Keywords point-of-care ultrasonography - venous dopplers - critical care - deep venous thrombosis
Introduction
Venous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and pulmonary
embolism (PE), is an important cause of morbidity and mortality in the intensive care
units (ICUs).[1 ]
[2 ] Studies have shown that the incidence of DVTs in the ICU is somewhere between 13%
to 21% in patients not receiving chemical prophylaxis[3 ]
[4 ]
[5 ]
[6 ] and 5% to 23% in patients receiving chemical prophylaxis.[1 ]
[7 ]
[8 ] In neurocritical care units (NCCUs), no reliable population-based estimates are
available; however, the incidence is perceived to be higher. For instance, the incidence
of DVTs in traumatic brain injury patients has been reported to be as high as 40%.[9 ] DVTs are often asymptomatic in ICU patients.[10 ]
[11 ]
[12 ]
[13 ]
[14 ] Screening of these asymptomatic patients can lead to a higher rate of DVT detection.[8 ]
[15 ]
[16 ] This may help in preventing life-threatening complications such as PE and in turn
reduce morbidity and mortality, although the evidence supporting this hypothesis is
inconsistent.[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
A typical lower extremity venous Doppler (LEVD) scan combines compression study, color
Doppler, and pulse wave Doppler along the entire length of the venous system. Its
routine use can be resource intensive and requires experienced ultrasound technicians.
In contrast, a two-point compression ultrasound (2-CUS) examination focuses on the
highest probability sites for DVTs instead of scanning the whole leg.[22 ] The shortened examination can be performed with minimal training and has been shown
to be a fast and reliable alternative for the diagnosis of DVTs in symptomatic ICU
patients with a diagnostic accuracy of 95%.[23 ] The examination relies solely on compression at two points and cuts down the total
scan time by skipping the Doppler exam and evaluation for calf DVTs which are of uncertain
clinical significance. The 2-CUS examination can be valuable in critically ill patients
with urgent need for evaluation of DVTs, which would otherwise be dependent on availability
of an ultrasound technician and an interpreting radiologist.[23 ] In NCCUs, the technique can be used for frequent screening of high-risk patients
such as those with paraplegia or patients unable to receive DVT prophylaxis due to
a contraindication. However, the feasibility and accuracy of the 2-CUS have not been
well established for screening of asymptomatic patients, particularly in the NCCUs.
The aim of this study is to evaluate the use of 2-CUS as surveillance for patients
in this patient population.
Methods
Study Design and Patient Population
We performed a retrospective analysis of prospectively collected data for quality
improvement purposes of consecutive patients admitted to the NCCU at Baylor College
of Medicine between August 2020 and February 2021. All patients admitted to the NCCU
older than 18 years were screened for inclusion. The exclusion criteria included an
expected length of stay (LOS) of less than 48 hours, based on the perceived lower
risk of DVTs with shorter hospital stays. Patients with a positive COVID-19 test were
excluded to minimize unnecessary exposure. Additionally, individuals with a preexisting
DVT diagnosis were excluded to reduce potential bias associated with a known DVT diagnosis.
Ultrasound Procedure
Clinical fellows of the Neurocritical Care Fellowship program (n = 4) at Baylor College of Medicine with no prior experience in lower extremity venous
ultrasonography received training to perform 2-CUS. The training included a video
lecture, which included both normal and abnormal cases, and a single practical demonstration
in human subjects following which the fellows were allowed to do the 2-CUS screening
examinations per protocol. There was no prespecified requirement for the fellows to
perform a minimum number of practice scans before screening the patients. The ultrasound
was performed in the supine position with the patients' hips slightly abducted and
externally rotated and the knees flexed as needed to approach the popliteal fossa.
We used Philips Sparq (Philips Ultrasound, Bothell, WA, United States) linear array
transducer (L12–4) or Fujifilm SonoSite Edge II Total (Fujifilm Ultrasound, Bothell,
WA, United States) linear array transducer (L10–5) to visualize lower extremity veins
at two sites. The common femoral vein (CFV) was insonated in the inguinal area below
the inguinal crease and the popliteal vein (PoV) was imaged in the popliteal fossa
along their transverse axis. The examination was limited to these two points and was
not aided by Doppler augmentation. Color Doppler was used at the operator's discretion
to aid in the identification of the vein but was not required for thrombus evaluation.
