Journal of Clinical Neuroscience 21 (2014) 765–768
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Efficacy of antibiotic-impregnated external ventricular drains in reducing ventriculostomy-associated infections Yana Mikhaylov, Thomas J. Wilson, Venkatakrishna Rajajee, B. Gregory Thompson, Cormac O. Maher, Stephen E. Sullivan, Teresa L. Jacobs, Mary J. Kocan, Aditya S. Pandey ⇑ Department of Neurosurgery, University of Michigan, 1500 E. Medical Center Drive, Room 3552 TC, Ann Arbor, MI 48109-5338, USA
a r t i c l e
i n f o
Article history: Received 11 September 2013 Accepted 12 September 2013
Keywords: Antibiotics External ventricular drain Ventriculitis Ventriculostomy-associated infection Ventriculostomy catheter
a b s t r a c t Use of an external ventricular drain (EVD) is essential for managing patients with hydrocephalus or intracranial hypertension. While this procedure is safe and efficacious, ventriculostomy-associated infections (VAI) continue to cause significant morbidity. In this study, we evaluated the efficacy of antibiotic-coated EVD (AC-EVD) in reducing the occurrence of VAI. Between July 2007 and July 2009, 203 patients underwent placement of an EVD. A total of 145 of these patients met the inclusion criteria, with 76 patients (52.4%) receiving AC-EVD and 69 patients (47.6%) receiving uncoated EVD. Ten patients (6.9%) developed VAI, of whom three were in the AC-EVD group and seven were in the uncoated EVD group (p = 0.19). The mean duration between catheter insertion and positive cerebrospinal fluid culture was significantly greater in the AC-EVD group versus the uncoated EVD group (15 ± 4 days versus 4 ± 2 days, respectively; p = 0.001). In the uncoated EVD group, 17 of 69 patients (24.6%) were dead at 3 years versus 12 of 76 (15.8%) patients in the AC-EVD group (p = 0.21). The overall VAI rate was 6.9% with a trend toward lower infection rates in the AC-EVD group compared to the uncoated EVD group (3.9% versus 10.1%, respectively; p > 0.05). Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction External ventricular drains (EVD) allow for diagnosis and treatment of intracranial hypertension. While their use is essential in neurocritical care, ventriculostomy-associated infections (VAI) and secondary ventriculitis lead to significant morbidity and costs. Previous clinical series have reported a 3–19% occurrence of VAI, leading to longer stays in the intensive care unit (ICU) as well as increased morbidity [1–3]. A retrospective analysis performed at the University of Michigan, USA showed that the infection rate in patients with EVD for longer than 10 days was approximately 8.6% [4]. Antibiotic-coated EVD (AC-EVD), as well as prophylactic antibiotics, are frequently used in an attempt to reduce the incidence of VAI. Poon et al. showed that antibiotic prophylaxis continued for the duration of EVD placement reduces the rate of VAI from 11% to 3% (p = 0.01) [5]. In addition, the advent of ACEVD may have reduced the occurrence of VAI to less than 3% [6,7]. Following a review of the literature, the University of Michigan protocol was modified in July 2008 to require placement of the Bactiseal AC-EVD (DePuy, Raynham, MA, USA) in all patients requiring an EVD. Our objective was to perform a retrospective
⇑ Corresponding author. Tel.: +1 734 615 4486; fax: +1 734 936 9294. E-mail address: [email protected] (A.S. Pandey). 0967-5868/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocn.2013.09.002
analysis of the impact of the use of AC-EVD on the incidence of VAI at our institution. 2. Methods 2.1. Study design This retrospective study evaluating the incidence of VAI was approved by the University of Michigan Institutional Review Board. All patients requiring EVD placement between July 2007 and July 2009 were eligible for inclusion in the study. Exclusion criteria consisted of known or suspected infection of cerebrospinal fluid (CSF), placement of more than one EVD, replacement of the EVD within a 30 day period, and/or EVD removal less than 24 hours after insertion. 2.2. Insertion and maintenance of EVD All EVD insertions were performed using standard sterile precautions in the operating room, angiography suite, emergency department, or neurosurgical/trauma-burn ICU. Uncoated EVD were employed prior to implementation of the July 2008 protocol. Thereafter, Bactiseal EVD impregnated with 0.15% clindamycin and 0.054% rifampicin were employed as the standard catheter. The EVD was introduced at a location approximately 1 cm anterior to
766
Y. Mikhaylov et al. / Journal of Clinical Neuroscience 21 (2014) 765–768
