SPECIAL ARTICLE

TOPIC SPECIAL ARTICLE

Meta-Analysis of Hemorrhagic Complications

From Ventriculostomy Placement
by Neurosurgeons
David F. Bauer, MD*
Shantanu N. Razdan, MBBS‡
Alfred A. Bartolucci, PhD§
James M. Markert, MD, MPH*
*Division of Neurosurgery, Departments
of ‡Health Care Organization and Policy
and §Biostatistics, School of Public
Health, University of Alabama at
Birmingham, Birmingham, Alabama
Correspondence:
James M. Markert, MD,
University of Alabama,
Birmingham, FOT #1060,
1530 3rd Avenue South,
Birmingham, AL 35294-3410.
E-mail: markert@uab.edu.
Received, February 7, 2010.
Accepted, January 14, 2011.
Published Online, April 5, 2011.
Copyright ª 2011 by the
Congress of Neurological Surgeons
BACKGROUND: Ventriculostomy placement is an important diagnostic and therapeutic
tool for neurosurgeons. Multiple authors have presented retrospective series of patients
evaluating periprocedure hemorrhage.
OBJECTIVE: We performed a meta-analysis of existing studies to determine a more
accurate rate of hemorrhage.
METHODS: A MEDLINE and PubMed search was performed to find all studies of 25 or
more patients conducted since 1970 that found a hemorrhagic complication rate from
placement of a ventriculostomy. Studies in which a non-neurosurgeon placed the
ventriculostomy and studies involving premature infants were excluded.
RESULTS: Sixteen studies were used to obtain data from 2428 ventriculostomy pro-
cedures. Hemorrhage was found after 203 procedures, and 52 of these hemorrhages
were deemed significant by the authors. The cumulative rate of hemorrhage was 7.0%
(95% confidence interval: 4.5%-9.4%), with P , .05. The cumulative rate of significant
hemorrhage was 0.8% (95% confidence interval: 0.2%-1.4%) with P , .05.
CONCLUSION: Based on our meta-analysis, the overall hemorrhagic complication rate
from ventriculostomy placement by neurosurgeons is approximately 7%. The rate of
significant hemorrhage from ventriculostomy placement is approximately 0.8%. Further
prospective studies are warranted to better address this question.
KEY WORDS: Hemorrhage, Meta-analysis, Ventriculostomy
Neurosurgery 69:255–260, 2011 DOI: 10.1227/NEU.0b013e31821a45ba www.neurosurgery-online.com

hemorrhage.2 Lack of clear methodology limits

V
entriculostomy placement is an important
diagnostic and therapeutic tool for neu-
rosurgeons. According to procedural data
published by the American Association of
Neurological Surgeons, 42 446 intracranial
pressure monitoring procedures were performed
during 2006.1 It is routine practice for neuro-
surgeons to place an intracranial pressure mon-
itoring device in a patient with hydrocephalus or
a traumatic brain injury and a poor neurological
examination.
Hemorrhage is a known complication of ven-
triculostomy. Multiple authors have presented
retrospective series showing the rate of hemor-
rhage from ventriculostomy placement ranging
between 1% and 41%. Only 1 meta-analysis
has been performed to try to find a ‘‘true’’ rate of
ABBREVIATION: CI, confidence interval
the results and confers bias to this study. The
study did not grade the quality of the included
studies. It only included 13 studies3-15 with
1790 ventriculostomy placements. One study
included data from ventriculostomies performed
by non-neurosurgeons, and 1 study included
data from ventriculostomies performed in pre-
mature infants.
In light of these data, we elected to perform
a more rigorous meta-analysis that included
additional studies from the literature and a novel
grading system evaluating each included study.
METHODS
A MEDLINE and PubMed search was performed
to find all studies of 25 or more patients conducted
since 1970 that found a hemorrhagic complication
rate from placement of a ventriculostomy. We chose

NEUROSURGERY VOLUME 69 | NUMBER 2 | AUGUST 2011 | 255

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.

