Original Article

Journal of Intensive Care Medicine

2020, Vol. 35(11) 1226-1234
Venous Thromboembolism in Neurocritical ª The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
Care Patients DOI: 10.1177/0885066619841547
journals.sagepub.com/home/jic

Tanuwong Viarasilpa, MD1,2, Nicha Panyavachiraporn, MD1,2,

Jack Jordan, MS3, Seyed Mani Marashi, PhD4, Meredith van Harn, MS5,
Noel O. Akioyamen, BScN1, Robert G. Kowalski, MBBCh, MS1,
and Stephan A. Mayer, MD1

Abstract
Background: Venous thromboembolism (VTE) is a potentially life-threatening complication among critically ill patients.
Neurocritical care patients are presumed to be at high risk for VTE; however, data regarding risk factors in this population are
limited. We designed this study to evaluate the frequency, risk factors, and clinical impact of VTE in neurocritical care patients.
Methods: We obtained data from the electronic medical record of all adult patients admitted to neurological intensive care unit
(NICU) at Henry Ford Hospital between January 2015 and March 2018. Venous thromboembolism was defined as deep vein
thrombosis, pulmonary embolism, or both diagnosed by Doppler, chest computed tomography (CT) angiography or ventilation–
perfusion scan >24 hours after admission. Patients with ICU length of stay <24 hours or who received therapeutic anticoagulants
or were diagnosed with VTE within 24 hours of admission were excluded. Results: Among 2188 consecutive NICU patients,
63 (2.9%) developed VTE. Prophylactic anticoagulant use was similar in patients with and without VTE (95% vs 92%; P ¼ .482).
Venous thromboembolism was associated with higher mortality (24% vs 13%, P ¼ .019), and longer ICU (12 [interquartile range,
IQR 5-23] vs 3 [IQR 2-8] days, P < .001) and hospital (22 [IQR 15-36] vs 8 [IQR 5-15] days, P < .001) length of stay. In a mul-
tivariable analysis, potentially modifiable predictors of VTE included central venous catheterization (odds ratio [OR] 3.01; 95%
confidence interval [CI], 1.69-5.38; P < .001) and longer duration of immobilization (Braden activity score <3, OR 1.07 per day;
95% CI, 1.05-1.09; P < .001). Nonmodifiable predictors included higher International Medical Prevention Registry on Venous
Thromboembolism (IMPROVE) scores (which accounts for age >60, prior VTE, cancer and thrombophilia; OR 1.66; 95% CI,
1.40-1.97; P < .001) and body mass index (OR 1.05; 95% CI, 1.01-1.08; P ¼ .007). Conclusions: Despite chemoprophylaxis, VTE
still occurred in 2.9% of neurocritical care patients. Longer duration of immobilization and central venous catheterization are
potentially modifiable risk factors for VTE in critically ill neurological patients.

Keywords
venous thromboembolism, neurocritical care, incidence, risk factor, outcome

Introduction
Venous thromboembolism (VTE) encompasses 2 clinical
1
entities, deep vein thrombosis (DVT) and pulmonary embo- Department of Neurology, Henry Ford Hospital, Detroit, MI, USA
2
Division of Critical Care, Department of Medicine, Siriraj Hospital, Mahidol
lism (PE), and is a major complication among critically ill
University, Bangkok, Thailand
patients, and still occurs despite widespread use of thrombo- 3
Department of Quality Administration, Henry Ford Hospital, Detroit, MI,
prophylaxis.1-5 The reported incidence of VTE has ranged USA
4
from 1.2% to 31.6% in various groups of neurological Department of Strategic and Operational Analytics, Henry Ford Hospital,
patients6-14; however, the data regarding risk factors in criti- Detroit, MI, USA
5
Department of Public Health Sciences, Henry Ford Hospital, Detroit, MI, USA
cally ill neurological patients are limited. In this population,
VTE risk is presumed to be high because of prolonged coma Received February 1, 2019. Received revised March 8, 2019. Accepted
and paralysis, as well as vascular endothelial activation from March 13, 2019.
neurologic diseases, and anticoagulation prophylaxis is rec-
Corresponding Author:
ommended.15-17 We designed this study to evaluate the inci- Stephan A. Mayer, Department of Neurology, Henry Ford Hospital, 2799 W
dence, risk factors, and impact on outcome of VTE in Grand Blvd, Detroit, MI 48202, USA.
neurocritical care patients. Email: [email protected]
Viarasilpa et al 1227

