Introduction
Infections in patients with a subarachnoid hemorrhage (SAH), mainly pneumonia, urinary tract infections, blood stream infections, or bacterial ventriculitis, are independently associated with a prolonged stay in the intensive care unit (ICU), a poor outcome and increased mortality [
1,
2]. Bacterial ventriculitis is reported in 3–29 % of SAH patients [
2‐
6] and is strongly associated with the placement of CSF (cerebrospinal fluid) catheters [
5,
7]. Although infections and fever after SAH have been studied, which includes ventriculitis, no specific study has paid attention specifically and solely to suspected bacterial ventriculitis in this subgroup [
1,
3‐
6,
8]. In studies on external catheter-related bacterial ventriculitis, subgroup analyses specifically for SAH patients have not been performed [
9‐
16]. Although the incidence of external catheter-related bacterial ventriculitis is not expected to be different in other brain injury patients, the clinical symptoms of SAH closely resemble those of bacterial ventriculitis and can thus influence the clinical diagnosis of a bacterial ventriculitis.
A clinical suspicion of nosocomial bacterial ventriculitis should prompt a diagnostic workup and antimicrobial therapy [
7]. The diagnostic workup in these patients typically consists of neuroimaging, cerebrospinal fluid analysis (cell counts, Gram’s staining, cultures, biochemical tests for glucose and protein), and cultures of blood. The interpretation of the numbers of white cells in cerebrospinal fluid is especially problematic in patients who have bacterial ventriculitis that develops after intraventricular and subarachnoid hemorrhage; although a formula has been proposed for interpretation [
13], the diagnostic accuracy is unknown. The diagnosis of nosocomial bacterial ventriculitis is made on the basis of the results of a cerebrospinal fluid culture, of which the results can be false negative. We performed a retrospective cohort study to investigate the diagnostic accuracy of clinical and laboratory characteristics in bacterial ventriculitis after SAH.
Materials and Methods
Participants
The Academic Medical Center (AMC) in Amsterdam (The Netherlands) acts as a tertiary referral center for patients with an aneurysmal SAH in a region of approximately 1.3 million people. Consecutive patients admitted with an aneurysmal SAH were included in this study. These patients were collected from a prospectively kept database of all SAH patients admitted in the AMC. There was no objection of the Medical Ethics Committee of the AMC to perform this study. Patients were excluded if there was no clinical data available (for example due to early referral to another hospital). Clinical records, from the electronic patient chart, were retrospectively examined. World Federation of Neurosurgical Societies (WFNS) grade, Fisher grade, and length of stay were documented. WFNS grade is a five category (I–V) neurological grading scale based upon the Glasgow Coma Score (GCS) in which a higher score indicates a worse clinical condition [
17]. Grades were categorized as a dichotomous variable: good (I–III) and poor (IV–V) grade. The Fisher grade is a four-point scale for classification of the amount of extravasated blood on the initial CT scan (grade 1: no visible hematoma, grade 4: intraventricular or intracerebral hematoma) [
18]. Grades were dichotomized into good (Fisher grade 1–3) and poor (Fisher grade 4) grades. Treatment procedures (i.e., coiling, surgical clipping, decompressive craniotomy), catheter placement (external ventricular, lumbar or cisternal catheter), number of catheters and duration of drainage were documented.
Placement and maintenance of the external ventricular and lumbar catheters was performed according to a standardized protocol to minimize the risk of infection [
19]. External ventricular catheter placement was performed under sterile circumstances in the operating theater. Standard external ventricular catheters without antibiotic or silver impregnation were used. The external ventricular catheters were tunneled subcutaneously for at least 5 cm. External lumbar catheters were placed at bedside under sterile conditions and also tunneled subcutaneously. In most patients who underwent a craniotomy for surgical clipping of an aneurysm, a cisternal catheter was placed for postoperative CSF drainage during five days. All types of catheters were tunneled subcutaneously. In all procedures, patients received prophylactic antibiotics at least 30 min before the start of the surgery: ceftriaxone 1000 mg for an external ventricular catheter or a craniotomy, flucloxacillin 1000 mg for an external lumbar catheter. The catheters were attended to according to a standardized nursing protocol. No routine microbiological or chemical tests of CSF were taken. External ventricular or lumbar catheters were not routinely changed, they were only changed if there was a dysfunction or if there was an intractable infection.
Selective decontamination of the digestive tract was given to all patients in the ICU who were expected to be ventilator-dependent for at least two days. SDD was used as prevention of secondary colonization of patients with Gram-negative bacteria in order to reduce mortality [
20]. It consisted of orabase (polymyxin E 100 mg, tobramycin 80 mg, amphotericin B 500 mg), four times daily orally and intravenous cefotaxime 3 g/24 h continuously during 4 days.
