Abstract
Introduction:
Blood loss following total knee arthroplasty can lead to substantial morbidity and the need for blood transfusions. Hemostatic agents have been used to minimize blood loss and to decrease transfusion rates. Floseal is a thrombin-based hemostatic agent with unknown efficacy for achieving these goals in patients undergoing total knee arthroplasty.
Methods:
We performed a prospective randomized controlled trial on the use of Floseal in patients undergoing total knee arthroplasty, with the primary end point being blood loss as measured through drain output. Demographic characteristics, operative side, diagnosis, intraoperative details, implant choice, hospital course, laboratory values, visual analog scale pain scores, knee range of motion, adverse events, transfusion rates, and deviations from protocol were recorded.
Results:
A total of 196 patients were enrolled, with ninety-seven patients being randomized to the Floseal group and ninety-nine patients being randomized to the control group. There were no significant differences between the Floseal and control groups in terms of drain output at twenty-four hours (711 compared with 702 mL; p = 0.823). No differences were noted between the groups in terms of operative side, diagnosis, intraoperative details, implant choice, hospital course, laboratory values, visual analog scale pain scores, knee range of motion, or transfusion rates. Complications occurred infrequently. In the acute postoperative period, there were two cases of cellulitis (one in each group), two deep venous thromboses (one in each group), and one paralytic ileus (in the control group), all of which resolved with nonoperative measures. At the six-week follow-up, one patient in the Floseal group had died from a cause unrelated to surgery, two patients (one in each group) had suture abscesses with cellulitis that resolved with postoperative antibiotics, and four patients (two in each group) underwent knee manipulation under anesthesia to achieve improved knee motion. With the numbers available, there was no significant association between Floseal use and the occurrence of these adverse events.
Conclusions:
The present study showed that Floseal had no demonstrable effect on blood loss as measured through drain output following total knee arthroplasty. There were also no notable adverse events associated with its use. The usefulness of Floseal as a hemostatic agent in total knee arthroplasty remains unclear.
Level of Evidence:
Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence.
Bone and soft-tissue bleeding following total knee arthroplasty is a major cause of postoperative morbidity, including pain, decreased knee motion, increased transfusion requirements, and, often, increased convalescence times1. After total knee arthroplasty, the cut and exposed bone ends that are not covered by the implants are a major contributor to postoperative bleeding. The control of bleeding from exposed bone is not amenable to electrocautery or ligature control2. The use of antiplatelet and anticoagulant agents to reduce thromboembolic incidents has complicated the issue by increasing bleeding complications. The development of locally applied hemostatic agents at the time of surgery may help to reduce blood loss and to improve surgical outcomes3,4.
The use of topical agents—including both fibrin-based and thrombin-based agents—to inhibit bleeding directly at its source is of interest3-7. Two prospective, randomized, controlled trials examined the efficacy of a fibrin sealant to reduce bleeding in patients managed with total knee arthroplasty3,4. Wang et al. reported that the decrease in the hemoglobin level was significantly (28.9%) less in the fibrin-sealant group compared with the control group3. Similarly, Levy et al. found that the decrease in the level was 25 g/L in the treatment group, compared with 37 g/L in the control group (p < 0.001)4. Both studies also demonstrated a substantial reduction in drain output and total measured blood loss.
Floseal (Baxter, Deerfield, Illinois), a thrombin-based sealant, is a high-viscosity hemostatic gel composed of specifically engineered collagen-derived granules mixed with topical thrombin. It has been found to stop bleeding in two minutes and functions on actively bleeding tissue5. It is both bioresorbable and biocompatible. The advantage of Floseal over fibrin-based agents is that it does not require an additional person in the operating room to draw blood from the patient for production of fibrin-rich cryoprecipitate. Floseal has been shown to be efficacious for reducing blood loss in both spinal and cardiac surgery5-7.
We are not aware of any prospective, randomized, controlled trial that has evaluated the efficacy of a thrombin-based hemostatic agent in reducing blood loss in patients managed with total knee arthroplasty. Therefore, we designed a randomized controlled study to evaluate the use of such an agent in patients undergoing total knee arthroplasty. The primary outcome measure was drain output, and a secondary outcome measure was the transfusion rate. We hypothesized that the application of a thrombin-based hemostatic sealant to uncovered bone areas in patients with fully cemented total knee implants would decrease bleeding and lead to lower transfusion rates.