Methods Table 1. Characteristics of In-Hospital Venous Thromboembolism,

n ¼ 63.a
Study Population
Type of VTE
We obtained data from Henry Ford Health System (HFHS) VTE Isolated DVT 29 (46)
database that was built from the electronic medical record Isolated PE 11 (17)
(EMR; www.EPIC.com) of adult patients admitted to all inten- Both DVT and PE 23 (37)
sive care (ICU) and step-down units in the HFHS between Jan- Anatomical Location of DVTb n ¼ 52
uary 2015 and March 2018. The VTE database was created as a Lower extremity DVT 45 (87)
quality assurance program in the HFHS Department of Quality Upper extremity DVTc 4 (8)
Both upper and lower extremity DVT 2 (4)
Administration. In this analysis, we included patients aged 18
Anatomical Location of PEd n ¼ 34
years who admitted to the Henry Ford Hospital (HFH) neurolo- Pulmonary trunk, right or left main pulmonary artery 11 (32)
gical intensive care unit (NICU; 16 beds) for at least 24 hours. In Lobar or segmental branches 19 (56)
our NICU, chemoprophylaxis against VTE is started on admis- Isolated subsegmental branchese 2 (6)
sion or 12 hours after uncomplicated surgery, clipping of a rup- Clinically significant VTEf 52 (83)
tured aneurysm, or demonstration of stability of intracranial Median time from admission to VTE diagnosis, daysg 9 [4-14]
blood. Patients diagnosed with VTE within 24 hours of admis- Abbreviations: DVT, deep vein thrombosis; IQR, interquartile range; PE, pul-
sion, who had VTE as an admission diagnosis, or who received monary embolism; VTE, venous thromboembolism.
a
therapeutic anticoagulants within 24 hours of hospital or NICU b
Data are n (% of total available data within each column) or median [IQR].
admission were excluded. This study was approved by HFHS With or without PE. No data for location of DVT in 1 patient. Thirty-six (69%)
patients had proximal DVT and 15 (29%) patients had isolated distal lower
Institutional Review Board. As all data were anonymized and extremity (distal to popliteal vein) DVT.
the study posed no significant risks, the requirement for written c
Upper extremity DVT was defined as thrombosis of superior vena cava,
informed consent was waived. internal jugular, subclavian, brachiocephalic, axillary or brachial veins.
d
With or without DVT. No data for location of pulmonary embolism in 2
patients.
e
Two patients with subsegmental PE did not have proximal DVT.
Definition of In-Hospital VTE f
Defined as thrombosis involving pulmonary trunk, right or left main, lobar, or
In-hospital VTE was defined as acute incident DVT (either upper segmental branches of pulmonary arteries or DVT involving popliteal or more
proximal segments of lower extremity veins or upper extremity veins.
or lower extremities), PE, or both that were newly diagnosed by g
The diagnosis date of in-hospital venous thromboembolism was the date of
duplex venous ultrasonography, computed tomographic venogra- vascular imaging confirmation.
phy (CTV), computed tomographic pulmonary angiography or
ventilation–perfusion scan (V/Q scan) of the lungs after 24 hours
of hospitalization. In the HFHS-VTE database, VTE events were (defined as VTE occurred before hospital admission), history
identified from discharge summary documents. To confirm of cancer, thrombophilia,26 International Medical Prevention
the diagnosis of in-hospital VTE, we verified all VTE Registry on Venous Thromboembolism (IMPROVE) VTE risk
events from the HFHS-VTE database by manually review- score (accounted for age >60 years, history of cancer, known
ing the reports of vascular imaging or V/Q scan in the thrombophilia and previous history of VTE; patients with
EMR; the date of VTE diagnosis and anatomical location IMPROVE score  2 were classified as high VTE risk; Supple-
of VTE were also documented. The diagnosis date of in- mental Table 1),27 discharge diagnoses according to diagnosis-
hospital VTE was defined as the date of vascular imaging related group (DRG; Supplemental Table 2), baseline laboratory
confirmation. Upper extremity DVT included DVT involv- results included platelets, prothrombin time (PT), international
ing superior vena cava, internal jugular, subclavian, bra- normalized ratio, activated partial thromboplastin time, and
chiocephalic, axillary, or brachial veins. Proximal DVT serum creatinine, and renal dysfunction (defined as baseline
was defined as DVT involving popliteal or more proximal glomerular filtration rate < 60 mL/min/1.73m2).
segments of lower extremity veins or involving upper In-hospital immobilization was defined as an activity sub-
extremity veins. Isolated distal DVT was defined as DVT score of 1 or 2 in Braden Scale, which is originally used for
involving the deep veins of lower extremity distal to popli- predicting risk for pressure ulcer. The Braden Scale consists of 6
teal veins in the absence of proximal DVT.18 Because the subscores that assess sensory perception, skin moisture, activity,
risk of progressive or recurrent VTE in subsegmental PE mobility, friction and shear, and nutrition status. For the activity
and isolated distal DVT is uncertain,19-24 we defined clini- subscore, 1 ¼ patients who are confined to bed, 2 ¼ patients who
cally significant VTE as PE involving the pulmonary have severely limited ability to walk, cannot bear own weight
trunk; right or left main, lobar or segmental branches of and must be assisted into chair or wheelchair, 3 ¼ patients who
the pulmonary artery; or proximal DVT. can walk occasionally, with or without assistance, but for very
short distances, and 4 ¼ patients who can walk frequently.28
This score was recorded daily in the EMR, and the activity
Data Collection subscores were abstracted as a part of HFHS-VTE database.
Data collected included patient demographics, body mass index Anticoagulant prophylaxis was defined as at least 1 prophy-
(BMI), Charlson comorbidity index,25 prior history of VTE lactic dose of unfractionated heparin, low-molecular-weight
1228 Journal of Intensive Care Medicine 35(11)

Table 2. Patient Characteristics and Treatments.a

Characteristics With VTE, n ¼ 63 Without VTE, n ¼ 2125 P Value

Demographics
Age, years, mean (SD) 57 (16) 59 (17) .469
Male 38 (60) 1056 (50) .097
BMI, kg/m2, mean (SD) 31.1 (9.4) 28.2 (7.3) .006
Race .226
White 31 (51) 1103 (56)
African American 29 (48) 772 (39)
Others 1 (2) 108 (5)
Charlson comorbidity index, mean (SD) 6.5 (3.5) 6.0 (3.9) .243
Previous history of VTE 37 (59) 330 (16) <.001
History of cancer 15 (24) 461 (22) .688
Thrombophilia 0 (0) 7 (0.3) >.999
Renal dysfunctionb 4 (13) 197 (22) .249
Duration of in-hospital immobilization, days,c mean (SD) 26 (18) 9 (9) <.001
VTE risk prediction score
IMPROVE VTE risk score d 3 [1-4] 1 [0 -1] <.001
Patients with IMPROVE score  2 39 (62) 531 (25) <.001
Laboratory data on admission
Platelet count,  103/mL, mean (SD) 247 (111) 226 (88) .072
PT, seconds, mean (SD) 16.0 (4.4) 15.1 (4.2) .021
INR, mean (SD) 1.3 (0.5) 1.2 (0.5) .028
APTT, seconds, mean (SD) 31.9 (8.4) 31.2 (10.9) .273
Creatinine, mg/dL, mean (SD) 1.1 (1.3) 1.1 (1.1) .419
Methods of VTE prophylaxis
Anticoagulant prophylaxise 60 (95) 1947 (92) .482
Mechanical prophylaxis 63 (100) 2097 (99) >.999
Duration of VTE prophylaxis
Duration of anticoagulant prophylaxis, days, mean (SD) 14 (11) 9 (9) <.001
Duration of mechanical prophylaxis, days, mean (SD) 25 (19) 10 (9) <.001
Treatments f
Central venous catheterization 38 (60) 378 (18) <.001
Mechanical ventilation 47 (75) 878 (41) <.001
Blood transfusion 8 (13) 305 (14) .712
RBC transfusion 7 (11) 274 (13) .677
Platelet transfusion 0 (0) 51 (2) .401
FFP transfusion 2 (3) 42 (2) .364

