Review Articles
Transcranial Doppler in the Diagnosis of Cerebral
Vasospasm: An Updated Meta-Analysis
Jean-Mathieu Mastantuono, MD1; Christophe Combescure, PhD2,3; Nadia Elia, MD, MSc1,3,4;
Martin R. Tramèr, MD, DPhil1,3; Christopher Lysakowski, MD1,3
Objectives: To evaluate the performance of transcranial Doppler
and transcranial color-coded duplex Doppler in patients with cere-
bral vasospasm due to aneurysm rupture. Angiography was con-
sidered as the gold standard comparator.
Data Sources: Search in MEDLINE, Embase, and Central from
January 2001 to October 2017, without language restriction. Bibli-
ographies of retrieved articles were screened for additional studies.
Study Selection: Randomized studies comparing transcranial
Doppler or transcranial color-coded duplex Doppler with angiog-
raphy in adults.
Data Extraction: Data were extracted independently by several
investigators. Sensitivity and specificity were combined across
studies using a bivariate model. Preferred Reporting Items for
Systematic Reviews and Meta-Analyses was used for reporting
and Quality Assessment of Diagnostic Accuracy Studies-2 for
quality assessment.
Data Synthesis: We included 18 studies. Fifteen tested tran-
scranial Doppler. For the middle cerebral artery (10 studies,
1,408 tests), the pooled sensitivity was 66.7% (95% CI, 55.9–
75.9) and specificity was 89.5% (80.3–94.7). Three studies
(278 tests) tested transcranial color-coded duplex Doppler for
1
Division of Anesthesiology, Department Anesthesiology, Pharmacol-
ogy & Intensive Care Medicine, Geneva University Hospitals, Geneva,
­Switzerland.
2
Clinical Trials Centre & Division of Clinical Epidemiology, Department of
Health and Community Medicine, University of Geneva & Geneva Univer-
sity Hospitals, Geneva, Switzerland.
3
Faculty of Medicine, University of Geneva, Geneva, Switzerland.
4
Institute of Global Health, Medical Faculty, University of Geneva, Geneva,
Switzerland.
Supplemental digital content is available for this article. Direct URL cita-
tions appear in the printed text and are provided in the HTML and PDF
versions of this article on the journal’s website (http://journals.lww.com/
ccmjournal).
Dr. Elia received funding from the European Journal of Anaesthesiology
(methods editor). The remaining authors have disclosed that they do not
have any potential conflicts of interest.
For information regarding this article, E-mail: christopher.lysakowski@
hcuge.ch
Copyright © 2018 by the Society of Critical Care Medicine and Wolters
Kluwer Health, Inc. All Rights Reserved.
DOI: 10.1097/CCM.0000000000003297
Critical Care Medicine www.ccmjournal.org 1665
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
the middle cerebral artery. The pooled sensitivity was 81.5%
(66.0–90.0), and specificity was 96.6% (93.0–98.0). For an
arbitrarily chosen prevalence of vasospasm of 70%, positive
and negative predictive values were 93.7% (88.9–96.6) and
53.4% (46.7–60.9) for transcranial Doppler and 98.2% (96.4–
99.1) and 69.1% (56.1–80.9) for transcranial color-coded
duplex Doppler.
Conclusions: Assuming a high prevalence of vasospasm of the
middle cerebral artery, both transcranial Doppler and transcra-
nial color-coded duplex Doppler are likely to detect it, but nei-
ther is useful to exclude it. There is no convincing evidence
that the accuracy of transcranial color-coded duplex Doppler
is any better than that of transcranial Doppler. For arteries
other than middle cerebral artery, there is a lack of evidence of
the usefulness of transcranial Doppler. (Crit Care Med 2018;
46:1665–1672)
Key Words: diagnostic equipment; intracranial; meta-analysis;
transcranial Doppler; ultrasonography; vasospasm
D
uring recent years, transcranial Doppler (TCD) has
become a widely used bedside tool to identify, or to
exclude, an emboli or a stenosis of a cerebral artery,
or a vasospasm from a subarachnoid hemorrhage. It is there-
fore of great importance to know accuracy and usefulness of
this noninvasive method as a screening tool, compared with,
for instance, angiography that may be considered the gold
standard in this context but which is an invasive technique.
In a previously published systematic review that included rel-
evant studies from 1984 to 2001, TCD was shown to detect
the presence of a spasm of the middle cerebral artery (MCA),
confirmed by angiography (1). When TCD did not indicate a
spasm, no conclusion could be drawn. It has been suggested
that transcranial color-coded duplex Doppler (TCCD) allowed
for a more precise detection and monitoring of cerebral vaso-
spasms (2–6).
This update including valid and relevant literature since
2001 had three aims. First, to reevaluate the diagnostic and
screening accuracy of TCD for spasms of the MCA. Second,
to check whether new studies had been published testing the
usefulness of TCD for the diagnosis of spasm of other arteries.
Mastantuono et al
And third, to evaluate the diagnostic accuracy of TCCD in that
setting.
METHODS
This work is reported according to the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses (7).
Searching Data Sources
The seven studies included in the previous review (1) and one
study published before 2001 were included in the present meta-
analysis. We performed a new search in MEDLINE, Embase,