The criteria for positive test for DVT on 2-CUS included the visualization of an echogenic
focus inside the vein (consistent with a thrombus) and inability to compress the vein
completely (so the superior and inferior walls of the vein touch and the vein is completely
collapsed; see [Fig. 1 ]). The scans were interpreted by the fellow performing the ultrasound. The screening
2-CUS examination was done on admission, every third day and when there was a clinical
suspicion of DVT until detection of a DVT, death, or discharge of the patient from
the ICU.
Fig. 1 Example of a normal operator-generated 2-CUS scan. The common femoral vein (CFV)
is visualized on the top images and the popliteal vein (PV) on the bottom. The veins
are visualized precompression (left) and postcompression (right). Notice the collapsing
of veins after applying compression.
An LEVD scan was requested by the treating team in patients based on a positive 2-CUS,
clinical suspicion of DVT, or on all patients on every seventh day of their NCCU admission.
This scan was done using Philips Epic (Philips Ultrasound) linear array transducer
(L9–3) or Philips iU22 (Philips Ultrasound) linear array transducer (L9–3) by a certified
vascular ultrasound technician and interpreted by a radiologist. It involved a more
extensive evaluation of both proximal and distal lower extremity venous system including
the CFV, femoral vein (FV), deep femoral vein (DFV), PoV, anterior tibial vein (ATV),
posterior tibial vein (PTV), and peroneal veins (PrV). The veins were visualized in
both transverse and longitudinal axes. The examination included compressibility, evaluation
for echogenic material within the vein, color flow, pulse wave Dopplers, and flow
augmentation.
Statistical Analysis
Statistical analysis was performed using SPSS 21 (Armonk, NY, United States). Descriptive
statistics were performed for baseline characteristics. Categorical variables were
expressed as percentages, while continuous variables were expressed as means and standard
deviations. We calculated the overall sensitivity, specificity, positive predictive
value, negative predictive value, and diagnostic accuracy for lower extremity DVT
detection using protocolized 2-CUS scans. Only patients who completed both 2-CUS scans
and LEVD were included in these analyses. Wilson score intervals were used to calculate
95% confidence intervals (CIs). Since the calculations involved diagnostic metrics
such as sensitivity and specificity based on categorical variables, that is, positive
or negative 2-CUS or LEVD scans, patients who were missing either of these scans were
excluded without any data imputation ([Fig. 2 ]). There were no missing data for baseline characteristics reported in the manuscript.
Fig. 2 Flowchart summarizing patient selection.
Results
A total of 298 patients were admitted to the NCCU. Forty-two patients were screened
out as 2-CUS was not performed either due to anticipated LOS of less than 48 hours
(n = 2) or due to patients' refusal or agitation (n = 2). Of the remaining 256 patients, 192 patients were excluded as no LEVD was done
due to LOS of between 2 and 6 days. Sixty-four patients underwent both 2-CUS and an
LEVD ([Fig. 2 ]). The baseline characteristics of the studied cohort are detailed in [Table 1 ]. Briefly, the study population had a mean age of 62.34 years and comprised 40.9%
females and 59.1% males. The majority of the patients identified as White (55.2%),
followed by African American (26.3%) and Hispanic (13.9%). Most common admission diagnosis
was acute ischemic stroke (35%), followed by intracerebral hemorrhage (15.6%), subdural
hemorrhage (11%), status epilepticus/seizures (9.8%), and subarachnoid hemorrhage
(9.8%).