Abbreviations: APTT, activated partial thromboplastin time; BMI, body mass index; FFP, fresh frozen plasma; INR, international normalized ratio; IQR, interquartile
range; PT, prothrombin time; RBC, red blood cells; SD, standard deviation; VTE, venous thromboembolism.
Bold P values are statistically significant (P < 0.05).
a
Data are n (% of total available data within each column), mean (SD) or median [IQR].
b
Baseline glomerular filtration rate (GFR) <60 mL/min/1.73 m2.
c
Total duration that patients had activity subscore of Braden Scale < 3 (1 ¼ confined to bed, 2 ¼ severely limited ability to walk, cannot bear own weight and must
be assisted into chair, 3 ¼ walk occasionally but for very short distances, 4 ¼ walk frequently).
d
International Medical Prevention Registry on Venous Thromboembolism (IMPROVE) risk score is accounted for age > 60 years, history of cancer, known
thrombophilia and previous history of VTE; patients with IMPROVE score  2 were classified as high VTE risk.
e
Unfractionated heparin was used in 1975 patients (99%), Low-molecular-weight heparin was used in 1%, and only 1 patient received apixaban in prophylactic
dosage.
f
Data for central venous catheterization, mechanical ventilation and blood transfusion were censored on or after the day of VTE diagnosis.

heparin, or a direct-acting oral anticoagulant given during hos- triple lumen, dialysis, or tunneled catheters placed into the
pitalization or for VTE patients before VTE diagnosis (Supple- internal jugular, subclavian, or femoral veins; pulmonary artery
mental Table 3).16,26,29-35 Mechanical prophylaxis included catheters; Port-A catheters; and peripherally inserted central
graduated compression stocking or intermittent pneumatic catheters (PICCs). Outcome assessments included in-hospital
compression use. Duration of anticoagulant and mechanical mortality, intensive care unit (ICU) and hospital length of stay,
prophylaxis was recorded. duration of mechanical ventilation, Glasgow Coma Scale
Treatments during hospitalization that we recorded included (GCS) at hospital discharge, and discharge disposition.
blood transfusions, mechanical ventilation, and central venous All data were abstracted from the EMR to the HFHS-VTE
catheterization (CVC), which were defined as single, double, or database. For patients with in-hospital VTE, we performed
Viarasilpa et al 1229

Table 3. Patient Status at Hospital Discharge.a infection, 10% had other neurological diseases, and 15% had
a non-neurological condition documented as a major discharge
With VTE, Without VTE, P
Characteristics n ¼ 63 n ¼ 2125 Value
diagnosis (Supplemental Table 2).

Mortality 15 (24) 286 (13) .019 Venous Thromboembolism Characteristics

GCS at discharge 14 [11-15] 15 [13-15] .050
Duration of mechanical 13 (13) 8 (8) .003 In-hospital VTE occurred in 63 patients (2.9%; Figure 1); 29
ventilation, days, mean (SD) (1.3%) patients had isolated DVT, 11 (0.5%) patients had
ICU length of stay, days 12 [5-23] 3 [2-8] <.001 isolated PE, and 23 (1.1%) patients had both DVT and PE.
Hospital length of stay, days 22 [15-36] 8 [5-15] <.001
Clinically significant VTE occurred in 52 patients (83% of
Discharge disposition .001
Home or Self care 1 (2) 11 (1) those with VTE). Among patients with DVT, 87% had isolated
Skilled nursing or long-term 17 (27) 387 (18) lower extremity involvement, 8% had isolated upper extremity
chronic care involvement, and 4% had both upper and lower extremity
Rehabilitation facility 27 (43) 904 (43) involvement. Median time from hospital admission to VTE
Transferred to other hospital 3 (5) 537 (25) diagnosis was 9 days (interquartile range [IQR], 4-14 days;
Abbreviations: GCS, Glasgow Coma Scale; IQR, interquartile range; SD, stan-
Table 1; Figure 2); 86% of VTE events occurred during che-
dard deviation; VTE, Venous thromboembolism. moprophylaxis. Among 38 patients with VTE and a CVC, 7 of
Bold P values are statistically significant (P < 0.05). them had isolated PE without DVT. Of the 30 patients with
a
Data are n (% of total available data within each column), mean (SD) or known location of DVT and CVC, DVT occurred at the loca-
median [IQR].
tion of the CVC in 12 (40%) patients and at a different location
in 18 patients (60%; Supplemental Table 4).
additional searches for the first date of CVC, mechanical ven-
tilation, and blood transfusion in the EMR. Data for CVC, Baseline Characteristics
mechanical ventilation, and blood transfusion on or after the
day of VTE diagnosis were censored. Mean age of all patients was 59 years; 55% were white and 39%
were African American. Patient demographics and the Charlson
comorbidity index total scores were not significantly different
Statistical Analysis between patients with and without VTE. In univariable analysis,
To identify risk factors for in-hospital VTE, we compared BMI was higher (31.1 [9.4] vs 28.2 [7.3] kg/m2, P ¼ .006), and a
data between patients with and without VTE. Continuous previous history of VTE was more frequent (59% vs 16%, P <
data were described using means, standard deviations, med- .001) in patients with VTE compared to those without. Patients
ians, 25th quartiles, and 75th quartiles as appropriate, while with VTE had longer duration of in-hospital immobilization (26
categorical data were described using counts and column [18] vs 9 [9] days, P < .001), longer PT (16.0 [4.4] vs 15.1 [4.2]
percentages. Univariate 2-group comparisons are carried out seconds, P ¼ .021), and higher IMPROVE VTE risk scores;
using Wilcoxon rank-sum tests for continuous variables due however, only 62% of patients with VTE were identified as high
to non-normal distributions and using w2 tests or Fisher risk by IMPROVE score (Table 2).
exact tests for categorical variables as appropriate based
on expected cell count. Multivariable logistic regression Treatment
modeling was carried out to identify independent predictors
of VTE, and adjusted odds ratios (ORs) with 95% confi- Almost all patients received mechanical prophylaxis. Prophy-
dence intervals (CIs) were reported. Statistical significance lactic anticoagulants were used in 92% of patients overall, with
is set at P < .05. All analyses were performed using SAS 9.4 no difference in utilization in those with and without VTE
(SAS Institute Inc, Cary, North Carolina). (Table 2). Unfractionated heparin was the most commonly
used anticoagulant for VTE prophylaxis (98%). Duration of
pharmacologic (14 [11] vs 9 [9] days, P < .001) and mechanical
Results (25 [19] vs 10 [9] days, P < .001) thromboprophylaxis was
Study Population longer in patients with VTE. Central venous catheterization
(60% vs 18%, P < .001) and mechanical ventilation (75% vs
From January 2015 to March 2018, 52 351 patients were admit- 41%, P < 0.001) were more frequently used in patients with
ted to all ICU and step-down units in the HFHS. From that VTE. All types of blood transfusion were not significantly
sample, 2781 patients were admitted to the HFH NICU. Of different between patients with and without VTE (Table 2).
these, 593 patients were excluded, leaving 2188 patients for
inclusion in the final analysis (Figure 1). Of these patients,
based on discharge DRG, 50% required a neurosurgical proce-
Patient Status at Hospital Discharge
dure or endovascular intervention, 16% had acute ischemic Mortality was significantly higher in patients with VTE com-
stroke or intracerebral hemorrhage, 8% had trauma-related pared to those without (24% vs 13%, P ¼ .019). In-hospital
brain or spinal cord injury, 1% had central nervous system VTE was associated with lower discharge GCS, longer
1230 Journal of Intensive Care Medicine 35(11)

