The present study was a prospective, controlled, randomized, single-center
clinical trial of a consecutive series of selected patients who
were scheduled to have a primary total hip arthroplasty. The main
criteria for exclusion from the study were a history or symptoms
of deep-vein thrombosis or pulmonary embolism, esophageal disease,
or metabolic bone disease or current, ongoing anticoagulation therapy.
A total of 161 consecutive patients with advanced osteoarthritis
were referred to the study. Thirty-one patients were not randomly assigned
to a study treatment: eleven (35%), because they withdrew
consent; five (16%), because of geographic inaccessibility;
five (16%), because of previous venous thrombosis; three
(10%), because of long-term anticoagulation therapy; two
(6%), because of previous pulmonary embolism; two (6%),
because of esophageal disease; and three (10%), for other
reasons. The 130 remaining patients were randomly assigned, with
use of computer-generated random-number sheets at the time of the
operation, to have the femoral component inserted with use of our
standard cementing technique or with use of the bone-vacuum cementing
technique. No restriction was applied to the assignment of the interventions.
The intervention assignment was concealed until the time of the operation.
The study was approved by the Investigational Review Committee of
the hospital. The 130 patients who were eligible for enrollment
in the clinical trial signed an informed-consent form. All operations
were performed or supervised by the senior author (R.P.P.).
The preoperative physical status of the patients was assessed according
to the criteria of the American Society of Anesthesiologists21. Clinical data were gathered by
chart review at admission and by the anesthesiologist’s
evaluation. All operations were performed with the patient under
general anesthesia22. Clinical
monitoring included a three-lead electrocardiogram, invasive measurement
of arterial pressure with a cannula placed into the radial artery,
and measurement of pulse oximetric oxygen saturation and end tidal
carbon dioxide tension. Patients with severe associated diseases
rated class 3 or 4 according to the system of the American Society
of Anesthesiologists had a central venous catheter placed into the right
jugular vein. Blood loss was replaced with predonated autologous
blood and with blood salvaged intraoperatively with use of the Cell
Saver (Haemonetics, Braintree, Massachusetts). Both types of blood-loss
management were used in all but two patients, who did not receive
predonated blood. The blood-loss index, a validated measure used
in clinical studies8, was calculated
with the equation: blood-loss index = preoperative level
of hemoglobin — level of hemoglobin before discharge + number
of units transfused. Postoperative bleeding events were classified
as minor bleeding from a wound (bleeding at the injection site,
epistaxis, or wound hematoma not requiring operative decompression),
major bleeding from a wound (wound hematoma requiring operative
decompression), or major bleeding not related to a wound (gastrointestinal
or intracerebral hemorrhage).
The hip joint was exposed through the direct lateral approach with
the patient in the supine position. Cement was used to fix an all-polyethylene
cup (Müller; Brehm, Weisendorf, Germany) in 102 hips, a
press-fit acetabular component (Phoenix; Brehm) was inserted without
cement in twenty hips, and a reinforcement ring (Müller;
Sulzer-Medica, Baar, Switzerland) was inserted in the remaining
eight hips.
In the control group of sixty-five hips, a femoral component was
cemented with use of our standard technique. The femoral canal was
opened with an osteotome, and the medullary cavity was prepared
with chipped-tooth broaches after aspiration of fat and bone marrow.
An autologous bone plug was placed into the distal part of the femur,
and the bone surface was washed with pulsatile high-volume lavage
with 1000 mL of saline solution. Vacuum-mixed standard-viscosity
polymethylmethacrylate (Palacos-R; Biomet Merck, Darmstadt, Germany)
was introduced in an antegrade direction from proximal to distal
and pressurized with a cement gun (Sterivap; Merck, Darmstadt, Germany)
before the insertion of the stem (Müller Straight-Stem;
Brehm). A Teflon (polytetrafluoroethylene) tube with an inner diameter
of 4 mm was used to vent the femoral canal at the level of the bone
plug during the introduction of the cement and insertion of the stem.
In the sixty-five hips managed with the bone-vacuum technique,
the femoral canal was prepared with cutting tools coated with a
fine, pure diamond surface (Scientific Development, Munich, Germany).