and Cochrane Library without language restriction, using
“intracranial aneurysm,” “ruptured intracranial aneurysm,”
“cerebral vasospasm,” “cerebral angiography,” “Transcranial
Doppler,” “Transcranial Color-Coded Duplex Doppler,” and
“subarachnoid haemorrhage” from January 2001 to October
2017. Bibliographies of retrieved articles were screened for
additional studies.
Inclusion/Exclusion Criteria
One author (J.-M.M.) assessed the eligibility of retrieved
articles. Only trials in adults were considered, and they were
restricted to full reports of randomized controlled trials com-
paring TCD or TCCD with cerebral angiography for the diag-
nosis of cerebral vasospasm after subarachnoid hemorrhage
due to a ruptured aneurysm. Two authors (J.-M.M., C.L.)
checked inclusion criteria independently.
Data Extraction
We extracted data from different cerebral arteries on true
and false positives, and true and false negatives with TCD
or TCCD compared with angiography. When only sensitivity
and specificity were reported, we contacted the authors and
asked for the raw data. When they did not answer, we tried to
compute dichotomous data from these values. We extracted
data on year of publication, unit of analysis (patient, artery),
diagnostic thresholds for cerebral vasospasm, and values
for clinically relevant vasospasm both for angiography and
Doppler. Data extraction was performed by one author
(J.-M.M.) and independently checked by another (C.L.).
Disagreements were resolved by discussion with the other
authors.
Quality and Risk of Bias Assessment
The Quality Assessment of Diagnostic Accuracy Studies-2 was
used to assess the quality (8). This methodologic quality was
assessed by two separate sections: the risk of bias and applicabil-
ity. The risk of bias was quantified as “low,” “high,” or “unclear”
using the specific questions in four different domains: 1) patient’s
selection; 2) index test (in this case TCCD and TCD); 3) gold
standard diagnostic test (in this case angiography); and 4) flow
of the patients (all patients included, the same gold standard
used) and the timing (intervals between the tests). Three of these
domains particularly, diagnostic test, gold standard, and flow of
the patients and timing were used to evaluate the methodologic
1666 www.ccmjournal.org October 2018 • Volume 46 • Number 10
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
quality. There was a pre hoc decision that trials were at high risk
of bias when the delay between Doppler and angiography was
more than 24 hours. Applicability was also rated as “low”, “high,”
or “unclear” and was taken into consideration patient’s selection,
diagnostic test, and the gold standard.
Statistics
Hierarchical summary receiver-operating characteristic curves,
pooled sensitivities and specificities, and overall likelihood ratios
were obtained using bivariate models with random effects. The
predictive values for a prevalence of 70% were derived from the
pooled sensitivities and specificities. Heterogeneity was mea-
sured with the I2 statistic and investigated by comparing sub-
groups (year of publication, unit of analysis [patient/artery],
and threshold of velocities) (9). A leave-one-out sensitivity
analysis was conducted to check the robustness of the findings.
Potential publication bias was investigated using the Deeks’ test
(10). We used R software version 3.1.3 (R Foundation for Sta-
tistical Computing, Vienna, Austria; www.R-project.org) with
the packages “mada” and “mvmeta” for meta-analysis (11). Sta-
tistical significance was defined as p less than 0.05 (two-sided)
except for the test on publication bias (p < 0.10). Predictive val-
ues were calculated using the pooled sensitivities and specifici-
ties for a 70% prevalence of angiographic cerebral vasospasm
based on literature (12, 13). The bivariate model was used to
compare the pooled sensitivity and specificity between studies
with TCD and studies with TCCD.
RESULTS
Study Selection
We identified and screened 214 potentially relevant trials
(Fig. 1). Of those, 180 were excluded for a variety of reasons.
Of the remaining 34 studies, 24 were disqualified because of
incomplete data for computation of 2 × 2 tables (2, 14–36).
Finally, 18 studies were included for analysis: 10 were pub-
lished since 2001 (37–46) and eight were published before
2001, seven (47–53) from the previously published system-
atic review (1) and one supplementary trial that had not been
included (3). Fifteen studies tested TCD (37, 38, 40, 41, 43–53)
and three tested TCCD (3, 39, 42).
Quality and Risk of Bias Assessment
In nine of 17 studies (the study by Kyoi et al [24] was excluded),
the quality was considered as good (Appendix 1, Supple-
mental Digital Content 1, http://links.lww.com/CCM/D658).
Applicability of included trials is presented in Appendix 2
(Supplemental Digital Content 2, http://links.lww.com/CCM/
D659). Risk of bias of one study published in Japanese was not
assessed (47).
TCD: Studies’ Description
Fifteen studies compared TCD with angiography in different
arteries (Table 1). Seven analyzed data by patient, and eight by
artery. Angiographic vasospasm was defined as lumen narrowing
in any artery of greater than 25% in nine studies (38, 40, 44, 46,
 Review Articles

