Predictors of extraventricular drain–associated bacterial ventriculitis

• Published in: Journal of critical care Vol. 29; no. 1; pp. 77 - 82
• Main Authors: Williamson, Rachel A., Phillips-Bute, Barbara G., McDonagh, David L., Gray, Marisa C., Zomorodi, Ali R., Olson, DaiWai M., Britz, Gavin W., Laskowitz, Daniel T., James, Michael L.
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.02.2014; Elsevier Limited
• Subjects: Adult; Age Factors; Aged; Bacterial infections; Bacterial Infections - etiology; Bacterial Infections - microbiology; Cerebral Ventriculitis - etiology; Cerebral Ventriculitis - microbiology; Cerebrospinal Fluid; Clinical neurology; Critical Care; Critical Care - statistics & numerical data; CSF; Drainage - adverse effects; Extraventricular drain; Female; Hospitalization; Hospitals, University - statistics & numerical data; Humans; Intensive care; Length of Stay; Male; Middle Aged; Mortality; Neurosurgery; Retrospective Studies; Sex Factors; Staphylococcus; Studies; Ventriculitis; United States > US; Intensive care; Neurosurgery; Clinical neurology; Ventriculitis; Extraventricular drain; CSF
• ISSN: 0883-9441; 1557-8615; 1557-8615
• DOI: 10.1016/j.jcrc.2013.08.012

Bacterial ventriculitis (BV) may develop in patients requiring external ventricular drains (EVDs). The purpose of this study was to determine predictors of EVD-associated BV onset. A retrospective review of Duke University Hospital patients with EVD device placement between January 2005 and May 2010 was conducted. Subject data were captured for predefined variables. Outcomes included in-hospital mortality, length of stay, and neurologic status at discharge. In 410 subjects with 420 EVDs, the BV rate was 10.2%. Univariate analysis indicated that age, sex, positive blood culture, duration of EVD placement, and the number of cerebrospinal fluid (CSF) samples taken were associated with BV. Of these, the number of CSF samples and sex retained significance in multivariable modeling (female: odds ratio, 0.47 [confidence interval, 0.23-0.97]; CSF samples: odds ratio, 1.08 [confidence interval 1.01-1.17]; P = .04; c index = 0.69). In this model, each CSF sample taken expanded the likelihood of BV by 8.3%. The most common pathogens were Staphylococcus or proprioniobacter (n = 26). Bacterial ventriculitis was associated with an increase in hospital length of stay (33 ± 22.9 days vs 24.6 ± 20.4 days; P = .04) but not mortality. An association exists between CSF sampling frequency and the development of EVD-associated BV. Larger prospective studies should be aimed at identifying causal relationships between these variables.