the coronal suture in the midpupillary line. All patients received antibiotic prophylaxis with intravenous cefazolin until the EVD was removed unless the patient had a penicillin allergy, in which case intravenous vancomycin was utilized. When a pre-existing infection was being treated with a broad-spectrum antibiotic, this agent was considered to function as prophylactic coverage for the EVD. Identical insertion and maintenance protocols were used for both coated and uncoated catheter types. EVD were inspected daily by the neurosurgical team for evidence of CSF leak and infection at the insertion site. Sampling and analysis of CSF was performed only when clinical suspicion of infection existed. Such analysis included cell count with differential, glucose, protein, Gram stain, and culture. 2.3. Data collection Inpatient progress notes, operative notes, outpatient notes, laboratory data, and radiological images were evaluated for each patient. We collected the following data points: catheter type used (coated versus uncoated), age, sex, diagnosis with reason for EVD implantation, performance of craniotomy, presence of subarachnoid or intraventricular hemorrhage, location of insertion, duration the EVD remained in place, history of diabetes, history of cardiovascular conditions including hypertension, antibiotic administered, use of steroids, complications associated with EVD insertion, number of insertion attempts, presence or absence of a CSF leak, presence of concomitant lumbar or subdural drain, volume of CSF drained per day, presenting Glasgow Coma Scale (GCS) score, and presence of stapled gauze versus adhesive transparent film dressing. The primary endpoint of this analysis was to evaluate the incidence of VAI as defined by the occurrence of positive CSF cultures. A positive CSF culture within 1 week of removal of the EVD with clinical evidence of ventriculitis was treated with a full course of antibiotics and was also considered to represent a VAI. Secondary outcome measures included mean number of days between EVD insertion and positive CSF culture, median change in GCS score between admission and removal of EVD, need for new CSF diverting shunt placement, length of stay within the ICU following
placement of EVD, and 3 year mortality. Cases of eosinophilic meningitis associated with AC-EVD were also recorded. 2.4. Statistical analysis All continuous variables were subjected to the D’Agostino– Pearson test for normal distribution. Mean and standard deviation with range was calculated for variables with normal distribution. For variables without normal distribution, the median with interquartile range (IQR) was calculated. Univariate assessment of categorical variables with outcomes of interest was accomplished using chi-squared or Fisher exact test as appropriate. Univariate assessment of normally distributed continuous variables was performed using Student’s t-test and non-normally distributed variables with the Mann–Whitney U test. The threshold for statistical significance was p < 0.05. All univariate associations that attained p < 0.2 were included, along with the exploratory variable (use of AC-EVD) in multivariate logistic regression models. 3. Results Between July 2007 and July 2009, 203 patients had EVD placed at the University of Michigan. Fifty-eight patients were excluded from study for the following reasons: EVD replacement within a 30 day period, EVD placed secondary to ventriculitis or meningitis, bilateral ventriculostomies, or unknown EVD type. Of the 145 patients evaluated in the final analysis, 76 (52.4%) patients received AC-EVD catheters, while 69 (47.6%) patients received uncoated EVD. Overall there were 10 positive CSF cultures (6.9%), of which nine patients had CSF drawn while the EVD was in place and one was drawn 6 days following removal of the EVD. This last patient was re-admitted with headache and neck stiffness leading to a lumbar puncture and positive CSF cultures. The distribution of variables in patients with and without VAI is shown in Table 1. Only the median volume of CSF drained attained statistical significance (p < 0.05) in univariate analysis with the primary outcome (VAI). Seventy-three of 135 patients (54.1%) without VAI and three of 10 patients (30.0%) with VAI received an AC-EVD (p = 0.19).
Table 1 Distribution of variables in patients with and without ventriculostomy-associated infection after external ventricular drain placement Variable
VAI absent (n = 135)
VAI present (n = 10)
p Value
Mean age (years) ± SD Female (%) Antibiotic-coated catheter used (%) Diagnosis (n) Hydrocephalus Intracerebral hemorrhage CSF leak SAH Ischemic stroke Traumatic brain injury Tumor Unruptured vascular malformation Craniotomy (%) SAH/IVH present (%) Diabetes (%) Cardiovascular comorbidity (%) Steroid use (%) Median initial GCS score (IQR) Total duration of EVD use, days (IQR) Multiple passes required during EVD placement (%) Site of EVD insertion (n) Angiography suite Emergency room Neurological ICU
54 ± 17 80 (59) 73 (54)
54 ± 14 7 (70) 3 (30)
0.96 0.96 0.19 0.85
4 20 3 65 4 9 24 6 85 (63) 76 (56) 19 (14) 61 (45) 7 (5) 15 (9–15) 7 (3–13) 8 (6)
0 2 0 7 0 0 1 0 5 (50) 8 (80) 0 (0) 6 (60) 1 (10) 15 (10–15) 11 (8–15) 1 (10)
5 35 29
0 2 3
0.50 0.19 0.36 0.51 0.48 0.85 0.06 0.48 0.87
767
Y. Mikhaylov et al. / Journal of Clinical Neuroscience 21 (2014) 765–768 Table 1 (continued) Variable
VAI absent (n = 135)
VAI present (n = 10)
Operating room Outside hospital Trauma ICU Antibiotic used for prophylaxis (n) Cefepime Ceftriaxone Clindamycin Cefazolin Vancomycin Piperacillin/Tazobactam None Median CSF volume drained, cc (IQR) Dressing stapled (%) CSF leak present (%) New ventriculoperitoneal shunt eventually placed (%) Mortality at 3 years (%)
59 5 2
4 1 0
2 1 1 108 21 1 1 558 (72–1440) 62 (46) 9 (7) 26 (19) 28 (21)
0 0 0 9 1 0 0 1424 (997–1940) 3 (30) 2 (20) 2 (20) 1 (10)
p Value
0.99
0.01 0.47 0.17 1.00 0.69
CSF = cerebrospinal fluid, EVD = external ventricular drain, ICU = intensive care unit, IQR = interquartile range, IVH = intraventricular hemorrhage, SAH = subarachnoid hemorrhage, SD = standard deviation, VAI = ventriculostomy-associated infection.