These instruments are designed to reduce the production of bone
debris and to prevent an increase in intramedullary pressure. A
proximal drainage cannula with a diameter of 4.5 mm was positioned
on the projection of the linea aspera into the cancellous bone of
the intertrochanteric region (Fig. 1). The cannula was connected with
a suction tube to a vacuum pump (Vacumix; Braun, Melsungen, Germany),
which provided suction of —800 mbar. The drainage system
was checked frequently to make sure it was functioning adequately.
No bleeding of the cancellous bone is visible if the suction system
is working correctly. The cannula drains the proximal part of the
femoral canal during the insertion of the stem, prevents an increase
in intramedullary pressure, and promotes intrusion of polymethylmethacrylate
into the cancellous bone. An autologous bone plug was placed into
the distal part of the femur, and the bone surface was washed with pulsatile
high-volume lavage with 1000 mL of saline solution. The same bone
cement, the same tube for venting of the femoral canal at the level
of the bone plug, the same bone-cement introduction technique, and
the same stem as used in the control group were used in the group
managed with the bone-vacuum technique. The drainage cannula was
removed after the bone cement had hardened (twelve to fourteen minutes
after mixing of the bone cement).
Intraoperative fat and bone-marrow embolism was determined by
echocardiography (Sonoline Versa Plus; Siemens, Erlangen, Germany)
with a 5.0-MHz multiplanar transesophageal probe. The intraoperative
monitoring was continuously videotaped, and the images were analyzed
by an observer who was blinded to the method of cementing of the
stem. The echogenic patterns detected during the operation were
rated with use of a four-grade score20,23.
They were considered to be grade 0 when there were no emboli or
small echogenic particles in the right atrium and right ventricle,
grade 1 when there was a cascade of fine emboli and the right atrium
and ventricle were partially opacified with echogenic material,
grade 2 when there was a cascade of fine emboli or embolic masses
with a diameter of £5 mm and the right atrium and ventricle
were completely opacified with echogenic material, and grade 3 when
there were fine emboli mixed with large embolic masses that had
a diameter of >5 mm or there were serpentine emboli.
Blood samples for hemodynamic measurements were drawn at the
time of the operative incision (event 1), five minutes after implantation
of the cup (event 2), after preparation of the femoral canal (event
3), five minutes after insertion of the stem (event 4), five minutes
after reduction of the hip joint (event 5), fifteen minutes after
relocation of the hip joint (event 6), and two hours after the end
of the operation (event 7). Arterial and mixed venous blood was
obtained for blood-gas analysis. The method of Ries et al.24 was used to derive the shunt values
(the proportion of the cardiac output that passes through the lungs
without participating in the gas exchange).
All patients received pharmacological prophylaxis against deep-vein
thrombosis. Low-molecular-weight heparin (Fraxiparine [nadroparin-calcium];
Sanofi-Synthelabo, Berlin, Germany) was administrated subcutaneously
(dose adjusted according to body weight, 0.2 to 0.6 mL; 0.1 mL = 950
IU of anti-Xa) twelve hours before the operation and was continued daily
until discharge. All patients wore bilateral thigh-high antithromboembolic
stockings and received instructions to actively dorsiflex and plantar
flex the ankles when resting in bed. Physical therapy, range of
motion exercises, and walking with full weight-bearing were usually
initiated on the second postoperative day.
Postoperatively, each lower extremity of all 130 patients was scanned
for the presence of deep-vein thrombosis, and the size and site
of any thrombus was determined, by duplex ultrasonography (Sonoline
Elegra; Siemens) with use of a 5.0 and a 7.5-MHz linear transducer.
The scans were performed from the inguinal ligament, beyond the
trifurcation of the popliteal vein to the level of the ankle, and
included visualization of the common femoral, deep and superficial
femoral, popliteal, peroneal, posterior tibial, anterior tibial,
and deep soleal veins. Bilateral duplex studies were performed preoperatively
and on postoperative days 4, 14, and 45. The duplex images were
analyzed by an observer who was blinded with regard to the method
of fixation of the femoral component. The criteria indicating normal
findings on duplex examination were (1) complete coaptability of
the vein through manual pressure with the transducer probe, (2)
the absence of echogenic thrombi, (3) normal venous Doppler signals
in response to respiration or to the Valsalva maneuver, and (4) good
visualization on ultrasound25.
