Mediastinal bleeding with significant blood loss is common during cardiac surgery, and bleeding may continue into the postoperative phase.1–3 In general, bleeding is associated with complex dysregulation of platelets and proteins, coagulation factor consumption, hemodilution, and/or consumption of fibrinogen (FIB) and other serine proteases.1–6 Historically, variables used to assess bleeding propensity and coagulation status have included platelet count (PLT), FIB concentration, prothrombin time (PT), international normalized ratio (INR), and activated partial thromboplastin time, and activated clotting time (ACT) for heparin monitoring. However, these measures do not accurately track patient coagulation status to enable accurate prediction of bleeding (<50% prediction) and show wide variation in absolute values corresponding to a given coagulation state.7 INR elevation after cardiac surgery is present in >80% of patients and does not adequately reflect bleeding risk.8 Other problems with current laboratory tests include the requirement for plasma rather than whole blood, limited information on platelet function, unacceptably long turnaround times, and moderate complexity in performance, which increases chances of human error.
Current clinical guidelines recommend the use of viscoelastic hemostatic analyzer (VHA) tests, thromboelastography (TEG™, Haemonetics Corporation, Braintree, MA), and rotational thromboelastometry (RoTEM™, Tem Systems, Inc, Research Triangle Park, NC) to assess coagulation status of cardiac surgery patients7,9–14 and guide transfusion of blood products.15 TEG™ assesses coagulation (clot strength over time) by mechanical oscillation, imitating conditions of slow or no blood flow, as occurs in surgery where blood flow is interrupted. Reaction time, or the time to clot formation, is assessed by the R value. TEG™ assesses clot strength by maximum amplitude (MA; mm), a transformed visual representation of transferred oscillatory torque, and G, an estimate of the elastic shear modulus of the clot (dynes/cm2), where G = (5000 × MA)/(100 − MA).13
Because clot formation is the change of state of blood from a viscous fluid to a semisolid gel, physical characteristics of that state change can be interrogated by sonic estimation of elasticity via resonance (SEER) sonorheometry (Quantra™, HemoSonics LLC, Charlottesville, VA). SEER sonorheometry uses high-frequency ultrasound pulses to characterize the dynamic changes in viscoelastic properties of a blood sample during coagulation.16,17 Coagulation status is assessed by variables such as clot time, heparinase clot time, clot time ratio, clot stiffness, and FIB and platelet contributions to clot stiffness.17 Clot time and heparinase clot time may be analogous to TEG™ reaction time R and TEG™ R in the presence of heparinase, respectively. Clot stiffness is measured at the clot shear modulus value 7 minutes after clot time and may be analogous to TEG™ G. Because platelets and fibrin/FIB are major determinants of whole blood clot stiffness, the relative contribution of these components can be assessed by measuring clot stiffness with and without a platelet inhibitor.
The goal of this study was to assess the potential of SEER sonorheometry technology compared with TEG™ (which is presently used clinically in our center) for precise, rapid, and near-point-of-care monitoring of hemostasis during cardiac surgery. Assessment was performed by examining paired performance measures of coagulation response during the course of cardiac surgery and cardiopulmonary bypass (CPB), as measured simultaneously by TEG™ and SEER sonorheometry. The specific aims were to (1) assess tracking of coagulation variables during the time course of cardiac surgery; (2) quantify the amount of agreement between the 2 methods, and (3) quantify testing times for the 2 technologies. This study was not designed to determine which method was superior or “best” nor was it set up to evaluate the 2 methods in terms of patient response to interventions designed to curb excessive bleeding.
The study protocol was approved by the Virginia Commonwealth University Health System IRB for human subject protection (Virginia Commonwealth University protocol number: HM 20003776; ClinicalTrials.gov identifier: NCT02392247; primary investigator: BDS; registered March 12, 2015). Fifty-two adult (≥18 years) elective patients scheduled to undergo cardiac surgery (coronary artery bypass grafting, valve procedures, combined procedures, or aortic surgery) with CPB were enrolled between May and September 2015. Patients were ineligible if they were an emergent case; received heparin or clopidogrel within 5 days of surgery; had a serum creatinine >1.5 mg/dL (indicative of renal disease); had a history of active liver disease; or were pregnant, minors, prisoners, or unable to understand and consent to the study. Informed written consent was obtained before study enrollment.
