There is the potential for serious bias by indication in observational studies of therapeutic interventions. In our case, the relationship between the motivation for TXA use and subsequent bleeding-related outcomes might have confounded any relationship between the actual effect of the drug and the outcomes of interest. In order to minimize the effect of differences between patients who did and did not receive TXA and to enhance causal inference, we created a series of IPT weights. IPTW has been shown to effectively balance identifiable risk factors between treatment groups32. The IPT weight was the inverse of the propensity score for patients receiving TXA, and the inverse of 1 minus the propensity score for patients not receiving TXA. The propensity score was the probability of receiving TXA treatment based on patient and hospital-related variables. Stabilized weights were used33. The adequacy of this approach for addressing confounding was checked by comparing standardized differences between patients receiving versus not receiving TXA in the weighted and unweighted samples34.
Because of the hierarchical nature of the data, we addressed hospital-level clustering (covariance) in the propensity score model and surgeon-level clustering in the outcome model35. The propensity model predicting TXA use was fitted with a mixed-effects logistic regression model with a random effect for hospital. A different propensity model was fitted to derive weights for the VTE outcome analysis, because this analysis excluded patients using anticoagulants before admission. Also, a hospital-level variable reflecting the frequency of VTE screening was added.
All outcome models were mixed-effects models with a random effect for surgeon, weighted using IPTW, and included the following surgical-level variables as covariates: general anesthesia, surgical approach, bilateral surgery, and type of VTE prophylaxis used. A linear model was fitted for the drop in hemoglobin. Logistic models were used for transfusion, readmission, and cardiovascular events. A negative binomial model was fitted for LOS. For VTE, a Cox proportional hazard model, with treatment with aspirin and/or anticoagulant agent included as a time-dependent covariate, was used. In the transfusion analysis, we included a covariate for the calendar time period (6-month intervals) in which the surgery was performed to account for quality-improvement initiatives during the study period that might have affected transfusion practices. The effect estimate in the LOS analysis was expressed as an incident rate ratio (IRR), comparing LOS in the TXA group with LOS in the no-TXA group. SAS software (version 9.4; SAS Institute) was used for all data analyses.
A power analysis was performed using PASS 13 software (NCSS). TXA was used in approximately 51% of the hip cases and 48% of the knee cases. Cardiovascular events within 7 days of surgery occurred in approximately 0.8% of the hip cases and 0.6% of the knee cases in which TXA was not administered. With 11,489 hip cases and 23,236 knee cases, and an r2 of approximately 0.03 for the association of the other surgical-level variables and the use of TXA, this study had 80% power to show a significant adverse cardiovascular effect of TXA if the true odds ratios (ORs) were at least 1.69 and 1.55 for total hip arthroplasty and total knee arthroplasty, respectively. Using the same logic, and given that the incidence of VTE within 90 days among patients not given TXA in this registry is approximately 0.8% in hip arthroplasty and 1.8% in knee arthroplasty, this study would have a power of 80% to show a significant adverse VTE effect if the true ORs were at least 1.58 and 1.30, respectively.
The frequencies of baseline patient and surgical-level variables for hip and knee arthroplasty are shown in Tables I and II, respectively. Standardized differences before and after sample weighting with IPTW are shown in Table III. The closer the standardized differences are to 0, the less impact the risk factors have on the outcomes, making the calculated TXA effect less confounded. It has been suggested that a difference of <0.1 provides acceptable risk adjustment between groups36. Similarly, good risk adjustment or balance was achieved for the VTE data set.
In primary elective knee arthroplasty, perioperative TXA use was associated with a smaller drop in hemoglobin (mean difference = −0.68 g/dL; 95% CI = −0.64 to −0.71 g/dL; p < 0.0001) and decreased odds of transfusion (OR = 0.26; 95% CI = 0.21 to 0.31; p < 0.0001). This beneficial outcome associated with TXA was accompanied by a slightly decreased relative LOS (IRR = 0.93; 95% CI = 0.92 to 0.95; p < 0.0001) and decreased risk of VTE within 90 days after surgery (HR = 0.56; 95% CI = 0.42 to 0.73; p < 0.0001). We found no evidence of increased odds of readmission (OR = 0.90; 95% CI = 0.79 to 1.04; p = 0.1511) or cardiovascular outcomes (OR = 1.12; 95% CI = 0.74 to 1.71; p = 0.5715).
