Prevention of Venous Thromboembolism in

Neurosurgery: A Metaanalysis
Jacob F. Collen, Jeffrey L. Jackson, Andrew F. Shorr and Lisa K. Moores

Chest 2008;134;237-249; Prepublished online July 18, 2008;

DOI 10.1378/chest.08-0023
The online version of this article, along with updated information
and services can be found online on the World Wide Web at:
http://chestjournal.org/cgi/content/abstract/134/2/237

CHEST is the official journal of the American College of Chest

Physicians. It has been published monthly since 1935. Copyright 2007
by the American College of Chest Physicians, 3300 Dundee Road,
Northbrook IL 60062. All rights reserved. No part of this article or PDF
may be reproduced or distributed without the prior written permission
of the copyright holder
(http://www.chestjournal.org/misc/reprints.shtml). ISSN: 0012-3692.

Downloaded from chestjournal.org on September 1, 2008

Copyright © 2008 by American College of Chest Physicians
 Original Research
PULMONARY EMBOLISM

Prevention of Venous
Thromboembolism in Neurosurgery*
A Metaanalysis
Jacob F. Collen, MD; Jeffrey L. Jackson, MD, MPH;
Andrew F. Shorr, MD, MPH, FCCP; and Lisa K. Moores, MD, FCCP

Background: Venous thromboembolism (VTE) is an important complication of neurosurgery.

Current guidelines recommend pharmacologic prophylaxis in this setting with either unfraction-
ated heparin or low-molecular-weight heparin (LMWH). We conducted a systematic review
asking, “Among patients undergoing neurosurgical procedures, how safe and effective is the
prophylactic use of heparin and mechanical devices?”
Methods: We searched the medical literature to identify prospective trials reporting on VTE
prevention (either mechanical or pharmacologic). The rates of VTE and bleeding were our
primary end points and were pooled using a random-effects model.
Results: We identified 30 studies reporting on 7,779 patients. There were 18 randomized
controlled trials and 12 cohort studies. The results of pooled relative risks (RRs) showed LMWH
and intermittent compression devices (ICDs) to be effective in reducing the rate of deep vein
thrombosis (LMWH: RR, 0.60; 95% confidence interval [CI], 0.44 to 0.81; ICD: RR, 0.41; 95% CI,
0.21 to 0.78). Similar results were seen when pooled rates from all 30 trials were analyzed. In
head-to-head trials, there was no statistical difference in the rate of intracranial hemorrhage
(ICH) between therapy with LMWH and nonpharmacologic methods (RR, 1.97; 95% CI, 0.64 to
6.09). The pooled rates of ICH and minor bleeding were generally higher with heparin therapy
than with non– heparin-based prophylactic modalities.
Conclusions: In a mixed neurosurgical population, LMWH and ICDs are both effective in the
prevention of VTE. Sensitivity analyses have suggested that isolated high-risk groups, such as
those with patients undergoing craniotomy for neoplasm, may benefit from a combination of
prophylactic methods, suggesting the need for a more individualized approach to these patients.
(CHEST 2008; 134:237–249)

Key words: hemorrhage; intracranial hemorrhages; neurosurgery; thromboembolism; thrombosis

Abbreviations: CI ⫽ confidence interval; CS ⫽ compression stocking; DVT ⫽ deep venous thrombosis; ICD ⫽ intermittent
compression device; ICH ⫽ intracranial hemorrhage; LMWH ⫽ low-molecular-weight heparin; PE ⫽ pulmonary embolism;
RCT ⫽ randomized controlled trial; RR ⫽ relative risk; UFH ⫽ unfractionated heparin; VTE ⫽ venous thromboembolism

N eurosurgery patients are at high risk of venous

thromboembolic events postoperatively, partic-
ularly patients undergoing intracranial surgery for
malignancy, the elderly, and those undergoing pro-
longed surgery. In patients undergoing elective pos-
terior lumbar spinal surgery, the known risk factors
include prolonged immobilization/bed rest, lengthy
operative procedures, prone positioning on frames
with flexion of the hips/knees, and distraction of the
spine (which may compress lower extremity venous
return).1,2 The current recommendations for throm-
boembolism prophylaxis include the use of intermit-
tent compression devices (ICDs) postoperatively,
with or without compression stockings (CSs), low-
dose unfractionated heparin (UFH) perioperatively,
or low-molecular-weight heparin (LMWH) postop-
eratively.3–7 However, neurosurgeons are concerned
about bleeding complications, and to date there has
been no systematic assessment of the data on phar-
macologic prophylaxis efficacy and safety in neuro-
surgery and spinal surgery patients. Our purpose was
to conduct a systematic review to answer the ques-
tion, “Among patients undergoing neurosurgical pro-
cedures, what is the relative efficacy of LMWH,