Figure 1. Study population. *Henry Ford Health System Venous Thromboembolism (HFHS-VTE) database was built from the electronic
medical record of adult patients admitted to all intensive care (ICU) and step-down unit in the HFHS between January 2015 and March 2018.
NICU indicates neurological intensive care unit; ICU, intensive care unit; VTE, venous thromboembolism.

Figure 2. Time from admission to venous thromboembolism (VTE) diagnosis. *No data for VTE diagnosis date in 2 patients.

Table 4. Multivariable Analysis for Predictors of In-hospital VTE. duration of mechanical ventilation (13 [13] vs 8 [8] days, P ¼
Adjusted Odds
.003) and longer ICU (12 [IQR 5-23] vs 3 [IQR 2-8] days, P <
Factor Ratios (95% CI) a P Value .001) and hospital (22 [IQR 15-36] vs 8 [IQR 5-15] days, P <
.001) length of stay (Table 3).
Central venous catheterization 3.01 (1.69-5.38) <.001
IMPROVE VTE risk score 1.66 (1.40-1.97) a <.001
Duration of in-hospital immobilization 1.07 (1.05-1.09) a <.001 Predictors of In-Hospital VTE
Body mass index (BMI) 1.05 (1.01-1.08) a .007
Mechanical ventilation 1.76 (0.90-3.47) .100 In multivariable logistic regression analysis, CVC (OR 3.01,
Prothrombin time at admission 1.01 (0.97-1.06)a .606 95% CI 1.69-5.38, P < .001), higher IMPROVE score (OR
1.66, 95% CI 1.40-1.97, P < .001), longer duration of immo-
Abbreviations: CI, confidence interval; IMPROVE, International Medical Preven-
tion Registry on Venous Thromboembolism; VTE, venous thromboembolism.
bilization (OR 1.07, 95% CI 1.05-1.09, P < .001), and higher
Bold P values are statistically significant (P < 0.05). BMI (OR 1.05, 95% CI 1.01-1.08, P ¼ 0.007) were indepen-
a
For continuous variables, odds ratios are applied for every 1-unit increase in dent predictors of in-hospital VTE. Mechanical ventilation and
each factor.
Viarasilpa et al 1231