Perioperative and Postoperative Methods
Details of the perioperative and operative procedures will be described in detail in an accompanying article and are only briefly summarized here. CPB support was provided during the time of cardiac repair. Patients were initially anticoagulated with heparin (300 IU/kg IV) to a target-activated ACT of 450 seconds. Heparin was reversed after weaning from bypass with a 1:1 ratio of protamine sulfate to the heparin dose administered initially. During surgery, blood products (packed red blood cells, plasma, fresh frozen plasma, cryoprecipitate, platelets) and fluid (crystalloid, colloid) were administered per cardiac team discretion using an in-place algorithm based on TEG™, PLT count, FIB, and PT/INR. After surgery, patients were transported to the intensive care unit (ICU) while still intubated and ventilated. Intraoperative blood loss was not measured routinely, because blood suctioned from the surgical field is returned to the bypass circuit. Postoperative blood loss through mediastinal and pleural drains was recorded at 1, 8, and 24 hours after surgery (after transfer to the ICU). Clinicians were blinded to results from the hemostatic assays not in routine use; treatments for bleeding, transfusion, and coagulation dysfunction were unaffected by this protocol.
Clinical and Demographic Data
Demographics (age, sex, height, weight, and ethnicity), type of surgery, surgical and bypass time, type and number of blood products, crystalloid and colloid volumes, chest tube drainage, and preoperative medications were obtained from retrospective chart review. Total intraoperative blood loss was not routinely estimated during cardiac surgery cases. Patient electronic records were maintained in the Virginia Commonwealth University Health System computerized health information system. All study data were collected and managed using REDCap electronic data capture tools hosted at Virginia Commonwealth University.18
Blood Sampling and Hemostatic Assay Function Tests
Blood was sampled at 4 time points during surgery: (1) following anesthetic induction (baseline); (2) end of CPB (on patient rewarming, approximately 20 minutes before weaning from CPB); (3) 10 minutes after protamine administration and weaning from CPB; and (4) immediately before patient transfer to the ICU. Blood was drawn from indwelling radial arterial lines placed for routine monitoring; lines were kept patent by constant-pressure normal saline infusion without heparin added. An initial 10 mL “waste” sample was drawn and reinjected into the venous lines and not used for analysis. Ten milliliters of whole blood was then drawn in citrated and EDTA vacutainers for routine hematology and comparative determinations. Blood was collected and processed by trained laboratory technicians. Time between sample collection and completion of processing by each method was recorded for each sample. All coagulation and hematology analyses were performed in the laboratory section of the Virginia Commonwealth University Health System Surgical Services operating room.
FIB (milligrams per deciliter) was measured on a STA Compact Max (Diagnostica Stago, Parsippany, NJ). ACT (seconds) were measured with a HEMOCHRON Response System (Accriva Diagnostics San Diego, CA). PLT counts (109/L) were obtained with a Beckman Coulter Platelet counter (LH 750, Beckman Coulter, Miami, FL).
