Case Report

A 47-year-old Caucasian female with unremarkable medical history was admitted to the neurological intensive care unit because of spontaneous subarachnoid hemorrhage grade 2 according to WFNS. Digital subtraction angiography revealed an aneurysm at the tip of the basilar artery. The patient underwent endovascular treatment with complete coil occlusion of the aneurysm. On the second day after admission, controlled CT scan revealed acute obstructive hydrocephalus and an external ventricular drain (EVD) catheter was placed. Because of mild intracranial hypertension due to incipient brain swelling, the patient remained on continuous sedation and analgesia. Despite administration of nimodipine and prompt initiation of hypervolemic and controlled hypertensive therapy the patient developed moderate vasospasm on the fifth day after the bleeding leading to a small infarct in the territory of the right posterior cerebral artery. In addition, the patient developed fever (bladder temperature >38.3°C) together with a rise of inflammatory parameters (increase of C-reactive protein [CRP] from 4.3 mg/l baseline to 21 mg/l). Empirical antibiotic treatment with an aminopenicillin plus beta-lactamase inhibitor (amoxicillin/clavulanate) was started after extensive screening for potential sources of infection (cerebrospinal fluid [CSF] analysis, repeated blood cultures, bronchoalveolar lavage, urine analysis). On day 7 post-admission CSF analysis yielded an approximately 10-fold increase in the cell index (CI) [1]. CSF culture remained sterile. Relevant CSF findings are summarized in Table 1. Because no other source of infection could be detected until then, a decision to administer vancomycin 10 mg once daily intrathecally was made. On the following day, multiresistant coagulase-negative Staphylococci were isolated from blood culture. Amoxicillin/clavulanate was discontinued after a 8-day course, and vancomycin was also given systemically. Intrathecal therapy with vancomycin was continued in parallel because the CI was still elevated above baseline level (Table 1). All indwelling devices were changed with the exception of the EVD because the risk of surgical complications of reinsertion in an extremely vulnerable brain secondary to brain edema and vasospasm was deemed inappropriately high by the neurosurgical consultant. With source control inflammatory parameters having rapidly declined both intrathecal and IV vancomycin were discontinued after a treatment course of 7 days. Further, brain edema and vasospasm resolved in the second week after admission, and the patient was weaned off the ventilator and eventually was extubated on post-admission day 12. After emergence from analgosedation, the patient presented fully responsive, only suffering from partial left hemianopsia. However, 1 week later her condition deteriorated. The patient appeared disoriented and drowsy and presented with septic fever. Diagnostic work-up showed purulent CSF (Table 1). In addition, neuroimaging revealed ependymal enhancement (Fig. 1) confirming the diagnosis of pyogenic ventriculomeningitis. Gram-negative rods were detected in CSF gram stain and CSF culture grew Klebsiella pneumoniae as causative agent. The same bacteria could also be isolated from blood culture 24 h later. Owing to the frequent access to the otherwise closed drainage system for vancomycin instillation, it can be speculated that this subsequent episode of gram-negative ventriculitis was related to EVD manipulation and irrigation. However, hematogenous seeding must be discussed as another possible mechanism of infection albeit the susceptibility of nonvascular devices to bacteremia is less clear. Antimicrobial therapy with high-dose carbapenem (meropenem 2 g IV qid) was instituted. CSF cultures turned sterile on day 4 after carbapenem initiation. EVD catheter exchange was planned, however, because repeat imaging demonstrated resolution of hydrocephalus and ventricular debris, the drainage catheter was removed without replacement after a 29-day period of external ventriculostomy. Microbiological analysis of the distal segment of the EVD-catheter applying semiquantitative roll-plate culture detected colonization with coagulase-negative Staphylococci. Resistance testing differed from the results obtained for the Staphylococcus strain isolated from blood culture on day 8 after admission indicating the possibility of independent infections with two distinct organisms. On examination of the plate, there was no evidence for gram-negative bacteria. Intravenous antimicrobial therapy with meropenem was continued for a total of 14 days. Control lumbar puncture yielded complete resolution of CSF infection. The patient made a full recovery and was discharged from hospital 6 weeks after admission.