Bacteria identified in VAI patients with AC-EVD included Enterobacter aerogenes in one patient, Corynebacterium afermentans in one patient, and Propionibacterium acnes in one patient. Bacteria identified in VAI patients without AC-EVD included Acinetobacter calcoaceticus in one patient, Streptococcus mitis in one patient, coagulase-negative Staphylococcus in one patient, and Propionibacterium acnes in four patients. When the presence of AC-EVD was tested in multivariate analysis with individual variables that attained p < 0.2 on univariate analysis (median CSF volume drained, duration of EVD use, presence of subarachnoid or intraventricular hemorrhage, or presence of CSF leak), presence of AC-EVD failed to attain statistical significance for association with VAI. When total duration of EVD use was included in multivariate analysis with median volume of CSF drained, there was no longer a statistically significant association of either variable with the primary outcome. Among the patients within the AC-EVD group, one patient had a lumbar drain and another patient had a subdural drain. No such drains were documented to be present in the uncoated EVD group. The mean (±standard deviation) number of days between EVD insertion and positive CSF culture was 4 ± 2 days for the uncoated EVD group versus 15 ± 4 days for the AC-EVD group (p = 0.001). Fifteen (21.7%) patients from the uncoated EVD group required new CSF diverting shunts compared to 13 patients (17.1%) from the AC-EVD group (p = 0.533). The median change in GCS score between the uncoated EVD (0; IQR 0–1.75) and AC-EVD (0; IQR 0– 1) groups was not statistically significant (p = 0.50). The median number of days spent in the ICU following placement of EVD was not significantly different; 13 (IQR 8–18) for the uncoated EVD group and 13 (IQR 6.25–19.75) for the AC-EVD group (p = 0.74).
Table 2 Results by external ventricular drain type
Infections, NNT = 16 (%) Average incubation period until positive culture (days) New shunts (not replaced) (%) Average change in GCS score ICU stay (days) 3 year mortality
Plain catheter (n = 69)
Bactiseala coated catheter (n = 76)
p Value
7 (10) 3.7
3 (4) 14.7
0.193 0.001
15 (22) 0.7 14 17 (24.6%)
13 (17) 0.7 13.5 12 (15.8%)
0.533 0.500 0.740 0.210
a DePuy, Raynham, MA, USA. GCS = Glasgow Coma Scale, ICU = intensive care unit, NNT = number needed to treat.
Seventeen of 69 patients (24.6%) in the uncoated EVD group were dead at 3 years versus 12 of 76 (15.8%) in the AC-EVD group (p = 0.21) (Table 2).
4. Discussion Insertion of EVD is essential in the management of neurosurgical patients with hydrocephalus and/or high intracranial pressure. While these procedures are common and safe, VAI continue to be a significant cause of morbidity. We performed a retrospective analysis of patients who received EVD at the University of Michigan between 2007 and 2009 to determine the rate of VAI in those patients with uncoated EVD versus those with AC-EVD. Of the 145 patients included in this study, 76 (52.4%) patients received Bactiseal coated EVD, while 69 (47.6%) patients received uncoated EVD. There were fewer VAI in the AC-EVD group than in the uncoated EVD group (4% versus 10%, respectively); however, this difference was not statistically significant (p = 0.193). In view of the relatively small number of VAI in our study and the trend toward fewer infections with AC-EVD, a true protective effect with AC-EVD remains possible. Age, sex, presenting GCS score, baseline co-morbidities, location of insertion (OR versus ICU), type of dressing, and presenting pathologies did not vary significantly between these two groups. Patients with VAI required EVD use for a longer duration and had a higher median volume of CSF output per day. It is unclear if the trend toward an association between longer duration of EVD use and VAI is cause or effect. While duration of EVD use was a risk factor for VAI in some studies [1,4], patients with ventriculitis may develop a non-communicating hydrocephalus requiring longer duration of CSF diversion. The association between higher median CSF volume drained and VAI may be a result of confounding from longer duration EVD use in patients with VAI (as suggested in our multivariate analysis) as well as a consequence of non-communicating hydrocephalus caused by VAI. On the other hand, it is plausible that the larger volume of drainage from a ventriculostomy necessitates the intraventricular space remain open to the environment for longer duration, thus increasing the probability of bacterial translocation. Of particular interest is our finding of a significantly longer duration between EVD placement and positive CSF culture in patients with AC-EVD compared to uncoated EVD (15 versus 4 days). This has not been previously documented; however, it is to be expected, given that it would take longer for bacteria to proliferate in the antibacterial environment of the AC-EVD catheter compared to the plain EVD. While it might be expected that AC-EVD result in
768
Y. Mikhaylov et al. / Journal of Clinical Neuroscience 21 (2014) 765–768