(Table 2). The diagnostic per-

formance of TCD varied across
studies. For instance, for the
MCA, the sensitivity of TCD
varied from 38.5% (50) to 100%
(38). The I2 statistics confirmed
heterogeneity across studies.
The widths of the 95% CIs indi-
cated a large uncertainty around
the point estimates for MCA
(Appendix 3, Supplemental
Digital Content 3, http://links.
lww.com/CCM/D660).
Likelihood ratios indicated
that information from the TCD
results was moderate (Table 2).
For an arbitrarily chosen
prevalence of 70%, the nega-
tive predictive values (NPVs)
corresponding to the pooled
sensitivities and specificities
were low and the positive pre-
dictive values (PPVs) were high
for MCA and ACA. Predictive
values for the MCA with other
prevalences are presented in
Appendix 4 (Supplemental
Digital Content 4, http://links.
lww.com/CCM/D661).

TCCD: Study Description

and Quality
Three studies (278 tests) com-
pared TCCD with angiogra-
phy (3, 39, 42). Their quality
was good, and the risk of bias
was lower than in TCD stud-
ies (Appendix 1, Supplemen-
tal Digital Content 1, http://
links.lww.com/CCM/D658;
and Appendix 2, Supplemen-
tal Digital Content 2, http://
links.lww.com/CCM/D659).
For the assessment of MCA,
Figure 1. Flow chart of retrieved and eventually analyzed trials. specific thresholds of mean
Adapted from Lysakowski et al (1). velocity were used (Table 1).
Angiographic vasospasm was
47, 49, 50, 52, 53), more than 1 mm in one (48) and of greater than defined as lumen narrowing in any artery of greater than
33% in one (43). Four studies did not report the threshold value 25% in two studies (3, 39) and of greater than 30% in one
for the angiographic diagnosis of vasospasm (37, 41, 45, 51). The (42). Data were computed by arteries in two studies (3, 42)
Doppler diagnosis of vasospasm was based on mean velocity. and by patient in one (39) (Table 1).

TCD: Diagnostic Accuracy TCCD: Diagnostic Accuracy

Pooled sensitivities for the MCA and the basilar cerebral artery Pooled sensitivity and specificity are presented in Table 2
were higher than for the anterior cerebral artery (ACA). The and illustrated in Appendix 3 (Supplemental Digital Con-
pooled specificities were nearly similar for these three arteries tent 3, http://links.lww.com/CCM/D660). For an arbitrarily

Critical Care Medicine www.ccmjournal.org 1667

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Mastantuono et al

TABLE 1. Description of Included Trials

Year of
Publi- TCD Mean True Sample Sensitivity Specificity
Author (Reference) cation Artery Velocity cm/s Positive Size Unit   % %

TCD
  Kyoi et al (47) 1989 MCA > 120 10 18 Patient 90.9 100.0
  Sloan et al (48) 1989 MCA > 120 17 34 Patient 58.6 100.0
  Langlois et al (49) 1992 MCA > 130 11 112 Artery 73.3 100.0
  Lennihan et al (51) 1993 MCA > 140 6 66 Artery 85.7 98.3
  Burch et al (50) 1996 MCA > 120 15 87 Artery 38.5 93.8
  Proust et al (46) 2002 MCA > 120 111 460 Patient 63.8 95.1
  Mascia et al (38) 2003 MCA > 160 13 33 Artery 100.0 75.0
  Gonzalez et al (40) 2007 MCA > 120 21 114 Artery 58.3 85.9
  Carrera et al (41) 2009 MCA > 90 11 199 Patient 73.3 64.7
  Sebastian et al (43) 2013 MCA > 150 23 285 Artery 79.3 85.9
  Kyoi et al (47) 1989 ACA > 90 9 18 Patient 81.8 71.4
  Lennihan et al (51) 1993 ACA > 120 2 66 Artery 13.3 100.0
  Wozniak et al (53) 1996 ACA > 120 9 87 Artery 18.0 64.9
  Sloan et al (52) 1994 BCA > 60 10 42 Patient 76.9 79.3
  Sviri et al (44) 2006 BCA > 90 34 144 Patient 50.0 88.2
  Burch et al (50) 1996 Internal carotid > 90 11 90 Artery 25.0 91.3
artery
  Wozniak et al (53) 1996 Posterior cer- > 90 11 84 Artery 47.8 68.9
ebral artery
  Wintermark et al (45) 2006 Any > 120 88 630 Artery 71.5 89.7
  Frontera et al (37) 2009 Any > 120 121 300 Patient 70.8 62.8
Transcranial color-coded duplex Doppler
  Proust et al (3) 1999 MCA > 120 5 38 Patient 71.4 100.0
  Krejza et al (39) 2005 MCA > 94 32 222 Artery 76.2 97.2
  Wang et al (42) 2012 MCA > 120 12 18 Patient 85.7 100.0
ACA = anterior cerebral artery, BCA = basilar cerebral artery, MCA = middle cerebral artery, TCD = Transcranial Doppler.
Unit = vasospasm considered on patient or on artery.