Table 1 Progression of CSF findings over time
Fig. 1
figure 1

Contrast-enhanced cranial CT scan showing ependymal enhancement (arrows) and diffuse brain swelling. External ventricular drain catheter in situ in the right lateral ventricle


The above clinical vignette highlights some of the important aspects that must be considered when assuming a nosocomial infectious complication in the critically ill. The most commonly documented nosocomial infections are ventilator-associated pneumonia, bloodstream infections secondary to indwelling vascular catheters, urinary tract infection, surgical site infections, and sinusitis [2]. Notably, patients with severe intracranial disease, such as subarachnoid hemorrhage, bear a significant risk for such nosocomial infections [3] associated with prolonged length of stay and poor outcome [4]. Expert guidelines from national organizations are available concerning practice parameters for the evaluation of adult patients who develop signs of acute inflammation in the intensive care unit [5]. Regarding nosocomial CSF infections, several potential pitfalls must be taken into consideration. In some instances, diagnosis of device-related ventriculitis is straightforward with the presence of purulent CSF, usually seen in gram-negative infections. In the majority of cases however, diagnosis is impaired by the presence of systemic inflammation due to the primary disease. A prospective study of fever in neurocritical care patients indicates that although fever occurs in about 25% of such patients, almost half of them are noninfectious in etiology [6]. Fever may be from extracerebral sources of infection. Neurocritical care patients may suffer from various complications of intensive care treatment and frequently present with systemic inflammatory response syndrome, multiorgan dysfunction, or even failure [4]. Therefore, monitoring of acute-phase proteins as CRP and serum procalcitonin (PCT) is not helpful for the differential diagnosis of ventriculitis in intensive care patients because these parameters do not allow discrimination between systemic or local infection [7]. Further, CSF analysis might not prove useful in identifying ventriculitis because of the presence of aseptic inflammation in case of hemorrhagic CSF [8]. In our case, ventriculitis was suspected because of a rise in the so-called cell index (CI, ratio of leucocytes and erythrocytes in CSF divided by the ratio of leucocytes and erythrocytes in peripheral blood). This CI was introduced by our group as an adjunct parameter for the diagnosis of EVD-related ventriculitis to compensate confounding of the CSF white cell count by intraventricular hemorrhage [1]. The CI is based on the hypothesis that in patients with intraventricular blood and no evidence of ependymal or systemic inflammation, the white and red blood cells are distributed equally in the CSF and peripheral blood. A significant (cut off at our institution: 5-fold) increase of the CI is highly indicative of EVD-related ventriculitis. Again, it should be emphasized that no single parameter can reliably predict or exclude EVD-related infection [9]. The same holds true for neuroimaging studies. Although distinct CT and MR imaging features of pyogenic ventriculitis have been described [10], performance of such examinations must not delay initiation of antiinfective treatment. Clinical suspicion of nosocomial ventriculitis suffices for prompt initiation of empiric antibiotic therapy.

As already mentioned, the diagnosis of a suspected ventriculostomy-related infection or ventriculitis warrants urgent initiation of adequate antimicrobial chemotherapy. The antiinfectives selected must penetrate the blood–brain and blood–CSF barriers. Unless isolation of a causative agent allowing targeted treatment, empiric pharmacotherapy must cover the most likely pathogens involved. Knowledge of local resistance patterns is of paramount importance since the microbiological profiles vary between centers due to differences in antibiotic usage. According to recent surveys [11, 12], gram-positive cocci consistent with skin flora account for more than two-thirds of nosocomial CSF infections. The most common isolated gram-negative species are enterobacteriaceae and Pseudomonas. Nosocomial ventriculitis caused by fungi are rare but can represent a problem in immunosuppressed patients.

Due to the frequent implication of Staphylococci in nosocomial ventriculitis initial therapy with an antistaphylococcal agent with good CSF penetration such as rifampicin and a cephalosporin may be considered first-line therapy for this infection [1315]. If Staphylococci indeed are isolated and the organism is methicillin susceptible, therapy with flucloxacillin is regarded sufficient. However, in the real-life ICU setting multidrug-resistant gram-positive nosocomial pathogens such as methicillin-resistant Staphylococcus aureus (MRSA) or Staphylococccus epidermidis (MRSE) are highly prevalent [16]. Therefore, initial empiric treatment with the glycopeptide antibiotic vancomycin is recommended [15]. On the other hand, therapy with vancomycin has some limitations. The penetration of IV vancomycin into CSF, even through inflamed meninges and ventricular ependyma, is poor [17]. To overcome this pharmacokinetic drawback, direct instillation of vancomycin into the ventricles has been used. Potential caveats of this therapeutical approach are the possible risk of the emergence of vancomycin-resistant pathogens due serum levels far below the minimally inhibitory concentration. Further, intrathecal administration requires EVD manipulation and catheter irrigation, both factors which have been identified to increase the risk of nosocomial ventriculitis [18]. Adverse effects, mainly its nephrotoxicity, may represent contraindications for vancomycin therapy in critically ill patients. Furthermore, strains with decreased susceptibility and resistance to vancomycin have emerged [19]. The oxazolidinone antibiotic linezolid represents an alternative to glycopeptides with a similar spectrum. The efficacy of linezolid for the treatment of nosocomial gram-positive ventriculomeningitis has been demonstrated recently [20, 21]. However, usage of this substance is, besides its economic impact, also not without risk, especially in intensive care patients. Linezolid is a reversible nonselective inhibitor of monoaminooxidase and therefore has the potential to interact with adrenergic substances, enhancing the pressor response [22]. Further, toxicity to the nervous system after prolonged therapy with linezolid has been described [23]. Suspected infections with resistant gram-negative pathogens require therapy with a cephalosporin with antipseudomonal activity or a carbapenem until culture results provide information to optimize therapy [15]. In the cases of fulminant gram-negative ventriculitis, intraventricular aminoglycosides also have been administered [24]. The same concerns remain as with direct instillation of vancomycin. Voriconazole and amphotericin B are options for the treatment of nosocomial fungal infections of the central nervous system [15].