www.chestjournal.org CHEST / 134 / 2 / AUGUST, 2008 237

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
UFH, and mechanical devices in preventing throm-
boembolism, and what are the relative bleeding
complications?”
Materials and Methods
Literature Search
While our goal was to analyze only randomized controlled trials
(RCTs) that were head-to-head comparisons of different methods
of prophylaxis, in order to systematically retrieve the literature
from around the world, we also searched for prospective cohort
trials of venous thromboembolism (VTE) prophylaxis. Two inves-
tigators, with the assistance of a medical librarian, independently
searched the published literature (from 1960 through August
2007) to identify published RCTs and prospective clinical trials of
VTE prophylaxis in neurosurgical patients, using either pharma-
cologic or mechanical methods. The search was not limited to
studies published in the English language. The following data-
bases were searched: MEDLINE; PubMed; Cochrane RCT;
Embase; Biosis; PASCAL; Sci Search; IPA; and Computer
Retrieval of Information on Scientific Projects. The search terms
included “neurosurgery,” “neurosurgical procedures,” “thrombo-
embolism,” “thromboprophylaxis,” “heparin,” “Lovenox,” and
“enoxaparin.” Full-text articles of all potentially appropriate
studies were reviewed, and a hand search of the bibliographies of
each retrieved article was conducted.
Study Selection Criteria
The inclusion criteria included the following: (1) a randomized
trial or prospective cohort study evaluating pharmacologic VTE
prophylaxis (with UFH or LMWH); or (2) a randomized trial or
prospective cohort study evaluating mechanical VTE prophylaxis
(with ICDs or CSs); (3) an objective assessment of deep venous
thrombosis (DVT) [ie, with Doppler compression sonography,
impedence plethysmography, radiofibrinogen uptake scanning,
autopsy, or venography] and pulmonary embolism (PE) [ie, with
CT angiography, ventilation perfusion scanning, or pulmonary
angiogram, or by autopsy] with the reporting of incidences; and
(4) a neurosurgical population. Studies were excluded if they
were not primarily related to neurosurgery patients, were not
prospective studies, were not specifically about VTE prophylaxis,
or if they were articles primarily about patients with penetrating
or closed head injuries, spinal cord injuries, or stroke.
*From the Department of Medicine (Dr. Collen), Walter Reed
Army Medical Center, Washington, DC; the Uniformed Services
University of the Health Sciences (Drs. Jackson and Moores),
Bethesda, MD; the Department of Pulmonary Medicine (Dr.
Shorr), Washington Hospital Center, Washington, DC.
The opinions or assertions herein are the private views of the
authors and are not to be construed as reflecting the views of the
Department of the Army or the Department of Defense.
The authors have reported to the ACCP that no significant
conflicts of interest exist with any companies/organizations whose
products or services may be discussed in this article.
Manuscript received January 4, 2008; revision accepted March
12, 2008.
Reproduction of this article is prohibited without written permission
from the American College of Chest Physicians (www.chestjournal.
org/misc/reprints.shtml).
Correspondence to: Jacob F. Collen, MD, 1672 North Twenty-
First St, Apartment 7, Arlington, VA 22209; e-mail: Jcollen2002@
hotmail.com
DOI: 10.1378/chest.08-0023
238
Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
Study Quality Assessment
RCTs were rated for eight elements of quality using the
methods of Jadad et al.8 In addition, we created a quality
assessment tool that was adapted from the criteria of McMaster9
for evaluating the validity of studies about prognosis, which we
also used for the evaluation of RCTs in addition to the prospec-
tive cohort trials. Studies were assessed for the presence of the
following eight features: a description of the characteristics of
the patient sample; a description of the inclusion and exclusion
criteria; potential selection bias; the completeness of the
follow-up; a description of the reasons for incomplete follow-up;
the definition of outcomes stated at the start of the study; and the
objectivity of outcomes. Two raters independently assessed qual-
ity (Jadad et al criteria: ␬, 0.56; McMaster criteria: ␬, 0.37;
p ⫽ 0.005). Disagreements were resolved by consensus.
Data Extraction
We extracted the following data from each study: study design
(ie, efficacy, safety, and prospective, randomized, or double-
blind); exclusion criteria; patient demographics (ie, number of
patients, sex, age, and type of neurosurgical intervention); DVT
prophylaxis modality (ie, mechanical [ICDs or CSs]; or pharma-
cologic [LMWH or UFH]); the method of DVT diagnosis (ie,
fibrinogen scanning, venography, or ultrasound); the method of
PE diagnosis (ie, CT scan, pulmonary angiogram, or ventilation
perfusion scan); the length of follow-up; and the number of
patients lost to follow-up and the reasons for being lost to
follow-up. The outcomes assessed were DVT, PE, minor bleed-
ing events, major bleeding events, intracranial hemorrhage
(ICH), reoperation for bleeding and deaths, and whether deaths
were study related.
Statistical Analysis
Our primary goal was to pool the relative risks (RRs) from RCTs
that included head-to-head comparisons of different modalities
of prophylaxis. These RRs were pooled using the DerSimonian
and Laird10 random-effects method. Because there were rela-
tively few such trials, and in order to exhaustively synthesize the
literature, we also calculated the overall rates as well as the
annualized rates from the data provided in each article for all
trials, including both prospective cohort trials and RCTs. The
variance for each outcome was calculated using exact binomial
methods11 and were pooled using a random-effects model.10 For
both RRs and rates, pooled heterogeneity was assessed visually
with Galbraith plots,12 Q statistics (␹2 test),13 and the I2 statistic.
Studies with an I2 statistic of 25 to 50% are considered to have
low heterogeneity, those with an I2 statistic of 50 to 75% to have
moderate heterogeneity, and those with an I2 statistic of ⬎ 75%
to have a high degree of heterogeneity.14 Publication bias was
assessed visually using funnel plots as well as statistically using the
methods of Begg and Berlin,15 Egger et al,16 and Duvall and
Tweedie.17 In addition, we performed a sensitivity analysis,
assessing the effects of study quality and various other study
characteristics using stratified analysis and metaregression. A
random-effects metaregression was used to adjust for the poten-
tial differences between studies. All analyses were performed
using a statistical software package (Stata, version 9.2; StataCorp;
College Station, TX).
Results
We identified 2,520 potential studies in our liter-
ature search. We excluded 2,490 studies, leaving 30
Original Research
 Online Literature Search conducted using key words:
Thromboembolism, Thromboprophylaxis, Neurosurgery,
Neurosurgical procedures,Low Molecular Weight Heparin,
And Heparin
Searched the following Databases:
PUBMED, SciSearch, Cochrane RCT,
Embase, Biosis, Pascal, MEDLINE, CRISP
The search was limited to RCT and
Clinical Trials, but no limits with regards to
age, publication type, language
2520 initial articles located online
2064 articles excluded because unrelated to Neurosurgery
142 articles excluded because not RCT or prospective trials
133 articles excluded because not related to VTE prophylaxis
65 articles excluded because dealt with cranial or spinal cord injury or stroke
86 duplicate articles excluded
30 articles remained for inclusion
Figure 1. Article search and selection process.
studies (Fig 1). These 30 studies reported on 7,779
patients who were undergoing neurosurgical proce-
dures (Table 1). Eleven trials18 –28 assessed LMWH, 12
trials21,23,27–36 assessed UFH, 18 trials2,19,21–23,33,35– 46
assessed ICD, and 10 trials18,20,25,27,28,37,41,44–46 assessed
CS. Nine articles18,22,23,25,27,28,30,33,37 reported both safety
and efficacy, 8 articles20,24,26,29,32,34–36 assessed safety, and
13 articles2,19,21,31,36,38–44,46 assessed efficacy. Of the 30
studies, 18 were RCTs18,19,21–23,25,27,28,30–32,34,36,37,41,42,44,46
(6 of these were double-blinded studies18,21,25,28,32,34)
and 12 were cohort studies.2,20,24,26,29,33,35,38 – 40,43,45
The timing of chemical prophylaxis differed across
studies (Table 2), as follows: nine studies19,27–32,34,36
utilized preoperative prophylaxis; eight stud-
ies18,20 –22,24 –26,33 utilized postoperative prophylaxis;
and two studies23,35 utilized intraoperative prophy-
laxis. Of these, preoperative prophylaxis was used in
one trial using LMWH alone,19 in six studies using
UFH alone,29 –32,34,36 and in two studies comparing
both methods. Postoperative prophylaxis was used in
one study using UFH alone,33 in seven studies using
LMWH alone,18 –20,22,24 –26 and in one study21 com-
paring both methods. Intraoperative prophylaxis was
used in two studies, one with both methods,23 and
one with UFH alone.25
Most of the studies we reviewed had problems
with quality. Of the RCTs, the average Jadad score
was 5.81 (range, 2 to 8). Only 25% of studies had
www.chestjournal.org
Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
concealed allocation; 44% were blind. On the prog-
nosis quality rating scheme, 93% had an adequate
outcome definition, 80% had an unbiased sample,
80% had adequate follow-up, 70% had adequate
inclusion/exclusion criteria, and 53% had an unbi-
ased selection process.
Of the 30 included articles, 18 were RCTs and 12
were prospective cohort studies. Among the 18
RCTs, 4 directly compared UFH to LMWH, 2
RCTs21,23 were conducted in craniotomy patients,
and 2 RCTs27,28 were conducted among spinal sur-
gery patients. Two RCTs18,25 compared LMWH to
CSs, mostly in craniotomy patients. Two RCTs19,22
compared LMWH to ICDs, also primarily in crani-
otomy patients. Three RCTs31,32,34 compared UFH
to placebo; 2 of these RCTs31,32 looked at craniotomy
patients and one RCT34 looked at spinal patients.
Seven of the RCTs evaluated mechanical prophy-
laxis. Five of these RCTs compared ICDs to CSs;
three of these RCTs37,44,45 were conducted mostly in
craniotomy patients, and two RCTs41,46 were exclu-
sively in spinal patients. One RCT42 compared ther-
apy with ICDs to placebo in a mixed neurosurgical
population. Another RCT30 compared the use of
ICDs to electrical calf stimulation, also in a mixed
population, and another RCT46 compared thigh
ICDs to foot ICDs in spinal patients. These latter
two studies30,46 were pooled as prospective cohorts.
Among the 12 prospective cohorts, 7 involved
pharmacologic prophylaxis and 5 involved mechani-
cal prophylaxis. Two articles24,26 looked at therapy
with LMWH alone in craniotomy patients. Four
articles studied UFH, one alone29 and three in
combination with ICDs,33,35,36 also primarily in cra-
niotomy patients.
There were five articles2,38 – 40,43 on mechanical
prophylaxis alone. These included cohort studies of
ICDs alone38 – 40 all in spinal cases, and ICDs and
CSs together,2,43 also primarily in spinal patients.
Efficacy
The results of the head-to-head pooled RRs are
given in Table 3 and Figure 2. For DVT prophylaxis,
ICDs were statistically better than placebo (RR,
0.41; 95% confidence interval [CI], 0.21 to 0.78), and
LMWH was better than CS (RR, 0.60; 95% CI, 0.44
to 0.81). None of the other head-to-head compari-
sons were statistically significant for reducing DVTs
or PEs. When the rates of all trials, including
prospective trials, were pooled, the five trials that
managed patients perioperatively without prophy-
laxis had high rates of DVT formation (15.5 per 100
patients; 95% CI, 3.9 to 27.2; n ⫽ 5; Q statistic, 32.1;
I2 statistic, 87.5%) [Fig 3]. CSs minimally reduced
this rate (11.6 per 100 patients; 95% CI, 3.4 to 19.8;
CHEST / 134 / 2 / AUGUST, 2008 239
 Table 1—Included Studies

