Abstract
BACKGROUND: Patients with hemorrhagic stroke and an external ventricular drain in situ are at risk for ventriculostomy-related-infections (VRI). Because of the contamination of the cerebrospinal fluid (CSF) with blood and the high frequency of false negative CSF culture, the diagnosis of VRI remains challenging. This study investigated the introduction of CSF broad range eubacterial polymerase chain reaction (ePCR) and its effect on frequency and duration of antibiotic therapy for VRI, neurocritical care unit (NCCU) length of stay, related costs, and outcome. METHODS: Between 2020 and 2022, we prospectively included 193 patients admitted to the NCCU of the University Hospital of Zürich with hemorrhagic stroke and an external ventricular drain for more than 48 h. Patient characteristics, serum inflammatory markers, white blood cell count in CSF, use and duration of antibiotic treatment for VRI, microbiological findings (CSF cultures and ePCR tests), and NCCU length of stay were compared in patients with no infection, noncerebral infection, suspected VRI, and confirmed VRI. Data of patients with suspected VRI of this cohort were compared with a retrospective cohort of patients with suspected VRI treated at our NCCU before the introduction of CSF ePCR testing (2013-2019). RESULTS: Out of 193 patients, 12 (6%) were diagnosed with a confirmed VRI, 66 (34%) with suspected VRI, 90 (47%) with a noncerebral infection, and 25 (13%) had no infection at all. Compared with the retrospective cohort of patients, the use of CSF ePCR resulted in a reduction of patients treated for suspected VRI for the whole duration of 14 days (from 51 to 11%). Furthermore, compared with the retrospective group of patients with suspected VRI (n = 67), after the introduction of CSF ePCR, patients with suspected VRI had shorter antibiotic treatment duration of almost 10 days and, hence, lower related costs with comparable outcome at 3 months. CONCLUSIONS: The use of CSF ePCR to identify VRI resulted in shorter antibiotic treatment duration without changing the outcome, as compared with a retrospective cohort of patients with suspected VRI.