for more informed compassionate discussions with patients’ families regarding prognosis and care planning after EVD placement.