We found a significant reduction in the risk of transfusion with the use of TXA in primary total hip and total knee replacement surgery in the clinical practice setting. Hip and knee patients were three-fourths and one-fourth as likely, respectively, to receive transfusions if TXA was administered. Additionally, TXA was associated with a reduction in the drop in hemoglobin by an average of 0.6 g/dL per case. These findings are consistent with those of several randomized controlled clinical trials and meta-analyses of the use of TXA in arthroplasty patients16,19-25.
However, clinical trials have inclusion and exclusion criteria that may not represent all contemporary patients undergoing total joint replacement in the United States. Rare thrombogenic events such as myocardial infarction, transient ischemic attack, or stroke are hard to detect in clinical trials because of the relatively small number of cases. For example, Sukeik et al. performed a meta-analysis of 11 randomized controlled trials investigating the use of TXA in total hip arthroplasty23; 10 of the studies had <50 cases and the other had 100, for a total of 505 cases. They reported that there was no significant difference in deep-vein thrombosis (DVT) or pulmonary embolism (PE) between TXA and no-TXA groups, but they did not report on the absolute numbers of DVTs, and there were only 3 cases with PE. Aguilera et al. randomized 150 patients to intravenous or topical administration of TXA or to placebo15. Exclusion criteria were an allergy to TXA, a history of coagulopathy or VTE events, previous bypass surgery, use of anticoagulant or contraceptive treatment, and a cardiovascular prosthesis. They did not attempt to identify thromboembolic complications because they excluded cases “with possible TXA related complications.” Our study was designed specifically to have sufficient statistical power to test the relationship of TXA and these untoward events. A study such as ours provides reassurance that the results of randomized controlled trials are relevant and generalizable to the broader, unselected population.
We showed a significantly decreased risk of VTE within 90 days of total knee arthroplasty associated with TXA, and we did not find an increase in VTE events in total hip arthroplasty. This is consistent with the findings of 1 study involving a large administrative database and 2 meta-analyses in which no increased risks of DVT or PE were identified37-39. Our finding of the association of TXA with a significantly decreased risk of VTE in total knee arthroplasty is preliminary and may be due to residual confounding not accounted for in the propensity model. As demonstrated in Table II, patients with more comorbidities may not be treated with TXA as often and may be at higher risk for VTE at baseline, biasing the results. Alternately, a possible physiologic explanation for this observation might be that decreased bleeding in and around the knee leads to decreased swelling and less venous obstruction. The key clinical point is that TXA was not associated with an increased risk.
We found no association between TXA and an increased risk of myocardial infarction, stroke, or transient ischemic attack in either hip or knee arthroplasty. In a meta-analysis of surgery trials, Ker et al. also found no difference in myocardial infarction or stroke events associated with the use of TXA38. In our study, 58 (0.5%) of the hip cases and 108 (0.5%) of the knee cases experienced cardiovascular events within 7 days of surgery. Because the outcomes are rare, the CIs around our estimates are wide enough to allow for a 50% to 70% increase in cardiovascular outcomes associated with treatment. We also found a decreased risk of readmission among the hip patients treated with TXA, along with strong evidence of no increase in LOS. Those outcomes were not investigated in other large studies, to our knowledge.
The current study had several limitations. We did not assess the effect of intravenous versus topical administration or TXA dosage on outcomes. We focused on the potential impact of TXA use on thrombogenic events and readmissions but did not explore relationships with all adverse events associated with total joint replacement. Patients were not randomly assigned to treatment, and bias by indication is a possibility. We also found that surgeons changed practice over time: the percentage of surgeons using TXA in <10% of their cases went from 76% in the first 3 months to 30% in the last 3 months of this study, while the percentage of surgeons using it in >90% of their cases went from 12% to 41% (Fig. 3). The change in TXA utilization at the surgeon level creates the potential for patient selection bias, resulting in an exaggerated benefit attributable to TXA. Our strategies to delineate the TXA effect more clearly given this potential selection bias were to balance patient-level variables using IPTW based on the patients’ propensity to receive TXA given their characteristics and to account for the potential correlation of outcomes among patients cared for in the same hospital or by the same surgeon. We were reassured because we were able to achieve excellent balance on covariates, indicating that we were able to adequately control for identified confounders.