A switch towards the appropriate antimicrobial agent tailored to the causative pathogen must be made as soon as microbiologic data from culture are available. Recommendations on the duration of antimicrobial therapy have not been rigorously studied. Mostly, treatment is continued for 10–14 days, although some experts have suggested shorter durations if repeated CSF cultures are negative.

The approaches to therapy of ventriculostomy-related infections in the published literature have included antimicrobial treatment with or without device removal [15]. No randomized, controlled study has been performed so far addressing this clinically relevant issue. In accordance with existing data for catheter-related bloodstream infections, the decision as to whether the catheter should be removed or retained largely depends on the causing organism [25]. Anyway, in the case of persistent CSF infection despite appropriate antimicrobial therapy removal of the device is an important measure. If the EVD is to be retained, a longer duration of therapy (10–14 days rather than 5–7 days if the catheter is exchanged) is recommended [13]. More robust data exist on the prophylactic EVD revision and according to the most recent studies there is little evidence to support the practice of elective catheter exchange at a predefined interval [26, 27].

At our institution we have implemented the following algorithm for the management of a febrile patient with suspected EVD-related ventriculitis [13]: All potential sources of nosocomial infections are examined. Special attention is given to ventilator-associated pneumonia, catheter-related bloodstream infections, urinary tract infection, and Clostridium difficile-induced colitis. Surgical sites are carefully inspected. Further, other potential causes of fever such as venous thromboembolic disease or drug-related fever are evaluated. Necessary examinations are CSF analysis including Gram stain, cell counts and differential, glucose, protein, and lactate concentrations, as well as CSF culture and calculation of the CI, together with analysis of simultaneously drawn blood samples. In case of primarily hemorrhagic CSF and a Gram stain positive for Staphylococci or an at least 5-fold increase in the CI and ventriculostomy catheter in place for more than 72 h, we start therapy with intrathecal vancomycin (10 mg daily). If blood culture also yields gram-positive cocci vancomycin (1 g IV tid to qid according to trough levels) is added. In patients where administration of vancomycin is contraindicated, especially in patients with impaired renal function, we administer linezolid (600 mg IV bid). If polymicrobial infection must be suspected or in immunocompromised patients, therapy with vancomycin or linezolid is supplemented with meropenem (2 g IV qid). Antimicrobial therapy is usually continued for 7 days unless a prolonged treatment period is judged appropriate. EVD is only replaced in patients with inadequate response to antimicrobial therapy. Whenever possible, antiinfective treatment is adapted according to resistance testing of an isolated pathogen (e.g., step down from vancomycin to flucloxacillin).

In the case of a primarily purulent CSF, we treat immediately with a broad-spectrum antimicrobial therapy covering resistant gram-positive and gram-negative pathogens (vancomycin IV 1 g tid plus meropenem 2 g IV qid). Note that prolonged therapy with broad-spectrum antibiotics may be complicated by Clostridium difficile (antibiotic-associated) colitis or fungal superinfection, primarily with Candida species. With proven pyogenic ventriculitis, we usually perform EVD exchange. It should be also noted that device-related infections by gram-negative rods are associated with a high frequency of relapse if the catheter is retained [28]. Again, therapy is promptly adapted according to resistance testing of an isolated pathogen. Antiinfective therapy is maintained for at least 2 weeks. Administration of intraventricular aminoglycosides (e.g., gentamicin 8 mg daily) remains an option in patients where repeated CSF cultures yield gram-negative bacteria despite appropriate IV antimicrobial therapy.