The efficacy of tranexamic acid in primary anatomic and reverse total shoulder arthroplasty: A systematic review and meta-analysis of level I randomized controlled trials

Abstract

Purpose

The purpose of this study was to systematically review the available level I evidence regarding the impact of tranexamic acid (TXA) on early postoperative outcomes in patients undergoing anatomic total shoulder arthroplasty (TSA) and reverse total shoulder arthroplasty (RTSA).

Methods

A systematic review of the literature through April 2023 was performed to identify level I RCTs examining the use of TXA at the time of primary TSA or RTSA.

Results

Among 5 included studies, a total of 435 patients (219 TXA, 216 control) were identified. Superior hematologic outcomes were observed among the TXA cohort, including lower 24-hour drain output (MD −112.70 mL: p < 0.001), lower pre- to postoperative change in hemoglobin (MD: −0.68 g/dL, p < 0.001), and less total perioperative blood loss (MD: −249.56 mL, p < 0.001). Postoperative Visual Analog Scale for pain (VAS-pain) scores were lower in the TXA group, but not significantly (MD: −0.46, p = 0.17). Postoperative blood transfusion was required in 3/219 TXA patients (1.4%) and 7/216 control patients (3.2%) (RR: 0.40, p = 0.16).

Conclusion

Perioperative TXA reduces drain output and total blood loss without increasing the risk of adverse events. TXA was not shown to decrease postoperative transfusion rates when compared to placebo controls.

Level of Evidence

Level I, meta-analysis.

Introduction

In recent years, tranexamic acid (TXA) has been increasingly used to reduce blood loss associated with total joint arthroplasty.^1,2 Following arthroplasty procedures, perioperative blood loss resulting in allogenic blood transfusion has been associated with adverse outcomes ranging from periprosthetic infections to myocardial infarction.^3,4 TXA is a lysine analog that reversibly binds plasminogen and prevents its cleavage into plasmin, thus stabilizing clots and promoting hemostasis.^5,6

The efficacy and safety profile of perioperative TXA administration is well established in the orthopedic literature, particularly in trauma and elective total hip and knee arthroplasty.^7–14 In hip and knee arthroplasty, there is a robust body of literature which demonstrates that TXA administration may reduce transfusion rates by up to 50% without a significant increase in associated complications.^15–19 These findings are independent of dosing, timing, type of anesthesia, and/or preoperative hemoglobin levels.^11–14,20 As allogenic blood transfusions have been associated with increased risks of transfusion reactions, infection, length of stay (LOS), and costs, recent clinical recommendations have endorsed the use of TXA in other areas of orthopedic surgery.^11–13,20 In shoulder arthroplasty numerous prospective and retrospective studies have begun to establish a similar safety/efficacy profile for perioperative TXA administration.^5,21–24

Although prior studies have attempted to synthesize the available data regarding the efficacy of TXA in total shoulder arthroplasty (TSA), all have incorporated level II and level III studies, and none have exclusively analyzed level I RCTs.^25–28 As defined by the Centre for Evidence-Based Medicine (CEBM), level I evidence refers to individual RCTs (with narrow confidence interval (CI) and >80% follow-up), level II includes individual cohort studies (including low quality RCT <80% follow-up) and level III being individual case-control studies. ²⁹ There is significant value in limiting meta-analysis to level I studies because these provide the strongest evidence appropriate for treatment recommendation. Therefore, the purpose of this study was to systematically review and synthesize the available level I evidence regarding the impact of TXA on the reduction of blood loss, transfusion rate, and postoperative outcomes including operative time, hospital LOS and adverse events in patients undergoing TSA.

Methods

Search strategy and study selection

A systematic review of the literature was performed according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). ³⁰ A comprehensive search of the PubMed Central, MEDLINE, Embase, and Scopus databases from inception through April 2023 was performed. References were manually reviewed for the addition of further studies. The search strategy was determined a priori and performed using the following search string: (“TXA” OR “tranexamic acid”) AND “shoulder.”

Inclusion and exclusion criteria

Studies were included if they met the following inclusion criteria: (a) regarded clinical outcomes of TXA use in patients undergoing primary anatomic (TSA) or reverse total shoulder arthroplasty (RTSA), (b) were of level I evidence, (c) included a non-TXA control, and (d) had available text written in the English language. Studies were excluded if they (a) were systematic reviews/meta-analyses, letters to the editor, elemental analyses, or case reports, and (b) were of level II–V evidence.

