Introduction

A TKA is a surgery that alleviates pain and improves function but involves substantial blood loss that might result in allogeneic blood transfusion [11, 32, 43]. Allogeneic blood transfusion exposes patients to risks like immunologic reactions and transmission of diseases [15], and is associated with substantial cost to the healthcare system. Numerous strategies have been suggested to address this issue [32], with the use of antifibrinolytics to control perioperative blood loss a popular method [36]. Tranexamic acid, a synthetic analog of the amino acid lysine that inhibits fibrinolysis by competitively blocking the lysine binding sites of plasminogen [12, 42], has found widespread use. Tranexamic acid has been found to reduce blood loss in TKA and THA without an increase in thromboembolic complications [23]. While there is consensus regarding the efficacy of tranexamic acid in reducing blood loss, numerous issues persist regarding the best regimen and route of administration [30].

Tranexamic acid can be administered during the perioperative period either intravenously (IV), intraarticularly (IA), or by a combination of these measures. IV administration of tranexamic acid is a common practice with evidence favoring its effectiveness in reducing blood loss and the rate of allogeneic transfusion use in patients undergoing TKA [7, 22, 49]. However, concerns regarding complications like thromboembolic events after systemic administration [40] and the need for a prolonged effect to reduce postoperative blood loss [8] have led to interest in the topical administration of tranexamic acid. In addition, several studies [24, 26, 33, 38] have explored the possibility of using combined administration of tranexamic acid considering the advantages of both methods, but to our knowledge, only one published study compared all three methods of tranexamic acid administration as a strategy for reducing blood loss after TKA [48].

The purpose of our study was to compare the efficacy of tranexamic acid in patients undergoing TKA in terms of total blood loss and the allogeneic transfusion rate among the three routes of administration: IV, IA, and combined IV and IA. In addition, we wanted to compare these regimens in terms of venous thromboembolism (VTE) and the frequency of wound complications.

Patients and Methods

This study was a randomized controlled study. After obtaining institutional review board approval, a total of 444 patients scheduled for elective unilateral primary TKA between March 2014 and March 2015 were assessed for eligibility (Fig. 1). Exclusion criteria were: a diagnosis other than primary osteoarthritis, an acquired or congenital coagulopathy, patients receiving current anticoagulation therapy, preoperative hepatic or renal dysfunction or severe ischemic heart disease, and a history of thromboembolic disease. A total of 396 patients were enrolled and randomized to one of the four tranexamic acid groups (IV-only, IA-only, low-dose combined, and high-dose combined groups) using a computer-generated randomization table with a permutation block of six (99 in each group). All the enrolled patients provided informed consent for this study.

Fig. 1
figure 1

The flow diagram shows the number of patients assessed and included at each stage of the trial. IV = intravenous; IA = intraarticular.

Full size image

Administration of Tranexamic Acid

Patients assigned to the IV-only group were given intraoperative tranexamic acid with a weight-adjusted dosage (10 mg/kg) 30 minutes before tourniquet deflation; the same dose was repeated 3 hours after surgery. The calculated dose of tranexamic acid was mixed in 100 mL of normal saline and given as a slow IV injection. For patients assigned to the IA-only group, 2 g of tranexamic acid in 30 mL of normal saline was injected in the joint after closure of the retinaculum and quadriceps tendon but before subcutaneous closure. Patients in the low-dose combined group were given an IA injection of 1 g tranexamic acid in 30 mL of normal saline in addition to weight-adjusted intraoperative and postoperative IV administrations. Similarly, patients in the high-dose combined group were given an IA injection of 2 g tranexamic acid in 30 mL of normal saline in addition to intraoperative and postoperative IV administration of tranexamic acid.

Perioperative Management and Surgical Procedures

All patients were managed using the same perioperative protocols. Patients underwent a systematic medical clearance protocol that identifies patients with anemia, which is defined as hemoglobin less than 12 g/dL for females and less than 13 g/dL for males. Patients with a diagnosis of anemia were prescribed oral iron unless other underlying diseases were suspected. Patients who were taking acetylsalicylic acid, antiplatelet agents, or nonselective cyclooxygenase inhibitors were asked to discontinue the medications 7 to 10 days before surgery.

