- Email: [email protected]
Comparative effectiveness of epsilon-aminocaproic acid and tranexamic acid on postoperative bleeding following cardiac surgery during a national medication shortage
Journal of Clinical Anesthesia (2016) 35, 516–523
Original Contribution
Comparative effectiveness of epsilon-aminocaproic acid and tranexamic acid on postoperative bleeding following cardiac surgery during a national medication shortage☆ Kevin P. Blaine MD, MPH (Resident Physician)⁎, Christopher Press MD (Resident Physician), Ken Lau MD (Resident Physician)1 , Jan Sliwa MD (Resident Physician)2 , Vidya K. Rao MD (Clinical Assistant Professor), Charles Hill MD (Clinical Assistant Professor and Associated Critical Care Director) Department of Anesthesia, Pain, and Perioperative Medicine, Stanford University Medical Center, 300 Pasteur Dr, H3580, Stanford, CA 97305 Received 15 December 2015; accepted 14 August 2016
Keywords: ε-aminocaproic acid; Tranexamic acid; Blood conservation; Perioperative bleeding; Cardiac surgery
Abstract Study objective: The aim of this study was to compare the effectiveness of epsilon-aminocaproic acid (εACA) and tranexamic acid (TXA) in contemporary clinical practice during a national medication shortage. Design: A retrospective cohort study. Setting: The study was performed in all consecutive cardiac surgery patients (n = 128) admitted to the cardiac-surgical intensive care unit after surgery at a single academic center immediately before and during a national medication shortage. Measurements: Demographic, clinical, and outcomes data were compared by descriptive statistics using χ2 and t test. Surgical drainage and transfusions were compared by multivariate linear regression for patients receiving εACA before the shortage and TXA during the shortage. Main results: In multivariate analysis, no statistical difference was found for surgical drain output (OR 1.10, CI 0.97–1.26, P = .460) or red blood cell transfusion requirement (OR 1.79, CI 0.79–2.73, P = .176). Patients receiving εACA were more likely to receive rescue hemostatic medications (OR 1.62, CI 1.02–2.55, P = .041). Conclusions: Substitution of εACA with TXA during a national medication shortage produced equivalent postoperative bleeding and red cell transfusions, although patients receiving εACA were more likely to require supplemental hemostatic agents. Published by Elsevier Inc.
☆
State of Disclosure: No authors have financial conflicts of interest related to this work. ⁎ Corresponding author at: Critical Care Medicine Department, National Institutes of Health Clinical Center, 9000 Rockville Pike. 2C145. Tel.: +1 240 762 1873. E-mail addresses: [email protected] (K.P. Blaine). 1 Present affiliation: Anesthesia Services Medical Group, 3626 Ruffin Rd., San Diego, CA 92123. 2 Present affiliation: Department of Anesthesiology, Brigham and Women's Hospital, 75 Francis St, Boston, MA 02115.
http://dx.doi.org/10.1016/j.jclinane.2016.08.037 0952-8180/Published by Elsevier Inc.
Comparative effectiveness of eACA and TXA
1. Introduction Nearly 1 in 5 blood products transfusions in the United States occurs in the setting of cardiac surgery [1]. Blood conservation techniques are used to minimize perioperative transfusions during and after cardiac surgery [2]. These techniques include intraoperative cell-salvage devices, improved surgical techniques, topical hemostatic agents, and conservative transfusion thresholds [3]. Anti-fibrinolytic agents are often used as a pharmacological adjunct to maintain clot stability and minimize bleeding. Contemporary antifibrinolytic agents include aprotinin, εaminocaproic acid (εACA), and tranexamic acid (TXA). In 2007 to 2008, the most thoroughly studied agent, aprotinin, was removed from the market in the US, UK, Canada, and European Union due to safety concerns [4,5]. Since 2012, aprotinin has been returned for use in Canada and Europe following a systematic review of multiple studies concluding with a favorable the risk/benefit ratio [6]. Aprotinin remains unavailable in the United States. The alternative antifibrinolytic agents, εACA and TXA, have been compared in small trials published between 1994 and 2001 [7-11] which did not establish the clinical superiority of either agent. The Blood Conservation Using Antifibrinolytics in a Randomized Trial (BART) was the only large study to include TXA (n = 770) and εACA (n = 768) arms, and demonstrated equivocal outcomes in mortality for a composite ‘massive transfusion’ endpoint (including greater than 1.5 L blood loss within 8 hours, N8 U red blood cells (RBCs) transfused, or surgical re-exploration for hemorrhage) [12]. The results are difficult to interpret as major hemorrhage following surgery is multifactorial and often involves compromised surgical hemostasis. It is unrealistic to expect an antifibrinolytic agent alone to prevent perioperative catastrophe. A recent Cochrane review reported insufficient evidence (particularly for εACA) to recommend either lysine analog over the other [13]. Studies included in that analysis were limited by small size, remote publication dates, and surgical indications often limited to primary coronary revascularization surgery (for which many patients today might receive percutaneous coronary angiography). Contemporary antifibrinolytic agents are used for a variety of cardiac procedures beyond coronary revascularization, such as valve surgery or ventricular assist device placement, for which they have not been well studied. Finally, since the 1990s when many of these trials were conducted, thresholds for transfusions have lowered such that today small differences in blood may be more relevant towards the decisions to transfuse than they would have been in an era where most patients would receive blood regardless. While some clinicians favor a particular agent, at present, the use of εACA and TXA remains in a state of equipoise. In 2012–2013, the United States experienced a national shortage of εACA following FDA-mandated upgrades to the manufacturing process (W. Fridich, Luitpold Pharmaceuticals, personal communication, August 8, 2015). The blood
517 conservation practice at our institution included eACA exclusively until stores of this medication were exhausted on March 30, 2013. There was an immediate substitution of TXA for all cardiac procedures from April 1, 2013, until supplies were replenished on June 4, 2013. The shortage presented an opportunity to compare retrospectively the effectiveness of εACA and TXA on postoperative bleeding in the ICU following cardiac surgery under contemporary surgical conditions and with minimal patient risk.
2. Materials and methods 2.1. Study population The study population includes all consecutive patients undergoing cardiac surgery with cardiopulmonary bypass between February 1 and June 3, 2013, who received either TXA or εACA. The protocol was approved by the Research Compliance Office and Institutional Review Board of Stanford University School of Medicine with a waiver of informed consent. Patients were included for the first cardiac or proximal aortic surgery of the hospitalization. Patients were required to have a planned post-operative admission to the cardiothoracic surgical intensive care unit with surgical drains in situ. Patients who died or required surgical exploration within 8 hours were excluded, as antifibrinolytic agents alone were unlikely to prevent these complications. Prior to April 1, 2013, all cardiac surgery patients at our institution received εACA (loading dose 10-15 mg/kg over 1015 min, 2-3 mg/kg bolus into the bypass priming solution, then 1-2 mg/kg per hour infusion for 6 hours). During the shortage, all patients received TXA (loading dose 10-15 mg/kg over 1015 min, 2-2.5 mg/kg bolus in the bypass priming solution, then 1-2 mg/kg per hour for 6 hours) until εACA was again available on June 4, 2013. Allotment of either εACA or TXA was assigned based on medication availability and was independent of clinician judgment.
2.2. Data collection Patient charts were accessed for the following data: patient demographics including age on admission, sex, and race; preoperative co-morbidities including hypertension, diabetes mellitus, chronic pulmonary obstructive disease, chronic renal insufficiency (baseline Cr N1.5 g/dL), congestive heart failure with New York Heart Association functional class; ASA Physical Classification; unstable pre-operative coronary artery disease (as evidenced by electrography, echocardiography, cardiac catheterization, perfusion scan, or unstable angina defined by chest pain at rest with electrocardiogram or echocardiographic changes); pre-operative atrial fibrillation; pre-operative hemoglobin, platelet count, and INR; type of surgical procedure; duration of surgical procedure; duration
518
K.P. Blaine et al.