Data extraction

Data were extracted from each individual study and organized into a spreadsheet for further analysis. Extracted data included study design, level of evidence, sample size, patient demographics, surgical details, and select clinical outcomes. Clinical outcomes of interest included drain output, change in hemoglobin, total blood loss, postoperative Visual Analog Scale for pain (VAS-pain) scores, operative time, hospital LOS, and complications. Within individual studies total perioperative blood loss was estimated using change in hemoglobin and total blood volume estimated via patient height, weight, and sex. This method is thought to be more accurate than others including estimated blood loss (EBL). ²²

Risk of bias assessment

Risk of bias within individual studies was assessed using Version 2.0 of the Cochrane Risk of Bias Tool (RoB 2). ³¹ The tool is specifically designed for the evaluation of RCTs, and includes a series of validated questions to screen for features relevant to risk of bias. Each domain was scored as low risk of bias, high risk of bias, or some concern of bias, and the results presented in risk of bias plots. ³²

Statistical analysis

The meta-analysis software Review Manager 5.4.1 (Cochrane Collaboration, London, UK) was used for quantitative synthesis. Each outcome measure was depicted in a forest plot detailing the mean difference (MD) for continuous variables and risk ratio (RR) for dichotomous variables. For continuous data, if the mean or standard deviation (SD) were not reported, the median, range, interquartile range (IQR), and sample size were used to estimate these measures according to the methodology recommended by the Cochrane Handbook for Systematic Reviews of Interventions ³³ and Hozo et al. ³⁴ Heterogeneity was assessed via the I² and Chi² (χ²) indices. To account for heterogeneity between studies, a random-effects meta-analysis was used due to its ability to weight studies more equally. Continuous variables were reported as mean ± SD or range, whereas categorical variables were reported as frequencies with percentages unless otherwise indicated. A p-value ≤ 0.05 was used to determine statistical significance.

Results

Study selection

The initial database search identified a total of 397 studies (Figure 1). After the removal of duplicates, the abstract and title of 174 articles were screened, of which 161 did not meet inclusion criteria, leaving 13 studies for full-text review. Following full-text review, a total of 5 studies met eligibility criteria and were included in the final analysis.^5,21–24 The reasons for exclusion were incorrect procedure (n = 3), incorrect level of evidence (n = 2), no control without TXA administration (n = 2), and incorrect language (n = 1).

Figure 1.

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram demonstrating study selection process. Abbreviations: TXA, tranexamic acid.

Study characteristics and patient demographics

All included studies were level I, double-blinded, placebo-controlled RCTs published between 2015 and 2021 (Table 1). Three studies were conducted in the United States,^5,22,24 while one was conducted in Austria, ²³ and another in Australia. ²¹ Four of the included studies were published within the Journal of Shoulder and Elbow Surgery (JSES) Family of Journals,^5,21,22,24 and one in The Bone & Joint Journal. ²³ A total of 435 patients were identified, of which 219 received TXA and 216 did not. Four studies included patients undergoing both anatomic TSA and RTSA,^21,21–23 while one included patients undergoing RTSA only. ²⁴ The weighted mean age of the TXA cohort was 68 years (range, 67–72), and 122 of the patients were female (55.7%). The weighted mean age of the control cohort was 68 years (range, 65–73), and 116 of the patients (53.7%) were female. The total dose of TXA was 2 g in three studies,^5,21,23 1 g in one study, ²² and weight based (20 mg/kg total) in one study. ²⁴ The route of administration was intravenous in four studies^21–24 and topical in one. ⁵ Control patients received a normal saline placebo in all studies. A transfusion threshold of 7 g/dL was utilized in four studies,^5,21,22,24 while a threshold of 8 g/dL was used in one. ²³ Additionally, among three studies, transfusions were initiated in patients with a Hb < 9 g/dL and symptoms of acute blood loss anemia (i.e. hypotension, tachycardia, dizziness).^5,21,24 One study used a threshold of 8 g/dL in symptomatic patients, ²³ while one did not specify a threshold. ²²

Table 1.

Study Characteristics of the Five Included RCTs.