Even with these efforts, there could still be other unidentified confounders responsible for the apparent lack of thrombogenic events with TXA use. For example, a surgeon may choose to withhold TXA from patients with cardiovascular risk factors that we did not measure, which in turn could obscure a true underlying increased risk of myocardial infarction or stroke. Although some level of uncertainty remains, techniques exist to estimate how strong the impact of such unmeasured confounders would need to be to overturn our conclusions40. Given reasonable assumptions about the differential prevalence of the unmeasured confounder, this seems unlikely. For example, a confounder would have to have an RR of close to 6 to reverse our conclusions about cardiovascular outcomes.
In summary, our registry-based study adds to the evidence that the use of TXA in primary total hip and knee arthroplasty patients is a safe and effective adjunct to reduce blood loss and the need for transfusion. Additionally, TXA was associated with reduced readmissions among total hip arthroplasty patients and reduced risk of VTE among total knee arthroplasty patients.
NOTE: The authors thank Steven Coon, MPH, for his assistance with data analysis and preparation of this manuscript. Support for the quality improvement work of MARCQI is provided by Blue Cross Blue Shield of Michigan and Blue Care Network as part of the BCBSM Value Partnerships program. No funding is provided for research related to MARCQI. Although BCBSM and MARCQI work collaboratively, the opinions, beliefs, and viewpoints expressed by the authors do not necessarily reflect those of BCBSM or any of its employees. To learn more about Value Partnerships, see http://http://www.valuepartnerships.com.
Investigation performed at the MARCQI Coordinating Center, University of Michigan, Ann Arbor, Michigan
1. Bolton-Maggs PH, Cohen H. Serious Hazards of Transfusion (SHOT) haemovigilance and progress is improving transfusion safety. Br J Haematol. 2013 ;163(3):303–14. Epub 2013 Sep 14.
2. Bernard AC, Davenport DL, Chang PK, Vaughan TB, Zwischenberger JB. Intraoperative transfusion of 1 U to 2 U packed red blood cells is associated with increased 30-day mortality, surgical-site infection, pneumonia, and sepsis in general surgery patients. J Am Coll Surg. 2009 ;208(5):931–7: 937.e1-2, discussion :938-9. Epub 2009 Mar 26.
3. Corwin HL, Gettinger A, Pearl RG, Fink MP, Levy MM, Abraham E, MacIntyre NR, Shabot MM, Duh MS, Shapiro MJ. The CRIT Study: Anemia and blood transfusion in the critically ill—current clinical practice in the United States. Crit Care Med. 2004 ;32(1):39–52.
4. Frisch NB, Wessell NM, Charters MA, Yu S, Jeffries JJ, Silverton CD. Predictors and complications of blood transfusion in total hip and knee arthroplasty. J Arthroplasty. 2014 ;29(9)(Suppl):189–92. Epub 2014 May 24.
5. Hart A, Khalil JA, Carli A, Huk O, Zukor D, Antoniou J. Blood transfusion in primary total hip and knee arthroplasty. Incidence, risk factors, and thirty-day complication rates. J Bone Joint Surg Am. 2014 ;96(23):1945–51.
6. Hill GE, Frawley WH, Griffith KE, Forestner JE, Minei JP. Allogeneic blood transfusion increases the risk of postoperative bacterial infection: a meta-analysis. J Trauma. 2003 ;54(5):908–14.
7. Innerhofer P, Klingler A, Klimmer C, Fries D, Nussbaumer W. Risk for postoperative infection after transfusion of white blood cell-filtered allogeneic or autologous blood components in orthopedic patients undergoing primary arthroplasty. Transfusion. 2005 ;45(1):103–10.
8. Koch CG, Li L, Duncan AI, Mihaljevic T, Loop FD, Starr NJ, Blackstone EH. Transfusion in coronary artery bypass grafting is associated with reduced long-term survival. Ann Thorac Surg. 2006 ;81(5):1650–7.