240
Minor
Craniotomy
Efficacy Male and Other Neoplasm Vascular
Study/Year vs Age, Patients, Gender, Craniotomy, Craniotomy, Craniotomy, Spine, Quality
(Country-Language) Intervention Design Safety yr No. % No. No. No. No. DVT Dx PE Dx Score Quality Problems
18
Agnelli et al /1998 LMWH vs placebo Double-blind Both 56 307 50 0 299 0 46 V, US A, V̇/Q̇ 8 None
(Italy-English) (all with CSs) PCT
Barnett et al29/1977 UFH Cohort Safety NS 150 NS 3 36 3 98 NS NS 3 Sample definition;
(United States- inclusion/exclusion
English) criteria; selection bias;
outcome definition
Bostrom et al30/1986 UFH vs electrical RCT Both 60 104 54 0 39 14 25 F, V NS 6 Follow-up
(Sweden-English) calf stimulation
Bucci et al37/1989 Mechanical (ICD RCT Both NS 70 NS 0 39 20 0 V A 6 Sample definition;
(United States- vs CS) selection bias
English)
Bynke et al38/1987 Mechanical (ICD) Cohort Efficacy 59 31 52 0 31 0 0 F, V NS 6 Inclusion/exclusion
(Sweden-English) criteria; selection bias
Cerrato et al31/1978 UFH vs placebo RCT Efficacy 52 100 51 0 100 0 0 F A 5 Inclusion/exclusion
(Italy-English) criteria; selection bias;
outcomes discussion
Constantini et al32/ UFH vs placebo Double-blind Safety 56 103 47 0 103 0 0 NS NS 8 None
2001 (Israel- PCT
English)
Dickinson et al19/ LMWH vs RCT Efficacy 47 66 NS 9 66 0 0 US NS 8 None
1998 (United mechanical
States-English) (ICD/CS vs
LMWH/CS vs
LMWH/CS/
ICD)
Epstein39/2005 Mechanical (ICD) Cohort Efficacy 50 200 61 0 0 0 200 US CT 6 Inclusion/exclusion
(United States- criteria; selection bias
English)
Epstein40/2006 Mechanical (ICD) Cohort Efficacy 53 139 56 0 0 0 139 US CT 6 Inclusion/exclusion
(United States- criteria; selection bias

Downloaded from chestjournal.org on September 1, 2008

English)

Copyright © 2008 by American College of Chest Physicians

Frim et al33/1992 UFH (with ICD) Cohort Both NS 138 NS 0 16 23 26 V, US A, V̇/Q̇ 5 Sample definition;
(United States- inclusion/exclusion
English) criteria; selection bias
Gerlach et al20/2003 LMWH (and CSs) Cohort Safety NS 2,823 NS 1,504 584 438 0 V, US NS 5 Sample definition;
(Germany- inclusion/exclusion
English) criteria; selection bias
Goldhaber et al21/ UFH vs LMWH Double-blind Efficacy 47 150 53 11 139 0 0 US CT 7 Follow-up
2002 (United (all with ICD PCT
States-English) and CSs)
Gruber et al34/1984 UFH vs placebo Double-blind Safety 46 50 50 0 0 0 50 V A 7 Selection bias
(Switzerland- PCT
English)
(Continued)

Original Research
 Table 1—Continued

Minor
Craniotomy
Efficacy Male and Other Neoplasm Vascular
Study/Year vs Age, Patients, Gender, Craniotomy, Craniotomy, Craniotomy, Spine, Quality
(Country-Language) Intervention Design Safety yr No. % No. No. No. No. DVT Dx PE Dx Score Quality Problems

www.chestjournal.org
22
Kurtoglu et al / LMWH vs RCT Both 37 120 39 0 0 119 11 US CT 8 None
2004 (Turkey- mechanical
English) (ICD)
MacDonald et al35/ UFH (with ICD Cohort Safety 49 106 48 0 58 38 0 V, US A, V̇/Q̇ 8 Follow-up
1999 (United and CSs)
States-English)
MacDonald et al23/ UFH vs LMWH RCT Both 50 97 47 0 63 24 0 US A, CT, 7 Follow-up
2003 (United (all with ICD) V̇/Q̇
States-English)
Nelson et al41/1996 Mechanical (ICD/ RCT Efficacy 52 117 48 0 0 0 117 U NS 6 Selection bias; follow-up
(United States- CS vs CS)
English)
Norwood et al24/ LMWH Cohort Safety 40 150 NS 25 NS NS NS US A, CT 8 None
2002 (United
States-English)
Nurmohamed et LMWH vs placebo Double-blind Both 52 485 54 0 400 51 7 V, US A, V 6 Follow-up
al25/1996 (all with CSs) PCT
(Netherlands-
English)
Paoletti et al26/1989 LMWH Cohort Safety 48 97 62 0 62 26 0 U, V NS 8 None
(France-French)
Prestar27/1992 UFH vs LMWH RCT Both 43 200 30 0 0 0 200 V V̇/Q̇ 8 None
(Germany- (all with CSs)
German)
Rokito et al2/1996 Mechanical (ICD Cohort Efficacy 45 75 NS 0 0 0 75 U NS 7 Follow-up
(United States- and CSs)
English)
Skillman et al42/1978 Mechanical vs RCT Efficacy 50 95 NS 0 31 7 50 F, V NS 5 Sample definition;
(United States- placebo (ICD vs inclusion/exclusion
English) placebo) criteria; selection bias

Downloaded from chestjournal.org on September 1, 2008

Copyright © 2008 by American College of Chest Physicians
Smith et al43/1994 Mechanical (ICD Cohort Efficacy 41 317 NS 0 0 0 317 US NS 6 Inclusion/exclusion
(United States- and CSs) criteria; selection bias
English)
Turpie et al44/1989 Mechanical vs RCT Efficacy 50 239 60 0 117 54 14 F, V NS 8 None
(Canada-English) placebo (ICD/
CS vs CSs vs
placebo)
Voth et al28/1992 UFH vs LMWH Double-blind Both 53 179 44 0 0 0 179 F NS 7 Selection bias
(Germany- (all with CSs) PCT
English)
Wautrecht et al45/ Mechanical (ICD/ RCT Efficacy NS 23 NS 0 23 0 0 V NS 8 None

CHEST / 134 / 2 / AUGUST, 2008

1996 (Belgium- CS vs CS)
English)
(Continued)

241
 n ⫽ 7; Q statistic, 118.3; I2 statistic, 94.9%). Among

*PCT ⫽ prospective controlled trial; NS ⫽ not stated; US ⫽ ultrasound; CT ⫽ CT angiogram; A ⫽ angiogram; V ⫽ venogram; V̇/Q̇ ⫽ ventilation-perfusion scan; F ⫽ 125I fibrinogen scan; Dx ⫽ diagnosis.
sample; selection bias
the 12 trials with patients managed exclusively with

Quality Problems

Lack well-defined
ICDs, the DVT rate was 1.9 per 100 patients (95%
CI, 0.6 to 3.3; n ⫽ 12; Q statistic, 20.51; I2 statistic,
46.4%). The pooled rate of DVT for studies using
LMWH was 4.1 per 100 patients (95% CI, 2.0 to 6.1;

None
n ⫽ 11; Q statistic, 77.1; I2 statistic, 85.7%), and for
Quality UFH it was 0.9 per 100 patients (95% CI, 0.0 to 1.8;
Score n ⫽ 12; Q statistic, 20.9; I2 statistic, 47.5%). There
6

8
were no statistical differences when heparin therapy
PE Dx

was combined with ICDs or CSs. The pooled rate of

NS

NS
DVT for LMWH alone was 0.98 per 100 patients
(95% CI, 0.0 to 2.8; n ⫽ 3; Q statistic, 2.8; I2 statistic,
DVT Dx

29%), the rate of DVT for LMWH combined with

NS

US

ICDs was 5.7 per 100 patients (95% CI, 2.6 to 8.8;
n ⫽ 9; Q statistic, 73.6; I2 statistic, 49%), the rate of
Spine,
No.