admission prothrombin time did not have statistical signifi- subclavian vein catheters were placed at the same location of
cance in the multivariate model (Table 4). DVT in 4 (67%) of 6 patients with upper extremity DVT.
However, in patients with lower extremity DVT (n ¼ 45), only
8 (20%) patients had femoral vein catheters placed at the same
Discussion side of DVT, 1 (2%) patient had femoral vein catheter place at
In this large cohort of neurocritical care patients, we found that the different side of DVT, 20 (44%) patients had internal jugu-
the overall frequency of in-hospital VTE was 2.9%. The major- lar or subclavian catheters, and 20 (44%) patients did not have
ity (83%) of these events were clinically significant. Anticoa- CVC (Supplemental Table 4). Since 60% of patients with DVT
gulant prophylaxis was used in the vast majority of patients and CVC had CVC placed at the different location of DVT, it
(92%) in the study cohort, most cases of VTE occurred during appears that CVC was an independent risk factor for VTE
chemoprophylaxis, and failure to anticoagulate was not identi- regardless of its position. Avoidance or early removal of CVC
fied as a risk factor for VTE. In multivariable analysis, CVC when it is no longer required may minimize the risk of VTE in
and longer duration of in-hospital immobilization were identi- neurocritical care patients.
fied as potentially modifiable risk factors for VTE, whereas The IMPROVE VTE risk score, which accounts for age >60
higher BMI and IMPROVE VTE risk scores (which accounts years, or a history of cancer, thrombophilia, or previous VTE,
for age >60 years, prior VTE, cancer and thrombophilia) were has been validated as a predictor of in-hospital or posthospita-
identified as nonmodifiable risk factors. In-hospital VTE was lization (up to 90 days) thromboembolism in acutely ill medical
associated with higher mortality, lower discharge GCS, and patients.27 In our study, an increased IMPROVE score was an
longer duration of mechanical ventilation and ICU and hospital independent predictor of in-hospital VTE. The univariable
length of stay. analysis suggests that this association is driven primarily by
The frequency of VTE in our neurocritical care population prior history of VTE. While 7% of patients (39 of 570) classi-
(2.9% of 2188 patients) was comparable to that reported in a fied as high risk (score  2) on the IMPROVE score developed
recent study (3.3% of 4632 patients).36 In other studies, VTE VTE, 24 (38%) of 63 patients with VTE were categorized as
incidence has ranged from 1.2% to 2.8% after acute ischemic
low VTE risk by the IMPROVE score. Therefore, the
stroke,6,7 2.6% to 2.9% after intracerebral hemorrhage,8,37 and
IMPROVE score alone is not sufficient to identify neurological
4.4% to 18% after subarachnoid hemorrhage.9,10,37 By contrast,
patients at risk of in-hospital VTE.
DVT rates exceeding 30% have been reported with routine
Prolonged duration of in-hospital immobilization was inde-
screening Doppler ultrasonography after moderate-to-severe
pendently associated with VTE in our cohort. Immobilization
traumatic brain injury (without chemoprophylaxis)13 or large
is a well-recognized risk factor for VTE; however, definitions
hemispheric infarction requiring decompressive hemicraniect-
of immobilization have varied across studies.27,51-53 In our
omy (with chemoprophylaxis).38 We found no association
study, we adapted the activity subscore of Braden Scale28
between VTE and age, gender, or any particular diagnostic
because it was assessed daily as part of pressure ulcer risk
group. Compared to the previous studies of medical–surgical
critically ill patients reporting VTE rates of 2% to 10%,2-5 the evaluation in our EMR; patients with the score of 1 or 2 (con-
frequency of VTE in neurocritical care patients appears to be fined to bed or cannot bear own weight) were classified as
on the lower side; a higher incidence of VTE in sepsis and immobilized. Intensive care unit mobilization has increasingly
septic shock patients might explain this finding.18 gained traction as a means to improve functional outcome and
Central venous catheterization was found to be 1 of 2 poten- reduce ICU and hospital length of stay in critically ill
tially modifiable risk factors for VTE in our study; 9% of patients.54-56 Our findings support the concept that reduced
patients with a CVC developed VTE. Many studies in critically VTE risk may be another benefit of early mobilization, at least
ill, non-critically ill, and patients with sepsis support this in neurological patients.
association.18,39,40 In patients with sepsis and critically ill In our study, high BMI was associated with an increased risk
medical patients receiving thromboprophylaxis, CVC of in-hospital VTE. Chronic inflammation and impaired fibri-
increased the risk of VTE by an OR of 2.82 (95% CI 1.07- nolysis promote prothrombotic states in obesity.57 High BMI
7.38) and 2.64 (95% CI 1.87-3.72), respectively.18,39 Central has been reported to be an independent risk factor for throm-
venous catheter–related thrombosis is an important complica- boprophylaxis failure in critically ill patients and has been
tion of CVC insertion41 and a risk factor for upper extremity associated with VTE in patients with subarachnoid hemorrhage
DVT.42,43 Regarding the types of CVC, PICCs have been asso- and large hemispheric infarction undergoing decompressive
ciated with a higher risk of VTE than centrally inserted central hemicraniectomy in previous studies. 5,9,38,39,58 Higher
catheters.44-46 Local vessel wall injury and stasis of blood flow weight-based doses of anticoagulant prophylaxis have been
caused by CVC is thought to contribute to the pathogenesis of reported to maintain peak plasma anti-Xa levels within or near
VTE.47,48 In addition, a hypercoagulable state induced by cen- the recommended range for thromboprophylaxis and to reduce
tral venous catheter or pulmonary artery catheter insertion has in-hospital VTE in morbidly obese patients without increasing
been demonstrated with thromboelastography in both animal the risk of bleeding.59-62 Further study is required to determine
and human studies.49,50 In our cohort, CVC most often con- the efficacy and safety of weight-based chemoprophylaxis in
tributed to DVT in an upper extremity. Internal jugular or obese patients.
1232 Journal of Intensive Care Medicine 35(11)

