Venous thromboembolic disease is a frequent complication
following blunt trauma1-6. Patients
with pelvic or acetabular fracture are a subpopulation of patients
with multiple trauma who have been identified as being at high risk
for the development of deep-vein thrombosis4.
Pulmonary embolism is the most common cause of deaths occurring
more than seven days after traumatic injury7.
Because of the risk of internal bleeding, especially in patients
with a head injury, several authors have expressed concerns regarding
aggressive anticoagulation as primary prophylaxis for patients who
have sustained traumatic injuries7-10. Some authors have utilized
inferior vena cava filters as either prophylaxis against or treatment
of venous thromboembolic disease2,7,
but others have expressed major concerns regarding the long-term
consequences of the use of such devices11.
Mechanical compression is an alternative to pharmacological prophylaxis
that avoids the risk of bleeding complications.
Mechanical compression to prevent deep-vein thrombosis is currently
available in two forms: thigh-calf low-pressure sequential-compression devices,
and high-pressure pulsatile-compression devices that are
available with either a calf-foot or a foot-only
wrap. Low-pressure sequential-compression devices inflate the thigh
and calf chambers sequentially to a pressure of approximately 45
mm Hg. Combination pulsatile-compression devices utilize calf and
foot chambers and inflate to a pressure of approximately 160 mm
Hg. These pumps differ from the standard foot pumps because they have
two bladders, one of which inflates over the calf and one of which
inflates over the foot. The pulsatile pumps inflate for only two
to three seconds in each twenty-second cycle. There have
been several recent articles describing the use of these two forms
of mechanical compression to prevent deep-vein thrombosis in patients
who have sustained traumatic injuries3,9,10,12.
The hypothesis of our study was that pulsatile compression is
more effective than sequential compression in preventing deep-vein
thrombosis in patients with pelvic or acetabular fracture requiring
internal fixation.
This study was designed as a prospective, randomized, and blinded
evaluation of sequential and pulsatile mechanical compression devices.
All patients admitted to the University of Alabama Hospital with
a pelvic or acetabular fracture due to blunt trauma between December
1, 1997, and May 1, 1999, were evaluated for possible inclusion
in the study. Inclusion criteria included blunt trauma causing a
pelvic or acetabular fracture with a pattern requiring surgical
fixation, an age of at least sixteen years, and an ability and willingness
to comply with both the mechanical prophylaxis protocol and the
screening studies for deep-vein thrombosis. Exclusion criteria included
a history of venous thromboembolic disease, initiation of mechanical
compression more than seventy-two hours following the injury, a
body habitus or weight that made it difficult for the patient to
fit in the magnetic resonance imaging scanner, or a stable injury
that did not require surgical treatment. The institutional review
board of the University of Alabama Hospital approved this study.
Patients who agreed to participate in this study were randomized
into one of two treatment groups for prophylaxis against deep-vein
thrombosis, and informed consent was obtained. Randomization was
accomplished by using a computer-generated randomization table.
Patients randomized into Group A were treated bilaterally with a
thigh-calf sequential-compression device (Kendall SCD; Kendall,
Mansfield, Massachusetts). Patients randomized into Group B were
treated bilaterally with a combination sequential pump that covers
the calf and foot (PlexiPulse; NuTech, Kinetic Concepts, San Antonio,
Texas). Patients were treated with the mechanical prophylaxis as
soon as possible following admission to the trauma service. The
patients and the nursing staff were encouraged to utilize the pumps
for the maximum number of hours possible per day, with removal allowed
only for nursing care and physical therapy. The nursing staff was instructed
to ensure proper application of wraps and to document the number
of hours of wear on a data sheet. No patient who remained in the
study received any form of pharmacological prophylaxis, and no patient
withdrew from the study voluntarily because of an inability to tolerate
the pumps.