BAUER ET AL
1970 as a starting date because Hounsfield invented the computed
tomography (CT) scanner in 1972, and we wanted to capture all
studies in which CT was used to evaluate the presence of hemorrhage.
We chose studies of 25 or more patients to decrease selection bias. The
Congress of Neurological Surgeons and American Association of
Neurological Surgeons Web sites were also searched for any
unpublished abstracts that fit our inclusion criteria (1995-2008 at
www.AANS.org; 1997-2008 at www.CNS.org). These respective years
cover the entire online abstract databases of both organizations. In
addition, we attempted to search international Web sites such as the
Web site for the Society of British Neurosurgeons and the European
Society of Neurosurgical Societies, but we were unable to find online
abstracts on any of these sites. Studies in which a non-neurosurgeon
placed the ventriculostomy or studies that involved ventriculostomy
placement in premature infants were excluded.
The Web of Science was used to find the number of times each
article had been cited in the published literature. A novel grading
system was derived in an attempt to quantify the quality of each study.
All studies were retrospective, but not all studies obtained routine
postventriculostomy imaging for all patients. In addition, not all
studies quantified the extent of hemorrhage caused by ventriculostomy
placement. We thought that these 2 criteria improved the quality of
hemorrhage detection and quantification. Our grading system con-
sisted of the following 3 categories: grade I, every patient underwent
a postprocedure CT scan and the extent of hemorrhage was described
by the authors; grade II, every patient underwent a postprocedure CT
scan, but the extent of hemorrhage was not described by the authors;
and grade III, every patient did not undergo a routine postprocedure
CT scan.
We carried out 2 main group analyses using the meta-analysis software
Comprehensive Meta-analysis, version 2.0 (Biostat, Englewood, New
Jersey). One group included the hemorrhage rate reported in each study,
and the other included the significant hemorrhage rate from each study.
Data were entered as ‘‘rate of hemorrhage’’ and ‘‘rate of significant
hemorrhage’’ from each study. The standard error was calculated
manually for each study using the formula: standard error = square root
{[proportion (1 2 proportion)]/n}. The random-effects model was used
for analysis and result reporting. We assumed that each study meth-
odology was unique and studies were not homogeneous.
To do calculations for studies that had no incidental or significant
hemorrhages, we first performed an analysis excluding these studies to
obtain a lower limit for the 95% confidence interval (CI) of the
cumulative hemorrhage rate. We used this lower limit as the
approximate rate of hemorrhage for these studies, and we manually
calculated the standard error using these data. This was done because of
the limitation of the software to ignore 0 as a proportion for calculating
standard error.
RESULTS
Sixteen studies3-13,16-20 were used to obtain data from 2428
ventriculostomy procedures. The number of times that each
article was cited in the literature ranged from 0 to 244 citations
(Table 1). Using our novel grading scale, 5 articles were classified
as grade I, 5 articles as grade II, and 6 articles were considered to
be grade III (Table 1). Hemorrhage was observed after 203 of the
procedures, and 52 of these hemorrhages were deemed significant
by the authors (Table 1). The cumulative rates of hemorrhage and
256 | VOLUME 69 | NUMBER 2 | AUGUST 2011
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
significant hemorrhage from all 16 studies are 7.0% (95% CI:
4.5%-9.4%) with P , .05 and 0.8% (95% CI: 0.2%-1.4%) with
P , .05, respectively (Table 2, Figures 1 and 2).
A re-examination of the meta-analysis data in patients who had
routine postoperative scanning (grade 1 plus grade 2) found
hemorrhage in 189 patients after 1513 procedures. The cumu-
lative rate of hemorrhage was 12.1% (95% CI: 7.1%-17.1%)
with P , .05. The cumulative rate of significant hemorrhage in
the same group was 1.4% (95% CI: 0.3%-2.5%) with P , .05.
In the subgroup in which every patient underwent a post-
procedure CT scan and the extent of hemorrhage was described
by the authors (grade I), the rate of hemorrhage was 20.3% (95%
CI: 6.5%-34.1%) with P , .05. The rate of significant hem-
orrhage in the same subgroup was 3.7% (95% CI: 0.8%-6.6%)
with P , .05.
In the subgroup in which every patient underwent a post-
procedure CT scan but the extent of hemorrhage was not de-
scribed by the authors (grade II), the rate of hemorrhage was
3.6% (95% CI: 0.7%-6.6%) with P , .05. The rate of significant
hemorrhage in the same subgroup was 0.5% (95% CI: 0%-1.1%)
with P = .06.
In the group that did not undergo a routine postoperative CT
scan (grade III), the rate of hemorrhage was 1.4% (95% CI:
0.5%-2.4%) with P , .05. The rate of significant hemorrhage in
the same subgroup was 0.5% (95% CI: 0%-0.9%) with P , .05.
A z test for proportions was used to compare the total number
of studies that reported hemorrhages in the group that underwent
a postprocedure CT scan (9 of 10 of these studies, or 90%,
reported hemorrhages) and the group of studies that did not
undergo routine postprocedure CT scans (5 of 6 of the studies,
or 83%, reported hemorrhages). There was no significant dif-
ference in hemorrhage rates seen between these 2 groups of
studies (P . .05).
For both overall and significant hemorrhage groups, the
P value was ,.05 for the test of heterogeneity, implying that
the assumptions made for using a random model to carry out the
analysis were accurate.
DISCUSSION
Communicating accurate procedural risks to patients and
family members is vital to the practicing neurosurgeon. Often it is
difficult to quantify risks for patients because of the lack of high-
quality, meaningful studies in the literature. Data from the
neurosurgical literature are often Level 3, in the form of non-
randomized, retrospective case series. A rigorous meta-analysis of
the risk of hemorrhage from a very common procedure, ven-
triculostomy placement, provides important data for neuro-
surgeons communicating with patients and their families about
the risks of this procedure.
One previous meta-analysis performed by Binz et al2 evalu-
ating the hemorrhagic complication rate of ventriculostomy has
been published. In the article, the authors did not de-
scribe their methodology in detail. Their reported results of 5.7%
www.neurosurgery-online.com
 META-ANALYSIS VENTRICULOSTOMY HEMORRHAGE