the selection of unusual or particularly resistant organisms such as yeast and multi-drug resistant Gram-negative organisms, this was not observed in the limited number of infections seen in our study. Bactiseal coated EVD allow for release of the antibiotics (0.15% clindamycin and 0.054% rifampicin) into the lumen of the EVD up to 50 days after insertion [8]. We evaluated whether the release of these antibiotics could potentially lead to scarring within CSF flow pathways and thus a higher incidence of hydrocephalus. There was, however, no significant difference between the number of patients requiring CSF diverting shunts in the AC-EVD and uncoated EVD groups. There was also no significant difference in the length of ICU stay and 3 year mortality between patients with and without AC-EVD. Our routine clinical practice is to place all patients with EVD on prophylactic antibiotics for the duration of catheter use. In a retrospective analysis performed by Alleyne et al., the incidence of VAI in patients receiving prophylactic antibiotics was 3.8% versus 4% in those who only received periprocedural antibiotics.[9] We were not able to determine a difference in the frequency of VAI as a function of the type of prophylactic antibiotic (p = 0.39). The need for prophylactic antibiotics when AC-EVD are used was addressed by Wong et al. in 2008 [10]. Patients requiring ventriculostomies were randomized to plain EVD with prophylactic antibiotics versus ACEVD without prophylactic antibiotics. Neither group in that series developed VAI and thus the authors concluded that the use of prophylactic antibiotics with plain EVD was equivalent to AC-EVD in preventing VAI. Pople et al. reported on an international randomized trial comparing AC-EVD and uncoated EVD [11]. The rate of VAI was not significantly different between the two groups; 2.3% in the AC-EVD group versus 2.8% in the uncoated EVD group (p = 0.193). While AC-EVD placement is proven to be safe, the use of AC-EVD did not reduce the incidence of VAI in that trial. The authors in that study did not identify whether patients received prophylactic antibiotics for the duration of EVD placement. This is an important determination, as prophylaxis with antibiotics could allow an uncoated EVD to have much reduced incidence of VAI. On the other hand, Abla et al. compared minocycline-coated catheters to rifampin-coated catheters in a randomized trial. [12] Of the 129 patients who received these EVD, none developed a VAI. The lack of VAI in this group is much less than the expected rate of 8.8% as determined by meta-analyses. Similarly, Zabramski et al. reported on a randomized controlled trial comparing AC-EVD versus uncoated EVD. [7] Positive CSF cultures were significantly increased in the plain EVD group compared to the AC-EVD group (9.4% versus 1.3%, respectively; p = 0.002). While use of AC-EVD is safe, current reports are conflicted in the ability of such catheters to reduce the incidence of VAI. We report a statistically insignificant reduction in the incidence of VAI with the use of AC-EVD (3.9% versus 10.1% for uncoated EVD; p > 0.05). This study is inherently limited secondary to its retrospective design. In addition, our patient care protocol prevents us
from sending routine CSF cultures without suspicion of infection. Thus, we only counted symptomatic infections and not potential positive CSF cultures without clinical symptoms. Given that the data represents EVD placed over a period of 2 years, infection rates could have varied according to changing practices; however, there were no changes to our surgical techniques/practices. Future trials are necessary to understand the role of AC-EVD and prophylactic antibiotics in reducing VAI. 5. Conclusions There was a trend toward fewer VAI in patients with AC-EVD compared to uncoated EVD, although the difference did not reach statistical significance. Duration between EVD insertion and positive culture was significantly longer in patients with AC-EVD. Further randomized trials are necessary to understand the role of AC-EVD in preventing VAI. Conflict of interest/disclosure The authors declare that they have no financial or other conflicts of interest in relation to this research and its publication. References [1] Arabi Y, Memish ZA, Balkhy HH, et al. Ventriculostomy-associated infections: incidence and risk factors. Am J Infect Control 2005;33:137–43. [2] Bota DP, Lefranc F, Vilallobos HR, et al. Ventriculostomy-related infections in critically ill patients: a 6-year experience. J Neurosurg 2005;103:468–72. [3] Rebuck JA, Murry KR, Rhoney DH, et al. Infection related to intracranial pressure monitors in adults: analysis of risk factors and antibiotic prophylaxis. J Neurol Neurosurg Psychiatry 2000;69:381–4. [4] Park P, Garton HJ, Kocan MJ, et al. Risk of infection with prolonged ventricular catheterization. Neurosurgery 2004;55:594–9. [5] Poon WS, Ng S, Wai S. CSF antibiotic prophylaxis for neurosurgical patients with ventriculostomy: a randomised study. Acta Neurochir Suppl 1998;71: 146–8. [6] Tamburrini G, Massimi L, Caldarelli M, et al. Antibiotic impregnated external ventricular drainage and third ventriculostomy in the management of hydrocephalus associated with posterior cranial fossa tumours. Acta Neurochir (Wien) 2008;150:1049–55 [discussion 1055–1046]. [7] Zabramski JM, Whiting D, Darouiche RO, et al. Efficacy of antimicrobialimpregnated external ventricular drain catheters: a prospective, randomized, controlled trial. J Neurosurg 2003;98:725–30. [8] Thomas R, Lee S, Patole S, et al. Antibiotic-impregnated catheters for the prevention of CSF shunt infections: a systematic review and meta-analysis. Br J Neurosurg 2012;26:175–84. [9] Alleyne Jr CH, Hassan M, Zabramski JM. The efficacy and cost of prophylactic and perioprocedural antibiotics in patients with external ventricular drains. Neurosurgery 2000;47:1124–7. [10] Wong GK, Poon WS, Ng SC, et al. The impact of ventricular catheter impregnated with antimicrobial agents on infections in patients with ventricular catheter: interim report. Acta Neurochir Suppl 2008;102:53–5. [11] Pople I, Poon W, Assaker R, et al. Comparison of infection rate with the use of antibiotic-impregnated vs standard extraventricular drainage devices: a prospective, randomized controlled trial. Neurosurgery 2012;71:6–13. [12] Abla AA, Zabramski JM, Jahnke HK, et al. Comparison of two antibioticimpregnated ventricular catheters: a prospective sequential series trial. Neurosurgery 2011;68:437–42.