Fig. 1
Cohort diagram.
of cardiopulmonary bypass; use of additional hemostatic agents during surgery (such as recombinant factor VII, factor eight binding inhibitor bypassing activity (FEIBA), or desmopressin); and use of circulatory arrest. Intraoperative transfusions were included as a surrogate estimation of intraoperative blood loss as intravascular circulating blood volume is impossible to measure reliably under cardiopulmonary bypass. The primary endpoint was the total quantitative blood loss from all surgical drains, in mL, recorded during the first 8 hour after surgery. Beyond 8 hour, additional surgical drain output is less likely to be related to antifibrinolytics and may contain serous fluid, confounding the measurements. The secondary endpoints were the number of units of red blood cells, plasma, or platelets transfused during the first 8 hour after surgery.
use in multivariate regression. Univariate and multivariate OR are presented with 95% CI. All statistical analyses were performed in R Statistical Software v3.2.1 (R Foundation for Statistical Computing, Vienna, Austria). A power analysis was performed using PASS Statistical Software [14] and found that 120 patients achieved 90% power using an F-Test with a significance level of 0.05 to detect a 200-mL difference in surgical drain output during the first 8 hour after surgery. A 200-mL difference was selected after discussions with senior anesthesiologists and intensivists in our group as the drain volume likely to trigger a management decision, such as whether to transfuse blood product.
2.3. Statistical analysis
3.1. Patient characteristics
Baseline characteristics between the TXA and εACA groups were compared by descriptive statistics. Univariate regression analysis was used to identify potential confounders and is presented as frequencies for categorical variables, means ± SD for continuous data, and median (intraquartile range) for non-normally distributed continuous data. Multivariate linear regression with identified confounders adjusted by stepwise subtraction was performed for surgical drain output and the number of units of blood products transfused within the first 8 hour after surgery. Logarithmic transformation was used for non-Gaussian continuous variables to permit their
A total of 158 consecutive surgeries patients were screened. No patients were excluded for failure to receive one of the study drugs. All patients between February 1 and March 31, 2013, received εACA, and all patients from April 1 to June 3, 2013, received TXA. One patient who met inclusion criteria was subsequently excluded from further analysis due to a preoperative history of hemophilia with significant perioperative bleeding and transfusion requirements likely unrelated to the use of antifibrinolytics. A total of 128 patients were included in the analysis, including 68 receiving εACA and 60 receiving TXA (see Fig. 1). Baseline characteristics are presented in Table 1.
3. Results
Comparative effectiveness of eACA and TXA Table 1
519
Baseline patient characteristics.
Age (y) Female sex Race/ethnicity Non-hispanic White Black East Asian South Asian Latino ASA Physical Class ASA 3 ASA 4 Pre-op Labs Hb (g/dL) Plts (103/dL) INR Non-cardiac co-morbidities Diabetes mellitus Obesity Chronic obstructive pulmonary disease Hypertension Chronic renal disease (baseline serum Cr N1.5 g/dL) Peripheral arterial disease Cardiac co-morbidities Unstable coronary artery disease Atrial fibrillation NYHA class 3–4 CHF Type of surgery Isolated coronary artery bypass graft Isolated aortic valve surgery Isolated mitral valve surgery Isolated surgery of the ascending aorta Combined bypass graft and aortic valve surgery Combined aortic and mitral valve surgery Combined aortic valve and ascending aorta surgery Heart transplantation Left ventricular assist device insertion Surgical conditions Emergency surgery Re-do sternotomy Use of circulatory arrest Surgical duration (h) Duration of cardiopulmonary bypass (h) Use of rescue hemostatic agent Intra-operative transfusions Units of RBCs transfused Units of FFP transfused 6-pack of platelets transfused Clinical outcomes Surgical re-exploration within 8 h Inpatient mortality 8-h Study outcomes Surgical drain output (mL) Units of RBCs transfused Units of FFP transfused 6-pack of platelets transfused
εACA (n = 68)
TXA (n = 60)
Total (n = 128)
63.7 ± 13.7 25 (36.8%)
59.7 ± 15.9 22 (36.7%)
61.8 ± 14.8 47 (36.7%)
.937 .991
46 (67.6%) 4 (5.9%) 9 (13.2%) 3 (4.4%) 6 (8.8%)
46 (76.7%) 2 (3.3%) 2 (3.3%) 0 10 (16.7%)
92 (71.9) 6 (4.7%) 11 (8.6%) 3 (2.3%) 16 (12.5%)
.350 .793 .093 .289 .284
24 (35.3%) 44 (64.7%)
21 (35.0%) 39 (65.0%)
45 (35.2%) 83 (64.8%)
.972 .972
12.1 ± 2.4 225 ± 70 1.21 ± 0.24
12.4 ± 2.3 209 ± 64 1.20 ± 0.24
.175 .009** .620
16 (23.5%) 14 (20.6%) 3 (4.4%) 46 (67.6%) 13 (19.1%) 1 (1.5%)
16 (26.7%) 23 (38.3%) 3 (5.0%) 40 (66.7%) 11 (18.3%) 3 (5.0%)
32 (25.0%) 37 (28.9%) 6 (4.7%) 86 (67.2%) 24 (18.8%) 4 (3.1%)
.838 .044* .875 .906 .910 .525
24 (35.3%) 19 (27.3%) 24 (35.3%)
18 (30.0%) 16 (26.7%) 16 (26.7%)
42 (32.8%) 35 (27.3%) 40 (31.3%)
.654 .872 .390
14 (20.6%) 13 (19.1%) 9 (13.2%) 6 (8.8%) 6 (8.8%) 3 (4.4%) 6 (8.8%) 5 (7.4%) 4 (5.9%)
12 (20.0%) 10 (16.7%) 8 (13.3%) 9 (15.0%) 4 (6.7%) 0 4 (6.7%) 3 (5.0%) 5 (8.3%)
26 (20.3%) 23 (18.0%) 17 (13.3%) 15 (11.7%) 10 (7.8%) 3 (2.3%) 10 (7.8%) 8 (6.3%) 9 (7.0%)
.583 .588 .934 .719 .987 .278 .650 .100 .650
11 (16.2%) 11 (16.2%) 8 (11.8%) 5.9 ± 1.6 2.8 ± 1.2 22 (32.4%)
13 (21.7%) 15 (25.0%) 7 (11.7%) 6.0 ± 2.4 2.5 ± 1.1 10 (16.7%)
24 (18.8%) 26 (20.3%) 15 (11.7%) 6.0 ± 2.0 2.7 ± 1.1 32 (25.0%)
.571 .309 .986 .913 .092 .039*
0 (0–17) 0 (0–17) 0 (0–6)
0 (0–20) 0 (0–10) 0 (0–4)
0 (0–20) 0 (0–17) 0 (0–6)
.956 .656 .985
0 3 (4.4%)
0 1 (1.7%)
0 4 (3.1%)
.703
12.7 ± 2.1 195 ± 55 1.19 ± 0.25
425 (235–2400) 0.65 ± 1.27 0.32 ± 0.78 0.16 ± 0.44
333 (188–2284) 0.32 ± 0.87 0.25 ± 0.75 0.12 ± 0.42
393 (211–2400) 0.49 ± 1.11 0.29 ± 0.76 0.14 ± 0.43
P
.217 .086 .588 .554
Normally distributed continuous data presented as means ± SD. Non-normal data presented as medians and interquartile ranges. Categorical data presented as frequencies. Abbreviations: ASA-PS, American Society of Anesthesiologists Physical Status; NYHA, New York Heart Association; CHF, Congestive heart failure; FFP, Fresh frozen plasma.
520 Patients receiving εACA and TXA differed in regards to preoperative platelet count and obesity. Patients receiving εACA were also more likely to receive rescue hemostatic agents.