Median time from hospital admission to VTE diagnosis was DRG only, the data for specific diagnosis of patients who
9 [IQR 4-14] days in our study. The median time to VTE underwent neurosurgical procedure (n ¼ 1086) were not avail-
diagnosis has varied from 6 to 18 days in previous studies of able in our data set. We did not have the information regarding
neurological patients.14,37,38 Based on these data, screening prothrombin complex concentrate treatment and were unable to
compression ultrasonography at about 1 week is reasonable obtain data on possible complications of anticoagulant prophy-
and might be beneficial, particularly in high-risk patients who laxis, such as bleeding or heparin-induced thrombocytopenia.
are obese, older than 60 years, immobilized, have central The single-center nature of our study may limit the general-
venous access, or have a history of cancer, thrombophilia, or izability of our findings. Finally, we did not record data on
prior VTE. VTE occurrence and clinical outcome after hospital discharge.
In-hospital VTE was associated with lower GCS scores,
higher mortality, and longer duration of mechanical ventila- Conclusion
tion and ICU and hospital length of stay in our study. These
associations have also been reported in previous studies of In neurocritical care patients, VTE still occurred despite throm-
medical–surgical critically ill, patients with sepsis and boprophylaxis. Our study has practical implications for patient
trauma.2,18,63,64 By contrast, there was no association between care. Early mobilization and efforts to minimize the use and
VTE and mortality in the recent large cohort of neurocritical expedite the removal of central venous catheters may reduce
care patients reported by Sauro et al.36 Although PE can be the risk of VTE in critically ill neurological patients.
fatal, it is a low-frequency event. This study was not primarily
Acknowledgments
designed to explore whether VTE is a risk factor for hospital
mortality or longer length of stay. It seems much more likely We thank Dr Ahmad Riad Ramadan, and the neuro-ICU nurses for the
excellent care provided to our patients, and for their support of
that high morbidity puts patients at risk of VTE rather than
research conducted in neurointensive care unit.
the converse.
Strengths of our study include the relatively large sample Authors’ Note
size. We performed in-depth medical record searches to verify
Tanuwong Viarasilpa had full access to all the data in the study and
all VTE events, document the anatomical location of VTE, takes responsibility for the integrity of the data and the accuracy of the
determine the date of VTE diagnosis, and to determine whether data analysis, performed manual searches for additional data collec-
CVC insertion, mechanical ventilation, and blood transfusion tion, analysis and interpretation, drafted and revised the manuscript.
was performed in patients with in-hospital VTE. We did not Nicha Panyavachiraporn performed manual searches for additional
consider treatments given to patients after the occurrence of data collection, analysis and interpretation, drafted, and revised the
VTE as risk factor for VTE; we censored the data for CVC, manuscript. Jack Jordan and Seyed Mani Marashi created the Henry
mechanical ventilation, and blood transfusion after VTE diag- Ford Health System Venous Thromboembolism Database from the
nosis. Although the incidence of VTE was low, presumably due electronic medical records and provided the data for research purpose.
to high compliance with thromboprophylaxis, the majority of Noel O. Akioyamen performed manual searches for additional data
VTE occurred during prophylaxis period were clinically sig- collection. Meredith van Harn performed the statistical analysis and
critically reviewed the manuscript. Robert G. Kowalski assisted in the
nificant, and impacted patient outcomes.
data analysis and critically reviewed the manuscript. Stephan A.
There are also limitations of this study. Perhaps most impor- Mayer designed the study and data analysis, performed data review
tantly, VTE is prone to underdiagnosis, a drawback that affects and interpretation, and drafted and revised the manuscript. All authors
all large cohort studies derived from large administrative data have read and approved the final manuscript and agreed to be accoun-
sets. We relied on clinicians establishing the diagnosis of VTE table for all aspects of the work.
based on diagnostic testing in response to clinical signs and The data sets used and/or analyzed during the current study are
symptoms, since routine surveillance with Doppler venous available from the corresponding author on reasonable request.
ultrasonography and CT pulmonary angiography was not per- This study was approved by the Henry Ford Health System Institu-
formed, as recommended by current guidelines.26 Because we tional Review Board. The requirement for written informed consent
did not routinely perform admission screening for VTE, it was waived because all data were anonymized and the study posed no
might be possible that some patients with VTE developed this significant risks.
prior to or at the time of admission. We counted even a single Declaration of Conflicting Interests
dose of prophylactic anticoagulant as an attempt to provide
The author(s) declared the following potential conflicts of interest
chemoprophylaxis, and skipped or missed doses of anticoagu-
with respect to the research, authorship, and/or publication of this
lants were not considered. Because 98% of our cohort received article: S.A.M. has received consulting fees from UCB Pharma. Other
unfractionated heparin for thromboprophylaxis, we could not authors declare that they have no competing interests.
compare the efficacy of different type of anticoagulants. The
position, type, and duration of CVC placement were available Funding
in patients with VTE only. We did not have data for standard The author(s) disclosed receipt of the following financial support for
ICU clinical physiologic severity scores or detailed data on use the research, authorship, and/or publication of this article: This study
of vasopressors or ventilator settings. Because we were able to was funded by Educational and Research Fund from Department of
acquire the data for discharge diagnosis from the principal Neurology of Henry Ford Hospital. Tanuwong Viarasilpa has received
Viarasilpa et al 1233