The present study received institutional review board approval, and informed consent was obtained from all patients. Patients were enrolled in a consecutive prospective manner on a voluntary basis. The study was registered in the public trial registry (clinicaltrials.gov identifier #NCT00990288).
Participants
Candidates for the study were patients undergoing primary total knee arthroplasty for the treatment of osteoarthritis. Patients were excluded if they had a history of allergies to bovine products or a known bleeding disorder. Patients with inflammatory arthritis were also excluded. Data were collected preoperatively, postoperatively in the recovery room, and throughout the hospital stay until the time of discharge.
Intervention
All patients received a fully cemented posterior stabilized total knee replacement with patellar resurfacing. Surgery was performed with an above-the-knee tourniquet that was inflated prior to the incision and released prior to wound closure. Randomization to the treatment or control group occurred at the time of surgery via a sealed envelope system. The surgeon was notified of randomization after the cementing of components but prior to the application of the sealant. If the envelope said “no,” Floseal was not applied and hemostasis was achieved in the usual fashion with electrocautery followed by wound closure. If the envelope said “yes,” hemostasis of the soft tissues was performed in the usual fashion with electrocautery but Floseal was applied to all exposed bone surfaces after the tourniquet had been released and the bone cement had polymerized. Specifically, Floseal was applied to all exposed bone areas around the tibial plateau, the tibial pinholes, and the back of the knee. The knee was hyperflexed to expose the posterior aspects of the femoral condyles as well as the posterior exposed bone of the tibial plateau. Floseal was then applied to the lateral and medial femoral bone surfaces that were exposed adjacent to the prosthesis, in the intercondylar notch, and on the cut anterior surface of the femur. Last, Floseal was applied around the margins of the patella. The hemostatic matrix was also applied to the intracapsular soft tissues. All of a 10-mL vial of the product was used. Floseal remained in place for two minutes, and then any excess product was gently rinsed from the knee with a bulb syringe. Two 3.175-mm-size deep drains (ConstaVac; Stryker, Kalamazoo, Michigan) were inserted into all knees following deflation of the tourniquet. These drains were left on low-flow suction (25 mm Hg) for the duration of their insertion time. Fascial layers were closed in the usual fashion, and staples were used superficially. A compressive bandage was applied to the limb following closure.
A single orthopaedic surgeon (M.P.F.) performed the arthroplasties and applied the Floseal. All patients underwent hypotensive anesthesia with spinal anesthesia and a femoral nerve block. All patients were managed with warfarin as well as with sequential pneumatic compressive boots for prophylaxis against venous thromboembolism. The first dose of warfarin was given on the night of surgery, and warfarin treatment was continued for six weeks. All patients had a preoperative international normalized ratio (INR) of <1.4, which is a set standard for epidural anesthesia at our institution. An INR from 2.0 to 2.5 was used for therapeutic dosage maintenance. Daily blood draws for cell counts, chemistry studies, and coagulation profile were taken in the immediate postoperative period and to the third postoperative day. The use of a controlled passive motion device, physical therapy, and mobilization all commenced on the first postoperative day. Total drain output was recorded at twenty-four hours. The time of insertion of each drain was recorded so that identical time measurements were recorded. Drains were then removed after twenty-four hours.
Outcome Measures
The primary end point was the total drain output at twenty-four hours postoperatively. Several secondary outcome measures were assessed. Visual analog pain scores for each knee were obtained on the third day postoperatively and at six weeks after discharge. Passive knee motion was also recorded at these time points. Transfusion requirements and the change in hematocrit and hemoglobin values were also recorded for the hospital stay. A record of adverse events both related to the wound and to any medical complications was recorded. Total blood volume loss was also calculated.
Sample Size
We are aware of no previous studies that have evaluated the differences in blood loss following total knee replacement between knees receiving and not receiving a thrombin-based hemostatic agent. Therefore, we used data from our pilot study to perform an a priori power analysis. We calculated our sample size with use of the following inputs: 80% power, a critical p value of 0.05, and a mean difference (and standard deviation) of 150 ± 350 mL. Our pilot data suggested a mean difference of 120 ± 350 mL, but we decided that a mean difference of 150 mL was more clinically relevant. This resulted in a necessary sample size of eighty-seven subjects per group, for a total required sample size of 174 subjects. We added a 10% “failure” rate for improper drain usage, giving us a total needed sample size of ninety-eight subjects per group, or 196 for the study. A post hoc power calculation using the actual number of subjects in the study and the expected drain output differences above demonstrated 85% power to detect those differences.