134
NS

DVT for UFH alone was 3.3 per 100 patients (95%
CI, 0.0 to 7.2; n ⫽ 5; Q statistic, 8.2; I2 statistic,
Craniotomy,

51%), and the rate of DVT for UFH combined with

Vascular

No.

NS

ICDs was 1.5 per 100 patients (95% CI, 0.0 to 3.1;
0

n ⫽ 7; Q statistic, 12.2; I2 statistic, 51%).

Pulmonary emboli were uncommon, and the pooled
Craniotomy,
Neoplasm

rates were similar between the different therapeutic

No.

NS

modalities. Patients receiving placebo had a PE rate

0

of 0.22 per 100 patients (95% CI, 0.00 to 1.81; n ⫽ 5;

Table 1—Continued

Q statistic, 1.3; I2 statistic, 0%), patients using ICDs

Craniotomy,
Craniotomy
and Other

had a pooled PE rate of 0.32 per 100 patients (95%

Minor

No.

CI, 0.0 to 0.81; Q statistic, 4.0; I2 statistic, 0%). For

0

LMWH therapy, the rate of PE was 0.11 to 100

patients (95% CI, 0 to 0.25; n ⫽ 11; Q statistic, 3.6;
Gender,

I2 statistic, 0%), and for UFH therapy the rate was

Male

NS

58
%

0.29 per 100 patients (95% CI, 0.00 to 0.23; n ⫽ 12;

Q statistic, 3.06; I2 statistic, 0%). None of these PE
Patients,

rates were significantly different. Similarly, there

No.

872

134

were no differences in PE rates when heparin

therapy was combined with ICDs or CSs. In a pooled
Age,

NS

40
yr

analysis of three studies37,41,45 comparing the efficacy

of CSs to mechanical compression devices (ie, ICDs)
Efficacy

Efficacy
Safety

Safety

for the prevention of DVT, there was no difference

vs

in the pooled RR (RR, 0.81; 95% CI, 0.32 to 1.78)

[Table 3], though the results tended to favor ICDs,
Design

which is consistent with the differences found when

Cohort

the rates from all studies were included.

RCT

Safety
ICD and CSs vs
Mechanical (thigh
UFH (with ICD

foot ICD and

Intervention

There were four RCTs18,19,22,25 comparing LMWH

and CSs)

to nonpharmacologic management, four RCTs21,23,27,28

CSs)

comparing LMWH to UFH, and three RCTs31,32,34

comparing UFH and placebo. None of the pooled
RRs for ICH, minor bleeding, major bleeding, or
(Country-Language)

Wood et al46/1997
(United States-
States-English)

death were statistically significant (Table 4, Fig 4),

Wen and Hall /
1988 (United
Study/Year

36

though the heparin modalities tended to have higher

English)

rates of ICH and minor bleeding.

The pooled rates (Fig 5) demonstrated that pa-
tients managed perioperatively without heparin had

242 Original Research

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
 Table 2—Timing of Pharmacologic Prophylaxis*
LMWH Agent, Dose, UFH Dose and Timing of Prophylaxis
Study/Year and Frequency Frequency Dose in Relation to Surgery Comment
23
MacDonald et al /2003 Dalteparin qd 5,000 U bid Intraoperatively With induction anesthesia
MacDonald et al35/1999 5,000 U bid Intraoperatively With induction anesthesia
Goldhaber et al21/2002 Enoxaparin 40 mg qd 5,000 U bid Postoperatively Postoperative day 1
Kurtoglu et al22/2004 Enoxaparin 40 mg qd Postoperatively 24 h after hospital admission
Agnelli et al18/1998 Enoxaparin 40 mg qd Postoperatively Within 12–24 h after surgery
Dickinson et al19/1998 Enoxaparin 30 mg bid Preoperatively Before anesthesia
Frim et al33/1992 5,000 U bid Postoperatively Postoperative day 1
Gerlach et al20/2003 Nadroparin qd Postoperatively 24 h after surgery
Wen and Hall36/1988 5,000 U bid Preoperatively Evening prior to surgery
Norwood et al24/2002 Enoxaparin 30 mg bid Postoperatively 24 h after surgery
Constantini et al32/2001 5,000 U bid Preoperatively 2 h prior to surgery
Barnett et al29/1977 5,000 U bid Preoperatively With preoperative medications
Bostrom et al30/1986 5,000 U bid Preoperatively 2 h prior to surgery
Paoletti et al26/1989 Fraxiparin qd Postoperatively 24 h after surgery
Cerrato et al31/1978 5,000 U bid Preoperatively 2 h prior to surgery
Voth et al28/1992 1,500 U APTT LMWH qd 5,000 U bid Preoperatively 2 h prior to surgery
Gruber et al34/1984 2,500 U bid Preoperatively 2 h prior to surgery
Prestar27/1992 1,500 U APTT LMWH qd 5,000 U tid Preoperatively Evening prior to surgery
Nurmohamed et al25/1996 Nadroparin qd Postoperatively 18–24 h after surgery
*APTT ⫽ activated partial thromboplastin time.

low rates of intracranial bleeding (0.04 per 1,000
patients; 95% CI, 0.00 to 3.7; n ⫽ 19; Q statistic, 1.9;
I2 statistic, 0%). ICH rates among patients receiving
LMWH was 1.52 per 1,000 patients (95% CI, 1.09 to
1.94; n ⫽ 12; Q statistic, 17.18; I2 statistic, 36%) and
that among patients receiving UFH was 0.35 per
1,000 patients (95% CI, 0.00 to 7.4; n ⫽ 12; Q
statistic, 6.12; I2 statistic, 0%). Patients receiving
LMWH had statistically significantly higher bleeding
rates than those receiving therapy with mechanical
modalities (p ⬍ 0.0005), though patients receiving
UFH did not have higher rates (p ⫽ 0.40) Minor
bleeding was more common, and its incidence was
statistically greater than zero for all modalities, in-
cluding placebo (p ⬍ 0.001 for all). Patients not
receiving heparin had minor bleeding at a rate of 1.7
per 1,000 patients (95% CI, 0.0 to 6.1; n ⫽ 19; Q
statistic, 21.0; I2 statistic, 19%), while the rate with
LMWH therapy was 12.3 per 1,000 patients (95%
CI, 0.22 to 24.4; n ⫽ 11; Q statistic, 38.2; I2 statistic,
74%) and with UFH therapy it was 5.3 per 1,000
(95% CI, 0 to 14.6; n ⫽ 12; Q statistic, 23.9; I2
statistic, 53%). There were no statistical differences
in the rates of minor bleeding.
Major bleeding rates were low. For patients not
receiving heparin, the rate of major bleeding was
0.59 per 1,000 patients (95% CI, 0.00 to 0.57;
n ⫽ 17; Q statistic, 0.97; I2 statistic, 0%), for patients
receiving UFH the rate was 0.82 per 1,000 patients
(95% CI, 0.0 to 2.1; n ⫽ 11; Q statistic, 1.93; I2
statistic, 0%), and for patients receiving LMWH the
rate was 0.16 per 1,000 patients (95% CI, 0.0 to 0.67;
n ⫽ 10; Q statistic, 1.58; I2 statistic, 0%). None of
these major bleeding rates were statistically different