3.2. Eight-hour surgical drain output Univariate predictors of postoperative drainage are presented in Table 2. A multivariate linear regression line demonstrated a correlation coefficient of 0.32. Significant predictors included female sex, mitral valve surgery, and the use of intraoperative hemostatic agents. Postoperative surgical drainage did not differ between patients receiving εACA and TXA (Table 3a).
3.3. 8-hour transfusion requirements Multivariate linear regression for allogenic RBC transfusion returned a weak correlation coefficient of 0.14. A statistically significant regression model could not be generated to predict the transfusion of fresh frozen plasma or platelets. The use of neither εACA nor TXA predicted RBC transfusion of any blood products at 8 hours (Table 3b).
4. Discussion Both εACA and TXA have established roles in reducing perioperative bleeding in cardiac surgery. Small randomized trials from the 1990s and early 2000s suggested equivalence between these agents and inferiority to aprotonin. Since the withdrawal of aprotinin, institutions in the United States have reverted back to the older lysine analogs, often interchangeably, assuming therapeutic equivalence. The clinical effectiveness of either agent has not been re-evaluated in contemporary practice. The present study used the opportunity of a medication shortage to examine postoperative bleeding in patients receiving recommended doses of either εACA or TXA. In a study adequately powered to detect a clinically meaningful change of 200 mL of surgical drainage after 8 hours, no difference was seen between these agents in multivariate analyses. Multiple studies have compared antifibrinolytic medications in cardiac surgery, but not definitive recommendation has yet been made. Since 2006, with the recognition that allogeneic blood may worsen outcomes in some patients [15], most centers have now adopted more conservative transfusion strategies to minimize the volume of blood transfused. Therefore, prior studies conducted with more liberal transfusion practices may have limited applications to modern clinical management. The only large trial with εACA and TXA arms, the BART trial, was designed to look for the clinical superiority of aprotonin versus 2 controls. The composite endpoint of ‘massive hemorrhage’ (based on chest tube output N1500 mL, transfusion N10 U of RBCs, re-operation, or death within 30 days) presumed that antifibrinolytic drugs were effective
K.P. Blaine et al. against catastrophic bleeding. We suspect that hyperfibrinolysis is more likely to manifest with slower venous oozing from surgical beds rather than acute, life-threatening hemorrhage. The BART trial was neither designed nor powered to detect differences in minor bleeding between εACA and TXA, and so did not completely address the question. Our results agree with a similar retrospective study performed independently by Falana and Patel [16], who also performed a single-center retrospective analysis during the 2012–2013 εACA shortage. The authors used the BART study composite endpoint of ‘massive hemorrhage’ with a univariate analysis and similarly report no difference between the two agents. As already noted, major hemorrhage in the immediate postoperative period is often related to surgical complications that antifibrinolytics would not correct. The prior study was confounded by indication, as treating clinicians could choose either εACA or TXA. The prior study was underpowered, with only 15 of 120 patients meeting the composite endpoint. The authors did measure adverse events, and reported no difference in safety outcomes. Importantly, they also included a cost-effectiveness analysis specifically considering blood transfusions, and found that εACA is the more affordable option for their institution. Cost-effectiveness may play a role in future work. Pharmaceutical prices paid by institutions fluctuate and can be difficult to predict. Given two agents with similar clinical applications and efficacies, but differing price, the decision to use one over the other may depend on cost-savings in another realm. For blood therapies, that often relates to the cost of blood transfusion. The cost of a single unit of red blood cells in the United States is estimated at $761 ± $294 [17], which is substantially more than the cost of either lysine analog. Although pharmaceutical costs within the United States can vary by time and institution, our institution typically paid US$13.85 for εACA per cardiac case immediately prior to the 2012 to 2013 shortage, and US$148 for TXA per cardiac case during the shortage. However, transfusions may not be the only marker of cost in this setting. Patients receiving εACA in our study were more likely to require additional hemostatic agents, which include recombinant factor VII, FEIBA, or desmopressin. At our institution, these agents are used as rescue therapy in patients already managed with antifibrinolytic agents. In multivariate regression, the use of these agents was a strong confounder for both drainage and transfusion. The present study is neither designed nor powered to explore the differential use of rescue therapies. These adjuncts can cost thousands of US dollars per dose, such that even when there is no difference in the cost of blood products, TXA may be the more affordable option if it saves a single dose of a rescue agent. Future studies investigating perioperative bleeding should consider the use of adjunctive therapies as markers of unsatisfactory clot stability, either separately or in a composite endpoint with bleeding or transfusions. Medication shortages are an ongoing problem in North America with important consequences for patient care [18]. However, there is emerging awareness that useful information
Comparative effectiveness of eACA and TXA Table 2
521
Unadjusted odds ratios for surgical drainage at 8 hours. OR (95% CI)
Age Female sex Race Non-hispanic White Black East Asian South Asian Latino ASA-Physical Class ASA 3 ASA 4 Pre-op labs Hb (g/L) Plts (103/dL) INR Non-cardiac co-morbidities Diabetes mellitus Obese COPD Hypertension Chronic renal disease (baseline serum Cr N1.5) Peripheral arterial disease Cardiac co-morbidities Unstable coronary artery disease Atrial fibrillation NYHA class 3–4 CHF Surgical procedure Isolated coronary artery bypass graft Isolated aortic valve surgery Isolated mitral valve surgery Isolated surgery of the ascending aorta Combined bypass graft and aortic valve surgery Combined aortic and mitral valve surgery Combined aortic valve and ascending aorta surgery Heart transplantation Left ventricular assist device insertion Surgical conditions Emergency surgery Re-do sternotomy Use of circulatory arrest Surgical duration (h) Duration of cardiopulmonary bypass (h) Use of rescue hemostatic agent Intra-operative Transfusions Units of RBCs transfused Units of FFP transfused 6-pack of platelets transfused
P-value
1.01 (1.00–1.01) 0.53 (0.40–0.71)
.290 b.001***
0.82 (0.59–1.13) 2.27 (1.15–4.47) 0.85 (0.50–1.44) 0.88 (0.33–2.31) 1.19 (0.77–1.86)
.232 .018* .547 .791 .436
0.87 (0.64–1.19) 1.15 (0.84–1.56)
.386 .386
0.98 (0.92–1.05) 1.00 (1.00–1.01) 2.16 (1.17–4.01)
.596 .038* .016*
1.13 (0.58–1.59) 1.06 (0.77–1.46) 1.62 (0.81–3.23) 1.14 (0.84–1.56) 1.76 (1.23–2.53) 0.91 (0.39–2.13)
.478 .735 .172 .400 .003** .838
1.50 (1.10–2.03) 1.21 (0.87–1.67) 1.44 (1.06–1.96)
.011* .262 .023*
1.26 (0.81–1.97) 0.94 (0.57–1.56) 1.23 (0.70–2.19) 0.56 (0.33–0.95) 1.05 (0.51–2.13) n/a 2.10 (1.06–4.18) 0.92 (0.50–1.69) 2.13 (1.22–3.72)
.305 .825 .474 .037* .902 n/a .038* .794 .009**
1.08 (0.74–1.57) 1.24 (0.86–1.78) 1.17 (0.74–1.84) 1.07 (1.00–1.15) 1.03 (0.90–1.17) 2.04 (1.49–2.80)
.698 .250 .506 .052 .672 b.001***
1.44 (1.22–1.70) 1.70 (0.16–2.02) 1.38 (1.25–1.52)
.032* .002** .001**
Data presented as OR (95% CI). A logarithmic transformation of surgical drainage at 8 hours was required to normalize the data.
emerges from the use of substitute medications during these events [16,19,20]. This is particularly true for comparing the effectiveness of generic medications. For example, one recent study reviewed pressor selection and hemodynamics during a
recent shortage of ephedrine [19]. Future studies may likewise be able to retrospectively review institutional experience from a shortage to contrast commonly used but little examined medications.