Deep-vein thrombosis was classified as proximal if there was involvement
of the deep veins proximal to the trifurcation in the calf (popliteal,
femoral, or iliac veins) and as distal if there was involvement
distal to the trifurcation (peroneal, posterior tibial, or anterior
tibial veins) and no proximal thrombosis was detected. Patients
with abnormal findings on duplex ultrasonography on postoperative
days 4 or 14 were examined thereafter with serial duplex ultrasonography
studies in order to detect propagation, stability, and/or
resolution of the deep-vein thrombosis. Detection of a deep-vein
thrombosis warranted therapy with low-molecular-weight heparin (dose adjusted
according to body weight, 0.6 to 1.2 mL [5700 to 11,400
IU of anti-Xa]) and use of compression stockings.
Reliability of Duplex Ultrasonography
Venography and duplex ultrasonography were performed on a separate
cohort of fifty-five patients (110 lower extremities) who were scheduled
to have a total hip arthroplasty, in order to determine the reliability
of ultrasonography at our institution. This investigation was needed
to validate the clinical relevance of the randomized clinical trial.
All patients signed an informed-consent form. Duplex ultrasonography
and venography were performed preoperatively and on postoperative
day 7 by the observer in the main study. Venography was performed
after duplex ultrasonography, and the results were analyzed by a
single experienced radiologist who was blinded to the findings of
the duplex ultrasonography. Venography was performed with use of
the method described by Rabinov and Paulin26,
except that the superficial veins were occluded by an ankle tourniquet.
Statistical Analysis
The continuous demographic data of the two groups of patients
were analyzed with use of a two-tailed, unpaired t test. Mean values
were given with the standard deviation. For rank-scaled data, median
values were given with the interquartile range. Relative frequencies
of unpaired samples were compared with use of Fisher’s
exact test. Unpaired groups of continuous data without assumption
of normal distribution were compared with use of the Mann-Whitney
U test. Two-sided p values of £0.05 were considered significant.
Correction for multiple comparisons was done with the method of Hommel
to control type-I error. All calculations were carried out with
use of SPSS for Windows (version 9; SPSS, Chicago, Illinois).
A total of 130 patients were eligible for the study. There was no
difference between the two groups of sixty-five patients with respect
to gender, age, body-mass index, duration of the operation, blood-loss
index, duration of hospitalization, or associated systemic disease
(Table I). Echocardiography
revealed a cascade of fine echogenic particles or embolic masses
with a diameter of £5 mm (grade 2) during the insertion
of the femoral component in fifty-nine patients (fifty-nine hips;
91%) treated with our standard technique and in ten patients
(ten hips; 15%) treated with the bone-vacuum technique
(Table II).
The difference between the two groups with respect to the intensity
and duration of the grade-2 embolic events was significant (p < 0.001).
Embolic masses with a diameter of >5 mm (grade 3) were
observed after the relocation of the hip joint in thirty-three patients
(51%) in the control group and in five patients (8%)
in the group managed with the bone-vacuum technique (Figs. 2-A and 2-B); the difference
was significant (p < 0.001). No embolic events were detected
during the operative approach to the hip joint, the osteotomy of
the femoral neck, or the preparation of the socket. The insertion
of the cup with cement caused a short-term grade-2 embolism in one
patient (2%) managed with the bone-vacuum technique. Eleven
patients (17%) in the control group had a transient grade-2
embolism that was probably caused by the preparation of the femur
with broaches. In contrast, none of the patients managed with the
bone-vacuum technique, which was performed with specialized cutting tools,
had a grade-2 embolism during the preparation of the femur.
The mean pulmonary shunt values increased only slightly, from
8.3% to 8.6%, during the operation in the patients
managed with the bone-vacuum cementing technique. In contrast, a
significant increase in the shunt values, from 8.2% to
10.1% (p < 0.05), indicating intraoperative cardiopulmonary
impairment, was observed after insertion of the stem with our standard
cementing technique. A nearly immediate cause-and-effect relationship
was observed between grade-2 and grade-3 embolic events detected
during surgery and signs of cardiopulmonary impairment. A grade-1
embolism was never followed by changes in laboratory and clinical
parameters. In the majority of the control patients, the pulmonary
shunt values returned to baseline only at the end of the operation.