9. McAlister FA, Clark HD, Wells PS, Laupacis A. Perioperative allogeneic blood transfusion does not cause adverse sequelae in patients with cancer: a meta-analysis of unconfounded studies. Br J Surg. 1998 ;85(2):171–8.
10. Weber WP, Zwahlen M, Reck S, Misteli H, Rosenthal R, Buser AS, Kaufmann M, Oertli D, Widmer AF, Marti WR. The association of preoperative anemia and perioperative allogeneic blood transfusion with the risk of surgical site infection. Transfusion. 2009 ;49(9):1964–70. Epub 2009 May 11.
11. Rohde JM, Dimcheff DE, Blumberg N, Saint S, Langa KM, Kuhn L, Hickner A, Rogers MA. Health care-associated infection after red blood cell transfusion: a systematic review and meta-analysis. JAMA. 2014 ;311(13):1317–26.
12. Harris RN, Moskal JT, Capps SG. Does tranexamic acid reduce blood transfusion cost for primary total hip arthroplasty? A case-control study. J Arthroplasty. 2015 ;30(2):192–5. Epub 2014 Sep 06.
13. Tuttle JR, Ritterman SA, Cassidy DB, Anazonwu WA, Froehlich JA, Rubin LE. Cost benefit analysis of topical tranexamic acid in primary total hip and knee arthroplasty. J Arthroplasty. 2014 ;29(8):1512–5. Epub 2014 Feb 03.
14. Yoshihara H, Yoneoka D. National trends in the utilization of blood transfusions in total hip and knee arthroplasty. J Arthroplasty. 2014 ;29(10):1932–7. Epub 2014 Apr 26.
15. Aguilera X, Martínez-Zapata MJ, Hinarejos P, Jordán M, Leal J, González JC, Monllau JC, Celaya F, Rodríguez-Arias A, Fernández JA, Pelfort X, Puig-Verdie Ll. Topical and intravenous tranexamic acid reduce blood loss compared to routine hemostasis in total knee arthroplasty: a multicenter, randomized, controlled trial. Arch Orthop Trauma Surg. 2015 ;135(7):1017–25. Epub 2015 May 07.
16. Alshryda S, Sukeik M, Sarda P, Blenkinsopp J, Haddad FS, Mason JM. A systematic review and meta-analysis of the topical administration of tranexamic acid in total hip and knee replacement. Bone Joint J. 2014 ;96-B(8):1005–15.
17. Charoencholvanich K, Siriwattanasakul P. Tranexamic acid reduces blood loss and blood transfusion after TKA: a prospective randomized controlled trial. Clin Orthop Relat Res. 2011 ;469(10):2874–80. Epub 2011 Apr 22.
18. Gomez-Barrena E, Ortega-Andreu M, Padilla-Eguiluz NG, Pérez-Chrzanowska H, Figueredo-Zalve R. Topical intra-articular compared with intravenous tranexamic acid to reduce blood loss in primary total knee replacement: a double-blind, randomized, controlled, noninferiority clinical trial. J Bone Joint Surg Am. 2014 ;96(23):1937–44.
19. Hiippala ST, Strid LJ, Wennerstrand MI, Arvela JV, Niemelä HM, Mäntylä SK, Kuisma RP, Ylinen JE. Tranexamic acid radically decreases blood loss and transfusions associated with total knee arthroplasty. Anesth Analg. 1997 ;84(4):839–44.
20. Huang F, Wu D, Ma G, Yin Z, Wang Q. The use of tranexamic acid to reduce blood loss and transfusion in major orthopedic surgery: a meta-analysis. J Surg Res. 2014 ;186(1):318–27. Epub 2013 Sep 13.
21. Husted H, Blønd L, Sonne-Holm S, Holm G, Jacobsen TW, Gebuhr P. Tranexamic acid reduces blood loss and blood transfusions in primary total hip arthroplasty: a prospective randomized double-blind study in 40 patients. Acta Orthop Scand. 2003 ;74(6):665–9.
22. Seo JG, Moon YW, Park SH, Kim SM, Ko KR. The comparative efficacies of intra-articular and IV tranexamic acid for reducing blood loss during total knee arthroplasty. Knee Surg Sports Traumatol Arthrosc. 2013 ;21(8):1869–74. Epub 2012 Jun 24.