522 Table 3a hours.
K.P. Blaine et al. Multivariate regression for surgical drain output at 8
εACA Female sex Mitral valve surgery Use of rescue hemostatic agent
OR (95% CI)
P
1.10 (0.97–1.26) 0.56 (0.49–0.64) 0.70 (0.59–0.84) 1.96 (1.69–2.28)
.460 b.001*** .049* b.001***
2
Multivariate regression analysis with r = 0.32. Data presented as OR (95% CI). A logarithmic transformation of surgical drainage at 8 hours was required to normalize the data. Abbreviations: epsilon-aminocaproic acid (εACA).
Limitations of the present study include a retrospective design, which is unable to identify all potential confounders. All consecutive surgeries were included, and so the control group may be considered an historical control. It is possible, though unlikely, that practice changes during the 4-month study period may account from some of the variability seen between the 2 arms. A sample size analysis suggests an adequate sample to detect a clinically relevant change, although a larger sample would have provided greater confidence in our ability to adjust for unmeasured confounders. The study also occurred at a single academic center, which may not represent practice at all institutions. Some readers may question the exclusion of the surgical estimated blood loss. That decision was justified by the uncertain nature of blood loss measurements for patients on bypass and the use of intraoperative transfusions as a surrogate measurement. The inclusion of a heterogeneous patient population better reflects modern clinical practice than a population narrowed to a single surgical indication, but also produced a large degree of unmeasured variability. The choice of postoperative surgical drainage is an appropriate endpoint that reflects the most likely clinical benefit from antifibrinolytic drugs, and the planned difference of 200 mL is meaningful to clinicians, although some may have preferred a smaller difference. One strength is that the study is free from confounding by indication because of a strictly protocolized blood conservation bundle. In summary, the present study is a single-center, retrospective review of the relative clinical effectiveness of εACA and TXA to minimize postoperative bleeding and transfusion under contemporary clinical practice. In multivariate analysis, no difference was observed for bleeding, however patients receiving εACA were more likely to require rescue hemostatic agents. The present study demonstrates how a medication shortage can be used opportunistically to review the use of older medication in current clinical practice with minimal patient risk and study expense.
5. Conclusion A retrospective analysis of the effect of εACA and TXA following cardiac surgery at a single center in the United States found no difference in chest tube drainage or transfusion
Table 3b Multivariate regression for red blood cell transfusion at 8 hours. OR (95% CI) εACA Rescue hemostatic agent Ventricular assist device insertion surgery
P
1.79 (0.79–2.73) 4.71 (2.88–7.70)
.176 .002**
12.79 (5.64–29.02)
.002**
2
Multivariate regression analysis with r = 0.14. Data presented as OR (95% CI).
requirements between the two medications, although patients receiving εACA were more likely to require rescue hemostatic agents.
Acknowledgements The authors would like to acknowledge the technical assistance of Fred Hurt, DPharm, and the operating room pharmacy staff at Stanford University Medical Center. This project was supported solely by departmental funds from the Stanford University School of Medicine Department of Anesthesiology, Pain, and Perioperative Medicine.
References [1] Stover EP, Siegel LC, Parks R, Levin J, Body SC, Maddi R, et al. Variability in transfusion practice for coronary artery bypass surgery persists despite national consensus guidelines: a 24-institution study. Institutions of the multicenter study of perioperative ischemia research group. Anesthesiology 1998;88:327-33. [2] Andreasen JJ. Pharmacologic methods to reduce postoperative bleeding in adult cardiac surgery. A mini-review. Curr Pharm Des 2013;19: 3992-5. [3] Nalla BP, Freedman J, Hare GM, Mazer CD. Update on blood conservation for cardiac surgery. J Cardiothorac Vasc Anesth 2012;26:117-33. [4] Mangano DT, Tudor IC, Dietzel C. Multicenter study of perioperative ischemia research G, ischemia R, education F the risk associated with aprotinin in cardiac surgery. N Engl J Med 2006;354:353-65. [5] Mangano DT, Miao Y, Vuylsteke A, Tudor IC, Juneja R, Filipescu D, et al. Mortality associated with aprotinin during 5 years following coronary artery bypass graft surgery. JAMA 2007;297:471-9. [6] McMullan V, Alston RP III. Aprotinin and cardiac surgery: a sorry tale of evidence misused. Br J Anaesth 2013;110:675-8. [7] Hardy JF, Belisle S, Dupont C, Harel F, Robitaille D, Roy M, et al. Prophylactic tranexamic acid and epsilon-aminocaproic acid for primary myocardial revascularization. Ann Thorac Surg 1998;65:371-6. [8] Maineri P, Covaia G, Realini M, Caccia G, Ucussich E, Luraschi M, et al. Postoperative bleeding after coronary revascularization. Comparison between tranexamic acid and epsilon-aminocaproic acid. Minerva Cardioangiol 2000;48:155-60. [9] Pinosky ML, Kennedy DJ, Fishman RL, Reeves ST, Alpert CC, Ecklund J, et al. Tranexamic acid reduces bleeding after cardiopulmonary bypass when compared to epsilon aminocaproic acid and placebo. J Card Surg 1997;12:330-8. [10] Penta de Peppo A, Pierri MD, Scafuri A, De Paulis R, Colantuono G, Caprara E, et al. Intraoperative antifibrinolysis and blood-saving
Comparative effectiveness of eACA and TXA
[11]
[12]
[13]
[14] [15]
techniques in cardiac surgery. Prospective trial of 3 antifibrinolytic drugs. Tex Heart Inst J 1995;22:231-6. Menichetti A, Tritapepe L, Ruvolo G, Speziale G, Cogliati A, Di Giovanni C, et al. Changes in coagulation patterns, blood loss and blood use after cardiopulmonary bypass: aprotinin vs tranexamic acid vs epsilon aminocaproic acid. J Cardiovasc Surg (Torino) 1996;37:401-7. Fergusson DA, Hebert PC, Mazer CD, Fremes S, MacAdams C, Murkin JM, et al. A comparison of aprotinin and lysine analogues in high-risk cardiac surgery. N Engl J Med 2008;358:2319-31. Henry DA, Carless PA, Moxey AJ, O'Connell D, Stokes BJ, Fergusson DA, et al. Anti-fibrinolytic use for minimising perioperative allogeneic blood transfusion. Cochrane Database Syst Rev 2011, CD001886. Hintze J. Pass 12. Kaysville, Utah, USA: NCSS, LLC; 2013. Koch CG, Li L, Duncan AI, Mihaljevic T, Cosgrove DM, Loop FD, et al. Morbidity and mortality risk associated with red blood cell and bloodcomponent transfusion in isolated coronary artery bypass grafting. Crit Care Med 2006;34:1608-16.
523 [16] Falana O, Patel G. Efficacy and safety of tranexamic acid versus -aminocaproic acid in cardiovascular surgery. Ann Pharmacother 2014; 48:1563-9. [17] Shander A, Hofmann A, Ozawa S, Theusinger OM, Gombotz H, Spahn DR. Activity-based costs of blood transfusions in surgical patients at four hospitals. Transfusion 2010;50:753-65. [18] McLaughlin M, Kotis D, Thomson K, Harrison M, Fennessy G, Postelnick M, et al. Effects on patient care caused by drug shortages: a survey. J Manag Care Pharm 2013;19:783-8. [19] Ladha KS, Nanji KC, Pierce E, Poon KT, Hyder JA. The impact of a shortage of pharmacy-prepared ephedrine syringes on intraoperative medication use. Anesth Analg 2015;121:404-9. [20] Thoma BN, Li J, McDaniel CM, Wordell CJ, Cavarocchi N, Pizzi LT. Clinical and economic impact of substituting dexmedetomidine for propofol due to a US drug shortage: examination of coronary artery bypass graft patients at an urban medical Centre. Pharmacoeconomics 2014;32: 149-57.