TABLE 1. Studies That Reported Hemorrhagic Complications of External Ventricular Draina

Significant Total Postprocedure Hemorrhage
Author, Year Hemorrhages Hemorrhages EVDs CT Scan Described Citations Grade
Friedman and Vries, 1980 1 0 100 No No 51 III
Narayan et al, 1982 4 1 207 Yes No 244 II
North and Reilly, 1986 2 1 199 No No 45 III
Paramore and Turner, 1994 2 2 253 Yes No 57 II
Khanna et al, 1995 0 0 106 No No 43 III
Guyot et al, 1998 9 2 274 No No 32 III
O’Leary et al, 2000 0 0 49 Yes No 12 II
Roitberg et al, 2001 1 0 103 No No 19 III
Wiesmann and Mayer, 2001 6 0 92 Yes Grade I: volume of lesion ,1 mL 16 I
Grade II: volume of lesion 1-15 mL
Grade III: volume of lesion .15 mL
Krotz et al, 2004 1 0 52 Yes No 6 II
Anderson et al, 2004 12 1 68 Yes Blaha et al21 grading system 15 I
Maniker et al, 2006 52 4 160 Yes New areas of hyperdensity immediately 3 I
adjacent to and/or including the
catheter trajectory; volume of new
intraparenchymal hematoma measured
Leung et al, 2006 1 0 133 No No 2 III
Huyette et al, 2008 18 0 98 Yes No 2 II
Kakarla et al, 2008 17 4 346 Yes Track hemorrhages/EDH/SDH/IVH 1 I
Gardner et al, 2009 77 37 188 Yes Significant hemorrhages classified as 0 I
,15 mL ICH/.15 mL ICH or IVH/surgical
Total 203 52 2428 I=5
II = 5
III = 6
a
Articles were graded by the following criteria: grade I, postprocedure CT, hemorrhage described; II, postprocedure CT, hemorrhage not described; III, no postprocedure CT.
EVD, external ventricular drain; CT, computed tomography; EDH, epidural hematoma; SDH, subdural hematoma; IVH, intraventricular hemorrhage; ICH, intracranial
hemorrhage.

(102/1790) as the rate of hemorrhage and 0.61% (11/1790) as in patient selection, indications for external ventricular drain
a rate of significant hemorrhage appeared to be a simple average placement, and treatment protocol. In addition, we included 5
(total number of hemorrhage cases divided by total number of additional studies16-20 and excluded 2 studies,14,15 one in which
ventriculostomies). We think that a weighted average, as we ventriculostomy was performed by a non-neurosurgeon14 and
performed in our study, is a more rigorous and meaningful ap- a second in which external ventricular drains were placed in
proach. Performing calculations without giving any significance premature infants.15 Studies were only included in this meta-
to the relative weight of each study can skew the analysis so that it analysis if a neurosurgeon performed the ventriculostomies. We
is overly influenced by each study’s sample size. Forest plots
(Figures 3 and 4) included in our study indicate the magnitude
and significance of effects seen in each study and the relative rate
of the hemorrhage rates for each study. For example, the size of TABLE 2. Summary of Hemorrhage and Significant Hemorrhage
each square is directly proportional to the weight assigned to that Rates
study for calculating the cumulative rate, which in turn depends
Significant
on the sample size of the study. The location of the square
Hemorrhage, % Hemorrhage, %
depends on the magnitude of the rate of hemorrhage for that
study. The P value shown for each study is for a test of the null. All 16 studies 7 0.8
The cumulative rate is represented by a diamond, whose location (grades I, II, and III combined)
Combined grade I and 12.1 1.4
depends on the magnitude, and its width is determined by the grade II studies (10 studies)
precision. Binz et al did not report whether they used a fixed- Grade III studies 1.4 0.5
effects model or a random-effects model. We used the random- only (6 studies)
effects model in our study because we expected each study to vary