defined prevalence of vasospasm of 70%, PPV was 98.2%
(96.4–99.1) and NPV was 69.1% (56.1–80.9) (Table 2). Pre-
dictive values for other prevalences are presented in Appen-
dix 4 (Supplemental Digital Content 4, http://links.lww.com/
CCM/D661). TCCD studies concerning MCA showed lower
heterogeneity than with TCD. No data were available for the
other arteries. TCCD appeared to present a better pooled
estimate than TCD, but this difference was not statistically
significant (p = 0.138) (Table 2).
Sources of Heterogeneity
Sources of heterogeneity were investigated only in TCD studies
and only for the MCA due to the limited number of studies
that assessed other arteries. Unit of analysis (patients or arter-
ies) was not associated with a difference in pooled sensitiv-
ity (p = 0.78) or specificity (p = 0.73). Pooled sensitivity was
higher in studies with a velocity threshold above 120 cm·s–1
(79.3% [65.6–88.5]) compared with studies with a threshold
of 120 cm·s–1 or lower (61.1% [49.2–71.9]); that difference was
statistically significant (p = 0.048).
There was no difference in pooled sensitivities between
five MCA studies published before 2001 (65.9% [44.8–
82.2]) (47–51) and five studies published after 2001 (69.1%
[62.1–75.4]) (38, 40, 41, 43, 46) (p = 0.41). However, the
difference in pooled specificities was statistically signifi-
cant and showed a decrease in specificity over time (97.5%
[92.6–99.2] before 2001 and 84.1% [70.1–92.2] after 2001;

1668 www.ccmjournal.org October 2018 • Volume 46 • Number 10

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
 Review Articles

TABLE 2. Diagnostis Accuracy of Transcranial Doppler and Transcranial Color-Coded

Duplex Doppler
Transcranial Color-Coded
Transcranial Doppler Duplex Doppler
Estimate and
Number of Middle Cerebral Anterior Cerebral Basilar Cerebral Middle ­Cerebral
Studies Artery Artery Artery All Arteries Artery pa

n studies 10 4 2 2 3
Pooled sensitivity 66.7% 32.7% 62.1% 71.2% 81.5% 0.138
(55.9–75.9) (10.9–65.7) (33.3–84.3) (65.7–76.1) (67.5–90.3)
I2, % 57.8 78 66 0 0.00
Pooled specificity 89.5% 89.6% 84.5% 79.4% 96.6% 0.127
(80.3–94.7) (48.2–98.7) (71.1–92.3) (43.5–95.1) (93.2–98.3)
I2, % 89 65 24 98 0.00
Positive LR 6.86 5.77 4.05 4.54 25.0
(3.57–12.50) (0.55–25.60) (2.28–6.76) (1.24–14.00) (11.9–49.9)
Negative LR 0.38 0.79 0.44 0.40 0.20
(0.27–0.50) (0.41–1.33) (0.20–0.74) (0.27–0.69) (0.11–0.34)
Positive predictive 93.7% 87.4% 90.0% 88.9% 98.2%
value (88.9–96.6) (57.8–98.3) (84.6–93.9) (74.3–97.1) (96.4–99.1)
Negative 53.4% 35.4% 48.6% 53.9% 69.1%
predictive value (46.7–60.9) (24.6–50.6) (36.5–67.6) (38.1–61.2) (56.1–80.9)
LR = likelihood ratio.
Positive predictive value and negative predictive value were calculated with a prevalence of 70%. Numbers in parenthesis represent 95% CI.
a
p = comparison between Transcranial Doppler and transcranial color-coded duplex Doppler for the middle cerebral artery.

p = 0.019). Cumulative meta-analysis suggested that pooled
sensitivities and specificities have regularly decreased since
1993 (Fig. 2).
Clinical Vasospasm
In five studies using TCD, velocity of MCA was measured in
493 patients with clinically symptomatic vasospasms defined
as the appearance of a new global or focal neurologic deteri-
oration or Glasgow Coma Scale degradation. The threshold
of mean velocity for the diagnosis of vasospasm was greater
than 120 cm·s–1 in three trials (19, 40, 48), and was greater
than 150 cm·s–1 (43) and greater than 160 cm·s–1 (38) in one
study each. Angiographic vasospasm was defined as lumen
narrowing in any artery of greater than 25% in two studies
(38, 40), of greater than 33% (43) or greater than 50% (19)
in one study each, and of more than 1 mm in one study (48).
Pooled sensitivity was 66.2% (59.3–72.6), and pooled speci-
ficity was 82.2% (74.9–87.8). These results were very close to
those obtained with all TCD studies.
Sensitivity Analyses and Biases
The leave-one-out sensitivity analysis conducted for studies using
TCD and assessing the MCA showed that the findings were not
drawn by a single study (Appendix 5, Supplemental Digital Con-
tent 5, http://links.lww.com/CCM/D662). In contrast, the pooled
sensitivity and/or specificity of TCD in ACA varied importantly
when any of the four studies was removed (Appendix 5, Supple-
mental Digital Content 5, http://links.lww.com/CCM/D662).
In low risk of bias TCD studies, the pooled specificity for
MCA was slightly lower than the specificity pooled over all
studies. This was not found for TCCD studies; they presented
only low risk of bias (Table 3).
When the interval between two tests was shorter than 24
hours (low risk of flow timing bias), the pooled sensitivity
decreased. The pooled sensitivity was higher in studies with a
good TCD performance and a prespecified threshold (low risk
of index test bias) (Table 3).
Small TCD studies tended to show a higher sensitivity or a
higher specificity (Appendix 3, Supplemental Digital Content 3,
http://links.lww.com/CCM/D660), but an association between
the diagnostic accuracy and study size was not detected (p = 0.40).
Global quality of TCCD studies was better than that of TCD
studies (Appendix 1, Supplemental Digital Content 1, http://
links.lww.com/CCM/D658).
DISCUSSION
Summary of Main Results
Several results emerge from this meta-analysis. First, this work
confirms that for MCA, and the assumption of a 70% preva-
lence of vasospasm, TCD may be used to identify patients with
vasospasm (high PPV), but it should not be used to rule-out
patients without vasospasm (low NPV and sensitivity). In
this context, TCCD did not show better accuracy. Although
the PPV and NPV values were higher, the difference was not
statistically significant (p = 0.138 for sensibility and p = 0.127