Original Contribution
Comparative effectiveness of epsilon-aminocaproic acid and tranexamic acid on postoperative bleeding following cardiac surgery during a national medication shortage☆ Kevin P. Blaine MD, MPH (Resident Physician)⁎, Christopher Press MD (Resident Physician), Ken Lau MD (Resident Physician)1 , Jan Sliwa MD (Resident Physician)2 , Vidya K. Rao MD (Clinical Assistant Professor), Charles Hill MD (Clinical Assistant Professor and Associated Critical Care Director) Department of Anesthesia, Pain, and Perioperative Medicine, Stanford University Medical Center, 300 Pasteur Dr, H3580, Stanford, CA 97305 Received 15 December 2015; accepted 14 August 2016
Keywords: ε-aminocaproic acid; Tranexamic acid; Blood conservation; Perioperative bleeding; Cardiac surgery
Abstract Study objective: The aim of this study was to compare the effectiveness of epsilon-aminocaproic acid (εACA) and tranexamic acid (TXA) in contemporary clinical practice during a national medication shortage. Design: A retrospective cohort study. Setting: The study was performed in all consecutive cardiac surgery patients (n = 128) admitted to the cardiac-surgical intensive care unit after surgery at a single academic center immediately before and during a national medication shortage. Measurements: Demographic, clinical, and outcomes data were compared by descriptive statistics using χ2 and t test. Surgical drainage and transfusions were compared by multivariate linear regression for patients receiving εACA before the shortage and TXA during the shortage. Main results: In multivariate analysis, no statistical difference was found for surgical drain output (OR 1.10, CI 0.97–1.26, P = .460) or red blood cell transfusion requirement (OR 1.79, CI 0.79–2.73, P = .176). Patients receiving εACA were more likely to receive rescue hemostatic medications (OR 1.62, CI 1.02–2.55, P = .041). Conclusions: Substitution of εACA with TXA during a national medication shortage produced equivalent postoperative bleeding and red cell transfusions, although patients receiving εACA were more likely to require supplemental hemostatic agents. Published by Elsevier Inc.
☆
State of Disclosure: No authors have financial conflicts of interest related to this work. ⁎ Corresponding author at: Critical Care Medicine Department, National Institutes of Health Clinical Center, 9000 Rockville Pike. 2C145. Tel.: +1 240 762 1873. E-mail addresses: [email protected] (K.P. Blaine). 1 Present affiliation: Anesthesia Services Medical Group, 3626 Ruffin Rd., San Diego, CA 92123. 2 Present affiliation: Department of Anesthesiology, Brigham and Women's Hospital, 75 Francis St, Boston, MA 02115.
http://dx.doi.org/10.1016/j.jclinane.2016.08.037 0952-8180/Published by Elsevier Inc.
Comparative effectiveness of eACA and TXA
1. Introduction Nearly 1 in 5 blood products transfusions in the United States occurs in the setting of cardiac surgery [1]. Blood conservation techniques are used to minimize perioperative transfusions during and after cardiac surgery [2]. These techniques include intraoperative cell-salvage devices, improved surgical techniques, topical hemostatic agents, and conservative transfusion thresholds [3]. Anti-fibrinolytic agents are often used as a pharmacological adjunct to maintain clot stability and minimize bleeding. Contemporary antifibrinolytic agents include aprotinin, εaminocaproic acid (εACA), and tranexamic acid (TXA). In 2007 to 2008, the most thoroughly studied agent, aprotinin, was removed from the market in the US, UK, Canada, and European Union due to safety concerns [4,5]. Since 2012, aprotinin has been returned for use in Canada and Europe following a systematic review of multiple studies concluding with a favorable the risk/benefit ratio [6]. Aprotinin remains unavailable in the United States. The alternative antifibrinolytic agents, εACA and TXA, have been compared in small trials published between 1994 and 2001 [7-11] which did not establish the clinical superiority of either agent. The Blood Conservation Using Antifibrinolytics in a Randomized Trial (BART) was the only large study to include TXA (n = 770) and εACA (n = 768) arms, and demonstrated equivocal outcomes in mortality for a composite ‘massive transfusion’ endpoint (including greater than 1.5 L blood loss within 8 hours, N8 U red blood cells (RBCs) transfused, or surgical re-exploration for hemorrhage) [12]. The results are difficult to interpret as major hemorrhage following surgery is multifactorial and often involves compromised surgical hemostasis. It is unrealistic to expect an antifibrinolytic agent alone to prevent perioperative catastrophe. A recent Cochrane review reported insufficient evidence (particularly for εACA) to recommend either lysine analog over the other [13]. Studies included in that analysis were limited by small size, remote publication dates, and surgical indications often limited to primary coronary revascularization surgery (for which many patients today might receive percutaneous coronary angiography). Contemporary antifibrinolytic agents are used for a variety of cardiac procedures beyond coronary revascularization, such as valve surgery or ventricular assist device placement, for which they have not been well studied. Finally, since the 1990s when many of these trials were conducted, thresholds for transfusions have lowered such that today small differences in blood may be more relevant towards the decisions to transfuse than they would have been in an era where most patients would receive blood regardless. While some clinicians favor a particular agent, at present, the use of εACA and TXA remains in a state of equipoise. In 2012–2013, the United States experienced a national shortage of εACA following FDA-mandated upgrades to the manufacturing process (W. Fridich, Luitpold Pharmaceuticals, personal communication, August 8, 2015). The blood
517 conservation practice at our institution included eACA exclusively until stores of this medication were exhausted on March 30, 2013. There was an immediate substitution of TXA for all cardiac procedures from April 1, 2013, until supplies were replenished on June 4, 2013. The shortage presented an opportunity to compare retrospectively the effectiveness of εACA and TXA on postoperative bleeding in the ICU following cardiac surgery under contemporary surgical conditions and with minimal patient risk.
2. Materials and methods 2.1. Study population The study population includes all consecutive patients undergoing cardiac surgery with cardiopulmonary bypass between February 1 and June 3, 2013, who received either TXA or εACA. The protocol was approved by the Research Compliance Office and Institutional Review Board of Stanford University School of Medicine with a waiver of informed consent. Patients were included for the first cardiac or proximal aortic surgery of the hospitalization. Patients were required to have a planned post-operative admission to the cardiothoracic surgical intensive care unit with surgical drains in situ. Patients who died or required surgical exploration within 8 hours were excluded, as antifibrinolytic agents alone were unlikely to prevent these complications. Prior to April 1, 2013, all cardiac surgery patients at our institution received εACA (loading dose 10-15 mg/kg over 1015 min, 2-3 mg/kg bolus into the bypass priming solution, then 1-2 mg/kg per hour infusion for 6 hours). During the shortage, all patients received TXA (loading dose 10-15 mg/kg over 1015 min, 2-2.5 mg/kg bolus in the bypass priming solution, then 1-2 mg/kg per hour for 6 hours) until εACA was again available on June 4, 2013. Allotment of either εACA or TXA was assigned based on medication availability and was independent of clinician judgment.
2.2. Data collection Patient charts were accessed for the following data: patient demographics including age on admission, sex, and race; preoperative co-morbidities including hypertension, diabetes mellitus, chronic pulmonary obstructive disease, chronic renal insufficiency (baseline Cr N1.5 g/dL), congestive heart failure with New York Heart Association functional class; ASA Physical Classification; unstable pre-operative coronary artery disease (as evidenced by electrography, echocardiography, cardiac catheterization, perfusion scan, or unstable angina defined by chest pain at rest with electrocardiogram or echocardiographic changes); pre-operative atrial fibrillation; pre-operative hemoglobin, platelet count, and INR; type of surgical procedure; duration of surgical procedure; duration
518
K.P. Blaine et al.