NEUROSURGERY VOLUME 69 | NUMBER 2 | AUGUST 2011 | 257

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.

BAUER ET AL

FIGURE 1. Forest plot of overall hemorrhage rate of all studies. CI, confidence interval.

also did not include the study that exclusively dealt with pre-
mature infants, because we believe that the high rate of bleeding
seen might be attributed to the immaturity of the brain
parenchyma.
We also evaluated the relationship between postoperative CT
scans and hemorrhage rates to see whether the group that un-
derwent routine CT scans postoperatively had a higher or lower
rate of hemorrhages than the group that did not undergo routine
postoperative CT scans. Nine of 10 studies (90%) (grade I and II
studies) in which a routine postoperative CT scan was performed
reported hemorrhage compared with 5 of 6 studies (83%) that
did not perform a routine postoperative CT scan but still reported
hemorrhage (grade III studies) (Table 1). To determine whether
the group that underwent routine postoperative CT scans actually

FIGURE 2. Forest plot of significant hemorrhage rate of all studies. CI, confidence interval.

258 | VOLUME 69 | NUMBER 2 | AUGUST 2011 www.neurosurgery-online.com

Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.


reported a higher percentage of studies with hemorrhage than the
group that did not undergo routine postoperative CT scans,
a comparison of these 2 groups (combined grade I and II studies
vs grade III studies) was performed using a z test; there was no
statistically significant difference between the 2 groups. It is also
important to note that the rate of hemorrhages and significant
hemorrhages in the group that underwent routine postoperative
CT scans (grades I and II) was 12.1% and 1.4%, respectively,
which was higher than similar rates in the group that did not
undergo routine CT scans (grade III) (1.4% and 0.5%, re-
spectively). It is likely that both minor and significant hemor-
rhages were missed in this second group of patients because of the
lack of routine postprocedure CT imaging. The initial group of
studies may represent a more accurate representation of the true
rates of hemorrhage.
We were surprised by the large difference in the rate of
hemorrhage and significant hemorrhage between grade I and
grade II studies. The rate of hemorrhage should have been similar
because routine imaging was performed on all patients. We at-
tribute the difference between these groups most likely to error in
study design. Because all studies were retrospective, it is possible
that studies specifically looking for hemorrhage may be more
likely to find hemorrhage, although this probably does not reflect
the entire difference. No conclusion can be made by this finding.
The best answer is probably somewhere between the hemorrhage
rates found in grade I and grade II studies.
Although the American Association of Neurological Surgeons
and the Congress of Neurological Surgeons online abstract ar-
chives were searched for the years 1995 through 2008, it is
possible that there were abstracts published before 1995 that we
were unable to locate. In addition, it is possible that we missed
abstracts published in print internationally despite our attempt to
locate them online. Publication bias was not assessed because we
believe that our sample size is too small to obtain an accurate
result.
Few studies have been performed that primarily looked at
hemorrhage as the main complication of ventriculostomy
placement. Many studies included in our analysis reported rates
of hemorrhage as a secondary endpoint to infection. Because
hemorrhage is a secondary endpoint, these articles usually pro-
vided few details regarding the hemorrhage. Characteristics such
as the volume of hemorrhage, patient symptomatology, and the
way in which the hemorrhage was diagnosed would have added
strength to our meta-analysis. Also, these studies included
patients of all age groups except premature infants, and the
indications for ventriculostomy in each study were mixed, which
could also influence results. Additional prospective studies eval-
uating postventriculostomy hemorrhage are warranted to better
understand this complication.
We attempted to look at other variables that we hypothesize
may make a difference in hemorrhage rate, including the location
of the ventriculostomy (ie, frontal or parietal), location where the
ventriculostomy was performed (operating room vs intensive care
unit), impact of neuronavigation, experience level of the operator,
NEUROSURGERY