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Efficacy of antibiotic-impregnated external ventricular drains in reducing ventriculostomy-associated infections Yana Mikhaylov, Thomas J. Wilson, Venkatakrishna Rajajee, B. Gregory Thompson, Cormac O. Maher, Stephen E. Sullivan, Teresa L. Jacobs, Mary J. Kocan, Aditya S. Pandey ⇑ Department of Neurosurgery, University of Michigan, 1500 E. Medical Center Drive, Room 3552 TC, Ann Arbor, MI 48109-5338, USA
a r t i c l e
i n f o
Article history: Received 11 September 2013 Accepted 12 September 2013
Keywords: Antibiotics External ventricular drain Ventriculitis Ventriculostomy-associated infection Ventriculostomy catheter
a b s t r a c t Use of an external ventricular drain (EVD) is essential for managing patients with hydrocephalus or intracranial hypertension. While this procedure is safe and efficacious, ventriculostomy-associated infections (VAI) continue to cause significant morbidity. In this study, we evaluated the efficacy of antibiotic-coated EVD (AC-EVD) in reducing the occurrence of VAI. Between July 2007 and July 2009, 203 patients underwent placement of an EVD. A total of 145 of these patients met the inclusion criteria, with 76 patients (52.4%) receiving AC-EVD and 69 patients (47.6%) receiving uncoated EVD. Ten patients (6.9%) developed VAI, of whom three were in the AC-EVD group and seven were in the uncoated EVD group (p = 0.19). The mean duration between catheter insertion and positive cerebrospinal fluid culture was significantly greater in the AC-EVD group versus the uncoated EVD group (15 ± 4 days versus 4 ± 2 days, respectively; p = 0.001). In the uncoated EVD group, 17 of 69 patients (24.6%) were dead at 3 years versus 12 of 76 (15.8%) patients in the AC-EVD group (p = 0.21). The overall VAI rate was 6.9% with a trend toward lower infection rates in the AC-EVD group compared to the uncoated EVD group (3.9% versus 10.1%, respectively; p > 0.05). Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction External ventricular drains (EVD) allow for diagnosis and treatment of intracranial hypertension. While their use is essential in neurocritical care, ventriculostomy-associated infections (VAI) and secondary ventriculitis lead to significant morbidity and costs. Previous clinical series have reported a 3–19% occurrence of VAI, leading to longer stays in the intensive care unit (ICU) as well as increased morbidity [1–3]. A retrospective analysis performed at the University of Michigan, USA showed that the infection rate in patients with EVD for longer than 10 days was approximately 8.6% [4]. Antibiotic-coated EVD (AC-EVD), as well as prophylactic antibiotics, are frequently used in an attempt to reduce the incidence of VAI. Poon et al. showed that antibiotic prophylaxis continued for the duration of EVD placement reduces the rate of VAI from 11% to 3% (p = 0.01) [5]. In addition, the advent of ACEVD may have reduced the occurrence of VAI to less than 3% [6,7]. Following a review of the literature, the University of Michigan protocol was modified in July 2008 to require placement of the Bactiseal AC-EVD (DePuy, Raynham, MA, USA) in all patients requiring an EVD. Our objective was to perform a retrospective
⇑ Corresponding author. Tel.: +1 734 615 4486; fax: +1 734 936 9294. E-mail address: [email protected] (A.S. Pandey). 0967-5868/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocn.2013.09.002
analysis of the impact of the use of AC-EVD on the incidence of VAI at our institution. 2. Methods 2.1. Study design This retrospective study evaluating the incidence of VAI was approved by the University of Michigan Institutional Review Board. All patients requiring EVD placement between July 2007 and July 2009 were eligible for inclusion in the study. Exclusion criteria consisted of known or suspected infection of cerebrospinal fluid (CSF), placement of more than one EVD, replacement of the EVD within a 30 day period, and/or EVD removal less than 24 hours after insertion. 2.2. Insertion and maintenance of EVD All EVD insertions were performed using standard sterile precautions in the operating room, angiography suite, emergency department, or neurosurgical/trauma-burn ICU. Uncoated EVD were employed prior to implementation of the July 2008 protocol. Thereafter, Bactiseal EVD impregnated with 0.15% clindamycin and 0.054% rifampicin were employed as the standard catheter. The EVD was introduced at a location approximately 1 cm anterior to
766
Y. Mikhaylov et al. / Journal of Clinical Neuroscience 21 (2014) 765–768
the coronal suture in the midpupillary line. All patients received antibiotic prophylaxis with intravenous cefazolin until the EVD was removed unless the patient had a penicillin allergy, in which case intravenous vancomycin was utilized. When a pre-existing infection was being treated with a broad-spectrum antibiotic, this agent was considered to function as prophylactic coverage for the EVD. Identical insertion and maintenance protocols were used for both coated and uncoated catheter types. EVD were inspected daily by the neurosurgical team for evidence of CSF leak and infection at the insertion site. Sampling and analysis of CSF was performed only when clinical suspicion of infection existed. Such analysis included cell count with differential, glucose, protein, Gram stain, and culture. 