Abnormally high proportions of the cardiac output that passes through
the lungs without participating in gas exchange persisted in seven
patients in the control group who had severe transatrial echogenic
events and preexisting severe systemic disease (class 3 according
to the system of the American Society of Anesthesiology).
Reliability of Duplex Ultrasonography
Thrombosis of the venous system was detected with use of duplex
ultrasonography on postoperative day 4 in twelve (18%)
of the sixty-five patients in the control group and in two (3%)
of the sixty-five patients managed with the bone-vacuum technique
(Table III);
the difference was significant (p = 0.009). Thrombosis
of the distal venous system was found in five of the twelve patients
in the control group and in both patients managed with the bone-vacuum
technique. Distal thrombosis was asymptomatic in all seven patients.
Thrombosis of the proximal venous system was observed in seven patients
managed with the standard cementing technique but was not observed in
the patients managed with the bone-vacuum technique. Proximal deep-vein
thrombosis was symptomatic in one patient. None of the patients
with proximal deep-vein thrombosis showed concomitant ipsilateral
distal thrombosis. Of the fourteen patients who had deep-vein thrombosis,
thirteen had deep-vein thrombosis detected in the involved extremity
and one (in the control group) had it in the uninvolved extremity. The
length of the thrombus averaged 19 cm (range, 6 to 38 cm) in the
affected patients in the control group and was 14 and 22 cm in the
two patients who were managed with the bone-vacuum technique. The
difference between the groups was not significant (p = 0.1).
Two patients (3%) in the control group had an asymptomatic
distal deep-vein thrombosis on postoperative day 14. Both had had
negative findings on ultrasonography on postoperative day 4. No
thrombosis of the venous system occurred in either group of patients
after discharge. Resolution of the deep-vein thrombosis was observed at
postoperative day 45 in fifteen of the sixteen patients with positive
findings on ultrasonography after the index operation. One patient
in the control group had persisting deep-vein thrombosis, and anticoagulation
was continued for that patient.
With the numbers available, multivariate analysis revealed that
gender, age, body-mass index, duration of the operation, preoperative
level of hemoglobin, level of hemoglobin before discharge, and number
of units reinfused and transfused did not affect the rate of postoperative
deep-vein thrombosis (p > 0.05 for all).
Minor bleeding from the wound was observed postoperatively in
four patients (6%) in the control group and in five patients (8%)
managed with the bone-vacuum technique. Major bleeding from a wound
or major bleeding not related to a wound did not occur. No patient
had symptoms consistent with pulmonary embolism within the first
forty-five days after the index operation.
Compared with venography, duplex ultrasonography showed 93% sensitivity,
98% specificity, and 95.5% accuracy with regard
to overall detection of deep-vein thrombosis in 110 lower extremities
in fifty-five patients (Table IV). The results of the venography
and duplex ultrasonography fully agreed with regard to proximal
deep-vein thrombosis (100% sensitivity, 100% specificity,
and 100% accuracy), and they agreed with regard to distal
deep-vein thrombosis for all but three limbs (83% sensitivity,
98% specificity, and 90.5% accuracy).
We hypothesized that fat and bone-marrow embolism is an important
cause of thrombogenesis during total hip arthroplasty and that prevention
of fat and bone-marrow embolism can reduce the incidence of postoperative
thrombosis of the deep venous system. To our knowledge, the present
study is the first randomized clinical trial designed to investigate
the relationship between intraoperative fat and bone-marrow embolism
and thrombosis of the deep venous system. The low incidence of postoperative
deep-vein thrombosis observed in the group of patients with the
lower rate of fat and bone-marrow embolism seems to confirm the
hypothesis of the study.
During the last three decades, substantial advances have been made
in the understanding of the pathophysiology of and prophylaxis against
venous thromboembolic disease. Nevertheless, thromboembolism remains
an important cause of morbidity and mortality during and after total
hip arthroplasty5,27-30. Although
it has been proposed that the initial stimulus for venous thrombosis
occurs during the operation9,12,20,31 and
that the period of maximal risk of thrombogenesis in patients managed
with a total hip arthroplasty occurs during the operation7,9,11-14, most treatment is currently
directed at retarding the extension of existing thrombi with the
use of postoperative therapies such as anticoagulation medications
and/or mechanical devices. When the femoral component is
inserted with use of standard surgical techniques, tissue thromboplastin
from the bone marrow is forced into the draining veins of the proximal part
of the femur, leading to activation of the clotting cascade, thromboembolism,
and cardiopulmonary impairment9,20,31,32.