Statistical Analysis Plan
Statistical analysis consisted of means and standard deviations for continuous data and frequency counts for discrete data. Inferential analysis was conducted with use of a Wilcoxon paired rank test to account for potentially non-normal data distributions for continuous variables and a chi-square test for discrete variables. A two-tailed p value of ≤0.05 was considered significant.
Randomization
A computer-generated randomization list was drawn up by the statistician with use of the SAS System for Windows 9.1 (SAS Institute, Cary, North Carolina). Each randomized selection was then put into a sealed envelope and numbered from 1 to 196 (Fig. 1). No one involved in the surgery or postoperative assessment was involved in the randomization process.
All study personnel and participants were blinded to the treatment assignment for the duration of the study. Only the study statistician (S.L.) and the data monitoring committee saw unblinded data, but none of those individuals had any contact with study participants.
Source of Funding
This study received product support for research from Baxter Healthcare Corporation (Baxter, Deerfield, Illinois). Specifically, the Floseal was provided free of charge from Baxter, and funds were used for research coordinator salary support, database and computer-related costs, institutional review board fees, Spanish translation of the informed-consent form, and Clinical Research Department and close-out fees. No one from the corporation was involved in the data collection or analysis or any other detail of the study, aside from study design, education about application of the product, and product delivery and availability for the purposes of this study.
The study enrollment period started in February 2009 and lasted until October 2010 in order to reach the sample size needed from our a priori power calculation. In this time period, 336 patients met the inclusion criteria to enroll in the study, and a total of 196 patients volunteered and gave consent for participation. Of the 196 patients in the study, ninety-seven were randomized to the Floseal group and ninety-nine were randomized to the control group. The diagnosis was degenerative joint disease for 98% of the patients and osteonecrosis for 2%. Of the patients with degenerative joint disease, eight had posttraumatic osteoarthritis and four had an anterior cruciate ligament-deficient knee. These diagnoses were distributed between the Floseal and control groups (p = 0.366). Forty-four percent of the knee replacements were performed on the left side, and 56% were performed on the right side. A lateral retinacular release was performed during 31% of the arthroplasties.
The average age was 72.7 years in the Floseal group, compared with 70.1 years in the control group. Table I and a table in the Appendix show additional demographic information as well as preoperative knee motion values, intraoperative details, transfusion details, and laboratory values. The drain output at twenty-four hours was 711.3 ± 287.2 mL in the Floseal group, compared with 701.6 ± 314.7 mL in the control group (p = 0.823). On the basis of these numbers, there was no significant difference between the two groups in terms of preoperative deformity, knee motion (Table II), intraoperative details, or visual analog scale pain scores (Table III). Assessment of randomization demonstrated that, with the numbers available, there were no significant differences between the two groups with regard to baseline characteristics.
A significant difference was noted between the groups in terms of the red blood-cell counts on the first and second postoperative days. On the first postoperative day, the average hemoglobin value was 10.55 g/dL for the Floseal group and 10.03 g/dL for the control group (p = 0.001). On the second postoperative day, the values were 9.85 and 9.49 g/dL, respectively (p = 0.02) (Table I).
To further assess transfusion rates, we also analyzed the number of patients who had donated autologous blood for transfusion purposes. This was an important point of scrutiny because one of our secondary outcome measures was the transfusion rate, and the threshold for the transfusion of autologous blood is very low. This threshold was demonstrated by our data on transfusion rates; in the Floseal group a mean of 0.83 units of autologous blood and 0.25 units of homologous blood were transfused, and in the control group a mean of 0.85 units of autologous blood and 0.34 units of homologous blood were transfused. There were no differences between the groups in terms of the rates of transfusion of autologous or homologous blood. In the Floseal group, a total of fifty-eight patients predonated, with twenty-six patients donating one unit of autologous blood and thirty-two patients donating two units. In the control group, a total of sixty-three patients predonated, with thirty patients donating one unit of autologous blood and thirty-three patients donating two units. This difference between the groups in terms of the frequency of autologous donation was not significant.
Patients who had donated autologous blood began with lower hemoglobin values than those who had not donated (11.5 compared with 12.3 g/dL; p = 0.006). However, by the second postoperative day, this difference was no longer significant (10.0 compared with 9.9 g/dL; p = 0.804).