Critical Care Medicine www.ccmjournal.org 1669

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Mastantuono et al

prevalences of vasospasm. As
Studies N Pooled specificity Pooled sensitivity
expected, when the prevalence
of vasospasm is high, TCD is
likely to identify it. However,
Kyoi et al, 1989 (47) 18
if TCD does not identify a
+ Sloan et al, 1989 (48) 52
vasospasm, it does not neces-
sarily mean that there is none.
+ Langlois et al, 1992 (49) 164 With decreasing prevalence
of vasospasm, the screening
+ Lennihan et al, 1993 (51) 230 performance of TCD further
decreases suggesting a lim-
+ Burch et al, 1996 (50) 317 ited usefulness as a screening
method in the context of low
+ Proust et al, 2002 (46) 777
prevalences of vasospasm.
These analyses were done
+ Mascia et al, 2003 (38) 810
with data of the MCA, as the
+ Gonzalez et al, 2007 (40) 924
other arteries were less well
documented.
+ Carrera et al, 2009 (41) 1123 TCCD is thought to improve
the performance of TCD and
+ Sebastian et al, 2013 (43) 1408 to decrease inter- and intrao-
bserver variability by correct-
ing the insonation angle (6).
0 50 100 0 50 100 We retrieved only three studies
Specificity (%) Sensitivity (%) testing TCCD on the MCA that
fulfilled our inclusion criteria
Figure 2. Cumulative meta-analyses of sensitivities and specificities of Transcranial Doppler of the middle (3, 39, 42). We found no evi-
cerebral artery. They were assessed by adding studies chronologically one-by-one. The squares represent the dence that color Doppler had
pooled estimates, and the horizontal lines represent the 95% CI. The cumulative number of tests is reported in
the column “N.” a better diagnostic accuracy
than conventional Doppler
for specificity). Second, the evidence regarding the suitability and cannot conclude that TCCD is better than TCD in this
of TCD and TCCD for the diagnosis of vasospasm of other setting.
cerebral arteries is still lacking. Third, there was evidence of a Our analyses are raising issues concerning the quality of
negative correlation between the methodologic quality of the data reporting in published articles. Reporting of results was
studies and the diagnostic accuracy of TCD. This could not often incomplete or unclear. One third of the studies did not
be confirmed for TCCD studies since only three trials were report on sensitivity and specificity, and the raw data to cal-
included and their quality was good. culate them were not accessible. Also, interpretation of esti-
To evaluate the capacity of TCD as a screening tool, we mates was limited due to the lack of reported uncertainty (the
modeled its use in different contexts, defining different counts of tests used to derive sensitivities and specificities or

TABLE 3. Risk for Bias in Selected Low Risk of Bias Studies

Pooled Estimates (95% CI)

Transcranial Doppler Studies Transcranial Color-Coded Duplex Doppler Studies

Bias n Sensitivity Specificity n Sensitivity Specificity

Selection bias 3 62.3% (55.3–68.7) 88.1% (71.8–95.6) 3 81.5% (67.5–90.3) 96.6% (93.2–98.3)
Index test bias 4 70.1% (38.4–89.9) 88.0% (77.9–93.8) 2 84.2% (69.4–92.6) 96.4% (92.8–98.3)
Reference 4 56.7% (39.6–72.3) 86.5% (74.5–93.3) 3 81.5% (67.5–90.3) 96.6% (93.2–98.3)
standard bias
Flow timing bias 5 56.4% (41.5–70.2) 88.8% (78.2–94.6) 1 84.0% (63.9–95.5)a 96.5% (92.9–98.6)
Global 4 56.7% (39.6–72.3) 86.5% (74.5–93.3) 3 81.5% (67.5–90.3) 96.6% (93.2–98.3)
Clopper-Person exact CI.
a

Index test bias describes how the test was conducted and interpreted. Reference standard bias describes how the reference was conducted and interpreted.
Flow timing bias describes the time interval and any interventions between index test and reference standard. Numbers in parenthesis represent 95% CI.