Furthermore, embolization of bone-marrow elements can produce direct
venous endothelial lesions. The study by Stewart et al.14 showed that total hip arthroplasty
in dogs was ultimately responsible for the production of frequent,
identifiable endothelial tears infiltrated with leukocytes and platelets
in veins around and distant to the hip joint. These changes were
not observed in veins in control animals. Stewart et al. suggested that
total hip arthroplasty causes the release of vasoactive substances
and that these substances enter the circulation and survive long
enough to influence veins distant from the surgical site. Kinking
and occlusion of the femoral vein caused by excessive flexion and
rotation of the leg during the operation also may disrupt the endothelium
and cause venous stasis9,10,33.
These observations support the recommendation that the prevention
of deep venous thrombosis associated with total hip arthroplasty
be focused on the period during the insertion of the femoral component
rather than during the postoperative period.
The increase in intramedullary pressure in the femur during insertion
of the stem is the most important pathogenic factor for the migration
of bone marrow, fat, and bone debris and the embolization of these
elements through the venous system located along the linea aspera11,15,17,34. Intraoperative embolic
phenomena can be reduced with use of the bone-vacuum technique19,20. With this technique, it is possible
to prevent an increase in intrafemoral pressure during the insertion
of the stem. This may result in less release of tissue thromboplastin,
less pronounced activation of the clotting cascade during the total
hip arthroplasty, and less damage of the endothelium of the peripheral
venous system. Experimental investigations have demonstrated an
improvement in the penetration of cement into cancellous bone with
use of the bone-vacuum technique19,35,
and a clinical study has shown satisfactory radiographic results
that were comparable with those achieved with use of a contemporary
cementing technique36. In the
present study, suction of —800 mbar applied to the bone-vacuum
cannula appeared to provide sufficient drainage in the proximal
portion of the femur to reduce migration of the medullary content
during insertion of the stem. The echocardiographic findings suggest
that a distal drainage system is subordinate to the proximal one
for prophylaxis against fat and bone-marrow embolism. Furthermore,
avoiding the use of the distal cannula eliminates the risk of fracture
of the femur due to concentration of stress surrounding the venting hole.
The clinical relevance of a single cannula placed along the projection
of the linea aspera for the prevention of intraoperative pulmonary
impairment was confirmed in a separate randomized trial34.
The present study has several potential limitations. No circulating
indices of thrombosis and fibrinolysis were determined during the
operation. Thus, quantification of thrombogenic activity and correlation
with intensity of intraoperative embolic events were not possible.
Sharrock et al.9 observed a significant
increase of D-dimer, fibrinopeptide A, thrombin-antithrombin complexes,
and prothrombin F1.2 after the insertion of femoral components with
cement. In our opinion, the timing of embolic events observed during
the insertion of the stem in the present study and the clotting
cascade activation reported by Sharrock et al. suggest a cause-and-effect
relationship.
In the present study, deep-vein thrombosis was detected with use
of serial duplex ultrasonography. This method has the advantage
of being noninvasive, safe, and repeatable. Nevertheless, much of
the research on thromboembolism in orthopaedic surgery has been
based on venograms7. Contrast
venography is more sensitive than ultrasonography, but, since it
is invasive, uncomfortable, and possibly thrombogenic, repeated
investigation with use of this technique is impractical. A single
venogram can only measure prevalence, the rate at the moment when
the test is carried out, rather than incidence, the total rate in
the postoperative period37. Duplex
ultrasonography is highly operator-dependent, with accuracy varying
widely depending on the expertise and the experience of the observer.