Two different screening studies were performed to evaluate patients
for the presence of deep-vein thrombosis. The studies were done
within twenty-four hours prior to the patient’s discharge,
or earlier if the patient demonstrated any signs or symptoms consistent
with venous thromboembolic disease. All screening studies were performed
bilaterally. Duplex ultrasound examinations were performed with
an Acuson 128 XP or Sequoia System (Mountain View, California) or
an ATL 3000HDI machine (Advanced Technology Laboratories, Bothell, Washington)
and high-resolution linear transducers with use of color and spectral
Doppler vessel interrogation. Standard compression and flow augmentation
techniques were utilized in the lower extremities from the groin
to the popliteal fossa. Ultrasound criteria for deep-vein thrombosis included
visualization of an intraluminal thrombus, loss of vessel compressibility,
and decreased blood flow. Magnetic resonance venograms of the pelvis and
lower extremities were made with a Signa 1.5-tesla magnet (General
Electric Medical Systems, Milwaukee, Wisconsin) and use of two-dimensional
time-of-flight sequences (SPGR [spoiled gradient-recalled
acquisition in the steady state]; repetition time, 47 msec;
echo time, 10.4 msec; 2-mm-thick axial images without gap; 256 ¥ 128 matrix;
and one excitation) with saturation of arterial signal. Two musculoskeletal
radiologists reviewed the coronal reformatted multiple-intensity projections
and source axial images. Discrepancies between the readings were
resolved by the radiologists reviewing the study in question together
and coming to a consensus.
Magnetic resonance venogram (MRV) criteria for deep-vein thrombosis
included intraluminal signal-flow voids with abrupt narrowing or
termination of luminal signal that could not be accounted for by flow
or magnetic susceptibility artifacts from internal fixation devices.
All screening studies were interpreted with the radiologists blinded
regarding the type of prophylaxis against deep-vein thrombosis and
the result of the other study. When the two screening tests had
conflicting results, the magnetic resonance venography was judged
to be valid if the deep-vein thrombosis was located in the pelvis.
If not, the staff radiologists settled the discrepancy by reviewing
both studies and ordering a repeat ultrasound or venogram if necessary.
The end point of the study was when the two imaging studies had been
performed to evaluate for deep-vein thrombosis, although patients
were followed clinically for the development of deep-vein thrombosis
and pulmonary embolism on an outpatient basis. A documented pulmonary
embolism was accepted as evidence of deep-vein thrombosis without
a corroborating magnetic resonance venogram because of the difficulty
involved in the performance of magnetic resonance venography on
critically ill patients.
Data collected included the presence or absence of deep-vein
thrombosis and pulmonary embolism, the type and location of deep-vein
thrombosis, the time of commencement of mechanical compression,
the time between the injury and the surgical stabilization, the
number of hours that the pump was worn, and the time between the
injury and the screening studies. Because some authors had suggested
that magnetic resonance venography may detect deep-vein thromboses
that are small and clinically irrelevant13,
we documented whether a clot was occlusive or nonocclusive and,
if it was nonocclusive, whether it was greater or less than 2 cm
in size. Additional data included the use of femoral intravenous
lines, the AO classification14 of
the fracture, the surgical approach, the injury severity score15, and the presence of additional
skeletal injuries.
A number of different statistical tests were used to analyze
the data in this study. The Fisher exact and chi-square tests were
employed to determine the significance of the difference between
the pulsatile and sequential forms of mechanical prophylaxis. Analysis
of variance was used to determine the significance of the difference,
with regard to age, injury severity, and time to surgery, between patients
with and those without deep-vein thrombosis.
All patients admitted to this study had sustained a pelvic or
acetabular fracture due to blunt trauma and required surgical fixation.
The average injury severity score was 19.8 points (range, 9 to 59 points)
in Group A and 16.1 points (range, 9 to 50 points) in Group B. The
screening tests were done at an average of 6.0 days following the
surgery and 10.8 days following the injury in Group A and at an average
of 6.5 days following the surgery and 10.8 days following the injury
in Group B. There was no significant difference between the two
groups with respect to the injury severity score or the time to
the screening studies (p > 0.05). There was also no significant
difference with regard to gender, weight, AO classification of the
fracture, associated long-bone fractures, number of patients with
femoral vein cannulation, or surgical approach. The patients in
Group A utilized the pumps for an average of 20.8 hours (range,
four to twenty-four hours) per day compared with 21.3 hours (range,
seven to twenty-four hours) per day in Group B. Because our data
regarding patient compliance were incomplete, no statistical analyses
correlating usage of the pump with the development of clots were
performed.
