We report a systematic review and meta-analysis of published randomised controlled trials evaluating the efficacy of tranexamic acid (TXA) in reducing blood loss and transfusion in total hip replacement (THR). The data were evaluated using the generic evaluation tool designed by the Cochrane Bone, Joint and Muscle Trauma Group. We identified 11 clinical trials which were suitable for detailed extraction of data. There were no trials that used TXA in revision THR. A total of seven studies (comprising 350 patients) were eligible for the blood loss outcome data. The use of TXA reduced intra-operative blood loss by a mean of 104 ml (95% confidence interval (CI) −164 to −44, p = 0.0006, heterogeneity I2 0%), postoperative blood loss by a mean of 172 ml (95% CI −263 to −81, p = 0.0002, heterogeneity I2 63%) and total blood loss by a mean of 289 ml (95% CI −440 to −138, p < 0.0002, heterogeneity I2 54%).
TXA led to a significant reduction in the proportion of patients requiring allogeneic blood transfusion (risk difference −0.20, 95% CI −0.29 to −0.11, p < 0.00001, I2 15%). There were no significant differences in deep-vein thrombosis, pulmonary embolism, infection rates or other complications among the study groups.
In total hip replacement (THR) there can be considerable blood loss, requiring allogeneic blood transfusion. This carries significant risks of immunological reactions, transmission of disease, intravascular haemolysis, transfusion-induced coagulopathy, renal failure, admission to intensive care and even death.1–3
A variety of blood-conserving techniques have been developed to reduce blood loss and post-operative transfusion rates, including controlled hypotension, regional anaesthesia, autologous blood transfusion, intra-operative blood salvage, and the use of erythropoietin and anti-fibrinolytic agents.3,4
The antifibrinolytics include tranexamic acid (TXA), aprotinin and ε-aminocaproic acid (EACA), which have different mechanisms to inhibit the dissolution of blood clots. They have been used successfully to stop bleeding after dental extraction, tonsillectomy, prostate surgery, heavy menstrual bleeding, cardiac surgery, and in patients with haemophilia. Numerous studies have investigated their efficacy in reducing blood loss and transfusion requirements in THR.5–15 However, some of these studies have been criticised for poor design, low power, inconclusive results and short follow-up. This has prejudiced the use of these potentially valuable agents in orthopaedic practice.
The purpose of this systematic review and meta-analysis was to investigate the evidence for the efficacy of TXA in reducing blood loss and transfusion requirements after THR, and the evidence for any increase in complication rates.
Materials and Methods
This review was based on Cochrane methodology for conducting systematic reviews and meta-analyses.16
Study selection criteria.
We only included randomised controlled trials in this study. The participants were adults who had undergone THR, regardless of the type or size of prosthesis used. We considered analysing cemented and uncemented implants, but there were insufficient data to allow this.
The intervention considered was the administration of intravenous (IV) TXA. Studies that involved oral or intramuscular treatment were excluded. Only those studies with a control group were considered. The control group received a placebo, another antifibrinolytic agent or no treatment. Those trials involving a significant intervention, such as heparin prophylaxis, which was not given to both groups, were also excluded.
The primary outcome measure was intra-operative, post-operative and total blood loss. The secondary outcome measures were the proportion of patients who had allogeneic blood transfusion, the amount of blood units transfused per patient, functional hip outcome measures (e.g. Oxford hip score17), general quality of life outcome measures (e.g. Short-form (SF) 12,18 SF-3619 or EUROQOL20) and complications.
Search methods for identification of studies.
The following exploded Medical Subject Headings (MeSH) terms were used for the initial literature search: ‘Antifibrinolytics’, ‘Tranexamic acid’, ‘Cyklokapron’, ‘Aprotinin’, ‘Trasylol’, ‘Epsilon Aminocaproic Acid’ and ‘Amicar’. The Medline search was then refined to include clinical trials and randomised, controlled trials in adult humans. The results were cross-checked with other databases, namely EMBASE, the Cochrane Controlled Trials Register, HealthSTAR and CINAHL, Google and Google Scholar for trials of antifibrinolytics and THR, published in any language between 1966 and December 2008.