TEG™ analysis was performed on a TEG 5000 Thromboelastograph hemostasis analyzer (Haemonetics). Kaolin-activated TEG™ determinations were performed at 37°C. TEG™ variables measured were R time (minutes), K (time to reach 20 mm amplitude, minutes), MA (mm) α (angle; degrees), G (dynes/cm2), and the coagulation index (CI; calculated from a weighted sum of R, K, MA, and α). TEG™ samples were run at time points 2 and 3 to assess for residual heparin. Patient coagulation state was defined as hypercoagulable if CI > 3, hypocoagulable if CI < −3, and normal otherwise.19
SEER sonorheometry analysis was performed on a Research Use Only version of the Quantra™ Hemostasis Analyzer (HemoSonics LLC). The Quantra™ was located in a room next to the cardiac operating rooms. The Quantra™ uses a 4-channel cartridge with embedded reagents. Details on the system are described elsewhere.17 Cartridges were provided in accordance with the company’s standards. Citrated blood was drawn into a 3-mL syringe for direct introduction into the Quantra™ cartridge interface. We obtained 2 measures of coagulation activation times: clot time (time to activation of coagulation, seconds) and heparinase clot time (clot time in the presence of heparinase; seconds). Clot time values were not available during bypass, because the patient was heparinized, and no heparinase was present in this particular reagent. The ratio of clot times, clot time/heparinase clot time, is a measure of the amount of residual heparin remaining in the system; a ratio >1.4 indicates the presence of residual heparin.20 We obtained 2 measures of clot stiffness: clot stiffness and FIB contribution to clot stiffness, both measured 7 minutes after time to clot. Platelet contribution was assessed as the difference of the 2 measures (clot stiffness − FIB contribution); this is a functional measure of platelet activity. All clot stiffness-derived variables were in units of hectopascals (1 Pascal × 10–2).
We measured laboratory times for both technologies, computed as the interval between sample drop-off and completion of analysis (minutes); we also calculated the additional time between the end of analysis and the time that results were made available in the patient’s electronic record.
We estimated that a cohort of 50 patients would be required to capture at least 7 “bleeding” patients with a probability of 95%, assuming a 25% incidence of bleeding, and SD of 20%, to determine maximum clot stiffness within ±10% of the mean.21
Clot strength and coagulation activation time trajectories were assessed for G, R, clot stiffness, FIB contribution, platelet contribution, clot time, and heparinase clot time with repeated-measures mixed models, with time as a fixed effect, patients as random effects, and an unstructured covariance matrix to model the time dependencies between observations. Planned comparisons were made between baseline and each subsequent time point, and familywise error rates were controlled by Bonferroni correction adjusted for multiple comparisons by setting statistical significance at P = 0.012. Correlations between the 2 methods were extracted from the corresponding response variance-covariance and correlation matrices; thus, the overall correlation between methods was expressed in terms of the variance components of the linear mixed model while accounting for the correlation between the repeated measures.22 Calculations were performed in SAS PROC MIXED (version 12, SAS Inc , Cary, NC).
Method comparisons were performed according to Clinical Laboratory Science Institute guidelines.23 The TEG™ metric R was compared with SEER clot time and heparinase clot time (minutes), and TEG™ metric G was compared with SEER clot stiffness metrics, clot stiffness and FIB contribution (hPa). G was used instead of MA because it is expressed in units of force per unit area (dynes per square centimeter), and thus, it is more readily converted to units comparable with those used to describe clot stiffness with the SEER Quantra (because 1 dyne/cm2 = 1 Pa; 1 hPa = 1 Pa × 10–2). Because of time dependencies in the data, each time period was assessed independently. However, data were also pooled over all 4 time periods and reanalyzed to assess strength of the association as influenced by pseudoreplication (i.e., the effect of the conflation of the number of observations with the number of independent patients24,25). Agreement between paired measures was assessed by Deming (orthogonal) regression, as recommended by Clinical Laboratory Science Institute guidelines.23 Deming regression quantifies the specific magnitude and direction of bias between the 2 methods; it assumes unequal variances between both measurement axes. The term “bias” in this context is in terms of location and scale shifts between methods and is quantified as the deviation of the intercept and slope of the paired line of relationship from the line of identity. The line of identity represents “perfect” agreement between methods, as indicated by a line at 45° through the origin. Proportional bias between methods was indicated by a statistically significant (P < 0.05) deviation of the slope from 1, and constant bias was indicated by statistically significant deviation of the intercept from 0.23 Confidence intervals were obtained by bootstrapping 1000 times on the point estimates for the slope, intercept, and correlation coefficient and obtaining the 2.5th and 97.5th percentiles from the resulting distributions.26 The relationships of FIB contribution with FIB and platelet contribution with PLT were also quantified by orthogonal regression. Calculations were performed in JMP Pro 12 (SAS Institute, Cary, NC).