1670 www.ccmjournal.org October 2018 • Volume 46 • Number 10

Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

95% CI were not systematically reported). Additionally, when
the diagnosis of TCD was based on the unit of patients and
not on the unit of arteries, authors rarely explained how they
combined the results derived from arteries to draw a diagnosis
per patient. Finally, statistical analyses rarely accounted for the
clustering of data (various arteries from one single patient).
Specific statistical methods are available for this type of data,
for instance, regression models with mixed effects or general-
ized estimating equations (54).
Strengths of This Analysis
Sensitivities and specificities were combined using a bivariate
model to account for potential between-study correlations,
as recommended for meta-analyses of studies assessing diag-
nostic accuracy (55). Sources of heterogeneity were investi-
gated and the robustness of the findings was checked using
various methods (leave-one-out sensitivity analyses, sensi-
tivity analyses conducted on studies with a low risk of bias).
Weaknesses of This Analysis
In some studies, relevant data for the meta-analysis could not
be obtained, and bias due to the selection of studies cannot be
excluded. Sensitivity and specificity data were combined sup-
posing the independence between observations because studies
reported findings assuming this independence. This assump-
tion is probably not justified as various arteries were assessed per
patient and sensitivities and specificities were assessed at the level
of arteries in most studies. Therefore, the uncertainty around the
sensitivities and specificities may be underestimated. Heteroge-
neity across studies was detected for TCD. In studies assessing
MCA, we identified only year of publication as a potential source
of heterogeneity. The number of studies assessing other arteries
than MCA was too low to investigate further potential sources of
heterogeneity. Significant heterogeneity was not found for TCCD
perhaps due to the limited number of available studies.
Further Research Orientation
We suggest focusing further research on TCCD. Future stud-
ies should compare TCCD with angiography, the gold stan-
dard method for the diagnosis of vasospasms, and should also
include tests on arteries other than MCA.
Researchers should use rigorous methodology. The risk of bias
in assessment of diagnostic performance can be limited by taking
appropriate precautions such as the minimization of the inter-
val between the Doppler investigation and angiography. Future
research on the TCCD should follow the Standards for Reporting
of Diagnostic Accuracy Studies recommendations (56).
CONCLUSIONS
For the MCA, and in the context of a high prevalence of vaso-
spasm, both conventional and color Doppler are likely to detect
a vasospasm, but neither is useful to exclude one. There is no
evidence that the accuracy of TCCD is any better than that of
TCD; however, more studies are needed to confirm this state-
ment. For other arteries, the use of TCD cannot yet be recom-
mended because there is a lack of relevant and valid data.
Critical Care Medicine www.ccmjournal.org 1671
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Review Articles
REFERENCES
1. Lysakowski C, Walder B, Costanza MC, et al: Transcranial Doppler
versus angiography in patients with vasospasm due to a ruptured cere-
bral aneurysm: A systematic review. Stroke 2001; 32:2292–2298
2. Arkuszewski M, Swiat M, Hurst RW, et al: Vertebral and basilar
arteries: Transcranial color-coded duplex ultrasonography versus
conventional TCD in detection of narrowings. Neuroradiol J 2012;
25:509–514
3. Proust F, Callonec F, Clavier E, et al: Usefulness of transcranial color-
coded sonography in the diagnosis of cerebral vasospasm. Stroke
1999; 30:1091–1098
4. Mariak Z, Krejza J, Swiercz M, et al: Accuracy of transcranial color
Doppler ultrasonography in the diagnosis of middle cerebral artery
spasm determined by receiver operating characteristic analysis. J
Neurosurg 2002; 96:323–330
5. Krejza J, Rudzinski W, Pawlak MA, et al: Angle-corrected imaging
transcranial Doppler sonography versus imaging and nonimaging
transcranial Doppler sonography in children with sickle cell disease.
AJNR Am J Neuroradiol 2007; 28:1613–1618
6. Krejza J, Mariak Z, Babikian VL: Importance of angle correction in the
measurement of blood flow velocity with transcranial Doppler sonog-
raphy. AJNR Am J Neuroradiol 2001; 22:1743–1747
7. Moher D, Liberati A, Tetzlaff J, et al; PRISMA Group: Preferred
Reporting Items for Systematic Reviews and Meta-Analyses: The
PRISMA statement. PLoS Med 2009; 6:e1000097
8. Whiting PF, Rutjes AW, Westwood ME, et al; QUADAS-2 Group:
QUADAS-2: A revised tool for the Quality Assessment of Diagnostic
Accuracy Studies. Ann Intern Med 2011; 155:529–536