The bibliographies of the retrieved trials and other relevant publications, including reviews and meta-analyses, were examined for additional articles. We also searched the following websites to identify additional unpublished and ongoing studies: Current Controlled Trials (www.controlled-trials.com), Centre Watch (www.centerwatch.com), TrialsCentral (www.trialscentral.org/ClinicalTrials.aspx) and the United Kingdom National Research Register (www.nrr.nhs.uk). We also searched The Journal of Bone and Joint Surgery - British Volume, The Journal of Bone and Joint Surgery - American Volume, and the Journal of the American Academy of Orthopaedic Surgeons for any relevant publications.
Data collection and analysis.
Two authors (MS, SA) independently applied the search strategy to selected references from these databases. The titles and abstracts of the articles were reviewed independently. When there was a doubt, the full article was retrieved for further scrutiny. The two authors independently assessed each full study report to see whether it met the review’s inclusion criteria, and authors were contacted for more information and clarification of data as necessary. Any disagreement was discussed with the senior authors (FSH, JMM), and when consensus could not be reached, that study was excluded.
In order to assess the methodological quality of included studies the review authors (MS, SA) used a modification of the generic evaluation tool used by the Cochrane Bone, Joint and Muscle Trauma Group21 (Table I⇓). Any disagreement was resolved by the senior authors. Although the total quality assessment score was reported for each study, it was not used to weight the studies in the meta-analysis.
A data extraction form was designed and agreed by the authors, and a pilot test of five articles was performed to ensure its consistency. Initially, two authors (MS, SA) independently extracted the data, which was later reviewed jointly to produce agreed accurate data. Disagreements were resolved by consensus or consultation with the senior authors. The authors of individual trials were contacted directly to provide further information when necessary.
The Review Manager Database (RevMan version 5.0, The Cochrane Collaboration 2008) was used to analyse the selected studies. Continuous data for each arm in a particular study were expressed as mean and sd, and the treatment effect as mean differences. Dichotomous data for each arm in a particular study were expressed as proportions or risks, and the treatment effect as risk differences. Missing data were sought from the authors. When this was not possible or data were missing through loss to follow-up, intention-to-treat principles were used.
Statistical heterogeneity was assessed using the value of I2 and the result of the chi-squared test. A p-value of < 0.1 and an I2 value > 50% were considered suggestive of statistical heterogeneity, prompting a random effects modelling estimate. Otherwise, a fixed effects approach was used. Conversely, a non-significant chi-squared test result (a p-value ≥ 0.1 and an I2 value ≤ 50%) only suggested that there was no evidence of heterogeneity: it did not imply that there was necessarily homogeneity, as there may have been insufficient power to be able to detect heterogeneity.
When the data allowed, we performed subgroup analysis of the trials according to the type of anaesthesia, dose regimen and timing of TXA delivery, transfusion trigger and administration of low-molecular-weight heparin (LMWH).
Following application, refinement and reapplication of the literature search strategy, 66 studies were identified as potentially relevant. However, subsequent scrutiny of the abstracts led to the exclusion of 46 of these. The full publications were obtained for the remainder. These were assessed and a further nine were excluded for various reasons according to the inclusion and exclusion criteria.
