RESUMO
BACKGROUND: Ventriculostomy-related infections (VRIs) are reported in about 10 % of patients with external ventricular drains (EVDs). VRIs are difficult to diagnose due to clinical and laboratory abnormalities caused by the primary neurological injury which led to insertion of the EVD. Polymerase chain reaction (PCR) of the cerebrospinal fluid (CSF) may enable more accurate diagnosis of VRI. We performed a prospective cohort study to measure the incidence of VRI as diagnosed by 16S rRNA PCR. METHODS: Patients admitted to intensive care with a primary diagnosis of subarachnoid haemorrhage (SAH), traumatic brain injury (TBI), or intracerebral haemorrhage (ICH), who required an EVD, were assessed for inclusion in this study. Data were extracted from the electronic medical record, bedside charts, or from a prospectively collected database, the Neuroscience Outcomes in Intensive CarE database (NOICE). 16S rRNA PCR was performed on routinely collected CSF as per laboratory protocol. VRI was also diagnosed based on pre-existing definitions. RESULTS: 237 CSF samples from 39 patients were enrolled in the study. The mean patient age was 55.7 years, and 56.4 % were female. The most common primary neurological diagnosis was SAH (61.5 %). The incidence of a positive PCR was 2.6 % of patients (1 in 39) and 0.8 % of CSF samples (2 in 237). The incidence of VRI according to pre-published diagnostic criteria was 2.6 % - 41 % of patients and 0.4 % - 17.6 % of CSF samples. 28.2 % of patients were treated for VRI. Pre-published definitions which relied on CSF culture results had higher specificity and lower false positive rates for predicting a PCR result when compared to definitions incorporating non-microbiological markers of VRI. In CSF samples with a negative 16S rRNA PCR, there was a high proportion of non-microbiological markers of infection, and a high incidence of fever on the day the CSF sample was taken. CONCLUSIONS: The incidence of VRI as defined as a positive PCR was lower than the incidence of VRI according to several published definitions, and lower than the incidence of VRI as defined as treatment by the clinical team. Non-microbiological markers of VRI may be less reliable than a positive CSF culture in diagnosing VRI.