2.3. Data collection Inpatient progress notes, operative notes, outpatient notes, laboratory data, and radiological images were evaluated for each patient. We collected the following data points: catheter type used (coated versus uncoated), age, sex, diagnosis with reason for EVD implantation, performance of craniotomy, presence of subarachnoid or intraventricular hemorrhage, location of insertion, duration the EVD remained in place, history of diabetes, history of cardiovascular conditions including hypertension, antibiotic administered, use of steroids, complications associated with EVD insertion, number of insertion attempts, presence or absence of a CSF leak, presence of concomitant lumbar or subdural drain, volume of CSF drained per day, presenting Glasgow Coma Scale (GCS) score, and presence of stapled gauze versus adhesive transparent film dressing. The primary endpoint of this analysis was to evaluate the incidence of VAI as defined by the occurrence of positive CSF cultures. A positive CSF culture within 1 week of removal of the EVD with clinical evidence of ventriculitis was treated with a full course of antibiotics and was also considered to represent a VAI. Secondary outcome measures included mean number of days between EVD insertion and positive CSF culture, median change in GCS score between admission and removal of EVD, need for new CSF diverting shunt placement, length of stay within the ICU following
placement of EVD, and 3 year mortality. Cases of eosinophilic meningitis associated with AC-EVD were also recorded. 2.4. Statistical analysis All continuous variables were subjected to the D’Agostino– Pearson test for normal distribution. Mean and standard deviation with range was calculated for variables with normal distribution. For variables without normal distribution, the median with interquartile range (IQR) was calculated. Univariate assessment of categorical variables with outcomes of interest was accomplished using chi-squared or Fisher exact test as appropriate. Univariate assessment of normally distributed continuous variables was performed using Student’s t-test and non-normally distributed variables with the Mann–Whitney U test. The threshold for statistical significance was p < 0.05. All univariate associations that attained p < 0.2 were included, along with the exploratory variable (use of AC-EVD) in multivariate logistic regression models. 3. Results Between July 2007 and July 2009, 203 patients had EVD placed at the University of Michigan. Fifty-eight patients were excluded from study for the following reasons: EVD replacement within a 30 day period, EVD placed secondary to ventriculitis or meningitis, bilateral ventriculostomies, or unknown EVD type. Of the 145 patients evaluated in the final analysis, 76 (52.4%) patients received AC-EVD catheters, while 69 (47.6%) patients received uncoated EVD. Overall there were 10 positive CSF cultures (6.9%), of which nine patients had CSF drawn while the EVD was in place and one was drawn 6 days following removal of the EVD. This last patient was re-admitted with headache and neck stiffness leading to a lumbar puncture and positive CSF cultures. The distribution of variables in patients with and without VAI is shown in Table 1. Only the median volume of CSF drained attained statistical significance (p < 0.05) in univariate analysis with the primary outcome (VAI). Seventy-three of 135 patients (54.1%) without VAI and three of 10 patients (30.0%) with VAI received an AC-EVD (p = 0.19).
Table 1 Distribution of variables in patients with and without ventriculostomy-associated infection after external ventricular drain placement Variable
VAI absent (n = 135)
VAI present (n = 10)
p Value
Mean age (years) ± SD Female (%) Antibiotic-coated catheter used (%) Diagnosis (n) Hydrocephalus Intracerebral hemorrhage CSF leak SAH Ischemic stroke Traumatic brain injury Tumor Unruptured vascular malformation Craniotomy (%) SAH/IVH present (%) Diabetes (%) Cardiovascular comorbidity (%) Steroid use (%) Median initial GCS score (IQR) Total duration of EVD use, days (IQR) Multiple passes required during EVD placement (%) Site of EVD insertion (n) Angiography suite Emergency room Neurological ICU
54 ± 17 80 (59) 73 (54)
54 ± 14 7 (70) 3 (30)
0.96 0.96 0.19 0.85
4 20 3 65 4 9 24 6 85 (63) 76 (56) 19 (14) 61 (45) 7 (5) 15 (9–15) 7 (3–13) 8 (6)
0 2 0 7 0 0 1 0 5 (50) 8 (80) 0 (0) 6 (60) 1 (10) 15 (10–15) 11 (8–15) 1 (10)
5 35 29
0 2 3
0.50 0.19 0.36 0.51 0.48 0.85 0.06 0.48 0.87
767
Y. Mikhaylov et al. / Journal of Clinical Neuroscience 21 (2014) 765–768 Table 1 (continued) Variable
VAI absent (n = 135)
VAI present (n = 10)
Operating room Outside hospital Trauma ICU Antibiotic used for prophylaxis (n) Cefepime Ceftriaxone Clindamycin Cefazolin Vancomycin Piperacillin/Tazobactam None Median CSF volume drained, cc (IQR) Dressing stapled (%) CSF leak present (%) New ventriculoperitoneal shunt eventually placed (%) Mortality at 3 years (%)
59 5 2
4 1 0
2 1 1 108 21 1 1 558 (72–1440) 62 (46) 9 (7) 26 (19) 28 (21)
0 0 0 9 1 0 0 1424 (997–1940) 3 (30) 2 (20) 2 (20) 1 (10)
p Value
0.99
0.01 0.47 0.17 1.00 0.69
CSF = cerebrospinal fluid, EVD = external ventricular drain, ICU = intensive care unit, IQR = interquartile range, IVH = intraventricular hemorrhage, SAH = subarachnoid hemorrhage, SD = standard deviation, VAI = ventriculostomy-associated infection.