Table 3—Efficacy (Pooled RRs of RCTs)*

Comparisons Studies, No. DVT PE

ICD vs CSs 3 0.81 (0.32–1.78) 0.49 (0.08–2.85)

关4.05; 26%兴 关0.26; 0%兴
ICD vs placebo 2 0.41 (0.21–0.78) 0.37 (0.03–4.06)
Q ⫽ 0.18, I2 ⫽ 0% 关0.42; 0%兴
LMWH vs CSs 2 0.60 (0.44–0.81) 0.29 (0.05–1.85)
Q ⫽ 0.48, I2 ⫽ 0% 关0.55; 0%兴
LMWH vs ICD 2 0.79 (0.30–2.12) 1.62 (0.35–7.46)
Q ⫽ 0.01, I2 ⫽ 0% Q ⫽ 0.40, I2 ⫽ 0%
LMWH vs UFH 4 1.46 (0.61–3.51) 0.43 (0.08–2.41)
Q ⫽ 2.32, I2 ⫽ 0% 关0.67; 0%兴
UFH vs placebo 3 0.50 (0.11–2.38) 0.96 (0.10–9.06)
Q ⫽ 3.98 I2 ⫽ 49% 关0.0; 0%兴
*Values are given as the mean (95% CI) 关Q statistic; I2 statistic兴.

www.chestjournal.org CHEST / 134 / 2 / AUGUST, 2008 243

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
 Efficacy: DVT
Favors LMWH Favors UFH Favors LMWH Favors CS
Risk ratio

Study (95% CI) % We

Goldhaber
Macdonald Agnelli
agnelli 0.53 ( 0.33, 0.84) 4

Prestar Nurmohamed
nurmohamed 0.66 ( 0.44, 0.98) 5

Voth
Overall
Overall 0.60 ( 0.44, 0.81) 10
Overall

0 1 100 .331245 1 3.01890

Risk ratio Risk ratio

LMWH vs UFH LMWH vs CS

Favors ICD Favors Placebo Favors UFH Favors Placebo
Risk Ratio (95% CI)

Skillman Cerrato

Constantini
Turpie
Gruber
Overall
Overall

0 1 8
0 1 70
Risk ratio
Risk ratio

ICD vs Placebo UFH vs Placebo

Figure 2. Peto plot of pooled RRs for DVT from RCTs.

than zero (nonheparin, p ⫽ 0.77; UFH, p ⫽ 0.93;
LMWH, p ⫽ 0.96) or differed from each other.
Sensitivity Analysis
We performed a sensitivity analysis exploring the
potential effects of several variables on our results,
including the year of publication, the type of study
(ie, prospective cohort vs controlled trial), the type of
procedure, the underlying disease (ie, neoplasm,
vascular, complex spine, or other), the length of
follow-up, the mean age, lost to follow-up, gender,
the timing of heparin administration, and the study
quality. Prospective cohort trials reported lower rates
of DVT than randomized trials (cohort trials: RR,
1.11; 95% CI, 0.23 to 1.99; RCTs: RR, 7.6; 95% CI,
5.5 to 10.01; p ⫽ 0.03) but had no differences in
bleeding rates. Older patients and those undergoing
procedures for treatment of neoplasms also had
higher rates of DVT formation. Patients receiving
heparin prophylaxis preoperatively had statistically
lower (p ⬍ 0.001) rates of DVT formation than those
patients receiving their first dose postoperatively
(preoperative dosing: DVT rate, 3.3 per 100 patients;
95% CI, 1.2 to 5.5; n ⫽ 11; Q statistic, 29.3; I2
statistic, 65.8%; postoperative dosing: DVT rate, 5.4
per 100 patients; 95% CI, 2.4 to 8.4; n ⫽ 10; Q
statistic, 87.2; I2 statistic, 89.7%). Patients undergo-
ing the procedure for vascular procedures had higher
rates of ICH. There was no difference in the rates of
ICH between heparin being administered preoper-
atively (0.28 per 100 patients; 95% CI, 0.0 to 0.59),
intraoperatively (0.39 per 100 patients; 95% CI, 0.0
to 1.25), or postoperatively (0.77 per 100 patients;
95% CI, 0.16 to 1.4). None of the variables were
related to our findings for major or minor bleeding
or PE rates. In addition, there was no effect of
quality ratings on the results we report for ICH,
major or minor bleeding, DVT, or PE, with either
the Jadad or the prospective cohort rating scale.
After adjusting for age, neoplasm, and study de-
sign, both heparin modalities (LMWH, p ⫽ 0.02;
UFH, p ⫽ 0.001) and ICDs (p ⫽ 0.04) had signifi-
cantly lower rates of DVT formation than patients
given either CS or placebo. There were no statistical
differences, even before adjustment between the
rates of DVT among LMWH, UFH, or ICD therapy.
After adjustment, patients receiving LMWH had a
higher incidence of ICH than those receiving non-
heparin modalities.
Discussion
In the general surgery and orthopedic patient
populations, pharmacologic prophylaxis has been

244 Original Research

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
 Rate/100 (95% CI)
Cerrato (1978) 34.0 (21.2-48.8)
Constantini (2001) 4.2 (0.5-14.3)
Placebo Gruber (1984) 0.0 (0.0-13.7)
Skillman (1978) 25.0 (13.6-39.6)
Turpie (1989) 19.8 (11.7-30.1)
Subtotal 15.5 (3.9-27.2)
Agnelli (1998) 27.4 (20.4-35.0)
Bucci (1989) 0.0 (0.0-10.9)
Nelson (1996) 0.0 (0.0-3.1)
Compression Normohamed (1996) 20.9 (15.9-26.6)
Stockings Rokito (1996) 0.0 (0.0-8.4)
Turpie (1989) 8.8 (3.7-17.6)
Wautrecht (1996) 40.0 (23.2-65.5)
Subtotal 11.6 (3.4-19.8)
Bucci (1989) 3.1 (0.1-13.8)
Dickinson (1998) 13.6 (2.9-34.9)
Epstein (2005) 4.0 (1.7-7.7)
Epstein (2006) 2.9 (0.8-7.2)
Intermittent Kurtoglu (2004) 6.7 (1.9-16.2)
Nelson (1996) 0.0 (0.0-3.1)
Compression Rokito (1996) 0.0 (0.0-10.6)
Skillman (1978) 8.3 (2.3-19.9)
Device Smith (1994) 0.6 (0.08-2.3)
Turpie (1989) 8.9 (3.6-17.2)
Wautrecht (1996) 0.0 (0.0-14.8)
Wood (1997) 0.8 (0.02-4.0)
Subtotal 1.9 (0.6-3.3)

Agnelli (1998) 14.4 (9.2-20.9)

Dickinson (1998) 11.4 (3.8-24.6)
Gerlach (2003) 0.14 (0.04-0.4)
Low Goldhaber (2002) 12.0 (5.6-21.6)
Molecular Kurtoglu (2004) 5.0 (1.0-13.9)
MacDonald (2003) 4.1 (0.5-13.9)
Weight Normuhomed (1996) 13.7 (9.6-18.7)
Heparin Norwood (2002) 1.3 (0.2-4.7)
Paoletti (1989) 0.0 (0.0-3.7)
Prestar (1992) 0.0 (0.0-3.6)
Voth (1992) 1.1 (0.03-6.2)
Subtotal 4.1 (2.0-6.1)

Barnett (1977) 0.0 (0.0-2.4)

Bostrom (1986) 10.2 (3.4-22.2)
Cerrato (1978) 6.0 (1.3-16.6)
Constantini (2001) 3.6 (0.4-12.5)
Frim (1992) 0.0 (0.0-2.6)
Unfractionated Goldhaber (2002) 6.6 (2.2-14.9)
Heparin Gruber (1984) 4.0 (0.1-20.4)
MacDonald (2003) 0.0 (0.0-7.4)
Macdonald (1999) 6.6 (2.7-13.1)
Prestar (1992) 1.0 (0.03-5.5)
Voth (1992) 2.2 (0.3-7.6)
Wen (1988) 0.0 (0.0-0.4)
Subtotal 0.9 (0.00-1.84)