Study	Journal of Publication	Study Design	LoE	Region	Total Patients	TXA						Control						Exclusion Criteria
						Patients	Female	Age (y)	Total Dose	Route	Timing	Patients (N)	Female	Age (y)	Control Treatment	Route	Total Dose
Gillespie 2015	JSES	RCT	I	USA	111	56	33 (58.9%)	67.59	2g	Topical	1g intraoperative, 1g just prior to wound closure	55	29 (52.7%)	66.45	Normal saline	Topical	100 mL	Allergy to TXA Cardiac disease History of bleeding or metabolic disorder History of joint infection History of liver disease History of thromboemolic event Preoperrative Hb <11.5 or Hct <35% Revision surgery Severe joint deformity Unwilling to accept blood transfusion
Pauzenberger 2017	The Bone & Joint Journal	RCT	I	Austria	54	27	7 (25.9%)	70.3 ± 9.3	2g	IV	1g 30 min prior to incision, 1g during wound closure	27	9 (33.3%)	71.3 ± 7.9	Normal saline	IV	200 mL (100 mL x 2)	Age < 40 Allergy to TXA Anticoagulant use Hematologic disorders Hemiarthroplasty History of coagulopathy History of thromboemolic event Pregnancy or breastfeeding Refusal to participate Retinopathy Revision surgery Severe comorbidities Unwilling to accept blood transfusion
Vara 2017	JSES	RCT	I	USA	102	53	33 (62.3%)	67 ± 9	20 mg/kg	IV	10 mg/kg 60 min prior to surgery, 10 mg/kg during wound closure	49	27 (55.1%)	66 ± 9	Normal saline	IV	Equal to TXA arm, volume not specified	Acute proximal humerus fracture Allergy to TXA Concomitant procedures History of coagulopathy History of thromboembolic event Major comorbidities Minors Preoperative Hb <11 g/dL in women, < 12 g/dL in men Refusal to give written consent Unwilling to accept blood transfusion
Cvetanovich 2018	JSES Open Access	RCT	I	USA	108	52	29 (55.8%)	67.7 ± 10.9	1g	IV	1g 10 min prior to incision	56	28 (50.0%)	65.2 ± 9.2	Normal saline	IV	10 mL	Acquired disturbances of color vision Acute humerus fracture Allergy to TXA Anticoagulant use within 5 days of surgery History of cardiac stent placement History of coagulopathy History of myocardial infarction within 6 months History of prior open shoulder surgery History of renal impairment History of thromboembolic event Pregnancy or breastfeeding Revision arthroplasty Unwilling to accept blood transfusion
Cunningham 2021	JSES	RCT	I	Australia	60	31	20 (64.5%)	72 ± 8	2g	IV	2g immediately prior to skin incision	29	23 (73.9%)	73 ± 9	Normal saline	IV	NR	Allergy to TXA History of coagulopathy History of seizure Preoperative use of anticoagulants Revision arthroplasty Unwilling to accept blood transfusion
TOTAL					435	219	122 (55.7%)	68 (67–72)				216	116 (53.7%)	68 (65–73)

Abbreviation: RCTs, randomized controlled trials.

Risk of bias

Study assessment with the Cochrane RoB 2 Tool revealed an overall low risk of bias among the majority of included studies (Figure 2). There was “some concern” for bias in the study published by Gillespie et al. due to two patients in the placebo group and five patients in the treatment group not having outcomes data. ⁵

Figure 2.

Traffic-light and summary plots for visualization of Cochrane Collaboration Risk of Bias Version 2.0 (RoB 2) grading for RCTs. Abbreviations: RCTs, randomized controlled trials.

Drain output

Drain output at 24 hours/postoperative day (POD) 1 was reported in four studies.^5,21,23,24 The weighted mean drain output was 107.3 mL (range, 50–153) in the TXA group and 219.9 mL (range, 170–294) in the control cohort (MD [95% CI], −112.70 [-152.75, −72.66]; I²= 63%; p < 0.001; Figure 3A).

Figure 3.

Forest plots demonstrating differences in (A) 24 hours/POD 1 drain output, (B) pre- to postoperative change in hemoglobin, and (C) total blood loss between TXA and control cohorts. Abbreviations: POD, postoperative day; TXA, tranexamic acid; VAS, Visual Analog Scale.

Change in hemoglobin

Pre- to postoperative change in hemoglobin levels was reported in three studies.^5,21,24 The weighted mean decrease was 2.4 g/dL (range, 1.7–3.3) in patients that received TXA and 3.1 g/dL (range, 2.3–3.9) in those that received saline placebo (MD [95% CI], −0.68 [-0.97, −0.39]; I²= 0%; p < 0.001; Figure 3B).