All surgical procedures were performed by one surgeon (TKK) using the standard medial parapatellar approach with application of the aforementioned blood conservation protocol. All patients were given spinal anesthesia. A pneumatic tourniquet was used on all patients and was inflated with a safe margin of approximately 150 mm Hg above preinflation systolic blood pressure. Before implant fixation, the femoral canal was plugged with a bone block and all prostheses were fixed with cement. The patella was resurfaced in all patients. After completing prosthesis fixation, the tourniquet was deflated and arterial bleeding was cauterized. A closed suction drain was placed subcutaneously with the intention to reduce blood loss and wound complication risk [44]. A compressive dressing then was applied with a CryoCuff (Howmedicare, Seoul, Korea) and knee immobilizer during the first 24 hours. The drain was removed approximately 24 hours after surgery. Allogeneic blood transfusion was given if the hemoglobin level fell below 7.0 mg/dL or if anemic symptoms such as dyspnea or tachycardia persisted even after volume replacement in patients with hemoglobin levels between 7.0 and 8.0 mg/dL. This transfusion trigger was chosen based on several previous studies that reported that the use of a restrictive hematologic algorithm with the hemoglobin level of 7.0 g/dL as the trigger does not increase adverse events related to postoperative anemia [14, 20, 39]. When transfusion was indicated based on those criteria, 1 unit of packed red blood cells was transfused at a time to increase the hemoglobin level to 8.0 g/dL. IV iron and erythropoietin were administered to patients without anemic symptoms but with a hemoglobin level between 7.0 and 8.0 mg/dL. Preoperative autologous blood donation was not performed in any patient during the study period. Our routine protocol of discharge was the fifth day after unilateral TKA during the study period.

All patients were given thromboprophylaxis according to our individualized approach protocol. Patients were assessed preoperatively for the risk of pulmonary embolism or bleeding and classified in four risk-stratified categories. A designated postoperative thromboprophylactic regimen was administered for each of the four categories: (1) standard risk for pulmonary embolism and bleeding: intermittent pneumatic compression during admission and aspirin 100 mg once a day for 5 weeks; (2) elevated risk for pulmonary embolism and standard risk for bleeding: intermittent pneumatic compression during admission and 10 mg rivaroxaban once a day for 10 days followed by 100 mg aspirin once a day for 25 days; (3) standard risk for pulmonary embolism and elevated risk for bleeding: intermittent pneumatic compression only during admission; and (4) elevated risk for pulmonary embolism and bleeding: intermittent pneumatic compression during admission and 100 mg aspirin once a day for 5 weeks. The risk-stratified categories were not considered in the randomization process to designate the four study groups.

Outcome Measures

The primary outcome measures were total blood loss, allogeneic transfusion rate, and incidence of VTE events. The total blood loss was calculated using a formula described in previous studies, which is based on patient blood volume and decrease in hemoglobin [18, 31, 37]. Blood volume was estimated according to the formula of Nadler et al. [37], which took body mass, height, and sex into account. Decrease in hemoglobin was defined as the difference between the preoperative hemoglobin level and the postoperative hemoglobin level on the fifth postoperative day [18], when the hemoglobin level was typically the lowest in our patient population or when the hemoglobin level was lowest before blood transfusion [29]. Patients who were given an allogeneic transfusion, which had been decided with strict transfusion criteria applied, were noted. All symptomatic VTE events were meticulously noted. Routine screening for deep vein thrombosis was not performed. CT angiography was done for patients with suspicious symptoms including pain, swelling, and tenderness in the thigh or calf and consultations with hematologists were sought to confirm the diagnosis and to plan treatment for diagnosed VTE events.

The secondary outcomes were postoperative decrease in hemoglobin, the proportions of patients with hemoglobin lower than 7.0, 8.0, and 9.0 g/dL, and the proportions of patients with wound complications, including deep periprosthetic joint infection.