0 65

Figure 3. Peto plot of overall DVT rates (all studies).

proven to safely and significantly reduce VTE. Ac-
cepted methods include mechanical prophylaxis, but
there is a clear emphasis toward pharmacologic
methods (ie, UFH and LMWH), with increasing
emphasis on LMWH and fondaparinux.6 Current
guidelines extrapolate these data to make similar
recommendations in neurosurgery patients. Based
on published surveys of practicing neurosurgeons,
however, attempts to apply these recommendations
have been met with resistance, primarily due to the
surgeons’ fear of CNS bleeding complications with
potentially catastrophic neurologic deficits.5
The primary data on the appropriate preventive
measures in neurosurgical patients are less estab-
lished. Two prior metaanalyses,7,47 evaluated four
and three articles, respectively, assessing LMWH
and UFH for VTE prophylaxis in neurosurgery
patients. They found that LMWH was safe and
effective, but lacked enough data to comment on
UFH. Three other articles have reviewed VTE pro-
phylaxis in neurosurgical patients. In 1998, Clagett et
al4 recommended ICD with or without CS in pa-
tients undergoing intracranial surgery, with LMWH
and UFH as acceptable alternatives, and noted that

www.chestjournal.org CHEST / 134 / 2 / AUGUST, 2008 245

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
 Table 4 —Safety (Pooled RRs of RCTs)*

Comparisons Studies, No. ICH Minor Major Death

LMWH vs nonpharmacologic management 4 1.97 (0.64–6.09) 2.06 (1.07–3.96) 0.95 (0.18–5.09) 0.96 (0.47–1.96)
关3.77; 0%兴 关0.87; 0%兴 关0.51; 0%兴 关3.27; 0%兴
UFH vs nonpharmacologic management 3 2.11 (0.39–11.31) 1.00 (0.48–2.11) 0.85 (0.12–5.99) 0.97 (0.13–7.37)
关0.01; 0%兴 关0.91; 0%兴 关0.99; 0%兴 关0.26; 0%兴
LMWH vs UFH 4 1.78 (0.37–8.50) 1.28 (0.64–2.59) 1.00 (0.18–5.74) 0.72 (0.11–4.42)
关0.27; 0%兴 关0.30; 0%兴 关0.00; 0%兴 关0.3; 0%兴
*Values are given as the mean (95% CI) 关Q statistic; I2 statistic兴.

combination therapy using pharmacologic and me- not find a statistically increased rate of ICH with the
chanical modalities was probably more effective than use of heparin modalities in head-to-head trials, trends
therapy with either modality alone.4 In 2004, Ag- in these and the rates seen when all trials were used for
nelli3 again performed a structured literature review a pooled rate seem to validate the surgeons’ concerns
and found that pooled data from patients receiving about ICH. Importantly, our results suggest that me-
LMWH prophylaxis showed increased rates of ICH chanical prophylaxis measures may be as effective as
(LMWH therapy, 2.1%; mechanical modalities, heparin therapy in this population, thus allowing the
1.1%). He recommended the use of LMWH in surgeon to avoid this possible small increased risk of
craniotomy patients with caution.3 In 2005, Danish bleeding. This key finding in our review is contrary to
et al5 performed a decision-tree analysis based on a the existing belief and practice. Mechanical methods
review of the literature and comparison to the are historically less efficacious than pharmacologic
experience of his institution, and recommended methods in other surgical populations, but appear to
mechanical prophylaxis in favor of LMWH, due to show a significant contribution in this population, with-
concerns about increased ICH.5 out bleeding complications. By convention, these are
Our study helps to quantify both the risks and the typically applied to the patient preoperatively or intra-
benefits of the various preventive measures in mixed operatively, and although not statistically well assessed,
neurosurgical populations. In addition, although we did this may have an advantage over postoperative applica-

Safety: ICH
Favors LMWH Favors UFH Favors LMWH Favors No Heparin
Risk Ratio (95% CI)

Risk Ratio (95% CI)

Goldhaber 3.00 (0.12-72.49)
Agnelli 0.75 (0.17-3.32)
Macdonald 2.00 (0.19-21.34)
Dickinson 5.00 (0.61-41.08)
Prestar 0.98 (0.02-48.93)
Kurtoglu 1.00 (0.06-15.62)
Voth 1.06 (0.02-52.69)
Nurmohamed 6.07 (0.74-50.08)

Overall 1.78 ( 0.37, 8.50) 1.97 (0.64-6.09)

Overall

0 1 72 0 1 50
Risk ratio Risk ratio

LMWH vs. UFH LMWH vs. no heparin

Favors UFH Favors No Heparin

Risk Ratio (95% CI)

Cerrato 1.96 (0.07-57.19)

Constantini 2.29 (0.21-24.49)

Gruber 1.93 (0.07-54.99)

Overall 2.11 (0.39-11.31)

0 1 57
Risk ratio
UFH vs. no heparin

Figure 4. Peto plot of pooled RRs of ICH from RCTs.

246 Original Research

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
 Rate/1000 (95% CI)
Agnelli (1998) 0.0 (0.0-23.7)
Bostrom (1986) 0.0 (0.0-77.1)
Bucci (1989) 0.0 (0.0-51.3)
Bynke (1987) 0.0 (0.0-112.2)
Cerrato (1978) 0.0 (0.0-71.1)
Constantini (2001) 41.7 (5.1-142.5)
Dickinson (1998) 0.0 (0.0-142.5)
Placebo Epstein (2005) 0.0 (0.0-18.3)
Gruber (1984) 0.0 (0.0-137.2)
Kurtoglu (2004) 0.0 (0.0-59.6)
Nelson (1996) 0.0 (0.0-31.0)
Normuhomed (1996) 4.1(0.1-22.6)
Rokito (1996) 0.0 (0.0-32.9)
Skillman (1978) 0.0 (0.0-38.1)
Smith (1994) 0.0 (0.0-11.6)
Turpie (1989) 0.0 (0.0-15.3)
Wautrecht (1996) 0.0 (0.0-148.2)
Wood (1997) 0.0 (0.0-26.7)

Subtotal 0.04 (0.0-3.7)

Agnelli (1998) 19.6 (4.0-56.2)
Dickinson (1998) 113.6 (37.9-245.6)
Gerlach (2003) 15.2 (11.0-20.4)
Goldhaber (2002) 13.3 (0.3-72.0)
Kurtoglu (2004) 16.7 (0.4-89.4)
LMWH MacDonald (2003) 40.8 (4.9-139.8)
Normuhomed (1996) 24.9 (9.2-53.4)
Norwood (2002) 40.0 (14.8-85.0)
Paoletti (1989) 82.5 (36.3-156.1)
Prestar (1992) 0.0 (0.0-35.9)
Voth (1992) 0.0 (0.0-41.5)
Subtotal 1.52 (1.09-1.94)
Barnett (1977) 6.7 (0.2-36.6)
Bostrom (1986) 51.2 (10.8-143.8)
Cerrato (1978) 0.0 (0.0-71.1)
Constantini (2001) 18.2 (0.5-97.2)
UFH Frim (1992) 0.0 (0.0-26.4
Goldhaber (2002) 0.0 (0.0-48.0)
Gruber (1984) 0.0 (0.0-137.2)
MacDonald (2003) 20.8 (0.5-110.7)
Macdonald (1999) 37.7 (10.3-93.8)
Prestar (1992) 0.0 (0.0-36.6)
Voth (1992) 0.0 (0.0-39.3)
Wen (1988) 3.4 (0.7-10.0)
Subtotal 0.35 (0.0-7.4)

0 25

Figure 5. Peto plot of overall ICH rates (all studies).