A consecutive nonrandomized cohort of 52 cardiac surgery patients was enrolled. Two patients were excluded, one because consent was withdrawn and the second because the surgery was cancelled after consent had been obtained.
The average patient age was 60 (SD, 14) years with an average body mass index of 30 (SD, 7) kg/m2; 28 of 50 (56%) were men, and 43 of 50 (86%) were taking aspirin and/or other preoperative medications known to affect coagulation. The median surgery time was 5 hours (interquartile range [IQR], 4.4–6.4 hours); median time on bypass was 2.3 hours (IQR, 1.8–2.9 hours), and 22 of 50 patients (44%) required blood product administration during bypass. The median chest tube drainage was 820 mL (IQR, 563–1226 mL) over 24 hours. “Bleeding” defined as chest tube drainage >1500 mL in 24 hours occurred in only 8 of 50 patients; however, no patient required a return to the operating room for reexploration.
Figure 1 shows clot stiffness trajectories as estimated by TEG™ and SEER. Both methods followed similar trajectories in clot strength during the course of bypass surgery; G, clot stiffness, and FIB contribution (Fig. 1A) all exhibited levels that were highest at baseline and then stabilized at levels 30% below baseline (P < 0.001 for all). SEER clot stiffness values averaged 10 hPa higher than corresponding G at all time points. The contribution of PLTs to clot stiffness (Fig. 1B) showed similar trajectories to those estimated by overall measures of clot stiffness. Correlations between methods, accounting for repeated measures, were 0.89 and 0.82 for clot stiffness with G and FIB contribution with G, respectively.
Reaction time R was 1 to 2.5 minutes longer than corresponding clot times (Fig. 2). R in the presence of heparinase increased during bypass from 4.5 minutes at baseline to 6.2 minutes (P < 0.001); R subsequently decreased by 1.3 minutes (P < 0.001) during weaning and on ICU transfer. Clot times were relatively uniform throughout, averaging 3.1 (SD, 0.6) minutes; heparinase clot times averaged 3.5 (SD, 0.8) minutes. Correlations between methods, accounting for repeated measures, were 0.55 and 0.67 for clot time with R (kaolin) and heparinase clot time with R (heparinase), respectively.
Clot time ratios were stable during the time course of surgery; median clot time ratios were 1.07 (IQR, 1.02–1.14) at baseline; 1.08 (IQR, 1.03–1.18) after heparin reversal, and 1.06 (IQR, 1.01–1.11) at ICU transfer; differences between baseline and time period 3 (10 minutes after heparin reversal) and 4 (transfer to ICU) were not statistically significant (P = 0.19 and P = 0.58, respectively). Clot ratios >1.4, the threshold indicative of residual heparin remaining after protamine administration, were not observed. Scoring of coagulation status on the basis of TEG™ CI indicated that 14 of 50 patients were hypercoagulable (CI > 3) at baseline, and 4 patients were hypocoagulable (TEG™ CI < −3) during bypass and weaning, with 2 still hypocoagulable at transfer to the ICU.
Figure 3 shows clot stiffness and FIB contribution to clot stiffness (measured by SEER sonorheometry) compared with G (measured by TEG™) with line of best fit computed by Deming (orthogonal) regression; corresponding regression coefficients and correlation coefficients are given in Table 1. The line of identity (perfect agreement) is indicated by the dashed line. Although correlation was good (r > 0.6 for all time periods, r > 0.78 for pooled data; Table 1), the slope of clot stiffness with G was >1, with a greater mean difference over G of 2 to 20 hPa and increasing by 3 hPa with every unit change in G (Table 1). In contrast, the slope of FIB contribution with G did not differ from 1, indicating similar unit changes of each variable; however, G consistently tracked 2 to 5 hPa lower than the corresponding FIB contribution measurements (Fig. 3; Table 1). FIB contribution to clot stiffness as measured by SEER increased by 2 to 3 Pa for each unit change in FIB; platelet contribution to clot stiffness increased by 16 Pa for each unit change in platelets (Fig. 4; Table 1).