To calculate the postoperative decrease in hemoglobin, we subtracted the hemoglobin level on postoperative Day 5 from the preoperative hemoglobin level unless patients were given an allogeneic transfusion. If patients received a transfusion, the lowest hemoglobin level before transfusion was used for the calculation. As certain cutoff values of the hemoglobin level are frequently used as transfusion triggers, the proportions of patients with hemoglobin levels less than 7.0, 8.0, and 9.0 g/dL were calculated. Wound complications, namely, oozing, superficial wound necrosis, and periprosthetic joint infection, were noted.

Statistical Analysis

The four groups were compared based on the primary and secondary outcome variables. ANOVA with post hoc Tukey’s honestly significant difference test was used for continuous variables (total blood loss, decrease in hemoglobin,), and the chi-square test was used for categorical variables (allogeneic transfusion incidence, proportions of patients with hemoglobin levels lower than 7.0, 8.0, and 9.0 g/dL, and the incidences of wound complications and VTE events). Statistical analyses were performed using SPSS Version 20.0 for Windows (IBM Corp, Armonk, NY, USA), and p values less than 0.05 were considered significant. The statistical power of the study was calculated by G-Power software (Version 3.1.9.2; Heinrich-Heine-Universität Düsseldorf, Dusseldorf, Germany) and found to be 0.98. The sample size was determined based on the difference in the primary outcome, which was calculated total blood loss. Based on preliminary data, the difference of calculated total blood loss between groups was 101 mL with a SD of 227 mL. Sample size estimation at an alpha of 0.05 and power of 80% indicated that 79 patients would be required for each arm. To compensate for the expected dropouts (15%), we needed 99 patients per group. The patients and a clinical investigator (SC) who prospectively collected all of the clinical information were unaware of the group identities until the final data analysis.

There were no differences across the groups in terms of demographic data or preoperative hematologic values (Table 1). Of the 444 patients assessed, 48 were excluded for the following reasons: 11 for diagnoses other than primary osteoarthritis, five owing to use of anticoagulation therapy, two with chronic renal disease, one with liver cirrhosis, and 29 who declined to participate (Fig. 1). A total of 396 patients (knees) undergoing unilateral TKA were enrolled. Twenty patients were excluded from final analysis owing to various reasons (eg, two patients because the dose was not given per protocol; 16 were lost to followup, and two pulled out the suction drain) leaving a total of 376 patients (knees) included in the final analysis: IV only (IV-only group, n = 93), IA only (IA-only group, n = 93), IV and IA injection of 1 g (low-dose combined group, n = 95), and IV and IA injection of 2 g (high-dose combined group, n = 95) tranexamic acid in 30 mL of normal saline.

Table 1 Demographic variables and preoperative hematologic values*
Full size table

Results

The combined use of IV and IA administration of tranexamic acid reduced blood loss more than IV use alone. IA use alone also resulted in improved blood loss reduction compared with IV use alone; these results were similar to those of the combined use groups (Table 2). The high-dose and low-dose combined groups and the IA-only group had less total blood loss (564 ± 242 mL, 642 ± 242 mL, and 633 ± 205 mL, respectively) than the IV-only group (764 ± 217 mL; mean differences = 199 mL, [95% CI, 116–283 mL], p < 0.001; 121 mL, [95% CI, 38–205 mL], p = 0.001]; 131 mL, [95% CI, 47–214 mL], p < 0.001), and no differences were found among the three groups that received IA tranexamic acid (p > 0.05) (Table 3). The high-dose and the low-dose combined groups and the IA-only group had smaller decreases in hemoglobin (2.0 ± 0.8 g/dL, 2.3 ± 0.8 g/dL, 2.4 ± 0.8 g/dL, respectively) than the IV-only group (2.9 ± 0.9 g/dL; mean differences = 0.9 g/dL, [95% CI, 0.5–1.2 g/dL], p < 0.001; 0.5 g/dL, [95% CI, 0.2–0.9 g/dL], p = 0.001; 0.5 g/dL, [95% CI, 0.2–0.8 g/dL], p < 0.001), and the high-dose combined group had a smaller decrease in hemoglobin than the low-dose combined group or the IA-alone group (mean differences = 0.3 g/dL, [95% CI, 0–0.6 g/dL], p = 0.047; 0.3 g/dL, [95% CI, 0–0.6 g/dL], p = 0.033) (Table 4). No patients in any of the four groups received an allogeneic transfusion and no differences were found among the groups in the proportions of patients with hemoglobin less than 7.0 g/dL, 8.0 g/dL, and 9.0 g/dL (Table 2).