Bacteria identified in VAI patients with AC-EVD included Enterobacter aerogenes in one patient, Corynebacterium afermentans in one patient, and Propionibacterium acnes in one patient. Bacteria identified in VAI patients without AC-EVD included Acinetobacter calcoaceticus in one patient, Streptococcus mitis in one patient, coagulase-negative Staphylococcus in one patient, and Propionibacterium acnes in four patients. When the presence of AC-EVD was tested in multivariate analysis with individual variables that attained p < 0.2 on univariate analysis (median CSF volume drained, duration of EVD use, presence of subarachnoid or intraventricular hemorrhage, or presence of CSF leak), presence of AC-EVD failed to attain statistical significance for association with VAI. When total duration of EVD use was included in multivariate analysis with median volume of CSF drained, there was no longer a statistically significant association of either variable with the primary outcome. Among the patients within the AC-EVD group, one patient had a lumbar drain and another patient had a subdural drain. No such drains were documented to be present in the uncoated EVD group. The mean (±standard deviation) number of days between EVD insertion and positive CSF culture was 4 ± 2 days for the uncoated EVD group versus 15 ± 4 days for the AC-EVD group (p = 0.001). Fifteen (21.7%) patients from the uncoated EVD group required new CSF diverting shunts compared to 13 patients (17.1%) from the AC-EVD group (p = 0.533). The median change in GCS score between the uncoated EVD (0; IQR 0–1.75) and AC-EVD (0; IQR 0– 1) groups was not statistically significant (p = 0.50). The median number of days spent in the ICU following placement of EVD was not significantly different; 13 (IQR 8–18) for the uncoated EVD group and 13 (IQR 6.25–19.75) for the AC-EVD group (p = 0.74).
Table 2 Results by external ventricular drain type
Infections, NNT = 16 (%) Average incubation period until positive culture (days) New shunts (not replaced) (%) Average change in GCS score ICU stay (days) 3 year mortality
Plain catheter (n = 69)
Bactiseala coated catheter (n = 76)
p Value
7 (10) 3.7
3 (4) 14.7
0.193 0.001
15 (22) 0.7 14 17 (24.6%)
13 (17) 0.7 13.5 12 (15.8%)
0.533 0.500 0.740 0.210
a DePuy, Raynham, MA, USA. GCS = Glasgow Coma Scale, ICU = intensive care unit, NNT = number needed to treat.
Seventeen of 69 patients (24.6%) in the uncoated EVD group were dead at 3 years versus 12 of 76 (15.8%) in the AC-EVD group (p = 0.21) (Table 2).
4. Discussion Insertion of EVD is essential in the management of neurosurgical patients with hydrocephalus and/or high intracranial pressure. While these procedures are common and safe, VAI continue to be a significant cause of morbidity. We performed a retrospective analysis of patients who received EVD at the University of Michigan between 2007 and 2009 to determine the rate of VAI in those patients with uncoated EVD versus those with AC-EVD. Of the 145 patients included in this study, 76 (52.4%) patients received Bactiseal coated EVD, while 69 (47.6%) patients received uncoated EVD. There were fewer VAI in the AC-EVD group than in the uncoated EVD group (4% versus 10%, respectively); however, this difference was not statistically significant (p = 0.193). In view of the relatively small number of VAI in our study and the trend toward fewer infections with AC-EVD, a true protective effect with AC-EVD remains possible. Age, sex, presenting GCS score, baseline co-morbidities, location of insertion (OR versus ICU), type of dressing, and presenting pathologies did not vary significantly between these two groups. Patients with VAI required EVD use for a longer duration and had a higher median volume of CSF output per day. It is unclear if the trend toward an association between longer duration of EVD use and VAI is cause or effect. While duration of EVD use was a risk factor for VAI in some studies [1,4], patients with ventriculitis may develop a non-communicating hydrocephalus requiring longer duration of CSF diversion. The association between higher median CSF volume drained and VAI may be a result of confounding from longer duration EVD use in patients with VAI (as suggested in our multivariate analysis) as well as a consequence of non-communicating hydrocephalus caused by VAI. On the other hand, it is plausible that the larger volume of drainage from a ventriculostomy necessitates the intraventricular space remain open to the environment for longer duration, thus increasing the probability of bacterial translocation. Of particular interest is our finding of a significantly longer duration between EVD placement and positive CSF culture in patients with AC-EVD compared to uncoated EVD (15 versus 4 days). This has not been previously documented; however, it is to be expected, given that it would take longer for bacteria to proliferate in the antibacterial environment of the AC-EVD catheter compared to the plain EVD. While it might be expected that AC-EVD result in
768
Y. Mikhaylov et al. / Journal of Clinical Neuroscience 21 (2014) 765–768