Median time to obtain TEG™ analysis results was 42 minutes (IQR, 36–50 minutes; minimum time, 22 minutes; maximum time, 160 minutes). An additional 2 minutes (IQR, 1–4 minutes; minimum time, <1 minutes; maximum time, 33 minutes) was required for the entry of TEG™ data into the hospital electronic medical record system. Median processing time for Quantra was 11 minutes (IQR, 10.5–11.5 minutes, minimum time, 9 minutes; maximum time, 17 minutes; Viola, personal communication); an additional 5 minutes (minimum time, <1 minutes; maximum time, 23 minutes) was incurred between specimen drop-off and the beginning of processing, indicating total elapsed time of approximately 16 to 34 minutes.
This study was conducted on a small cohort of cardiac surgery patients during the course of 5 hours of cardiac surgery and weaning from bypass. We tracked historically used coagulation test measures together with markers evaluated by 2 quite distinct whole blood VHA technologies, TEG™ and SEER sonorheometry. There is no “gold standard” test available as yet, and all currently used coagulation testing technologies have differences and artifact introduced from the respective methodologies. Therefore, methods were compared only to document similarity of trajectories of clotting variables during the time course of cardiac surgery and quantify strength of association between paired measurements. This study was not designed to determine which method is “superior,” and in any case, differences in physical properties of the techniques precluded strict comparisons.
In general, whole-blood VHA tests are appealing because they capture the biology of clot formation as a comprehensive cellular event. Separation of PLTs, erythrocytes, and white cells from plasma contributes to the lack of bleeding-predictive capability of historical standard tests such as PT/INR and activated partial thromboplastin time.7 Current VHA technologies monitor clotting dynamics in a low-shear environment, and therefore, lack the flow and shear characteristics required for normal biologic activation of coagulation. Current systems also lack the capacity to monitor endothelial cells, which sense shear, along with platelets. Clot strength as measured by TEG™ (MA, G) infers clot shear modulus based on measures of oscillatory torque.27 Determination of coagulation properties through SEER sonorheometry is a new assessment tool for whole-blood VHA testing. The propagating shear wave is induced by ultrasound radiation force, and then the shear modulus of the evolving blood clot is measured using low-energy ultrasound pulses to quantify displacement associated with the resonance of the propagating shear wave.16,17
Results from the method comparisons analyses in this study must be interpreted with caution. Although TEG™ and SEER measures show positive association in measures of clotting function, relational differences (quantified as bias) show that measurements are not substitutable. In this study, there is no gold standard, so bias in this context refers to the amount by which SEER measurements deviate from the TEG™ comparison set. In all likelihood, the deviations observed between techniques result from the very different underlying mathematical and physical relationships governing the devices. First, as described earlier, measurement techniques used in this study were technically distinct, with different physical principles of operation. Second, the variables compared in this study, although expressed in the same measurement units, also implicitly assume equivalent measures of clot strength or stiffness, which may not be valid in practice. For example, although Quantra™ clot stiffness is inferred from shear wave displacement after an ultrasound pulse, MA and G are essentially measures of clot strength based on mechanical stretch. Biochemical differences in test substrates and reagents may be a factor. It is also possible that FIB contribution measured by SEER determines a more “functional” FIB activity. The plasma-based Clauss method of FIB testing has been criticized as being inaccurate in determining patient status during and immediately after CPB.10,28 Biochemical reasons for these inaccuracies are not completely determined, but such artifacts make the use of plasma-based coagulation tests of little clinical value.7
Treatment of bleeding patients requires speed and accuracy in determining coagulation function. The speed of SEER testing was in the time range of routine ACT performed by perfusionists. However, the TEG™ system used in this study was not automated, and few physicians using TEG™ wait for full development; clinically useful information with R/K and A/MA is usually available within 10 minutes. Many centers are using tissue factor-activated rapid TEG™ instead. Therefore, the differences in time noted here between technologies will need to be reassessed with more recent generation TEG™ devices.