9. Higgins JP, Thompson SG, Deeks JJ, et al: Measuring inconsistency
in meta-analyses. BMJ 2003; 327:557–560
10. Deeks JJ, Macaskill P, Irwig L: The performance of tests of publication
bias and other sample size effects in systematic reviews of diagnostic
test accuracy was assessed. J Clin Epidemiol 2005; 58:882–893
11. Doebler P, Holling H, Böhning D: A mixed model approach to meta-
analysis of diagnostic studies with binary test outcome. Psychol
Methods 2012; 17:418–436
12. Connolly ES Jr, Rabinstein AA, Carhuapoma JR, et al; American Heart
Association Stroke Council; Council on Cardiovascular Radiology
and Intervention; Council on Cardiovascular Nursing; Council
on Cardiovascular Surgery and Anesthesia; Council on Clinical
Cardiology: Guidelines for the management of aneurysmal subarach-
noid hemorrhage: A guideline for healthcare professionals from the
American Heart Association/American Stroke Association. Stroke
2012; 43:1711–1737
13. Wintermark M, Dillon WP, Smith WS, et al: Visual grading system for
vasospasm based on perfusion CT imaging: Comparisons with con-
ventional angiography and quantitative perfusion CT. Cerebrovasc
Dis 2008; 26:163–170
14. Vora YY, Suarez-Almazor M, Steinke DE, et al: Role of transcranial
Doppler monitoring in the diagnosis of cerebral vasospasm after sub-
arachnoid hemorrhage. Neurosurgery 1999; 44:1237–1247
15. Sviri GE, Newell DW, Lewis DH, et al: Impact of basilar artery vaso-
spasm on outcome in patients with severe cerebral vasospasm after
aneurysmal subarachnoid hemorrhage. Stroke 2006; 37:2738–2743
16. Karnchanapandh K: Effect of increased intracranial pressure on cere-
bral vasospasm in SAH. Acta Neurochir Suppl 2008; 102:307–310
17. Sviri GE, Britz GW, Lewis DH, et al: Dynamic perfusion computed
tomography in the diagnosis of cerebral vasospasm. Neurosurgery
2006; 59:319–325; discussion 319
18. Oskouian RJ Jr, Martin NA, Lee JH, et al: Multimodal quantitation of
the effects of endovascular therapy for vasospasm on cerebral blood
flow, transcranial Doppler ultrasonographic velocities, and cerebral
artery diameters. Neurosurgery 2002; 51:30–41
19. Suarez JI, Qureshi AI, Yahia AB, et al: Symptomatic vasospasm
diagnosis after subarachnoid hemorrhage: Evaluation of transcranial
Doppler ultrasound and cerebral angiography as related to compro-
mised vascular distribution. Crit Care Med 2002; 30:1348–1355
20. Veyna RS, Seyfried D, Burke DG, et al: Magnesium sulfate therapy
after aneurysmal subarachnoid hemorrhage. J Neurosurg 2002;
96:510–514
Mastantuono et al
21. Horn P, Vajkoczy P, Bauhuf C, et al: Quantitative regional cerebral
blood flow measurement techniques improve noninvasive detection
of cerebrovascular vasospasm after aneurysmal subarachnoid hemor-
rhage. Cerebrovasc Dis 2001; 12:197–202
22. Fontanella M, Valfrè W, Benech F, et al: Vasospasm after SAH due to
aneurysm rupture of the anterior circle of Willis: Value of TCD monitor-
ing. Neurol Res 2008; 30:256–261
23. Staalsø JM, Edsen T, Romner B, et al: Transcranial Doppler velocime-
try in aneurysmal subarachnoid haemorrhage: Intra- and interobserver
agreement and relation to angiographic vasospasm and mortality. Br
J Anaesth 2013; 110:577–585
24. Bellapart J, Jones L, Bandeshe H, et al: Plasma endothelin-1 as
screening marker for cerebral vasospasm after subarachnoid hemor-
rhage. Neurocrit Care 2014; 20:77–83
25. Cooke D, Koppula B, Seiler D, et al: Semiquantitative software
SPECT analysis in aneurysmal subarachnoid hemorrhage-related
vasospasm. Nucl Med Commun 2010; 31:53–58
26. Kincaid MS, Souter MJ, Treggiari MM, et al: Accuracy of transcra-
nial Doppler ultrasonography and single-photon emission computed
tomography in the diagnosis of angiographically demonstrated cere-
bral vasospasm. J Neurosurg 2009; 110:67–72
27. Badjatia N, Topcuoglu MA, Pryor JC, et al: Preliminary experience with
intra-arterial nicardipine as a treatment for cerebral vasospasm. AJNR
Am J Neuroradiol 2004; 25:819–826
28. Unterberg AW, Sakowitz OW, Sarrafzadeh AS, et al: Role of bedside
microdialysis in the diagnosis of cerebral vasospasm following aneu-
rysmal subarachnoid hemorrhage. J Neurosurg 2001; 94:740–749
29. Schmid-Elsaesser R, Kunz M, Zausinger S, et al: Intravenous magne-
sium versus nimodipine in the treatment of patients with aneurysmal
subarachnoid hemorrhage: A randomized study. Neurosurgery 2006;
58:1054–1065; discussion 1054
30. Hänggi D, Eicker S, Beseoglu K, et al: A multimodal concept in
patients after severe aneurysmal subarachnoid hemorrhage: Results
of a controlled single centre prospective randomized multimodal
phase I/II trial on cerebral vasospasm. Cent Eur Neurosurg 2009;
70:61–67
31. Hänggi D, Liersch J, Turowski B, et al: The effect of lumboventricular
lavage and simultaneous low-frequency head-motion therapy after
severe subarachnoid hemorrhage: Results of a single center prospec-