Fig. 1
Cohort diagram.
of cardiopulmonary bypass; use of additional hemostatic agents during surgery (such as recombinant factor VII, factor eight binding inhibitor bypassing activity (FEIBA), or desmopressin); and use of circulatory arrest. Intraoperative transfusions were included as a surrogate estimation of intraoperative blood loss as intravascular circulating blood volume is impossible to measure reliably under cardiopulmonary bypass. The primary endpoint was the total quantitative blood loss from all surgical drains, in mL, recorded during the first 8 hour after surgery. Beyond 8 hour, additional surgical drain output is less likely to be related to antifibrinolytics and may contain serous fluid, confounding the measurements. The secondary endpoints were the number of units of red blood cells, plasma, or platelets transfused during the first 8 hour after surgery.
use in multivariate regression. Univariate and multivariate OR are presented with 95% CI. All statistical analyses were performed in R Statistical Software v3.2.1 (R Foundation for Statistical Computing, Vienna, Austria). A power analysis was performed using PASS Statistical Software [14] and found that 120 patients achieved 90% power using an F-Test with a significance level of 0.05 to detect a 200-mL difference in surgical drain output during the first 8 hour after surgery. A 200-mL difference was selected after discussions with senior anesthesiologists and intensivists in our group as the drain volume likely to trigger a management decision, such as whether to transfuse blood product.
2.3. Statistical analysis
3.1. Patient characteristics
Baseline characteristics between the TXA and εACA groups were compared by descriptive statistics. Univariate regression analysis was used to identify potential confounders and is presented as frequencies for categorical variables, means ± SD for continuous data, and median (intraquartile range) for non-normally distributed continuous data. Multivariate linear regression with identified confounders adjusted by stepwise subtraction was performed for surgical drain output and the number of units of blood products transfused within the first 8 hour after surgery. Logarithmic transformation was used for non-Gaussian continuous variables to permit their
A total of 158 consecutive surgeries patients were screened. No patients were excluded for failure to receive one of the study drugs. All patients between February 1 and March 31, 2013, received εACA, and all patients from April 1 to June 3, 2013, received TXA. One patient who met inclusion criteria was subsequently excluded from further analysis due to a preoperative history of hemophilia with significant perioperative bleeding and transfusion requirements likely unrelated to the use of antifibrinolytics. A total of 128 patients were included in the analysis, including 68 receiving εACA and 60 receiving TXA (see Fig. 1). Baseline characteristics are presented in Table 1.
3. Results
Comparative effectiveness of eACA and TXA Table 1
519
Baseline patient characteristics.
Age (y) Female sex Race/ethnicity Non-hispanic White Black East Asian South Asian Latino ASA Physical Class ASA 3 ASA 4 Pre-op Labs Hb (g/dL) Plts (103/dL) INR Non-cardiac co-morbidities Diabetes mellitus Obesity Chronic obstructive pulmonary disease Hypertension Chronic renal disease (baseline serum Cr N1.5 g/dL) Peripheral arterial disease Cardiac co-morbidities Unstable coronary artery disease Atrial fibrillation NYHA class 3–4 CHF Type of surgery Isolated coronary artery bypass graft Isolated aortic valve surgery Isolated mitral valve surgery Isolated surgery of the ascending aorta Combined bypass graft and aortic valve surgery Combined aortic and mitral valve surgery Combined aortic valve and ascending aorta surgery Heart transplantation Left ventricular assist device insertion Surgical conditions Emergency surgery Re-do sternotomy Use of circulatory arrest Surgical duration (h) Duration of cardiopulmonary bypass (h) Use of rescue hemostatic agent Intra-operative transfusions Units of RBCs transfused Units of FFP transfused 6-pack of platelets transfused Clinical outcomes Surgical re-exploration within 8 h Inpatient mortality 8-h Study outcomes Surgical drain output (mL) Units of RBCs transfused Units of FFP transfused 6-pack of platelets transfused
εACA (n = 68)
TXA (n = 60)
Total (n = 128)
63.7 ± 13.7 25 (36.8%)
59.7 ± 15.9 22 (36.7%)
61.8 ± 14.8 47 (36.7%)
.937 .991
46 (67.6%) 4 (5.9%) 9 (13.2%) 3 (4.4%) 6 (8.8%)
46 (76.7%) 2 (3.3%) 2 (3.3%) 0 10 (16.7%)
92 (71.9) 6 (4.7%) 11 (8.6%) 3 (2.3%) 16 (12.5%)
.350 .793 .093 .289 .284
24 (35.3%) 44 (64.7%)
21 (35.0%) 39 (65.0%)
45 (35.2%) 83 (64.8%)
.972 .972
12.1 ± 2.4 225 ± 70 1.21 ± 0.24
12.4 ± 2.3 209 ± 64 1.20 ± 0.24
.175 .009** .620
16 (23.5%) 14 (20.6%) 3 (4.4%) 46 (67.6%) 13 (19.1%) 1 (1.5%)
16 (26.7%) 23 (38.3%) 3 (5.0%) 40 (66.7%) 11 (18.3%) 3 (5.0%)
32 (25.0%) 37 (28.9%) 6 (4.7%) 86 (67.2%) 24 (18.8%) 4 (3.1%)
.838 .044* .875 .906 .910 .525
24 (35.3%) 19 (27.3%) 24 (35.3%)
18 (30.0%) 16 (26.7%) 16 (26.7%)
42 (32.8%) 35 (27.3%) 40 (31.3%)
.654 .872 .390
14 (20.6%) 13 (19.1%) 9 (13.2%) 6 (8.8%) 6 (8.8%) 3 (4.4%) 6 (8.8%) 5 (7.4%) 4 (5.9%)
12 (20.0%) 10 (16.7%) 8 (13.3%) 9 (15.0%) 4 (6.7%) 0 4 (6.7%) 3 (5.0%) 5 (8.3%)
26 (20.3%) 23 (18.0%) 17 (13.3%) 15 (11.7%) 10 (7.8%) 3 (2.3%) 10 (7.8%) 8 (6.3%) 9 (7.0%)
.583 .588 .934 .719 .987 .278 .650 .100 .650
11 (16.2%) 11 (16.2%) 8 (11.8%) 5.9 ± 1.6 2.8 ± 1.2 22 (32.4%)
13 (21.7%) 15 (25.0%) 7 (11.7%) 6.0 ± 2.4 2.5 ± 1.1 10 (16.7%)
24 (18.8%) 26 (20.3%) 15 (11.7%) 6.0 ± 2.0 2.7 ± 1.1 32 (25.0%)
.571 .309 .986 .913 .092 .039*
0 (0–17) 0 (0–17) 0 (0–6)
0 (0–20) 0 (0–10) 0 (0–4)
0 (0–20) 0 (0–17) 0 (0–6)
.956 .656 .985
0 3 (4.4%)
0 1 (1.7%)
0 4 (3.1%)
.703
12.7 ± 2.1 195 ± 55 1.19 ± 0.25
425 (235–2400) 0.65 ± 1.27 0.32 ± 0.78 0.16 ± 0.44
333 (188–2284) 0.32 ± 0.87 0.25 ± 0.75 0.12 ± 0.42
393 (211–2400) 0.49 ± 1.11 0.29 ± 0.76 0.14 ± 0.43
P
.217 .086 .588 .554
Normally distributed continuous data presented as means ± SD. Non-normal data presented as medians and interquartile ranges. Categorical data presented as frequencies. Abbreviations: ASA-PS, American Society of Anesthesiologists Physical Status; NYHA, New York Heart Association; CHF, Congestive heart failure; FFP, Fresh frozen plasma.
520 Patients receiving εACA and TXA differed in regards to preoperative platelet count and obesity. Patients receiving εACA were also more likely to receive rescue hemostatic agents.