Total blood loss

Total blood loss was reported in four studies^21–24 and was significantly lower in the TXA cohort (912.3 mL; range, 550–1122.4) compared to the control cohort (1161.9 mL; range, 740–1472.6) (MD [95% CI], −249.56 [-347.60, −151.52]; I²= 16%; p < 0.001; Figure 3C).

Postoperative pain

Postoperative VAS-pain scores were reported in two studies.^21,23 Overall, the weighted mean score was lower in the TXA group (2.2; range, 1.3–4) compared to the control cohort (2.7; range, 2–4), however, the difference was not significant (MD [95% CI], −0.46 [-1.11, −0.19]; I²= 32%; p = 0.17; Figure 4A).

Figure 4.

Forest plots demonstrating differences in (A) postoperative VAS-pain scores, (B) operative time, and (C) hospital LOS between TXA and control cohorts. Abbreviations: LOS, length of stay; TXA, tranexamic acid; VAS, Visual Analog Scale.

Operative time

Among the three studies reporting operative time,^21,22,24 there was no significant difference between the study group (98.7 min; range, 83–101.1) and the control group (101.6 min; range, 86–104) (MD [95% CI], −2.90 [-8.13, 2.33]; I²= 0%; p = 0.28; Figure 4B).

Hospital length of stay

Hospital LOS was analyzed in three studies^21,22,24 and was 2.4 days (range, 1.8–5.1) in the TXA group, and 2.5 days (range, 1.8–4.8) in the control group (MD [95% CI], −0.06 [-0.32, 0.20]; I²= 0%; p = 0.65; Figure 4C).

Hematoma formation

Three studies reported the incidence of postoperative hematoma.^21,23,24 Utilizing a study-specific scale based on visibility and pain, Pauzenberger et al. found a significantly greater incidence of overall hematoma formation in the placebo group (n = 16, 59.3%) compared to the TXA group (n = 6, 25.9%; p = 0.027). ²³ Cunningham et al. reported one hematoma in the control cohort and zero in the study group. ²¹ Vara et al. reported an incidence of zero in both cohorts. ²⁴ Overall, pooled analysis revealed that the risk of postoperative hematoma was significantly lower in the TXA cohort (RR [95% CI], 0.37 [0.18,0.79]; I² = 0%; p = 0.010).

Blood transfusion

Among the five included studies, 3/219 TXA patients (1.4%) and 7/216 control patients (3.2%) required a postoperative blood transfusion.^5,21–24 Pooled analysis revealed no significant difference in transfusion rate between cohorts (RR [95% CI], 0.40 [0.11, 1.45]; p = 0.16).

Complications

Adverse events were reported in 1 TXA patient (0.46%) and 3 control patients (1.4%).^5,21–24 Vara et al. reported a syncopal fall in a TXA patient that was effectively treated with fluids. ²⁴ The suspected etiology of the fall was not articulated, and the authors did not state whether there was a potential association with TXA administration. Among control patients, Vara et al. reported one skin allergy and one non-ST-segment elevation myocardial infraction, ²⁴ and Cvetanovich et al. reported one deep vein thrombosis (DVT). ²² No other thromboembolic events were reported in either cohort in any study. Pooled analysis revealed no difference in the risk of postoperative complications between groups (RR [95% CI], −0.01 [-0.03, 0.02]; I² = 0%; p = 0.58).

Discussion

This systematic review and meta-analysis aimed to synthesize the existing level I evidence regarding the utility of TXA in patients undergoing primary reverse or anatomic TSA. This study found that, compared to patients receiving normal saline placebo, patients receiving TXA experienced superior hematologic outcomes including lower 24-hour drain output (MD 112.7 mL), lower change in hemoglobin levels (MD 0.68 g/dL), and less total blood loss (MD 249.6 mL). TXA administration did not prolong operative times (MD 2.9 min), increase hospital LOS (MD 0.06 days), or increase the risk of adverse events (0.46% TXA, 1.4% control). The incidence of postoperative anemia requiring blood transfusion was similar in both the TXA (1.4%) and control cohorts (3.2%).