Table 1
Patient characteristics
Age (y), mean (SD)
62.34 (17.67)
Female, n (%)
Male, n (%)
122 (40.9%)
176 (59.1)
White, n (%)
African American, n (%)
Hispanic, n (%)
Asian, n (%)
155 (55.2)
74 (26.3)
39 (13.9)
13 (4.6)
ICU LOS in days, mean (SD)
5.31 (5.44)
Obesity, n (%)
103 (34.6)
History of smoking, n (%)
49 (16.4)
History of malignancy, n (%)
47 (15.7)
History of DVT, n (%)
5 (1.7)
Limb paresis, n (%)
70 (23.5)
Recent surgery, n (%)
116(38.8)
Admission diagnosis,
n
(%)
AIS
SAH
SDH
ICH
SCI
SE/seizures
TBI
Postneurosurgery
Other[a ]
87 (35)
25 (9.8)
28 (11)
40 (15.6)
11 (4.3)
25 (9.8)
8 (3.1)
29 (11.3)
32 (12.5)
Abbreviations: AIS, acute ischemic stroke; DVT, deep vein thrombosis; ICH, intracranial
hemorrhage; ICU, intensive care unit; LOS, length of stay; SAH, subarachnoid hemorrhage;
SCI, spinal cord injury; SD, standard deviation; SDH, subdural hematoma; SE, status
epilepticus; TBI, traumatic brain injury.
a Other includes encephalopathy and neuromuscular, infectious/inflammatory, and neoplastic
processes.
The incidence of DVT in patients who underwent LEVD was 8 (10.6%). 2-CUS was done
in 256 patients; 9 patients were positive, of whom 6 were confirmed by LEVD and 3
were false positives. Two patients had false-negative 2-CUS. When compared with LEVD,
the sensitivity of 2-CUS was 75% (95% CI: 0.41–0.93) and specificity was 93.1% (95%
CI: 0.84–0.97). The positive predictive value was 66.7% (95% CI: 0.35–0.88) and the
negative predictive value was 96.4% (95% CI: 0.88–0.99; see [Table 2 ]). Among the nine patients with DVT, 5 had a diagnosis of acute ischemic stroke,
2 patients had subarachnoid hemorrhage, 1 was status post neurosurgery, and 1 had
spinal cord pathology. The diagnostic accuracy increased from 82.6% during the first
month to 97.6% after the first month of the study with an overall diagnostic accuracy
of 92.6% (95% CI: 0.83–0.97), suggesting a learning curve. The sensitivity for proximal
above-knee DVTs, that is, CFV and PoV, was 100%. The two DVTs missed on 2-CUS were
located in the PTVs, which were not scanned with 2-CUS.
Table 2
Screening test calculations
Doppler positive
Doppler negative
Total
Sensitivity = 75% (95% CI: 0.41–0.93)
Specificity = 94.64% (95% CI: 0.85–0.98)
PPV = 66.7% (95% CI: 0.35–0.87)
NPV = 96.36% (95% CI: 0.88–0.99)
Accuracy = 92.2% (95% CI: 0.83–0.97)
2-CUS positive
TP: 6
FP: 3
9
2-CUS negative
FN: 2
TN: 53
59
Total
8
56
64
Abbreviations: CI, confidence interval; FN, false positive; NPV, negative predictive
value; PPV, positive predictive value; TP, true positive.
Discussion
To our knowledge, this is the first study evaluating the use of serial 2-CUS as a
screening tool for DVTs in NCCUs. The sensitivity for all DVTs in this study was 75%
as it missed two distal PTV DVTs. Our study was able to capture 100% of all proximal
DVTs, which are typically associated with higher risk of life-threatening VTE or PE.[24 ]
[25 ]
[26 ] The clinical significance of distal DVTs is unclear as data suggest that they may
not extend proximally, remain uneventful, and might not benefit from anticoagulation.[24 ]
[25 ]
[26 ] Specificity for all DVTs reported in our study was 95%. The three false positives
occurred in the first month of the study, indicating a learning curve and a better
diagnostic accuracy with time and experience with this diagnostic tool. It should
also be noted that 2-CUS and LEVDs were not performed simultaneously and the differences
in the result might be secondary to a new clot formation, clot propagation, or clot
migration.[23 ]
The diagnostic accuracy measures reported in our study are similar to the three previously
reported ICU studies that compared the performance of point-of-care ultrasound with
LEVD for the detection of DVTs. Kory et al performed a three-point compression ultrasound
(CFV, SFV, and PV) in medical, surgical, and cardiothoracic ICUs and reported a sensitivity
of 86% and specificity of 96%.[23 ] Caronia et al reported a lower sensitivity of 63% and comparable specificity of
97% for 2-CUS ultrasound scans (CFV and PV) performed by a medical resident. Although
residents showed 97% agreement with LEVDs, a higher incidence of FV DVTs that were
not scanned by 2-CUS resulted in a much lower sensitivity.[27 ] It should be noted that in our study, no FV DVTs were identified, which resulted
in a higher sensitivity for proximal DVTs. Recently, a prospective observational study
reported sensitivity of 69% (95% CI: 41.3–89) and specificity of 99% (95% CI: 96.7–100)
with an intensivist-performed three-point CUS (CFV, FV, and PV) in trauma patients.