3.2. Eight-hour surgical drain output Univariate predictors of postoperative drainage are presented in Table 2. A multivariate linear regression line demonstrated a correlation coefficient of 0.32. Significant predictors included female sex, mitral valve surgery, and the use of intraoperative hemostatic agents. Postoperative surgical drainage did not differ between patients receiving εACA and TXA (Table 3a).
3.3. 8-hour transfusion requirements Multivariate linear regression for allogenic RBC transfusion returned a weak correlation coefficient of 0.14. A statistically significant regression model could not be generated to predict the transfusion of fresh frozen plasma or platelets. The use of neither εACA nor TXA predicted RBC transfusion of any blood products at 8 hours (Table 3b).
4. Discussion Both εACA and TXA have established roles in reducing perioperative bleeding in cardiac surgery. Small randomized trials from the 1990s and early 2000s suggested equivalence between these agents and inferiority to aprotonin. Since the withdrawal of aprotinin, institutions in the United States have reverted back to the older lysine analogs, often interchangeably, assuming therapeutic equivalence. The clinical effectiveness of either agent has not been re-evaluated in contemporary practice. The present study used the opportunity of a medication shortage to examine postoperative bleeding in patients receiving recommended doses of either εACA or TXA. In a study adequately powered to detect a clinically meaningful change of 200 mL of surgical drainage after 8 hours, no difference was seen between these agents in multivariate analyses. Multiple studies have compared antifibrinolytic medications in cardiac surgery, but not definitive recommendation has yet been made. Since 2006, with the recognition that allogeneic blood may worsen outcomes in some patients [15], most centers have now adopted more conservative transfusion strategies to minimize the volume of blood transfused. Therefore, prior studies conducted with more liberal transfusion practices may have limited applications to modern clinical management. The only large trial with εACA and TXA arms, the BART trial, was designed to look for the clinical superiority of aprotonin versus 2 controls. The composite endpoint of ‘massive hemorrhage’ (based on chest tube output N1500 mL, transfusion N10 U of RBCs, re-operation, or death within 30 days) presumed that antifibrinolytic drugs were effective
K.P. Blaine et al. against catastrophic bleeding. We suspect that hyperfibrinolysis is more likely to manifest with slower venous oozing from surgical beds rather than acute, life-threatening hemorrhage. The BART trial was neither designed nor powered to detect differences in minor bleeding between εACA and TXA, and so did not completely address the question. Our results agree with a similar retrospective study performed independently by Falana and Patel [16], who also performed a single-center retrospective analysis during the 2012–2013 εACA shortage. The authors used the BART study composite endpoint of ‘massive hemorrhage’ with a univariate analysis and similarly report no difference between the two agents. As already noted, major hemorrhage in the immediate postoperative period is often related to surgical complications that antifibrinolytics would not correct. The prior study was confounded by indication, as treating clinicians could choose either εACA or TXA. The prior study was underpowered, with only 15 of 120 patients meeting the composite endpoint. The authors did measure adverse events, and reported no difference in safety outcomes. Importantly, they also included a cost-effectiveness analysis specifically considering blood transfusions, and found that εACA is the more affordable option for their institution. Cost-effectiveness may play a role in future work. Pharmaceutical prices paid by institutions fluctuate and can be difficult to predict. Given two agents with similar clinical applications and efficacies, but differing price, the decision to use one over the other may depend on cost-savings in another realm. For blood therapies, that often relates to the cost of blood transfusion. The cost of a single unit of red blood cells in the United States is estimated at $761 ± $294 [17], which is substantially more than the cost of either lysine analog. Although pharmaceutical costs within the United States can vary by time and institution, our institution typically paid US$13.85 for εACA per cardiac case immediately prior to the 2012 to 2013 shortage, and US$148 for TXA per cardiac case during the shortage. However, transfusions may not be the only marker of cost in this setting. Patients receiving εACA in our study were more likely to require additional hemostatic agents, which include recombinant factor VII, FEIBA, or desmopressin. At our institution, these agents are used as rescue therapy in patients already managed with antifibrinolytic agents. In multivariate regression, the use of these agents was a strong confounder for both drainage and transfusion. The present study is neither designed nor powered to explore the differential use of rescue therapies. These adjuncts can cost thousands of US dollars per dose, such that even when there is no difference in the cost of blood products, TXA may be the more affordable option if it saves a single dose of a rescue agent. Future studies investigating perioperative bleeding should consider the use of adjunctive therapies as markers of unsatisfactory clot stability, either separately or in a composite endpoint with bleeding or transfusions. Medication shortages are an ongoing problem in North America with important consequences for patient care [18]. However, there is emerging awareness that useful information
Comparative effectiveness of eACA and TXA Table 2
521
Unadjusted odds ratios for surgical drainage at 8 hours. OR (95% CI)
Age Female sex Race Non-hispanic White Black East Asian South Asian Latino ASA-Physical Class ASA 3 ASA 4 Pre-op labs Hb (g/L) Plts (103/dL) INR Non-cardiac co-morbidities Diabetes mellitus Obese COPD Hypertension Chronic renal disease (baseline serum Cr N1.5) Peripheral arterial disease Cardiac co-morbidities Unstable coronary artery disease Atrial fibrillation NYHA class 3–4 CHF Surgical procedure Isolated coronary artery bypass graft Isolated aortic valve surgery Isolated mitral valve surgery Isolated surgery of the ascending aorta Combined bypass graft and aortic valve surgery Combined aortic and mitral valve surgery Combined aortic valve and ascending aorta surgery Heart transplantation Left ventricular assist device insertion Surgical conditions Emergency surgery Re-do sternotomy Use of circulatory arrest Surgical duration (h) Duration of cardiopulmonary bypass (h) Use of rescue hemostatic agent Intra-operative Transfusions Units of RBCs transfused Units of FFP transfused 6-pack of platelets transfused
P-value
1.01 (1.00–1.01) 0.53 (0.40–0.71)
.290 b.001***
0.82 (0.59–1.13) 2.27 (1.15–4.47) 0.85 (0.50–1.44) 0.88 (0.33–2.31) 1.19 (0.77–1.86)
.232 .018* .547 .791 .436
0.87 (0.64–1.19) 1.15 (0.84–1.56)
.386 .386
0.98 (0.92–1.05) 1.00 (1.00–1.01) 2.16 (1.17–4.01)
.596 .038* .016*
1.13 (0.58–1.59) 1.06 (0.77–1.46) 1.62 (0.81–3.23) 1.14 (0.84–1.56) 1.76 (1.23–2.53) 0.91 (0.39–2.13)
.478 .735 .172 .400 .003** .838
1.50 (1.10–2.03) 1.21 (0.87–1.67) 1.44 (1.06–1.96)
.011* .262 .023*
1.26 (0.81–1.97) 0.94 (0.57–1.56) 1.23 (0.70–2.19) 0.56 (0.33–0.95) 1.05 (0.51–2.13) n/a 2.10 (1.06–4.18) 0.92 (0.50–1.69) 2.13 (1.22–3.72)
.305 .825 .474 .037* .902 n/a .038* .794 .009**
1.08 (0.74–1.57) 1.24 (0.86–1.78) 1.17 (0.74–1.84) 1.07 (1.00–1.15) 1.03 (0.90–1.17) 2.04 (1.49–2.80)
.698 .250 .506 .052 .672 b.001***
1.44 (1.22–1.70) 1.70 (0.16–2.02) 1.38 (1.25–1.52)
.032* .002** .001**
Data presented as OR (95% CI). A logarithmic transformation of surgical drainage at 8 hours was required to normalize the data.
emerges from the use of substitute medications during these events [16,19,20]. This is particularly true for comparing the effectiveness of generic medications. For example, one recent study reviewed pressor selection and hemodynamics during a
recent shortage of ephedrine [19]. Future studies may likewise be able to retrospectively review institutional experience from a shortage to contrast commonly used but little examined medications.