Table 2 Efficacy of outcome variables*
Full size table
Table 3 Total blood loss comparisons
Full size table
Table 4 Hemoglobin drop comparisons
Full size table

The proportions of patients with symptomatic deep vein thrombosis, pulmonary embolism, and wound complication were extremely low in all study groups and no notable group differences were observed (Table 5). Only one patient in the IV-only group had a symptomatic pulmonary embolism, which was treated with low molecular weight heparin followed by warfarin without any residual sequelae. No other patients had symptomatic deep vein thrombosis or other thromboembolic events. Only minor wound complications (superficial wound necrosis and oozing) occurred, and these were uncommon in all groups (range, 3%–7%), with no between-group differences. No patients in any group had a deep periprosthetic joint infection develop during the surveillance period.

Table 5 Venous thromboembolism, wound complications, and deep infection*
Full size table

Discussion

The efficacy of tranexamic acid in reducing blood loss and allogeneic transfusion rate has been reported in numerous studies but there has been debate regarding the ideal route of administration [26, 30, 47, 48, 53]. Our study was conducted to compare the effects of tranexamic acid, in terms of total blood loss and allogeneic blood transfusion rate, when administered by IV, IA, or combined IV and IA routes. Several studies have compared the effects of topical and IV administration of tranexamic acid [8, 19, 35, 50], while others have compared the effects of combined versus IV or IA administration of tranexamic acid in TKA [24, 26, 33]. Our study is unique as we attempted to compare three different modalities of tranexamic acid administration (ie, IV alone, IA alone, combined IV and IA) at the same time. Our major finding is that IA tranexamic acid administration further reduces blood loss after TKA compared with IV use alone, and a combination of IV and IA administration will not provide additional advantages in reducing blood loss.

Our findings need to be interpreted in light of the following limitations. First, our study population had a high female predominance (92%), which might be a confounding factor. If there is sex-dependent difference in bleeding risk, it may restrict our findings from being extrapolated to a population with a different sex composition. However, only 11% of eligible patients were excluded from this study during the study period, and there was no difference in the sex ratio between the four groups. The gender composition of all patients in our TKA database is similar to the gender composition of the study cohort and also is similar to national results in South Korean patients [10, 28]. Second, we did not have a placebo group, as there is already ample published evidence in favor of tranexamic acid in this setting [1, 52, 54], and because to do so would be depriving the patients of its beneficial effects and subjecting them to the potential risks of increased blood loss and transfusion. Third, we did not include patients who underwent simultaneous or staged bilateral TKAs, and excluded patients presently taking anticoagulants as they might be expected to have greater blood loss; therefore, our findings may not to be applicable to these patients. Fourth, the vacuum drain was placed in the subcutaneous space, which is different from the typical intraarticular placement. This factor needs to be taken into account when comparing the amount of blood loss in our patients with those reported in other studies. Our practice is supported by a previous study that found that the use of a subcutaneous vacuum drain reduces blood loss without increasing wound complications [44]. Fifth, no blood analyses were performed to estimate serum tranexamic acid levels and no ultrasonography was done to assess asymptomatic deep vein thrombosis and pulmonary embolism as there is substantial evidence supporting the safety of tranexamic acid administration and a low incidence of complications [52]. Sixth, because we excluded a small number of patients who were receiving anticoagulants, we cannot rule out the possibility that combined administration is more effective in those patients, compared with IA or IV administration. However, as only five patients receiving anticoagulants were excluded from this study, the effect of this exclusion would not be substantial.