23. Sukeik M, Alshryda S, Haddad FS, Mason JM. Systematic review and meta-analysis of the use of tranexamic acid in total hip replacement. J Bone Joint Surg Br. 2011 ;93(1):39–46.
24. Whiting DR, Gillette BP, Duncan C, Smith H, Pagnano MW, Sierra RJ. Preliminary results suggest tranexamic acid is safe and effective in arthroplasty patients with severe comorbidities. Clin Orthop Relat Res. 2014 ;472(1):66–72.
25. Wu Q, Zhang HA, Liu SL, Meng T, Zhou X, Wang P. Is tranexamic acid clinically effective and safe to prevent blood loss in total knee arthroplasty? A meta-analysis of 34 randomized controlled trials. Eur J Orthop Surg Traumatol. 2015 ;25(3):525–41. Epub 2014 Nov 28.
26. Bruce-Brand R, Dragomir R, Baker J, Harty J. Cerebrovascular infarction following bilateral total knee arthroplasty and tranexamic acid administration. Acta Orthop Belg. 2013 ;79(3):351–4.
27. Gupta PN, Mullamalla UR, Sabin P, Vellappan P. Acute MI in a young hypertensive woman: could it be due to tranexamic acid? BMJ Case Rep. 2013;2013. Epub 2013 May 27.
28. Sundström A, Seaman H, Kieler H, Alfredsson L. The risk of venous thromboembolism associated with the use of tranexamic acid and other drugs used to treat menorrhagia: a case-control study using the General Practice Research Database. BJOG. 2009 ;116(1):91–7. Epub 2008 Nov 11.
29. Hughes R, Hallstrom B, Cowen M, Igrisan R, Singal B, Share D. Michigan Arthroplasty Registry Collaborative Quality Initiative (MARCQI) as a model for regional registries in the United States. Orthop Res Rev. 2015 ;7:47–56.
30. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998 ;36(1):8–27.
31. Bilimoria KY, Chung J, Ju MH, Haut ER, Bentrem DJ, Ko CY, Baker DW. Evaluation of surveillance bias and the validity of the venous thromboembolism quality measure. JAMA. 2013 ;310(14):1482–9.
32. Lunceford JK, Davidian M. Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study. Stat Med. 2004 ;23(19):2937–60.
33. Xu S, Ross C, Raebel MA, Shetterly S, Blanchette C, Smith D. Use of stabilized inverse propensity scores as weights to directly estimate relative risk and its confidence intervals. Value Health. 2010 ;13(2):273–7. Epub 2009 Nov 12.
34. Austin PC, Mamdani MM, van Walraven C, Tu JV. Quantifying the impact of survivor treatment bias in observational studies. J Eval Clin Pract. 2006 ;12(6):601–12.
35. Li F, Zaslavsky AM, Landrum MB. Propensity score weighting with multilevel data. Stat Med. 2013 ;32(19):3373–87. Epub 2013 Mar 24.
36. Normand ST, Landrum MB, Guadagnoli E, Ayanian JZ, Ryan TJ, Cleary PD, McNeil BJ. Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: a matched analysis using propensity scores. J Clin Epidemiol. 2001 ;54(4):387–98.
37. Gandhi R, Evans HMK, Mahomed SR, Mahomed NN. Tranexamic acid and the reduction of blood loss in total knee and hip arthroplasty: a meta-analysis. BMC Res Notes. 2013;6:184. Epub 2013 May 07.
38. Ker K, Edwards P, Perel P, Shakur H, Roberts I. Effect of tranexamic acid on surgical bleeding: systematic review and cumulative meta-analysis. BMJ. 2012;344:e3054. Epub 2012 May 17.
39. Poeran J, Rasul R, Suzuki S, Danninger T, Mazumdar M, Opperer M, Boettner F, Memtsoudis SG. Tranexamic acid use and postoperative outcomes in patients undergoing total hip or knee arthroplasty in the United States: retrospective analysis of effectiveness and safety. BMJ. 2014;349:g4829. Epub 2014 Aug 12.
40. Vanderweele TJ, Arah OA. Bias formulas for sensitivity analysis of unmeasured confounding for general outcomes, treatments, and confounders. Epidemiology. 2011 ;22(1):42–52.