The false-negative scans were reported for nonocclusive DVTs and a single false-positive
scan was due to misinterpretation of the femoral artery as a vein.[28 ] The first two studies were done in patients with a clinical suspicion of DVT[23 ]
[27 ] and the last study utilized routine twice-weekly surveillance for high-risk trauma
patients.[28 ] Please see [Table 3 ] for a comparison of these studies with our study.
Table 3
Comparison of the current study with previously published studies reporting the use
of point-of-care ultrasound as a screening tool for lower extremity DVTs in critical
care setting
Study
Imaging protocol
Reason for examination
Incidence of lower extremity DVTs
Sensitivity
Specificity
Accuracy
Kory et al[23 ] (N = 128)
3 compressions along the CFV, 2 compressions along the PoV, and sequential compression
along the FV in 2-cm increments if the CFV and PoV sites were fully compressible
Clinical suspicion. No routine surveillance
20%
86%
96%
95%
Caronia et al[27 ] (N = 75)
Compression at the CFV and popliteal veins
Clinical suspicion. No routine surveillance
16%
63%
97%
n/a
Roberts et al[28 ] (N = 117)
3 compressions along the CFV, 2 compressions along the PoV, and sequential compression
along the FV in 2-cm increments if the CFV and PoV sites were fully compressible
Twice weekly screening in prespecified high-risk patient population
22.2
69% (95% CI: 41.3–89)
99% (95% CI: 96.7–100)
n/a
Current study (N = 64)
Compression at the CFV and popliteal veins
Clinical suspicion and routine surveillance twice weekly
10.1%
75% (95% CI: 41–93)
94.6% (95% CI: 85–98)
92.2% (95 CI: 83–97
Abbreviations: CFV, common femoral vein; CI, confidence interval; FV, femoral vein;
n/a, not applicable; PoV, popliteal vein.
The incidence of 10.6% for DVTs in our study is comparable to the previously reported
incidence by Dickerson et al in NCCU patients when weekly LEVDs were used for surveillance.[21 ] However, other studies report a significant variation in the incidence of DVTs ranging
from 2.5% in neurosurgical patients[29 ] to 40% in traumatic brain injury patients.[9 ] Both studies used LEVDs twice weekly as a surveillance method.[9 ]
[29 ] There is also a notable variation in the incidence of DVTs in non-neurological critical
care patients, ranging from 2.8 to 23% with different surveillance screening protocols.[1 ]
[3 ]
[4 ]
[5 ]
[6 ]
[7 ]
[8 ]
[20 ]
[28 ] This variation can be explained by the heterogenous patient population, smaller
cohort sizes, differences in the surveillance protocols, and frequency of screening
examinations in these studies.[28 ]
[29 ]
DVT surveillance is not the standard of practice and more studies are needed to evaluate
its benefits and appropriate protocols. Studies looking at routine surveillance of
DVTs have shown increased rates of detection for DVTs, but have failed to consistently
show a decrease in life-threatening PE, LOS, morbidity, or mortality.[9 ]
[18 ]
[19 ]
[20 ]
[21 ]
[29 ]
[30 ]
[31 ] Surveillance studies have been done with LEVD in a twice-weekly or every 3 days
frequency.[9 ]
[19 ]
[29 ]
[30 ] Recently, Arabi et al in a preplanned substudy of the Pneumatic Compression for
Preventing Venous Thrombo-embolism Trial (PREVENT) compared a surveilled group with
LEVD (n = 1,682) with standard of care showing an increase in DVT detection (n = 383), an earlier diagnosis (4 vs. 20 days, respectively), and a lower 90-day mortality.[19 ] The overall frequency of PE detection was similar in both groups; however, surveillance
was associated with a quicker diagnosis of PE by 4 days.[19 ] Another similar study in an ICU population (n = 128) showed that surveillance was associated with increased DVT diagnosis (2.8
vs. 1.3% in nonsurveilled patients) and a decrease in the incidence of PE (0.7 vs.