Bacterial Ventriculitis and Diagnostic Testing
Bacterial ventriculitis was suspected in SAH patients if they experienced a (new) period of fever (see below for definition) and/or altered mental status and/or (new) nuchal rigidity. Other causes of infection were excluded by performing blood, bronchial and urinary cultures, plain chest X-rays and physical examination. Other causes of an altered mental status, like hydrocephalus or cerebral ischemia, were excluded by performing a plain CT scan.
Clinical records were evaluated to identify patients who were treated for bacterial ventriculitis. If antibiotics were started specifically for ventriculitis, this patient was defined as ‘clinically suspected bacterial ventriculitis’. All patients who were started on these antibiotics were included irrespective if they underwent transcranial surgery or had an external ventricular or lumbar catheter placed. Data from the chemical analyses of blood [C-reactive protein (mg/L), white blood cell count (10E9/L) and glucose levels (mmol/L)] and CSF [red blood cells (cells/μL), white blood cell count (cells/μL), protein (g/L) and glucose levels (mmol/L)] and microbiological analyses of CSF were collected. CSF was collected at the time of clinical suspicion of a bacterial ventriculitis from the external ventricular or lumbar catheter in place. If there was no CSF catheter in place, a lumbar puncture was performed. Bacterial ventriculitis was defined by a positive CSF culture for bacteria; however, if cultures revealed Staphylococcus epidermidis, two consecutive positive cultures were need to rule out contamination.
Of all patients, the daily maximum temperature (T
max) was collected. Fever was identified as a T
max ≥ 38.5 °C. For the calculation of the T
max in patients with suspected bacterial ventriculitis the T
max until, and including, the day the clinical suspicion of bacterial ventriculitis rose (see definition of bacterial ventriculitis) was used. This was done in order to exclude the days after the (suspected) bacterial ventriculitis, which are not relevant for comparison. For the patients not suspected of bacterial ventriculitis, the T
max during the whole admission period was used. The number of fever days was defined as the sum of days with a T
max ≥ 38.5 °C during the total period of hospitalization. The percentage of fever days was determined by dividing the amount of fever days by the total amount of days when a temperature was taken during the total hospitalization period.
Patients were divided into three groups: (1) no suspicion for bacterial ventriculitis; (2) clinical suspicion for bacterial ventriculitis, defined as initiation of empirical antibiotic treatment aimed specifically at bacterial ventriculitis, but negative CSF cultures; and (3) CSF culture-positive bacterial ventriculitis. Antibiotic regimen for bacterial ventriculitis consisted of ceftriaxone 2 g twice a day plus vancomycin 2 g twice a day, or ceftazidime 2 g three times a day plus vancomycin 2 g twice a day if an external CSF catheter was in place. Antibiotics were discontinued when the results of the CSF culture were negative (usually after 72 h) [
6]. The antibiotic regimen was continued for two weeks if CSF cultures were positive.
To define a possible effect of SDD on the occurrence of bacterial ventriculitis and on the results of the CSF cultures, patients with suspected bacterial ventriculitis were divided into four groups: CSF culture-negative ventriculitis patients who receive SDD (group 1) or who do not receive SDD (group 2), and patients who have a CSF culture-positive bacterial ventriculitis who do (group 3) or do not receive (group 4) SDD.
Statistical Methods
The obtained data were analyzed using Statistical Package for the Social Sciences 16.0 Software (IBM SPSS 19.0). Categorical variables were tested using a χ
2 or Fisher’s exact test. Continuous variables were tested with the Kolmogorov–Smirnov test for normal distribution. Normally distributed continuous variables are represented as a mean with a 95 % confidence interval (CI), and continuous variables which are not normally distributed are represented as a median with a 95 % confidence interval (CI). Normally distributed variables were tested with the Student’s t test (two group comparison) or a one-way ANOVA with Bonferroni post hoc analysis (multiple group comparison). Data that were not normally distributed were tested with the Mann–Whitney U test (two group comparison) or Kruskal–Wallis test (multiple group comparison). Values of p < 0.05 were considered as statistically significant.
Discussion
This study shows that almost a quarter of all patients admitted with an aneurysmal subarachnoid hemorrhage were treated with antibiotics for a clinical suspicion of bacterial ventriculitis, but that only in a small minority of these patients the CSF culture proved positive. A clinical suspicion of nosocomial bacterial ventriculitis should prompt a diagnostic workup and antimicrobial therapy. Our data show that clinical signs of ventriculitis are nonspecific and difficult to recognize in SAH patients who are sedated, who have recently undergone neurosurgery, or have a sterile inflammatory response in the CSF due to the SAH. Therefore, improvement of diagnostics for nosocomial bacterial ventriculitis in patients with aneurysmal SAH are needed to improve patients care and to reduce the administration of antibiotics, in order to lower the risk of antibiotic resistance development, and concomitantly to decrease health care costs.