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
META-ANALYSIS VENTRICULOSTOMY HEMORRHAGE
time of day of the procedure, and impact on coagulation status
and platelet function at the time of the procedure. Unfortunately,
we were unable to locate enough data on any of these variables to
make a statement on the significance of the variable. Further
study of the impact of these variables on the hemorrhage rate of
ventriculostomy would be useful in answering these questions.
We also attempted to look at infection as a primary endpoint.
Many studies did not use infection as a primary endpoint, and
this selection bias would preclude a rigorous meta-analysis of this
subject. Evaluating infection was difficult because many authors
did not explicitly define the criteria that they used to diagnose
infection. One group may believe that a single positive culture
defines infection. Another group may define infection as an
increase in white blood cell count in the cerebrospinal fluid. Also,
use of antibiotic impregnated ventriculostomy may add
additional difficulty in diagnostic criteria for infection. Further
prospective studies of infection of ventriculostomy would help
shed more light on this topic.
The overall limitation to our study is the lack of good-quality,
prospective, randomized trials evaluating hemorrhage as a com-
plication of ventriculostomy as a primary endpoint. Perhaps
future high-quality studies will provide us with more definitive
data.
CONCLUSION
A rigorous meta-analysis of 16 studies including 2328 patients
found a rate of hemorrhagic complication after ventriculostomy
placement by neurosurgeons of 7%. The rate of significant
hemorrhage was 0.8%. The rate of incidental hemorrhage is
probably slightly higher than 7% (perhaps approximating 12.1%,
the rate seen when just studies with routine postprocedure CT
scans were analyzed) because not every patient in every study in
the analysis underwent a postoperative CT scan. Further pro-
spective studies are warranted to produce definitive data.
Disclosure
The authors have no personal financial or institutional interest in any of the
drugs, materials, or devices described in this article.
REFERENCES
1. American Association of Neurological Surgeons. National Neurosurgical Procedural
Statistics. American Association of Neurological Surgeons Survey. Meadows, IL:
AANS; 2006.
2. Binz DD, Toussaint LG III, Friedman JA. Hemorrhagic complications of ven-
triculostomy placement: a meta-analysis. Neurocrit Care. 2009;10(2):253-256.
3. Anderson RC, Kan P, Klimo P, Brockmeyer DL, Walker ML, Kestle JR.
Complications of intracranial pressure monitoring in children with head trauma.
J Neurosurg. 2004;101(1 suppl):53-58.
4. Friedman WA, Vries JK. Percutaneous tunnel ventriculostomy: summary of 100
procedures. J Neurosurg. 1980;53(5):662-665.
5. Khanna RK, Rosenblum ML, Rock JP, Malik GM. Prolonged external ventricular
drainage with percutaneous long-tunnel ventriculostomies. J Neurosurg.
1995;83(5):791-794.
6. Roitberg BZ, Khan N, Alp MS, Hersonskey T, Charbel FT, Ausman JI. Bedside
external ventricular drain placement for the treatment of acute hydrocephalus. Br J
Neurosurg. 2001;15(4):324-327.
VOLUME 69 | NUMBER 2 | AUGUST 2011 | 259
BAUER ET AL
7. Maniker AH, Vaynman AY, Karimi RJ, Sabit AO, Holland B. Hemorrhagic
complications of external ventricular drainage. Neurosurgery. 2006;59(4 suppl 2):
ONS419–424; discussion ONS424-425.
8. Leung GK, Ng KB, Taw BB, Fan YW. Extended subcutaneous tunneling tech-
nique for external ventricular drainage. Br J Neurosurg. 2007;21(4):359-364.
9. Guyot LL, Dowling C, Diaz FG, Michael DB. Cerebral monitoring devices:
analysis of complications. Acta Neurochir Suppl. 1998;71:47-49.
10. Paramore CG, Turner DA. Relative risks of ventriculostomy infection and
morbidity. Acta Neurochir (Wien). 1994;127(1-2):79-84.
11. North B, Reilly P. Comparison among three methods of intracranial pressure
recording. Neurosurgery. 1986;18(6):730-732.
12. Narayan RK, Kishore PR, Becker DP, et al. Intracranial pressure: to monitor or not
to monitor? A review of our experience with severe head injury. J Neurosurg.
1982;56(5):650-659.
13. Wiesmann M, Mayer TE. Intracranial bleeding rates associated with two methods
of external ventricular drainage. J Clin Neurosci. 2001;8(2):126-128.
14. Ehtisham A, Taylor S, Bayless L, Klein MW, Janzen JM. Placement of external
ventricular drains and intracranial pressure monitors by neurointensivists. Neu-
rocrit Care. 2009;10(2):241-247.