Current VHA technologies have a well-established literature demonstrating utility for coagulopathy guidance.7,9 The Quantra™ system is a new way of assessing whole blood coagulation, relying on pulsed ultrasound rather than mechanical oscillation, as is the case for TEG™ systems. However, at present, the Quantra™ is not Food and Drug Administration approved and not available for purchase or clinical use either in the United State or Europe. Future research using next-generation Quantra™ systems should be focused on further assessments of clinical utility for prediction of bleeding and the development of algorithm-based coagulation treatment schemes.
Name: Penny S. Reynolds, PhD.
Contribution: This author helped design the study, analyzed the data, co-wrote the manuscript, and approved the final manuscript.
Conflicts of Interest: Penny S. Reynolds has no conflicts of interest to declare.
Name: Paul Middleton, MD.
Contribution: This author has seen the study data, helped design the study, collected the data, and helped write the manuscript.
Conflicts of Interest: Paul Middleton has no conflicts of interest to declare.
Name: Harry McCarthy, CCP.
Contribution: This author helped design the study, collect the data, and write the manuscript.
Conflicts of Interest: Harry McCarthy has no conflicts of interest to declare.
Name: Bruce D. Spiess, MD.
Contribution: This author has seen the study data, helped design the study, reviewed the analysis of the data, and co-wrote the manuscript, and approved the final manuscript.
Conflicts of Interest: Bruce D. Spiess has received speaking honoraria from Haemonetics in the past regarding coagulation/monitoring lecturing. Dr. Spiess declares that this project was sponsored by a research grant from HemoSonics to VCUHS in support of the research. Dr. Spiess has stock options in HemoSonics LLC and serves on their medical advisory board.
This manuscript was handled by: Roman M. Sniecinski, MD.
1. Bessen MW, Kelin AAThe coagulopathy of cardiopulmonary bypass.Crit Rev Clin Lab Sci201047197–212
2. Yavari M, Becker RCCoagulation and fibrinolytic protein kinetics in cardiopulmonary bypass.J Thromb Thrombolysis20092795–104
3. Levy JHSpiess BDThe relationships between coagulation, inflammation and endothelium: Inflammation responds.The Relationship Between Coagulation, Inflammation and Endothelium—A Pyramid Towards Outcome. Society of Cardiovascular Anesthesiologists Monograph2000Baltimore, MDLippincott Williams and Wilkins91–106
4. LaPar DJ, Isbell JM, Mulloy DP, Stone ML, Kern JA, Ailawadi G, Kron ILPlanned cardiac reexploration in the intensive care unit is a safe procedure.Ann Thorac Surg2014981645–52
5. Moulton MJ, Creswell LL, Mackey ME, Cox JL, Rosenbloom MReexploration for bleeding is a risk factor for adverse outcomes after cardiac operations.J Thorac Cardiovasc Surg19961111037–46
6. Stokes ME, Ye X, Shah M, Mercaldi K, Reynolds MW, Rupnow MF, Hammond JImpact of bleeding-related complications and/or blood product transfusions on hospital costs in inpatient surgical patients.BMC Health Serv Res201111135
7. Welsby IJ, Jiao K, Ortel TL, Brudney CS, Roche AM, Bennett-Guerrero E, Gan TJThe kaolin-activated thrombelastograph predicts bleeding after cardiac surgery.J Cardiothorac Vasc Anesth200620531–5
8. Meesters MI, Vonk AB, van de Weerdt EK, Kamminga S, Boer CLevel of agreement between laboratory and point-of-care prothrombin time before and after cardiopulmonary bypass in cardiac surgery.Thromb Res20141331141–4
9. Johansson PI, Sølbeck S, Genet G, Stensballe J, Ostrowski SRCoagulopathy and hemostatic monitoring in cardiac surgery: an update.Scand Cardiovasc J201246194–202