tion.1,2 In our analysis, the administration of heparin
preoperatively was statistically more effective in pre-
venting VTE than postoperative administration. It is
important to note that this interpretation is limited by
the fact that there are only two direct comparison
studies of LMWH and ICDs, involving only 186 total
patients.
Sensitivity analysis revealed a higher rate of DVT
in elderly patients and those undergoing craniotomy
for neoplasm. Unfortunately, the smaller numbers of
these patients and the lack of primary data preclude
our ability to ascertain whether this select group
might benefit from chemical prophylaxis. Given that
LMWH appears to be more efficacious and have a
mortality benefit in cancer patients,48 –51 further
studies in these select high-risk groups are needed.
Based on our results, an RCT comparing therapy
with ICDs with LMWH in these patients would be
ethically possible.
Limitations
There are several limitations to our analysis. First,
the quality of the articles reviewed varied widely, and

www.chestjournal.org CHEST / 134 / 2 / AUGUST, 2008 247

Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
many suffered from potential selection bias. In ad-
dition, the interrater agreement on study quality was
low (although we analyzed our data using consensus
scores). The effect of quality on study findings in
metaanalyses has been varied. Most, like our study,
have found little effect. Second, many of the studies
were small trials. Particularly for rare events, such as
pulmonary emboli, the studies may be too small to
give reliable incidence estimates. For example, in
our pooled analysis of the risk of PEs, there were
only seven trials18,25,31,32,34,42,44 that used no prophy-
laxis (placebo groups) and reported PE rates. Six of
the seven trials18,25,31,32,34,44 reported no PEs, and
one trial42 had two PEs from among 48 placebo
patients, resulting in only two PEs among 650 pa-
tients (0.3%). While one benefit of metaanalysis is to
create potentially larger sample sizes, including
many small studies with no events may underesti-
mate the true incidence. In addition, we included
several prospective cohort trials. The rate of VTE
events in these studies was uniformly lower than the
rates seen in the RCTs, suggesting the likely under-
reporting of events in theses series. Third, the pooled
rates of DVT had high degrees of heterogeneity for
placebo, CS, and LMWH groups, though little het-
erogeneity for ICD and UFH groups. Our pooled
analysis of rates suggested that both ICD and hepa-
rin-based modalities were better than CS or placebo.
We performed sensitivity analyses to attempt to
explain this heterogeneity and found four factors that
explained much of the heterogeneity in DVT rates,
as follows: study design (RCTs vs prospective trials);
average patient age; the timing of the administration
of heparin; and whether or not the operation was for
treatment of a neoplasm. After adjustment, we found
that heparin and ICD still reduced DVTs more than
CS or placebo. Fourth, and most importantly, the best
data on relative efficacy and safety come from a direct
comparison among treatment modalities in RCTs. Our
data have suggested that heparin-based modalities and
ICDs are equally effective in prophylaxis, with patients
utilizing ICDs having slightly less ICH and minor
bleeding. While we tried to adjust for differences
between the trials, these analyses should be viewed
with caution since statistical adjustment is no substitute
for direct comparisons in RCTs. Finally, the clinical
relevance of the bleeding events was inadequately
addressed in most trials. Further study with well-
defined definitions of bleeding and other complications
within RCTs are needed.
Conclusions
Current guidelines6 recommend multimodality
prophylaxis with mechanical methods augmented by
either LMWH or UFH. Based on our analysis,
248
Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
however, it appears that prophylaxis with ICDs alone
or LMWH alone are equally efficacious, and the
rates of ICH may be lower with the use of ICDs.
Even a potentially small reduction in bleeding com-
plications is especially relevant in this population.
There may be a role for the combination of mechan-
ical and pharmacologic methods in high-risk pa-
tients, such as the elderly or those undergoing
craniotomy for the treatment of malignancy, but
further studies are needed in this population of
patients. In the meantime, a more individualized
discussion about the risks and benefits between the
surgeon and the higher risk patient should guide the
ultimate decision.
References
1 Ferree BA. Deep venous thrombosis following lumbar lami-
notomy and laminectomy. Orthopedics 1994; 17:35–38
2 Rokito SE, Schwartz MC, Neuwirth MG. Deep vein throm-
bosis after major reconstructive spinal surgery. Spine 1996;
21:853– 858
3 Agnelli G. Prevention of venous thromboembolism in surgical
patients. Circulation 2004; 110:IV4 –IV12
4 Clagett GP, Anderson FA Jr, Geerts W, et al. Prevention of
venous thromboembolism. Chest 1998; 114(suppl):531S–560S
5 Danish SF, Burnett MG, Ong JG, et al. Prophylaxis for deep
venous thrombosis in craniotomy patients: a decision analysis.
Neurosurgery 2005; 56:1286 –1292
6 Geerts WH, Pineo GF, Heit JA, et al. Prevention of venous
thromboembolism: the Seventh ACCP Conference on Anti-
thrombotic and Thrombolytic Therapy. Chest 2004; 126(suppl):
338S– 400S
7 Iorio A, Agnelli G. Low-molecular-weight and unfractionated
heparin for prevention of venous thromboembolism in neu-
rosurgery: a meta-analysis. Arch Intern Med 2000; 160:2327–
2332
8 Jadad AR, Moore RA, Carroll D, et al. Assessing the quality
of reports of randomized clinical trials: is blinding necessary?
Control Clin Trials 1996; 17:1–12
9 Laupacis A, Wells G, Richardson WS, et al. Users’ guides to
the medical literature: V. How to use an article about
prognosis: Evidence-Based Medicine Working Group: JAMA
1994; 272:234 –237
10 DerSimonian R, Laird N. Meta-analysis in clinical trials.
Control Clin Trials 1986; 7:177–188
11 Akçakaya HR. Estimating the variance of survival rates and
fecundities. Animal Conservation 2002; 5:333–336
12 Galbraith RF. A note on graphical presentation of estimated
odds ratios from several clinical trials. Stat Med 1988; 7:889 –
894
13 Mantel N, Haenszel W. Statistical aspects of the analysis of
data from retrospective studies of disease. J Natl Cancer Inst
1959; 22:719 –748
14 Higgins JP, Thompson SG, Deeks JJ, et al. Measuring
inconsistency in meta-analyses. BMJ 2003; 327:557–560
15 Begg CB, Berlin JA. Publication bias and dissemination of
clinical research. J Natl Cancer Inst 1989; 81:107–115
16 Egger M, Davey Smith G, Schneider M, et al. Bias in
meta-analysis detected by a simple, graphical test. BMJ 1997;
315:629 – 634
17 Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based
method of testing and adjusting for publication bias in
meta-analysis. Biometrics 2000; 56:455– 463
Original Research
18 Agnelli G, Piovella F, Buoncristiani P, et al. Enoxaparin plus
compression stockings compared with compression stockings
alone in the prevention of venous thromboembolism after
elective neurosurgery. N Engl J Med 1998; 339:80 – 85
19 Dickinson LD, Miller LD, Patel CP, et al. Enoxaparin
increases the incidence of postoperative intracranial hemor-
rhage when initiated preoperatively for deep venous throm-
bosis prophylaxis in patients with brain tumors. Neurosurgery
1998; 43:1074 –1081
20 Gerlach R, Scheuer T, Beck J, et al. Risk of postoperative
hemorrhage after intracranial surgery after early nadroparin
administration: results of a prospective study. Neurosurgery
2003; 53:1028 –1034
21 Goldhaber SZ, Dunn K, Gerhard-Herman M, et al. Low rate