In addition, information on preoperative anticoagulation use was also collected. In the entire study, forty-four patients were taking aspirin, twenty were taking warfarin, fourteen were taking Plavix (clopidogrel bisulfate; Bristol-Myers Squibb, New York, NY; Sanofi-Aventis, Bridgewater, New Jersey), and two patients were taking warfarin and Plavix. Patients who were taking Plavix preoperatively discontinued taking Plavix one week preoperatively and did not resume taking it until warfarin therapy was stopped. Patients who were taking aspirin for heart disease continued to take aspirin throughout the perioperative period. This variable was also well distributed between the Floseal and control groups (p = 0.652).
Complications and adverse events occurred infrequently. In the acute postoperative period, there were two cases of cellulitis (one in each group) that resolved with antibiotic treatment, two deep venous thromboses (one in each group) that were diagnosed with ultrasound and treated with full-dose anticoagulation, and one case of paralytic ileus (in the control group) that resolved without further complications after nonoperative treatment. At the time of the six-week follow-up, one patient in the Floseal group had died from a cause unrelated to surgery, two patients (one in each group) had suture abscesses with cellulitis that resolved with postoperative antibiotics, and four patients (two in each group) underwent knee manipulation under anesthesia to achieve improved knee motion. There were no significant associations between Floseal use and the occurrence of these adverse events.
No significant differences in demographic characteristics were noted between the Floseal and control groups. No significant differences in drain output or transfusion rates were noted between the Floseal and control groups (see Appendix). However, there were significant differences in hemoglobin on the first and second postoperative days. The differences were small (10.6 compared with 10.0 g/dL on the first postoperative day [p = 0.001] and 9.9 compared with 9.5 g/dL on the second postoperative day [p = 0.01]) and, although significant, most likely were not of any clinical importance. By the third postoperative day, there were no longer any significant differences in laboratory values (Table I).
The major strength of the present study is the design and methodology. We are aware of no reports regarding the use of a thrombin-based hemostatic agent in total knee arthroplasty; therefore, a pilot study was our initial step in the design of the present study. Floseal was initially used by the senior author (M.P.F.) after noting an anecdotal effect in reducing bleeding in a small cohort of hemophiliacs undergoing knee replacement. Our pilot study demonstrated a 25.8% reduction in drain output at twenty-four hours in the Floseal arm as compared with the control arm. This pilot study was conducted among patients undergoing bilateral total knee arthroplasty with the aim of using the contralateral side undergoing surgical treatment as an internal control. This result was seen only when Floseal was randomized to the first side being operated on, and the reason for this result was not clear. It was postulated that hemodynamic differences in patients undergoing bilateral total knee arthroplasty surgery under the setting of hypotensive anesthesia may have played a role.
Our pilot study was elemental in designing the current study, conducting our power calculation, and making improvements in the protocol. Our pilot study involved patients undergoing bilateral total knee arthroplasty performed by three different arthroplasty surgeons from one institution. The present study involved patients undergoing unilateral total knee arthroplasty performed by a single surgeon from one institution. The single-surgeon design ensured a uniform method of applying the hemostatic agent and surgical technique, and the single-institution design also ensured a uniform protocol for postoperative care that collectively minimized systematic variables for the study.
Infection is always a concern after total joint arthroplasty because of the catastrophic implications of such a complication on the outcome of elective arthroplasty surgery. In addition, thromboembolic disease is also an important concern. Although there is a concern about infection because of the use of a foreign agent, concern about thromboembolic disease is less obvious. It might be considered that the use of a thrombin-based agent, although intra-articular, may lead to some material permeating into the systemic circulation, leading to an alteration in the coagulation cascade and subsequent clot formation. Our study showed that there were no significant differences between the Floseal and control groups in terms of adverse outcomes or complications such as infection or thromboembolic disease, although these events were rare. However, with the complication rate after knee arthroplasty being so low, no definitive conclusions on the safety of Floseal with regard to increasing the prevalence of infection or thromboembolic events can be reported. This is an important consideration in evaluating the safety of Floseal use.
Our concern about the effect of preoperative donation of autologous blood on transfusion rates was based on the observation that the threshold for reinfusion was lower in patients who had autologous blood available. Indeed, patients who had donated autologous blood had higher transfusion rates than patients who were receiving homologous blood. However, the higher transfusion rates seen in those patients who had donated autologous blood preclude any major conclusions, for two reasons. First, patients who donated autologous units most likely had lower starting hemoglobin values and were more likely in need of transfusion. Second, transfusion with autologous blood is less likely to exhibit transfusion reactions, disease transmission, and complications associated with homologous blood transfusion. As a result, it is carried out with less hesitation and is generally better accepted by patients8. We recognize that this is a limitation of the study, and it may have been more fruitful to make the study population more homogeneous by ensuring that all study patients either donated autologous blood preoperatively or did not. We also attempted to have the same transfusion trigger for autologous and homologous blood, but in practice, this did not occur. However, analysis showed no difference in the frequency of autologous blood donation between the Floseal and control groups and no differences in overall transfusion rates (see Appendix).