Thirty-three patients who were initially enrolled in the
study did not complete it. The reasons for withdrawal from the study
included claustrophobia (six patients); death (six) (no patient
died because of a thromboembolic event); refusal to undergo magnetic
resonance venography (five); inadvertent initiation of anticoagulation
by another service (five); discharge before the appropriate studies
had been performed (three); inability of the patient to fit in the
magnetic resonance scanner (two); inability of the patient to remain
immobile during the magnetic resonance imaging secondary to a closed
head injury (two); prior venous thromboembolic disease missed on
the initial screening (two); inadvertent switching of the pump types
(one); and pregnancy (one). The demographics of these patients were
not significantly different from those of the patients who successfully
completed the study. The patients who withdrew had a total of three
deep-vein thromboses, which were not included in the statistical analysis.
One hundred and seven patients completed the protocol and had
satisfactory screening studies. Deep-vein thrombosis developed in
ten (19%) of the fifty-four patients who completed the
protocol in Group A; seven (13%) had a large or occlusive
clot and one had a pulmonary embolus. Deep-vein thrombosis developed
in five (9%) of the fifty-three patients who completed
the study in Group B; two (4%) had a large or occlusive
clot and none had a pulmonary embolus. There were no deaths due
to venous thromboembolic disease during the study. The overall incidence
of deep-vein thrombosis in the study was fifteen (14%)
of 107. The difference in the rate of deep-vein thrombosis between
Group A and Group B was not significant (p = 0.265), although
there was a trend toward more large or occlusive clots in Group
A than in Group B (p = 0.16) .
The primary anatomic locations of the deep-vein thromboses in
the fourteen patients in whom the locations were identified included
the pelvis (six patients), the thigh (seven), and the calf (one).
Four patients had thrombosis in more than one vein. Three thromboses
involved the common iliac vein; five, the external iliac vein; one,
the internal iliac vein; nine, the common femoral vein; and one,
the popliteal vein.
Analysis of the pooled data from both groups demonstrated a significant
difference, in terms of age and time to surgery, between the patients
with occlusive deep-vein thrombosis and those with no deep-vein
thrombosis or with nonocclusive deep-vein thrombosis. The average
age was forty-six years for the patients with occlusive deep-vein thrombosis
compared with thirty-five years for those with no deep-vein thrombosis
and twenty-six years for those with nonocclusive deep-vein thrombosis
(p = 0.03, analysis of variance). The average time between
the injury and the surgery was ten days for the patients with occlusive
deep-vein thrombosis compared with five days for those with no deep-vein
thrombosis and four days for those with nonocclusive deep-vein thrombosis
(p = 0.0004, analysis of variance). Patients with an occlusive
clot had an average injury severity score of 23.0 points compared
with 13.7 and 17.7 points for the patients with nonocclusive and
no deep-vein thrombosis, respectively. The difference between the
groups represents a trend but was not significant with our sample
size (p = 0.19).
Multivariate analysis of variance demonstrated a trend toward
the development of deep-vein thrombosis with the Kocher-Langenbeck
and extensile or combined approaches. All nine occlusive deep-vein thromboses,
and thirteen of the fifteen deep-vein thromboses detected overall,
were in patients treated with one of these approaches. One nonocclusive
deep-vein thrombosis was associated with an ilioinguinal approach,
and one was associated with a Pfannenstiel approach.
In one patient in Group A, a late deep-vein thrombosis developed
following negative screening studies. Leg pain and increased swelling
developed two months following surgery, at which time the patient had
a positive ultrasound examination documenting the deep-vein thrombosis.
Screening studies were not routinely performed on patients following
discharge from the hospital.
Deep-vein thrombosis has been recognized as a major cause of
morbidity and mortality following major blunt trauma3-6,9,10,12,16-20. Geerts et al. documented
a 61% prevalence of deep-vein thrombosis following pelvic
fractures in patients who had received no prophylaxis against it4. While it is not clear that all of
their patients had fractures requiring surgical stabilization, their series
was clearly composed of patients with severe traumatic injuries.