The primary goal of TXA administration in TSA is to reduce total blood loss and decrease requirements for allogenic blood transfusion. These findings demonstrate that the use of TXA in shoulder arthroplasty resulted in lower 24 hours drain output, lower change in hemoglobin levels, and less total blood loss. There was a trend towards lower rates of transfusion with TXA, but this was not statistically significant. These findings may be due to a slight difference in blood transfusion threshold criteria and small sample size.

TXA dosing and timing is another variable that can affect blood loss. In this study, there were some differences between route of administration (IV vs. topical), and amount given (weight based vs. standardized dosing). Among patients undergoing RTSA and receiving TXA, in a prospective RCT, Yoon et al. found no difference in hemoglobin drop, total drain volume, and total blood loss when comparing IV, topical, and combined administration routes. ³⁵ Despite finding no difference in overall efficacy, the authors stated that because topical administration could reduce the risk of complications due to lower peripheral blood TXA concentration, that the topical route may be preferable. Similarly, when comparing oral and IV administration, Gao et al. found no difference in efficacy and concluded that use of the oral form should be encouraged to minimize overall healthcare expenditure. ¹⁹ Prior studies in the hip and knee literature have examined route of administration and dosing and found no differences between IV, topical, single doses, or multiple doses.^1,36 The effect of timing of TXA on blood loss is more variable. In a large meta-analysis in the hip and knee literature, pre-incision TXA was favored on primary TKA, but there were no differences found during THA.^1,2 Further study is needed to determine optimal timing and dosing of TXA during shoulder arthroplasty procedures.

Several of the included studies assessed relevant non-hematologic outcomes in their comparison groups. These outcomes included postoperative VAS-pain scores, operative time, and postoperative LOS. Though average postoperative pain was lower in the TXA group, the difference was found to be insignificant compared to the control cohort. With regards to operative time and postoperative LOS, it is not surprising that no difference was found between the comparison groups. The only conceivable ways that TXA could impact operative time was if bleeding secondarily impacted intraoperative visualization requiring added effort to achieve hemostasis or if the process of administering TXA added to the procedure time. Cunningham et al., reported on subjective intraoperative visibility and found no significant difference between groups. ²¹ It is more likely that TXA's impact is more cumulative and decreases blood loss over time. Additionally, with no difference in postoperative transfusion rates or complications in the reported studies, it would be unlikely for TXA to significantly affect the length patients’ hospital stays.

Despite the perceived blood-sparing effects of TXA, owing to its antifibrinolytic nature, there remains some concern that TXA administration may promote a hypercoagulable state and increase the risk of postoperative thromboembolic events or death. ³⁷ Because postoperative thromboembolic events remain relatively rare, the cost and logistical limitations of performing large scale RCTs may preclude individual studies from detecting relevant differences in risk. Nonetheless, despite meta-analysis of five level I RCTs, this study again demonstrated that among patients undergoing primary TSA there is no higher incidence of adverse events or thrombosis with the use of TXA. This is consistent with the more well-studied hip and knee arthroplasty literature, and strengthens our understanding of TXA saftey. ³⁸

This study is not without limitation. First, the included studies were limited to those published in the English language and indexed in the PubMed Central, MEDLINE, Embase, and Scopus databases. Second, the number of included RCTs was relatively small which may have resulted in this study being underpowered to detect differences in the rates of rare outcomes including blood transfusion and thromboembolic events. Despite similarities in demographic characteristics and eligibility criteria across included studies, it should be noted that the dose, timing, and route of TXA administration, specific surgical technique, and drain management protocols were not standardized across the included studies. Further, our understanding of the risk profile associated with TXA use in TSA is limited by a lack of standardization regarding the definition of postoperative complications. Although prior studies have demonstrated equivalence regarding the efficacy of topical and IV TXA administration,^35,39 these differences added a layer of heterogeneity and represent important areas for continued investigation. Despite these limitations, this study adds significant value to the existing literature as it is the only level I study synthesizing the available data regarding the efficacy of perioperative TXA administration in patients undergoing TSA. All included studies were double-blinded, placebo-controlled RCTs, the gold standard for evaluating the safety and efficacy of therapeutic interventions.

Conclusion

Among patients undergoing primary TSA, compared with placebo control, perioperative TXA reduces drain output and total blood loss without increasing the risk of adverse events or loss of operative time. However, TXA was not shown to decrease postoperative transfusion rates when compared to placebo controls.