10. Whiting P, Al M, Westwood M, Ramos IC, Ryder S, Armstrong N, Misso K, Ross J, Severens J, Kleijnen JViscoelastic point-of-care testing to assist with the diagnosis, management and monitoring of haemostasis: a systematic review and cost-effectiveness analysis.Health Technol Assess2015191–228, v–vi
11. Meyer AS, Meyer MA, Sørensen AM, Rasmussen LS, Hansen MB, Holcomb JB, Cotton BA, Wade CE, Ostrowski SR, Johansson PIThrombelastography and rotational thromboelastometry early amplitudes in 182 trauma patients with clinical suspicion of severe injury.J Trauma Acute Care Surg201476682–90
12. Haas T, Görlinger K, Grassetto A, Agostini V, Simioni P, Nardi G, Ranucci MThromboelastometry for guiding bleeding management of the critically ill patient: a systematic review of the literature.Minerva Anestesiol2014801320–35
13. Solomon C, Ranucci M, Hochleitner G, Schöchl H, Schlimp CJAssessing the methodology for calculating platelet contribution to clot strength (platelet component) in thromboelastometry and thrombelastography.Anesth Analg2015121868–78
14. Weber CF, Görlinger K, Meininger D, Herrmann E, Bingold T, Moritz A, Cohn LH, Zacharowski KPoint-of-care testing: a prospective, randomized clinical trial of efficacy in coagulopathic cardiac surgery patients.Anesthesiology2012117531–47
15. Ferraris VA, Brown JR, Despotis GJ, Hammon JW, Reece B, Saha SP, Song HK, Clough ER2011 Update to The Society of Thoracic Surgeons and the Society of Cardiovascular Anesthesiologists blood conservation clinical practice guidelines.Ann Thorac Surg201191944–82
16. Corey FS, Walker WFSonic estimation of elasticity via resonance: a new method of assessing hemostasis.Ann Biomed Eng20151–20
17. Ferrante EA, Givens TB, Lloyd CA, Fischer TJ, Viola FA novel device for the evaluation of hemostatic function in critical care settings.Anesth Analg.20161231372–1379.
18. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JGResearch electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support.J Biomed Inform200942377–81
19. Thromboelastography™ (TEG5000) System User’s ManualBraintree, MAHaemonetics Inc2006-P/N-510-US-Manual Revision AD
20. Mauldin FW Jr, Viola F, Hamer TC, Ahmed EM, Crawford SB, Haverstick DM, Lawrence MB, Walker WFAdaptive force sonorheometry for assessment of whole blood coagulation.Clin Chim Acta2010411638–44
21. Crow EL, Davis FA, Maxfield MWChapter 8. Acceptance sampling.Statistics Manual with Examples Taken from Ordnance Development1960Dover Publications Inc209–26
22. Hamlett A, Ryan L, Serrano-Trespalacios P, Wolfinger RMixed models for assessing correlation in the presence of replication.J Air Waste Manag Assoc200353442–50
23. Budd JR, Durham AP, Gwise TE, Iriarte B, Kallner A, Linnet K, Magari R, Vaks JEMeasurement Procedure Comparison and Bias Estimation Using Patient Samples. Approved Guideline—3rd Edition201333Clinical Laboratory Standards Institute1–79
24. Lazic SEThe problem of pseudoreplication in neuroscientific studies: is it affecting your analysis?BMC Neurosci2010115
25. Hurlbert SHPseudoreplication and the design of ecological field experiments.Ecol Monogr198454187–211
26. Carpenter J, Bithell JBootstrap confidence intervals: when, which, what? A practical guide for medical statisticians.Stat Med2000191141–64
27. Hochleitner G, Sutor K, Levett C, Leyser H, Schlimp CJ, Solomon CRevisiting Hartert’s 1962 calculation of the physical constants of thrombelastography.Clin Appl Thromb Hemost20151–10
28. Erdoes G, Gerster G, Colucci G, Kaiser H, Alberio L, Eberle BPrediction of post-weaning fibrinogen status during cardiopulmonary bypass: an observational study in 110 patients.PLoS One201510e0126692