Using contrast venography, Geerts et al. also documented an 80% prevalence
of deep-vein thrombosis associated with femoral fractures. In patients
receiving prophylaxis following major trauma, the prevalence of
deep-vein thrombosis has ranged from 2%16 to
33%18, depending on the
patient population studied, the type of screening study, and the
type of prophylaxis against the deep-vein thrombosis.
Mechanical prophylaxis against deep-vein thrombosis provides
protection without increasing the risk of blood loss. Mechanical
devices work by improving venous blood flow as well as by stimulating
endogenous fibrinolytic activity21.
The effectiveness of mechanical prophylaxis in patients undergoing
elective total joint replacement has been well documented. However,
the literature regarding mechanical prophylaxis after multiple trauma
is sparse3,9,12, with reported
rates of deep-vein thrombosis ranging from 4% to 21%.
Spain et al. evaluated 184 consecutive patients in a retrospective
study comparing sequential and pulsatile pumps10.
Sequential compression was used when possible, and pulsatile compression
was used when the patient had a lower-extremity fracture or another contraindication
to sequential compression. The authors found a 7% prevalence
of deep-vein thrombosis with sequential compression and a 3% prevalence
with pulsatile compression. The retrospective, nonrandomized design
as well as the variety of injury types were weaknesses of that study.
Ultrasound is notoriously poor at detecting pelvic clots. This
problem is exacerbated in the setting of trauma with the presence
of a pelvic hematoma. Ascending venography is invasive and also
lacks sensitivity for detecting pelvic deep-vein thromboses. Montgomery
et al. documented a false-negative rate of 58% for ascending
venography compared with magnetic resonance venography18. A major difficulty in evaluating
these different studies is that it is not clear what test, if any,
represents a "gold standard" for detecting pelvic
deep-vein thromboses. Ascending venography clearly is not accurate6,18. While Stover et al.13 advocated direct cannulation venography,
other authors have noted that it is not accurate for the internal
iliac system6,18. The accuracy
of magnetic resonance venography depends upon experienced interpretation
by radiologists familiar with the technique in patients with traumatic
injury. It has a high sensitivity and may detect clots that are
small and not at risk for embolization. We attempted to address
this problem by presenting our results in terms of the total number of
deep-vein thromboses as well as the number of large (2-cm) or occlusive
deep-vein thromboses. We are not certain of the importance of small
nonocclusive deep-vein thromboses in the pelvic veins; however,
because of their proximal location and the potential for clot propagation,
we elected to treat these patients for deep-vein thrombosis.
Magnetic resonance venography has the major advantage of being
able to detect thromboses in the pelvic veins as well as in the
thigh. In our study, six of fourteen patients with identified deep-vein thrombosis
had involvement of a pelvic vein, a finding that is consistent with
the results reported by Montgomery et al.18.
Numerous previous studies have documented pulmonary embolism despite
normal findings on ultrasound or ascending venography5,6,16,18. A probable source for these
emboli is the pelvic venous system.
The primary weakness of our study is that the sample size was
not adequate to prove superiority of one type of mechanical prophylaxis
over the other. Statistical evaluation demonstrated that approximately
350 patients would be necessary if current trends continued. Another
weakness of the study centers on the reliability of the data on
patient compliance. Some patients find mechanical pumps uncomfortable,
and compliance by patients and by hospital personnel remains an
acknowledged problem in this and other studies8.
This study confirms that patient age is a critical risk factor
in the development of deep-vein thrombosis following trauma. The
patients in whom occlusive deep-vein thrombosis developed were an
average of eleven years older than those in whom it did not develop.
There was also a significant association between the time to surgery
and the development of thrombosis, with the delay being more than
twice as long (9.8 compared with 4.7 days) for patients in whom
an occlusive clot developed compared with those in whom no deep-vein
thrombosis developed. Both of these factors are associated with
increased severity of injury or increased complications following
injury, or both.