The clinical symptoms of SAH, including headache, nuchal rigidity and altered mental status, closely resemble those of bacterial ventriculitis. These symptoms are thus non-specific and probably not useful to distinguish between both conditions. Fever, the most consistently reported feature in bacterial ventriculitis, occurs in 40 % of the patients after SAH with or without an infection [
21‐
24]. The cell count in CSF may be helpful but has a low sensitivity and specificity [
7]. The interpretation of white blood cells (WBC) in CSF is especially problematic in patients who develop bacterial ventriculitis after SAH as the presence of red blood cells (RBC) in CSF in itself causes an aseptic ventriculitis [
9,
10,
13,
14]. The proposed cell index for external ventricular drainage-related ventriculitis in patients with intraventricular hemorrhage was not confirmed in subsequent series [
13].
In the absence of better discriminative tests, treatment is usually initiated on first suspicion. Because nosocomial bacterial ventriculitis is associated with high mortality and morbidity [
10,
14,
25,
26]. With this treatment strategy we aimed in this study that as few as possible bacterial ventriculitis patients were missed and as much as possible were treated, thus reducing the morbidity and mortality. Furthermore, the antibiotics were discontinued as soon as the cultures come back negative, which was normally after three days. Although the CSF culture (the most validated test for bacterial meningitis and ventriculitis) was used as gold standard for diagnosis, it is known that false negative tests may occur [
21,
27,
28]. In the population under study, this could be further enhanced by the routine use of selective decontamination of the digestive tract in patients who were expected to be ventilator-dependent for at least two days. Patients who were treated for a (suspected) bacterial ventriculitis were more often in a worse clinical condition, hence the expectation to be ventilator-dependent. These poor condition patients are also the patients who are more prone to develop a bacterial ventriculitis [
2]. Prophylactic antibiotics may mask a positive CSF culture leading to a negative CSF culture in patients with true bacterial ventriculitis. However, only one of the patients clinically suspected of bacterial ventriculitis in whom antibiotics were discontinued after a negative CSF culture experienced a second period of clinically suspected and culture-positive bacterial ventriculitis. In this patient, the first suspicion in retrospect was low. Fever reoccurred five days after discontinuation of the antibiotics for the first suspicion and was preceded by a drain blockage for which surgical revision was necessary.
CSF drainage procedures are common after a SAH and a longer duration of external drainage makes patients more susceptible to bacterial ventriculitis [
7,
29,
30]. Not surprisingly, duration was significantly longer in patients with CSF culture-positive bacterial ventriculitis than in patients with CSF culture-negative bacterial ventriculitis. This is in concurrence with what is reported in the literature [
7,
9,
15,
16].
Three patients had a clinical suspicion after a lumbar puncture as sole procedure and three did not even had any procedure (shunting, surgery or lumbar puncture) at all. Despite this fact, they did show clinical signs of bacterial ventriculitis, emphasizing the difficulty in clinical diagnosis.
We found no specific CSF features that distinguish in the early diagnosis of ventriculitis after SAH, although the CSF red blood cell (RBC) count was higher in patients with CSF culture-negative bacterial ventriculitis than in patients with CSF culture-positive bacterial ventriculitis. The influx of RBCs in the CSF after SAH causes CSF disturbances, and these have been shown to have a poor discriminatory value [
10,
13,
31‐
33]. Other substrates, such as CSF lactate, cytokine levels, and serum procalcitonin, are also disturbed after SAH. Although procalcitonin has been shown to be able to discriminate between a systemic inflammatory response syndrome and a systemic infection [
34], the value to differentiate between aseptic and bacterial ventriculitis of this and other substrates is limited [
33‐
38]. Polymerase chain reaction of the CSF for the detection of bacterial pathogens has been shown to have a low sensitivity in patients with an external catheter-related bacterial ventriculitis and patients with aseptic ventriculitis after surgery, but assays are getting more sensitive [
39‐
42].
S. epidermidis was the most common pathogen found in the CSF cultures, as in other studies on external catheter-related bacterial ventriculitis [
7,
9,
10,
14‐
16,
43]. However, the frequency of
S. aureus as causative pathogen was lower than described in the literature, possibly the result of selective SDD which is aimed at reducing infection with
S. Aureus [
19,
44].
Limitations of this study are its retrospective nature and the small sample of patients with CSF culture-positive bacterial ventriculitis. Data were collected from clinical records, and underreporting of clinical suspicion and features may have biased the results. However, patients were only categorized as CSF culture-negative ventriculitis if antibiotics were started for this suspicion, which is a well-defined criterion with a computerized medication prescription system used in our facility. The choice of antibiotics is also quite specific and thus readily traceable to a tentative diagnosis of bacterial ventriculitis.
In this retrospective series of SAH patients, nosocomial bacterial ventriculitis is often suspected but confirmed by culture in a minority of cases. The clinical signs of ventriculitis are nonspecific and difficult to recognize in SAH patients. Improvement of diagnostics for nosocomial bacterial ventriculitis in patients with aneurysmal SAH is needed.