of venous thromboembolism after craniotomy for brain tumor
using multimodality prophylaxis. Chest 2002; 122:1933–1937
22 Kurtoglu M, Yanar H, Bilsel Y, et al. Venous thromboembo-
lism prophylaxis after head and spinal trauma: intermittent
pneumatic compression devices versus low molecular weight
heparin. World J Surg 2004; 28:807– 811
23 Macdonald RL, Amidei C, Baron J, et al. Randomized, pilot
study of intermittent pneumatic compression devices plus
dalteparin versus intermittent pneumatic compression de-
vices plus heparin for prevention of venous thromboembo-
lism in patients undergoing craniotomy. Surg Neurol 2003;
59:363–372
24 Norwood SH, McAuley CE, Berne JD, et al. Prospective
evaluation of the safety of enoxaparin prophylaxis for venous
thromboembolism in patients with intracranial hemorrhagic
injuries. Arch Surg 2002; 137:696 –701
25 Nurmohamed MT, van Riel AM, Henkens CM, et al. Low
molecular weight heparin and compression stockings in the
prevention of venous thromboembolism in neurosurgery.
Thromb Haemost 1996; 75:233–238
26 Paoletti C, Maubec E, Raggueneau JL, et al. Clinical toler-
ance of CY 216 (Fraxiparin) in the prevention of thrombo-
embolic accidents after neurosurgery. Agressologie 1989;
30:363–366
27 Prestar FJ. Prevention of thromboembolism complications
with low molecular weight heparin in microneurosurgical
lumbar intervertebral disk operations. Neurochirurgia
(Stuttg) 1992; 35:1– 4
28 Voth D, Schwarz M, Hahn K, et al. Prevention of deep vein
thrombosis in neurosurgical patients: a prospective double-
blind comparison of two prophylactic regimen. Neurosurg
Rev 1992; 15:289 –294
29 Barnett HG, Clifford JR, Llewellyn RC. Safety of mini-dose
heparin administration for neurosurgical patients. J Neuro-
surg 1977; 47:27–30
30 Bostrom S, Holmgren E, Jonsson O, et al. Post-operative
thromboembolism in neurosurgery: a study on the prophy-
lactic effect of calf muscle stimulation plus dextran compared
to low-dose heparin. Acta Neurochir (Wien) 1986; 80:83– 89
31 Cerrato D, Ariano C, Fiacchino F. Deep vein thrombosis and
low-dose heparin prophylaxis in neurosurgical patients.
J Neurosurg 1978; 49:378 –381
32 Constantini S, Kanner A, Friedman A, et al. Safety of
perioperative minidose heparin in patients undergoing brain
tumor surgery: a prospective, randomized, double-blind
study. J Neurosurg 2001; 94:918 –921
33 Frim DM, Barker FG II, Poletti CE, et al. Postoperative
low-dose heparin decreases thromboembolic complications in
neurosurgical patients. Neurosurgery 1992; 30:830 – 832
34 Gruber UF, Rem J, Meisner C, et al. Prevention of thrombo-
embolic complications with miniheparin-dihydroergotamine in
patients undergoing lumbar disc operations. Eur Arch Psychiatry
Neurol Sci 1984; 234:157–161
www.chestjournal.org
Downloaded from chestjournal.org on September 1, 2008
Copyright © 2008 by American College of Chest Physicians
35 Macdonald RL, Amidei C, Lin G, et al. Safety of periopera-
tive subcutaneous heparin for prophylaxis of venous throm-
boembolism in patients undergoing craniotomy. Neurosur-
gery 1999; 45:245–251
36 Wen DY, Hall WA. Complications of subcutaneous low-dose
heparin therapy in neurosurgical patients. Surg Neurol 1998;
50:521–525
37 Bucci MN, Papadopoulos SM, Chen JC, et al. Mechanical
prophylaxis of venous thrombosis in patients undergoing crani-
otomy: a randomized trial. Surg Neurol 1989; 32:285–288
38 Bynke O, Hillman J, Lassvik C. Does perioperative external
pneumatic leg muscle compression prevent post-operative
venous thrombosis in neurosurgery? Acta Neurochir (Wien)
1987; 88:46 – 48
39 Epstein NE. Intermittent pneumatic compression stocking
prophylaxis against deep venous thrombosis in anterior cer-
vical spinal surgery: a prospective efficacy study in 200
patients and literature review. Spine 2005; 30:2538 –2543
40 Epstein NE. Efficacy of pneumatic compression stocking pro-
phylaxis in the prevention of deep venous thrombosis and
pulmonary embolism following 139 lumbar laminectomies with
instrumented fusions. J Spinal Disord Tech 2006; 19:28 –31
41 Nelson LD Jr, Montgomery SP, Dameron TB Jr, et al. Deep
vein thrombosis in lumbar spinal fusion: a prospective study
of antiembolic and pneumatic compression stockings. J South
Orthop Assoc 1996; 5:181–184
42 Skillman JJ, Collins RE, Coe NP, et al. Prevention of deep
vein thrombosis in neurosurgical patients: a controlled, ran-
domized trial of external pneumatic compression boots. Sur-
gery 1978; 83:354 –358
43 Smith MD, Bressler EL, Lonstein JE, et al. Deep venous
thrombosis and pulmonary embolism after major reconstruc-
tive operations on the spine: a prospective analysis of three
hundred and seventeen patients. J Bone Joint Surg Am 1994;
76:980 –985
44 Turpie AG, Hirsh J, Gent M, et al. Prevention of deep vein
thrombosis in potential neurosurgical patients: a randomized
trial comparing graduated compression stockings alone or
graduated compression stockings plus intermittent pneumatic
compression with control. Arch Intern Med 1989; 149:679 –
681
45 Wautrecht J, Macquaire V, Vandesteene A, et al. Prevention
of deep vein thrombosis in neurosurgical patients with brain
tumors: a controlled, randomized study comparing graded
compression stockings alone and with intermittent sequential
compression: correlation with pre- and postoperative fibrino-
lysis; preliminary results. Int Angiol 1996; 15:5–10
46 Wood KB, Kos PB, Abnet JK, et al. Prevention of deep-vein
thrombosis after major spinal surgery: a comparison study of
external devices. J Spinal Disord 1997; 10:209 –214
47 Agnelli G. Prevention of venous thromboembolism after
neurosurgery. Thromb Haemost 1999; 82:925–930
48 Lyman GH, Khorana AA, Falanga A, et al. American Society
of Clinical Oncology guideline: recommendations for venous
thromboembolism prophylaxis and treatment in patients with
cancer. J Clin Oncol 2007; 25:5490 –5505
49 Lee AY. Cancer and venous thromboembolism: prevention,
treatment and survival. J Thromb Thrombolysis 2008; 25:
33–36
50 Nishioka J, Goodin S. Low-molecular-weight heparin in
cancer-associated thrombosis: treatment, secondary preven-
tion, and survival. J Oncol Pharm Pract 2007; 13:85–97
51 Ferretti G, Bria E, Giannarelli D, et al. Is recurrent venous
thromboembolism after therapy reduced by low-molecular-
weight heparin compared with oral anticoagulants? Chest
2006; 130:1808 –1816
CHEST / 134 / 2 / AUGUST, 2008 249
 Prevention of Venous Thromboembolism in Neurosurgery: A
Metaanalysis
Jacob F. Collen, Jeffrey L. Jackson, Andrew F. Shorr and Lisa K. Moores
Chest 2008;134;237-249; Prepublished online July 18, 2008;
DOI 10.1378/chest.08-0023
This information is current as of September 1, 2008

Updated Information Updated information and services, including

& Services high-resolution figures, can be found at:
http://chestjournal.org/cgi/content/full/134/2/237
References This article cites 51 articles, 13 of which you can access
for free at:
http://chestjournal.org/cgi/content/full/134/2/237#BIBL
Permissions & Licensing Information about reproducing this article in parts
(figures, tables) or in its entirety can be found online at:
http://chestjournal.org/misc/reprints.shtml
Reprints Information about ordering reprints can be found online:
http://chestjournal.org/misc/reprints.shtml
Email alerting service Receive free email alerts when new articles cite this
article sign up in the box at the top right corner of the
online article.
Images in PowerPoint format Figures that appear in CHEST articles can be
downloaded for teaching purposes in PowerPoint slide
format. See any online article figure for directions.

Downloaded from chestjournal.org on September 1, 2008

Copyright © 2008 by American College of Chest Physicians