The remaining 11 randomised, controlled trials were included in the meta-analysis5–15 (Table II⇓). The majority were small studies with between 36 and 100 participants. However, they were relatively well designed and the quality assessment score was high in most of them, with a mode of 24, the highest possible score and a range of 14 to 24.7–10,12,13,15 Only one study had a score less than 20.6
The trials performed were all for primary THR, and osteoarthritis was the most common diagnosis. The patients’ characteristics were comparable within each study group, including the pre-operative haemoglobin and haematocrit levels. A placebo (normal saline) was given in nine studies, and only two studies used controls who did not receive any treatment.6,15 Different doses, ranging from 10 mg/kg to 30 mg/kg, and modes of TXA delivery were used. A single IV bolus given pre-operatively was used in five studies.5,8,10,13,15 Three studies used repeated boluses6,7,9 and the remainder a prolonged infusion.11,12,14 All studies except those of Ido et al,6 Yamasaki et al15 and Garneti and Field8 used LMWH for prophylaxis against deep-vein thrombosis (DVT). The former two did not use chemical prophylaxis and the latter used mechanical prophylaxis only. In seven studies, screening for DVT relied on clinical assessment followed by ultrasound examination or venography in suspected cases.6,7,9,10,12,13,15 In four studies an ultrasound examination or venogram was performed routinely.5,8,11,14 Six studies stated a transfusion trigger, which was related to a fall in either haemoglobin or haematocrit levels.5,7,11–14 Drains were used in the majority of studies with four using two drains (subfascial and subcutaneous), three using one drain (subfascial) and one using three drains (intra-articular, subfascial and subcutaneous). Post-operative drainage was measured between 24 and 48 hours, when the drains were removed in most cases. There were seven trials that used regional anaesthesia alone,5,8,11–15 one did not mention the type of anaesthetic6 and the remaining three used a combination of general and regional anaesthesia.7,9,10 The prostheses were cemented in seven trials;6–11,13 cementless in one,15 hybrid in another5 and a combination of the above in the remaining two.12,14 The amount of blood units transfused per patient, functional hip measures and general quality of life outcome measures were not analysed, as there were insufficient data. No studies reported on mortality. A cost comparison between TXA and blood products was analysed in five studies5,7,10,12,13 and favoured the use of TXA.
Effects of interventions.
Intra-operative: Five studies5,11,13–15 with a total of 259 patients were eligible for this outcome. These trials randomised 127 patients to receive TXA and 132 patients as controls. The use of TXA significantly reduced intra-operative blood loss, measured by weighing sponges and suction drainage, by a mean of 104.4 ml (95% CI −164.2 to −44.7, p < 0.001, heterogeneity I2 = 0%) (Fig. 1⇓).
Seven studies5,6,8,11,13–15 including 350 patients were eligible for this outcome. The use of TXA significantly reduced post-operative blood loss, measured by drainage volume, by a mean of 172.4 ml (95% CI −263.6 to −81.3, p < 0.001). However, there was significant heterogeneity among the studies included (I2 = 63%) (Fig. 2⇓).
Blood transfusion rate.
Seven studies5,7–10,12,13 with a total of 346 patients were eligible to measure the effect of TXA on blood transfusion after THR. The use of TXA led to a 20% reduction in the proportion of patients who required blood transfusion (risk difference −0.20, 95% CI −0.29 to −0.11, p < 0.001, I2 = 15%, Fig. 4⇓).
ten trials5,7–15 with a total of 464 patients reported on DVT, 229 of whom received TXA. There was no significant difference among the study groups in relation to a higher risk of developing DVT (Mantel-Haenszel, p = 0.46).
Pulmonary embolism: There were three reported events of pulmonary embolism in the 11 trials,8,10 two in the TXA group and one in the control group. However, there was no statistical significance in the risk of developing pulmonary embolism among the groups (Mantel-Haenszel, p = 0.76).
There were two reported infections in each group. Johansson et al13 noted two superficial wound infections in the control group which were successfully treated with antibiotics without further complications. In the TXA group, one patient developed a superficial infection7 and the other a deep infection, which required re-operation five months after the primary procedure10 (Mantel-Haenszel, p = 0.97).