Finally, a critical analysis of our data focused on variables that may have affected blood loss and patient coagulability. One such factor was a diagnosis of posttraumatic osteoarthritis. Due to changes in the soft tissues in the posttraumatic knee, we thought that it may lead to altered blood loss. However, diagnosis distribution was not significantly different between the groups. Preoperative anticoagulation regimens were also scrutinized. Although the preoperative INR was not different between the groups, we thought that other agents such as Plavix and aspirin may have altered our results. However, differences in the use of aspirin and Plavix between groups was not noted between the groups, and patients were evenly distributed between the groups.
In conclusion, Floseal use did not lead to decreased blood loss as measured on the basis of drain output during total knee arthroplasty in the present study. However, an effect on hemoglobin levels was observed on the first postoperative day (difference, 0.52 g/dL) and on the second postoperative day (difference, 0.36 g/dL). Floseal use appears safe and was not associated with any adverse outcomes or events; however, its ability to minimize blood loss and to decrease transfusion rates in patients undergoing total knee arthroplasty remains inconclusive.
A table showing demographic characteristics, intraoperative details, and transfusion data is available with the online version of this article as a data supplement at jbjs.org.
Birkmeyer
JD;
Goodnough
LT;
AuBuchon
JP;
Noordsij
PG;
Littenberg
B. The cost-effectiveness of preoperative autologous blood donation for total hip and knee replacement. Transfusion.
1993;33(
7):544-51.[PubMed][CrossRef]
Raut
VV;
Stone
MH;
Wroblewski
BM. Reduction of postoperative blood loss after press-fit condylar knee arthroplasty with use of a femoral intramedullary plug. J Bone Joint Surg Am.
1993;75(
9):1356-7.[PubMed]
Wang
GJ;
Hungerford
DS;
Savory
CG;
Rosenberg
AG;
Mont
MA;
Burks
SG;
Mayers
SL;
Spotnitz
WD. Use of fibrin sealant to reduce bloody drainage and hemoglobin loss after total knee arthroplasty: a brief note on a randomized prospective trial. J Bone Joint Surg Am.
2001;83-A(
10):1503-5.[PubMed]
Levy
O;
Martinowitz
U;
Oran
A;
Tauber
C;
Horoszowski
H. The use of fibrin tissue adhesive to reduce blood loss and the need for blood transfusion after total knee arthroplasty. A prospective, randomized, multicenter study. J Bone Joint Surg Am.
1999;81(
11):1580-8.[PubMed]
Oz
MC;
Cosgrove
DM
3rd;
Badduke
BR;
Hill
JD;
Flannery
MR;
Palumbo
R;
Topic
N. Controlled clinical trial of a novel hemostatic agent in cardiac surgery. The Fusion Matrix Study Group. Ann Thorac Surg.
2000;69(
5):1376-82.[PubMed][CrossRef]
Renkens
KL
Jr;
Payner
TD;
Leipzig
TJ;
Feuer
H;
Morone
MA;
Koers
JM;
Lawson
KJ;
Lentz
R;
Shuey
H
Jr;
Conaway
GL;
Andersson
GB;
An
HS;
Hickey
M;
Rondinone
JF;
Shargill
NS. A multicenter, prospective, randomized trial evaluating a new hemostatic agent for spinal surgery. Spine (Phila Pa 1976).
2001;26(
15):1645-50.[PubMed][CrossRef]
Rousou
J;
Levitsky
S;
Gonzalez-Lavin
L;
Cosgrove
D;
Magilligan
D;
Weldon
C;
Hiebert
C;
Hess
P;
Joyce
L;
Bergsland
J. Randomized clinical trial of fibrin sealant in patients undergoing resternotomy or reoperation after cardiac operations. A multicenter study. J Thorac Cardiovasc Surg.
1989;97(
2):194-203.[PubMed]
Schreiber
GB;
Busch
MP;
Kleinman
SH;
Korelitz
JJ. The risk of transfusion-transmitted viral infections. The Retrovirus Epidemiology Donor Study. N Engl J Med.
1996;334(
26):1685-90.[PubMed][CrossRef]