tive Phase II trial. J Neurosurg 2008; 108:1192–1199
32. Turowski B, Haenggi D, Wittsack J, et al: Cerebral perfusion com-
puterized tomography in vasospasm after subarachnoid hemorrhage:
Diagnostic value of MTT. Rofo 2007; 179:847–854
33. Rajendran JG, Lewis DH, Newell DW, et al: Brain SPECT used to
evaluate vasospasm after subarachnoid hemorrhage: Correlation
with angiography and transcranial Doppler. Clin Nucl Med
2001;26:125–130
34. Yokose N, Sakatani K, Murata Y, et al: Bedside monitoring of cerebral
blood oxygenation and hemodynamics after aneurysmal subarachnoid
hemorrhage by quantitative time-resolved near-infrared spectroscopy.
World Neurosurg 2010; 73:508–513
35. Rabinstein AA, Friedman JA, Weigand SD, et al: Predictors of cere-
bral infarction in aneurysmal subarachnoid hemorrhage. Stroke 2004;
35:1862–1866
36. Smith ML, Abrahams JM, Chandela S, et al: Subarachnoid hemor-
rhage on computed tomography scanning and the development
of cerebral vasospasm: The Fisher grade revisited. Surg Neurol
2005;63:229–234
37. Frontera JA, Fernandez A, Schmidt JM, et al: Defining vasospasm
after subarachnoid hemorrhage: What is the most clinically relevant
definition? Stroke 2009; 40:1963–1968
1672 www.ccmjournal.org October 2018 • Volume 46 • Number 10
Copyright © 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
38. Mascia L, Fedorko L, terBrugge K, et al: The accuracy of transcranial
Doppler to detect vasospasm in patients with aneurysmal subarach-
noid hemorrhage. Intensive Care Med 2003; 29:1088–1094
39. Krejza J, Kochanowicz J, Mariak Z, et al: Middle cerebral artery spasm
after subarachnoid hemorrhage: Detection with transcranial color-
coded duplex US. Radiology 2005; 236:621–629
40. Gonzalez NR, Boscardin WJ, Glenn T, et al: Vasospasm probabil-
ity index: A combination of transcranial Doppler velocities, cerebral
blood flow, and clinical risk factors to predict cerebral vasospasm
after aneurysmal subarachnoid hemorrhage. J Neurosurg 2007;
107:1101–1112
41. Carrera E, Schmidt JM, Oddo M, et al: Transcranial Doppler ultra-
sound in the acute phase of aneurysmal subarachnoid hemorrhage.
Cerebrovasc Dis 2009; 27:579–584
42. Wang HC, Lin WC, Yang TM, et al: Time course of cerebral hemody-
namics in aneurysmal subarachnoid hemorrhage. J Clin Ultrasound
2012; 40:91–98
43. Sebastian J, Derksen C, Khan K, et al: Derivation of transcra-
nial Doppler criteria for angiographically proven middle cerebral
artery vasospasm after aneurysmal subarachnoid hemorrhage.
J Neuroimaging 2013; 23:489–494
44. Sviri GE, Ghodke B, Britz GW, et al: Transcranial Doppler grading cri-
teria for basilar artery vasospasm. Neurosurgery 2006; 59:360–366;
discussion 360
45. Wintermark M, Ko NU, Smith WS, et al: Vasospasm after subarach-
noid hemorrhage: Utility of perfusion CT and CT angiography on diag-
nosis and management. AJNR Am J Neuroradiol 2006; 27:26–34
46. Proust F, Debono B, Gérardin E, et al: Angiographic cerebral vaso-
spasm and delayed ischemic deficit on anterior part of the circle
of Willis. Usefulness of transcranial Doppler. Neurochirurgie 2002;
48:489–499
47. Kyoi K, Hashimoto H, Tokunaga H, et al: Time course of blood veloc-
ity changes and clinical symptoms related to cerebral vasospasm and
prognosis after aneurysmal surgery. No Shinkei Geka 1989; 17:21–30
48. Sloan MA, Haley EC Jr, Kassell NF, et al: Sensitivity and specificity of
transcranial Doppler ultrasonography in the diagnosis of vasospasm fol-
lowing subarachnoid hemorrhage. Neurology 1989; 39:1514–1518
4 9. Langlois O, Rabehenoina C, Proust F, et al: Diagnosis of vaso-
spasm: Comparison between arteriography and transcranial
Doppler. A series of 112 comparative tests. Neurochirurgie 1992;
38:138–140
50. Burch CM, Wozniak MA, Sloan MA, et al: Detection of intracranial
internal carotid artery and middle cerebral artery vasospasm following
subarachnoid hemorrhage. J Neuroimaging 1996; 6:8–15
51. Lennihan L, Petty GW, Fink ME, et al: Transcranial Doppler detection
of anterior cerebral artery vasospasm. J Neurol Neurosurg Psychiatry
1993; 56:906–909
52. Sloan MA, Burch CM, Wozniak MA, et al: Transcranial Doppler detec-
tion of vertebrobasilar vasospasm following subarachnoid hemor-
rhage. Stroke 1994; 25:2187–2197
53. Wozniak MA, Sloan MA, Rothman MI, et al: Detection of vasospasm
by transcranial Doppler sonography. The challenges of the anterior
and posterior cerebral arteries. J Neuroimaging 1996; 6:87–93
54. Genders TS, Spronk S, Stijnen T, et al: Methods for calculating sen-
sitivity and specificity of clustered data: A tutorial. Radiology 2012;
265:910–916
55. Leeflang MM, Deeks JJ, Takwoingi Y, et al: Cochrane diagnostic test
accuracy reviews. Syst Rev 2013; 2:82
56. Bossuyt PM, Reitsma JB, Bruns DE, et al; STARD Group: STARD
2015: An updated list of essential items for reporting diagnostic
accuracy studies. Radiology 2015; 277:826–832