In this section we compared all other reported adverse events among the groups, such as formation of haematoma, secretion from drain sites and systemic complications. Systemic complications included one case of a brief respiratory arrest related to delay in volume replacement from early post-operative blood loss which was treated without complications,14 a case of post-operative delirium from unrecognised alcohol withdrawal, necessitating re-intubation,14 a patient with urinary retention who was treated with a suprapubic catheter,7 a patient with transient chest discomfort and fever four weeks after operation which resolved spontaneously,9 and a patient in the control group who had nausea on administration of the saline.10 In the TXA group, one patient developed transient dyspnoea on the third post-operative day,7 another had pyelonephritis one month post-operatively7 and one developed slight hemiparesis 58 days postoperatively. Her initial CT brain scan was normal, but a further CT scan three months later, after another episode, showed signs of older infarctions in the right hemisphere.10 Overall, the results showed that the use of TXA was associated with fewer such complications. However, this did not reach a statistically significant level (Mantel-Haenszel, p = 0.16).
There are several issues related to quality control in conducting a meta-analysis, in particular study selection and the homogeneity of the studies. A systematic review and meta-analysis with homogeneity is regarded as level Ia evidence.22 Hence, our study focused on the use of TXA and THR as a single group to reduce heterogeneity related to other antifibrinolytic agents and other types of operation.
TXA has gained popularity in reducing peri-operative blood loss, particularly after the publication of a trial in high-risk cardiac surgery.23 It is cheaper and safer than aprotinin and much more potent than EACA, with overall good penetration into major joints.24,25
The most significant result of this review is the consistency of TXA in reducing blood loss and transfusion rates after THR in the majority of studies. However, with the exception of intra-operative blood loss, there was significant heterogeneity among the studies when post-operative and total blood losses were evaluated. The variations that may have accounted for such heterogeneity include the difference in sample sizes, the variations in patient characteristics, inclusion and exclusion criteria, differences in management protocols and logistics between treating centres, and different strategies for measuring outcomes.
When the weighted effect estimates for reduction of postoperative blood loss are plotted against the total amount of TXA administered, the best-fitting line showed a direct relationship between the amount of TXA given and the effect estimate. However, no such relationship was detected in the reduction of intra-operative and total blood loss or blood transfusion rates (Fig. 5⇓).
Regardless of the amount of blood loss, TXA reduced allogeneic blood transfusion requirements by 20% with no significant heterogeneity (I2 = 15%). Furthermore, subgroup analysis showed that this positive effect persisted regardless of whether TXA was delivered in single or multiple doses, whether the patient had LMWH, regional anaesthesia or a cemented THR, and whether a transfusion protocol existed (Table III⇓). Interestingly, multiple-dose regimen and the use of transfusion protocols produced a more profound risk reduction of blood transfusion (−0.25 vs −0.17 and −0.24 vs −0.14, respectively). Similar effects were also evident on subgroup analysis of intra-operative, post-operative and total blood loss outcomes.
Two studies in the meta-analysis did not support the routine use of TXA in THR. Benoni et al9 performed a randomised double-blinded study on 39 THRs where TXA was given at the end of the operation and three hours later in 20 patients and an equivalent protocol of normal saline was given to 19 patients. The results showed that TXA did not significantly reduce intra- or post-operative blood loss (550 ml vs 500 ml and 440 ml vs 450 ml, respectively). However, these authors and the results of our meta-analysis attribute these findings to the fact that TXA was given too late to show a significant effect, as most of the other studies administered TXA pre-operatively, with overall good results. Conversely, Garneti and Field8 randomised 50 patients to either a single dose of TXA or a similar volume of saline as a pre-operative bolus. The results were in favour of the placebo group, with a mean post-operative blood loss of 353 ml (sd 311) vs 411 ml (sd 220) for the TXA group and 1340 ml (sd 665) vs 1443 ml (sd 809) total blood loss for each group, respectively. The reasons for these differences from other studies are unclear. Patient numbers, duration of the operation, dose, time of administration of the drug in relation to the surgery and number of times the drug was administered were proposed as possible contributing factors by the authors. Also, more patients in the TXA group (64%) required transfusion than in the placebo group (56%). However, this could be attributed to the different transfusion strategies of the anaesthetists, one of whom transfused most patients unless they were young and healthy, and the fact that there was no defined transfusion protocol, which could have been a source of bias.