1.5% in nonsurveilled patients).[20 ] Surveillance was found to be an independent predictor of higher DVT and lower PE
incidence. In the mentioned study, the associated cost increment was $509,091 per
life saved, suggesting the use of 2-CUS instead of LEVD could lower the associated
cost significantly.[20 ] Among neurosurgical patients, Dickerson et al[21 ] (n = 485) and Samuel et al[18 ] (n = 147) failed to demonstrate a reduction in PE and mortality rate when routine surveillance
was performed through LEVDs. Larger multicentric and randomized controlled studies
are warranted to investigate if 2-CUS surveillance scans can lead to earlier detection
of DVTs, prevent clinically significant VTEs and PEs, and affect the overall morbidity
and mortality associated with these events.
The notable strength of this study is evaluation of 2-CUS as a screening tool in neurocritical
care patient population, which has previously been not well studied. The study emphasizes
that 2-CUS scans are feasible in NCCU with relatively quick training of previously
unexperienced practitioners and can be used for faster screening and surveillance.
The technique can be quickly learnt, is faster, and is cost-effective. However, the
technique does not necessarily replace the need for LEVD when a strong clinical suspicion
is present as it carries a risk of detecting false positives and can have false negatives.
A positive 2-CUS should be followed up with LEVD to avoid overtreatment. Although
this study detected proximal DVTs with 100% sensitivity, the technique still has a
potential to miss out on proximal DVTs that are typically not insonated with the 2-CUS
such as FV and DFVs.[27 ] In addition, it can miss out on distal DVTs as seen in our study. Thus, a strong
clinical suspicion should prompt LEVD despite a negative 2-CUS study.
The study has several limitations. The single-center design of the study with a small
sample size consisting of a specific patient population limits the generalizability
of this study. Certain patient populations such as patients with traumatic brain injury
are under-represented in our study. The study was done in an academic center where
a large treatment team comprising attending physicians, neurocritical care fellows,
and residents were responsible for taking care of the patients. The routine screening
examinations may not be feasible for a smaller team taking care of a high volume of
patients. The study does not explore the metrics for the individual operators and
the impact of inter-rater variability. The small sample size can lead to potentially
imprecise estimates of sensitivity and specificity. However, the overall diagnostic
accuracy of this study is similar to the previously reported literature.[19 ]
[20 ]
[21 ] The neurocritical care fellows, ultrasound technicians, and interpreting radiologists
were not strictly blinded to other's sonographic findings. However, the 2-CUS scans
were generally performed before the LEVDs and the results of 2-CUS scans were not
communicated to the ultrasound technicians or radiologists. Another limitation is
that the 2-CUS scans and LEVDs were not performed simultaneously. Per protocol, a
positive 2-CUS examination led to an immediate request for LEVD, which was performed
within a day. However, for negative screening examinations, the time interval between
2-CUS and LEVD ranged from 0 to 3 days. This introduces a potential bias as clot formation
between the two examinations could have potentially impacted the false-negative rate.
However, in our study, the two observed false-negative scans were for distal PTVs
not scanned during 2-CUS examination. Finally, the study did not address if the routine
surveillance with 2-CUS leads to a more rapid detection of DVTs and how it can affect
the overall morbidity and mortality of the patients.
Conclusion
2-CUS is a quick bedside tool to screen for DVTs and can be quickly learnt by providers
with no formal training in ultrasonography. Our study supports that 2-CUS is feasible,
quick, and cost-effective tool for screening clinically significant lower extremity
proximal DVTs in NCCU patients. Future studies are needed to validate these findings
in larger cohorts and to assess if such screening protocols can effectively increase
the rate of DVT detection and lower the life-threatening thromboembolic complications
such as PE and overall mortality in this patient population.