the selection of unusual or particularly resistant organisms such as yeast and multi-drug resistant Gram-negative organisms, this was not observed in the limited number of infections seen in our study. Bactiseal coated EVD allow for release of the antibiotics (0.15% clindamycin and 0.054% rifampicin) into the lumen of the EVD up to 50 days after insertion [8]. We evaluated whether the release of these antibiotics could potentially lead to scarring within CSF flow pathways and thus a higher incidence of hydrocephalus. There was, however, no significant difference between the number of patients requiring CSF diverting shunts in the AC-EVD and uncoated EVD groups. There was also no significant difference in the length of ICU stay and 3 year mortality between patients with and without AC-EVD. Our routine clinical practice is to place all patients with EVD on prophylactic antibiotics for the duration of catheter use. In a retrospective analysis performed by Alleyne et al., the incidence of VAI in patients receiving prophylactic antibiotics was 3.8% versus 4% in those who only received periprocedural antibiotics.[9] We were not able to determine a difference in the frequency of VAI as a function of the type of prophylactic antibiotic (p = 0.39). The need for prophylactic antibiotics when AC-EVD are used was addressed by Wong et al. in 2008 [10]. Patients requiring ventriculostomies were randomized to plain EVD with prophylactic antibiotics versus ACEVD without prophylactic antibiotics. Neither group in that series developed VAI and thus the authors concluded that the use of prophylactic antibiotics with plain EVD was equivalent to AC-EVD in preventing VAI. Pople et al. reported on an international randomized trial comparing AC-EVD and uncoated EVD [11]. The rate of VAI was not significantly different between the two groups; 2.3% in the AC-EVD group versus 2.8% in the uncoated EVD group (p = 0.193). While AC-EVD placement is proven to be safe, the use of AC-EVD did not reduce the incidence of VAI in that trial. The authors in that study did not identify whether patients received prophylactic antibiotics for the duration of EVD placement. This is an important determination, as prophylaxis with antibiotics could allow an uncoated EVD to have much reduced incidence of VAI. On the other hand, Abla et al. compared minocycline-coated catheters to rifampin-coated catheters in a randomized trial. [12] Of the 129 patients who received these EVD, none developed a VAI. The lack of VAI in this group is much less than the expected rate of 8.8% as determined by meta-analyses. Similarly, Zabramski et al. reported on a randomized controlled trial comparing AC-EVD versus uncoated EVD. [7] Positive CSF cultures were significantly increased in the plain EVD group compared to the AC-EVD group (9.4% versus 1.3%, respectively; p = 0.002). While use of AC-EVD is safe, current reports are conflicted in the ability of such catheters to reduce the incidence of VAI. We report a statistically insignificant reduction in the incidence of VAI with the use of AC-EVD (3.9% versus 10.1% for uncoated EVD; p > 0.05). This study is inherently limited secondary to its retrospective design. In addition, our patient care protocol prevents us
from sending routine CSF cultures without suspicion of infection. Thus, we only counted symptomatic infections and not potential positive CSF cultures without clinical symptoms. Given that the data represents EVD placed over a period of 2 years, infection rates could have varied according to changing practices; however, there were no changes to our surgical techniques/practices. Future trials are necessary to understand the role of AC-EVD and prophylactic antibiotics in reducing VAI. 5. Conclusions There was a trend toward fewer VAI in patients with AC-EVD compared to uncoated EVD, although the difference did not reach statistical significance. Duration between EVD insertion and positive culture was significantly longer in patients with AC-EVD. Further randomized trials are necessary to understand the role of AC-EVD in preventing VAI. Conflict of interest/disclosure The authors declare that they have no financial or other conflicts of interest in relation to this research and its publication. References [1] Arabi Y, Memish ZA, Balkhy HH, et al. Ventriculostomy-associated infections: incidence and risk factors. Am J Infect Control 2005;33:137–43. [2] Bota DP, Lefranc F, Vilallobos HR, et al. Ventriculostomy-related infections in critically ill patients: a 6-year experience. J Neurosurg 2005;103:468–72. [3] Rebuck JA, Murry KR, Rhoney DH, et al. Infection related to intracranial pressure monitors in adults: analysis of risk factors and antibiotic prophylaxis. J Neurol Neurosurg Psychiatry 2000;69:381–4. [4] Park P, Garton HJ, Kocan MJ, et al. Risk of infection with prolonged ventricular catheterization. Neurosurgery 2004;55:594–9. [5] Poon WS, Ng S, Wai S. CSF antibiotic prophylaxis for neurosurgical patients with ventriculostomy: a randomised study. Acta Neurochir Suppl 1998;71: 146–8. [6] Tamburrini G, Massimi L, Caldarelli M, et al. Antibiotic impregnated external ventricular drainage and third ventriculostomy in the management of hydrocephalus associated with posterior cranial fossa tumours. Acta Neurochir (Wien) 2008;150:1049–55 [discussion 1055–1046]. [7] Zabramski JM, Whiting D, Darouiche RO, et al. Efficacy of antimicrobialimpregnated external ventricular drain catheters: a prospective, randomized, controlled trial. J Neurosurg 2003;98:725–30. [8] Thomas R, Lee S, Patole S, et al. Antibiotic-impregnated catheters for the prevention of CSF shunt infections: a systematic review and meta-analysis. Br J Neurosurg 2012;26:175–84. [9] Alleyne Jr CH, Hassan M, Zabramski JM. The efficacy and cost of prophylactic and perioprocedural antibiotics in patients with external ventricular drains. Neurosurgery 2000;47:1124–7. [10] Wong GK, Poon WS, Ng SC, et al. The impact of ventricular catheter impregnated with antimicrobial agents on infections in patients with ventricular catheter: interim report. Acta Neurochir Suppl 2008;102:53–5. [11] Pople I, Poon W, Assaker R, et al. Comparison of infection rate with the use of antibiotic-impregnated vs standard extraventricular drainage devices: a prospective, randomized controlled trial. Neurosurgery 2012;71:6–13. [12] Abla AA, Zabramski JM, Jahnke HK, et al. Comparison of two antibioticimpregnated ventricular catheters: a prospective sequential series trial. Neurosurgery 2011;68:437–42.