Other meta-analyses examined the relationship between TXA and blood loss and/or transfusion after THR. Zufferey et al26 analysed the effect of intravenous anti-fibrinolytics, including TXA, aprotinin and EACA, on blood transfusion in various orthopaedic procedures. Only studies with a transfusion protocol were included, which resulted in five studies for analysis. Their results on blood transfusion were similar to ours, especially when considering a multiple-dose regimen for TXA. However, the effect of TXA on blood loss was only briefly discussed as part of ‘other efficacy endpoints’, and was evaluated as a single group under ‘peri-operative blood loss’, with no clear definition of the blood loss.
Ho and Ismail27 studied the effect of TXA on reducing blood transfusion after total hip and knee arthroplasties. However, most of the studies were on knee arthroplasties. Blood loss was again collectively defined as ‘peri-operative blood loss’, despite including the results of total blood loss as well as post-operative blood loss under this definition.
Gill and Rosenstein28 performed another meta-analysis where the effects of TXA and aprotinin in THR were evaluated. Only six trials were eligible for this meta-analysis, which showed that TXA was effective in reducing intra-operative and total blood loss, but failed to produce statistically significant results for reducing the proportion of patients who received allogeneic blood transfusions. However, although they contradict our findings and those of others, a detailed analysis of how those results were achieved is not included.
Kagoma et al29 reviewed the evidence of using TXA, EACA and aprotinin on total blood loss and transfusion rates in knee and hip arthroplasties. Despite similar trends to our study, all three antifibrinolytics were either analysed as a single group or their effects evaluated for hip and knee arthroplasties combined. Also, where sd was not reported in the original studies, an estimate was calculated using different formulae in all of the above meta-analyses, leading to discrepancies in the values of sds and eventually the outcomes measured. Therefore, we only included studies where an sd was already reported when measuring continuous outcomes.
There are several strengths in this meta-analysis. First, we conducted a thorough literature search of randomised, controlled trials, including publications in any language as well as unpublished abstracts. Secondly, the quality assessment score for most of the studies included was high, which contributes to the strength of point estimates and conclusions drawn from the meta-analysis. Thirdly, the meta-analysis showed favourable outcomes when using TXA, especially in reducing intra-operative blood loss and blood transfusion requirements, with no significant heterogeneity among the studies or an increased risk of thromboembolic events. Subgroup analysis showed similar findings with better outcomes when using a multiple-dose regimen and a transfusion protocol, which is consistent with evidence from previous studies.26,27 Fourth, analyses of cost-effectiveness in five studies were all in favour of using TXA over blood transfusion products.
The limitations of this meta-analysis included insufficient data to support the analysis of functional outcome scores or quality of life outcome measures as originally planned. Furthermore, with the exception of post-operative blood loss, our findings do not support the assumption that a higher dose of TXA necessarily reduces intra-operative and total blood loss or blood transfusion rates. Also, trials included in our study were designed to assess the efficacy and safety of TXA in primary THR, where high-risk factors were excluded. These included patients with a history of cardiovascular disease, thromboembolic events, bleeding diathesis, allergy to TXA, renal failure with creatinine level > 250 μmol/l, pregnancy, and those on warfarin or a therapeutic dose of LMWH. Therefore, no definite conclusions regarding the safety of TXA can be derived from our meta-analysis in relation to revision hip arthroplasty or in high-risk patients, where TXA is contraindicated.
In summary, we conclude that TXA significantly reduced intra-operative blood loss and transfusion requirements after primary THR, with no significant increase in complication rates. Favourable results were also suggested for post-operative and total blood loss, but with significant heterogeneity, which necessitates careful interpretation as to the effect of TXA in this context. Also, future randomised trials of sufficient power should be designed to examine the efficacy and safety of TXA in revision hip surgery and to compare this with other methods of blood conservation.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
- Received April 7, 2010.
- Accepted September 23, 2010.
- © 2011 British Editorial Society of Bone and Joint Surgery