Thus, in order to substantiate the reliability of duplex ultrasonography
for the detection of postoperative deep-vein thrombosis at our institution,
we investigated a separate cohort of fifty-five patients who underwent total
hip arthroplasty. The study was needed to validate the findings
of the randomized trial and was performed by the same observer as
in that trial. On comparison with venography, duplex ultrasonography
demonstrated 93% sensitivity, 98% specificity,
and 95.5% accuracy with regard to overall detection of
deep-vein thrombosis. On the other hand, the sensitivity of duplex
ultrasonography for the detection of distal deep-vein thrombosis
was only 83%. These findings agree with data reported by
other investigators25,37,38, and
we considered them sufficient to confirm the value of serial duplex
sonography for the diagnosis of deep-vein thrombosis in the present
study. Nevertheless, this noninvasive method is still somewhat limited
with regard to the detection of distal deep-vein thrombosis.
The results in the present study were influenced by our use of pharmacological
prophylaxis. In Germany, administration of heparin is required before
and after all total hip arthroplasties. It can be postulated that
the incidence of thrombosis of the venous system is substantially
increased in patients managed with the bone-vacuum cementing technique
but not treated with heparin or other drugs. While the period of
maximal thrombogenesis occurs during the insertion of the femoral component,
it is not known how long the risk of thrombotic stimulus persists
after the operation9,12,23. Furthermore,
other risk factors for thrombogenesis, such as hypercoagulability,
may not be assessed with intraoperative measures alone.
The present randomized study has several major strengths. A large
number of patients were studied; the gender, age, body-mass index,
and physical status were similar in the two groups; and all operations
were performed or supervised by the same surgeon. Transesophageal
echocardiography has been shown to be a reliable method for the
detection of fat and bone-marrow elements passing through the heart
during orthopaedic and trauma-related operative procedures12,20,39. The four-grade scoring system
for the quantification of the embolic events is similar to scoring
systems used in other studies40,41 and
provides the same reproducibility and validation limits. In the
present study, although we did not perform intraobserver and interobserver
tests, one blinded, experienced observer quantified echocardiographic
findings off-line. A computer-assisted analysis probably is a more
accurate method for assessment of echogenic events, but, in this
setting, the reproducibility and validation have not yet been defined42,43.
The overall incidence of thrombosis of the venous system (22%)
detected in the control group of patients is higher than that reported
in other randomized studies in which low-molecular-weight heparin
was administered for prophylaxis against deep-vein thrombosis after
total hip arthroplasty1,3,8,44.
The high incidence of deep-vein thrombosis in the control group
of the present study could be related to the management of the patients
with general anesthesia. Many case studies have demonstrated that
epidural anesthesia reduces the risk of postoperative deep-vein
thrombosis45,46.
Interestingly, in the present study, resolution of deep-vein thrombosis
was observed at postoperative day 45 in fifteen of sixteen patients
with positive postoperative ultrasonography. There is currently
a paucity of information about venous recanalization after deep-vein
thrombosis.
Most investigations related to the prevalence of deep-vein thrombosis
have involved cemented components47,48.
It is possible that cement may have inherent thrombogenic properties
and thus influence the prevalence of thrombosis of the deep venous
system. Reports concerning this possibility are conflicting; activation
of the clotting cascade and concurrent fibrinolysis have been described
during and after total hip and total knee arthroplasty, but it is
uncertain whether this activation differs according to whether the
prosthesis is inserted with or without cement31,32,49.
A previous echocardiographic investigation showed a low risk of
fat and bone-marrow embolism during total hip arthroplasty performed
without cement23. While the majority
of studies on the rate of deep-vein thrombosis associated with total
hip arthroplasty performed with and without cement were not randomized50,51, Laupacis et al.52, who randomized 250 patients undergoing
total hip arthroplasty to receive a femoral component either with
or without cement, found no marked difference between the two groups with
respect to the frequency of deep-vein thrombosis (50% compared
with 47%, respectively).
In conclusion, the results of the present study show that surgical
prophylaxis against fat and bone-marrow embolism is effective in
reducing the incidence of postoperative thrombosis of the deep venous
system. While the decision regarding the best means of postoperative
prophylaxis remains controversial, we advocate the use of intraoperative
prophylaxis as well as postoperative prophylaxis to lower the risk.
Preparation of the femoral canal with diamond cutting tools requires the
same amount of time that is generally necessary with use of conventional
broaches. Positioning of the cannula, connection to the drainage
pump, and checking for adequate function require three to five minutes.
As a result of the advantageous prophylactic effect of the bone-vacuum
technique, we now use it routinely in all patients managed with
cementing of a femoral component.