funding from Siriraj Hospital, Mahidol University, Bangkok, Thai- 16. Gould MK, Garcia DA, Wren SM, et al. Prevention of VTE in
land. The funding sources had no role in the collection, analysis, or nonorthopedic surgical patients: Antithrombotic Therapy and Pre-
interpretation of the data. vention of Thrombosis, 9th ed: American College of Chest Phy-
sicians Evidence-Based Clinical Practice Guidelines. Chest.
Supplemental Material 2012;141(suppl 2):e227S-e77S.
Supplemental material for this article is available online. 17. Nyquist P, Bautista C, Jichici D, et al. Prophylaxis of venous
thrombosis in neurocritical care patients: an evidence-based
guideline: a statement for healthcare professionals from the neu-
References rocritical care society. Neurocrit Care. 2016;24(1):47-60.
1. Minet C, Potton L, Bonadona A, et al. Venous thromboembolism 18. Kaplan D, Casper TC, Elliott CG, et al. VTE Incidence and risk
in the ICU: main characteristics, diagnosis and thromboprophy- factors in patients with severe sepsis and septic shock. Chest.
laxis. Crit Care. 2015;19(1):287. 2015;148(5):1224-1230.
2. Cook D, Crowther M, Meade M, et al. Deep venous thrombosis in 19. Kearon C, Akl EA, Ornelas J, et al. Antithrombotic therapy for
medical-surgical critically ill patients: prevalence, incidence, and VTE disease: CHEST guideline and expert panel report. Chest.
risk factors. Crit Care Med. 2005;33(7):1565-1571. 2016;149(2):315-352.
3. Patel R, Cook DJ, Meade MO, et al. Burden of illness in venous 20. Stein PD, Goodman LR, Hull RD, Dalen JE, Matta F. Diagnosis
thromboembolism in critical care: a multicenter observational and management of isolated subsegmental pulmonary embolism:
study. J Crit Care. 2005;20(4):341-347. review and assessment of the options. Clin Appl Thromb Hemost.
4. Muscedere JG, Heyland DK, Cook D. Venous thromboembolism 2012;18(1):20-26.
in critical illness in a community intensive care unit. J Crit Care. 21. Carrier M, Righini M, Le Gal G. Symptomatic subsegmental
2007;22(4):285-289. pulmonary embolism: what is the next step? J Thromb Haemost.
5. Lim W, Meade M, Lauzier F, et al. Failure of anticoagulant 2012;10(8):1486-1490.
thromboprophylaxis: risk factors in medical-surgical critically ill 22. Robert-Ebadi H, Righini M. Management of distal deep vein
patients. Crit Care Med. 2015;43(2):401-410.
thrombosis. Thromb Res. 2017;149:48-55.
6. Douds GL, Hellkamp AS, Olson DM, et al. Venous thromboem-
23. Righini M, Bounameaux H. Clinical relevance of distal deep vein
bolism in the get with the guidelines-stroke acute ischemic stroke
thrombosis. Curr Opin Pulm Med. 2008;14(5):408-413.
population: incidence and patterns of prophylaxis. J Stroke Cere-
24. Righini M, Paris S, Le Gal G, Laroche JP, Perrier A, Bounameaux
brovasc Dis. 2014;23(1):123-129.
H. Clinical relevance of distal deep vein thrombosis. Review of
7. Skaf E, Stein PD, Beemath A, Sanchez J, Bustamante MA, Olson
literature data. Thromb Haemost. 2006;95(1):56-64.
RE. Venous thromboembolism in patients with ischemic and
25. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method
hemorrhagic stroke. Am J Cardiol. 2005;96(12):1731-1733.
of classifying prognostic comorbidity in longitudinal studies:
8. Goldstein JN, Fazen LE, Wendell L, et al. Risk of thromboembo-
development and validation. J Chronic Dis. 1987;40(5):373-383.
lism following acute intracerebral hemorrhage. Neurocrit Care.
26. Kahn SR, Lim W, Dunn AS, et al. Prevention of VTE in non-
2009;10(1):28-34.
surgical patients: Antithrombotic Therapy and Prevention of
9. Kshettry VR, Rosenbaum BP, Seicean A, Kelly ML, Schiltz NK,
Thrombosis, 9th ed: American College of Chest Physicians
Weil RJ. Incidence and risk factors associated with in-hospital
Evidence-Based Clinical Practice Guidelines. Chest. 2012;
venous thromboembolism after aneurysmal subarachnoid hemor-
rhage. J Clin Neurosci. 2014;21(2):282-286. 141(suppl 2):e195S-e226S.
10. Ray WZ, Strom RG, Blackburn SL, Ashley WW, Sicard GA, Rich 27. Spyropoulos AC, Anderson FA, Jr., FitzGerald G, et al. Predic-
KM. Incidence of deep venous thrombosis after subarachnoid tive and associative models to identify hospitalized medical
hemorrhage. J Neurosurg. 2009;110(5):1010-1014. patients at risk for VTE. Chest. 2011;140(3):706-714.
11. Smith TR, Nanney AD III, Lall RR, et al. Development of venous 28. Bergstrom N, Braden BJ, Laguzza A, Holman V. The Braden
thromboembolism (VTE) in patients undergoing surgery for brain scale for predicting pressure sore risk. Nurs Res. 1987;36(4):
tumors: results from a single center over a 10-year period. J Clin 205-210.
Neurosci. 2015;22(3):519-525. 29. Falck-Ytter Y, Francis CW, Johanson NA, et al. Prevention of
12. Khaldi A, Helo N, Schneck MJ, Origitano TC. Venous throm- VTE in orthopedic surgery patients: Antithrombotic Therapy and
boembolism: deep venous thrombosis and pulmonary embolism Prevention of Thrombosis, 9th ed: American College of Chest
in a neurosurgical population. J Neurosurg. 2011;114(1):40-46. Physicians Evidence-Based Clinical Practice Guidelines. Chest.
13. Ekeh AP, Dominguez KM, Markert RJ, McCarthy MC. Incidence 2012;141(suppl 2):e278S-e325S.
and risk factors for deep venous thrombosis after moderate and 30. Samama MM, Cohen AT, Darmon JY, et al. A comparison of
severe brain injury. J Trauma. 2010;68(4):912-915. enoxaparin with placebo for the prevention of venous throm-
14. Skrifvars MB, Bailey M, Presneill J, et al. Venous thromboem- boembolism in acutely ill medical patients. Prophylaxis in med-
bolic events in critically ill traumatic brain injury patients. Inten- ical patients with enoxaparin study group. N Engl J Med. 1999;
sive Care Med. 2017;43(3):419-428. 341(11):793-800.
15. Perry JR. Thromboembolic disease in patients with high-grade 31. sanofi-aventis U.S. LLC. Lovenox®[package insert]. Bridge-
glioma. Neuro Oncol. 2012;14(suppl 4):73-80. water, NJ: sanofi-aventis U.S. LLC; 2017.
1234 Journal of Intensive Care Medicine 35(11)