522 Table 3a hours.
K.P. Blaine et al. Multivariate regression for surgical drain output at 8
εACA Female sex Mitral valve surgery Use of rescue hemostatic agent
OR (95% CI)
P
1.10 (0.97–1.26) 0.56 (0.49–0.64) 0.70 (0.59–0.84) 1.96 (1.69–2.28)
.460 b.001*** .049* b.001***
2
Multivariate regression analysis with r = 0.32. Data presented as OR (95% CI). A logarithmic transformation of surgical drainage at 8 hours was required to normalize the data. Abbreviations: epsilon-aminocaproic acid (εACA).
Limitations of the present study include a retrospective design, which is unable to identify all potential confounders. All consecutive surgeries were included, and so the control group may be considered an historical control. It is possible, though unlikely, that practice changes during the 4-month study period may account from some of the variability seen between the 2 arms. A sample size analysis suggests an adequate sample to detect a clinically relevant change, although a larger sample would have provided greater confidence in our ability to adjust for unmeasured confounders. The study also occurred at a single academic center, which may not represent practice at all institutions. Some readers may question the exclusion of the surgical estimated blood loss. That decision was justified by the uncertain nature of blood loss measurements for patients on bypass and the use of intraoperative transfusions as a surrogate measurement. The inclusion of a heterogeneous patient population better reflects modern clinical practice than a population narrowed to a single surgical indication, but also produced a large degree of unmeasured variability. The choice of postoperative surgical drainage is an appropriate endpoint that reflects the most likely clinical benefit from antifibrinolytic drugs, and the planned difference of 200 mL is meaningful to clinicians, although some may have preferred a smaller difference. One strength is that the study is free from confounding by indication because of a strictly protocolized blood conservation bundle. In summary, the present study is a single-center, retrospective review of the relative clinical effectiveness of εACA and TXA to minimize postoperative bleeding and transfusion under contemporary clinical practice. In multivariate analysis, no difference was observed for bleeding, however patients receiving εACA were more likely to require rescue hemostatic agents. The present study demonstrates how a medication shortage can be used opportunistically to review the use of older medication in current clinical practice with minimal patient risk and study expense.
5. Conclusion A retrospective analysis of the effect of εACA and TXA following cardiac surgery at a single center in the United States found no difference in chest tube drainage or transfusion
Table 3b Multivariate regression for red blood cell transfusion at 8 hours. OR (95% CI) εACA Rescue hemostatic agent Ventricular assist device insertion surgery
P
1.79 (0.79–2.73) 4.71 (2.88–7.70)
.176 .002**
12.79 (5.64–29.02)
.002**
2
Multivariate regression analysis with r = 0.14. Data presented as OR (95% CI).
requirements between the two medications, although patients receiving εACA were more likely to require rescue hemostatic agents.
Acknowledgements The authors would like to acknowledge the technical assistance of Fred Hurt, DPharm, and the operating room pharmacy staff at Stanford University Medical Center. This project was supported solely by departmental funds from the Stanford University School of Medicine Department of Anesthesiology, Pain, and Perioperative Medicine.
References [1] Stover EP, Siegel LC, Parks R, Levin J, Body SC, Maddi R, et al. Variability in transfusion practice for coronary artery bypass surgery persists despite national consensus guidelines: a 24-institution study. Institutions of the multicenter study of perioperative ischemia research group. Anesthesiology 1998;88:327-33. [2] Andreasen JJ. Pharmacologic methods to reduce postoperative bleeding in adult cardiac surgery. A mini-review. Curr Pharm Des 2013;19: 3992-5. [3] Nalla BP, Freedman J, Hare GM, Mazer CD. Update on blood conservation for cardiac surgery. J Cardiothorac Vasc Anesth 2012;26:117-33. [4] Mangano DT, Tudor IC, Dietzel C. Multicenter study of perioperative ischemia research G, ischemia R, education F the risk associated with aprotinin in cardiac surgery. N Engl J Med 2006;354:353-65. [5] Mangano DT, Miao Y, Vuylsteke A, Tudor IC, Juneja R, Filipescu D, et al. Mortality associated with aprotinin during 5 years following coronary artery bypass graft surgery. JAMA 2007;297:471-9. [6] McMullan V, Alston RP III. Aprotinin and cardiac surgery: a sorry tale of evidence misused. Br J Anaesth 2013;110:675-8. [7] Hardy JF, Belisle S, Dupont C, Harel F, Robitaille D, Roy M, et al. Prophylactic tranexamic acid and epsilon-aminocaproic acid for primary myocardial revascularization. Ann Thorac Surg 1998;65:371-6. [8] Maineri P, Covaia G, Realini M, Caccia G, Ucussich E, Luraschi M, et al. Postoperative bleeding after coronary revascularization. Comparison between tranexamic acid and epsilon-aminocaproic acid. Minerva Cardioangiol 2000;48:155-60. [9] Pinosky ML, Kennedy DJ, Fishman RL, Reeves ST, Alpert CC, Ecklund J, et al. Tranexamic acid reduces bleeding after cardiopulmonary bypass when compared to epsilon aminocaproic acid and placebo. J Card Surg 1997;12:330-8. [10] Penta de Peppo A, Pierri MD, Scafuri A, De Paulis R, Colantuono G, Caprara E, et al. Intraoperative antifibrinolysis and blood-saving
Comparative effectiveness of eACA and TXA
[11]
[12]
[13]
[14] [15]
techniques in cardiac surgery. Prospective trial of 3 antifibrinolytic drugs. Tex Heart Inst J 1995;22:231-6. Menichetti A, Tritapepe L, Ruvolo G, Speziale G, Cogliati A, Di Giovanni C, et al. Changes in coagulation patterns, blood loss and blood use after cardiopulmonary bypass: aprotinin vs tranexamic acid vs epsilon aminocaproic acid. J Cardiovasc Surg (Torino) 1996;37:401-7. Fergusson DA, Hebert PC, Mazer CD, Fremes S, MacAdams C, Murkin JM, et al. A comparison of aprotinin and lysine analogues in high-risk cardiac surgery. N Engl J Med 2008;358:2319-31. Henry DA, Carless PA, Moxey AJ, O'Connell D, Stokes BJ, Fergusson DA, et al. Anti-fibrinolytic use for minimising perioperative allogeneic blood transfusion. Cochrane Database Syst Rev 2011, CD001886. Hintze J. Pass 12. Kaysville, Utah, USA: NCSS, LLC; 2013. Koch CG, Li L, Duncan AI, Mihaljevic T, Cosgrove DM, Loop FD, et al. Morbidity and mortality risk associated with red blood cell and bloodcomponent transfusion in isolated coronary artery bypass grafting. Crit Care Med 2006;34:1608-16.
523 [16] Falana O, Patel G. Efficacy and safety of tranexamic acid versus -aminocaproic acid in cardiovascular surgery. Ann Pharmacother 2014; 48:1563-9. [17] Shander A, Hofmann A, Ozawa S, Theusinger OM, Gombotz H, Spahn DR. Activity-based costs of blood transfusions in surgical patients at four hospitals. Transfusion 2010;50:753-65. [18] McLaughlin M, Kotis D, Thomson K, Harrison M, Fennessy G, Postelnick M, et al. Effects on patient care caused by drug shortages: a survey. J Manag Care Pharm 2013;19:783-8. [19] Ladha KS, Nanji KC, Pierce E, Poon KT, Hyder JA. The impact of a shortage of pharmacy-prepared ephedrine syringes on intraoperative medication use. Anesth Analg 2015;121:404-9. [20] Thoma BN, Li J, McDaniel CM, Wordell CJ, Cavarocchi N, Pizzi LT. Clinical and economic impact of substituting dexmedetomidine for propofol due to a US drug shortage: examination of coronary artery bypass graft patients at an urban medical Centre. Pharmacoeconomics 2014;32: 149-57.