15. Rhodes TT, Edwards WH, Saunders RL, et al. External ventricular drainage for
initial treatment of neonatal posthemorrhagic hydrocephalus: surgical and neu-
rodevelopmental outcome. Pediatr Neurosci. 1987;13(5):255-262.
16. Krotz M, Linsenmaier U, Kanz KG, Pfeifer KJ, Mutschler W, Reiser M. Eval-
uation of minimally invasive percutaneous CT-controlled ventriculostomy in
patients with severe head trauma. Eur Radiol. 2004;14(2):227-233.
17. O’Leary ST, Kole MK, Hoover DA, Hysell SE, Thomas A, Shaffrey CI. Efficacy
of the Ghajar Guide revisited: A prospective study. J Neurosurg. 2000;92(5):
801-803.
18. Gardner PA, Engh J, Atteberry D, Moossy JJ. Hemorrhage rates after external
ventricular drain placement. J Neurosurg. 2009;110(5):1021-1025.
19. Kakarla UK, Kim LJ, Chang SW, Theodore N, Spetzler RF. Safety and accuracy of
bedside external ventricular drain placement. Neurosurgery. 2008;63(1 suppl 1):
ONS162–166; discussion ONS166-167.
20. Huyette DR, Turnbow BJ, Kaufman C, Vaslow DF, Whiting BB, Oh MY.
Accuracy of the freehand pass technique for ventriculostomy catheter placement:
retrospective assessment using computed tomography scans. J Neurosurg.
2008;108(1):88-91.
21. Blaha M, Lazar D, Winn RH, Ghatan S. Hemorrhagic complications of in-
tracranial pressure monitors in children. Pediatr Neurosurg. 2003;39(1):27-31.
Acknowledgments
We thank Dr Beverly Walters for her guidance, expertise, and editorial as-
sistance on this article.
COMMENTS
V entriculostomy is one of the most common neurosurgical proce-
dures, but it is associated with a number of complications including
hemorrhage and infection. This article describes a meta-analysis of
hemorrhagic complications after ventriculostomy. A total of 2428 pa-
tients were identified from 16 studies with 203 patients described
as having a hemorrhage and 52 of these hemorrhages were deemed
260 | VOLUME 69 | NUMBER 2 | AUGUST 2011
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
‘‘significant.’’ The clinical implications of significant, however, require
further explanation. These are important data and open an argument as
to whether patients who require intracranial pressure monitoring should
undergo intraparenchymal sensor insertion as the primary procedure
with ventriculostomy reserved for those who require drainage of cere-
brospinal fluid as a therapeutic maneuver. Overall, the study was well
conducted and presented but raises a number of important points.
Peter J. Hutchinson
Cambridge, United Kingdom
T he authors provide a very comprehensive and well-executed meta-
analysis of studies of hemorrhage after external ventricular drain
placement. As expected, they found that the incidence of hemorrhage is
very low, especially the incidence of ‘‘clinically significant’’ hemorrhage.
Unfortunately, they were unable to determine what that term means in
many cases because many of the articles that they included in their
analysis did not define the term. Presumably those patients had neu-
rological signs or symptoms as a result of the hemorrhage. It is not clear,
however, what proportion required surgery. It also would be of interest to
know whether there is an association between intracranial pressure and
the risk of hemorrhage. My suspicion is that those with increased
intracranial pressure are less at risk of hemorrhage, but again, this
information was not available. This study is an important contribution
as a contemporary standard for the risk of ventriculostomy-related
hemorrhage.
Donald Marion
Washington, DC
T he authors presented their results of a meta-analysis evaluating the
hemorrhagic complications from ventriculostomy placement. The
authors have been extensive in their statistical analysis of the related and
identified questions of concern and have thus presented a thorough
review of the issue. This is the second meta-analysis on the subject and as
such has noted differences from the earlier meta-analysis in the number
of articles included, the use of weighted averages in the analyses, and
a novel grading system introduced by the authors. However, this article
remains unclear as to how it differs substantially in its conclusions from
the previously published article. In addition, there are several issues that
limit the generalizability of the conclusions made, including the defi-
nition of ‘‘significant’’ hemorrhage. Nonetheless, this article is well
presented and adds to the important ongoing discussions regarding the
use and associated risk of ventriculostomies.
Odette Harris
Stanford, California
www.neurosurgery-online.com