The major finding of our study was that the IA group and the high- and low-dose combined groups had less total blood loss than IV-only group, and there was no difference between the three groups that included IA administration in the regimen. Our findings suggest that an IA injection of tranexamic acid could be viewed as a more effective mode of administration of tranexamic acid, compared with IV administration. This finding is supported by a systematic review which found better efficacy of topical administration of tranexamic acid than IV use in an indirect comparison of the two methods [2]. However, several recent meta-analyses showed that IA and IV administration are comparable in effectiveness in decreasing blood loss and blood transfusion rate during TKA [9, 16, 17, 35]. We found less blood loss in the IA-only group than the IV-only group, but the difference was not sufficiently great to be clinically meaningful and the transfusion rate was not different. Therefore the efficacy of the two methods may be equivalent from this clinical standpoint. However, IA administration is a simple, surgeon-directed method, and it may have other benefits as well; a previous study showed that IA administration of tranexamic acid also reduced knee swelling in addition to reducing blood loss after TKA [25]. Another study showed that IA administration of tranexamic acid may improve early functional outcomes after TKA [46]. Theoretically, IA administration may lessen potential systemic adverse risk of tranexamic acid, as the topical application leads to 70% lower systemic absorption [51]. Considering the comparable (or even better) efficacy of IA administration, along with the potential benefits, we believe IA administration of tranexamic acid could be the preferred route in patients undergoing primary unilateral TKA.

Our findings also suggest that there is no compelling reason to combine IV administration with IA use, in terms of the blood loss and transfusion rate. Although the decrease in hemoglobin was lower in the high-dose combined group than the IA-only group, the difference was only 0.4 g/dL. This difference did not result in a lower transfusion rate or total blood loss. Enhanced efficacy of combined administration was reported in several studies [24, 26, 38]. However, there seems to be little published evidence comparing IA with combined administration of tranexamic acid [33, 48]. A recent study in which 120 patients were randomized to a topical, combined, or control group showed that the regimen used in the combined group was more effective than that used in the topical group in lowering the postoperative decrease in hemoglobin (1.9 g/dL vs 2.4 g/dL) [33]. However, there were no differences in the mean calculated blood loss and transfusion rate, which agrees with our results. In another study which compared topical, IV, combined, and placebo groups, no additional advantage of combined administration over topical or IV administration was found [48]. Our findings are consistent with the studies that question the necessity of combination approaches to administration [33, 48]. Again, combined administration did not show any remarkable advantage over IA administration of tranexamic acid alone in patients undergoing unilateral TKA.

Although our findings suggest that combined administration of tranexamic acid is not more beneficial than IA use alone in patients undergoing primary unilateral TKA, it may be considered in patients undergoing simultaneous bilateral TKAs. Kim et al. [29] reported that the decrease in hemoglobin in patients having simultaneous TKAs (5.1 g/dL) was much larger than that in patients having a unilateral TKA (3.8 g/dL) without tranexamic acid administration. Therefore the combined use of tranexamic acid may be helpful in reducing the decrease in hemoglobin and transfusion rate in this clinical scenario, although further studies will be required.

Tranexamic acid, as an antifibrinolytic agent, causes concern in terms of a possible increase in VTE complications after TKA. Although it is established that tranexamic acid does not inhibit fibrinolytic activity in the vein wall [4], its effect on a disrupted endothelium is unknown. We found no symptomatic deep vein thrombosis and only one pulmonary embolism among all of our study patients. Our findings are consistent with those of a systematic review [30] which found that no previous studies have reported an increased incidence of symptomatic deep vein thrombosis or pulmonary embolism with the use of tranexamic acid in TKA at any dose, timing, or route of administration [3, 5, 6, 13, 18, 21, 25, 27, 34, 41, 45, 51, 55]. The safety of this drug also is supported by several meta-analyses [1, 52, 54]. However, we did not routinely screen patients for deep vein thrombosis or pulmonary embolism unless the patient presented with clinical symptoms, and we also did not perform any specific tests looking for adverse effects related to the use of tranexamic acid. Therefore, we cannot exclude the possibility that the use of tranexamic acid increases the incidence of subclinical adverse events. Although we did not observe a difference in symptomatic thromboembolic events, our study was far too small to make safety claims on this point; future meta-analyses will need to aggregate the results of many randomized trials to evaluate the safety of these regimens.

IA tranexamic acid administration further reduces blood loss after unilateral TKA compared with IV use alone, regardless of its combination with IV administration. In this study, no additional effect of IV combination was found in further reducing blood loss compared with IA alone. IA administration is a more-effective mode of administration of tranexamic acid than systemic use for this purpose, and combined IV administration need not be considered additionally for primary unilateral TKA.