32. Lassen MR, Raskob GE, Gallus A, Pineo G, Chen D, Portman RJ. 49. King DR, Cohn SM, Feinstein AJ, Proctor KG. Systemic coagu-
Apixaban or enoxaparin for thromboprophylaxis after knee lation changes caused by pulmonary artery catheters: laboratory
replacement. N Engl J Med. 2009;361(6):594-604. findings and clinical correlation. J Trauma. 2005;59(4):853-857;
33. Lassen MR, Raskob GE, Gallus A, Pineo G, Chen D, Hornick P. discussion 7-9.
Apixaban versus enoxaparin for thromboprophylaxis after knee 50. Ryan ML, Thorson CM, King DR, et al. Insertion of central
replacement (ADVANCE-2): a randomised double-blind trial. venous catheters induces a hypercoagulable state. J Trauma Acute
Lancet. 2010;375(9717):807-815. Care. 2012;73(2):385-390.
34. Lassen MR, Gallus A, Raskob GE, Pineo G, Chen D, Ramirez 51. Anderson FA Jr, Spencer FA. Risk factors for venous thromboem-
LM. Apixaban versus enoxaparin for thromboprophylaxis after bolism. Circulation. 2003;107(23 suppl 1):I9-16.
hip replacement. N Engl J Med. 2010;363(26):2487-2498. 52. Pottier P, Hardouin JB, Lejeune S, Jolliet P, Gillet B, Planchon B.
35. ELIQUIS® [Package Insert]. Bristol-Myers Squibb Company. Immobilization and the risk of venous thromboembolism. A
Princeton, NJ, and New York, NY: Pfizer Inc; 2016. meta-analysis on epidemiological studies. Thromb Res. 2009;
36. Sauro KM, Soo A, Kramer A, et al. Venous thromboembolism 124(4):468-476.
prophylaxis in neurocritical care patients: are current practices, 53. Hull RD. Relevance of immobility and importance of risk assess-
best practices? Neurocrit Care. 2019;30(2):355-363. doi:10.1007/ ment management for medically ill patients. Clin Appl Thromb
s12028-018-0614-9. Hemost. 2013;19(3):268-276.
37. Kim KS, Brophy GM. Symptomatic venous thromboembolism: 54. Hashem MD, Parker AM, Needham DM. Early mobilization and
incidence and risk factors in patients with spontaneous or trau- rehabilitation of patients who are critically ill. Chest. 2016;
matic intracranial hemorrhage. Neurocrit Care. 2009;11(1): 150(3):722-731.
28-33. 55. Engel HJ, Needham DM, Morris PE, Gropper MA. ICU early
38. Chalouhi N, Daou B, Rincon F, et al. Risk of venous thromboem- mobilization: from recommendation to implementation at three
bolism in patients with large hemispheric infarction undergoing medical centers. Crit Care Med. 2013;41(9 suppl 1):S69-S80.
decompressive hemicraniectomy. Neurocrit Care. 2016;25(1): 56. Hodgson CL, Stiller K, Needham DM, et al. Expert consensus and
105-109. recommendations on safety criteria for active mobilization of
39. Fontaine GV, Vigil E, Wohlt PD, et al. Venous thromboembolism mechanically ventilated critically ill adults. Crit Care. 2014;
in critically ill medical patients receiving chemoprophylaxis: a 18(6):658.
focus on obesity and other risk factors. Clin Appl Thromb Hemost. 57. Blokhin IO, Lentz SR. Mechanisms of thrombosis in obesity.
2016;22(3):265-273. Curr Opin Hematol. 2013;20(5):437-444.
40. Wang TF, Wong CA, Milligan PE, Thoelke MS, Woeltje KF, 58. Lee YR, Blanco DD. Efficacy of standard dose unfractionated
Gage BF. Risk factors for inpatient venous thromboembolism heparin for venous thromboembolism prophylaxis in morbidly
despite thromboprophylaxis. Thromb Res. 2014;133(1): obese and non-morbidly obese critically Ill patients. J Thromb
25-29. Thrombolysis. 2017;44(3):386-391.
41. Parienti JJ, Mongardon N, Megarbane B, et al. Intravascular com- 59. Rondina MT, Wheeler M, Rodgers GM, Draper L, Pendleton RC.
plications of central venous catheterization by insertion site. Weight-based dosing of enoxaparin for VTE prophylaxis in mor-
N Engl J Med. 2015;373(13):1220-1229. bidly obese, medically-Ill patients. Thromb Res. 2010;125(3):
42. Kucher N.Clinical practice. Deep-vein thrombosis of the upper 220-223.
extremities. N Engl J Med. 2011;364(9):861-869. 60. Ikesaka R, Delluc A, Le Gal G, Carrier M. Efficacy and safety of
43. Kraaijpoel N, van Es N, Porreca E, Buller HR, Di Nisio M. The weight-adjusted heparin prophylaxis for the prevention of acute
diagnostic management of upper extremity deep vein thrombosis: venous thromboembolism among obese patients undergoing bar-
a review of the literature. Thromb Res. 2017;156:54-59. iatric surgery: a systematic review and meta-analysis. Thromb
44. Chopra V, Anand S, Hickner A, et al. Risk of venous thromboem- Res. 2014;133(4):682-687.
bolism associated with peripherally inserted central catheters: a 61. Wang TF, Milligan PE, Wong CA, Deal EN, Thoelke MS, Gage
systematic review and meta-analysis. Lancet. 2013;382(9889): BF. Efficacy and safety of high-dose thromboprophylaxis in mor-
311-325. bidly obese inpatients. Thromb Haemost. 2014;111(1):88-93.
45. Fletcher JJ, Wilson TJ, Rajajee V, et al. A randomized trial of 62. Miranda S, Le Cam-Duchez V, Benichou J, et al. Adjusted value
central venous catheter type and thrombosis in critically ill neu- of thromboprophylaxis in hospitalized obese patients: a compara-
rologic patients. Neurocrit Care. 2016;25(1):20-28. tive study of two regimens of enoxaparin: the ITOHENOX study.
46. Saber W, Moua T, Williams EC, et al. Risk factors for catheter- Thromb Res. 2017;155:1-5.
related thrombosis (CRT) in cancer patients: a patient-level data 63. Malato A, Dentali F, Siragusa S, et al. The impact of deep vein
(IPD) meta-analysis of clinical trials and prospective studies. thrombosis in critically ill patients: a meta-analysis of major clin-
J Thromb Haemost. 2011;9(2):312-319. ical outcomes. Blood Transfus. 2015;13(4):559-568.
47. Geerts W. Central venous catheter-related thrombosis. Hematol- 64. Velmahos GC, Nigro J, Tatevossian R, et al. Inability of an
ogy Am Soc Hematol Educ Program. 2014;2014(1):306-311. aggressive policy of thromboprophylaxis to prevent deep venous
48. Rajasekhar A, Streiff MB. How I treat central venous access thrombosis (DVT) in critically injured patients: are current meth-
device-related upper extremity deep vein thrombosis. Blood. ods of DVT prophylaxis insufficient? J Am Coll Surg. 1998;
2017;129(20):2727-2736. 187(5):529-533.